From e3f47f89a57ef115755184a8b3f03a47ee227418 Mon Sep 17 00:00:00 2001 From: Daniel Drake Date: Wed, 10 Dec 2008 23:28:25 +0200 Subject: [PATCH 001/123] USB: unusual_devs.h additions for Pentax K10D Jaak Ristioja reported problems with his Pentax K10D camera: https://bugs.gentoo.org/show_bug.cgi?id=250406 /proc/bus/usb/devices: T: Bus=02 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 2 Spd=12 MxCh= 0 D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs= 1 P: Vendor=0a17 ProdID=006e Rev= 1.00 S: Manufacturer=PENTAX Corporation S: Product=K10D C:* #Ifs= 1 Cfg#= 1 Atr=c0 MxPwr= 2mA I:* If#= 0 Alt= 0 #EPs= 3 Cls=08(stor.) Sub=06 Prot=50 Driver=usb-storage E: Ad=81(I) Atr=02(Bulk) MxPS= 64 Ivl=0ms E: Ad=02(O) Atr=02(Bulk) MxPS= 64 Ivl=0ms E: Ad=83(I) Atr=03(Int.) MxPS= 64 Ivl=100ms The number of reported sectors is off-by-one. Signed-off-by: Daniel Drake Cc: Kadianakis George Cc: stable Signed-off-by: Phil Dibowitz --- drivers/usb/storage/unusual_devs.h | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index bfcc1fe8251..a54212f0a51 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -1425,6 +1425,13 @@ UNUSUAL_DEV( 0x0a17, 0x006, 0x0000, 0xffff, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_INQUIRY ), +/* Reported by Jaak Ristioja */ +UNUSUAL_DEV( 0x0a17, 0x006e, 0x0100, 0x0100, + "Pentax", + "K10D", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_FIX_CAPACITY ), + /* These are virtual windows driver CDs, which the zd1211rw driver * automatically converts into WLAN devices. */ UNUSUAL_DEV( 0x0ace, 0x2011, 0x0101, 0x0101, From e2673b28911a43257265523e3672861be6e44093 Mon Sep 17 00:00:00 2001 From: Nguyen Anh Quynh Date: Thu, 11 Dec 2008 15:04:11 -0800 Subject: [PATCH 002/123] USB: another unusual_devs entry for another bad Argosy storage device I have another Argosy USB storage device, which has the same problem with the Argosy USB storage device already fixed in 2.6.27.7. But this device has another product ID (840:84), so this patch adds a new entry into unusual_devs to fix the mount problem. I enclose here two patches: one against 2.6.27.8, and another against the latest linus-git tree. The information about the Argosy device is like below: #lsusb -v -d 840:84 Bus 005 Device 005: ID 0840:0084 Argosy Research, Inc. Device Descriptor: bLength 18 bDescriptorType 1 bcdUSB 2.00 bDeviceClass 0 (Defined at Interface level) bDeviceSubClass 0 bDeviceProtocol 0 bMaxPacketSize0 64 idVendor 0x0840 Argosy Research, Inc. idProduct 0x0084 bcdDevice 0.01 iManufacturer 1 Generic iProduct 2 USB 2.0 Storage Device iSerial 3 8400000000002549 bNumConfigurations 1 Configuration Descriptor: bLength 9 bDescriptorType 2 wTotalLength 32 bNumInterfaces 1 bConfigurationValue 1 iConfiguration 0 bmAttributes 0xc0 Self Powered MaxPower 2mA Interface Descriptor: bLength 9 bDescriptorType 4 bInterfaceNumber 0 bAlternateSetting 0 bNumEndpoints 2 bInterfaceClass 8 Mass Storage bInterfaceSubClass 6 SCSI bInterfaceProtocol 80 Bulk (Zip) iInterface 0 Endpoint Descriptor: bLength 7 bDescriptorType 5 bEndpointAddress 0x01 EP 1 OUT bmAttributes 2 Transfer Type Bulk Synch Type None Usage Type Data wMaxPacketSize 0x0200 1x 512 bytes bInterval 0 Endpoint Descriptor: bLength 7 bDescriptorType 5 bEndpointAddress 0x82 EP 2 IN bmAttributes 2 Transfer Type Bulk Synch Type None Usage Type Data wMaxPacketSize 0x0200 1x 512 bytes bInterval 0 Device Qualifier (for other device speed): bLength 10 bDescriptorType 6 bcdUSB 2.00 bDeviceClass 0 (Defined at Interface level) bDeviceSubClass 0 bDeviceProtocol 0 bMaxPacketSize0 64 bNumConfigurations 1 Device Status: 0x0000 (Bus Powered) Before the patch, dmesg returns a lot of information like below (my dmesg is overflown): .... [ 138.833390] sd 7:0:0:0: [sdb] Add. Sense: No additional sense information [ 138.877631] sd 7:0:0:0: [sdb] Sense Key : No Sense [current] [ 138.877643] sd 7:0:0:0: [sdb] Add. Sense: No additional sense information [ 138.921906] sd 7:0:0:0: [sdb] Sense Key : No Sense [current] [ 138.921923] sd 7:0:0:0: [sdb] Add. Sense: No additional sense information .... After the fix, dmesg returns below information: .... usb 5-1: new high speed USB device using ehci_hcd and address 5 usb 5-1: configuration #1 chosen from 1 choice scsi7 : SCSI emulation for USB Mass Storage devices usb-storage: device found at 5 usb-storage: waiting for device to settle before scanning usb-storage: device scan complete scsi 7:0:0:0: Direct-Access HTS54808 0M9AT00 MG4O PQ: 0 ANSI: 0 sd 7:0:0:0: [sdb] 156301488 512-byte hardware sectors (80026 MB) sd 7:0:0:0: [sdb] Write Protect is off sd 7:0:0:0: [sdb] Mode Sense: 03 00 00 00 sd 7:0:0:0: [sdb] Assuming drive cache: write through sd 7:0:0:0: [sdb] 156301488 512-byte hardware sectors (80026 MB) sd 7:0:0:0: [sdb] Write Protect is off sd 7:0:0:0: [sdb] Mode Sense: 03 00 00 00 sd 7:0:0:0: [sdb] Assuming drive cache: write through sdb: sdb1 sd 7:0:0:0: [sdb] Attached SCSI disk sd 7:0:0:0: Attached scsi generic sg1 type 0 kjournald starting. Commit interval 5 seconds EXT3 FS on sdb1, internal journal EXT3-fs: recovery complete. EXT3-fs: mounted filesystem with ordered data mode. Cc: Kuniyasu Suzaki Signed-off-by: Nguyen Anh Quynh Signed-off-by: Andrew Morton Cc: stable Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/unusual_devs.h | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index a54212f0a51..af345042294 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -1320,6 +1320,13 @@ UNUSUAL_DEV( 0x0840, 0x0082, 0x0001, 0x0001, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_CAPACITY), +/* Reported and patched by Nguyen Anh Quynh */ +UNUSUAL_DEV( 0x0840, 0x0084, 0x0001, 0x0001, + "Argosy", + "Storage", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_FIX_CAPACITY), + /* Entry and supporting patch by Theodore Kilgore . * Flag will support Bulk devices which use a standards-violating 32-byte * Command Block Wrapper. Here, the "DC2MEGA" cameras (several brands) with From b16363991414a6025beb7269f9c1dd294f9b241f Mon Sep 17 00:00:00 2001 From: Oliver Neukum Date: Fri, 12 Dec 2008 11:01:45 +0100 Subject: [PATCH 003/123] USB: storage: extend unusual range for 067b:3507 This device has been released in a new revision which is still buggy. Signed-off-by: Oliver Neukum Signed-off-by: Phil Dibowitz Cc: stable Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/unusual_devs.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index af345042294..a7b2db73964 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -1040,7 +1040,7 @@ UNUSUAL_DEV( 0x067b, 0x2507, 0x0100, 0x0100, US_FL_FIX_CAPACITY | US_FL_GO_SLOW ), /* Reported by Alex Butcher */ -UNUSUAL_DEV( 0x067b, 0x3507, 0x0001, 0x0001, +UNUSUAL_DEV( 0x067b, 0x3507, 0x0001, 0x0101, "Prolific Technology Inc.", "ATAPI-6 Bridge Controller", US_SC_DEVICE, US_PR_DEVICE, NULL, From b8d23491f127aa0cd1863bd6cb58e771c558b762 Mon Sep 17 00:00:00 2001 From: Paulo Afonso Graner Fessel Date: Fri, 12 Dec 2008 12:05:18 +0100 Subject: [PATCH 004/123] USB: storage: recognizing and enabling Nokia 5200 cell phoes This patch corrects the issue when one connects a Nokia 5200 cell phone in data storage mode. If one uses an unpatched unusual_devs.h, the following messages appear on /var/log/messages: Dec 12 01:03:24 alberich kernel: usb 4-2: new full speed USB device using uhci_hcd and address 3 Dec 12 01:03:25 alberich kernel: usb 4-2: configuration #1 chosen from 1 choice Dec 12 01:03:25 alberich kernel: scsi10 : SCSI emulation for USB Mass Storage devices Dec 12 01:03:25 alberich kernel: usb 4-2: New USB device found, idVendor=0421, idProduct=04bd Dec 12 01:03:25 alberich kernel: usb 4-2: New USB device strings: Mfr=1, Product=2, SerialNumber=3 Dec 12 01:03:25 alberich kernel: usb 4-2: Product: Nokia 5200 Dec 12 01:03:25 alberich kernel: usb 4-2: Manufacturer: Nokia Dec 12 01:03:25 alberich kernel: usb 4-2: SerialNumber: 353930018354523 Dec 12 01:03:25 alberich kernel: usbcore: registered new interface driver ub Dec 12 01:03:30 alberich kernel: scsi 10:0:0:0: Direct-Access Nokia Nokia 5200 0000 PQ: 0 AN SI: 4 Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] 3985409 512-byte hardware sectors (2041 MB) Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Write Protect is off Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Assuming drive cache: write through Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] 3985409 512-byte hardware sectors (2041 MB) Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Write Protect is off Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Assuming drive cache: write through Dec 12 01:03:30 alberich kernel: sdg: sdg1 Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Attached SCSI removable disk Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: Attached scsi generic sg9 type 0 Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Sense Key : No Sense [current] Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Add. Sense: No additional sense information Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Sense Key : No Sense [current] Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Add. Sense: No additional sense information Dec 12 01:03:30 alberich kernel: sd 10:0:0:0: [sdg] Sense Key : No Sense [current] (...) The MicroSD card in the phone remains inaccessible and finally the cell phone turns itself off. The patch solves this problem and makes the cell phone fully accessible: [root@alberich kernel-linus-2.6.27.5-1mdv]# df -h Sist. Arq. Tam Usad Disp Uso% Montado em /dev/sda6 31G 5,2G 26G 17% / /dev/sda1 92M 27M 61M 31% /boot /dev/mapper/homevg-homelv 240G 237G 3,5G 99% /home /dev/sda3 21G 7,9G 13G 40% /mnt/windows /dev/sdg1 2,0G 287M 1,7G 15% /media/disk <-------- I've found necessary to use the FL_US_CAPACITY_FIX switch, as without it the cell phone is recognized but it went berserk when performing low-level functions on it (a fdisk -l /dev/uba for example). lsusb -v output follows: Bus 004 Device 004: ID 0421:04bd Nokia Mobile Phones Device Descriptor: bLength 18 bDescriptorType 1 bcdUSB 2.00 bDeviceClass 0 (Defined at Interface level) bDeviceSubClass 0 bDeviceProtocol 0 bMaxPacketSize0 64 idVendor 0x0421 Nokia Mobile Phones idProduct 0x04bd bcdDevice 6.03 iManufacturer 1 Nokia iProduct 2 Nokia 5200 iSerial 3 353930018354523 bNumConfigurations 1 Configuration Descriptor: bLength 9 bDescriptorType 2 wTotalLength 32 bNumInterfaces 1 bConfigurationValue 1 iConfiguration 0 bmAttributes 0xc0 Self Powered MaxPower 100mA Interface Descriptor: bLength 9 bDescriptorType 4 bInterfaceNumber 0 bAlternateSetting 0 bNumEndpoints 2 bInterfaceClass 8 Mass Storage bInterfaceSubClass 6 SCSI bInterfaceProtocol 80 Bulk (Zip) iInterface 0 Endpoint Descriptor: bLength 7 bDescriptorType 5 bEndpointAddress 0x81 EP 1 IN bmAttributes 2 Transfer Type Bulk Synch Type None Usage Type Data wMaxPacketSize 0x0040 1x 64 bytes bInterval 0 Endpoint Descriptor: bLength 7 bDescriptorType 5 bEndpointAddress 0x01 EP 1 OUT bmAttributes 2 Transfer Type Bulk Synch Type None Usage Type Data wMaxPacketSize 0x0040 1x 64 bytes bInterval 0 Device Status: 0x0001 Self Powered Signed-off-by: Paulo Afonso Graner Fessel Signed-off-by: Phil Dibowitz Cc: stable Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/unusual_devs.h | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index a7b2db73964..a0785404659 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -260,6 +260,13 @@ UNUSUAL_DEV( 0x0421, 0x04b9, 0x0500, 0x0551, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_CAPACITY ), +/* Reported by Paulo Fessel */ +UNUSUAL_DEV( 0x0421, 0x04bd, 0x0000, 0x9999, + "Nokia", + "5200", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_FIX_CAPACITY ), + /* Reported by Richard Nauber */ UNUSUAL_DEV( 0x0421, 0x04fa, 0x0550, 0x0660, "Nokia", From 3b438e30c686b1e904c759d3354d335050ab33f9 Mon Sep 17 00:00:00 2001 From: Sergey Ovcharenko Date: Wed, 17 Dec 2008 19:28:06 +0300 Subject: [PATCH 005/123] USB: storage: unusual_devs.h additions for Macpower MasterBox Jason Johnston reported these problems with his external USB hard drive: http://bugs.gentoo.org/show_bug.cgi?id=250789 The number of reported sectors is off-by-one. /proc/bus/usb/devices: T: Bus=01 Lev=02 Prnt=08 Port=00 Cnt=01 Dev#= 9 Spd=480 MxCh= 0 D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs= 1 P: Vendor=0dc4 ProdID=0073 Rev= 0.00 S: Manufacturer=Macpower Technology Co.LTD. S: Product=USB 2.0 3.5" DEVICE S: SerialNumber=03006C C:* #Ifs= 1 Cfg#= 1 Atr=c0 MxPwr=100mA I:* If#= 0 Alt= 0 #EPs= 2 Cls=08(stor.) Sub=06 Prot=50 Driver=usb-storage E: Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms E: Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms Signed-off-by: Sergey Ovcharenko Signed-off-by: Phil Dibowitz Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/unusual_devs.h | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index a0785404659..8a0eb00024d 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -1537,6 +1537,13 @@ UNUSUAL_DEV( 0x0d96, 0x5200, 0x0001, 0x0200, "JD 5200 z3", US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_INQUIRY), +/* Reported by Jason Johnston */ +UNUSUAL_DEV( 0x0dc4, 0x0073, 0x0000, 0x0000, + "Macpower Technology Co.LTD.", + "USB 2.0 3.5\" DEVICE", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_FIX_CAPACITY), + /* Reported by Lubomir Blaha * I _REALLY_ don't know what 3rd, 4th number and all defines mean, but this * works for me. Can anybody correct these values? (I able to test corrected From e64a5219296fff11a0da0322b43012265cd3b327 Mon Sep 17 00:00:00 2001 From: roel kluin Date: Tue, 21 Oct 2008 00:47:23 -0400 Subject: [PATCH 006/123] USB: clean up redundant tests on unsigned temp, bytes and param->{length,sglen,vary} are unsigned so these tests do nothing. Signed-off-by: Roel Kluin Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/ehci-dbg.c | 8 ++------ drivers/usb/misc/usbtest.c | 2 -- 2 files changed, 2 insertions(+), 8 deletions(-) diff --git a/drivers/usb/host/ehci-dbg.c b/drivers/usb/host/ehci-dbg.c index 0cb53ca8d34..7f4ace73d44 100644 --- a/drivers/usb/host/ehci-dbg.c +++ b/drivers/usb/host/ehci-dbg.c @@ -455,9 +455,7 @@ static void qh_lines ( (scratch >> 16) & 0x7fff, scratch, td->urb); - if (temp < 0) - temp = 0; - else if (size < temp) + if (size < temp) temp = size; size -= temp; next += temp; @@ -466,9 +464,7 @@ static void qh_lines ( } temp = snprintf (next, size, "\n"); - if (temp < 0) - temp = 0; - else if (size < temp) + if (size < temp) temp = size; size -= temp; next += temp; diff --git a/drivers/usb/misc/usbtest.c b/drivers/usb/misc/usbtest.c index 444c69c447b..5f1a19d1497 100644 --- a/drivers/usb/misc/usbtest.c +++ b/drivers/usb/misc/usbtest.c @@ -192,8 +192,6 @@ static struct urb *simple_alloc_urb ( { struct urb *urb; - if (bytes < 0) - return NULL; urb = usb_alloc_urb (0, GFP_KERNEL); if (!urb) return urb; From 8e8ce4b642ce4beb865fb1d67cd8ea883d9ae7d7 Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Mon, 20 Oct 2008 21:46:01 -0700 Subject: [PATCH 007/123] USB: use pci_ioremap_bar() in drivers/usb Use the newly introduced pci_ioremap_bar() function in drivers/usb. pci_ioremap_bar() just takes a pci device and a bar number, with the goal of making it really hard to get wrong, while also having a central place to stick sanity checks. Signed-off-by: Arjan van de Ven Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/pci-quirks.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/drivers/usb/host/pci-quirks.c b/drivers/usb/host/pci-quirks.c index ae6e70edd74..15e4f1d92c7 100644 --- a/drivers/usb/host/pci-quirks.c +++ b/drivers/usb/host/pci-quirks.c @@ -172,9 +172,9 @@ static void __devinit quirk_usb_handoff_ohci(struct pci_dev *pdev) if (!mmio_resource_enabled(pdev, 0)) return; - base = ioremap_nocache(pci_resource_start(pdev, 0), - pci_resource_len(pdev, 0)); - if (base == NULL) return; + base = pci_ioremap_bar(pdev, 0); + if (base == NULL) + return; /* On PA-RISC, PDC can leave IR set incorrectly; ignore it there. */ #ifndef __hppa__ @@ -221,9 +221,9 @@ static void __devinit quirk_usb_disable_ehci(struct pci_dev *pdev) if (!mmio_resource_enabled(pdev, 0)) return; - base = ioremap_nocache(pci_resource_start(pdev, 0), - pci_resource_len(pdev, 0)); - if (base == NULL) return; + base = pci_ioremap_bar(pdev, 0); + if (base == NULL) + return; cap_length = readb(base); op_reg_base = base + cap_length; From 3a4e72cbf2ac4435630a2b03bd25e60ef5967e99 Mon Sep 17 00:00:00 2001 From: Vikram Pandita Date: Fri, 24 Oct 2008 23:41:30 +0530 Subject: [PATCH 008/123] USB: Avoid 20ms delay in EHCI resume For function ehci_bus_resume() - Added flag resume_needed No need to wait for 20ms if no port was suspended - Change mdelay to msleep - release and reacquire the spinlock around mdelay Signed-off-by: vikram pandita Acked-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/ehci-hub.c | 18 +++++++++++++++--- 1 file changed, 15 insertions(+), 3 deletions(-) diff --git a/drivers/usb/host/ehci-hub.c b/drivers/usb/host/ehci-hub.c index 218f9660d7e..38650707dfe 100644 --- a/drivers/usb/host/ehci-hub.c +++ b/drivers/usb/host/ehci-hub.c @@ -194,6 +194,7 @@ static int ehci_bus_resume (struct usb_hcd *hcd) u32 temp; u32 power_okay; int i; + u8 resume_needed = 0; if (time_before (jiffies, ehci->next_statechange)) msleep(5); @@ -228,7 +229,9 @@ static int ehci_bus_resume (struct usb_hcd *hcd) /* Some controller/firmware combinations need a delay during which * they set up the port statuses. See Bugzilla #8190. */ - mdelay(8); + spin_unlock_irq(&ehci->lock); + msleep(8); + spin_lock_irq(&ehci->lock); /* manually resume the ports we suspended during bus_suspend() */ i = HCS_N_PORTS (ehci->hcs_params); @@ -236,12 +239,21 @@ static int ehci_bus_resume (struct usb_hcd *hcd) temp = ehci_readl(ehci, &ehci->regs->port_status [i]); temp &= ~(PORT_RWC_BITS | PORT_WAKE_BITS); if (test_bit(i, &ehci->bus_suspended) && - (temp & PORT_SUSPEND)) + (temp & PORT_SUSPEND)) { temp |= PORT_RESUME; + resume_needed = 1; + } ehci_writel(ehci, temp, &ehci->regs->port_status [i]); } + + /* msleep for 20ms only if code is trying to resume port */ + if (resume_needed) { + spin_unlock_irq(&ehci->lock); + msleep(20); + spin_lock_irq(&ehci->lock); + } + i = HCS_N_PORTS (ehci->hcs_params); - mdelay (20); while (i--) { temp = ehci_readl(ehci, &ehci->regs->port_status [i]); if (test_bit(i, &ehci->bus_suspended) && From b92a78e582b1a45649143dc86e526f5824092478 Mon Sep 17 00:00:00 2001 From: Rodolfo Giometti Date: Thu, 23 Oct 2008 10:08:07 +0200 Subject: [PATCH 009/123] usb host: Oxford OXU210HP HCD driver. This driver implements the support for Oxford OXU210HP USB high-speed host, no peripheral nor OTG. Signed-off-by: Rodolfo Giometti Cc: Kan Liu Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/Kconfig | 13 + drivers/usb/host/Makefile | 1 + drivers/usb/host/oxu210hp-hcd.c | 3985 +++++++++++++++++++++++++++++++ drivers/usb/host/oxu210hp.h | 447 ++++ include/linux/oxu210hp.h | 7 + 5 files changed, 4453 insertions(+) create mode 100644 drivers/usb/host/oxu210hp-hcd.c create mode 100644 drivers/usb/host/oxu210hp.h create mode 100644 include/linux/oxu210hp.h diff --git a/drivers/usb/host/Kconfig b/drivers/usb/host/Kconfig index f3a75a929e0..2b476b6b3d4 100644 --- a/drivers/usb/host/Kconfig +++ b/drivers/usb/host/Kconfig @@ -96,6 +96,19 @@ config USB_EHCI_HCD_PPC_OF Enables support for the USB controller present on the PowerPC OpenFirmware platform bus. +config USB_OXU210HP_HCD + tristate "OXU210HP HCD support" + depends on USB + ---help--- + The OXU210HP is an USB host/OTG/device controller. Enable this + option if your board has this chip. If unsure, say N. + + This driver does not support isochronous transfers and doesn't + implement OTG nor USB device controllers. + + To compile this driver as a module, choose M here: the + module will be called oxu210hp-hcd. + config USB_ISP116X_HCD tristate "ISP116X HCD support" depends on USB diff --git a/drivers/usb/host/Makefile b/drivers/usb/host/Makefile index 23be2222404..e5f3f20787e 100644 --- a/drivers/usb/host/Makefile +++ b/drivers/usb/host/Makefile @@ -13,6 +13,7 @@ obj-$(CONFIG_USB_WHCI_HCD) += whci/ obj-$(CONFIG_PCI) += pci-quirks.o obj-$(CONFIG_USB_EHCI_HCD) += ehci-hcd.o +obj-$(CONFIG_USB_OXU210HP_HCD) += oxu210hp-hcd.o obj-$(CONFIG_USB_ISP116X_HCD) += isp116x-hcd.o obj-$(CONFIG_USB_OHCI_HCD) += ohci-hcd.o obj-$(CONFIG_USB_UHCI_HCD) += uhci-hcd.o diff --git a/drivers/usb/host/oxu210hp-hcd.c b/drivers/usb/host/oxu210hp-hcd.c new file mode 100644 index 00000000000..251123c29d8 --- /dev/null +++ b/drivers/usb/host/oxu210hp-hcd.c @@ -0,0 +1,3985 @@ +/* + * Copyright (c) 2008 Rodolfo Giometti + * Copyright (c) 2008 Eurotech S.p.A. + * + * This code is *strongly* based on EHCI-HCD code by David Brownell since + * the chip is a quasi-EHCI compatible. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2 of the License, or (at your + * option) any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY + * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License + * for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "../core/hcd.h" + +#include +#include +#include + +#include +#include + +#include "oxu210hp.h" + +#define DRIVER_VERSION "0.0.50" + +/* + * Main defines + */ + +#define oxu_dbg(oxu, fmt, args...) \ + dev_dbg(oxu_to_hcd(oxu)->self.controller , fmt , ## args) +#define oxu_err(oxu, fmt, args...) \ + dev_err(oxu_to_hcd(oxu)->self.controller , fmt , ## args) +#define oxu_info(oxu, fmt, args...) \ + dev_info(oxu_to_hcd(oxu)->self.controller , fmt , ## args) + +static inline struct usb_hcd *oxu_to_hcd(struct oxu_hcd *oxu) +{ + return container_of((void *) oxu, struct usb_hcd, hcd_priv); +} + +static inline struct oxu_hcd *hcd_to_oxu(struct usb_hcd *hcd) +{ + return (struct oxu_hcd *) (hcd->hcd_priv); +} + +/* + * Debug stuff + */ + +#undef OXU_URB_TRACE +#undef OXU_VERBOSE_DEBUG + +#ifdef OXU_VERBOSE_DEBUG +#define oxu_vdbg oxu_dbg +#else +#define oxu_vdbg(oxu, fmt, args...) /* Nop */ +#endif + +#ifdef DEBUG + +static int __attribute__((__unused__)) +dbg_status_buf(char *buf, unsigned len, const char *label, u32 status) +{ + return scnprintf(buf, len, "%s%sstatus %04x%s%s%s%s%s%s%s%s%s%s", + label, label[0] ? " " : "", status, + (status & STS_ASS) ? " Async" : "", + (status & STS_PSS) ? " Periodic" : "", + (status & STS_RECL) ? " Recl" : "", + (status & STS_HALT) ? " Halt" : "", + (status & STS_IAA) ? " IAA" : "", + (status & STS_FATAL) ? " FATAL" : "", + (status & STS_FLR) ? " FLR" : "", + (status & STS_PCD) ? " PCD" : "", + (status & STS_ERR) ? " ERR" : "", + (status & STS_INT) ? " INT" : "" + ); +} + +static int __attribute__((__unused__)) +dbg_intr_buf(char *buf, unsigned len, const char *label, u32 enable) +{ + return scnprintf(buf, len, "%s%sintrenable %02x%s%s%s%s%s%s", + label, label[0] ? " " : "", enable, + (enable & STS_IAA) ? " IAA" : "", + (enable & STS_FATAL) ? " FATAL" : "", + (enable & STS_FLR) ? " FLR" : "", + (enable & STS_PCD) ? " PCD" : "", + (enable & STS_ERR) ? " ERR" : "", + (enable & STS_INT) ? " INT" : "" + ); +} + +static const char *const fls_strings[] = + { "1024", "512", "256", "??" }; + +static int dbg_command_buf(char *buf, unsigned len, + const char *label, u32 command) +{ + return scnprintf(buf, len, + "%s%scommand %06x %s=%d ithresh=%d%s%s%s%s period=%s%s %s", + label, label[0] ? " " : "", command, + (command & CMD_PARK) ? "park" : "(park)", + CMD_PARK_CNT(command), + (command >> 16) & 0x3f, + (command & CMD_LRESET) ? " LReset" : "", + (command & CMD_IAAD) ? " IAAD" : "", + (command & CMD_ASE) ? " Async" : "", + (command & CMD_PSE) ? " Periodic" : "", + fls_strings[(command >> 2) & 0x3], + (command & CMD_RESET) ? " Reset" : "", + (command & CMD_RUN) ? "RUN" : "HALT" + ); +} + +static int dbg_port_buf(char *buf, unsigned len, const char *label, + int port, u32 status) +{ + char *sig; + + /* signaling state */ + switch (status & (3 << 10)) { + case 0 << 10: + sig = "se0"; + break; + case 1 << 10: + sig = "k"; /* low speed */ + break; + case 2 << 10: + sig = "j"; + break; + default: + sig = "?"; + break; + } + + return scnprintf(buf, len, + "%s%sport %d status %06x%s%s sig=%s%s%s%s%s%s%s%s%s%s", + label, label[0] ? " " : "", port, status, + (status & PORT_POWER) ? " POWER" : "", + (status & PORT_OWNER) ? " OWNER" : "", + sig, + (status & PORT_RESET) ? " RESET" : "", + (status & PORT_SUSPEND) ? " SUSPEND" : "", + (status & PORT_RESUME) ? " RESUME" : "", + (status & PORT_OCC) ? " OCC" : "", + (status & PORT_OC) ? " OC" : "", + (status & PORT_PEC) ? " PEC" : "", + (status & PORT_PE) ? " PE" : "", + (status & PORT_CSC) ? " CSC" : "", + (status & PORT_CONNECT) ? " CONNECT" : "" + ); +} + +#else + +static inline int __attribute__((__unused__)) +dbg_status_buf(char *buf, unsigned len, const char *label, u32 status) +{ return 0; } + +static inline int __attribute__((__unused__)) +dbg_command_buf(char *buf, unsigned len, const char *label, u32 command) +{ return 0; } + +static inline int __attribute__((__unused__)) +dbg_intr_buf(char *buf, unsigned len, const char *label, u32 enable) +{ return 0; } + +static inline int __attribute__((__unused__)) +dbg_port_buf(char *buf, unsigned len, const char *label, int port, u32 status) +{ return 0; } + +#endif /* DEBUG */ + +/* functions have the "wrong" filename when they're output... */ +#define dbg_status(oxu, label, status) { \ + char _buf[80]; \ + dbg_status_buf(_buf, sizeof _buf, label, status); \ + oxu_dbg(oxu, "%s\n", _buf); \ +} + +#define dbg_cmd(oxu, label, command) { \ + char _buf[80]; \ + dbg_command_buf(_buf, sizeof _buf, label, command); \ + oxu_dbg(oxu, "%s\n", _buf); \ +} + +#define dbg_port(oxu, label, port, status) { \ + char _buf[80]; \ + dbg_port_buf(_buf, sizeof _buf, label, port, status); \ + oxu_dbg(oxu, "%s\n", _buf); \ +} + +/* + * Module parameters + */ + +/* Initial IRQ latency: faster than hw default */ +static int log2_irq_thresh; /* 0 to 6 */ +module_param(log2_irq_thresh, int, S_IRUGO); +MODULE_PARM_DESC(log2_irq_thresh, "log2 IRQ latency, 1-64 microframes"); + +/* Initial park setting: slower than hw default */ +static unsigned park; +module_param(park, uint, S_IRUGO); +MODULE_PARM_DESC(park, "park setting; 1-3 back-to-back async packets"); + +/* For flakey hardware, ignore overcurrent indicators */ +static int ignore_oc; +module_param(ignore_oc, bool, S_IRUGO); +MODULE_PARM_DESC(ignore_oc, "ignore bogus hardware overcurrent indications"); + + +static void ehci_work(struct oxu_hcd *oxu); +static int oxu_hub_control(struct usb_hcd *hcd, + u16 typeReq, u16 wValue, u16 wIndex, + char *buf, u16 wLength); + +/* + * Local functions + */ + +/* Low level read/write registers functions */ +static inline u32 oxu_readl(void *base, u32 reg) +{ + return readl(base + reg); +} + +static inline void oxu_writel(void *base, u32 reg, u32 val) +{ + writel(val, base + reg); +} + +static inline void timer_action_done(struct oxu_hcd *oxu, + enum ehci_timer_action action) +{ + clear_bit(action, &oxu->actions); +} + +static inline void timer_action(struct oxu_hcd *oxu, + enum ehci_timer_action action) +{ + if (!test_and_set_bit(action, &oxu->actions)) { + unsigned long t; + + switch (action) { + case TIMER_IAA_WATCHDOG: + t = EHCI_IAA_JIFFIES; + break; + case TIMER_IO_WATCHDOG: + t = EHCI_IO_JIFFIES; + break; + case TIMER_ASYNC_OFF: + t = EHCI_ASYNC_JIFFIES; + break; + case TIMER_ASYNC_SHRINK: + default: + t = EHCI_SHRINK_JIFFIES; + break; + } + t += jiffies; + /* all timings except IAA watchdog can be overridden. + * async queue SHRINK often precedes IAA. while it's ready + * to go OFF neither can matter, and afterwards the IO + * watchdog stops unless there's still periodic traffic. + */ + if (action != TIMER_IAA_WATCHDOG + && t > oxu->watchdog.expires + && timer_pending(&oxu->watchdog)) + return; + mod_timer(&oxu->watchdog, t); + } +} + +/* + * handshake - spin reading hc until handshake completes or fails + * @ptr: address of hc register to be read + * @mask: bits to look at in result of read + * @done: value of those bits when handshake succeeds + * @usec: timeout in microseconds + * + * Returns negative errno, or zero on success + * + * Success happens when the "mask" bits have the specified value (hardware + * handshake done). There are two failure modes: "usec" have passed (major + * hardware flakeout), or the register reads as all-ones (hardware removed). + * + * That last failure should_only happen in cases like physical cardbus eject + * before driver shutdown. But it also seems to be caused by bugs in cardbus + * bridge shutdown: shutting down the bridge before the devices using it. + */ +static int handshake(struct oxu_hcd *oxu, void __iomem *ptr, + u32 mask, u32 done, int usec) +{ + u32 result; + + do { + result = readl(ptr); + if (result == ~(u32)0) /* card removed */ + return -ENODEV; + result &= mask; + if (result == done) + return 0; + udelay(1); + usec--; + } while (usec > 0); + return -ETIMEDOUT; +} + +/* Force HC to halt state from unknown (EHCI spec section 2.3) */ +static int ehci_halt(struct oxu_hcd *oxu) +{ + u32 temp = readl(&oxu->regs->status); + + /* disable any irqs left enabled by previous code */ + writel(0, &oxu->regs->intr_enable); + + if ((temp & STS_HALT) != 0) + return 0; + + temp = readl(&oxu->regs->command); + temp &= ~CMD_RUN; + writel(temp, &oxu->regs->command); + return handshake(oxu, &oxu->regs->status, + STS_HALT, STS_HALT, 16 * 125); +} + +/* Put TDI/ARC silicon into EHCI mode */ +static void tdi_reset(struct oxu_hcd *oxu) +{ + u32 __iomem *reg_ptr; + u32 tmp; + + reg_ptr = (u32 __iomem *)(((u8 __iomem *)oxu->regs) + 0x68); + tmp = readl(reg_ptr); + tmp |= 0x3; + writel(tmp, reg_ptr); +} + +/* Reset a non-running (STS_HALT == 1) controller */ +static int ehci_reset(struct oxu_hcd *oxu) +{ + int retval; + u32 command = readl(&oxu->regs->command); + + command |= CMD_RESET; + dbg_cmd(oxu, "reset", command); + writel(command, &oxu->regs->command); + oxu_to_hcd(oxu)->state = HC_STATE_HALT; + oxu->next_statechange = jiffies; + retval = handshake(oxu, &oxu->regs->command, + CMD_RESET, 0, 250 * 1000); + + if (retval) + return retval; + + tdi_reset(oxu); + + return retval; +} + +/* Idle the controller (from running) */ +static void ehci_quiesce(struct oxu_hcd *oxu) +{ + u32 temp; + +#ifdef DEBUG + if (!HC_IS_RUNNING(oxu_to_hcd(oxu)->state)) + BUG(); +#endif + + /* wait for any schedule enables/disables to take effect */ + temp = readl(&oxu->regs->command) << 10; + temp &= STS_ASS | STS_PSS; + if (handshake(oxu, &oxu->regs->status, STS_ASS | STS_PSS, + temp, 16 * 125) != 0) { + oxu_to_hcd(oxu)->state = HC_STATE_HALT; + return; + } + + /* then disable anything that's still active */ + temp = readl(&oxu->regs->command); + temp &= ~(CMD_ASE | CMD_IAAD | CMD_PSE); + writel(temp, &oxu->regs->command); + + /* hardware can take 16 microframes to turn off ... */ + if (handshake(oxu, &oxu->regs->status, STS_ASS | STS_PSS, + 0, 16 * 125) != 0) { + oxu_to_hcd(oxu)->state = HC_STATE_HALT; + return; + } +} + +static int check_reset_complete(struct oxu_hcd *oxu, int index, + u32 __iomem *status_reg, int port_status) +{ + if (!(port_status & PORT_CONNECT)) { + oxu->reset_done[index] = 0; + return port_status; + } + + /* if reset finished and it's still not enabled -- handoff */ + if (!(port_status & PORT_PE)) { + oxu_dbg(oxu, "Failed to enable port %d on root hub TT\n", + index+1); + return port_status; + } else + oxu_dbg(oxu, "port %d high speed\n", index + 1); + + return port_status; +} + +static void ehci_hub_descriptor(struct oxu_hcd *oxu, + struct usb_hub_descriptor *desc) +{ + int ports = HCS_N_PORTS(oxu->hcs_params); + u16 temp; + + desc->bDescriptorType = 0x29; + desc->bPwrOn2PwrGood = 10; /* oxu 1.0, 2.3.9 says 20ms max */ + desc->bHubContrCurrent = 0; + + desc->bNbrPorts = ports; + temp = 1 + (ports / 8); + desc->bDescLength = 7 + 2 * temp; + + /* two bitmaps: ports removable, and usb 1.0 legacy PortPwrCtrlMask */ + memset(&desc->bitmap[0], 0, temp); + memset(&desc->bitmap[temp], 0xff, temp); + + temp = 0x0008; /* per-port overcurrent reporting */ + if (HCS_PPC(oxu->hcs_params)) + temp |= 0x0001; /* per-port power control */ + else + temp |= 0x0002; /* no power switching */ + desc->wHubCharacteristics = (__force __u16)cpu_to_le16(temp); +} + + +/* Allocate an OXU210HP on-chip memory data buffer + * + * An on-chip memory data buffer is required for each OXU210HP USB transfer. + * Each transfer descriptor has one or more on-chip memory data buffers. + * + * Data buffers are allocated from a fix sized pool of data blocks. + * To minimise fragmentation and give reasonable memory utlisation, + * data buffers are allocated with sizes the power of 2 multiples of + * the block size, starting on an address a multiple of the allocated size. + * + * FIXME: callers of this function require a buffer to be allocated for + * len=0. This is a waste of on-chip memory and should be fix. Then this + * function should be changed to not allocate a buffer for len=0. + */ +static int oxu_buf_alloc(struct oxu_hcd *oxu, struct ehci_qtd *qtd, int len) +{ + int n_blocks; /* minium blocks needed to hold len */ + int a_blocks; /* blocks allocated */ + int i, j; + + /* Don't allocte bigger than supported */ + if (len > BUFFER_SIZE * BUFFER_NUM) { + oxu_err(oxu, "buffer too big (%d)\n", len); + return -ENOMEM; + } + + spin_lock(&oxu->mem_lock); + + /* Number of blocks needed to hold len */ + n_blocks = (len + BUFFER_SIZE - 1) / BUFFER_SIZE; + + /* Round the number of blocks up to the power of 2 */ + for (a_blocks = 1; a_blocks < n_blocks; a_blocks <<= 1) + ; + + /* Find a suitable available data buffer */ + for (i = 0; i < BUFFER_NUM; + i += max(a_blocks, (int)oxu->db_used[i])) { + + /* Check all the required blocks are available */ + for (j = 0; j < a_blocks; j++) + if (oxu->db_used[i + j]) + break; + + if (j != a_blocks) + continue; + + /* Allocate blocks found! */ + qtd->buffer = (void *) &oxu->mem->db_pool[i]; + qtd->buffer_dma = virt_to_phys(qtd->buffer); + + qtd->qtd_buffer_len = BUFFER_SIZE * a_blocks; + oxu->db_used[i] = a_blocks; + + spin_unlock(&oxu->mem_lock); + + return 0; + } + + /* Failed */ + + spin_unlock(&oxu->mem_lock); + + return -ENOMEM; +} + +static void oxu_buf_free(struct oxu_hcd *oxu, struct ehci_qtd *qtd) +{ + int index; + + spin_lock(&oxu->mem_lock); + + index = (qtd->buffer - (void *) &oxu->mem->db_pool[0]) + / BUFFER_SIZE; + oxu->db_used[index] = 0; + qtd->qtd_buffer_len = 0; + qtd->buffer_dma = 0; + qtd->buffer = NULL; + + spin_unlock(&oxu->mem_lock); + + return; +} + +static inline void ehci_qtd_init(struct ehci_qtd *qtd, dma_addr_t dma) +{ + memset(qtd, 0, sizeof *qtd); + qtd->qtd_dma = dma; + qtd->hw_token = cpu_to_le32(QTD_STS_HALT); + qtd->hw_next = EHCI_LIST_END; + qtd->hw_alt_next = EHCI_LIST_END; + INIT_LIST_HEAD(&qtd->qtd_list); +} + +static inline void oxu_qtd_free(struct oxu_hcd *oxu, struct ehci_qtd *qtd) +{ + int index; + + if (qtd->buffer) + oxu_buf_free(oxu, qtd); + + spin_lock(&oxu->mem_lock); + + index = qtd - &oxu->mem->qtd_pool[0]; + oxu->qtd_used[index] = 0; + + spin_unlock(&oxu->mem_lock); + + return; +} + +static struct ehci_qtd *ehci_qtd_alloc(struct oxu_hcd *oxu) +{ + int i; + struct ehci_qtd *qtd = NULL; + + spin_lock(&oxu->mem_lock); + + for (i = 0; i < QTD_NUM; i++) + if (!oxu->qtd_used[i]) + break; + + if (i < QTD_NUM) { + qtd = (struct ehci_qtd *) &oxu->mem->qtd_pool[i]; + memset(qtd, 0, sizeof *qtd); + + qtd->hw_token = cpu_to_le32(QTD_STS_HALT); + qtd->hw_next = EHCI_LIST_END; + qtd->hw_alt_next = EHCI_LIST_END; + INIT_LIST_HEAD(&qtd->qtd_list); + + qtd->qtd_dma = virt_to_phys(qtd); + + oxu->qtd_used[i] = 1; + } + + spin_unlock(&oxu->mem_lock); + + return qtd; +} + +static void oxu_qh_free(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + int index; + + spin_lock(&oxu->mem_lock); + + index = qh - &oxu->mem->qh_pool[0]; + oxu->qh_used[index] = 0; + + spin_unlock(&oxu->mem_lock); + + return; +} + +static void qh_destroy(struct kref *kref) +{ + struct ehci_qh *qh = container_of(kref, struct ehci_qh, kref); + struct oxu_hcd *oxu = qh->oxu; + + /* clean qtds first, and know this is not linked */ + if (!list_empty(&qh->qtd_list) || qh->qh_next.ptr) { + oxu_dbg(oxu, "unused qh not empty!\n"); + BUG(); + } + if (qh->dummy) + oxu_qtd_free(oxu, qh->dummy); + oxu_qh_free(oxu, qh); +} + +static struct ehci_qh *oxu_qh_alloc(struct oxu_hcd *oxu) +{ + int i; + struct ehci_qh *qh = NULL; + + spin_lock(&oxu->mem_lock); + + for (i = 0; i < QHEAD_NUM; i++) + if (!oxu->qh_used[i]) + break; + + if (i < QHEAD_NUM) { + qh = (struct ehci_qh *) &oxu->mem->qh_pool[i]; + memset(qh, 0, sizeof *qh); + + kref_init(&qh->kref); + qh->oxu = oxu; + qh->qh_dma = virt_to_phys(qh); + INIT_LIST_HEAD(&qh->qtd_list); + + /* dummy td enables safe urb queuing */ + qh->dummy = ehci_qtd_alloc(oxu); + if (qh->dummy == NULL) { + oxu_dbg(oxu, "no dummy td\n"); + oxu->qh_used[i] = 0; + + return NULL; + } + + oxu->qh_used[i] = 1; + } + + spin_unlock(&oxu->mem_lock); + + return qh; +} + +/* to share a qh (cpu threads, or hc) */ +static inline struct ehci_qh *qh_get(struct ehci_qh *qh) +{ + kref_get(&qh->kref); + return qh; +} + +static inline void qh_put(struct ehci_qh *qh) +{ + kref_put(&qh->kref, qh_destroy); +} + +static void oxu_murb_free(struct oxu_hcd *oxu, struct oxu_murb *murb) +{ + int index; + + spin_lock(&oxu->mem_lock); + + index = murb - &oxu->murb_pool[0]; + oxu->murb_used[index] = 0; + + spin_unlock(&oxu->mem_lock); + + return; +} + +static struct oxu_murb *oxu_murb_alloc(struct oxu_hcd *oxu) + +{ + int i; + struct oxu_murb *murb = NULL; + + spin_lock(&oxu->mem_lock); + + for (i = 0; i < MURB_NUM; i++) + if (!oxu->murb_used[i]) + break; + + if (i < MURB_NUM) { + murb = &(oxu->murb_pool)[i]; + + oxu->murb_used[i] = 1; + } + + spin_unlock(&oxu->mem_lock); + + return murb; +} + +/* The queue heads and transfer descriptors are managed from pools tied + * to each of the "per device" structures. + * This is the initialisation and cleanup code. + */ +static void ehci_mem_cleanup(struct oxu_hcd *oxu) +{ + kfree(oxu->murb_pool); + oxu->murb_pool = NULL; + + if (oxu->async) + qh_put(oxu->async); + oxu->async = NULL; + + del_timer(&oxu->urb_timer); + + oxu->periodic = NULL; + + /* shadow periodic table */ + kfree(oxu->pshadow); + oxu->pshadow = NULL; +} + +/* Remember to add cleanup code (above) if you add anything here. + */ +static int ehci_mem_init(struct oxu_hcd *oxu, gfp_t flags) +{ + int i; + + for (i = 0; i < oxu->periodic_size; i++) + oxu->mem->frame_list[i] = EHCI_LIST_END; + for (i = 0; i < QHEAD_NUM; i++) + oxu->qh_used[i] = 0; + for (i = 0; i < QTD_NUM; i++) + oxu->qtd_used[i] = 0; + + oxu->murb_pool = kcalloc(MURB_NUM, sizeof(struct oxu_murb), flags); + if (!oxu->murb_pool) + goto fail; + + for (i = 0; i < MURB_NUM; i++) + oxu->murb_used[i] = 0; + + oxu->async = oxu_qh_alloc(oxu); + if (!oxu->async) + goto fail; + + oxu->periodic = (__le32 *) &oxu->mem->frame_list; + oxu->periodic_dma = virt_to_phys(oxu->periodic); + + for (i = 0; i < oxu->periodic_size; i++) + oxu->periodic[i] = EHCI_LIST_END; + + /* software shadow of hardware table */ + oxu->pshadow = kcalloc(oxu->periodic_size, sizeof(void *), flags); + if (oxu->pshadow != NULL) + return 0; + +fail: + oxu_dbg(oxu, "couldn't init memory\n"); + ehci_mem_cleanup(oxu); + return -ENOMEM; +} + +/* Fill a qtd, returning how much of the buffer we were able to queue up. + */ +static int qtd_fill(struct ehci_qtd *qtd, dma_addr_t buf, size_t len, + int token, int maxpacket) +{ + int i, count; + u64 addr = buf; + + /* one buffer entry per 4K ... first might be short or unaligned */ + qtd->hw_buf[0] = cpu_to_le32((u32)addr); + qtd->hw_buf_hi[0] = cpu_to_le32((u32)(addr >> 32)); + count = 0x1000 - (buf & 0x0fff); /* rest of that page */ + if (likely(len < count)) /* ... iff needed */ + count = len; + else { + buf += 0x1000; + buf &= ~0x0fff; + + /* per-qtd limit: from 16K to 20K (best alignment) */ + for (i = 1; count < len && i < 5; i++) { + addr = buf; + qtd->hw_buf[i] = cpu_to_le32((u32)addr); + qtd->hw_buf_hi[i] = cpu_to_le32((u32)(addr >> 32)); + buf += 0x1000; + if ((count + 0x1000) < len) + count += 0x1000; + else + count = len; + } + + /* short packets may only terminate transfers */ + if (count != len) + count -= (count % maxpacket); + } + qtd->hw_token = cpu_to_le32((count << 16) | token); + qtd->length = count; + + return count; +} + +static inline void qh_update(struct oxu_hcd *oxu, + struct ehci_qh *qh, struct ehci_qtd *qtd) +{ + /* writes to an active overlay are unsafe */ + BUG_ON(qh->qh_state != QH_STATE_IDLE); + + qh->hw_qtd_next = QTD_NEXT(qtd->qtd_dma); + qh->hw_alt_next = EHCI_LIST_END; + + /* Except for control endpoints, we make hardware maintain data + * toggle (like OHCI) ... here (re)initialize the toggle in the QH, + * and set the pseudo-toggle in udev. Only usb_clear_halt() will + * ever clear it. + */ + if (!(qh->hw_info1 & cpu_to_le32(1 << 14))) { + unsigned is_out, epnum; + + is_out = !(qtd->hw_token & cpu_to_le32(1 << 8)); + epnum = (le32_to_cpup(&qh->hw_info1) >> 8) & 0x0f; + if (unlikely(!usb_gettoggle(qh->dev, epnum, is_out))) { + qh->hw_token &= ~__constant_cpu_to_le32(QTD_TOGGLE); + usb_settoggle(qh->dev, epnum, is_out, 1); + } + } + + /* HC must see latest qtd and qh data before we clear ACTIVE+HALT */ + wmb(); + qh->hw_token &= __constant_cpu_to_le32(QTD_TOGGLE | QTD_STS_PING); +} + +/* If it weren't for a common silicon quirk (writing the dummy into the qh + * overlay, so qh->hw_token wrongly becomes inactive/halted), only fault + * recovery (including urb dequeue) would need software changes to a QH... + */ +static void qh_refresh(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + struct ehci_qtd *qtd; + + if (list_empty(&qh->qtd_list)) + qtd = qh->dummy; + else { + qtd = list_entry(qh->qtd_list.next, + struct ehci_qtd, qtd_list); + /* first qtd may already be partially processed */ + if (cpu_to_le32(qtd->qtd_dma) == qh->hw_current) + qtd = NULL; + } + + if (qtd) + qh_update(oxu, qh, qtd); +} + +static void qtd_copy_status(struct oxu_hcd *oxu, struct urb *urb, + size_t length, u32 token) +{ + /* count IN/OUT bytes, not SETUP (even short packets) */ + if (likely(QTD_PID(token) != 2)) + urb->actual_length += length - QTD_LENGTH(token); + + /* don't modify error codes */ + if (unlikely(urb->status != -EINPROGRESS)) + return; + + /* force cleanup after short read; not always an error */ + if (unlikely(IS_SHORT_READ(token))) + urb->status = -EREMOTEIO; + + /* serious "can't proceed" faults reported by the hardware */ + if (token & QTD_STS_HALT) { + if (token & QTD_STS_BABBLE) { + /* FIXME "must" disable babbling device's port too */ + urb->status = -EOVERFLOW; + } else if (token & QTD_STS_MMF) { + /* fs/ls interrupt xfer missed the complete-split */ + urb->status = -EPROTO; + } else if (token & QTD_STS_DBE) { + urb->status = (QTD_PID(token) == 1) /* IN ? */ + ? -ENOSR /* hc couldn't read data */ + : -ECOMM; /* hc couldn't write data */ + } else if (token & QTD_STS_XACT) { + /* timeout, bad crc, wrong PID, etc; retried */ + if (QTD_CERR(token)) + urb->status = -EPIPE; + else { + oxu_dbg(oxu, "devpath %s ep%d%s 3strikes\n", + urb->dev->devpath, + usb_pipeendpoint(urb->pipe), + usb_pipein(urb->pipe) ? "in" : "out"); + urb->status = -EPROTO; + } + /* CERR nonzero + no errors + halt --> stall */ + } else if (QTD_CERR(token)) + urb->status = -EPIPE; + else /* unknown */ + urb->status = -EPROTO; + + oxu_vdbg(oxu, "dev%d ep%d%s qtd token %08x --> status %d\n", + usb_pipedevice(urb->pipe), + usb_pipeendpoint(urb->pipe), + usb_pipein(urb->pipe) ? "in" : "out", + token, urb->status); + } +} + +static void ehci_urb_done(struct oxu_hcd *oxu, struct urb *urb) +__releases(oxu->lock) +__acquires(oxu->lock) +{ + if (likely(urb->hcpriv != NULL)) { + struct ehci_qh *qh = (struct ehci_qh *) urb->hcpriv; + + /* S-mask in a QH means it's an interrupt urb */ + if ((qh->hw_info2 & __constant_cpu_to_le32(QH_SMASK)) != 0) { + + /* ... update hc-wide periodic stats (for usbfs) */ + oxu_to_hcd(oxu)->self.bandwidth_int_reqs--; + } + qh_put(qh); + } + + urb->hcpriv = NULL; + switch (urb->status) { + case -EINPROGRESS: /* success */ + urb->status = 0; + default: /* fault */ + break; + case -EREMOTEIO: /* fault or normal */ + if (!(urb->transfer_flags & URB_SHORT_NOT_OK)) + urb->status = 0; + break; + case -ECONNRESET: /* canceled */ + case -ENOENT: + break; + } + +#ifdef OXU_URB_TRACE + oxu_dbg(oxu, "%s %s urb %p ep%d%s status %d len %d/%d\n", + __func__, urb->dev->devpath, urb, + usb_pipeendpoint(urb->pipe), + usb_pipein(urb->pipe) ? "in" : "out", + urb->status, + urb->actual_length, urb->transfer_buffer_length); +#endif + + /* complete() can reenter this HCD */ + spin_unlock(&oxu->lock); + usb_hcd_giveback_urb(oxu_to_hcd(oxu), urb, urb->status); + spin_lock(&oxu->lock); +} + +static void start_unlink_async(struct oxu_hcd *oxu, struct ehci_qh *qh); +static void unlink_async(struct oxu_hcd *oxu, struct ehci_qh *qh); + +static void intr_deschedule(struct oxu_hcd *oxu, struct ehci_qh *qh); +static int qh_schedule(struct oxu_hcd *oxu, struct ehci_qh *qh); + +#define HALT_BIT __constant_cpu_to_le32(QTD_STS_HALT) + +/* Process and free completed qtds for a qh, returning URBs to drivers. + * Chases up to qh->hw_current. Returns number of completions called, + * indicating how much "real" work we did. + */ +static unsigned qh_completions(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + struct ehci_qtd *last = NULL, *end = qh->dummy; + struct list_head *entry, *tmp; + int stopped; + unsigned count = 0; + int do_status = 0; + u8 state; + struct oxu_murb *murb = NULL; + + if (unlikely(list_empty(&qh->qtd_list))) + return count; + + /* completions (or tasks on other cpus) must never clobber HALT + * till we've gone through and cleaned everything up, even when + * they add urbs to this qh's queue or mark them for unlinking. + * + * NOTE: unlinking expects to be done in queue order. + */ + state = qh->qh_state; + qh->qh_state = QH_STATE_COMPLETING; + stopped = (state == QH_STATE_IDLE); + + /* remove de-activated QTDs from front of queue. + * after faults (including short reads), cleanup this urb + * then let the queue advance. + * if queue is stopped, handles unlinks. + */ + list_for_each_safe(entry, tmp, &qh->qtd_list) { + struct ehci_qtd *qtd; + struct urb *urb; + u32 token = 0; + + qtd = list_entry(entry, struct ehci_qtd, qtd_list); + urb = qtd->urb; + + /* Clean up any state from previous QTD ...*/ + if (last) { + if (likely(last->urb != urb)) { + if (last->urb->complete == NULL) { + murb = (struct oxu_murb *) last->urb; + last->urb = murb->main; + if (murb->last) { + ehci_urb_done(oxu, last->urb); + count++; + } + oxu_murb_free(oxu, murb); + } else { + ehci_urb_done(oxu, last->urb); + count++; + } + } + oxu_qtd_free(oxu, last); + last = NULL; + } + + /* ignore urbs submitted during completions we reported */ + if (qtd == end) + break; + + /* hardware copies qtd out of qh overlay */ + rmb(); + token = le32_to_cpu(qtd->hw_token); + + /* always clean up qtds the hc de-activated */ + if ((token & QTD_STS_ACTIVE) == 0) { + + if ((token & QTD_STS_HALT) != 0) { + stopped = 1; + + /* magic dummy for some short reads; qh won't advance. + * that silicon quirk can kick in with this dummy too. + */ + } else if (IS_SHORT_READ(token) && + !(qtd->hw_alt_next & EHCI_LIST_END)) { + stopped = 1; + goto halt; + } + + /* stop scanning when we reach qtds the hc is using */ + } else if (likely(!stopped && + HC_IS_RUNNING(oxu_to_hcd(oxu)->state))) { + break; + + } else { + stopped = 1; + + if (unlikely(!HC_IS_RUNNING(oxu_to_hcd(oxu)->state))) + urb->status = -ESHUTDOWN; + + /* ignore active urbs unless some previous qtd + * for the urb faulted (including short read) or + * its urb was canceled. we may patch qh or qtds. + */ + if (likely(urb->status == -EINPROGRESS)) + continue; + + /* issue status after short control reads */ + if (unlikely(do_status != 0) + && QTD_PID(token) == 0 /* OUT */) { + do_status = 0; + continue; + } + + /* token in overlay may be most current */ + if (state == QH_STATE_IDLE + && cpu_to_le32(qtd->qtd_dma) + == qh->hw_current) + token = le32_to_cpu(qh->hw_token); + + /* force halt for unlinked or blocked qh, so we'll + * patch the qh later and so that completions can't + * activate it while we "know" it's stopped. + */ + if ((HALT_BIT & qh->hw_token) == 0) { +halt: + qh->hw_token |= HALT_BIT; + wmb(); + } + } + + /* Remove it from the queue */ + qtd_copy_status(oxu, urb->complete ? + urb : ((struct oxu_murb *) urb)->main, + qtd->length, token); + if ((usb_pipein(qtd->urb->pipe)) && + (NULL != qtd->transfer_buffer)) + memcpy(qtd->transfer_buffer, qtd->buffer, qtd->length); + do_status = (urb->status == -EREMOTEIO) + && usb_pipecontrol(urb->pipe); + + if (stopped && qtd->qtd_list.prev != &qh->qtd_list) { + last = list_entry(qtd->qtd_list.prev, + struct ehci_qtd, qtd_list); + last->hw_next = qtd->hw_next; + } + list_del(&qtd->qtd_list); + last = qtd; + } + + /* last urb's completion might still need calling */ + if (likely(last != NULL)) { + if (last->urb->complete == NULL) { + murb = (struct oxu_murb *) last->urb; + last->urb = murb->main; + if (murb->last) { + ehci_urb_done(oxu, last->urb); + count++; + } + oxu_murb_free(oxu, murb); + } else { + ehci_urb_done(oxu, last->urb); + count++; + } + oxu_qtd_free(oxu, last); + } + + /* restore original state; caller must unlink or relink */ + qh->qh_state = state; + + /* be sure the hardware's done with the qh before refreshing + * it after fault cleanup, or recovering from silicon wrongly + * overlaying the dummy qtd (which reduces DMA chatter). + */ + if (stopped != 0 || qh->hw_qtd_next == EHCI_LIST_END) { + switch (state) { + case QH_STATE_IDLE: + qh_refresh(oxu, qh); + break; + case QH_STATE_LINKED: + /* should be rare for periodic transfers, + * except maybe high bandwidth ... + */ + if ((__constant_cpu_to_le32(QH_SMASK) + & qh->hw_info2) != 0) { + intr_deschedule(oxu, qh); + (void) qh_schedule(oxu, qh); + } else + unlink_async(oxu, qh); + break; + /* otherwise, unlink already started */ + } + } + + return count; +} + +/* High bandwidth multiplier, as encoded in highspeed endpoint descriptors */ +#define hb_mult(wMaxPacketSize) (1 + (((wMaxPacketSize) >> 11) & 0x03)) +/* ... and packet size, for any kind of endpoint descriptor */ +#define max_packet(wMaxPacketSize) ((wMaxPacketSize) & 0x07ff) + +/* Reverse of qh_urb_transaction: free a list of TDs. + * used for cleanup after errors, before HC sees an URB's TDs. + */ +static void qtd_list_free(struct oxu_hcd *oxu, + struct urb *urb, struct list_head *qtd_list) +{ + struct list_head *entry, *temp; + + list_for_each_safe(entry, temp, qtd_list) { + struct ehci_qtd *qtd; + + qtd = list_entry(entry, struct ehci_qtd, qtd_list); + list_del(&qtd->qtd_list); + oxu_qtd_free(oxu, qtd); + } +} + +/* Create a list of filled qtds for this URB; won't link into qh. + */ +static struct list_head *qh_urb_transaction(struct oxu_hcd *oxu, + struct urb *urb, + struct list_head *head, + gfp_t flags) +{ + struct ehci_qtd *qtd, *qtd_prev; + dma_addr_t buf; + int len, maxpacket; + int is_input; + u32 token; + void *transfer_buf = NULL; + int ret; + + /* + * URBs map to sequences of QTDs: one logical transaction + */ + qtd = ehci_qtd_alloc(oxu); + if (unlikely(!qtd)) + return NULL; + list_add_tail(&qtd->qtd_list, head); + qtd->urb = urb; + + token = QTD_STS_ACTIVE; + token |= (EHCI_TUNE_CERR << 10); + /* for split transactions, SplitXState initialized to zero */ + + len = urb->transfer_buffer_length; + is_input = usb_pipein(urb->pipe); + if (!urb->transfer_buffer && urb->transfer_buffer_length && is_input) + urb->transfer_buffer = phys_to_virt(urb->transfer_dma); + + if (usb_pipecontrol(urb->pipe)) { + /* SETUP pid */ + ret = oxu_buf_alloc(oxu, qtd, sizeof(struct usb_ctrlrequest)); + if (ret) + goto cleanup; + + qtd_fill(qtd, qtd->buffer_dma, sizeof(struct usb_ctrlrequest), + token | (2 /* "setup" */ << 8), 8); + memcpy(qtd->buffer, qtd->urb->setup_packet, + sizeof(struct usb_ctrlrequest)); + + /* ... and always at least one more pid */ + token ^= QTD_TOGGLE; + qtd_prev = qtd; + qtd = ehci_qtd_alloc(oxu); + if (unlikely(!qtd)) + goto cleanup; + qtd->urb = urb; + qtd_prev->hw_next = QTD_NEXT(qtd->qtd_dma); + list_add_tail(&qtd->qtd_list, head); + + /* for zero length DATA stages, STATUS is always IN */ + if (len == 0) + token |= (1 /* "in" */ << 8); + } + + /* + * Data transfer stage: buffer setup + */ + + ret = oxu_buf_alloc(oxu, qtd, len); + if (ret) + goto cleanup; + + buf = qtd->buffer_dma; + transfer_buf = urb->transfer_buffer; + + if (!is_input) + memcpy(qtd->buffer, qtd->urb->transfer_buffer, len); + + if (is_input) + token |= (1 /* "in" */ << 8); + /* else it's already initted to "out" pid (0 << 8) */ + + maxpacket = max_packet(usb_maxpacket(urb->dev, urb->pipe, !is_input)); + + /* + * buffer gets wrapped in one or more qtds; + * last one may be "short" (including zero len) + * and may serve as a control status ack + */ + for (;;) { + int this_qtd_len; + + this_qtd_len = qtd_fill(qtd, buf, len, token, maxpacket); + qtd->transfer_buffer = transfer_buf; + len -= this_qtd_len; + buf += this_qtd_len; + transfer_buf += this_qtd_len; + if (is_input) + qtd->hw_alt_next = oxu->async->hw_alt_next; + + /* qh makes control packets use qtd toggle; maybe switch it */ + if ((maxpacket & (this_qtd_len + (maxpacket - 1))) == 0) + token ^= QTD_TOGGLE; + + if (likely(len <= 0)) + break; + + qtd_prev = qtd; + qtd = ehci_qtd_alloc(oxu); + if (unlikely(!qtd)) + goto cleanup; + if (likely(len > 0)) { + ret = oxu_buf_alloc(oxu, qtd, len); + if (ret) + goto cleanup; + } + qtd->urb = urb; + qtd_prev->hw_next = QTD_NEXT(qtd->qtd_dma); + list_add_tail(&qtd->qtd_list, head); + } + + /* unless the bulk/interrupt caller wants a chance to clean + * up after short reads, hc should advance qh past this urb + */ + if (likely((urb->transfer_flags & URB_SHORT_NOT_OK) == 0 + || usb_pipecontrol(urb->pipe))) + qtd->hw_alt_next = EHCI_LIST_END; + + /* + * control requests may need a terminating data "status" ack; + * bulk ones may need a terminating short packet (zero length). + */ + if (likely(urb->transfer_buffer_length != 0)) { + int one_more = 0; + + if (usb_pipecontrol(urb->pipe)) { + one_more = 1; + token ^= 0x0100; /* "in" <--> "out" */ + token |= QTD_TOGGLE; /* force DATA1 */ + } else if (usb_pipebulk(urb->pipe) + && (urb->transfer_flags & URB_ZERO_PACKET) + && !(urb->transfer_buffer_length % maxpacket)) { + one_more = 1; + } + if (one_more) { + qtd_prev = qtd; + qtd = ehci_qtd_alloc(oxu); + if (unlikely(!qtd)) + goto cleanup; + qtd->urb = urb; + qtd_prev->hw_next = QTD_NEXT(qtd->qtd_dma); + list_add_tail(&qtd->qtd_list, head); + + /* never any data in such packets */ + qtd_fill(qtd, 0, 0, token, 0); + } + } + + /* by default, enable interrupt on urb completion */ + qtd->hw_token |= __constant_cpu_to_le32(QTD_IOC); + return head; + +cleanup: + qtd_list_free(oxu, urb, head); + return NULL; +} + +/* Each QH holds a qtd list; a QH is used for everything except iso. + * + * For interrupt urbs, the scheduler must set the microframe scheduling + * mask(s) each time the QH gets scheduled. For highspeed, that's + * just one microframe in the s-mask. For split interrupt transactions + * there are additional complications: c-mask, maybe FSTNs. + */ +static struct ehci_qh *qh_make(struct oxu_hcd *oxu, + struct urb *urb, gfp_t flags) +{ + struct ehci_qh *qh = oxu_qh_alloc(oxu); + u32 info1 = 0, info2 = 0; + int is_input, type; + int maxp = 0; + + if (!qh) + return qh; + + /* + * init endpoint/device data for this QH + */ + info1 |= usb_pipeendpoint(urb->pipe) << 8; + info1 |= usb_pipedevice(urb->pipe) << 0; + + is_input = usb_pipein(urb->pipe); + type = usb_pipetype(urb->pipe); + maxp = usb_maxpacket(urb->dev, urb->pipe, !is_input); + + /* Compute interrupt scheduling parameters just once, and save. + * - allowing for high bandwidth, how many nsec/uframe are used? + * - split transactions need a second CSPLIT uframe; same question + * - splits also need a schedule gap (for full/low speed I/O) + * - qh has a polling interval + * + * For control/bulk requests, the HC or TT handles these. + */ + if (type == PIPE_INTERRUPT) { + qh->usecs = NS_TO_US(usb_calc_bus_time(USB_SPEED_HIGH, + is_input, 0, + hb_mult(maxp) * max_packet(maxp))); + qh->start = NO_FRAME; + + if (urb->dev->speed == USB_SPEED_HIGH) { + qh->c_usecs = 0; + qh->gap_uf = 0; + + qh->period = urb->interval >> 3; + if (qh->period == 0 && urb->interval != 1) { + /* NOTE interval 2 or 4 uframes could work. + * But interval 1 scheduling is simpler, and + * includes high bandwidth. + */ + dbg("intr period %d uframes, NYET!", + urb->interval); + goto done; + } + } else { + struct usb_tt *tt = urb->dev->tt; + int think_time; + + /* gap is f(FS/LS transfer times) */ + qh->gap_uf = 1 + usb_calc_bus_time(urb->dev->speed, + is_input, 0, maxp) / (125 * 1000); + + /* FIXME this just approximates SPLIT/CSPLIT times */ + if (is_input) { /* SPLIT, gap, CSPLIT+DATA */ + qh->c_usecs = qh->usecs + HS_USECS(0); + qh->usecs = HS_USECS(1); + } else { /* SPLIT+DATA, gap, CSPLIT */ + qh->usecs += HS_USECS(1); + qh->c_usecs = HS_USECS(0); + } + + think_time = tt ? tt->think_time : 0; + qh->tt_usecs = NS_TO_US(think_time + + usb_calc_bus_time(urb->dev->speed, + is_input, 0, max_packet(maxp))); + qh->period = urb->interval; + } + } + + /* support for tt scheduling, and access to toggles */ + qh->dev = urb->dev; + + /* using TT? */ + switch (urb->dev->speed) { + case USB_SPEED_LOW: + info1 |= (1 << 12); /* EPS "low" */ + /* FALL THROUGH */ + + case USB_SPEED_FULL: + /* EPS 0 means "full" */ + if (type != PIPE_INTERRUPT) + info1 |= (EHCI_TUNE_RL_TT << 28); + if (type == PIPE_CONTROL) { + info1 |= (1 << 27); /* for TT */ + info1 |= 1 << 14; /* toggle from qtd */ + } + info1 |= maxp << 16; + + info2 |= (EHCI_TUNE_MULT_TT << 30); + info2 |= urb->dev->ttport << 23; + + /* NOTE: if (PIPE_INTERRUPT) { scheduler sets c-mask } */ + + break; + + case USB_SPEED_HIGH: /* no TT involved */ + info1 |= (2 << 12); /* EPS "high" */ + if (type == PIPE_CONTROL) { + info1 |= (EHCI_TUNE_RL_HS << 28); + info1 |= 64 << 16; /* usb2 fixed maxpacket */ + info1 |= 1 << 14; /* toggle from qtd */ + info2 |= (EHCI_TUNE_MULT_HS << 30); + } else if (type == PIPE_BULK) { + info1 |= (EHCI_TUNE_RL_HS << 28); + info1 |= 512 << 16; /* usb2 fixed maxpacket */ + info2 |= (EHCI_TUNE_MULT_HS << 30); + } else { /* PIPE_INTERRUPT */ + info1 |= max_packet(maxp) << 16; + info2 |= hb_mult(maxp) << 30; + } + break; + default: + dbg("bogus dev %p speed %d", urb->dev, urb->dev->speed); +done: + qh_put(qh); + return NULL; + } + + /* NOTE: if (PIPE_INTERRUPT) { scheduler sets s-mask } */ + + /* init as live, toggle clear, advance to dummy */ + qh->qh_state = QH_STATE_IDLE; + qh->hw_info1 = cpu_to_le32(info1); + qh->hw_info2 = cpu_to_le32(info2); + usb_settoggle(urb->dev, usb_pipeendpoint(urb->pipe), !is_input, 1); + qh_refresh(oxu, qh); + return qh; +} + +/* Move qh (and its qtds) onto async queue; maybe enable queue. + */ +static void qh_link_async(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + __le32 dma = QH_NEXT(qh->qh_dma); + struct ehci_qh *head; + + /* (re)start the async schedule? */ + head = oxu->async; + timer_action_done(oxu, TIMER_ASYNC_OFF); + if (!head->qh_next.qh) { + u32 cmd = readl(&oxu->regs->command); + + if (!(cmd & CMD_ASE)) { + /* in case a clear of CMD_ASE didn't take yet */ + (void)handshake(oxu, &oxu->regs->status, + STS_ASS, 0, 150); + cmd |= CMD_ASE | CMD_RUN; + writel(cmd, &oxu->regs->command); + oxu_to_hcd(oxu)->state = HC_STATE_RUNNING; + /* posted write need not be known to HC yet ... */ + } + } + + /* clear halt and/or toggle; and maybe recover from silicon quirk */ + if (qh->qh_state == QH_STATE_IDLE) + qh_refresh(oxu, qh); + + /* splice right after start */ + qh->qh_next = head->qh_next; + qh->hw_next = head->hw_next; + wmb(); + + head->qh_next.qh = qh; + head->hw_next = dma; + + qh->qh_state = QH_STATE_LINKED; + /* qtd completions reported later by interrupt */ +} + +#define QH_ADDR_MASK __constant_cpu_to_le32(0x7f) + +/* + * For control/bulk/interrupt, return QH with these TDs appended. + * Allocates and initializes the QH if necessary. + * Returns null if it can't allocate a QH it needs to. + * If the QH has TDs (urbs) already, that's great. + */ +static struct ehci_qh *qh_append_tds(struct oxu_hcd *oxu, + struct urb *urb, struct list_head *qtd_list, + int epnum, void **ptr) +{ + struct ehci_qh *qh = NULL; + + qh = (struct ehci_qh *) *ptr; + if (unlikely(qh == NULL)) { + /* can't sleep here, we have oxu->lock... */ + qh = qh_make(oxu, urb, GFP_ATOMIC); + *ptr = qh; + } + if (likely(qh != NULL)) { + struct ehci_qtd *qtd; + + if (unlikely(list_empty(qtd_list))) + qtd = NULL; + else + qtd = list_entry(qtd_list->next, struct ehci_qtd, + qtd_list); + + /* control qh may need patching ... */ + if (unlikely(epnum == 0)) { + + /* usb_reset_device() briefly reverts to address 0 */ + if (usb_pipedevice(urb->pipe) == 0) + qh->hw_info1 &= ~QH_ADDR_MASK; + } + + /* just one way to queue requests: swap with the dummy qtd. + * only hc or qh_refresh() ever modify the overlay. + */ + if (likely(qtd != NULL)) { + struct ehci_qtd *dummy; + dma_addr_t dma; + __le32 token; + + /* to avoid racing the HC, use the dummy td instead of + * the first td of our list (becomes new dummy). both + * tds stay deactivated until we're done, when the + * HC is allowed to fetch the old dummy (4.10.2). + */ + token = qtd->hw_token; + qtd->hw_token = HALT_BIT; + wmb(); + dummy = qh->dummy; + + dma = dummy->qtd_dma; + *dummy = *qtd; + dummy->qtd_dma = dma; + + list_del(&qtd->qtd_list); + list_add(&dummy->qtd_list, qtd_list); + list_splice(qtd_list, qh->qtd_list.prev); + + ehci_qtd_init(qtd, qtd->qtd_dma); + qh->dummy = qtd; + + /* hc must see the new dummy at list end */ + dma = qtd->qtd_dma; + qtd = list_entry(qh->qtd_list.prev, + struct ehci_qtd, qtd_list); + qtd->hw_next = QTD_NEXT(dma); + + /* let the hc process these next qtds */ + dummy->hw_token = (token & ~(0x80)); + wmb(); + dummy->hw_token = token; + + urb->hcpriv = qh_get(qh); + } + } + return qh; +} + +static int submit_async(struct oxu_hcd *oxu, struct urb *urb, + struct list_head *qtd_list, gfp_t mem_flags) +{ + struct ehci_qtd *qtd; + int epnum; + unsigned long flags; + struct ehci_qh *qh = NULL; + int rc = 0; + + qtd = list_entry(qtd_list->next, struct ehci_qtd, qtd_list); + epnum = urb->ep->desc.bEndpointAddress; + +#ifdef OXU_URB_TRACE + oxu_dbg(oxu, "%s %s urb %p ep%d%s len %d, qtd %p [qh %p]\n", + __func__, urb->dev->devpath, urb, + epnum & 0x0f, (epnum & USB_DIR_IN) ? "in" : "out", + urb->transfer_buffer_length, + qtd, urb->ep->hcpriv); +#endif + + spin_lock_irqsave(&oxu->lock, flags); + if (unlikely(!test_bit(HCD_FLAG_HW_ACCESSIBLE, + &oxu_to_hcd(oxu)->flags))) { + rc = -ESHUTDOWN; + goto done; + } + + qh = qh_append_tds(oxu, urb, qtd_list, epnum, &urb->ep->hcpriv); + if (unlikely(qh == NULL)) { + rc = -ENOMEM; + goto done; + } + + /* Control/bulk operations through TTs don't need scheduling, + * the HC and TT handle it when the TT has a buffer ready. + */ + if (likely(qh->qh_state == QH_STATE_IDLE)) + qh_link_async(oxu, qh_get(qh)); +done: + spin_unlock_irqrestore(&oxu->lock, flags); + if (unlikely(qh == NULL)) + qtd_list_free(oxu, urb, qtd_list); + return rc; +} + +/* The async qh for the qtds being reclaimed are now unlinked from the HC */ + +static void end_unlink_async(struct oxu_hcd *oxu) +{ + struct ehci_qh *qh = oxu->reclaim; + struct ehci_qh *next; + + timer_action_done(oxu, TIMER_IAA_WATCHDOG); + + qh->qh_state = QH_STATE_IDLE; + qh->qh_next.qh = NULL; + qh_put(qh); /* refcount from reclaim */ + + /* other unlink(s) may be pending (in QH_STATE_UNLINK_WAIT) */ + next = qh->reclaim; + oxu->reclaim = next; + oxu->reclaim_ready = 0; + qh->reclaim = NULL; + + qh_completions(oxu, qh); + + if (!list_empty(&qh->qtd_list) + && HC_IS_RUNNING(oxu_to_hcd(oxu)->state)) + qh_link_async(oxu, qh); + else { + qh_put(qh); /* refcount from async list */ + + /* it's not free to turn the async schedule on/off; leave it + * active but idle for a while once it empties. + */ + if (HC_IS_RUNNING(oxu_to_hcd(oxu)->state) + && oxu->async->qh_next.qh == NULL) + timer_action(oxu, TIMER_ASYNC_OFF); + } + + if (next) { + oxu->reclaim = NULL; + start_unlink_async(oxu, next); + } +} + +/* makes sure the async qh will become idle */ +/* caller must own oxu->lock */ + +static void start_unlink_async(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + int cmd = readl(&oxu->regs->command); + struct ehci_qh *prev; + +#ifdef DEBUG + assert_spin_locked(&oxu->lock); + if (oxu->reclaim || (qh->qh_state != QH_STATE_LINKED + && qh->qh_state != QH_STATE_UNLINK_WAIT)) + BUG(); +#endif + + /* stop async schedule right now? */ + if (unlikely(qh == oxu->async)) { + /* can't get here without STS_ASS set */ + if (oxu_to_hcd(oxu)->state != HC_STATE_HALT + && !oxu->reclaim) { + /* ... and CMD_IAAD clear */ + writel(cmd & ~CMD_ASE, &oxu->regs->command); + wmb(); + /* handshake later, if we need to */ + timer_action_done(oxu, TIMER_ASYNC_OFF); + } + return; + } + + qh->qh_state = QH_STATE_UNLINK; + oxu->reclaim = qh = qh_get(qh); + + prev = oxu->async; + while (prev->qh_next.qh != qh) + prev = prev->qh_next.qh; + + prev->hw_next = qh->hw_next; + prev->qh_next = qh->qh_next; + wmb(); + + if (unlikely(oxu_to_hcd(oxu)->state == HC_STATE_HALT)) { + /* if (unlikely(qh->reclaim != 0)) + * this will recurse, probably not much + */ + end_unlink_async(oxu); + return; + } + + oxu->reclaim_ready = 0; + cmd |= CMD_IAAD; + writel(cmd, &oxu->regs->command); + (void) readl(&oxu->regs->command); + timer_action(oxu, TIMER_IAA_WATCHDOG); +} + +static void scan_async(struct oxu_hcd *oxu) +{ + struct ehci_qh *qh; + enum ehci_timer_action action = TIMER_IO_WATCHDOG; + + if (!++(oxu->stamp)) + oxu->stamp++; + timer_action_done(oxu, TIMER_ASYNC_SHRINK); +rescan: + qh = oxu->async->qh_next.qh; + if (likely(qh != NULL)) { + do { + /* clean any finished work for this qh */ + if (!list_empty(&qh->qtd_list) + && qh->stamp != oxu->stamp) { + int temp; + + /* unlinks could happen here; completion + * reporting drops the lock. rescan using + * the latest schedule, but don't rescan + * qhs we already finished (no looping). + */ + qh = qh_get(qh); + qh->stamp = oxu->stamp; + temp = qh_completions(oxu, qh); + qh_put(qh); + if (temp != 0) + goto rescan; + } + + /* unlink idle entries, reducing HC PCI usage as well + * as HCD schedule-scanning costs. delay for any qh + * we just scanned, there's a not-unusual case that it + * doesn't stay idle for long. + * (plus, avoids some kind of re-activation race.) + */ + if (list_empty(&qh->qtd_list)) { + if (qh->stamp == oxu->stamp) + action = TIMER_ASYNC_SHRINK; + else if (!oxu->reclaim + && qh->qh_state == QH_STATE_LINKED) + start_unlink_async(oxu, qh); + } + + qh = qh->qh_next.qh; + } while (qh); + } + if (action == TIMER_ASYNC_SHRINK) + timer_action(oxu, TIMER_ASYNC_SHRINK); +} + +/* + * periodic_next_shadow - return "next" pointer on shadow list + * @periodic: host pointer to qh/itd/sitd + * @tag: hardware tag for type of this record + */ +static union ehci_shadow *periodic_next_shadow(union ehci_shadow *periodic, + __le32 tag) +{ + switch (tag) { + default: + case Q_TYPE_QH: + return &periodic->qh->qh_next; + } +} + +/* caller must hold oxu->lock */ +static void periodic_unlink(struct oxu_hcd *oxu, unsigned frame, void *ptr) +{ + union ehci_shadow *prev_p = &oxu->pshadow[frame]; + __le32 *hw_p = &oxu->periodic[frame]; + union ehci_shadow here = *prev_p; + + /* find predecessor of "ptr"; hw and shadow lists are in sync */ + while (here.ptr && here.ptr != ptr) { + prev_p = periodic_next_shadow(prev_p, Q_NEXT_TYPE(*hw_p)); + hw_p = here.hw_next; + here = *prev_p; + } + /* an interrupt entry (at list end) could have been shared */ + if (!here.ptr) + return; + + /* update shadow and hardware lists ... the old "next" pointers + * from ptr may still be in use, the caller updates them. + */ + *prev_p = *periodic_next_shadow(&here, Q_NEXT_TYPE(*hw_p)); + *hw_p = *here.hw_next; +} + +/* how many of the uframe's 125 usecs are allocated? */ +static unsigned short periodic_usecs(struct oxu_hcd *oxu, + unsigned frame, unsigned uframe) +{ + __le32 *hw_p = &oxu->periodic[frame]; + union ehci_shadow *q = &oxu->pshadow[frame]; + unsigned usecs = 0; + + while (q->ptr) { + switch (Q_NEXT_TYPE(*hw_p)) { + case Q_TYPE_QH: + default: + /* is it in the S-mask? */ + if (q->qh->hw_info2 & cpu_to_le32(1 << uframe)) + usecs += q->qh->usecs; + /* ... or C-mask? */ + if (q->qh->hw_info2 & cpu_to_le32(1 << (8 + uframe))) + usecs += q->qh->c_usecs; + hw_p = &q->qh->hw_next; + q = &q->qh->qh_next; + break; + } + } +#ifdef DEBUG + if (usecs > 100) + oxu_err(oxu, "uframe %d sched overrun: %d usecs\n", + frame * 8 + uframe, usecs); +#endif + return usecs; +} + +static int enable_periodic(struct oxu_hcd *oxu) +{ + u32 cmd; + int status; + + /* did clearing PSE did take effect yet? + * takes effect only at frame boundaries... + */ + status = handshake(oxu, &oxu->regs->status, STS_PSS, 0, 9 * 125); + if (status != 0) { + oxu_to_hcd(oxu)->state = HC_STATE_HALT; + return status; + } + + cmd = readl(&oxu->regs->command) | CMD_PSE; + writel(cmd, &oxu->regs->command); + /* posted write ... PSS happens later */ + oxu_to_hcd(oxu)->state = HC_STATE_RUNNING; + + /* make sure ehci_work scans these */ + oxu->next_uframe = readl(&oxu->regs->frame_index) + % (oxu->periodic_size << 3); + return 0; +} + +static int disable_periodic(struct oxu_hcd *oxu) +{ + u32 cmd; + int status; + + /* did setting PSE not take effect yet? + * takes effect only at frame boundaries... + */ + status = handshake(oxu, &oxu->regs->status, STS_PSS, STS_PSS, 9 * 125); + if (status != 0) { + oxu_to_hcd(oxu)->state = HC_STATE_HALT; + return status; + } + + cmd = readl(&oxu->regs->command) & ~CMD_PSE; + writel(cmd, &oxu->regs->command); + /* posted write ... */ + + oxu->next_uframe = -1; + return 0; +} + +/* periodic schedule slots have iso tds (normal or split) first, then a + * sparse tree for active interrupt transfers. + * + * this just links in a qh; caller guarantees uframe masks are set right. + * no FSTN support (yet; oxu 0.96+) + */ +static int qh_link_periodic(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + unsigned i; + unsigned period = qh->period; + + dev_dbg(&qh->dev->dev, + "link qh%d-%04x/%p start %d [%d/%d us]\n", + period, le32_to_cpup(&qh->hw_info2) & (QH_CMASK | QH_SMASK), + qh, qh->start, qh->usecs, qh->c_usecs); + + /* high bandwidth, or otherwise every microframe */ + if (period == 0) + period = 1; + + for (i = qh->start; i < oxu->periodic_size; i += period) { + union ehci_shadow *prev = &oxu->pshadow[i]; + __le32 *hw_p = &oxu->periodic[i]; + union ehci_shadow here = *prev; + __le32 type = 0; + + /* skip the iso nodes at list head */ + while (here.ptr) { + type = Q_NEXT_TYPE(*hw_p); + if (type == Q_TYPE_QH) + break; + prev = periodic_next_shadow(prev, type); + hw_p = &here.qh->hw_next; + here = *prev; + } + + /* sorting each branch by period (slow-->fast) + * enables sharing interior tree nodes + */ + while (here.ptr && qh != here.qh) { + if (qh->period > here.qh->period) + break; + prev = &here.qh->qh_next; + hw_p = &here.qh->hw_next; + here = *prev; + } + /* link in this qh, unless some earlier pass did that */ + if (qh != here.qh) { + qh->qh_next = here; + if (here.qh) + qh->hw_next = *hw_p; + wmb(); + prev->qh = qh; + *hw_p = QH_NEXT(qh->qh_dma); + } + } + qh->qh_state = QH_STATE_LINKED; + qh_get(qh); + + /* update per-qh bandwidth for usbfs */ + oxu_to_hcd(oxu)->self.bandwidth_allocated += qh->period + ? ((qh->usecs + qh->c_usecs) / qh->period) + : (qh->usecs * 8); + + /* maybe enable periodic schedule processing */ + if (!oxu->periodic_sched++) + return enable_periodic(oxu); + + return 0; +} + +static void qh_unlink_periodic(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + unsigned i; + unsigned period; + + /* FIXME: + * IF this isn't high speed + * and this qh is active in the current uframe + * (and overlay token SplitXstate is false?) + * THEN + * qh->hw_info1 |= __constant_cpu_to_le32(1 << 7 "ignore"); + */ + + /* high bandwidth, or otherwise part of every microframe */ + period = qh->period; + if (period == 0) + period = 1; + + for (i = qh->start; i < oxu->periodic_size; i += period) + periodic_unlink(oxu, i, qh); + + /* update per-qh bandwidth for usbfs */ + oxu_to_hcd(oxu)->self.bandwidth_allocated -= qh->period + ? ((qh->usecs + qh->c_usecs) / qh->period) + : (qh->usecs * 8); + + dev_dbg(&qh->dev->dev, + "unlink qh%d-%04x/%p start %d [%d/%d us]\n", + qh->period, + le32_to_cpup(&qh->hw_info2) & (QH_CMASK | QH_SMASK), + qh, qh->start, qh->usecs, qh->c_usecs); + + /* qh->qh_next still "live" to HC */ + qh->qh_state = QH_STATE_UNLINK; + qh->qh_next.ptr = NULL; + qh_put(qh); + + /* maybe turn off periodic schedule */ + oxu->periodic_sched--; + if (!oxu->periodic_sched) + (void) disable_periodic(oxu); +} + +static void intr_deschedule(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + unsigned wait; + + qh_unlink_periodic(oxu, qh); + + /* simple/paranoid: always delay, expecting the HC needs to read + * qh->hw_next or finish a writeback after SPLIT/CSPLIT ... and + * expect khubd to clean up after any CSPLITs we won't issue. + * active high speed queues may need bigger delays... + */ + if (list_empty(&qh->qtd_list) + || (__constant_cpu_to_le32(QH_CMASK) & qh->hw_info2) != 0) + wait = 2; + else + wait = 55; /* worst case: 3 * 1024 */ + + udelay(wait); + qh->qh_state = QH_STATE_IDLE; + qh->hw_next = EHCI_LIST_END; + wmb(); +} + +static int check_period(struct oxu_hcd *oxu, + unsigned frame, unsigned uframe, + unsigned period, unsigned usecs) +{ + int claimed; + + /* complete split running into next frame? + * given FSTN support, we could sometimes check... + */ + if (uframe >= 8) + return 0; + + /* + * 80% periodic == 100 usec/uframe available + * convert "usecs we need" to "max already claimed" + */ + usecs = 100 - usecs; + + /* we "know" 2 and 4 uframe intervals were rejected; so + * for period 0, check _every_ microframe in the schedule. + */ + if (unlikely(period == 0)) { + do { + for (uframe = 0; uframe < 7; uframe++) { + claimed = periodic_usecs(oxu, frame, uframe); + if (claimed > usecs) + return 0; + } + } while ((frame += 1) < oxu->periodic_size); + + /* just check the specified uframe, at that period */ + } else { + do { + claimed = periodic_usecs(oxu, frame, uframe); + if (claimed > usecs) + return 0; + } while ((frame += period) < oxu->periodic_size); + } + + return 1; +} + +static int check_intr_schedule(struct oxu_hcd *oxu, + unsigned frame, unsigned uframe, + const struct ehci_qh *qh, __le32 *c_maskp) +{ + int retval = -ENOSPC; + + if (qh->c_usecs && uframe >= 6) /* FSTN territory? */ + goto done; + + if (!check_period(oxu, frame, uframe, qh->period, qh->usecs)) + goto done; + if (!qh->c_usecs) { + retval = 0; + *c_maskp = 0; + goto done; + } + +done: + return retval; +} + +/* "first fit" scheduling policy used the first time through, + * or when the previous schedule slot can't be re-used. + */ +static int qh_schedule(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + int status; + unsigned uframe; + __le32 c_mask; + unsigned frame; /* 0..(qh->period - 1), or NO_FRAME */ + + qh_refresh(oxu, qh); + qh->hw_next = EHCI_LIST_END; + frame = qh->start; + + /* reuse the previous schedule slots, if we can */ + if (frame < qh->period) { + uframe = ffs(le32_to_cpup(&qh->hw_info2) & QH_SMASK); + status = check_intr_schedule(oxu, frame, --uframe, + qh, &c_mask); + } else { + uframe = 0; + c_mask = 0; + status = -ENOSPC; + } + + /* else scan the schedule to find a group of slots such that all + * uframes have enough periodic bandwidth available. + */ + if (status) { + /* "normal" case, uframing flexible except with splits */ + if (qh->period) { + frame = qh->period - 1; + do { + for (uframe = 0; uframe < 8; uframe++) { + status = check_intr_schedule(oxu, + frame, uframe, qh, + &c_mask); + if (status == 0) + break; + } + } while (status && frame--); + + /* qh->period == 0 means every uframe */ + } else { + frame = 0; + status = check_intr_schedule(oxu, 0, 0, qh, &c_mask); + } + if (status) + goto done; + qh->start = frame; + + /* reset S-frame and (maybe) C-frame masks */ + qh->hw_info2 &= __constant_cpu_to_le32(~(QH_CMASK | QH_SMASK)); + qh->hw_info2 |= qh->period + ? cpu_to_le32(1 << uframe) + : __constant_cpu_to_le32(QH_SMASK); + qh->hw_info2 |= c_mask; + } else + oxu_dbg(oxu, "reused qh %p schedule\n", qh); + + /* stuff into the periodic schedule */ + status = qh_link_periodic(oxu, qh); +done: + return status; +} + +static int intr_submit(struct oxu_hcd *oxu, struct urb *urb, + struct list_head *qtd_list, gfp_t mem_flags) +{ + unsigned epnum; + unsigned long flags; + struct ehci_qh *qh; + int status = 0; + struct list_head empty; + + /* get endpoint and transfer/schedule data */ + epnum = urb->ep->desc.bEndpointAddress; + + spin_lock_irqsave(&oxu->lock, flags); + + if (unlikely(!test_bit(HCD_FLAG_HW_ACCESSIBLE, + &oxu_to_hcd(oxu)->flags))) { + status = -ESHUTDOWN; + goto done; + } + + /* get qh and force any scheduling errors */ + INIT_LIST_HEAD(&empty); + qh = qh_append_tds(oxu, urb, &empty, epnum, &urb->ep->hcpriv); + if (qh == NULL) { + status = -ENOMEM; + goto done; + } + if (qh->qh_state == QH_STATE_IDLE) { + status = qh_schedule(oxu, qh); + if (status != 0) + goto done; + } + + /* then queue the urb's tds to the qh */ + qh = qh_append_tds(oxu, urb, qtd_list, epnum, &urb->ep->hcpriv); + BUG_ON(qh == NULL); + + /* ... update usbfs periodic stats */ + oxu_to_hcd(oxu)->self.bandwidth_int_reqs++; + +done: + spin_unlock_irqrestore(&oxu->lock, flags); + if (status) + qtd_list_free(oxu, urb, qtd_list); + + return status; +} + +static inline int itd_submit(struct oxu_hcd *oxu, struct urb *urb, + gfp_t mem_flags) +{ + oxu_dbg(oxu, "iso support is missing!\n"); + return -ENOSYS; +} + +static inline int sitd_submit(struct oxu_hcd *oxu, struct urb *urb, + gfp_t mem_flags) +{ + oxu_dbg(oxu, "split iso support is missing!\n"); + return -ENOSYS; +} + +static void scan_periodic(struct oxu_hcd *oxu) +{ + unsigned frame, clock, now_uframe, mod; + unsigned modified; + + mod = oxu->periodic_size << 3; + + /* + * When running, scan from last scan point up to "now" + * else clean up by scanning everything that's left. + * Touches as few pages as possible: cache-friendly. + */ + now_uframe = oxu->next_uframe; + if (HC_IS_RUNNING(oxu_to_hcd(oxu)->state)) + clock = readl(&oxu->regs->frame_index); + else + clock = now_uframe + mod - 1; + clock %= mod; + + for (;;) { + union ehci_shadow q, *q_p; + __le32 type, *hw_p; + unsigned uframes; + + /* don't scan past the live uframe */ + frame = now_uframe >> 3; + if (frame == (clock >> 3)) + uframes = now_uframe & 0x07; + else { + /* safe to scan the whole frame at once */ + now_uframe |= 0x07; + uframes = 8; + } + +restart: + /* scan each element in frame's queue for completions */ + q_p = &oxu->pshadow[frame]; + hw_p = &oxu->periodic[frame]; + q.ptr = q_p->ptr; + type = Q_NEXT_TYPE(*hw_p); + modified = 0; + + while (q.ptr != NULL) { + union ehci_shadow temp; + int live; + + live = HC_IS_RUNNING(oxu_to_hcd(oxu)->state); + switch (type) { + case Q_TYPE_QH: + /* handle any completions */ + temp.qh = qh_get(q.qh); + type = Q_NEXT_TYPE(q.qh->hw_next); + q = q.qh->qh_next; + modified = qh_completions(oxu, temp.qh); + if (unlikely(list_empty(&temp.qh->qtd_list))) + intr_deschedule(oxu, temp.qh); + qh_put(temp.qh); + break; + default: + dbg("corrupt type %d frame %d shadow %p", + type, frame, q.ptr); + q.ptr = NULL; + } + + /* assume completion callbacks modify the queue */ + if (unlikely(modified)) + goto restart; + } + + /* Stop when we catch up to the HC */ + + /* FIXME: this assumes we won't get lapped when + * latencies climb; that should be rare, but... + * detect it, and just go all the way around. + * FLR might help detect this case, so long as latencies + * don't exceed periodic_size msec (default 1.024 sec). + */ + + /* FIXME: likewise assumes HC doesn't halt mid-scan */ + + if (now_uframe == clock) { + unsigned now; + + if (!HC_IS_RUNNING(oxu_to_hcd(oxu)->state)) + break; + oxu->next_uframe = now_uframe; + now = readl(&oxu->regs->frame_index) % mod; + if (now_uframe == now) + break; + + /* rescan the rest of this frame, then ... */ + clock = now; + } else { + now_uframe++; + now_uframe %= mod; + } + } +} + +/* On some systems, leaving remote wakeup enabled prevents system shutdown. + * The firmware seems to think that powering off is a wakeup event! + * This routine turns off remote wakeup and everything else, on all ports. + */ +static void ehci_turn_off_all_ports(struct oxu_hcd *oxu) +{ + int port = HCS_N_PORTS(oxu->hcs_params); + + while (port--) + writel(PORT_RWC_BITS, &oxu->regs->port_status[port]); +} + +static void ehci_port_power(struct oxu_hcd *oxu, int is_on) +{ + unsigned port; + + if (!HCS_PPC(oxu->hcs_params)) + return; + + oxu_dbg(oxu, "...power%s ports...\n", is_on ? "up" : "down"); + for (port = HCS_N_PORTS(oxu->hcs_params); port > 0; ) + (void) oxu_hub_control(oxu_to_hcd(oxu), + is_on ? SetPortFeature : ClearPortFeature, + USB_PORT_FEAT_POWER, + port--, NULL, 0); + msleep(20); +} + +/* Called from some interrupts, timers, and so on. + * It calls driver completion functions, after dropping oxu->lock. + */ +static void ehci_work(struct oxu_hcd *oxu) +{ + timer_action_done(oxu, TIMER_IO_WATCHDOG); + if (oxu->reclaim_ready) + end_unlink_async(oxu); + + /* another CPU may drop oxu->lock during a schedule scan while + * it reports urb completions. this flag guards against bogus + * attempts at re-entrant schedule scanning. + */ + if (oxu->scanning) + return; + oxu->scanning = 1; + scan_async(oxu); + if (oxu->next_uframe != -1) + scan_periodic(oxu); + oxu->scanning = 0; + + /* the IO watchdog guards against hardware or driver bugs that + * misplace IRQs, and should let us run completely without IRQs. + * such lossage has been observed on both VT6202 and VT8235. + */ + if (HC_IS_RUNNING(oxu_to_hcd(oxu)->state) && + (oxu->async->qh_next.ptr != NULL || + oxu->periodic_sched != 0)) + timer_action(oxu, TIMER_IO_WATCHDOG); +} + +static void unlink_async(struct oxu_hcd *oxu, struct ehci_qh *qh) +{ + /* if we need to use IAA and it's busy, defer */ + if (qh->qh_state == QH_STATE_LINKED + && oxu->reclaim + && HC_IS_RUNNING(oxu_to_hcd(oxu)->state)) { + struct ehci_qh *last; + + for (last = oxu->reclaim; + last->reclaim; + last = last->reclaim) + continue; + qh->qh_state = QH_STATE_UNLINK_WAIT; + last->reclaim = qh; + + /* bypass IAA if the hc can't care */ + } else if (!HC_IS_RUNNING(oxu_to_hcd(oxu)->state) && oxu->reclaim) + end_unlink_async(oxu); + + /* something else might have unlinked the qh by now */ + if (qh->qh_state == QH_STATE_LINKED) + start_unlink_async(oxu, qh); +} + +/* + * USB host controller methods + */ + +static irqreturn_t oxu210_hcd_irq(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + u32 status, pcd_status = 0; + int bh; + + spin_lock(&oxu->lock); + + status = readl(&oxu->regs->status); + + /* e.g. cardbus physical eject */ + if (status == ~(u32) 0) { + oxu_dbg(oxu, "device removed\n"); + goto dead; + } + + status &= INTR_MASK; + if (!status) { /* irq sharing? */ + spin_unlock(&oxu->lock); + return IRQ_NONE; + } + + /* clear (just) interrupts */ + writel(status, &oxu->regs->status); + readl(&oxu->regs->command); /* unblock posted write */ + bh = 0; + +#ifdef OXU_VERBOSE_DEBUG + /* unrequested/ignored: Frame List Rollover */ + dbg_status(oxu, "irq", status); +#endif + + /* INT, ERR, and IAA interrupt rates can be throttled */ + + /* normal [4.15.1.2] or error [4.15.1.1] completion */ + if (likely((status & (STS_INT|STS_ERR)) != 0)) + bh = 1; + + /* complete the unlinking of some qh [4.15.2.3] */ + if (status & STS_IAA) { + oxu->reclaim_ready = 1; + bh = 1; + } + + /* remote wakeup [4.3.1] */ + if (status & STS_PCD) { + unsigned i = HCS_N_PORTS(oxu->hcs_params); + pcd_status = status; + + /* resume root hub? */ + if (!(readl(&oxu->regs->command) & CMD_RUN)) + usb_hcd_resume_root_hub(hcd); + + while (i--) { + int pstatus = readl(&oxu->regs->port_status[i]); + + if (pstatus & PORT_OWNER) + continue; + if (!(pstatus & PORT_RESUME) + || oxu->reset_done[i] != 0) + continue; + + /* start 20 msec resume signaling from this port, + * and make khubd collect PORT_STAT_C_SUSPEND to + * stop that signaling. + */ + oxu->reset_done[i] = jiffies + msecs_to_jiffies(20); + oxu_dbg(oxu, "port %d remote wakeup\n", i + 1); + mod_timer(&hcd->rh_timer, oxu->reset_done[i]); + } + } + + /* PCI errors [4.15.2.4] */ + if (unlikely((status & STS_FATAL) != 0)) { + /* bogus "fatal" IRQs appear on some chips... why? */ + status = readl(&oxu->regs->status); + dbg_cmd(oxu, "fatal", readl(&oxu->regs->command)); + dbg_status(oxu, "fatal", status); + if (status & STS_HALT) { + oxu_err(oxu, "fatal error\n"); +dead: + ehci_reset(oxu); + writel(0, &oxu->regs->configured_flag); + /* generic layer kills/unlinks all urbs, then + * uses oxu_stop to clean up the rest + */ + bh = 1; + } + } + + if (bh) + ehci_work(oxu); + spin_unlock(&oxu->lock); + if (pcd_status & STS_PCD) + usb_hcd_poll_rh_status(hcd); + return IRQ_HANDLED; +} + +static irqreturn_t oxu_irq(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + int ret = IRQ_HANDLED; + + u32 status = oxu_readl(hcd->regs, OXU_CHIPIRQSTATUS); + u32 enable = oxu_readl(hcd->regs, OXU_CHIPIRQEN_SET); + + /* Disable all interrupt */ + oxu_writel(hcd->regs, OXU_CHIPIRQEN_CLR, enable); + + if ((oxu->is_otg && (status & OXU_USBOTGI)) || + (!oxu->is_otg && (status & OXU_USBSPHI))) + oxu210_hcd_irq(hcd); + else + ret = IRQ_NONE; + + /* Enable all interrupt back */ + oxu_writel(hcd->regs, OXU_CHIPIRQEN_SET, enable); + + return ret; +} + +static void oxu_watchdog(unsigned long param) +{ + struct oxu_hcd *oxu = (struct oxu_hcd *) param; + unsigned long flags; + + spin_lock_irqsave(&oxu->lock, flags); + + /* lost IAA irqs wedge things badly; seen with a vt8235 */ + if (oxu->reclaim) { + u32 status = readl(&oxu->regs->status); + if (status & STS_IAA) { + oxu_vdbg(oxu, "lost IAA\n"); + writel(STS_IAA, &oxu->regs->status); + oxu->reclaim_ready = 1; + } + } + + /* stop async processing after it's idled a bit */ + if (test_bit(TIMER_ASYNC_OFF, &oxu->actions)) + start_unlink_async(oxu, oxu->async); + + /* oxu could run by timer, without IRQs ... */ + ehci_work(oxu); + + spin_unlock_irqrestore(&oxu->lock, flags); +} + +/* One-time init, only for memory state. + */ +static int oxu_hcd_init(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + u32 temp; + int retval; + u32 hcc_params; + + spin_lock_init(&oxu->lock); + + init_timer(&oxu->watchdog); + oxu->watchdog.function = oxu_watchdog; + oxu->watchdog.data = (unsigned long) oxu; + + /* + * hw default: 1K periodic list heads, one per frame. + * periodic_size can shrink by USBCMD update if hcc_params allows. + */ + oxu->periodic_size = DEFAULT_I_TDPS; + retval = ehci_mem_init(oxu, GFP_KERNEL); + if (retval < 0) + return retval; + + /* controllers may cache some of the periodic schedule ... */ + hcc_params = readl(&oxu->caps->hcc_params); + if (HCC_ISOC_CACHE(hcc_params)) /* full frame cache */ + oxu->i_thresh = 8; + else /* N microframes cached */ + oxu->i_thresh = 2 + HCC_ISOC_THRES(hcc_params); + + oxu->reclaim = NULL; + oxu->reclaim_ready = 0; + oxu->next_uframe = -1; + + /* + * dedicate a qh for the async ring head, since we couldn't unlink + * a 'real' qh without stopping the async schedule [4.8]. use it + * as the 'reclamation list head' too. + * its dummy is used in hw_alt_next of many tds, to prevent the qh + * from automatically advancing to the next td after short reads. + */ + oxu->async->qh_next.qh = NULL; + oxu->async->hw_next = QH_NEXT(oxu->async->qh_dma); + oxu->async->hw_info1 = cpu_to_le32(QH_HEAD); + oxu->async->hw_token = cpu_to_le32(QTD_STS_HALT); + oxu->async->hw_qtd_next = EHCI_LIST_END; + oxu->async->qh_state = QH_STATE_LINKED; + oxu->async->hw_alt_next = QTD_NEXT(oxu->async->dummy->qtd_dma); + + /* clear interrupt enables, set irq latency */ + if (log2_irq_thresh < 0 || log2_irq_thresh > 6) + log2_irq_thresh = 0; + temp = 1 << (16 + log2_irq_thresh); + if (HCC_CANPARK(hcc_params)) { + /* HW default park == 3, on hardware that supports it (like + * NVidia and ALI silicon), maximizes throughput on the async + * schedule by avoiding QH fetches between transfers. + * + * With fast usb storage devices and NForce2, "park" seems to + * make problems: throughput reduction (!), data errors... + */ + if (park) { + park = min(park, (unsigned) 3); + temp |= CMD_PARK; + temp |= park << 8; + } + oxu_dbg(oxu, "park %d\n", park); + } + if (HCC_PGM_FRAMELISTLEN(hcc_params)) { + /* periodic schedule size can be smaller than default */ + temp &= ~(3 << 2); + temp |= (EHCI_TUNE_FLS << 2); + } + oxu->command = temp; + + return 0; +} + +/* Called during probe() after chip reset completes. + */ +static int oxu_reset(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + int ret; + + spin_lock_init(&oxu->mem_lock); + INIT_LIST_HEAD(&oxu->urb_list); + oxu->urb_len = 0; + + /* FIMXE */ + hcd->self.controller->dma_mask = 0UL; + + if (oxu->is_otg) { + oxu->caps = hcd->regs + OXU_OTG_CAP_OFFSET; + oxu->regs = hcd->regs + OXU_OTG_CAP_OFFSET + \ + HC_LENGTH(readl(&oxu->caps->hc_capbase)); + + oxu->mem = hcd->regs + OXU_SPH_MEM; + } else { + oxu->caps = hcd->regs + OXU_SPH_CAP_OFFSET; + oxu->regs = hcd->regs + OXU_SPH_CAP_OFFSET + \ + HC_LENGTH(readl(&oxu->caps->hc_capbase)); + + oxu->mem = hcd->regs + OXU_OTG_MEM; + } + + oxu->hcs_params = readl(&oxu->caps->hcs_params); + oxu->sbrn = 0x20; + + ret = oxu_hcd_init(hcd); + if (ret) + return ret; + + return 0; +} + +static int oxu_run(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + int retval; + u32 temp, hcc_params; + + hcd->uses_new_polling = 1; + hcd->poll_rh = 0; + + /* EHCI spec section 4.1 */ + retval = ehci_reset(oxu); + if (retval != 0) { + ehci_mem_cleanup(oxu); + return retval; + } + writel(oxu->periodic_dma, &oxu->regs->frame_list); + writel((u32) oxu->async->qh_dma, &oxu->regs->async_next); + + /* hcc_params controls whether oxu->regs->segment must (!!!) + * be used; it constrains QH/ITD/SITD and QTD locations. + * pci_pool consistent memory always uses segment zero. + * streaming mappings for I/O buffers, like pci_map_single(), + * can return segments above 4GB, if the device allows. + * + * NOTE: the dma mask is visible through dma_supported(), so + * drivers can pass this info along ... like NETIF_F_HIGHDMA, + * Scsi_Host.highmem_io, and so forth. It's readonly to all + * host side drivers though. + */ + hcc_params = readl(&oxu->caps->hcc_params); + if (HCC_64BIT_ADDR(hcc_params)) + writel(0, &oxu->regs->segment); + + oxu->command &= ~(CMD_LRESET | CMD_IAAD | CMD_PSE | + CMD_ASE | CMD_RESET); + oxu->command |= CMD_RUN; + writel(oxu->command, &oxu->regs->command); + dbg_cmd(oxu, "init", oxu->command); + + /* + * Start, enabling full USB 2.0 functionality ... usb 1.1 devices + * are explicitly handed to companion controller(s), so no TT is + * involved with the root hub. (Except where one is integrated, + * and there's no companion controller unless maybe for USB OTG.) + */ + hcd->state = HC_STATE_RUNNING; + writel(FLAG_CF, &oxu->regs->configured_flag); + readl(&oxu->regs->command); /* unblock posted writes */ + + temp = HC_VERSION(readl(&oxu->caps->hc_capbase)); + oxu_info(oxu, "USB %x.%x started, quasi-EHCI %x.%02x, driver %s%s\n", + ((oxu->sbrn & 0xf0)>>4), (oxu->sbrn & 0x0f), + temp >> 8, temp & 0xff, DRIVER_VERSION, + ignore_oc ? ", overcurrent ignored" : ""); + + writel(INTR_MASK, &oxu->regs->intr_enable); /* Turn On Interrupts */ + + return 0; +} + +static void oxu_stop(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + + /* Turn off port power on all root hub ports. */ + ehci_port_power(oxu, 0); + + /* no more interrupts ... */ + del_timer_sync(&oxu->watchdog); + + spin_lock_irq(&oxu->lock); + if (HC_IS_RUNNING(hcd->state)) + ehci_quiesce(oxu); + + ehci_reset(oxu); + writel(0, &oxu->regs->intr_enable); + spin_unlock_irq(&oxu->lock); + + /* let companion controllers work when we aren't */ + writel(0, &oxu->regs->configured_flag); + + /* root hub is shut down separately (first, when possible) */ + spin_lock_irq(&oxu->lock); + if (oxu->async) + ehci_work(oxu); + spin_unlock_irq(&oxu->lock); + ehci_mem_cleanup(oxu); + + dbg_status(oxu, "oxu_stop completed", readl(&oxu->regs->status)); +} + +/* Kick in for silicon on any bus (not just pci, etc). + * This forcibly disables dma and IRQs, helping kexec and other cases + * where the next system software may expect clean state. + */ +static void oxu_shutdown(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + + (void) ehci_halt(oxu); + ehci_turn_off_all_ports(oxu); + + /* make BIOS/etc use companion controller during reboot */ + writel(0, &oxu->regs->configured_flag); + + /* unblock posted writes */ + readl(&oxu->regs->configured_flag); +} + +/* Non-error returns are a promise to giveback() the urb later + * we drop ownership so next owner (or urb unlink) can get it + * + * urb + dev is in hcd.self.controller.urb_list + * we're queueing TDs onto software and hardware lists + * + * hcd-specific init for hcpriv hasn't been done yet + * + * NOTE: control, bulk, and interrupt share the same code to append TDs + * to a (possibly active) QH, and the same QH scanning code. + */ +static int __oxu_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, + gfp_t mem_flags) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + struct list_head qtd_list; + + INIT_LIST_HEAD(&qtd_list); + + switch (usb_pipetype(urb->pipe)) { + case PIPE_CONTROL: + case PIPE_BULK: + default: + if (!qh_urb_transaction(oxu, urb, &qtd_list, mem_flags)) + return -ENOMEM; + return submit_async(oxu, urb, &qtd_list, mem_flags); + + case PIPE_INTERRUPT: + if (!qh_urb_transaction(oxu, urb, &qtd_list, mem_flags)) + return -ENOMEM; + return intr_submit(oxu, urb, &qtd_list, mem_flags); + + case PIPE_ISOCHRONOUS: + if (urb->dev->speed == USB_SPEED_HIGH) + return itd_submit(oxu, urb, mem_flags); + else + return sitd_submit(oxu, urb, mem_flags); + } +} + +/* This function is responsible for breaking URBs with big data size + * into smaller size and processing small urbs in sequence. + */ +static int oxu_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, + gfp_t mem_flags) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + int num, rem; + int transfer_buffer_length; + void *transfer_buffer; + struct urb *murb; + int i, ret; + + /* If not bulk pipe just enqueue the URB */ + if (!usb_pipebulk(urb->pipe)) + return __oxu_urb_enqueue(hcd, urb, mem_flags); + + /* Otherwise we should verify the USB transfer buffer size! */ + transfer_buffer = urb->transfer_buffer; + transfer_buffer_length = urb->transfer_buffer_length; + + num = urb->transfer_buffer_length / 4096; + rem = urb->transfer_buffer_length % 4096; + if (rem != 0) + num++; + + /* If URB is smaller than 4096 bytes just enqueue it! */ + if (num == 1) + return __oxu_urb_enqueue(hcd, urb, mem_flags); + + /* Ok, we have more job to do! :) */ + + for (i = 0; i < num - 1; i++) { + /* Get free micro URB poll till a free urb is recieved */ + + do { + murb = (struct urb *) oxu_murb_alloc(oxu); + if (!murb) + schedule(); + } while (!murb); + + /* Coping the urb */ + memcpy(murb, urb, sizeof(struct urb)); + + murb->transfer_buffer_length = 4096; + murb->transfer_buffer = transfer_buffer + i * 4096; + + /* Null pointer for the encodes that this is a micro urb */ + murb->complete = NULL; + + ((struct oxu_murb *) murb)->main = urb; + ((struct oxu_murb *) murb)->last = 0; + + /* This loop is to guarantee urb to be processed when there's + * not enough resources at a particular time by retrying. + */ + do { + ret = __oxu_urb_enqueue(hcd, murb, mem_flags); + if (ret) + schedule(); + } while (ret); + } + + /* Last urb requires special handling */ + + /* Get free micro URB poll till a free urb is recieved */ + do { + murb = (struct urb *) oxu_murb_alloc(oxu); + if (!murb) + schedule(); + } while (!murb); + + /* Coping the urb */ + memcpy(murb, urb, sizeof(struct urb)); + + murb->transfer_buffer_length = rem > 0 ? rem : 4096; + murb->transfer_buffer = transfer_buffer + (num - 1) * 4096; + + /* Null pointer for the encodes that this is a micro urb */ + murb->complete = NULL; + + ((struct oxu_murb *) murb)->main = urb; + ((struct oxu_murb *) murb)->last = 1; + + do { + ret = __oxu_urb_enqueue(hcd, murb, mem_flags); + if (ret) + schedule(); + } while (ret); + + return ret; +} + +/* Remove from hardware lists. + * Completions normally happen asynchronously + */ +static int oxu_urb_dequeue(struct usb_hcd *hcd, struct urb *urb, int status) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + struct ehci_qh *qh; + unsigned long flags; + + spin_lock_irqsave(&oxu->lock, flags); + switch (usb_pipetype(urb->pipe)) { + case PIPE_CONTROL: + case PIPE_BULK: + default: + qh = (struct ehci_qh *) urb->hcpriv; + if (!qh) + break; + unlink_async(oxu, qh); + break; + + case PIPE_INTERRUPT: + qh = (struct ehci_qh *) urb->hcpriv; + if (!qh) + break; + switch (qh->qh_state) { + case QH_STATE_LINKED: + intr_deschedule(oxu, qh); + /* FALL THROUGH */ + case QH_STATE_IDLE: + qh_completions(oxu, qh); + break; + default: + oxu_dbg(oxu, "bogus qh %p state %d\n", + qh, qh->qh_state); + goto done; + } + + /* reschedule QH iff another request is queued */ + if (!list_empty(&qh->qtd_list) + && HC_IS_RUNNING(hcd->state)) { + int status; + + status = qh_schedule(oxu, qh); + spin_unlock_irqrestore(&oxu->lock, flags); + + if (status != 0) { + /* shouldn't happen often, but ... + * FIXME kill those tds' urbs + */ + err("can't reschedule qh %p, err %d", + qh, status); + } + return status; + } + break; + } +done: + spin_unlock_irqrestore(&oxu->lock, flags); + return 0; +} + +/* Bulk qh holds the data toggle */ +static void oxu_endpoint_disable(struct usb_hcd *hcd, + struct usb_host_endpoint *ep) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + unsigned long flags; + struct ehci_qh *qh, *tmp; + + /* ASSERT: any requests/urbs are being unlinked */ + /* ASSERT: nobody can be submitting urbs for this any more */ + +rescan: + spin_lock_irqsave(&oxu->lock, flags); + qh = ep->hcpriv; + if (!qh) + goto done; + + /* endpoints can be iso streams. for now, we don't + * accelerate iso completions ... so spin a while. + */ + if (qh->hw_info1 == 0) { + oxu_vdbg(oxu, "iso delay\n"); + goto idle_timeout; + } + + if (!HC_IS_RUNNING(hcd->state)) + qh->qh_state = QH_STATE_IDLE; + switch (qh->qh_state) { + case QH_STATE_LINKED: + for (tmp = oxu->async->qh_next.qh; + tmp && tmp != qh; + tmp = tmp->qh_next.qh) + continue; + /* periodic qh self-unlinks on empty */ + if (!tmp) + goto nogood; + unlink_async(oxu, qh); + /* FALL THROUGH */ + case QH_STATE_UNLINK: /* wait for hw to finish? */ +idle_timeout: + spin_unlock_irqrestore(&oxu->lock, flags); + schedule_timeout_uninterruptible(1); + goto rescan; + case QH_STATE_IDLE: /* fully unlinked */ + if (list_empty(&qh->qtd_list)) { + qh_put(qh); + break; + } + /* else FALL THROUGH */ + default: +nogood: + /* caller was supposed to have unlinked any requests; + * that's not our job. just leak this memory. + */ + oxu_err(oxu, "qh %p (#%02x) state %d%s\n", + qh, ep->desc.bEndpointAddress, qh->qh_state, + list_empty(&qh->qtd_list) ? "" : "(has tds)"); + break; + } + ep->hcpriv = NULL; +done: + spin_unlock_irqrestore(&oxu->lock, flags); + return; +} + +static int oxu_get_frame(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + + return (readl(&oxu->regs->frame_index) >> 3) % + oxu->periodic_size; +} + +/* Build "status change" packet (one or two bytes) from HC registers */ +static int oxu_hub_status_data(struct usb_hcd *hcd, char *buf) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + u32 temp, mask, status = 0; + int ports, i, retval = 1; + unsigned long flags; + + /* if !USB_SUSPEND, root hub timers won't get shut down ... */ + if (!HC_IS_RUNNING(hcd->state)) + return 0; + + /* init status to no-changes */ + buf[0] = 0; + ports = HCS_N_PORTS(oxu->hcs_params); + if (ports > 7) { + buf[1] = 0; + retval++; + } + + /* Some boards (mostly VIA?) report bogus overcurrent indications, + * causing massive log spam unless we completely ignore them. It + * may be relevant that VIA VT8235 controlers, where PORT_POWER is + * always set, seem to clear PORT_OCC and PORT_CSC when writing to + * PORT_POWER; that's surprising, but maybe within-spec. + */ + if (!ignore_oc) + mask = PORT_CSC | PORT_PEC | PORT_OCC; + else + mask = PORT_CSC | PORT_PEC; + + /* no hub change reports (bit 0) for now (power, ...) */ + + /* port N changes (bit N)? */ + spin_lock_irqsave(&oxu->lock, flags); + for (i = 0; i < ports; i++) { + temp = readl(&oxu->regs->port_status[i]); + + /* + * Return status information even for ports with OWNER set. + * Otherwise khubd wouldn't see the disconnect event when a + * high-speed device is switched over to the companion + * controller by the user. + */ + + if (!(temp & PORT_CONNECT)) + oxu->reset_done[i] = 0; + if ((temp & mask) != 0 || ((temp & PORT_RESUME) != 0 && + time_after_eq(jiffies, oxu->reset_done[i]))) { + if (i < 7) + buf[0] |= 1 << (i + 1); + else + buf[1] |= 1 << (i - 7); + status = STS_PCD; + } + } + /* FIXME autosuspend idle root hubs */ + spin_unlock_irqrestore(&oxu->lock, flags); + return status ? retval : 0; +} + +/* Returns the speed of a device attached to a port on the root hub. */ +static inline unsigned int oxu_port_speed(struct oxu_hcd *oxu, + unsigned int portsc) +{ + switch ((portsc >> 26) & 3) { + case 0: + return 0; + case 1: + return 1 << USB_PORT_FEAT_LOWSPEED; + case 2: + default: + return 1 << USB_PORT_FEAT_HIGHSPEED; + } +} + +#define PORT_WAKE_BITS (PORT_WKOC_E|PORT_WKDISC_E|PORT_WKCONN_E) +static int oxu_hub_control(struct usb_hcd *hcd, u16 typeReq, + u16 wValue, u16 wIndex, char *buf, u16 wLength) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + int ports = HCS_N_PORTS(oxu->hcs_params); + u32 __iomem *status_reg = &oxu->regs->port_status[wIndex - 1]; + u32 temp, status; + unsigned long flags; + int retval = 0; + unsigned selector; + + /* + * FIXME: support SetPortFeatures USB_PORT_FEAT_INDICATOR. + * HCS_INDICATOR may say we can change LEDs to off/amber/green. + * (track current state ourselves) ... blink for diagnostics, + * power, "this is the one", etc. EHCI spec supports this. + */ + + spin_lock_irqsave(&oxu->lock, flags); + switch (typeReq) { + case ClearHubFeature: + switch (wValue) { + case C_HUB_LOCAL_POWER: + case C_HUB_OVER_CURRENT: + /* no hub-wide feature/status flags */ + break; + default: + goto error; + } + break; + case ClearPortFeature: + if (!wIndex || wIndex > ports) + goto error; + wIndex--; + temp = readl(status_reg); + + /* + * Even if OWNER is set, so the port is owned by the + * companion controller, khubd needs to be able to clear + * the port-change status bits (especially + * USB_PORT_FEAT_C_CONNECTION). + */ + + switch (wValue) { + case USB_PORT_FEAT_ENABLE: + writel(temp & ~PORT_PE, status_reg); + break; + case USB_PORT_FEAT_C_ENABLE: + writel((temp & ~PORT_RWC_BITS) | PORT_PEC, status_reg); + break; + case USB_PORT_FEAT_SUSPEND: + if (temp & PORT_RESET) + goto error; + if (temp & PORT_SUSPEND) { + if ((temp & PORT_PE) == 0) + goto error; + /* resume signaling for 20 msec */ + temp &= ~(PORT_RWC_BITS | PORT_WAKE_BITS); + writel(temp | PORT_RESUME, status_reg); + oxu->reset_done[wIndex] = jiffies + + msecs_to_jiffies(20); + } + break; + case USB_PORT_FEAT_C_SUSPEND: + /* we auto-clear this feature */ + break; + case USB_PORT_FEAT_POWER: + if (HCS_PPC(oxu->hcs_params)) + writel(temp & ~(PORT_RWC_BITS | PORT_POWER), + status_reg); + break; + case USB_PORT_FEAT_C_CONNECTION: + writel((temp & ~PORT_RWC_BITS) | PORT_CSC, status_reg); + break; + case USB_PORT_FEAT_C_OVER_CURRENT: + writel((temp & ~PORT_RWC_BITS) | PORT_OCC, status_reg); + break; + case USB_PORT_FEAT_C_RESET: + /* GetPortStatus clears reset */ + break; + default: + goto error; + } + readl(&oxu->regs->command); /* unblock posted write */ + break; + case GetHubDescriptor: + ehci_hub_descriptor(oxu, (struct usb_hub_descriptor *) + buf); + break; + case GetHubStatus: + /* no hub-wide feature/status flags */ + memset(buf, 0, 4); + break; + case GetPortStatus: + if (!wIndex || wIndex > ports) + goto error; + wIndex--; + status = 0; + temp = readl(status_reg); + + /* wPortChange bits */ + if (temp & PORT_CSC) + status |= 1 << USB_PORT_FEAT_C_CONNECTION; + if (temp & PORT_PEC) + status |= 1 << USB_PORT_FEAT_C_ENABLE; + if ((temp & PORT_OCC) && !ignore_oc) + status |= 1 << USB_PORT_FEAT_C_OVER_CURRENT; + + /* whoever resumes must GetPortStatus to complete it!! */ + if (temp & PORT_RESUME) { + + /* Remote Wakeup received? */ + if (!oxu->reset_done[wIndex]) { + /* resume signaling for 20 msec */ + oxu->reset_done[wIndex] = jiffies + + msecs_to_jiffies(20); + /* check the port again */ + mod_timer(&oxu_to_hcd(oxu)->rh_timer, + oxu->reset_done[wIndex]); + } + + /* resume completed? */ + else if (time_after_eq(jiffies, + oxu->reset_done[wIndex])) { + status |= 1 << USB_PORT_FEAT_C_SUSPEND; + oxu->reset_done[wIndex] = 0; + + /* stop resume signaling */ + temp = readl(status_reg); + writel(temp & ~(PORT_RWC_BITS | PORT_RESUME), + status_reg); + retval = handshake(oxu, status_reg, + PORT_RESUME, 0, 2000 /* 2msec */); + if (retval != 0) { + oxu_err(oxu, + "port %d resume error %d\n", + wIndex + 1, retval); + goto error; + } + temp &= ~(PORT_SUSPEND|PORT_RESUME|(3<<10)); + } + } + + /* whoever resets must GetPortStatus to complete it!! */ + if ((temp & PORT_RESET) + && time_after_eq(jiffies, + oxu->reset_done[wIndex])) { + status |= 1 << USB_PORT_FEAT_C_RESET; + oxu->reset_done[wIndex] = 0; + + /* force reset to complete */ + writel(temp & ~(PORT_RWC_BITS | PORT_RESET), + status_reg); + /* REVISIT: some hardware needs 550+ usec to clear + * this bit; seems too long to spin routinely... + */ + retval = handshake(oxu, status_reg, + PORT_RESET, 0, 750); + if (retval != 0) { + oxu_err(oxu, "port %d reset error %d\n", + wIndex + 1, retval); + goto error; + } + + /* see what we found out */ + temp = check_reset_complete(oxu, wIndex, status_reg, + readl(status_reg)); + } + + /* transfer dedicated ports to the companion hc */ + if ((temp & PORT_CONNECT) && + test_bit(wIndex, &oxu->companion_ports)) { + temp &= ~PORT_RWC_BITS; + temp |= PORT_OWNER; + writel(temp, status_reg); + oxu_dbg(oxu, "port %d --> companion\n", wIndex + 1); + temp = readl(status_reg); + } + + /* + * Even if OWNER is set, there's no harm letting khubd + * see the wPortStatus values (they should all be 0 except + * for PORT_POWER anyway). + */ + + if (temp & PORT_CONNECT) { + status |= 1 << USB_PORT_FEAT_CONNECTION; + /* status may be from integrated TT */ + status |= oxu_port_speed(oxu, temp); + } + if (temp & PORT_PE) + status |= 1 << USB_PORT_FEAT_ENABLE; + if (temp & (PORT_SUSPEND|PORT_RESUME)) + status |= 1 << USB_PORT_FEAT_SUSPEND; + if (temp & PORT_OC) + status |= 1 << USB_PORT_FEAT_OVER_CURRENT; + if (temp & PORT_RESET) + status |= 1 << USB_PORT_FEAT_RESET; + if (temp & PORT_POWER) + status |= 1 << USB_PORT_FEAT_POWER; + +#ifndef OXU_VERBOSE_DEBUG + if (status & ~0xffff) /* only if wPortChange is interesting */ +#endif + dbg_port(oxu, "GetStatus", wIndex + 1, temp); + put_unaligned(cpu_to_le32(status), (__le32 *) buf); + break; + case SetHubFeature: + switch (wValue) { + case C_HUB_LOCAL_POWER: + case C_HUB_OVER_CURRENT: + /* no hub-wide feature/status flags */ + break; + default: + goto error; + } + break; + case SetPortFeature: + selector = wIndex >> 8; + wIndex &= 0xff; + if (!wIndex || wIndex > ports) + goto error; + wIndex--; + temp = readl(status_reg); + if (temp & PORT_OWNER) + break; + + temp &= ~PORT_RWC_BITS; + switch (wValue) { + case USB_PORT_FEAT_SUSPEND: + if ((temp & PORT_PE) == 0 + || (temp & PORT_RESET) != 0) + goto error; + if (device_may_wakeup(&hcd->self.root_hub->dev)) + temp |= PORT_WAKE_BITS; + writel(temp | PORT_SUSPEND, status_reg); + break; + case USB_PORT_FEAT_POWER: + if (HCS_PPC(oxu->hcs_params)) + writel(temp | PORT_POWER, status_reg); + break; + case USB_PORT_FEAT_RESET: + if (temp & PORT_RESUME) + goto error; + /* line status bits may report this as low speed, + * which can be fine if this root hub has a + * transaction translator built in. + */ + oxu_vdbg(oxu, "port %d reset\n", wIndex + 1); + temp |= PORT_RESET; + temp &= ~PORT_PE; + + /* + * caller must wait, then call GetPortStatus + * usb 2.0 spec says 50 ms resets on root + */ + oxu->reset_done[wIndex] = jiffies + + msecs_to_jiffies(50); + writel(temp, status_reg); + break; + + /* For downstream facing ports (these): one hub port is put + * into test mode according to USB2 11.24.2.13, then the hub + * must be reset (which for root hub now means rmmod+modprobe, + * or else system reboot). See EHCI 2.3.9 and 4.14 for info + * about the EHCI-specific stuff. + */ + case USB_PORT_FEAT_TEST: + if (!selector || selector > 5) + goto error; + ehci_quiesce(oxu); + ehci_halt(oxu); + temp |= selector << 16; + writel(temp, status_reg); + break; + + default: + goto error; + } + readl(&oxu->regs->command); /* unblock posted writes */ + break; + + default: +error: + /* "stall" on error */ + retval = -EPIPE; + } + spin_unlock_irqrestore(&oxu->lock, flags); + return retval; +} + +#ifdef CONFIG_PM + +static int oxu_bus_suspend(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + int port; + int mask; + + oxu_dbg(oxu, "suspend root hub\n"); + + if (time_before(jiffies, oxu->next_statechange)) + msleep(5); + + port = HCS_N_PORTS(oxu->hcs_params); + spin_lock_irq(&oxu->lock); + + /* stop schedules, clean any completed work */ + if (HC_IS_RUNNING(hcd->state)) { + ehci_quiesce(oxu); + hcd->state = HC_STATE_QUIESCING; + } + oxu->command = readl(&oxu->regs->command); + if (oxu->reclaim) + oxu->reclaim_ready = 1; + ehci_work(oxu); + + /* Unlike other USB host controller types, EHCI doesn't have + * any notion of "global" or bus-wide suspend. The driver has + * to manually suspend all the active unsuspended ports, and + * then manually resume them in the bus_resume() routine. + */ + oxu->bus_suspended = 0; + while (port--) { + u32 __iomem *reg = &oxu->regs->port_status[port]; + u32 t1 = readl(reg) & ~PORT_RWC_BITS; + u32 t2 = t1; + + /* keep track of which ports we suspend */ + if ((t1 & PORT_PE) && !(t1 & PORT_OWNER) && + !(t1 & PORT_SUSPEND)) { + t2 |= PORT_SUSPEND; + set_bit(port, &oxu->bus_suspended); + } + + /* enable remote wakeup on all ports */ + if (device_may_wakeup(&hcd->self.root_hub->dev)) + t2 |= PORT_WKOC_E|PORT_WKDISC_E|PORT_WKCONN_E; + else + t2 &= ~(PORT_WKOC_E|PORT_WKDISC_E|PORT_WKCONN_E); + + if (t1 != t2) { + oxu_vdbg(oxu, "port %d, %08x -> %08x\n", + port + 1, t1, t2); + writel(t2, reg); + } + } + + /* turn off now-idle HC */ + del_timer_sync(&oxu->watchdog); + ehci_halt(oxu); + hcd->state = HC_STATE_SUSPENDED; + + /* allow remote wakeup */ + mask = INTR_MASK; + if (!device_may_wakeup(&hcd->self.root_hub->dev)) + mask &= ~STS_PCD; + writel(mask, &oxu->regs->intr_enable); + readl(&oxu->regs->intr_enable); + + oxu->next_statechange = jiffies + msecs_to_jiffies(10); + spin_unlock_irq(&oxu->lock); + return 0; +} + +/* Caller has locked the root hub, and should reset/reinit on error */ +static int oxu_bus_resume(struct usb_hcd *hcd) +{ + struct oxu_hcd *oxu = hcd_to_oxu(hcd); + u32 temp; + int i; + + if (time_before(jiffies, oxu->next_statechange)) + msleep(5); + spin_lock_irq(&oxu->lock); + + /* Ideally and we've got a real resume here, and no port's power + * was lost. (For PCI, that means Vaux was maintained.) But we + * could instead be restoring a swsusp snapshot -- so that BIOS was + * the last user of the controller, not reset/pm hardware keeping + * state we gave to it. + */ + temp = readl(&oxu->regs->intr_enable); + oxu_dbg(oxu, "resume root hub%s\n", temp ? "" : " after power loss"); + + /* at least some APM implementations will try to deliver + * IRQs right away, so delay them until we're ready. + */ + writel(0, &oxu->regs->intr_enable); + + /* re-init operational registers */ + writel(0, &oxu->regs->segment); + writel(oxu->periodic_dma, &oxu->regs->frame_list); + writel((u32) oxu->async->qh_dma, &oxu->regs->async_next); + + /* restore CMD_RUN, framelist size, and irq threshold */ + writel(oxu->command, &oxu->regs->command); + + /* Some controller/firmware combinations need a delay during which + * they set up the port statuses. See Bugzilla #8190. */ + mdelay(8); + + /* manually resume the ports we suspended during bus_suspend() */ + i = HCS_N_PORTS(oxu->hcs_params); + while (i--) { + temp = readl(&oxu->regs->port_status[i]); + temp &= ~(PORT_RWC_BITS + | PORT_WKOC_E | PORT_WKDISC_E | PORT_WKCONN_E); + if (test_bit(i, &oxu->bus_suspended) && (temp & PORT_SUSPEND)) { + oxu->reset_done[i] = jiffies + msecs_to_jiffies(20); + temp |= PORT_RESUME; + } + writel(temp, &oxu->regs->port_status[i]); + } + i = HCS_N_PORTS(oxu->hcs_params); + mdelay(20); + while (i--) { + temp = readl(&oxu->regs->port_status[i]); + if (test_bit(i, &oxu->bus_suspended) && (temp & PORT_SUSPEND)) { + temp &= ~(PORT_RWC_BITS | PORT_RESUME); + writel(temp, &oxu->regs->port_status[i]); + oxu_vdbg(oxu, "resumed port %d\n", i + 1); + } + } + (void) readl(&oxu->regs->command); + + /* maybe re-activate the schedule(s) */ + temp = 0; + if (oxu->async->qh_next.qh) + temp |= CMD_ASE; + if (oxu->periodic_sched) + temp |= CMD_PSE; + if (temp) { + oxu->command |= temp; + writel(oxu->command, &oxu->regs->command); + } + + oxu->next_statechange = jiffies + msecs_to_jiffies(5); + hcd->state = HC_STATE_RUNNING; + + /* Now we can safely re-enable irqs */ + writel(INTR_MASK, &oxu->regs->intr_enable); + + spin_unlock_irq(&oxu->lock); + return 0; +} + +#else + +static int oxu_bus_suspend(struct usb_hcd *hcd) +{ + return 0; +} + +static int oxu_bus_resume(struct usb_hcd *hcd) +{ + return 0; +} + +#endif /* CONFIG_PM */ + +static const struct hc_driver oxu_hc_driver = { + .description = "oxu210hp_hcd", + .product_desc = "oxu210hp HCD", + .hcd_priv_size = sizeof(struct oxu_hcd), + + /* + * Generic hardware linkage + */ + .irq = oxu_irq, + .flags = HCD_MEMORY | HCD_USB2, + + /* + * Basic lifecycle operations + */ + .reset = oxu_reset, + .start = oxu_run, + .stop = oxu_stop, + .shutdown = oxu_shutdown, + + /* + * Managing i/o requests and associated device resources + */ + .urb_enqueue = oxu_urb_enqueue, + .urb_dequeue = oxu_urb_dequeue, + .endpoint_disable = oxu_endpoint_disable, + + /* + * Scheduling support + */ + .get_frame_number = oxu_get_frame, + + /* + * Root hub support + */ + .hub_status_data = oxu_hub_status_data, + .hub_control = oxu_hub_control, + .bus_suspend = oxu_bus_suspend, + .bus_resume = oxu_bus_resume, +}; + +/* + * Module stuff + */ + +static void oxu_configuration(struct platform_device *pdev, void *base) +{ + u32 tmp; + + /* Initialize top level registers. + * First write ever + */ + oxu_writel(base, OXU_HOSTIFCONFIG, 0x0000037D); + oxu_writel(base, OXU_SOFTRESET, OXU_SRESET); + oxu_writel(base, OXU_HOSTIFCONFIG, 0x0000037D); + + tmp = oxu_readl(base, OXU_PIOBURSTREADCTRL); + oxu_writel(base, OXU_PIOBURSTREADCTRL, tmp | 0x0040); + + oxu_writel(base, OXU_ASO, OXU_SPHPOEN | OXU_OVRCCURPUPDEN | + OXU_COMPARATOR | OXU_ASO_OP); + + tmp = oxu_readl(base, OXU_CLKCTRL_SET); + oxu_writel(base, OXU_CLKCTRL_SET, tmp | OXU_SYSCLKEN | OXU_USBOTGCLKEN); + + /* Clear all top interrupt enable */ + oxu_writel(base, OXU_CHIPIRQEN_CLR, 0xff); + + /* Clear all top interrupt status */ + oxu_writel(base, OXU_CHIPIRQSTATUS, 0xff); + + /* Enable all needed top interrupt except OTG SPH core */ + oxu_writel(base, OXU_CHIPIRQEN_SET, OXU_USBSPHLPWUI | OXU_USBOTGLPWUI); +} + +static int oxu_verify_id(struct platform_device *pdev, void *base) +{ + u32 id; + char *bo[] = { + "reserved", + "128-pin LQFP", + "84-pin TFBGA", + "reserved", + }; + + /* Read controller signature register to find a match */ + id = oxu_readl(base, OXU_DEVICEID); + dev_info(&pdev->dev, "device ID %x\n", id); + if ((id & OXU_REV_MASK) != (OXU_REV_2100 << OXU_REV_SHIFT)) + return -1; + + dev_info(&pdev->dev, "found device %x %s (%04x:%04x)\n", + id >> OXU_REV_SHIFT, + bo[(id & OXU_BO_MASK) >> OXU_BO_SHIFT], + (id & OXU_MAJ_REV_MASK) >> OXU_MAJ_REV_SHIFT, + (id & OXU_MIN_REV_MASK) >> OXU_MIN_REV_SHIFT); + + return 0; +} + +static const struct hc_driver oxu_hc_driver; +static struct usb_hcd *oxu_create(struct platform_device *pdev, + unsigned long memstart, unsigned long memlen, + void *base, int irq, int otg) +{ + struct device *dev = &pdev->dev; + + struct usb_hcd *hcd; + struct oxu_hcd *oxu; + int ret; + + /* Set endian mode and host mode */ + oxu_writel(base + (otg ? OXU_OTG_CORE_OFFSET : OXU_SPH_CORE_OFFSET), + OXU_USBMODE, + OXU_CM_HOST_ONLY | OXU_ES_LITTLE | OXU_VBPS); + + hcd = usb_create_hcd(&oxu_hc_driver, dev, + otg ? "oxu210hp_otg" : "oxu210hp_sph"); + if (!hcd) + return ERR_PTR(-ENOMEM); + + hcd->rsrc_start = memstart; + hcd->rsrc_len = memlen; + hcd->regs = base; + hcd->irq = irq; + hcd->state = HC_STATE_HALT; + + oxu = hcd_to_oxu(hcd); + oxu->is_otg = otg; + + ret = usb_add_hcd(hcd, irq, IRQF_SHARED); + if (ret < 0) + return ERR_PTR(ret); + + return hcd; +} + +static int oxu_init(struct platform_device *pdev, + unsigned long memstart, unsigned long memlen, + void *base, int irq) +{ + struct oxu_info *info = platform_get_drvdata(pdev); + struct usb_hcd *hcd; + int ret; + + /* First time configuration at start up */ + oxu_configuration(pdev, base); + + ret = oxu_verify_id(pdev, base); + if (ret) { + dev_err(&pdev->dev, "no devices found!\n"); + return -ENODEV; + } + + /* Create the OTG controller */ + hcd = oxu_create(pdev, memstart, memlen, base, irq, 1); + if (IS_ERR(hcd)) { + dev_err(&pdev->dev, "cannot create OTG controller!\n"); + ret = PTR_ERR(hcd); + goto error_create_otg; + } + info->hcd[0] = hcd; + + /* Create the SPH host controller */ + hcd = oxu_create(pdev, memstart, memlen, base, irq, 0); + if (IS_ERR(hcd)) { + dev_err(&pdev->dev, "cannot create SPH controller!\n"); + ret = PTR_ERR(hcd); + goto error_create_sph; + } + info->hcd[1] = hcd; + + oxu_writel(base, OXU_CHIPIRQEN_SET, + oxu_readl(base, OXU_CHIPIRQEN_SET) | 3); + + return 0; + +error_create_sph: + usb_remove_hcd(info->hcd[0]); + usb_put_hcd(info->hcd[0]); + +error_create_otg: + return ret; +} + +static int oxu_drv_probe(struct platform_device *pdev) +{ + struct resource *res; + void *base; + unsigned long memstart, memlen; + int irq, ret; + struct oxu_info *info; + + if (usb_disabled()) + return -ENODEV; + + /* + * Get the platform resources + */ + res = platform_get_resource(pdev, IORESOURCE_IRQ, 0); + if (!res) { + dev_err(&pdev->dev, + "no IRQ! Check %s setup!\n", pdev->dev.bus_id); + return -ENODEV; + } + irq = res->start; + dev_dbg(&pdev->dev, "IRQ resource %d\n", irq); + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!res) { + dev_err(&pdev->dev, "no registers address! Check %s setup!\n", + pdev->dev.bus_id); + return -ENODEV; + } + memstart = res->start; + memlen = res->end - res->start + 1; + dev_dbg(&pdev->dev, "MEM resource %lx-%lx\n", memstart, memlen); + if (!request_mem_region(memstart, memlen, + oxu_hc_driver.description)) { + dev_dbg(&pdev->dev, "memory area already in use\n"); + return -EBUSY; + } + + ret = set_irq_type(irq, IRQF_TRIGGER_FALLING); + if (ret) { + dev_err(&pdev->dev, "error setting irq type\n"); + ret = -EFAULT; + goto error_set_irq_type; + } + + base = ioremap(memstart, memlen); + if (!base) { + dev_dbg(&pdev->dev, "error mapping memory\n"); + ret = -EFAULT; + goto error_ioremap; + } + + /* Allocate a driver data struct to hold useful info for both + * SPH & OTG devices + */ + info = kzalloc(sizeof(struct oxu_info), GFP_KERNEL); + if (!info) { + dev_dbg(&pdev->dev, "error allocating memory\n"); + ret = -EFAULT; + goto error_alloc; + } + platform_set_drvdata(pdev, info); + + ret = oxu_init(pdev, memstart, memlen, base, irq); + if (ret < 0) { + dev_dbg(&pdev->dev, "cannot init USB devices\n"); + goto error_init; + } + + dev_info(&pdev->dev, "devices enabled and running\n"); + platform_set_drvdata(pdev, info); + + return 0; + +error_init: + kfree(info); + platform_set_drvdata(pdev, NULL); + +error_alloc: + iounmap(base); + +error_set_irq_type: +error_ioremap: + release_mem_region(memstart, memlen); + + dev_err(&pdev->dev, "init %s fail, %d\n", pdev->dev.bus_id, ret); + return ret; +} + +static void oxu_remove(struct platform_device *pdev, struct usb_hcd *hcd) +{ + usb_remove_hcd(hcd); + usb_put_hcd(hcd); +} + +static int oxu_drv_remove(struct platform_device *pdev) +{ + struct oxu_info *info = platform_get_drvdata(pdev); + unsigned long memstart = info->hcd[0]->rsrc_start, + memlen = info->hcd[0]->rsrc_len; + void *base = info->hcd[0]->regs; + + oxu_remove(pdev, info->hcd[0]); + oxu_remove(pdev, info->hcd[1]); + + iounmap(base); + release_mem_region(memstart, memlen); + + kfree(info); + platform_set_drvdata(pdev, NULL); + + return 0; +} + +static void oxu_drv_shutdown(struct platform_device *pdev) +{ + oxu_drv_remove(pdev); +} + +#if 0 +/* FIXME: TODO */ +static int oxu_drv_suspend(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct usb_hcd *hcd = dev_get_drvdata(dev); + + return 0; +} + +static int oxu_drv_resume(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct usb_hcd *hcd = dev_get_drvdata(dev); + + return 0; +} +#else +#define oxu_drv_suspend NULL +#define oxu_drv_resume NULL +#endif + +static struct platform_driver oxu_driver = { + .probe = oxu_drv_probe, + .remove = oxu_drv_remove, + .shutdown = oxu_drv_shutdown, + .suspend = oxu_drv_suspend, + .resume = oxu_drv_resume, + .driver = { + .name = "oxu210hp-hcd", + .bus = &platform_bus_type + } +}; + +static int __init oxu_module_init(void) +{ + int retval = 0; + + retval = platform_driver_register(&oxu_driver); + if (retval < 0) + return retval; + + return retval; +} + +static void __exit oxu_module_cleanup(void) +{ + platform_driver_unregister(&oxu_driver); +} + +module_init(oxu_module_init); +module_exit(oxu_module_cleanup); + +MODULE_DESCRIPTION("Oxford OXU210HP HCD driver - ver. " DRIVER_VERSION); +MODULE_AUTHOR("Rodolfo Giometti "); +MODULE_LICENSE("GPL"); diff --git a/drivers/usb/host/oxu210hp.h b/drivers/usb/host/oxu210hp.h new file mode 100644 index 00000000000..8910e271cc7 --- /dev/null +++ b/drivers/usb/host/oxu210hp.h @@ -0,0 +1,447 @@ +/* + * Host interface registers + */ + +#define OXU_DEVICEID 0x00 + #define OXU_REV_MASK 0xffff0000 + #define OXU_REV_SHIFT 16 + #define OXU_REV_2100 0x2100 + #define OXU_BO_SHIFT 8 + #define OXU_BO_MASK (0x3 << OXU_BO_SHIFT) + #define OXU_MAJ_REV_SHIFT 4 + #define OXU_MAJ_REV_MASK (0xf << OXU_MAJ_REV_SHIFT) + #define OXU_MIN_REV_SHIFT 0 + #define OXU_MIN_REV_MASK (0xf << OXU_MIN_REV_SHIFT) +#define OXU_HOSTIFCONFIG 0x04 +#define OXU_SOFTRESET 0x08 + #define OXU_SRESET (1 << 0) + +#define OXU_PIOBURSTREADCTRL 0x0C + +#define OXU_CHIPIRQSTATUS 0x10 +#define OXU_CHIPIRQEN_SET 0x14 +#define OXU_CHIPIRQEN_CLR 0x18 + #define OXU_USBSPHLPWUI 0x00000080 + #define OXU_USBOTGLPWUI 0x00000040 + #define OXU_USBSPHI 0x00000002 + #define OXU_USBOTGI 0x00000001 + +#define OXU_CLKCTRL_SET 0x1C + #define OXU_SYSCLKEN 0x00000008 + #define OXU_USBSPHCLKEN 0x00000002 + #define OXU_USBOTGCLKEN 0x00000001 + +#define OXU_ASO 0x68 + #define OXU_SPHPOEN 0x00000100 + #define OXU_OVRCCURPUPDEN 0x00000800 + #define OXU_ASO_OP (1 << 10) + #define OXU_COMPARATOR 0x000004000 + +#define OXU_USBMODE 0x1A8 + #define OXU_VBPS 0x00000020 + #define OXU_ES_LITTLE 0x00000000 + #define OXU_CM_HOST_ONLY 0x00000003 + +/* + * Proper EHCI structs & defines + */ + +/* Magic numbers that can affect system performance */ +#define EHCI_TUNE_CERR 3 /* 0-3 qtd retries; 0 == don't stop */ +#define EHCI_TUNE_RL_HS 4 /* nak throttle; see 4.9 */ +#define EHCI_TUNE_RL_TT 0 +#define EHCI_TUNE_MULT_HS 1 /* 1-3 transactions/uframe; 4.10.3 */ +#define EHCI_TUNE_MULT_TT 1 +#define EHCI_TUNE_FLS 2 /* (small) 256 frame schedule */ + +struct oxu_hcd; + +/* EHCI register interface, corresponds to EHCI Revision 0.95 specification */ + +/* Section 2.2 Host Controller Capability Registers */ +struct ehci_caps { + /* these fields are specified as 8 and 16 bit registers, + * but some hosts can't perform 8 or 16 bit PCI accesses. + */ + u32 hc_capbase; +#define HC_LENGTH(p) (((p)>>00)&0x00ff) /* bits 7:0 */ +#define HC_VERSION(p) (((p)>>16)&0xffff) /* bits 31:16 */ + u32 hcs_params; /* HCSPARAMS - offset 0x4 */ +#define HCS_DEBUG_PORT(p) (((p)>>20)&0xf) /* bits 23:20, debug port? */ +#define HCS_INDICATOR(p) ((p)&(1 << 16)) /* true: has port indicators */ +#define HCS_N_CC(p) (((p)>>12)&0xf) /* bits 15:12, #companion HCs */ +#define HCS_N_PCC(p) (((p)>>8)&0xf) /* bits 11:8, ports per CC */ +#define HCS_PORTROUTED(p) ((p)&(1 << 7)) /* true: port routing */ +#define HCS_PPC(p) ((p)&(1 << 4)) /* true: port power control */ +#define HCS_N_PORTS(p) (((p)>>0)&0xf) /* bits 3:0, ports on HC */ + + u32 hcc_params; /* HCCPARAMS - offset 0x8 */ +#define HCC_EXT_CAPS(p) (((p)>>8)&0xff) /* for pci extended caps */ +#define HCC_ISOC_CACHE(p) ((p)&(1 << 7)) /* true: can cache isoc frame */ +#define HCC_ISOC_THRES(p) (((p)>>4)&0x7) /* bits 6:4, uframes cached */ +#define HCC_CANPARK(p) ((p)&(1 << 2)) /* true: can park on async qh */ +#define HCC_PGM_FRAMELISTLEN(p) ((p)&(1 << 1)) /* true: periodic_size changes*/ +#define HCC_64BIT_ADDR(p) ((p)&(1)) /* true: can use 64-bit addr */ + u8 portroute[8]; /* nibbles for routing - offset 0xC */ +} __attribute__ ((packed)); + + +/* Section 2.3 Host Controller Operational Registers */ +struct ehci_regs { + /* USBCMD: offset 0x00 */ + u32 command; +/* 23:16 is r/w intr rate, in microframes; default "8" == 1/msec */ +#define CMD_PARK (1<<11) /* enable "park" on async qh */ +#define CMD_PARK_CNT(c) (((c)>>8)&3) /* how many transfers to park for */ +#define CMD_LRESET (1<<7) /* partial reset (no ports, etc) */ +#define CMD_IAAD (1<<6) /* "doorbell" interrupt async advance */ +#define CMD_ASE (1<<5) /* async schedule enable */ +#define CMD_PSE (1<<4) /* periodic schedule enable */ +/* 3:2 is periodic frame list size */ +#define CMD_RESET (1<<1) /* reset HC not bus */ +#define CMD_RUN (1<<0) /* start/stop HC */ + + /* USBSTS: offset 0x04 */ + u32 status; +#define STS_ASS (1<<15) /* Async Schedule Status */ +#define STS_PSS (1<<14) /* Periodic Schedule Status */ +#define STS_RECL (1<<13) /* Reclamation */ +#define STS_HALT (1<<12) /* Not running (any reason) */ +/* some bits reserved */ + /* these STS_* flags are also intr_enable bits (USBINTR) */ +#define STS_IAA (1<<5) /* Interrupted on async advance */ +#define STS_FATAL (1<<4) /* such as some PCI access errors */ +#define STS_FLR (1<<3) /* frame list rolled over */ +#define STS_PCD (1<<2) /* port change detect */ +#define STS_ERR (1<<1) /* "error" completion (overflow, ...) */ +#define STS_INT (1<<0) /* "normal" completion (short, ...) */ + +#define INTR_MASK (STS_IAA | STS_FATAL | STS_PCD | STS_ERR | STS_INT) + + /* USBINTR: offset 0x08 */ + u32 intr_enable; + + /* FRINDEX: offset 0x0C */ + u32 frame_index; /* current microframe number */ + /* CTRLDSSEGMENT: offset 0x10 */ + u32 segment; /* address bits 63:32 if needed */ + /* PERIODICLISTBASE: offset 0x14 */ + u32 frame_list; /* points to periodic list */ + /* ASYNCLISTADDR: offset 0x18 */ + u32 async_next; /* address of next async queue head */ + + u32 reserved[9]; + + /* CONFIGFLAG: offset 0x40 */ + u32 configured_flag; +#define FLAG_CF (1<<0) /* true: we'll support "high speed" */ + + /* PORTSC: offset 0x44 */ + u32 port_status[0]; /* up to N_PORTS */ +/* 31:23 reserved */ +#define PORT_WKOC_E (1<<22) /* wake on overcurrent (enable) */ +#define PORT_WKDISC_E (1<<21) /* wake on disconnect (enable) */ +#define PORT_WKCONN_E (1<<20) /* wake on connect (enable) */ +/* 19:16 for port testing */ +#define PORT_LED_OFF (0<<14) +#define PORT_LED_AMBER (1<<14) +#define PORT_LED_GREEN (2<<14) +#define PORT_LED_MASK (3<<14) +#define PORT_OWNER (1<<13) /* true: companion hc owns this port */ +#define PORT_POWER (1<<12) /* true: has power (see PPC) */ +#define PORT_USB11(x) (((x)&(3<<10)) == (1<<10)) /* USB 1.1 device */ +/* 11:10 for detecting lowspeed devices (reset vs release ownership) */ +/* 9 reserved */ +#define PORT_RESET (1<<8) /* reset port */ +#define PORT_SUSPEND (1<<7) /* suspend port */ +#define PORT_RESUME (1<<6) /* resume it */ +#define PORT_OCC (1<<5) /* over current change */ +#define PORT_OC (1<<4) /* over current active */ +#define PORT_PEC (1<<3) /* port enable change */ +#define PORT_PE (1<<2) /* port enable */ +#define PORT_CSC (1<<1) /* connect status change */ +#define PORT_CONNECT (1<<0) /* device connected */ +#define PORT_RWC_BITS (PORT_CSC | PORT_PEC | PORT_OCC) +} __attribute__ ((packed)); + +/* Appendix C, Debug port ... intended for use with special "debug devices" + * that can help if there's no serial console. (nonstandard enumeration.) + */ +struct ehci_dbg_port { + u32 control; +#define DBGP_OWNER (1<<30) +#define DBGP_ENABLED (1<<28) +#define DBGP_DONE (1<<16) +#define DBGP_INUSE (1<<10) +#define DBGP_ERRCODE(x) (((x)>>7)&0x07) +# define DBGP_ERR_BAD 1 +# define DBGP_ERR_SIGNAL 2 +#define DBGP_ERROR (1<<6) +#define DBGP_GO (1<<5) +#define DBGP_OUT (1<<4) +#define DBGP_LEN(x) (((x)>>0)&0x0f) + u32 pids; +#define DBGP_PID_GET(x) (((x)>>16)&0xff) +#define DBGP_PID_SET(data, tok) (((data)<<8)|(tok)) + u32 data03; + u32 data47; + u32 address; +#define DBGP_EPADDR(dev, ep) (((dev)<<8)|(ep)) +} __attribute__ ((packed)); + + +#define QTD_NEXT(dma) cpu_to_le32((u32)dma) + +/* + * EHCI Specification 0.95 Section 3.5 + * QTD: describe data transfer components (buffer, direction, ...) + * See Fig 3-6 "Queue Element Transfer Descriptor Block Diagram". + * + * These are associated only with "QH" (Queue Head) structures, + * used with control, bulk, and interrupt transfers. + */ +struct ehci_qtd { + /* first part defined by EHCI spec */ + __le32 hw_next; /* see EHCI 3.5.1 */ + __le32 hw_alt_next; /* see EHCI 3.5.2 */ + __le32 hw_token; /* see EHCI 3.5.3 */ +#define QTD_TOGGLE (1 << 31) /* data toggle */ +#define QTD_LENGTH(tok) (((tok)>>16) & 0x7fff) +#define QTD_IOC (1 << 15) /* interrupt on complete */ +#define QTD_CERR(tok) (((tok)>>10) & 0x3) +#define QTD_PID(tok) (((tok)>>8) & 0x3) +#define QTD_STS_ACTIVE (1 << 7) /* HC may execute this */ +#define QTD_STS_HALT (1 << 6) /* halted on error */ +#define QTD_STS_DBE (1 << 5) /* data buffer error (in HC) */ +#define QTD_STS_BABBLE (1 << 4) /* device was babbling (qtd halted) */ +#define QTD_STS_XACT (1 << 3) /* device gave illegal response */ +#define QTD_STS_MMF (1 << 2) /* incomplete split transaction */ +#define QTD_STS_STS (1 << 1) /* split transaction state */ +#define QTD_STS_PING (1 << 0) /* issue PING? */ + __le32 hw_buf[5]; /* see EHCI 3.5.4 */ + __le32 hw_buf_hi[5]; /* Appendix B */ + + /* the rest is HCD-private */ + dma_addr_t qtd_dma; /* qtd address */ + struct list_head qtd_list; /* sw qtd list */ + struct urb *urb; /* qtd's urb */ + size_t length; /* length of buffer */ + + u32 qtd_buffer_len; + void *buffer; + dma_addr_t buffer_dma; + void *transfer_buffer; + void *transfer_dma; +} __attribute__ ((aligned(32))); + +/* mask NakCnt+T in qh->hw_alt_next */ +#define QTD_MASK __constant_cpu_to_le32 (~0x1f) + +#define IS_SHORT_READ(token) (QTD_LENGTH(token) != 0 && QTD_PID(token) == 1) + +/* Type tag from {qh, itd, sitd, fstn}->hw_next */ +#define Q_NEXT_TYPE(dma) ((dma) & __constant_cpu_to_le32 (3 << 1)) + +/* values for that type tag */ +#define Q_TYPE_QH __constant_cpu_to_le32 (1 << 1) + +/* next async queue entry, or pointer to interrupt/periodic QH */ +#define QH_NEXT(dma) (cpu_to_le32(((u32)dma)&~0x01f)|Q_TYPE_QH) + +/* for periodic/async schedules and qtd lists, mark end of list */ +#define EHCI_LIST_END __constant_cpu_to_le32(1) /* "null pointer" to hw */ + +/* + * Entries in periodic shadow table are pointers to one of four kinds + * of data structure. That's dictated by the hardware; a type tag is + * encoded in the low bits of the hardware's periodic schedule. Use + * Q_NEXT_TYPE to get the tag. + * + * For entries in the async schedule, the type tag always says "qh". + */ +union ehci_shadow { + struct ehci_qh *qh; /* Q_TYPE_QH */ + __le32 *hw_next; /* (all types) */ + void *ptr; +}; + +/* + * EHCI Specification 0.95 Section 3.6 + * QH: describes control/bulk/interrupt endpoints + * See Fig 3-7 "Queue Head Structure Layout". + * + * These appear in both the async and (for interrupt) periodic schedules. + */ + +struct ehci_qh { + /* first part defined by EHCI spec */ + __le32 hw_next; /* see EHCI 3.6.1 */ + __le32 hw_info1; /* see EHCI 3.6.2 */ +#define QH_HEAD 0x00008000 + __le32 hw_info2; /* see EHCI 3.6.2 */ +#define QH_SMASK 0x000000ff +#define QH_CMASK 0x0000ff00 +#define QH_HUBADDR 0x007f0000 +#define QH_HUBPORT 0x3f800000 +#define QH_MULT 0xc0000000 + __le32 hw_current; /* qtd list - see EHCI 3.6.4 */ + + /* qtd overlay (hardware parts of a struct ehci_qtd) */ + __le32 hw_qtd_next; + __le32 hw_alt_next; + __le32 hw_token; + __le32 hw_buf[5]; + __le32 hw_buf_hi[5]; + + /* the rest is HCD-private */ + dma_addr_t qh_dma; /* address of qh */ + union ehci_shadow qh_next; /* ptr to qh; or periodic */ + struct list_head qtd_list; /* sw qtd list */ + struct ehci_qtd *dummy; + struct ehci_qh *reclaim; /* next to reclaim */ + + struct oxu_hcd *oxu; + struct kref kref; + unsigned stamp; + + u8 qh_state; +#define QH_STATE_LINKED 1 /* HC sees this */ +#define QH_STATE_UNLINK 2 /* HC may still see this */ +#define QH_STATE_IDLE 3 /* HC doesn't see this */ +#define QH_STATE_UNLINK_WAIT 4 /* LINKED and on reclaim q */ +#define QH_STATE_COMPLETING 5 /* don't touch token.HALT */ + + /* periodic schedule info */ + u8 usecs; /* intr bandwidth */ + u8 gap_uf; /* uframes split/csplit gap */ + u8 c_usecs; /* ... split completion bw */ + u16 tt_usecs; /* tt downstream bandwidth */ + unsigned short period; /* polling interval */ + unsigned short start; /* where polling starts */ +#define NO_FRAME ((unsigned short)~0) /* pick new start */ + struct usb_device *dev; /* access to TT */ +} __attribute__ ((aligned(32))); + +/* + * Proper OXU210HP structs + */ + +#define OXU_OTG_CORE_OFFSET 0x00400 +#define OXU_OTG_CAP_OFFSET (OXU_OTG_CORE_OFFSET + 0x100) +#define OXU_SPH_CORE_OFFSET 0x00800 +#define OXU_SPH_CAP_OFFSET (OXU_SPH_CORE_OFFSET + 0x100) + +#define OXU_OTG_MEM 0xE000 +#define OXU_SPH_MEM 0x16000 + +/* Only how many elements & element structure are specifies here. */ +/* 2 host controllers are enabled - total size <= 28 kbytes */ +#define DEFAULT_I_TDPS 1024 +#define QHEAD_NUM 16 +#define QTD_NUM 32 +#define SITD_NUM 8 +#define MURB_NUM 8 + +#define BUFFER_NUM 8 +#define BUFFER_SIZE 512 + +struct oxu_info { + struct usb_hcd *hcd[2]; +}; + +struct oxu_buf { + u8 buffer[BUFFER_SIZE]; +} __attribute__ ((aligned(BUFFER_SIZE))); + +struct oxu_onchip_mem { + struct oxu_buf db_pool[BUFFER_NUM]; + + u32 frame_list[DEFAULT_I_TDPS]; + struct ehci_qh qh_pool[QHEAD_NUM]; + struct ehci_qtd qtd_pool[QTD_NUM]; +} __attribute__ ((aligned(4 << 10))); + +#define EHCI_MAX_ROOT_PORTS 15 /* see HCS_N_PORTS */ + +struct oxu_murb { + struct urb urb; + struct urb *main; + u8 last; +}; + +struct oxu_hcd { /* one per controller */ + unsigned int is_otg:1; + + u8 qh_used[QHEAD_NUM]; + u8 qtd_used[QTD_NUM]; + u8 db_used[BUFFER_NUM]; + u8 murb_used[MURB_NUM]; + + struct oxu_onchip_mem __iomem *mem; + spinlock_t mem_lock; + + struct timer_list urb_timer; + + struct ehci_caps __iomem *caps; + struct ehci_regs __iomem *regs; + + __u32 hcs_params; /* cached register copy */ + spinlock_t lock; + + /* async schedule support */ + struct ehci_qh *async; + struct ehci_qh *reclaim; + unsigned reclaim_ready:1; + unsigned scanning:1; + + /* periodic schedule support */ + unsigned periodic_size; + __le32 *periodic; /* hw periodic table */ + dma_addr_t periodic_dma; + unsigned i_thresh; /* uframes HC might cache */ + + union ehci_shadow *pshadow; /* mirror hw periodic table */ + int next_uframe; /* scan periodic, start here */ + unsigned periodic_sched; /* periodic activity count */ + + /* per root hub port */ + unsigned long reset_done[EHCI_MAX_ROOT_PORTS]; + /* bit vectors (one bit per port) */ + unsigned long bus_suspended; /* which ports were + * already suspended at the + * start of a bus suspend + */ + unsigned long companion_ports;/* which ports are dedicated + * to the companion controller + */ + + struct timer_list watchdog; + unsigned long actions; + unsigned stamp; + unsigned long next_statechange; + u32 command; + + /* SILICON QUIRKS */ + struct list_head urb_list; /* this is the head to urb + * queue that didn't get enough + * resources + */ + struct oxu_murb *murb_pool; /* murb per split big urb */ + unsigned urb_len; + + u8 sbrn; /* packed release number */ +}; + +#define EHCI_IAA_JIFFIES (HZ/100) /* arbitrary; ~10 msec */ +#define EHCI_IO_JIFFIES (HZ/10) /* io watchdog > irq_thresh */ +#define EHCI_ASYNC_JIFFIES (HZ/20) /* async idle timeout */ +#define EHCI_SHRINK_JIFFIES (HZ/200) /* async qh unlink delay */ + +enum ehci_timer_action { + TIMER_IO_WATCHDOG, + TIMER_IAA_WATCHDOG, + TIMER_ASYNC_SHRINK, + TIMER_ASYNC_OFF, +}; + +#include diff --git a/include/linux/oxu210hp.h b/include/linux/oxu210hp.h new file mode 100644 index 00000000000..0bf96eae538 --- /dev/null +++ b/include/linux/oxu210hp.h @@ -0,0 +1,7 @@ +/* platform data for the OXU210HP HCD */ + +struct oxu210hp_platform_data { + unsigned int bus16:1; + unsigned int use_hcd_otg:1; + unsigned int use_hcd_sph:1; +}; From aa459e6a2e8c9c5447b951f8f93e2ea76f65d104 Mon Sep 17 00:00:00 2001 From: Anton Vorontsov Date: Wed, 29 Oct 2008 14:25:50 -0700 Subject: [PATCH 010/123] USB: protect hcd.h from multiple inclusions This will let us use this header in other header files. Will be needed for the FHCI USB Host driver. Signed-off-by: Anton Vorontsov Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/hcd.h | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/usb/core/hcd.h b/drivers/usb/core/hcd.h index 9465e70f4dd..d202342b36b 100644 --- a/drivers/usb/core/hcd.h +++ b/drivers/usb/core/hcd.h @@ -16,6 +16,8 @@ * Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ +#ifndef __USB_CORE_HCD_H +#define __USB_CORE_HCD_H #ifdef __KERNEL__ @@ -490,3 +492,5 @@ extern struct rw_semaphore ehci_cf_port_reset_rwsem; extern unsigned long usb_hcds_loaded; #endif /* __KERNEL__ */ + +#endif /* __USB_CORE_HCD_H */ From da2bbdcc3838ce75c30bda8c3f9a6e55ece47ee1 Mon Sep 17 00:00:00 2001 From: Harvey Harrison Date: Wed, 29 Oct 2008 14:25:51 -0700 Subject: [PATCH 011/123] USB: avoid needless address-taking of function parameters There's no need to take the address of the function params or local variables when the direct value byteswapping routines are available. Signed-off-by: Harvey Harrison Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman --- drivers/net/usb/kaweth.c | 6 +++--- drivers/net/usb/pegasus.c | 12 ++++++------ drivers/usb/core/message.c | 6 +++--- drivers/usb/gadget/net2280.c | 2 +- 4 files changed, 13 insertions(+), 13 deletions(-) diff --git a/drivers/net/usb/kaweth.c b/drivers/net/usb/kaweth.c index 2ee034f70d1..3073ca25a0b 100644 --- a/drivers/net/usb/kaweth.c +++ b/drivers/net/usb/kaweth.c @@ -283,9 +283,9 @@ static int kaweth_control(struct kaweth_device *kaweth, dr->bRequestType= requesttype; dr->bRequest = request; - dr->wValue = cpu_to_le16p(&value); - dr->wIndex = cpu_to_le16p(&index); - dr->wLength = cpu_to_le16p(&size); + dr->wValue = cpu_to_le16(value); + dr->wIndex = cpu_to_le16(index); + dr->wLength = cpu_to_le16(size); return kaweth_internal_control_msg(kaweth->dev, pipe, diff --git a/drivers/net/usb/pegasus.c b/drivers/net/usb/pegasus.c index 166880c113d..d9241f1c080 100644 --- a/drivers/net/usb/pegasus.c +++ b/drivers/net/usb/pegasus.c @@ -150,8 +150,8 @@ static int get_registers(pegasus_t * pegasus, __u16 indx, __u16 size, pegasus->dr.bRequestType = PEGASUS_REQT_READ; pegasus->dr.bRequest = PEGASUS_REQ_GET_REGS; pegasus->dr.wValue = cpu_to_le16(0); - pegasus->dr.wIndex = cpu_to_le16p(&indx); - pegasus->dr.wLength = cpu_to_le16p(&size); + pegasus->dr.wIndex = cpu_to_le16(indx); + pegasus->dr.wLength = cpu_to_le16(size); pegasus->ctrl_urb->transfer_buffer_length = size; usb_fill_control_urb(pegasus->ctrl_urb, pegasus->usb, @@ -208,8 +208,8 @@ static int set_registers(pegasus_t * pegasus, __u16 indx, __u16 size, pegasus->dr.bRequestType = PEGASUS_REQT_WRITE; pegasus->dr.bRequest = PEGASUS_REQ_SET_REGS; pegasus->dr.wValue = cpu_to_le16(0); - pegasus->dr.wIndex = cpu_to_le16p(&indx); - pegasus->dr.wLength = cpu_to_le16p(&size); + pegasus->dr.wIndex = cpu_to_le16(indx); + pegasus->dr.wLength = cpu_to_le16(size); pegasus->ctrl_urb->transfer_buffer_length = size; usb_fill_control_urb(pegasus->ctrl_urb, pegasus->usb, @@ -261,7 +261,7 @@ static int set_register(pegasus_t * pegasus, __u16 indx, __u8 data) pegasus->dr.bRequestType = PEGASUS_REQT_WRITE; pegasus->dr.bRequest = PEGASUS_REQ_SET_REG; pegasus->dr.wValue = cpu_to_le16(data); - pegasus->dr.wIndex = cpu_to_le16p(&indx); + pegasus->dr.wIndex = cpu_to_le16(indx); pegasus->dr.wLength = cpu_to_le16(1); pegasus->ctrl_urb->transfer_buffer_length = 1; @@ -476,7 +476,7 @@ static inline void get_node_id(pegasus_t * pegasus, __u8 * id) for (i = 0; i < 3; i++) { read_eprom_word(pegasus, i, &w16); - ((__le16 *) id)[i] = cpu_to_le16p(&w16); + ((__le16 *) id)[i] = cpu_to_le16(w16); } } diff --git a/drivers/usb/core/message.c b/drivers/usb/core/message.c index 6d1048faf08..cc47d36798b 100644 --- a/drivers/usb/core/message.c +++ b/drivers/usb/core/message.c @@ -139,9 +139,9 @@ int usb_control_msg(struct usb_device *dev, unsigned int pipe, __u8 request, dr->bRequestType = requesttype; dr->bRequest = request; - dr->wValue = cpu_to_le16p(&value); - dr->wIndex = cpu_to_le16p(&index); - dr->wLength = cpu_to_le16p(&size); + dr->wValue = cpu_to_le16(value); + dr->wIndex = cpu_to_le16(index); + dr->wLength = cpu_to_le16(size); /* dbg("usb_control_msg"); */ diff --git a/drivers/usb/gadget/net2280.c b/drivers/usb/gadget/net2280.c index 8ae70de2c37..12c6d83b218 100644 --- a/drivers/usb/gadget/net2280.c +++ b/drivers/usb/gadget/net2280.c @@ -669,7 +669,7 @@ fill_dma_desc (struct net2280_ep *ep, struct net2280_request *req, int valid) /* 2280 may be polling VALID_BIT through ep->dma->dmadesc */ wmb (); - td->dmacount = cpu_to_le32p (&dmacount); + td->dmacount = cpu_to_le32(dmacount); } static const u32 dmactl_default = From 92b0da1571353a39d07e0a592c4dda294cc2e04c Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Wed, 29 Oct 2008 15:18:50 -0400 Subject: [PATCH 012/123] USB: straighten out inline code in sysfs.c This patch (as1156) straightens out some code in usbcore. The usb_create_intf_ep_files() and usb_remove_intf_ep_files() routines don't need to be separate inlines; they should be moved bodily into the places where they get used. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/sysfs.c | 35 +++++++++-------------------------- 1 file changed, 9 insertions(+), 26 deletions(-) diff --git a/drivers/usb/core/sysfs.c b/drivers/usb/core/sysfs.c index 4fb65fdc9dc..8c65fa75b5c 100644 --- a/drivers/usb/core/sysfs.c +++ b/drivers/usb/core/sysfs.c @@ -812,32 +812,11 @@ struct attribute_group *usb_interface_groups[] = { NULL }; -static inline void usb_create_intf_ep_files(struct usb_interface *intf, - struct usb_device *udev) -{ - struct usb_host_interface *iface_desc; - int i; - - iface_desc = intf->cur_altsetting; - for (i = 0; i < iface_desc->desc.bNumEndpoints; ++i) - usb_create_ep_files(&intf->dev, &iface_desc->endpoint[i], - udev); -} - -static inline void usb_remove_intf_ep_files(struct usb_interface *intf) -{ - struct usb_host_interface *iface_desc; - int i; - - iface_desc = intf->cur_altsetting; - for (i = 0; i < iface_desc->desc.bNumEndpoints; ++i) - usb_remove_ep_files(&iface_desc->endpoint[i]); -} - int usb_create_sysfs_intf_files(struct usb_interface *intf) { struct usb_device *udev = interface_to_usbdev(intf); struct usb_host_interface *alt = intf->cur_altsetting; + int i; int retval; if (intf->sysfs_files_created || intf->unregistering) @@ -851,18 +830,22 @@ int usb_create_sysfs_intf_files(struct usb_interface *intf) alt->string = usb_cache_string(udev, alt->desc.iInterface); if (alt->string) retval = device_create_file(&intf->dev, &dev_attr_interface); - usb_create_intf_ep_files(intf, udev); + for (i = 0; i < alt->desc.bNumEndpoints; ++i) + usb_create_ep_files(&intf->dev, &alt->endpoint[i], udev); intf->sysfs_files_created = 1; return 0; } void usb_remove_sysfs_intf_files(struct usb_interface *intf) { - struct device *dev = &intf->dev; + struct usb_host_interface *alt = intf->cur_altsetting; + int i; if (!intf->sysfs_files_created) return; - usb_remove_intf_ep_files(intf); - device_remove_file(dev, &dev_attr_interface); + + for (i = 0; i < alt->desc.bNumEndpoints; ++i) + usb_remove_ep_files(&alt->endpoint[i]); + device_remove_file(&intf->dev, &dev_attr_interface); intf->sysfs_files_created = 0; } From d859bffc662482f06ba5603e6f592de4e18cf135 Mon Sep 17 00:00:00 2001 From: Steven Noonan Date: Wed, 5 Nov 2008 12:41:24 -0800 Subject: [PATCH 013/123] USB: EHCI pci-quirks.c: don't wait so long for BIOS handoff Instead of waiting a painful 5000ms, quirk_usb_disable_ehci() now does a 1000ms loop to wait for the BIOS to acknowledge the handoff. The five second delay is really quite irritating to have to deal with every boot up, and I very seriously doubt any non-broken bios takes more than a second to do the actual handoff. Signed-off-by: Steven Noonan Cc: Alan Stern Cc: David Brownell Cc: Jeff Garzik Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/pci-quirks.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/host/pci-quirks.c b/drivers/usb/host/pci-quirks.c index 15e4f1d92c7..75b69847918 100644 --- a/drivers/usb/host/pci-quirks.c +++ b/drivers/usb/host/pci-quirks.c @@ -271,7 +271,7 @@ static void __devinit quirk_usb_disable_ehci(struct pci_dev *pdev) /* if boot firmware now owns EHCI, spin till * it hands it over. */ - msec = 5000; + msec = 1000; while ((cap & EHCI_USBLEGSUP_BIOS) && (msec > 0)) { tried_handoff = 1; msleep(10); From d767d888750a8e15656b7ee15d68f90a151b8936 Mon Sep 17 00:00:00 2001 From: Harvey Harrison Date: Thu, 6 Nov 2008 22:32:15 -0800 Subject: [PATCH 014/123] USB: wusb: annotate association types withe proper endianness Also a trivial annotation in rh.c for: drivers/usb/wusbcore/rh.c:366:9: warning: incorrect type in assignment (different base types) drivers/usb/wusbcore/rh.c:366:9: expected unsigned short [unsigned] [short] [usertype] drivers/usb/wusbcore/rh.c:366:9: got restricted __le16 [usertype] drivers/usb/wusbcore/rh.c:367:9: warning: incorrect type in assignment (different base types) drivers/usb/wusbcore/rh.c:367:9: expected unsigned short [unsigned] [short] [usertype] drivers/usb/wusbcore/rh.c:367:9: got restricted __le16 [usertype] Association types annotation fixes piles of warnings similar to: drivers/usb/wusbcore/cbaf.c:238:30: warning: incorrect type in initializer (different base types) drivers/usb/wusbcore/cbaf.c:238:30: expected restricted __le16 [usertype] id drivers/usb/wusbcore/cbaf.c:238:30: got int drivers/usb/wusbcore/cbaf.c:238:30: warning: incorrect type in initializer (different base types) drivers/usb/wusbcore/cbaf.c:238:30: expected restricted __le16 [usertype] len drivers/usb/wusbcore/cbaf.c:238:30: got int Signed-off-by: Harvey Harrison Cc: David Vrabel Cc: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/usb/wusbcore/rh.c | 2 +- include/linux/usb/association.h | 22 +++++++++++----------- 2 files changed, 12 insertions(+), 12 deletions(-) diff --git a/drivers/usb/wusbcore/rh.c b/drivers/usb/wusbcore/rh.c index 95c6fa3bf6b..3937bf6f8ce 100644 --- a/drivers/usb/wusbcore/rh.c +++ b/drivers/usb/wusbcore/rh.c @@ -326,7 +326,7 @@ static int wusbhc_rh_clear_port_feat(struct wusbhc *wusbhc, u16 feature, static int wusbhc_rh_get_port_status(struct wusbhc *wusbhc, u16 port_idx, u32 *_buf, u16 wLength) { - u16 *buf = (u16 *) _buf; + __le16 *buf = (__le16 *)_buf; if (port_idx > wusbhc->ports_max) return -EINVAL; diff --git a/include/linux/usb/association.h b/include/linux/usb/association.h index 07c5e3cf589..0a4a18b3c1b 100644 --- a/include/linux/usb/association.h +++ b/include/linux/usb/association.h @@ -28,17 +28,17 @@ struct wusb_am_attr { }; /* Different fields defined by the spec */ -#define WUSB_AR_AssociationTypeId { .id = 0x0000, .len = 2 } -#define WUSB_AR_AssociationSubTypeId { .id = 0x0001, .len = 2 } -#define WUSB_AR_Length { .id = 0x0002, .len = 4 } -#define WUSB_AR_AssociationStatus { .id = 0x0004, .len = 4 } -#define WUSB_AR_LangID { .id = 0x0008, .len = 2 } -#define WUSB_AR_DeviceFriendlyName { .id = 0x000b, .len = 64 } /* max */ -#define WUSB_AR_HostFriendlyName { .id = 0x000c, .len = 64 } /* max */ -#define WUSB_AR_CHID { .id = 0x1000, .len = 16 } -#define WUSB_AR_CDID { .id = 0x1001, .len = 16 } -#define WUSB_AR_ConnectionContext { .id = 0x1002, .len = 48 } -#define WUSB_AR_BandGroups { .id = 0x1004, .len = 2 } +#define WUSB_AR_AssociationTypeId { .id = cpu_to_le16(0x0000), .len = cpu_to_le16(2) } +#define WUSB_AR_AssociationSubTypeId { .id = cpu_to_le16(0x0001), .len = cpu_to_le16(2) } +#define WUSB_AR_Length { .id = cpu_to_le16(0x0002), .len = cpu_to_le16(4) } +#define WUSB_AR_AssociationStatus { .id = cpu_to_le16(0x0004), .len = cpu_to_le16(4) } +#define WUSB_AR_LangID { .id = cpu_to_le16(0x0008), .len = cpu_to_le16(2) } +#define WUSB_AR_DeviceFriendlyName { .id = cpu_to_le16(0x000b), .len = cpu_to_le16(64) } /* max */ +#define WUSB_AR_HostFriendlyName { .id = cpu_to_le16(0x000c), .len = cpu_to_le16(64) } /* max */ +#define WUSB_AR_CHID { .id = cpu_to_le16(0x1000), .len = cpu_to_le16(16) } +#define WUSB_AR_CDID { .id = cpu_to_le16(0x1001), .len = cpu_to_le16(16) } +#define WUSB_AR_ConnectionContext { .id = cpu_to_le16(0x1002), .len = cpu_to_le16(48) } +#define WUSB_AR_BandGroups { .id = cpu_to_le16(0x1004), .len = cpu_to_le16(2) } /* CBAF Control Requests (AMS1.0[T4-1] */ enum { From ff8973d9468ea07e61ef492dd8c806a6e1a76ac1 Mon Sep 17 00:00:00 2001 From: Phil Endecott Date: Wed, 12 Nov 2008 15:37:00 +0000 Subject: [PATCH 015/123] USB: Remove restrictions on signal numbers in devio.c Just over a year ago (!) I had this brief exchange with Alan Stern: >> It seems that the signal that can be used with USBDEVFS_DISCSIGNAL and >> in usbdevfs_urb.signr is limited to the real-time signals SIGRTMIN to >> SIGRTMAX. What's the rationale for this restriction? I believe that a >> process can kill() itself with any signal number, can't it? I was >> planning to use SIGIO for usbdevfs_urb.signr and SIGTERM (uncaught) for >> USBDEVFS_DISCSIGNAL. I don't think I'll have a problem with using >> SIGRTMIN+n instead, but I'm curious to know if there's some subtle >> problem with the non-real-time signals that I should be aware of. > > I don't know of any reason for this restriction. Since no-one else could think of a reason either, I offer the following patch which allows any signal to be used with USBDEVFS_DISCSIGNAL and usbdevfs_urb.signr. Signed-off-by: Phil Endecott Cc: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/devio.c | 5 ----- 1 file changed, 5 deletions(-) diff --git a/drivers/usb/core/devio.c b/drivers/usb/core/devio.c index aa79280df15..c85e29381a8 100644 --- a/drivers/usb/core/devio.c +++ b/drivers/usb/core/devio.c @@ -981,9 +981,6 @@ static int proc_do_submiturb(struct dev_state *ps, struct usbdevfs_urb *uurb, return -EINVAL; if (!uurb->buffer) return -EINVAL; - if (uurb->signr != 0 && (uurb->signr < SIGRTMIN || - uurb->signr > SIGRTMAX)) - return -EINVAL; if (!(uurb->type == USBDEVFS_URB_TYPE_CONTROL && (uurb->endpoint & ~USB_ENDPOINT_DIR_MASK) == 0)) { ifnum = findintfep(ps->dev, uurb->endpoint); @@ -1401,8 +1398,6 @@ static int proc_disconnectsignal(struct dev_state *ps, void __user *arg) if (copy_from_user(&ds, arg, sizeof(ds))) return -EFAULT; - if (ds.signr != 0 && (ds.signr < SIGRTMIN || ds.signr > SIGRTMAX)) - return -EINVAL; ps->discsignr = ds.signr; ps->disccontext = ds.context; return 0; From b404299f6bfdf5aa7315e521b570e33f082b2f9e Mon Sep 17 00:00:00 2001 From: Oliver Neukum Date: Fri, 1 Feb 2008 22:44:40 +0100 Subject: [PATCH 016/123] USB: power availability check for berry_charge this introduces a sanity check into berry_charge to give up before damage is done if we lack juice. Signed-off-by: Oliver Neukum Signed-off-by: Greg Kroah-Hartman --- drivers/usb/misc/berry_charge.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/drivers/usb/misc/berry_charge.c b/drivers/usb/misc/berry_charge.c index 24e2dc3148a..c05a85bc592 100644 --- a/drivers/usb/misc/berry_charge.c +++ b/drivers/usb/misc/berry_charge.c @@ -123,6 +123,11 @@ static int berry_probe(struct usb_interface *intf, { struct usb_device *udev = interface_to_usbdev(intf); + if (udev->bus_mA < 500) { + dbg(&udev->dev, "Not enough power to charge available\n"); + return -ENODEV; + } + dbg(&udev->dev, "Power is set to %dmA\n", udev->actconfig->desc.bMaxPower * 2); From 857cc4dfb6420ec0a67b3cda559aaa7c429ddce7 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ilpo=20J=C3=A4rvinen?= Date: Thu, 30 Oct 2008 13:56:47 +0200 Subject: [PATCH 017/123] USB: usbtmc: indent & braces disagree, something else is desired MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit It seems that there's rather involved way to say something which is commonly written in a plain simple form. Some type changes would probably be necessary to get gcc to do bitops instead of divide but it's no worse after my change than before I think. Signed-off-by: Ilpo Järvinen Signed-off-by: Greg Kroah-Hartman --- drivers/usb/class/usbtmc.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) diff --git a/drivers/usb/class/usbtmc.c b/drivers/usb/class/usbtmc.c index 43a863c5cc4..0f5c05f6f9d 100644 --- a/drivers/usb/class/usbtmc.c +++ b/drivers/usb/class/usbtmc.c @@ -21,6 +21,7 @@ #include #include +#include #include #include #include @@ -482,7 +483,6 @@ static ssize_t usbtmc_write(struct file *filp, const char __user *buf, int retval; int actual; unsigned long int n_bytes; - int n; int remaining; int done; int this_part; @@ -526,11 +526,8 @@ static ssize_t usbtmc_write(struct file *filp, const char __user *buf, goto exit; } - n_bytes = 12 + this_part; - if (this_part % 4) - n_bytes += 4 - this_part % 4; - for (n = 12 + this_part; n < n_bytes; n++) - buffer[n] = 0; + n_bytes = roundup(12 + this_part, 4); + memset(buffer + 12 + this_part, 0, n_bytes - (12 + this_part)); retval = usb_bulk_msg(data->usb_dev, usb_sndbulkpipe(data->usb_dev, From 011b15df465745474e3ec85482633685933ed5a7 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Tue, 4 Nov 2008 11:29:27 -0500 Subject: [PATCH 018/123] USB: change interface to usb_lock_device_for_reset() This patch (as1161) changes the interface to usb_lock_device_for_reset(). The existing interface is apparently not very clear, judging from the fact that several of its callers don't use it correctly. The new interface always returns 0 for success and it always requires the caller to unlock the device afterward. The new routine will not return immediately if it is called while the driver's probe method is running. Instead it will wait until the probe is over and the device has been unlocked. This shouldn't cause any problems; I don't know of any cases where drivers call usb_lock_device_for_reset() during probe. Signed-off-by: Alan Stern Cc: Pete Zaitcev Signed-off-by: Greg Kroah-Hartman --- drivers/block/ub.c | 11 +++++------ drivers/hid/usbhid/hid-core.c | 9 ++++----- drivers/media/video/pvrusb2/pvrusb2-hdw.c | 2 +- drivers/usb/core/usb.c | 23 +++++++---------------- drivers/usb/image/microtek.c | 11 +++++------ drivers/usb/storage/transport.c | 8 +++----- 6 files changed, 25 insertions(+), 39 deletions(-) diff --git a/drivers/block/ub.c b/drivers/block/ub.c index 048d71d244d..12fb816db7b 100644 --- a/drivers/block/ub.c +++ b/drivers/block/ub.c @@ -1579,7 +1579,7 @@ static void ub_reset_task(struct work_struct *work) struct ub_dev *sc = container_of(work, struct ub_dev, reset_work); unsigned long flags; struct ub_lun *lun; - int lkr, rc; + int rc; if (!sc->reset) { printk(KERN_WARNING "%s: Running reset unrequested\n", @@ -1597,10 +1597,11 @@ static void ub_reset_task(struct work_struct *work) } else if (sc->dev->actconfig->desc.bNumInterfaces != 1) { ; } else { - if ((lkr = usb_lock_device_for_reset(sc->dev, sc->intf)) < 0) { + rc = usb_lock_device_for_reset(sc->dev, sc->intf); + if (rc < 0) { printk(KERN_NOTICE "%s: usb_lock_device_for_reset failed (%d)\n", - sc->name, lkr); + sc->name, rc); } else { rc = usb_reset_device(sc->dev); if (rc < 0) { @@ -1608,9 +1609,7 @@ static void ub_reset_task(struct work_struct *work) "usb_lock_device_for_reset failed (%d)\n", sc->name, rc); } - - if (lkr) - usb_unlock_device(sc->dev); + usb_unlock_device(sc->dev); } } diff --git a/drivers/hid/usbhid/hid-core.c b/drivers/hid/usbhid/hid-core.c index 03cb494af1c..f0a0f72238a 100644 --- a/drivers/hid/usbhid/hid-core.c +++ b/drivers/hid/usbhid/hid-core.c @@ -102,7 +102,7 @@ static void hid_reset(struct work_struct *work) struct usbhid_device *usbhid = container_of(work, struct usbhid_device, reset_work); struct hid_device *hid = usbhid->hid; - int rc_lock, rc = 0; + int rc = 0; if (test_bit(HID_CLEAR_HALT, &usbhid->iofl)) { dev_dbg(&usbhid->intf->dev, "clear halt\n"); @@ -113,11 +113,10 @@ static void hid_reset(struct work_struct *work) else if (test_bit(HID_RESET_PENDING, &usbhid->iofl)) { dev_dbg(&usbhid->intf->dev, "resetting device\n"); - rc = rc_lock = usb_lock_device_for_reset(hid_to_usb_dev(hid), usbhid->intf); - if (rc_lock >= 0) { + rc = usb_lock_device_for_reset(hid_to_usb_dev(hid), usbhid->intf); + if (rc == 0) { rc = usb_reset_device(hid_to_usb_dev(hid)); - if (rc_lock) - usb_unlock_device(hid_to_usb_dev(hid)); + usb_unlock_device(hid_to_usb_dev(hid)); } clear_bit(HID_RESET_PENDING, &usbhid->iofl); } diff --git a/drivers/media/video/pvrusb2/pvrusb2-hdw.c b/drivers/media/video/pvrusb2/pvrusb2-hdw.c index 8fb92ac78c7..fa304e5f252 100644 --- a/drivers/media/video/pvrusb2/pvrusb2-hdw.c +++ b/drivers/media/video/pvrusb2/pvrusb2-hdw.c @@ -3655,7 +3655,7 @@ void pvr2_hdw_device_reset(struct pvr2_hdw *hdw) int ret; pvr2_trace(PVR2_TRACE_INIT,"Performing a device reset..."); ret = usb_lock_device_for_reset(hdw->usb_dev,NULL); - if (ret == 1) { + if (ret == 0) { ret = usb_reset_device(hdw->usb_dev); usb_unlock_device(hdw->usb_dev); } else { diff --git a/drivers/usb/core/usb.c b/drivers/usb/core/usb.c index 399e15fc505..400fa4cc9a3 100644 --- a/drivers/usb/core/usb.c +++ b/drivers/usb/core/usb.c @@ -513,10 +513,7 @@ EXPORT_SYMBOL_GPL(usb_put_intf); * disconnect; in some drivers (such as usb-storage) the disconnect() * or suspend() method will block waiting for a device reset to complete. * - * Returns a negative error code for failure, otherwise 1 or 0 to indicate - * that the device will or will not have to be unlocked. (0 can be - * returned when an interface is given and is BINDING, because in that - * case the driver already owns the device lock.) + * Returns a negative error code for failure, otherwise 0. */ int usb_lock_device_for_reset(struct usb_device *udev, const struct usb_interface *iface) @@ -527,16 +524,9 @@ int usb_lock_device_for_reset(struct usb_device *udev, return -ENODEV; if (udev->state == USB_STATE_SUSPENDED) return -EHOSTUNREACH; - if (iface) { - switch (iface->condition) { - case USB_INTERFACE_BINDING: - return 0; - case USB_INTERFACE_BOUND: - break; - default: - return -EINTR; - } - } + if (iface && (iface->condition == USB_INTERFACE_UNBINDING || + iface->condition == USB_INTERFACE_UNBOUND)) + return -EINTR; while (usb_trylock_device(udev) != 0) { @@ -550,10 +540,11 @@ int usb_lock_device_for_reset(struct usb_device *udev, return -ENODEV; if (udev->state == USB_STATE_SUSPENDED) return -EHOSTUNREACH; - if (iface && iface->condition != USB_INTERFACE_BOUND) + if (iface && (iface->condition == USB_INTERFACE_UNBINDING || + iface->condition == USB_INTERFACE_UNBOUND)) return -EINTR; } - return 1; + return 0; } EXPORT_SYMBOL_GPL(usb_lock_device_for_reset); diff --git a/drivers/usb/image/microtek.c b/drivers/usb/image/microtek.c index 885867a86de..4541dfcea88 100644 --- a/drivers/usb/image/microtek.c +++ b/drivers/usb/image/microtek.c @@ -350,17 +350,16 @@ static int mts_scsi_abort(struct scsi_cmnd *srb) static int mts_scsi_host_reset(struct scsi_cmnd *srb) { struct mts_desc* desc = (struct mts_desc*)(srb->device->host->hostdata[0]); - int result, rc; + int result; MTS_DEBUG_GOT_HERE(); mts_debug_dump(desc); - rc = usb_lock_device_for_reset(desc->usb_dev, desc->usb_intf); - if (rc < 0) - return FAILED; - result = usb_reset_device(desc->usb_dev); - if (rc) + result = usb_lock_device_for_reset(desc->usb_dev, desc->usb_intf); + if (result == 0) { + result = usb_reset_device(desc->usb_dev); usb_unlock_device(desc->usb_dev); + } return result ? FAILED : SUCCESS; } diff --git a/drivers/usb/storage/transport.c b/drivers/usb/storage/transport.c index 79108d5d317..2058db41618 100644 --- a/drivers/usb/storage/transport.c +++ b/drivers/usb/storage/transport.c @@ -1173,10 +1173,9 @@ int usb_stor_Bulk_reset(struct us_data *us) */ int usb_stor_port_reset(struct us_data *us) { - int result, rc_lock; + int result; - result = rc_lock = - usb_lock_device_for_reset(us->pusb_dev, us->pusb_intf); + result = usb_lock_device_for_reset(us->pusb_dev, us->pusb_intf); if (result < 0) US_DEBUGP("unable to lock device for reset: %d\n", result); else { @@ -1189,8 +1188,7 @@ int usb_stor_port_reset(struct us_data *us) US_DEBUGP("usb_reset_device returns %d\n", result); } - if (rc_lock) - usb_unlock_device(us->pusb_dev); + usb_unlock_device(us->pusb_dev); } return result; } From cd40c4c45eaedc289d2e1cc33b18a49a211f0f82 Mon Sep 17 00:00:00 2001 From: Anton Vorontsov Date: Sat, 8 Nov 2008 20:51:48 +0300 Subject: [PATCH 019/123] USB: fsl_qe_udc: Check for muram allocation errors The QE UDC doesn't check for cpm_muram_alloc() return values, this might cause all sorts of misbehaviour when cpm_muram_alloc() failed to allocate the muram memory. While at at, change few dev_dbg() calls to dev_err(), so that the driver would not die silently. Signed-off-by: Anton Vorontsov Cc: David Brownell Cc: Kumar Gala Cc: Li Yang Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/fsl_qe_udc.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/drivers/usb/gadget/fsl_qe_udc.c b/drivers/usb/gadget/fsl_qe_udc.c index f4027256509..d6c5bcd4006 100644 --- a/drivers/usb/gadget/fsl_qe_udc.c +++ b/drivers/usb/gadget/fsl_qe_udc.c @@ -26,6 +26,7 @@ #include #include #include +#include #include #include #include @@ -370,6 +371,9 @@ static int qe_ep_bd_init(struct qe_udc *udc, unsigned char pipe_num) /* alloc multi-ram for BD rings and set the ep parameters */ tmp_addr = cpm_muram_alloc(sizeof(struct qe_bd) * (bdring_len + USB_BDRING_LEN_TX), QE_ALIGNMENT_OF_BD); + if (IS_ERR_VALUE(tmp_addr)) + return -ENOMEM; + out_be16(&epparam->rbase, (u16)tmp_addr); out_be16(&epparam->tbase, (u16)(tmp_addr + (sizeof(struct qe_bd) * bdring_len))); @@ -689,7 +693,7 @@ en_done2: en_done1: spin_unlock_irqrestore(&udc->lock, flags); en_done: - dev_dbg(udc->dev, "failed to initialize %s\n", ep->ep.name); + dev_err(udc->dev, "failed to initialize %s\n", ep->ep.name); return -ENODEV; } @@ -2408,6 +2412,8 @@ static struct qe_udc __devinit *qe_udc_config(struct of_device *ofdev) tmp_addr = cpm_muram_alloc((USB_MAX_ENDPOINTS * sizeof(struct usb_ep_para)), USB_EP_PARA_ALIGNMENT); + if (IS_ERR_VALUE(tmp_addr)) + goto cleanup; for (i = 0; i < USB_MAX_ENDPOINTS; i++) { out_be16(&usbpram->epptr[i], (u16)tmp_addr); @@ -2513,7 +2519,7 @@ static int __devinit qe_udc_probe(struct of_device *ofdev, /* Initialize the udc structure including QH member and other member */ udc_controller = qe_udc_config(ofdev); if (!udc_controller) { - dev_dbg(&ofdev->dev, "udc_controll is NULL\n"); + dev_err(&ofdev->dev, "failed to initialize\n"); return -ENOMEM; } @@ -2568,7 +2574,7 @@ static int __devinit qe_udc_probe(struct of_device *ofdev, /* create a buf for ZLP send, need to remain zeroed */ udc_controller->nullbuf = kzalloc(256, GFP_KERNEL); if (udc_controller->nullbuf == NULL) { - dev_dbg(udc_controller->dev, "cannot alloc nullbuf\n"); + dev_err(udc_controller->dev, "cannot alloc nullbuf\n"); ret = -ENOMEM; goto err3; } From 796bcae7361c28cf825780f6f1aac9dd3411394e Mon Sep 17 00:00:00 2001 From: Vitaly Bordug Date: Sun, 9 Nov 2008 19:43:30 +0100 Subject: [PATCH 020/123] USB: powerpc: Workaround for the PPC440EPX USBH_23 errata [take 3] A published errata for ppc440epx states, that when running Linux with both EHCI and OHCI modules loaded, the EHCI module experiences a fatal error when a high-speed device is connected to the USB2.0, and functions normally if OHCI module is not loaded. There used to be recommendation to use only hi-speed or full-speed devices with specific conditions, when respective module was unloaded. Later, it was observed that ohci suspend is enough to keep things going, and it was turned into workaround, as explained below. Quote from original descriprion: The 440EPx USB 2.0 Host controller is an EHCI compliant controller. In USB 2.0 Host controllers, each EHCI controller has one or more companion controllers, which may be OHCI or UHCI. An USB 2.0 Host controller will contain one or more ports. For each port, only one of the controllers is connected at any one time. In the 440EPx, there is only one OHCI companion controller, and only one USB 2.0 Host port. All ports on an USB 2.0 controller default to the companion controller. If you load only an ohci driver, it will have control of the ports and any deviceplugged in will operate, although high speed devices will be forced to operate at full speed. When an ehci driver is loaded, it explicitly takes control of the ports. If there is a device connected, and / or every time there is a new device connected, the ehci driver determines if the device is high speed or not. If it is high speed, the driver retains control of the port. If it is not, the driver explicitly gives the companion controller control of the port. The is a software workaround that uses Initial version of the software workaround was posted to linux-usb-devel: http://www.mail-archive.com/linux-usb-devel@lists.sourceforge.net/msg54019.html and later available from amcc.com: http://www.amcc.com/Embedded/Downloads/download.html?cat=1&family=15&ins=2 The patch below is generally based on the latter, but reworked to powerpc/of_device USB drivers, and uses a few devicetree inquiries to get rid of (some) hardcoded defines. Signed-off-by: Vitaly Bordug Signed-off-by: Stefan Roese Cc: David Brownell Signed-off-by: Greg Kroah-Hartman --- arch/powerpc/boot/dts/sequoia.dts | 2 +- drivers/usb/host/ehci-hub.c | 9 ++++++- drivers/usb/host/ehci-ppc-of.c | 45 ++++++++++++++++++++++++++++++- drivers/usb/host/ehci.h | 34 +++++++++++++++++++++++ drivers/usb/host/ohci-ppc-of.c | 25 +++++++++++++++++ 5 files changed, 112 insertions(+), 3 deletions(-) diff --git a/arch/powerpc/boot/dts/sequoia.dts b/arch/powerpc/boot/dts/sequoia.dts index 3b295e8df53..43cc68bd319 100644 --- a/arch/powerpc/boot/dts/sequoia.dts +++ b/arch/powerpc/boot/dts/sequoia.dts @@ -134,7 +134,7 @@ }; USB1: usb@e0000400 { - compatible = "ohci-be"; + compatible = "ibm,usb-ohci-440epx", "ohci-be"; reg = <0x00000000 0xe0000400 0x00000060>; interrupt-parent = <&UIC0>; interrupts = <0x15 0x8>; diff --git a/drivers/usb/host/ehci-hub.c b/drivers/usb/host/ehci-hub.c index 38650707dfe..97a53a48a3d 100644 --- a/drivers/usb/host/ehci-hub.c +++ b/drivers/usb/host/ehci-hub.c @@ -434,8 +434,15 @@ static int check_reset_complete ( port_status &= ~PORT_RWC_BITS; ehci_writel(ehci, port_status, status_reg); - } else + /* ensure 440EPX ohci controller state is operational */ + if (ehci->has_amcc_usb23) + set_ohci_hcfs(ehci, 1); + } else { ehci_dbg (ehci, "port %d high speed\n", index + 1); + /* ensure 440EPx ohci controller state is suspended */ + if (ehci->has_amcc_usb23) + set_ohci_hcfs(ehci, 0); + } return port_status; } diff --git a/drivers/usb/host/ehci-ppc-of.c b/drivers/usb/host/ehci-ppc-of.c index b018deed2e8..ef732b704f5 100644 --- a/drivers/usb/host/ehci-ppc-of.c +++ b/drivers/usb/host/ehci-ppc-of.c @@ -107,11 +107,13 @@ ehci_hcd_ppc_of_probe(struct of_device *op, const struct of_device_id *match) { struct device_node *dn = op->node; struct usb_hcd *hcd; - struct ehci_hcd *ehci; + struct ehci_hcd *ehci = NULL; struct resource res; int irq; int rv; + struct device_node *np; + if (usb_disabled()) return -ENODEV; @@ -149,6 +151,20 @@ ehci_hcd_ppc_of_probe(struct of_device *op, const struct of_device_id *match) } ehci = hcd_to_ehci(hcd); + np = of_find_compatible_node(NULL, NULL, "ibm,usb-ohci-440epx"); + if (np != NULL) { + /* claim we really affected by usb23 erratum */ + if (!of_address_to_resource(np, 0, &res)) + ehci->ohci_hcctrl_reg = ioremap(res.start + + OHCI_HCCTRL_OFFSET, OHCI_HCCTRL_LEN); + else + pr_debug(__FILE__ ": no ohci offset in fdt\n"); + if (!ehci->ohci_hcctrl_reg) { + pr_debug(__FILE__ ": ioremap for ohci hcctrl failed\n"); + } else { + ehci->has_amcc_usb23 = 1; + } + } if (of_get_property(dn, "big-endian", NULL)) { ehci->big_endian_mmio = 1; @@ -181,6 +197,9 @@ err_ioremap: irq_dispose_mapping(irq); err_irq: release_mem_region(hcd->rsrc_start, hcd->rsrc_len); + + if (ehci->has_amcc_usb23) + iounmap(ehci->ohci_hcctrl_reg); err_rmr: usb_put_hcd(hcd); @@ -191,6 +210,11 @@ err_rmr: static int ehci_hcd_ppc_of_remove(struct of_device *op) { struct usb_hcd *hcd = dev_get_drvdata(&op->dev); + struct ehci_hcd *ehci = hcd_to_ehci(hcd); + + struct device_node *np; + struct resource res; + dev_set_drvdata(&op->dev, NULL); dev_dbg(&op->dev, "stopping PPC-OF USB Controller\n"); @@ -201,6 +225,25 @@ static int ehci_hcd_ppc_of_remove(struct of_device *op) irq_dispose_mapping(hcd->irq); release_mem_region(hcd->rsrc_start, hcd->rsrc_len); + /* use request_mem_region to test if the ohci driver is loaded. if so + * ensure the ohci core is operational. + */ + if (ehci->has_amcc_usb23) { + np = of_find_compatible_node(NULL, NULL, "ibm,usb-ohci-440epx"); + if (np != NULL) { + if (!of_address_to_resource(np, 0, &res)) + if (!request_mem_region(res.start, + 0x4, hcd_name)) + set_ohci_hcfs(ehci, 1); + else + release_mem_region(res.start, 0x4); + else + pr_debug(__FILE__ ": no ohci offset in fdt\n"); + of_node_put(np); + } + + iounmap(ehci->ohci_hcctrl_reg); + } usb_put_hcd(hcd); return 0; diff --git a/drivers/usb/host/ehci.h b/drivers/usb/host/ehci.h index c7d4b5a06bd..fb7054ccf4f 100644 --- a/drivers/usb/host/ehci.h +++ b/drivers/usb/host/ehci.h @@ -120,6 +120,16 @@ struct ehci_hcd { /* one per controller */ unsigned has_fsl_port_bug:1; /* FreeScale */ unsigned big_endian_mmio:1; unsigned big_endian_desc:1; + unsigned has_amcc_usb23:1; + + /* required for usb32 quirk */ + #define OHCI_CTRL_HCFS (3 << 6) + #define OHCI_USB_OPER (2 << 6) + #define OHCI_USB_SUSPEND (3 << 6) + + #define OHCI_HCCTRL_OFFSET 0x4 + #define OHCI_HCCTRL_LEN 0x4 + __hc32 *ohci_hcctrl_reg; u8 sbrn; /* packed release number */ @@ -636,6 +646,30 @@ static inline void ehci_writel(const struct ehci_hcd *ehci, #endif } +/* + * On certain ppc-44x SoC there is a HW issue, that could only worked around with + * explicit suspend/operate of OHCI. This function hereby makes sense only on that arch. + * Other common bits are dependant on has_amcc_usb23 quirk flag. + */ +#ifdef CONFIG_44x +static inline void set_ohci_hcfs(struct ehci_hcd *ehci, int operational) +{ + u32 hc_control; + + hc_control = (readl_be(ehci->ohci_hcctrl_reg) & ~OHCI_CTRL_HCFS); + if (operational) + hc_control |= OHCI_USB_OPER; + else + hc_control |= OHCI_USB_SUSPEND; + + writel_be(hc_control, ehci->ohci_hcctrl_reg); + (void) readl_be(ehci->ohci_hcctrl_reg); +} +#else +static inline void set_ohci_hcfs(struct ehci_hcd *ehci, int operational) +{ } +#endif + /*-------------------------------------------------------------------------*/ /* diff --git a/drivers/usb/host/ohci-ppc-of.c b/drivers/usb/host/ohci-ppc-of.c index 7ac53264ead..68a30171029 100644 --- a/drivers/usb/host/ohci-ppc-of.c +++ b/drivers/usb/host/ohci-ppc-of.c @@ -91,6 +91,7 @@ ohci_hcd_ppc_of_probe(struct of_device *op, const struct of_device_id *match) int rv; int is_bigendian; + struct device_node *np; if (usb_disabled()) return -ENODEV; @@ -147,6 +148,30 @@ ohci_hcd_ppc_of_probe(struct of_device *op, const struct of_device_id *match) if (rv == 0) return 0; + /* by now, 440epx is known to show usb_23 erratum */ + np = of_find_compatible_node(NULL, NULL, "ibm,usb-ehci-440epx"); + + /* Work around - At this point ohci_run has executed, the + * controller is running, everything, the root ports, etc., is + * set up. If the ehci driver is loaded, put the ohci core in + * the suspended state. The ehci driver will bring it out of + * suspended state when / if a non-high speed USB device is + * attached to the USB Host port. If the ehci driver is not + * loaded, do nothing. request_mem_region is used to test if + * the ehci driver is loaded. + */ + if (np != NULL) { + if (!of_address_to_resource(np, 0, &res)) { + if (!request_mem_region(res.start, 0x4, hcd_name)) { + writel_be((readl_be(&ohci->regs->control) | + OHCI_USB_SUSPEND), &ohci->regs->control); + (void) readl_be(&ohci->regs->control); + } else + release_mem_region(res.start, 0x4); + } else + pr_debug(__FILE__ ": cannot get ehci offset from fdt\n"); + } + iounmap(hcd->regs); err_ioremap: irq_dispose_mapping(irq); From 427c4f333474f5447f62387c1fb060e586c1a781 Mon Sep 17 00:00:00 2001 From: Kay Sievers Date: Fri, 7 Nov 2008 01:52:53 +0100 Subject: [PATCH 021/123] usb: struct device - replace bus_id with dev_name(), dev_set_name() Signed-off-by: Kay Sievers Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/ohci-tmio.c | 2 +- drivers/usb/musb/musb_core.c | 4 ++-- drivers/usb/musb/musb_gadget.c | 2 +- drivers/usb/musb/musbhsdma.c | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/usb/host/ohci-tmio.c b/drivers/usb/host/ohci-tmio.c index f9f134af0bd..8dabe8e31d8 100644 --- a/drivers/usb/host/ohci-tmio.c +++ b/drivers/usb/host/ohci-tmio.c @@ -201,7 +201,7 @@ static int __devinit ohci_hcd_tmio_drv_probe(struct platform_device *dev) if (!cell) return -EINVAL; - hcd = usb_create_hcd(&ohci_tmio_hc_driver, &dev->dev, dev->dev.bus_id); + hcd = usb_create_hcd(&ohci_tmio_hc_driver, &dev->dev, dev_name(&dev->dev)); if (!hcd) { ret = -ENOMEM; goto err_usb_create_hcd; diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c index 5280dba9b1f..0b742ebe5c8 100644 --- a/drivers/usb/musb/musb_core.c +++ b/drivers/usb/musb/musb_core.c @@ -1781,7 +1781,7 @@ allocate_instance(struct device *dev, #ifdef CONFIG_USB_MUSB_HDRC_HCD struct usb_hcd *hcd; - hcd = usb_create_hcd(&musb_hc_driver, dev, dev->bus_id); + hcd = usb_create_hcd(&musb_hc_driver, dev, dev_name(dev)); if (!hcd) return NULL; /* usbcore sets dev->driver_data to hcd, and sometimes uses that... */ @@ -1984,7 +1984,7 @@ bad_config: INIT_WORK(&musb->irq_work, musb_irq_work); /* attach to the IRQ */ - if (request_irq(nIrq, musb->isr, 0, dev->bus_id, musb)) { + if (request_irq(nIrq, musb->isr, 0, dev_name(dev), musb)) { dev_err(dev, "request_irq %d failed!\n", nIrq); status = -ENODEV; goto fail2; diff --git a/drivers/usb/musb/musb_gadget.c b/drivers/usb/musb/musb_gadget.c index d6a802c224f..6197daeab8f 100644 --- a/drivers/usb/musb/musb_gadget.c +++ b/drivers/usb/musb/musb_gadget.c @@ -1633,7 +1633,7 @@ int __init musb_gadget_setup(struct musb *musb) musb->g.speed = USB_SPEED_UNKNOWN; /* this "gadget" abstracts/virtualizes the controller */ - strcpy(musb->g.dev.bus_id, "gadget"); + dev_set_name(&musb->g.dev, "gadget"); musb->g.dev.parent = musb->controller; musb->g.dev.dma_mask = musb->controller->dma_mask; musb->g.dev.release = musb_gadget_release; diff --git a/drivers/usb/musb/musbhsdma.c b/drivers/usb/musb/musbhsdma.c index 8c734ef2c1e..d193b784908 100644 --- a/drivers/usb/musb/musbhsdma.c +++ b/drivers/usb/musb/musbhsdma.c @@ -424,7 +424,7 @@ dma_controller_create(struct musb *musb, void __iomem *base) controller->controller.channel_abort = dma_channel_abort; if (request_irq(irq, dma_controller_irq, IRQF_DISABLED, - musb->controller->bus_id, &controller->controller)) { + dev_name(musb->controller), &controller->controller)) { dev_err(dev, "request_irq %d failed!\n", irq); dma_controller_destroy(&controller->controller); From 74c71ebd8d7c7a513022851a02bb52b9fa7e0dcb Mon Sep 17 00:00:00 2001 From: Kay Sievers Date: Fri, 7 Nov 2008 01:47:15 +0100 Subject: [PATCH 022/123] oxu210hp-hcd.c: struct device - replace bus_id with dev_name(), dev_set_name() Signed-off-by: Kay Sievers Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/oxu210hp-hcd.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/usb/host/oxu210hp-hcd.c b/drivers/usb/host/oxu210hp-hcd.c index 251123c29d8..75548f7c716 100644 --- a/drivers/usb/host/oxu210hp-hcd.c +++ b/drivers/usb/host/oxu210hp-hcd.c @@ -3825,7 +3825,7 @@ static int oxu_drv_probe(struct platform_device *pdev) res = platform_get_resource(pdev, IORESOURCE_IRQ, 0); if (!res) { dev_err(&pdev->dev, - "no IRQ! Check %s setup!\n", pdev->dev.bus_id); + "no IRQ! Check %s setup!\n", dev_name(&pdev->dev)); return -ENODEV; } irq = res->start; @@ -3834,7 +3834,7 @@ static int oxu_drv_probe(struct platform_device *pdev) res = platform_get_resource(pdev, IORESOURCE_MEM, 0); if (!res) { dev_err(&pdev->dev, "no registers address! Check %s setup!\n", - pdev->dev.bus_id); + dev_name(&pdev->dev)); return -ENODEV; } memstart = res->start; @@ -3893,7 +3893,7 @@ error_set_irq_type: error_ioremap: release_mem_region(memstart, memlen); - dev_err(&pdev->dev, "init %s fail, %d\n", pdev->dev.bus_id, ret); + dev_err(&pdev->dev, "init %s fail, %d\n", dev_name(&pdev->dev), ret); return ret; } From d4f373e57d3916814110968c5ea1155a8d972b5a Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Mon, 10 Nov 2008 14:07:45 -0500 Subject: [PATCH 023/123] USB: usb-storage: add "quirks=" module parameter This patch (as1163b) adds a "quirks=" module parameter to usb-storage. This will allow people to make short-term changes to their unusual_devs list without rebuilding the entire driver. Testing will become much easier, and less-sophisticated users will be able to access their buggy devices after a simple config-file change instead of having to wait for a new kernel release. The patch also adds a documentation entry for usb-storage's "delay_use" parameter, which has been around for years but but was never listed among the kernel parameters. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- Documentation/kernel-parameters.txt | 29 +++++++ drivers/usb/storage/usb.c | 113 ++++++++++++++++++++++++++++ 2 files changed, 142 insertions(+) diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index 0b3f6711d2f..8eb6e35405c 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt @@ -91,6 +91,7 @@ parameter is applicable: SUSPEND System suspend states are enabled. FTRACE Function tracing enabled. TS Appropriate touchscreen support is enabled. + UMS USB Mass Storage support is enabled. USB USB support is enabled. USBHID USB Human Interface Device support is enabled. V4L Video For Linux support is enabled. @@ -2383,6 +2384,34 @@ and is between 256 and 4096 characters. It is defined in the file usbhid.mousepoll= [USBHID] The interval which mice are to be polled at. + usb-storage.delay_use= + [UMS] The delay in seconds before a new device is + scanned for Logical Units (default 5). + + usb-storage.quirks= + [UMS] A list of quirks entries to supplement or + override the built-in unusual_devs list. List + entries are separated by commas. Each entry has + the form VID:PID:Flags where VID and PID are Vendor + and Product ID values (4-digit hex numbers) and + Flags is a set of characters, each corresponding + to a common usb-storage quirk flag as follows: + c = FIX_CAPACITY (decrease the reported + device capacity by one sector); + i = IGNORE_DEVICE (don't bind to this + device); + l = NOT_LOCKABLE (don't try to lock and + unlock ejectable media); + m = MAX_SECTORS_64 (don't transfer more + than 64 sectors = 32 KB at a time); + r = IGNORE_RESIDUE (the device reports + bogus residue values); + s = SINGLE_LUN (the device has only one + Logical Unit); + w = NO_WP_DETECT (don't test whether the + medium is write-protected). + Example: quirks=0419:aaf5:rl,0421:0433:rc + add_efi_memmap [EFI; x86-32,X86-64] Include EFI memory map in kernel's map of available physical RAM. diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index 27016fd2cad..eb1a53a3e5c 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -113,6 +113,16 @@ static unsigned int delay_use = 5; module_param(delay_use, uint, S_IRUGO | S_IWUSR); MODULE_PARM_DESC(delay_use, "seconds to delay before using a new device"); +static char *quirks; +module_param(quirks, charp, S_IRUGO); +MODULE_PARM_DESC(quirks, "supplemental list of device IDs and their quirks"); + +struct quirks_entry { + u16 vid, pid; + u32 fflags; +}; +static struct quirks_entry *quirks_list, *quirks_end; + /* * The entries in this table correspond, line for line, @@ -473,6 +483,30 @@ static int associate_dev(struct us_data *us, struct usb_interface *intf) return 0; } +/* Adjust device flags based on the "quirks=" module parameter */ +static void adjust_quirks(struct us_data *us) +{ + u16 vid, pid; + struct quirks_entry *q; + unsigned int mask = (US_FL_FIX_CAPACITY | US_FL_IGNORE_DEVICE | + US_FL_NOT_LOCKABLE | US_FL_MAX_SECTORS_64 | + US_FL_IGNORE_RESIDUE | US_FL_SINGLE_LUN | + US_FL_NO_WP_DETECT); + + vid = le16_to_cpu(us->pusb_dev->descriptor.idVendor); + pid = le16_to_cpu(us->pusb_dev->descriptor.idProduct); + + for (q = quirks_list; q != quirks_end; ++q) { + if (q->vid == vid && q->pid == pid) { + us->fflags = (us->fflags & ~mask) | q->fflags; + dev_info(&us->pusb_intf->dev, "Quirks match for " + "vid %04x pid %04x: %x\n", + vid, pid, q->fflags); + break; + } + } +} + /* Find an unusual_dev descriptor (always succeeds in the current code) */ static struct us_unusual_dev *find_unusual(const struct usb_device_id *id) { @@ -497,6 +531,7 @@ static int get_device_info(struct us_data *us, const struct usb_device_id *id) idesc->bInterfaceProtocol : unusual_dev->useTransport; us->fflags = USB_US_ORIG_FLAGS(id->driver_info); + adjust_quirks(us); if (us->fflags & US_FL_IGNORE_DEVICE) { printk(KERN_INFO USB_STORAGE "device ignored\n"); @@ -1061,10 +1096,88 @@ static struct usb_driver usb_storage_driver = { .soft_unbind = 1, }; +/* Works only for digits and letters, but small and fast */ +#define TOLOWER(x) ((x) | 0x20) + +static void __init parse_quirks(void) +{ + int n, i; + char *p; + + if (!quirks) + return; + + /* Count the ':' characters to get 2 * the number of entries */ + n = 0; + for (p = quirks; *p; ++p) { + if (*p == ':') + ++n; + } + n /= 2; + if (n == 0) + return; /* Don't allocate 0 bytes */ + + quirks_list = kmalloc(n * sizeof(*quirks_list), GFP_KERNEL); + if (!quirks_list) + return; + + p = quirks; + quirks_end = quirks_list; + for (i = 0; i < n && *p; ++i) { + unsigned f = 0; + + /* Each entry consists of VID:PID:flags */ + quirks_end->vid = simple_strtoul(p, &p, 16); + if (*p != ':') + goto skip_to_next; + quirks_end->pid = simple_strtoul(p+1, &p, 16); + if (*p != ':') + goto skip_to_next; + + while (*++p && *p != ',') { + switch (TOLOWER(*p)) { + case 'c': + f |= US_FL_FIX_CAPACITY; + break; + case 'i': + f |= US_FL_IGNORE_DEVICE; + break; + case 'l': + f |= US_FL_NOT_LOCKABLE; + break; + case 'm': + f |= US_FL_MAX_SECTORS_64; + break; + case 'r': + f |= US_FL_IGNORE_RESIDUE; + break; + case 's': + f |= US_FL_SINGLE_LUN; + break; + case 'w': + f |= US_FL_NO_WP_DETECT; + break; + /* Ignore unrecognized flag characters */ + } + } + quirks_end->fflags = f; + ++quirks_end; + + skip_to_next: + /* Entries are separated by commas */ + while (*p) { + if (*p++ == ',') + break; + } + } /* for (i = 0; ...) */ +} + static int __init usb_stor_init(void) { int retval; + printk(KERN_INFO "Initializing USB Mass Storage driver...\n"); + parse_quirks(); /* register the driver, return usb_register return code if error */ retval = usb_register(&usb_storage_driver); From 9ac39f28b5237a629e41ccfc1f73d3a55723045c Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Wed, 12 Nov 2008 16:19:49 -0500 Subject: [PATCH 024/123] USB: add asynchronous autosuspend/autoresume support This patch (as1160b) adds support routines for asynchronous autosuspend and autoresume, with accompanying documentation updates. There already are several potential users of this interface, and others are likely to arise as autosuspend support becomes more widespread. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- Documentation/usb/power-management.txt | 22 +++++-- drivers/usb/core/driver.c | 86 ++++++++++++++++++++++++++ drivers/usb/core/hub.c | 5 +- drivers/usb/core/usb.c | 1 + drivers/usb/core/usb.h | 1 + include/linux/usb.h | 9 +++ 6 files changed, 117 insertions(+), 7 deletions(-) diff --git a/Documentation/usb/power-management.txt b/Documentation/usb/power-management.txt index e48ea1d5101..ad642615ad4 100644 --- a/Documentation/usb/power-management.txt +++ b/Documentation/usb/power-management.txt @@ -313,11 +313,13 @@ three of the methods listed above. In addition, a driver indicates that it supports autosuspend by setting the .supports_autosuspend flag in its usb_driver structure. It is then responsible for informing the USB core whenever one of its interfaces becomes busy or idle. The -driver does so by calling these three functions: +driver does so by calling these five functions: int usb_autopm_get_interface(struct usb_interface *intf); void usb_autopm_put_interface(struct usb_interface *intf); int usb_autopm_set_interface(struct usb_interface *intf); + int usb_autopm_get_interface_async(struct usb_interface *intf); + void usb_autopm_put_interface_async(struct usb_interface *intf); The functions work by maintaining a counter in the usb_interface structure. When intf->pm_usage_count is > 0 then the interface is @@ -330,10 +332,12 @@ associated with the device itself rather than any of its interfaces. This field is used only by the USB core.) The driver owns intf->pm_usage_count; it can modify the value however -and whenever it likes. A nice aspect of the usb_autopm_* routines is -that the changes they make are protected by the usb_device structure's -PM mutex (udev->pm_mutex); however drivers may change pm_usage_count -without holding the mutex. +and whenever it likes. A nice aspect of the non-async usb_autopm_* +routines is that the changes they make are protected by the usb_device +structure's PM mutex (udev->pm_mutex); however drivers may change +pm_usage_count without holding the mutex. Drivers using the async +routines are responsible for their own synchronization and mutual +exclusion. usb_autopm_get_interface() increments pm_usage_count and attempts an autoresume if the new value is > 0 and the @@ -348,6 +352,14 @@ without holding the mutex. is suspended, and it attempts an autosuspend if the value is <= 0 and the device isn't suspended. + usb_autopm_get_interface_async() and + usb_autopm_put_interface_async() do almost the same things as + their non-async counterparts. The differences are: they do + not acquire the PM mutex, and they use a workqueue to do their + jobs. As a result they can be called in an atomic context, + such as an URB's completion handler, but when they return the + device will not generally not yet be in the desired state. + There also are a couple of utility routines drivers can use: usb_autopm_enable() sets pm_usage_cnt to 0 and then calls diff --git a/drivers/usb/core/driver.c b/drivers/usb/core/driver.c index 8c081308b0e..23b3c7e79d4 100644 --- a/drivers/usb/core/driver.c +++ b/drivers/usb/core/driver.c @@ -1341,6 +1341,19 @@ void usb_autosuspend_work(struct work_struct *work) usb_autopm_do_device(udev, 0); } +/* usb_autoresume_work - callback routine to autoresume a USB device */ +void usb_autoresume_work(struct work_struct *work) +{ + struct usb_device *udev = + container_of(work, struct usb_device, autoresume); + + /* Wake it up, let the drivers do their thing, and then put it + * back to sleep. + */ + if (usb_autopm_do_device(udev, 1) == 0) + usb_autopm_do_device(udev, -1); +} + /** * usb_autosuspend_device - delayed autosuspend of a USB device and its interfaces * @udev: the usb_device to autosuspend @@ -1491,6 +1504,45 @@ void usb_autopm_put_interface(struct usb_interface *intf) } EXPORT_SYMBOL_GPL(usb_autopm_put_interface); +/** + * usb_autopm_put_interface_async - decrement a USB interface's PM-usage counter + * @intf: the usb_interface whose counter should be decremented + * + * This routine does essentially the same thing as + * usb_autopm_put_interface(): it decrements @intf's usage counter and + * queues a delayed autosuspend request if the counter is <= 0. The + * difference is that it does not acquire the device's pm_mutex; + * callers must handle all synchronization issues themselves. + * + * Typically a driver would call this routine during an URB's completion + * handler, if no more URBs were pending. + * + * This routine can run in atomic context. + */ +void usb_autopm_put_interface_async(struct usb_interface *intf) +{ + struct usb_device *udev = interface_to_usbdev(intf); + int status = 0; + + if (intf->condition == USB_INTERFACE_UNBOUND) { + status = -ENODEV; + } else { + udev->last_busy = jiffies; + --intf->pm_usage_cnt; + if (udev->autosuspend_disabled || udev->autosuspend_delay < 0) + status = -EPERM; + else if (intf->pm_usage_cnt <= 0 && + !timer_pending(&udev->autosuspend.timer)) { + queue_delayed_work(ksuspend_usb_wq, &udev->autosuspend, + round_jiffies_relative( + udev->autosuspend_delay)); + } + } + dev_vdbg(&intf->dev, "%s: status %d cnt %d\n", + __func__, status, intf->pm_usage_cnt); +} +EXPORT_SYMBOL_GPL(usb_autopm_put_interface_async); + /** * usb_autopm_get_interface - increment a USB interface's PM-usage counter * @intf: the usb_interface whose counter should be incremented @@ -1536,6 +1588,37 @@ int usb_autopm_get_interface(struct usb_interface *intf) } EXPORT_SYMBOL_GPL(usb_autopm_get_interface); +/** + * usb_autopm_get_interface_async - increment a USB interface's PM-usage counter + * @intf: the usb_interface whose counter should be incremented + * + * This routine does much the same thing as + * usb_autopm_get_interface(): it increments @intf's usage counter and + * queues an autoresume request if the result is > 0. The differences + * are that it does not acquire the device's pm_mutex (callers must + * handle all synchronization issues themselves), and it does not + * autoresume the device directly (it only queues a request). After a + * successful call, the device will generally not yet be resumed. + * + * This routine can run in atomic context. + */ +int usb_autopm_get_interface_async(struct usb_interface *intf) +{ + struct usb_device *udev = interface_to_usbdev(intf); + int status = 0; + + if (intf->condition == USB_INTERFACE_UNBOUND) + status = -ENODEV; + else if (udev->autoresume_disabled) + status = -EPERM; + else if (++intf->pm_usage_cnt > 0 && udev->state == USB_STATE_SUSPENDED) + queue_work(ksuspend_usb_wq, &udev->autoresume); + dev_vdbg(&intf->dev, "%s: status %d cnt %d\n", + __func__, status, intf->pm_usage_cnt); + return status; +} +EXPORT_SYMBOL_GPL(usb_autopm_get_interface_async); + /** * usb_autopm_set_interface - set a USB interface's autosuspend state * @intf: the usb_interface whose state should be set @@ -1563,6 +1646,9 @@ EXPORT_SYMBOL_GPL(usb_autopm_set_interface); void usb_autosuspend_work(struct work_struct *work) {} +void usb_autoresume_work(struct work_struct *work) +{} + #endif /* CONFIG_USB_SUSPEND */ /** diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c index b19cbfcd51d..95fb3104ba4 100644 --- a/drivers/usb/core/hub.c +++ b/drivers/usb/core/hub.c @@ -1374,8 +1374,9 @@ static void usb_stop_pm(struct usb_device *udev) usb_autosuspend_device(udev->parent); usb_pm_unlock(udev); - /* Stop any autosuspend requests already submitted */ - cancel_rearming_delayed_work(&udev->autosuspend); + /* Stop any autosuspend or autoresume requests already submitted */ + cancel_delayed_work_sync(&udev->autosuspend); + cancel_work_sync(&udev->autoresume); } #else diff --git a/drivers/usb/core/usb.c b/drivers/usb/core/usb.c index 400fa4cc9a3..44f2fc750b6 100644 --- a/drivers/usb/core/usb.c +++ b/drivers/usb/core/usb.c @@ -402,6 +402,7 @@ struct usb_device *usb_alloc_dev(struct usb_device *parent, #ifdef CONFIG_PM mutex_init(&dev->pm_mutex); INIT_DELAYED_WORK(&dev->autosuspend, usb_autosuspend_work); + INIT_WORK(&dev->autoresume, usb_autoresume_work); dev->autosuspend_delay = usb_autosuspend_delay * HZ; dev->connect_time = jiffies; dev->active_duration = -jiffies; diff --git a/drivers/usb/core/usb.h b/drivers/usb/core/usb.h index 9a1a45ac3ad..b60ebb4de1a 100644 --- a/drivers/usb/core/usb.h +++ b/drivers/usb/core/usb.h @@ -45,6 +45,7 @@ extern int usb_suspend(struct device *dev, pm_message_t msg); extern int usb_resume(struct device *dev); extern void usb_autosuspend_work(struct work_struct *work); +extern void usb_autoresume_work(struct work_struct *work); extern int usb_port_suspend(struct usb_device *dev); extern int usb_port_resume(struct usb_device *dev); extern int usb_external_suspend_device(struct usb_device *udev, diff --git a/include/linux/usb.h b/include/linux/usb.h index f72aa51f7bc..859a88e6ce9 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -398,6 +398,7 @@ struct usb_tt; * @urbnum: number of URBs submitted for the whole device * @active_duration: total time device is not suspended * @autosuspend: for delayed autosuspends + * @autoresume: for autoresumes requested while in_interrupt * @pm_mutex: protects PM operations * @last_busy: time of last use * @autosuspend_delay: in jiffies @@ -476,6 +477,7 @@ struct usb_device { #ifdef CONFIG_PM struct delayed_work autosuspend; + struct work_struct autoresume; struct mutex pm_mutex; unsigned long last_busy; @@ -513,6 +515,8 @@ extern struct usb_device *usb_find_device(u16 vendor_id, u16 product_id); extern int usb_autopm_set_interface(struct usb_interface *intf); extern int usb_autopm_get_interface(struct usb_interface *intf); extern void usb_autopm_put_interface(struct usb_interface *intf); +extern int usb_autopm_get_interface_async(struct usb_interface *intf); +extern void usb_autopm_put_interface_async(struct usb_interface *intf); static inline void usb_autopm_enable(struct usb_interface *intf) { @@ -539,8 +543,13 @@ static inline int usb_autopm_set_interface(struct usb_interface *intf) static inline int usb_autopm_get_interface(struct usb_interface *intf) { return 0; } +static inline int usb_autopm_get_interface_async(struct usb_interface *intf) +{ return 0; } + static inline void usb_autopm_put_interface(struct usb_interface *intf) { } +static inline void usb_autopm_put_interface_async(struct usb_interface *intf) +{ } static inline void usb_autopm_enable(struct usb_interface *intf) { } static inline void usb_autopm_disable(struct usb_interface *intf) From dc023dceec861c60bc1d1a17a2c6496ddac26ee7 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Thu, 13 Nov 2008 10:31:35 -0800 Subject: [PATCH 025/123] USB: Introduce usb_queue_reset() to do resets from atomic contexts This patch introduces a new call to be able to do a USB reset from an atomic contect. This is quite helpful in USB callbacks to handle errors (when the only thing that can be done is to do a device reset). It is done queuing a work struct that will do the actual reset. The struct is "attached" to an interface so pending requests from an interface are removed when said interface is unbound from the driver. The call flow then becomes: usb_queue_reset_device() __usb_queue_reset_device() [workqueue] usb_reset_device() usb_probe_interface() usb_cancel_queue_reset() [error path] usb_unbind_interface() usb_cancel_queue_reset() usb_driver_release_interface() usb_cancel_queue_reset() Note usb_cancel_queue_reset() needs smarts to try not to unqueue when it is actually being executed. This happens when we run the reset from the workqueue: usb_reset_device() is called and on interface unbind time, usb_cancel_queue_reset() would be called. That would deadlock on cancel_work_sync(). To avoid that, we set (before running usb_reset_device()) usb_intf->reset_running and clear it inmediately after returning. Patch is against 2.6.28-rc2 and depends on http://marc.info/?l=linux-usb&m=122581634925308&w=2 (as submitted by Alan Stern). Signed-off-by: Inaky Perez-Gonzalez Cc: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/driver.c | 23 ++++++++++++++++++-- drivers/usb/core/hub.c | 43 ++++++++++++++++++++++++++++++++++++++ drivers/usb/core/message.c | 41 ++++++++++++++++++++++++++++++++++++ include/linux/usb.h | 8 +++++++ 4 files changed, 113 insertions(+), 2 deletions(-) diff --git a/drivers/usb/core/driver.c b/drivers/usb/core/driver.c index 23b3c7e79d4..7e26fb3c275 100644 --- a/drivers/usb/core/driver.c +++ b/drivers/usb/core/driver.c @@ -184,6 +184,20 @@ static int usb_unbind_device(struct device *dev) return 0; } +/* + * Cancel any pending scheduled resets + * + * [see usb_queue_reset_device()] + * + * Called after unconfiguring / when releasing interfaces. See + * comments in __usb_queue_reset_device() regarding + * udev->reset_running. + */ +static void usb_cancel_queued_reset(struct usb_interface *iface) +{ + if (iface->reset_running == 0) + cancel_work_sync(&iface->reset_ws); +} /* called from driver core with dev locked */ static int usb_probe_interface(struct device *dev) @@ -242,6 +256,7 @@ static int usb_probe_interface(struct device *dev) mark_quiesced(intf); intf->needs_remote_wakeup = 0; intf->condition = USB_INTERFACE_UNBOUND; + usb_cancel_queued_reset(intf); } else intf->condition = USB_INTERFACE_BOUND; @@ -272,6 +287,7 @@ static int usb_unbind_interface(struct device *dev) usb_disable_interface(udev, intf); driver->disconnect(intf); + usb_cancel_queued_reset(intf); /* Reset other interface state. * We cannot do a Set-Interface if the device is suspended or @@ -380,8 +396,10 @@ void usb_driver_release_interface(struct usb_driver *driver, if (device_is_registered(dev)) { iface->condition = USB_INTERFACE_UNBINDING; device_release_driver(dev); + } else { + iface->condition = USB_INTERFACE_UNBOUND; + usb_cancel_queued_reset(iface); } - dev->driver = NULL; usb_set_intfdata(iface, NULL); @@ -942,7 +960,8 @@ static int usb_suspend_interface(struct usb_device *udev, if (udev->state == USB_STATE_NOTATTACHED || !is_active(intf)) goto done; - if (intf->condition == USB_INTERFACE_UNBOUND) /* This can't happen */ + /* This can happen; see usb_driver_release_interface() */ + if (intf->condition == USB_INTERFACE_UNBOUND) goto done; driver = to_usb_driver(intf->dev.driver); diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c index 95fb3104ba4..e65881899c8 100644 --- a/drivers/usb/core/hub.c +++ b/drivers/usb/core/hub.c @@ -3518,3 +3518,46 @@ int usb_reset_device(struct usb_device *udev) return ret; } EXPORT_SYMBOL_GPL(usb_reset_device); + + +/** + * usb_queue_reset_device - Reset a USB device from an atomic context + * @iface: USB interface belonging to the device to reset + * + * This function can be used to reset a USB device from an atomic + * context, where usb_reset_device() won't work (as it blocks). + * + * Doing a reset via this method is functionally equivalent to calling + * usb_reset_device(), except for the fact that it is delayed to a + * workqueue. This means that any drivers bound to other interfaces + * might be unbound, as well as users from usbfs in user space. + * + * Corner cases: + * + * - Scheduling two resets at the same time from two different drivers + * attached to two different interfaces of the same device is + * possible; depending on how the driver attached to each interface + * handles ->pre_reset(), the second reset might happen or not. + * + * - If a driver is unbound and it had a pending reset, the reset will + * be cancelled. + * + * - This function can be called during .probe() or .disconnect() + * times. On return from .disconnect(), any pending resets will be + * cancelled. + * + * There is no no need to lock/unlock the @reset_ws as schedule_work() + * does its own. + * + * NOTE: We don't do any reference count tracking because it is not + * needed. The lifecycle of the work_struct is tied to the + * usb_interface. Before destroying the interface we cancel the + * work_struct, so the fact that work_struct is queued and or + * running means the interface (and thus, the device) exist and + * are referenced. + */ +void usb_queue_reset_device(struct usb_interface *iface) +{ + schedule_work(&iface->reset_ws); +} +EXPORT_SYMBOL_GPL(usb_queue_reset_device); diff --git a/drivers/usb/core/message.c b/drivers/usb/core/message.c index cc47d36798b..aadf29f09c4 100644 --- a/drivers/usb/core/message.c +++ b/drivers/usb/core/message.c @@ -1441,6 +1441,46 @@ static struct usb_interface_assoc_descriptor *find_iad(struct usb_device *dev, return retval; } + +/* + * Internal function to queue a device reset + * + * This is initialized into the workstruct in 'struct + * usb_device->reset_ws' that is launched by + * message.c:usb_set_configuration() when initializing each 'struct + * usb_interface'. + * + * It is safe to get the USB device without reference counts because + * the life cycle of @iface is bound to the life cycle of @udev. Then, + * this function will be ran only if @iface is alive (and before + * freeing it any scheduled instances of it will have been cancelled). + * + * We need to set a flag (usb_dev->reset_running) because when we call + * the reset, the interfaces might be unbound. The current interface + * cannot try to remove the queued work as it would cause a deadlock + * (you cannot remove your work from within your executing + * workqueue). This flag lets it know, so that + * usb_cancel_queued_reset() doesn't try to do it. + * + * See usb_queue_reset_device() for more details + */ +void __usb_queue_reset_device(struct work_struct *ws) +{ + int rc; + struct usb_interface *iface = + container_of(ws, struct usb_interface, reset_ws); + struct usb_device *udev = interface_to_usbdev(iface); + + rc = usb_lock_device_for_reset(udev, iface); + if (rc >= 0) { + iface->reset_running = 1; + usb_reset_device(udev); + iface->reset_running = 0; + usb_unlock_device(udev); + } +} + + /* * usb_set_configuration - Makes a particular device setting be current * @dev: the device whose configuration is being updated @@ -1611,6 +1651,7 @@ free_interfaces: intf->dev.type = &usb_if_device_type; intf->dev.groups = usb_interface_groups; intf->dev.dma_mask = dev->dev.dma_mask; + INIT_WORK(&intf->reset_ws, __usb_queue_reset_device); device_initialize(&intf->dev); mark_quiesced(intf); dev_set_name(&intf->dev, "%d-%s:%d.%d", diff --git a/include/linux/usb.h b/include/linux/usb.h index 859a88e6ce9..c8e55aa979d 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -120,6 +120,11 @@ enum usb_interface_condition { * to the sysfs representation for that device. * @pm_usage_cnt: PM usage counter for this interface; autosuspend is not * allowed unless the counter is 0. + * @reset_ws: Used for scheduling resets from atomic context. + * @reset_running: set to 1 if the interface is currently running a + * queued reset so that usb_cancel_queued_reset() doesn't try to + * remove from the workqueue when running inside the worker + * thread. See __usb_queue_reset_device(). * * USB device drivers attach to interfaces on a physical device. Each * interface encapsulates a single high level function, such as feeding @@ -168,10 +173,12 @@ struct usb_interface { unsigned needs_remote_wakeup:1; /* driver requires remote wakeup */ unsigned needs_altsetting0:1; /* switch to altsetting 0 is pending */ unsigned needs_binding:1; /* needs delayed unbind/rebind */ + unsigned reset_running:1; struct device dev; /* interface specific device info */ struct device *usb_dev; int pm_usage_cnt; /* usage counter for autosuspend */ + struct work_struct reset_ws; /* for resets in atomic context */ }; #define to_usb_interface(d) container_of(d, struct usb_interface, dev) #define interface_to_usbdev(intf) \ @@ -507,6 +514,7 @@ extern int usb_lock_device_for_reset(struct usb_device *udev, /* USB port reset for device reinitialization */ extern int usb_reset_device(struct usb_device *dev); +extern void usb_queue_reset_device(struct usb_interface *dev); extern struct usb_device *usb_find_device(u16 vendor_id, u16 product_id); From 3a407e7391767898027eb7030c41c78ebc7a785d Mon Sep 17 00:00:00 2001 From: Jean Delvare Date: Thu, 13 Nov 2008 18:57:53 +0100 Subject: [PATCH 026/123] USB: Convert ohci-pnx4008 to a new-style i2c driver The legacy i2c binding model will go away soon, convert ohci-pnx4008 to use the new binding model instead. Signed-off-by: Jean Delvare Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/ohci-pnx4008.c | 85 ++++++++++++++------------------- 1 file changed, 36 insertions(+), 49 deletions(-) diff --git a/drivers/usb/host/ohci-pnx4008.c b/drivers/usb/host/ohci-pnx4008.c index e306ca6aef3..100bf3d8437 100644 --- a/drivers/usb/host/ohci-pnx4008.c +++ b/drivers/usb/host/ohci-pnx4008.c @@ -106,65 +106,34 @@ extern int ocpi_enable(void); static struct clk *usb_clk; -static int isp1301_probe(struct i2c_adapter *adap); -static int isp1301_detach(struct i2c_client *client); - static const unsigned short normal_i2c[] = { ISP1301_I2C_ADDR, ISP1301_I2C_ADDR + 1, I2C_CLIENT_END }; -static const unsigned short dummy_i2c_addrlist[] = { I2C_CLIENT_END }; -static struct i2c_client_address_data addr_data = { - .normal_i2c = normal_i2c, - .probe = dummy_i2c_addrlist, - .ignore = dummy_i2c_addrlist, +static int isp1301_probe(struct i2c_client *client, + const struct i2c_device_id *id) +{ + return 0; +} + +static int isp1301_remove(struct i2c_client *client) +{ + return 0; +} + +const struct i2c_device_id isp1301_id[] = { + { "isp1301_pnx", 0 }, + { } }; struct i2c_driver isp1301_driver = { .driver = { .name = "isp1301_pnx", }, - .attach_adapter = isp1301_probe, - .detach_client = isp1301_detach, + .probe = isp1301_probe, + .remove = isp1301_remove, + .id_table = isp1301_id, }; -static int isp1301_attach(struct i2c_adapter *adap, int addr, int kind) -{ - struct i2c_client *c; - int err; - - c = kzalloc(sizeof(*c), GFP_KERNEL); - if (!c) - return -ENOMEM; - - strlcpy(c->name, "isp1301_pnx", I2C_NAME_SIZE); - c->flags = 0; - c->addr = addr; - c->adapter = adap; - c->driver = &isp1301_driver; - - err = i2c_attach_client(c); - if (err) { - kfree(c); - return err; - } - - isp1301_i2c_client = c; - - return 0; -} - -static int isp1301_probe(struct i2c_adapter *adap) -{ - return i2c_probe(adap, &addr_data, isp1301_attach); -} - -static int isp1301_detach(struct i2c_client *client) -{ - i2c_detach_client(client); - kfree(isp1301_i2c_client); - return 0; -} - static void i2c_write(u8 buf, u8 subaddr) { char tmpbuf[2]; @@ -328,6 +297,8 @@ static int __devinit usb_hcd_pnx4008_probe(struct platform_device *pdev) struct usb_hcd *hcd = 0; struct ohci_hcd *ohci; const struct hc_driver *driver = &ohci_pnx4008_hc_driver; + struct i2c_adapter *i2c_adap; + struct i2c_board_info i2c_info; int ret = 0, irq; @@ -351,9 +322,20 @@ static int __devinit usb_hcd_pnx4008_probe(struct platform_device *pdev) ret = i2c_add_driver(&isp1301_driver); if (ret < 0) { - err("failed to connect I2C to ISP1301 USB Transceiver"); + err("failed to add ISP1301 driver"); goto out; } + i2c_adap = i2c_get_adapter(2); + memset(&i2c_info, 0, sizeof(struct i2c_board_info)); + strlcpy(i2c_info.name, "isp1301_pnx", I2C_NAME_SIZE); + isp1301_i2c_client = i2c_new_probed_device(i2c_adap, &i2c_info, + normal_i2c); + i2c_put_adapter(i2c_adap); + if (!isp1301_i2c_client) { + err("failed to connect I2C to ISP1301 USB Transceiver"); + ret = -ENODEV; + goto out_i2c_driver; + } isp1301_configure(); @@ -429,6 +411,9 @@ out3: out2: clk_put(usb_clk); out1: + i2c_unregister_client(isp1301_i2c_client); + isp1301_i2c_client = NULL; +out_i2c_driver: i2c_del_driver(&isp1301_driver); out: return ret; @@ -445,6 +430,8 @@ static int usb_hcd_pnx4008_remove(struct platform_device *pdev) pnx4008_unset_usb_bits(); clk_disable(usb_clk); clk_put(usb_clk); + i2c_unregister_client(isp1301_i2c_client); + isp1301_i2c_client = NULL; i2c_del_driver(&isp1301_driver); platform_set_drvdata(pdev, NULL); From 0a2e5b9b8a3d205b565dec18d6fe39ef1aed75cc Mon Sep 17 00:00:00 2001 From: Magnus Damm Date: Thu, 13 Nov 2008 15:46:37 +0900 Subject: [PATCH 027/123] USB: m66592 and r8a66597 resource changes Use the more common platform_get_resource() together with index instead of depending on the resource name and platform_get_resource_by_name(). Replace the resource_len() implementation with resource_size(). Signed-off-by: Magnus Damm Acked-by: Yoshihiro Shimoda Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/m66592-udc.c | 9 +++------ drivers/usb/host/r8a66597-hcd.c | 8 +++----- 2 files changed, 6 insertions(+), 11 deletions(-) diff --git a/drivers/usb/gadget/m66592-udc.c b/drivers/usb/gadget/m66592-udc.c index 3a8879ec206..43dcf9e1af6 100644 --- a/drivers/usb/gadget/m66592-udc.c +++ b/drivers/usb/gadget/m66592-udc.c @@ -1546,8 +1546,6 @@ static void nop_completion(struct usb_ep *ep, struct usb_request *r) { } -#define resource_len(r) (((r)->end - (r)->start) + 1) - static int __init m66592_probe(struct platform_device *pdev) { struct resource *res; @@ -1560,11 +1558,10 @@ static int __init m66592_probe(struct platform_device *pdev) int ret = 0; int i; - res = platform_get_resource_byname(pdev, IORESOURCE_MEM, - (char *)udc_name); + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); if (!res) { ret = -ENODEV; - pr_err("platform_get_resource_byname error.\n"); + pr_err("platform_get_resource error.\n"); goto clean_up; } @@ -1575,7 +1572,7 @@ static int __init m66592_probe(struct platform_device *pdev) goto clean_up; } - reg = ioremap(res->start, resource_len(res)); + reg = ioremap(res->start, resource_size(res)); if (reg == NULL) { ret = -ENOMEM; pr_err("ioremap error.\n"); diff --git a/drivers/usb/host/r8a66597-hcd.c b/drivers/usb/host/r8a66597-hcd.c index c21f14e0666..319041205b5 100644 --- a/drivers/usb/host/r8a66597-hcd.c +++ b/drivers/usb/host/r8a66597-hcd.c @@ -2275,7 +2275,6 @@ static int __init_or_module r8a66597_remove(struct platform_device *pdev) return 0; } -#define resource_len(r) (((r)->end - (r)->start) + 1) static int __init r8a66597_probe(struct platform_device *pdev) { #if defined(CONFIG_SUPERH_ON_CHIP_R8A66597) && defined(CONFIG_HAVE_CLK) @@ -2296,11 +2295,10 @@ static int __init r8a66597_probe(struct platform_device *pdev) goto clean_up; } - res = platform_get_resource_byname(pdev, IORESOURCE_MEM, - (char *)hcd_name); + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); if (!res) { ret = -ENODEV; - dev_err(&pdev->dev, "platform_get_resource_byname error.\n"); + dev_err(&pdev->dev, "platform_get_resource error.\n"); goto clean_up; } @@ -2315,7 +2313,7 @@ static int __init r8a66597_probe(struct platform_device *pdev) irq = ires->start; irq_trigger = ires->flags & IRQF_TRIGGER_MASK; - reg = ioremap(res->start, resource_len(res)); + reg = ioremap(res->start, resource_size(res)); if (reg == NULL) { ret = -ENOMEM; dev_err(&pdev->dev, "ioremap error.\n"); From 1c53e6cb80ae0442c5ee02ae29b1dec3d96a497e Mon Sep 17 00:00:00 2001 From: Qinghuang Feng Date: Fri, 14 Nov 2008 20:24:05 +0800 Subject: [PATCH 028/123] USB: gadget: pxa27x_udc.c: cleanup kernel-doc no argument named @index in pio_irq_disable, and no argument named @req in inc_ep_stats_bytes, remove them. Signed-off-by: Qinghuang Feng Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/pxa27x_udc.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/drivers/usb/gadget/pxa27x_udc.c b/drivers/usb/gadget/pxa27x_udc.c index 65110d02a20..990f40f988d 100644 --- a/drivers/usb/gadget/pxa27x_udc.c +++ b/drivers/usb/gadget/pxa27x_udc.c @@ -430,7 +430,6 @@ static void pio_irq_enable(struct pxa_ep *ep) /** * pio_irq_disable - Disables irq generation for one endpoint * @ep: udc endpoint - * @index: endpoint number */ static void pio_irq_disable(struct pxa_ep *ep) { @@ -586,7 +585,6 @@ static void inc_ep_stats_reqs(struct pxa_ep *ep, int is_in) * inc_ep_stats_bytes - Update ep stats counts * @ep: physical endpoint * @count: bytes transfered on endpoint - * @req: usb request * @is_in: ep direction (USB_DIR_IN or 0) */ static void inc_ep_stats_bytes(struct pxa_ep *ep, int count, int is_in) From 5e5b0e5d6a354f6f457368149eb2e9a61e5576cf Mon Sep 17 00:00:00 2001 From: Qinghuang Feng Date: Fri, 14 Nov 2008 20:24:11 +0800 Subject: [PATCH 029/123] USB: serial: ipw.c: mark {__init|__exit} for usb_ipw_{init|exit} mark {__init|__exit} for usb_ipw_{init|exit} () Signed-off-by: Qinghuang Feng Signed-off-by: Greg Kroah-Hartman --- drivers/usb/serial/ipw.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/usb/serial/ipw.c b/drivers/usb/serial/ipw.c index 3ac59a8a980..f530032ed93 100644 --- a/drivers/usb/serial/ipw.c +++ b/drivers/usb/serial/ipw.c @@ -473,7 +473,7 @@ static struct usb_serial_driver ipw_device = { -static int usb_ipw_init(void) +static int __init usb_ipw_init(void) { int retval; @@ -490,7 +490,7 @@ static int usb_ipw_init(void) return 0; } -static void usb_ipw_exit(void) +static void __exit usb_ipw_exit(void) { usb_deregister(&usb_ipw_driver); usb_serial_deregister(&ipw_device); From f150fa1afbf69a87f54752579ff2bb769aad88b3 Mon Sep 17 00:00:00 2001 From: Pete Zaitcev Date: Thu, 13 Nov 2008 21:31:21 -0700 Subject: [PATCH 030/123] USB: Allow usbmon as a module even if usbcore is builtin usbmon can only be built as a module if usbcore is a module too. Trivial changes to the relevant Kconfig and Makefile (and a few trivial changes elsewhere) allow usbmon to be built as a module even if usbcore is builtin. This is verified to work in all 9 permutations (3 correctly prohibited by Kconfig, 6 build a suitable result). Signed-off-by: Paul Bolle Signed-off-by: Pete Zaitcev Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/hcd.c | 4 ++-- drivers/usb/core/hcd.h | 4 ++-- drivers/usb/mon/Kconfig | 13 ++++++------- drivers/usb/mon/Makefile | 3 +-- include/linux/usb.h | 2 +- 5 files changed, 12 insertions(+), 14 deletions(-) diff --git a/drivers/usb/core/hcd.c b/drivers/usb/core/hcd.c index e1b42626d04..7403ed871ab 100644 --- a/drivers/usb/core/hcd.c +++ b/drivers/usb/core/hcd.c @@ -2028,7 +2028,7 @@ EXPORT_SYMBOL_GPL(usb_hcd_platform_shutdown); /*-------------------------------------------------------------------------*/ -#if defined(CONFIG_USB_MON) +#if defined(CONFIG_USB_MON) || defined(CONFIG_USB_MON_MODULE) struct usb_mon_operations *mon_ops; @@ -2064,4 +2064,4 @@ void usb_mon_deregister (void) } EXPORT_SYMBOL_GPL (usb_mon_deregister); -#endif /* CONFIG_USB_MON */ +#endif /* CONFIG_USB_MON || CONFIG_USB_MON_MODULE */ diff --git a/drivers/usb/core/hcd.h b/drivers/usb/core/hcd.h index d202342b36b..0aaa9cea6b3 100644 --- a/drivers/usb/core/hcd.h +++ b/drivers/usb/core/hcd.h @@ -421,7 +421,7 @@ static inline void usbfs_cleanup(void) { } /*-------------------------------------------------------------------------*/ -#if defined(CONFIG_USB_MON) +#if defined(CONFIG_USB_MON) || defined(CONFIG_USB_MON_MODULE) struct usb_mon_operations { void (*urb_submit)(struct usb_bus *bus, struct urb *urb); @@ -463,7 +463,7 @@ static inline void usbmon_urb_submit_error(struct usb_bus *bus, struct urb *urb, static inline void usbmon_urb_complete(struct usb_bus *bus, struct urb *urb, int status) {} -#endif /* CONFIG_USB_MON */ +#endif /* CONFIG_USB_MON || CONFIG_USB_MON_MODULE */ /*-------------------------------------------------------------------------*/ diff --git a/drivers/usb/mon/Kconfig b/drivers/usb/mon/Kconfig index deb9ddffa40..f28f350cd96 100644 --- a/drivers/usb/mon/Kconfig +++ b/drivers/usb/mon/Kconfig @@ -3,14 +3,13 @@ # config USB_MON - bool "USB Monitor" - depends on USB!=n - default y + tristate "USB Monitor" + depends on USB + default y if USB=y + default m if USB=m help - If you say Y here, a component which captures the USB traffic + If you select this option, a component which captures the USB traffic between peripheral-specific drivers and HC drivers will be built. For more information, see . - This is somewhat experimental at this time, but it should be safe. - - If unsure, say Y. + If unsure, say Y (if allowed), otherwise M. diff --git a/drivers/usb/mon/Makefile b/drivers/usb/mon/Makefile index 0f76ed5e161..c6516b56673 100644 --- a/drivers/usb/mon/Makefile +++ b/drivers/usb/mon/Makefile @@ -4,5 +4,4 @@ usbmon-objs := mon_main.o mon_stat.o mon_text.o mon_bin.o mon_dma.o -# This does not use CONFIG_USB_MON because we want this to use a tristate. -obj-$(CONFIG_USB) += usbmon.o +obj-$(CONFIG_USB_MON) += usbmon.o diff --git a/include/linux/usb.h b/include/linux/usb.h index c8e55aa979d..8bc81bffc19 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -336,7 +336,7 @@ struct usb_bus { #endif struct device *dev; /* device for this bus */ -#if defined(CONFIG_USB_MON) +#if defined(CONFIG_USB_MON) || defined(CONFIG_USB_MON_MODULE) struct mon_bus *mon_bus; /* non-null when associated */ int monitored; /* non-zero when monitored */ #endif From 6cd132015d92356b1287f6f3377a3ad955b6cf2f Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Thu, 13 Nov 2008 15:08:30 -0500 Subject: [PATCH 031/123] USB: announce new devices earlier This patch (as1166) changes usb_new_device(). Now new devices will be announced in the log _prior_ to being registered; this way the "new device" lines will appear before all the output from driver probing, which seems much more logical. Also, the patch adds a call to usb_stop_pm() to the failure pathway, so that the parent's count of unsuspended children will remain correct if registration fails. In order for this to work properly, the code to increment that count has to be moved forward, before the first point where a failure can occur. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/hub.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c index e65881899c8..ff066edf4dc 100644 --- a/drivers/usb/core/hub.c +++ b/drivers/usb/core/hub.c @@ -1636,6 +1636,10 @@ int usb_new_device(struct usb_device *udev) { int err; + /* Increment the parent's count of unsuspended children */ + if (udev->parent) + usb_autoresume_device(udev->parent); + usb_detect_quirks(udev); /* Determine quirks */ err = usb_configure_device(udev); /* detect & probe dev/intfs */ if (err < 0) @@ -1644,9 +1648,8 @@ int usb_new_device(struct usb_device *udev) udev->dev.devt = MKDEV(USB_DEVICE_MAJOR, (((udev->bus->busnum-1) * 128) + (udev->devnum-1))); - /* Increment the parent's count of unsuspended children */ - if (udev->parent) - usb_autoresume_device(udev->parent); + /* Tell the world! */ + announce_device(udev); /* Register the device. The device driver is responsible * for adding the device files to sysfs and for configuring @@ -1660,13 +1663,11 @@ int usb_new_device(struct usb_device *udev) /* put device-specific files into sysfs */ usb_create_sysfs_dev_files(udev); - - /* Tell the world! */ - announce_device(udev); return err; fail: usb_set_device_state(udev, USB_STATE_NOTATTACHED); + usb_stop_pm(udev); return err; } From f9dc8f99e5846606b1f475b3905eaf5ae1017c50 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Tue, 18 Nov 2008 14:08:07 -0500 Subject: [PATCH 032/123] usb-storage: clean up unusual_devs.h This patch (as1170) removes some duplicate entries in unusual_devs.h and rearranges a few others to put the list in proper numerical order. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/unusual_devs.h | 96 ++++++++++++------------------ 1 file changed, 39 insertions(+), 57 deletions(-) diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index 8a0eb00024d..0e0ebf275a2 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -188,6 +188,14 @@ UNUSUAL_DEV( 0x0421, 0x006a, 0x0000, 0x0701, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_CAPACITY ), +/* Submitted by Ricky Wong Yung Fei */ +/* Nokia 7610 Supernova - Too many sectors reported in usb storage mode */ +UNUSUAL_DEV( 0x0421, 0x00f5, 0x0000, 0x0470, + "Nokia", + "7610 Supernova", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_FIX_CAPACITY ), + /* Reported by Mario Rettig */ UNUSUAL_DEV( 0x0421, 0x042e, 0x0100, 0x0100, "Nokia", @@ -274,21 +282,6 @@ UNUSUAL_DEV( 0x0421, 0x04fa, 0x0550, 0x0660, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_CAPACITY ), -/* Patch for Nokia 5310 capacity */ -UNUSUAL_DEV( 0x0421, 0x006a, 0x0000, 0x0591, - "Nokia", - "5310", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - -/* Submitted by Ricky Wong Yung Fei */ -/* Nokia 7610 Supernova - Too many sectors reported in usb storage mode */ -UNUSUAL_DEV( 0x0421, 0x00f5, 0x0000, 0x0470, - "Nokia", - "7610 Supernova", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - /* Reported by Olaf Hering from novell bug #105878 */ UNUSUAL_DEV( 0x0424, 0x0fdc, 0x0210, 0x0210, "SMSC", @@ -395,6 +388,15 @@ UNUSUAL_DEV( 0x04a4, 0x0004, 0x0001, 0x0001, "DVD-CAM DZ-MV100A Camcorder", US_SC_SCSI, US_PR_CB, NULL, US_FL_SINGLE_LUN), +/* BENQ DC5330 + * Reported by Manuel Fombuena and + * Frank Copeland */ +UNUSUAL_DEV( 0x04a5, 0x3010, 0x0100, 0x0100, + "Tekom Technologies, Inc", + "300_CAMERA", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_IGNORE_RESIDUE ), + /* Patch for Nikon coolpix 2000 * Submitted by Fabien Cosse */ UNUSUAL_DEV( 0x04b0, 0x0301, 0x0010, 0x0010, @@ -487,15 +489,6 @@ UNUSUAL_DEV( 0x04b3, 0x4001, 0x0110, 0x0110, US_SC_DEVICE, US_PR_CB, NULL, US_FL_MAX_SECTORS_MIN), -/* BENQ DC5330 - * Reported by Manuel Fombuena and - * Frank Copeland */ -UNUSUAL_DEV( 0x04a5, 0x3010, 0x0100, 0x0100, - "Tekom Technologies, Inc", - "300_CAMERA", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_IGNORE_RESIDUE ), - #ifdef CONFIG_USB_STORAGE_CYPRESS_ATACB /* CY7C68300 : support atacb */ UNUSUAL_DEV( 0x04b4, 0x6830, 0x0000, 0x9999, @@ -814,15 +807,15 @@ UNUSUAL_DEV( 0x054c, 0x006d, 0x0000, 0x9999, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_INQUIRY ), -/* Submitted by Mike Alborn */ -UNUSUAL_DEV( 0x054c, 0x016a, 0x0000, 0x9999, +/* Submitted by Frank Engel */ +UNUSUAL_DEV( 0x054c, 0x0099, 0x0000, 0x9999, "Sony", "PEG Mass Storage", US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_INQUIRY ), - -/* Submitted by Frank Engel */ -UNUSUAL_DEV( 0x054c, 0x0099, 0x0000, 0x9999, + +/* Submitted by Mike Alborn */ +UNUSUAL_DEV( 0x054c, 0x016a, 0x0000, 0x9999, "Sony", "PEG Mass Storage", US_SC_DEVICE, US_PR_DEVICE, NULL, @@ -1357,17 +1350,6 @@ UNUSUAL_DEV( 0x0851, 0x1543, 0x0200, 0x0200, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_NOT_LOCKABLE), -/* Andrew Lunn - * PanDigital Digital Picture Frame. Does not like ALLOW_MEDIUM_REMOVAL - * on LUN 4. - * Note: Vend:Prod clash with "Ltd Maxell WS30 Slim Digital Camera" -*/ -UNUSUAL_DEV( 0x0851, 0x1543, 0x0200, 0x0200, - "PanDigital", - "Photo Frame", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_NOT_LOCKABLE), - /* Submitted by Jan De Luyck */ UNUSUAL_DEV( 0x08bd, 0x1100, 0x0000, 0x0000, "CITIZEN", @@ -1493,6 +1475,16 @@ UNUSUAL_DEV( 0x0c0b, 0xa109, 0x0000, 0xffff, US_FL_SINGLE_LUN ), #endif +/* + * Pete Zaitcev , bz#164688. + * The device blatantly ignores LUN and returns 1 in GetMaxLUN. + */ +UNUSUAL_DEV( 0x0c45, 0x1060, 0x0100, 0x0100, + "Unknown", + "Unknown", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_SINGLE_LUN ), + /* Submitted by: Nick Sillik * Needed for OneTouch extension to usb-storage * @@ -1510,16 +1502,6 @@ UNUSUAL_DEV( 0x0c0b, 0xa109, 0x0000, 0xffff, 0), #endif -/* - * Pete Zaitcev , bz#164688. - * The device blatantly ignores LUN and returns 1 in GetMaxLUN. - */ -UNUSUAL_DEV( 0x0c45, 0x1060, 0x0100, 0x0100, - "Unknown", - "Unknown", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_SINGLE_LUN ), - /* Submitted by Joris Struyve */ UNUSUAL_DEV( 0x0d96, 0x410a, 0x0001, 0xffff, "Medion", @@ -1666,13 +1648,6 @@ UNUSUAL_DEV( 0x0fce, 0xe030, 0x0000, 0x0000, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_CAPACITY | US_FL_IGNORE_RESIDUE ), -/* Reported by Ricardo Barberis */ -UNUSUAL_DEV( 0x0fce, 0xe092, 0x0000, 0x0000, - "Sony Ericsson", - "P1i", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_IGNORE_RESIDUE ), - /* Reported by Emmanuel Vasilakis */ UNUSUAL_DEV( 0x0fce, 0xe031, 0x0000, 0x0000, "Sony Ericsson", @@ -1680,6 +1655,13 @@ UNUSUAL_DEV( 0x0fce, 0xe031, 0x0000, 0x0000, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_IGNORE_RESIDUE | US_FL_FIX_CAPACITY ), +/* Reported by Ricardo Barberis */ +UNUSUAL_DEV( 0x0fce, 0xe092, 0x0000, 0x0000, + "Sony Ericsson", + "P1i", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_IGNORE_RESIDUE ), + /* Reported by Kevin Cernekee * Tested on hardware version 1.10. * Entry is needed only for the initializer function override. From a658367dae9dc572480f41817dbe1088a1a049ee Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Tue, 18 Nov 2008 14:08:38 -0500 Subject: [PATCH 033/123] USB: usb-storage: remove us->sensebuf This patch (as1171) removes us->sensebuf, since it isn't used anywhere. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/usb.c | 8 -------- drivers/usb/storage/usb.h | 1 - 2 files changed, 9 deletions(-) diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index eb1a53a3e5c..099e07c6af7 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -474,12 +474,6 @@ static int associate_dev(struct us_data *us, struct usb_interface *intf) US_DEBUGP("I/O buffer allocation failed\n"); return -ENOMEM; } - - us->sensebuf = kmalloc(US_SENSE_SIZE, GFP_KERNEL); - if (!us->sensebuf) { - US_DEBUGP("Sense buffer allocation failed\n"); - return -ENOMEM; - } return 0; } @@ -875,8 +869,6 @@ static void dissociate_dev(struct us_data *us) { US_DEBUGP("-- %s\n", __func__); - kfree(us->sensebuf); - /* Free the device-related DMA-mapped buffers */ if (us->cr) usb_buffer_free(us->pusb_dev, sizeof(*us->cr), us->cr, diff --git a/drivers/usb/storage/usb.h b/drivers/usb/storage/usb.h index a4ad73bd832..e4674fc715e 100644 --- a/drivers/usb/storage/usb.h +++ b/drivers/usb/storage/usb.h @@ -138,7 +138,6 @@ struct us_data { struct usb_ctrlrequest *cr; /* control requests */ struct usb_sg_request current_sg; /* scatter-gather req. */ unsigned char *iobuf; /* I/O buffer */ - unsigned char *sensebuf; /* sense data buffer */ dma_addr_t cr_dma; /* buffer DMA addresses */ dma_addr_t iobuf_dma; struct task_struct *ctl_thread; /* the control thread */ From 1537e0ad944acf3a4c2b311a646d7993b89499f7 Mon Sep 17 00:00:00 2001 From: Ben Efros Date: Tue, 18 Nov 2008 13:31:13 -0800 Subject: [PATCH 034/123] USB: storage devices and SAT Add the SANE SENSE flag to indicate that a device is capable of handling more than 18-bytes of sense data. This functionality is required for USB-ATA bridges implementing SAT. A future patch will actually enable this function for several devices. The logic behind this is that we can detect support for SANE_SENSE in a few ways: 1) ATA PASS THROUGH (12) or (16) execute successfully 2) SPC-3 or higher is in use 3) A previous CHECK CONDITION occurred with sense format 70-73 and had a length greater than 18-bytes total Signed-off-by: Ben Efros Signed-off-by: Matthew Dharm Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/scsiglue.c | 4 ++++ drivers/usb/storage/transport.c | 40 ++++++++++++++++++++++++++++++++- include/linux/usb_usual.h | 5 +++-- 3 files changed, 46 insertions(+), 3 deletions(-) diff --git a/drivers/usb/storage/scsiglue.c b/drivers/usb/storage/scsiglue.c index 09779f6a817..1b35e011a34 100644 --- a/drivers/usb/storage/scsiglue.c +++ b/drivers/usb/storage/scsiglue.c @@ -170,6 +170,10 @@ static int slave_configure(struct scsi_device *sdev) if (us->fflags & US_FL_CAPACITY_HEURISTICS) sdev->guess_capacity = 1; + /* assume SPC3 or latter devices support sense size > 18 */ + if (sdev->scsi_level > SCSI_SPC_2) + us->fflags |= US_FL_SANE_SENSE; + /* Some devices report a SCSI revision level above 2 but are * unable to handle the REPORT LUNS command (for which * support is mandatory at level 3). Since we already have diff --git a/drivers/usb/storage/transport.c b/drivers/usb/storage/transport.c index 2058db41618..f584e72cc68 100644 --- a/drivers/usb/storage/transport.c +++ b/drivers/usb/storage/transport.c @@ -578,6 +578,20 @@ void usb_stor_invoke_transport(struct scsi_cmnd *srb, struct us_data *us) need_auto_sense = 1; } + /* + * Determine if this device is SAT by seeing if the + * command executed successfully. Otherwise we'll have + * to wait for at least one CHECK_CONDITION to determine + * SANE_SENSE support + */ + if ((srb->cmnd[0] == ATA_16 || srb->cmnd[0] == ATA_12) && + result == USB_STOR_TRANSPORT_GOOD && + !(us->fflags & US_FL_SANE_SENSE) && + !(srb->cmnd[2] & 0x20)) { + US_DEBUGP("-- SAT supported, increasing auto-sense\n"); + us->fflags |= US_FL_SANE_SENSE; + } + /* * A short transfer on a command where we don't expect it * is unusual, but it doesn't mean we need to auto-sense. @@ -595,10 +609,15 @@ void usb_stor_invoke_transport(struct scsi_cmnd *srb, struct us_data *us) if (need_auto_sense) { int temp_result; struct scsi_eh_save ses; + int sense_size = US_SENSE_SIZE; + + /* device supports and needs bigger sense buffer */ + if (us->fflags & US_FL_SANE_SENSE) + sense_size = ~0; US_DEBUGP("Issuing auto-REQUEST_SENSE\n"); - scsi_eh_prep_cmnd(srb, &ses, NULL, 0, US_SENSE_SIZE); + scsi_eh_prep_cmnd(srb, &ses, NULL, 0, sense_size); /* FIXME: we must do the protocol translation here */ if (us->subclass == US_SC_RBC || us->subclass == US_SC_SCSI || @@ -632,6 +651,25 @@ void usb_stor_invoke_transport(struct scsi_cmnd *srb, struct us_data *us) return; } + /* If the sense data returned is larger than 18-bytes then we + * assume this device supports requesting more in the future. + * The response code must be 70h through 73h inclusive. + */ + if (srb->sense_buffer[7] > (US_SENSE_SIZE - 8) && + !(us->fflags & US_FL_SANE_SENSE) && + (srb->sense_buffer[0] & 0x7C) == 0x70) { + US_DEBUGP("-- SANE_SENSE support enabled\n"); + us->fflags |= US_FL_SANE_SENSE; + + /* Indicate to the user that we truncated their sense + * because we didn't know it supported larger sense. + */ + US_DEBUGP("-- Sense data truncated to %i from %i\n", + US_SENSE_SIZE, + srb->sense_buffer[7] + 8); + srb->sense_buffer[7] = (US_SENSE_SIZE - 8); + } + US_DEBUGP("-- Result from auto-sense is %d\n", temp_result); US_DEBUGP("-- code: 0x%x, key: 0x%x, ASC: 0x%x, ASCQ: 0x%x\n", srb->sense_buffer[0], diff --git a/include/linux/usb_usual.h b/include/linux/usb_usual.h index d9a3bbe38e6..998e5cbbf29 100644 --- a/include/linux/usb_usual.h +++ b/include/linux/usb_usual.h @@ -52,8 +52,9 @@ US_FLAG(MAX_SECTORS_MIN,0x00002000) \ /* Sets max_sectors to arch min */ \ US_FLAG(BULK_IGNORE_TAG,0x00004000) \ - /* Ignore tag mismatch in bulk operations */ - + /* Ignore tag mismatch in bulk operations */ \ + US_FLAG(SANE_SENSE, 0x00008000) + /* Sane Sense (> 18 bytes) */ #define US_FLAG(name, value) US_FL_##name = value , enum { US_DO_ALL_FLAGS }; From dbe6e0c023578dc7b13932f73991ed92b65f3811 Mon Sep 17 00:00:00 2001 From: Ben Efros Date: Sun, 23 Nov 2008 20:06:38 -0800 Subject: [PATCH 035/123] USB: storage: Flag devices known to support SANE_SENSE Add a few devices known to have support for larger sense buffers. Supporting SANE_SENSE does not necessarily mean SAT-1 or SAT-2 is fully supported. Depends on SANE_SENSE patch [1]. Incorporates the Maxtor and Western Digital devices originally submitted by Matthieu CASTET [2]. [1] https://lists.one-eyed-alien.net/pipermail/usb-storage/2008-November/004181.html [2] http://marc.info/?l=linux-usb&m=121762869915609&w=2 Signed-off-by: Ben Efros Signed-off-by: Matthew Dharm Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/unusual_devs.h | 41 +++++++++++++++++++++++++++--- 1 file changed, 37 insertions(+), 4 deletions(-) diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index 0e0ebf275a2..74580977831 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -85,6 +85,13 @@ UNUSUAL_DEV( 0x03f0, 0x0307, 0x0001, 0x0001, US_SC_8070, US_PR_USBAT, init_usbat_cd, 0), #endif +/* Reported by Ben Efros */ +UNUSUAL_DEV( 0x03f0, 0x070c, 0x0000, 0x0000, + "HP", + "Personal Media Drive", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_SANE_SENSE ), + /* Reported by Grant Grundler * HP r707 camera in "Disk" mode with 2.00.23 or 2.00.24 firmware. */ @@ -1004,6 +1011,13 @@ UNUSUAL_DEV( 0x05e3, 0x0702, 0x0000, 0xffff, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_GO_SLOW | US_FL_MAX_SECTORS_64 ), +/* Reported by Ben Efros */ +UNUSUAL_DEV( 0x05e3, 0x0723, 0x9451, 0x9451, + "Genesys Logic", + "USB to SATA", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_SANE_SENSE ), + /* Reported by Hanno Boeck * Taken from the Lycoris Kernel */ UNUSUAL_DEV( 0x0636, 0x0003, 0x0000, 0x9999, @@ -1452,6 +1466,13 @@ UNUSUAL_DEV( 0x0af0, 0x7401, 0x0000, 0x0000, US_SC_DEVICE, US_PR_DEVICE, NULL, 0 ), +/* Reported by Ben Efros */ +UNUSUAL_DEV( 0x0bc2, 0x3010, 0x0000, 0x0000, + "Seagate", + "FreeAgent Pro", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_SANE_SENSE ), + #ifdef CONFIG_USB_STORAGE_ISD200 UNUSUAL_DEV( 0x0bf6, 0xa001, 0x0100, 0x0110, "ATI", @@ -1475,6 +1496,12 @@ UNUSUAL_DEV( 0x0c0b, 0xa109, 0x0000, 0xffff, US_FL_SINGLE_LUN ), #endif +UNUSUAL_DEV( 0x0d49, 0x7310, 0x0000, 0x9999, + "Maxtor", + "USB to SATA", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_SANE_SENSE), + /* * Pete Zaitcev , bz#164688. * The device blatantly ignores LUN and returns 1 in GetMaxLUN. @@ -1674,6 +1701,12 @@ UNUSUAL_DEV( 0x1019, 0x0c55, 0x0000, 0x0110, US_SC_DEVICE, US_PR_DEVICE, usb_stor_ucr61s2b_init, 0 ), +UNUSUAL_DEV( 0x1058, 0x0704, 0x0000, 0x9999, + "Western Digital", + "External HDD", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_SANE_SENSE), + /* Reported by Fabio Venturi * The device reports a vendor-specific bDeviceClass. */ @@ -2063,10 +2096,10 @@ UNUSUAL_DEV( 0x14cd, 0x6600, 0x0201, 0x0201, * JMicron responds to USN and several other SCSI ioctls with a * residue that causes subsequent I/O requests to fail. */ UNUSUAL_DEV( 0x152d, 0x2329, 0x0100, 0x0100, - "JMicron", - "USB to ATA/ATAPI Bridge", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_IGNORE_RESIDUE ), + "JMicron", + "USB to ATA/ATAPI Bridge", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_IGNORE_RESIDUE | US_FL_SANE_SENSE ), /* Reported by Robert Schedel * Note: this is a 'super top' device like the above 14cd/6600 device */ From 12aae68a203e97a58d3f8237fc389201a4d9282d Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Thu, 20 Nov 2008 14:13:12 -0500 Subject: [PATCH 036/123] USB: g_file_storage: add CD-ROM emulation This patch (as1172) adds the ability to emulate a CD-ROM drive to g_file_storage. The emulation is limited, since it presents as a disc containing a single data track and no audio tracks. Still, it may come in useful on occasion. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/file_storage.c | 176 ++++++++++++++++++++++++++---- 1 file changed, 154 insertions(+), 22 deletions(-) diff --git a/drivers/usb/gadget/file_storage.c b/drivers/usb/gadget/file_storage.c index 2e71368f45b..93933155e81 100644 --- a/drivers/usb/gadget/file_storage.c +++ b/drivers/usb/gadget/file_storage.c @@ -1,7 +1,7 @@ /* * file_storage.c -- File-backed USB Storage Gadget, for USB development * - * Copyright (C) 2003-2007 Alan Stern + * Copyright (C) 2003-2008 Alan Stern * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -38,16 +38,17 @@ /* * The File-backed Storage Gadget acts as a USB Mass Storage device, - * appearing to the host as a disk drive. In addition to providing an - * example of a genuinely useful gadget driver for a USB device, it also - * illustrates a technique of double-buffering for increased throughput. - * Last but not least, it gives an easy way to probe the behavior of the - * Mass Storage drivers in a USB host. + * appearing to the host as a disk drive or as a CD-ROM drive. In addition + * to providing an example of a genuinely useful gadget driver for a USB + * device, it also illustrates a technique of double-buffering for increased + * throughput. Last but not least, it gives an easy way to probe the + * behavior of the Mass Storage drivers in a USB host. * * Backing storage is provided by a regular file or a block device, specified * by the "file" module parameter. Access can be limited to read-only by - * setting the optional "ro" module parameter. The gadget will indicate that - * it has removable media if the optional "removable" module parameter is set. + * setting the optional "ro" module parameter. (For CD-ROM emulation, + * access is always read-only.) The gadget will indicate that it has + * removable media if the optional "removable" module parameter is set. * * The gadget supports the Control-Bulk (CB), Control-Bulk-Interrupt (CBI), * and Bulk-Only (also known as Bulk-Bulk-Bulk or BBB) transports, selected @@ -64,7 +65,12 @@ * The default number of LUNs is taken from the number of "file" elements; * it is 1 if "file" is not given. If "removable" is not set then a backing * file must be specified for each LUN. If it is set, then an unspecified - * or empty backing filename means the LUN's medium is not loaded. + * or empty backing filename means the LUN's medium is not loaded. Ideally + * each LUN would be settable independently as a disk drive or a CD-ROM + * drive, but currently all LUNs have to be the same type. The CD-ROM + * emulation includes a single data track and no audio tracks; hence there + * need be only one backing file per LUN. Note also that the CD-ROM block + * length is set to 512 rather than the more common value 2048. * * Requirements are modest; only a bulk-in and a bulk-out endpoint are * needed (an interrupt-out endpoint is also needed for CBI). The memory @@ -91,6 +97,8 @@ * USB device controller (usually true), * boolean to permit the driver to halt * bulk endpoints + * cdrom Default false, boolean for whether to emulate + * a CD-ROM drive * transport=XXX Default BBB, transport name (CB, CBI, or BBB) * protocol=YYY Default SCSI, protocol name (RBC, 8020 or * ATAPI, QIC, UFI, 8070, or SCSI; @@ -103,15 +111,16 @@ * PAGE_CACHE_SIZE) * * If CONFIG_USB_FILE_STORAGE_TEST is not set, only the "file", "ro", - * "removable", "luns", and "stall" options are available; default values - * are used for everything else. + * "removable", "luns", "stall", and "cdrom" options are available; default + * values are used for everything else. * * The pathnames of the backing files and the ro settings are available in * the attribute files "file" and "ro" in the lun subdirectory of the * gadget's sysfs directory. If the "removable" option is set, writing to * these files will simulate ejecting/loading the medium (writing an empty * line means eject) and adjusting a write-enable tab. Changes to the ro - * setting are not allowed when the medium is loaded. + * setting are not allowed when the medium is loaded or if CD-ROM emulation + * is being used. * * This gadget driver is heavily based on "Gadget Zero" by David Brownell. * The driver's SCSI command interface was based on the "Information @@ -261,7 +270,7 @@ #define DRIVER_DESC "File-backed Storage Gadget" #define DRIVER_NAME "g_file_storage" -#define DRIVER_VERSION "7 August 2007" +#define DRIVER_VERSION "20 November 2008" static const char longname[] = DRIVER_DESC; static const char shortname[] = DRIVER_NAME; @@ -341,6 +350,7 @@ static struct { int removable; int can_stall; + int cdrom; char *transport_parm; char *protocol_parm; @@ -359,6 +369,7 @@ static struct { .protocol_parm = "SCSI", .removable = 0, .can_stall = 1, + .cdrom = 0, .vendor = DRIVER_VENDOR_ID, .product = DRIVER_PRODUCT_ID, .release = 0xffff, // Use controller chip type @@ -382,6 +393,9 @@ MODULE_PARM_DESC(removable, "true to simulate removable media"); module_param_named(stall, mod_data.can_stall, bool, S_IRUGO); MODULE_PARM_DESC(stall, "false to prevent bulk stalls"); +module_param_named(cdrom, mod_data.cdrom, bool, S_IRUGO); +MODULE_PARM_DESC(cdrom, "true to emulate cdrom instead of disk"); + /* In the non-TEST version, only the module parameters listed above * are available. */ @@ -411,6 +425,10 @@ MODULE_PARM_DESC(buflen, "I/O buffer size"); /*-------------------------------------------------------------------------*/ +/* SCSI device types */ +#define TYPE_DISK 0x00 +#define TYPE_CDROM 0x05 + /* USB protocol value = the transport method */ #define USB_PR_CBI 0x00 // Control/Bulk/Interrupt #define USB_PR_CB 0x01 // Control/Bulk w/o interrupt @@ -487,6 +505,8 @@ struct interrupt_data { #define SC_READ_12 0xa8 #define SC_READ_CAPACITY 0x25 #define SC_READ_FORMAT_CAPACITIES 0x23 +#define SC_READ_HEADER 0x44 +#define SC_READ_TOC 0x43 #define SC_RELEASE 0x17 #define SC_REQUEST_SENSE 0x03 #define SC_RESERVE 0x16 @@ -2006,23 +2026,28 @@ static int do_inquiry(struct fsg_dev *fsg, struct fsg_buffhd *bh) u8 *buf = (u8 *) bh->buf; static char vendor_id[] = "Linux "; - static char product_id[] = "File-Stor Gadget"; + static char product_disk_id[] = "File-Stor Gadget"; + static char product_cdrom_id[] = "File-CD Gadget "; if (!fsg->curlun) { // Unsupported LUNs are okay fsg->bad_lun_okay = 1; memset(buf, 0, 36); buf[0] = 0x7f; // Unsupported, no device-type + buf[4] = 31; // Additional length return 36; } - memset(buf, 0, 8); // Non-removable, direct-access device + memset(buf, 0, 8); + buf[0] = (mod_data.cdrom ? TYPE_CDROM : TYPE_DISK); if (mod_data.removable) buf[1] = 0x80; buf[2] = 2; // ANSI SCSI level 2 buf[3] = 2; // SCSI-2 INQUIRY data format buf[4] = 31; // Additional length // No special options - sprintf(buf + 8, "%-8s%-16s%04x", vendor_id, product_id, + sprintf(buf + 8, "%-8s%-16s%04x", vendor_id, + (mod_data.cdrom ? product_cdrom_id : + product_disk_id), mod_data.release); return 36; } @@ -2101,6 +2126,75 @@ static int do_read_capacity(struct fsg_dev *fsg, struct fsg_buffhd *bh) } +static void store_cdrom_address(u8 *dest, int msf, u32 addr) +{ + if (msf) { + /* Convert to Minutes-Seconds-Frames */ + addr >>= 2; /* Convert to 2048-byte frames */ + addr += 2*75; /* Lead-in occupies 2 seconds */ + dest[3] = addr % 75; /* Frames */ + addr /= 75; + dest[2] = addr % 60; /* Seconds */ + addr /= 60; + dest[1] = addr; /* Minutes */ + dest[0] = 0; /* Reserved */ + } else { + /* Absolute sector */ + put_be32(dest, addr); + } +} + +static int do_read_header(struct fsg_dev *fsg, struct fsg_buffhd *bh) +{ + struct lun *curlun = fsg->curlun; + int msf = fsg->cmnd[1] & 0x02; + u32 lba = get_be32(&fsg->cmnd[2]); + u8 *buf = (u8 *) bh->buf; + + if ((fsg->cmnd[1] & ~0x02) != 0) { /* Mask away MSF */ + curlun->sense_data = SS_INVALID_FIELD_IN_CDB; + return -EINVAL; + } + if (lba >= curlun->num_sectors) { + curlun->sense_data = SS_LOGICAL_BLOCK_ADDRESS_OUT_OF_RANGE; + return -EINVAL; + } + + memset(buf, 0, 8); + buf[0] = 0x01; /* 2048 bytes of user data, rest is EC */ + store_cdrom_address(&buf[4], msf, lba); + return 8; +} + + +static int do_read_toc(struct fsg_dev *fsg, struct fsg_buffhd *bh) +{ + struct lun *curlun = fsg->curlun; + int msf = fsg->cmnd[1] & 0x02; + int start_track = fsg->cmnd[6]; + u8 *buf = (u8 *) bh->buf; + + if ((fsg->cmnd[1] & ~0x02) != 0 || /* Mask away MSF */ + start_track > 1) { + curlun->sense_data = SS_INVALID_FIELD_IN_CDB; + return -EINVAL; + } + + memset(buf, 0, 20); + buf[1] = (20-2); /* TOC data length */ + buf[2] = 1; /* First track number */ + buf[3] = 1; /* Last track number */ + buf[5] = 0x16; /* Data track, copying allowed */ + buf[6] = 0x01; /* Only track is number 1 */ + store_cdrom_address(&buf[8], msf, 0); + + buf[13] = 0x16; /* Lead-out track is data */ + buf[14] = 0xAA; /* Lead-out track number */ + store_cdrom_address(&buf[16], msf, curlun->num_sectors); + return 20; +} + + static int do_mode_sense(struct fsg_dev *fsg, struct fsg_buffhd *bh) { struct lun *curlun = fsg->curlun; @@ -2848,6 +2942,26 @@ static int do_scsi_command(struct fsg_dev *fsg) reply = do_read_capacity(fsg, bh); break; + case SC_READ_HEADER: + if (!mod_data.cdrom) + goto unknown_cmnd; + fsg->data_size_from_cmnd = get_be16(&fsg->cmnd[7]); + if ((reply = check_command(fsg, 10, DATA_DIR_TO_HOST, + (3<<7) | (0x1f<<1), 1, + "READ HEADER")) == 0) + reply = do_read_header(fsg, bh); + break; + + case SC_READ_TOC: + if (!mod_data.cdrom) + goto unknown_cmnd; + fsg->data_size_from_cmnd = get_be16(&fsg->cmnd[7]); + if ((reply = check_command(fsg, 10, DATA_DIR_TO_HOST, + (7<<6) | (1<<1), 1, + "READ TOC")) == 0) + reply = do_read_toc(fsg, bh); + break; + case SC_READ_FORMAT_CAPACITIES: fsg->data_size_from_cmnd = get_be16(&fsg->cmnd[7]); if ((reply = check_command(fsg, 10, DATA_DIR_TO_HOST, @@ -2933,6 +3047,7 @@ static int do_scsi_command(struct fsg_dev *fsg) // Fall through default: + unknown_cmnd: fsg->data_size_from_cmnd = 0; sprintf(unknown, "Unknown x%02x", fsg->cmnd[0]); if ((reply = check_command(fsg, fsg->cmnd_size, @@ -3498,6 +3613,7 @@ static int open_backing_file(struct lun *curlun, const char *filename) struct inode *inode = NULL; loff_t size; loff_t num_sectors; + loff_t min_sectors; /* R/W if we can, R/O if we must */ ro = curlun->ro; @@ -3541,8 +3657,19 @@ static int open_backing_file(struct lun *curlun, const char *filename) rc = (int) size; goto out; } - num_sectors = size >> 9; // File size in 512-byte sectors - if (num_sectors == 0) { + num_sectors = size >> 9; // File size in 512-byte blocks + min_sectors = 1; + if (mod_data.cdrom) { + num_sectors &= ~3; // Reduce to a multiple of 2048 + min_sectors = 300*4; // Smallest track is 300 frames + if (num_sectors >= 256*60*75*4) { + num_sectors = (256*60*75 - 1) * 4; + LINFO(curlun, "file too big: %s\n", filename); + LINFO(curlun, "using only first %d blocks\n", + (int) num_sectors); + } + } + if (num_sectors < min_sectors) { LINFO(curlun, "file too small: %s\n", filename); rc = -ETOOSMALL; goto out; @@ -3845,9 +3972,12 @@ static int __init fsg_bind(struct usb_gadget *gadget) goto out; if (mod_data.removable) { // Enable the store_xxx attributes - dev_attr_ro.attr.mode = dev_attr_file.attr.mode = 0644; - dev_attr_ro.store = store_ro; + dev_attr_file.attr.mode = 0644; dev_attr_file.store = store_file; + if (!mod_data.cdrom) { + dev_attr_ro.attr.mode = 0644; + dev_attr_ro.store = store_ro; + } } /* Find out how many LUNs there should be */ @@ -3872,6 +4002,8 @@ static int __init fsg_bind(struct usb_gadget *gadget) for (i = 0; i < fsg->nluns; ++i) { curlun = &fsg->luns[i]; curlun->ro = mod_data.ro[i]; + if (mod_data.cdrom) + curlun->ro = 1; curlun->dev.release = lun_release; curlun->dev.parent = &gadget->dev; curlun->dev.driver = &fsg_driver.driver; @@ -4031,9 +4163,9 @@ static int __init fsg_bind(struct usb_gadget *gadget) mod_data.protocol_name, mod_data.protocol_type); DBG(fsg, "VendorID=x%04x, ProductID=x%04x, Release=x%04x\n", mod_data.vendor, mod_data.product, mod_data.release); - DBG(fsg, "removable=%d, stall=%d, buflen=%u\n", + DBG(fsg, "removable=%d, stall=%d, cdrom=%d, buflen=%u\n", mod_data.removable, mod_data.can_stall, - mod_data.buflen); + mod_data.cdrom, mod_data.buflen); DBG(fsg, "I/O thread pid: %d\n", task_pid_nr(fsg->thread_task)); set_bit(REGISTERED, &fsg->atomic_bitflags); From 64648a9dc4d7ac0189364188207310ec6bc75bbe Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Thu, 20 Nov 2008 14:20:03 -0500 Subject: [PATCH 037/123] USB: usb-storage: merge CB and CBI transport routines This patch (as1173) merges usb-storage's CB and CBI transports into a single routine. So much of their code is common, it's silly to keep them separate. Signed-off-by: Alan Stern CC: Matthew Dharm Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/transport.c | 61 +++++---------------------------- drivers/usb/storage/transport.h | 2 -- drivers/usb/storage/usb.c | 2 +- 3 files changed, 10 insertions(+), 55 deletions(-) diff --git a/drivers/usb/storage/transport.c b/drivers/usb/storage/transport.c index f584e72cc68..9cc30afd6d3 100644 --- a/drivers/usb/storage/transport.c +++ b/drivers/usb/storage/transport.c @@ -756,10 +756,10 @@ void usb_stor_stop_transport(struct us_data *us) } /* - * Control/Bulk/Interrupt transport + * Control/Bulk and Control/Bulk/Interrupt transport */ -int usb_stor_CBI_transport(struct scsi_cmnd *srb, struct us_data *us) +int usb_stor_CB_transport(struct scsi_cmnd *srb, struct us_data *us) { unsigned int transfer_length = scsi_bufflen(srb); unsigned int pipe = 0; @@ -801,6 +801,13 @@ int usb_stor_CBI_transport(struct scsi_cmnd *srb, struct us_data *us) } /* STATUS STAGE */ + + /* NOTE: CB does not have a status stage. Silly, I know. So + * we have to catch this at a higher level. + */ + if (us->protocol != US_PR_CBI) + return USB_STOR_TRANSPORT_GOOD; + result = usb_stor_intr_transfer(us, us->iobuf, 2); US_DEBUGP("Got interrupt data (0x%x, 0x%x)\n", us->iobuf[0], us->iobuf[1]); @@ -854,56 +861,6 @@ int usb_stor_CBI_transport(struct scsi_cmnd *srb, struct us_data *us) return USB_STOR_TRANSPORT_FAILED; } -/* - * Control/Bulk transport - */ -int usb_stor_CB_transport(struct scsi_cmnd *srb, struct us_data *us) -{ - unsigned int transfer_length = scsi_bufflen(srb); - int result; - - /* COMMAND STAGE */ - /* let's send the command via the control pipe */ - result = usb_stor_ctrl_transfer(us, us->send_ctrl_pipe, - US_CBI_ADSC, - USB_TYPE_CLASS | USB_RECIP_INTERFACE, 0, - us->ifnum, srb->cmnd, srb->cmd_len); - - /* check the return code for the command */ - US_DEBUGP("Call to usb_stor_ctrl_transfer() returned %d\n", result); - - /* if we stalled the command, it means command failed */ - if (result == USB_STOR_XFER_STALLED) { - return USB_STOR_TRANSPORT_FAILED; - } - - /* Uh oh... serious problem here */ - if (result != USB_STOR_XFER_GOOD) { - return USB_STOR_TRANSPORT_ERROR; - } - - /* DATA STAGE */ - /* transfer the data payload for this command, if one exists*/ - if (transfer_length) { - unsigned int pipe = srb->sc_data_direction == DMA_FROM_DEVICE ? - us->recv_bulk_pipe : us->send_bulk_pipe; - result = usb_stor_bulk_srb(us, pipe, srb); - US_DEBUGP("CB data stage result is 0x%x\n", result); - - /* if we stalled the data transfer it means command failed */ - if (result == USB_STOR_XFER_STALLED) - return USB_STOR_TRANSPORT_FAILED; - if (result > USB_STOR_XFER_STALLED) - return USB_STOR_TRANSPORT_ERROR; - } - - /* STATUS STAGE */ - /* NOTE: CB does not have a status stage. Silly, I know. So - * we have to catch this at a higher level. - */ - return USB_STOR_TRANSPORT_GOOD; -} - /* * Bulk only transport */ diff --git a/drivers/usb/storage/transport.h b/drivers/usb/storage/transport.h index e70b88182f0..242ff5e791a 100644 --- a/drivers/usb/storage/transport.h +++ b/drivers/usb/storage/transport.h @@ -113,8 +113,6 @@ struct bulk_cs_wrap { #define US_CBI_ADSC 0 -extern int usb_stor_CBI_transport(struct scsi_cmnd *, struct us_data*); - extern int usb_stor_CB_transport(struct scsi_cmnd *, struct us_data*); extern int usb_stor_CB_reset(struct us_data*); diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index 099e07c6af7..cdd009fae3c 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -591,7 +591,7 @@ static int get_transport(struct us_data *us) case US_PR_CBI: us->transport_name = "Control/Bulk/Interrupt"; - us->transport = usb_stor_CBI_transport; + us->transport = usb_stor_CB_transport; us->transport_reset = usb_stor_CB_reset; us->max_lun = 7; break; From 3dae5345311271fe598a61bd01f563fc835b4217 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Thu, 20 Nov 2008 14:22:18 -0500 Subject: [PATCH 038/123] USB: usb-storage: merge ATAPI and QIC-157 protocol routines This patch (as1174) merges usb-storage's QIC-157 and ATAPI protocol routines. Since the two functions are identical, there's no reason to keep them separate. Signed-off-by: Alan Stern Cc: Matthew Dharm Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/protocol.c | 24 ++---------------------- drivers/usb/storage/protocol.h | 3 +-- drivers/usb/storage/usb.c | 6 +++--- 3 files changed, 6 insertions(+), 27 deletions(-) diff --git a/drivers/usb/storage/protocol.c b/drivers/usb/storage/protocol.c index 3b3357e20ea..be441d84bc6 100644 --- a/drivers/usb/storage/protocol.c +++ b/drivers/usb/storage/protocol.c @@ -56,9 +56,9 @@ * Protocol routines ***********************************************************************/ -void usb_stor_qic157_command(struct scsi_cmnd *srb, struct us_data *us) +void usb_stor_pad12_command(struct scsi_cmnd *srb, struct us_data *us) { - /* Pad the ATAPI command with zeros + /* Pad the SCSI command with zeros out to 12 bytes * * NOTE: This only works because a scsi_cmnd struct field contains * a unsigned char cmnd[16], so we know we have storage available @@ -73,26 +73,6 @@ void usb_stor_qic157_command(struct scsi_cmnd *srb, struct us_data *us) usb_stor_invoke_transport(srb, us); } -void usb_stor_ATAPI_command(struct scsi_cmnd *srb, struct us_data *us) -{ - /* Pad the ATAPI command with zeros - * - * NOTE: This only works because a scsi_cmnd struct field contains - * a unsigned char cmnd[16], so we know we have storage available - */ - - /* Pad the ATAPI command with zeros */ - for (; srb->cmd_len<12; srb->cmd_len++) - srb->cmnd[srb->cmd_len] = 0; - - /* set command length to 12 bytes */ - srb->cmd_len = 12; - - /* send the command to the transport layer */ - usb_stor_invoke_transport(srb, us); -} - - void usb_stor_ufi_command(struct scsi_cmnd *srb, struct us_data *us) { /* fix some commands -- this is a form of mode translation diff --git a/drivers/usb/storage/protocol.h b/drivers/usb/storage/protocol.h index 487056ffb51..ffc3e2af015 100644 --- a/drivers/usb/storage/protocol.h +++ b/drivers/usb/storage/protocol.h @@ -40,8 +40,7 @@ #define _PROTOCOL_H_ /* Protocol handling routines */ -extern void usb_stor_ATAPI_command(struct scsi_cmnd*, struct us_data*); -extern void usb_stor_qic157_command(struct scsi_cmnd*, struct us_data*); +extern void usb_stor_pad12_command(struct scsi_cmnd*, struct us_data*); extern void usb_stor_ufi_command(struct scsi_cmnd*, struct us_data*); extern void usb_stor_transparent_scsi_command(struct scsi_cmnd*, struct us_data*); diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index cdd009fae3c..06c735703f4 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -704,19 +704,19 @@ static int get_protocol(struct us_data *us) case US_SC_8020: us->protocol_name = "8020i"; - us->proto_handler = usb_stor_ATAPI_command; + us->proto_handler = usb_stor_pad12_command; us->max_lun = 0; break; case US_SC_QIC: us->protocol_name = "QIC-157"; - us->proto_handler = usb_stor_qic157_command; + us->proto_handler = usb_stor_pad12_command; us->max_lun = 0; break; case US_SC_8070: us->protocol_name = "8070i"; - us->proto_handler = usb_stor_ATAPI_command; + us->proto_handler = usb_stor_pad12_command; us->max_lun = 0; break; From 96983d2d861bf94b7f70bc47ac3c5b289f519a2d Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Fri, 21 Nov 2008 11:46:17 -0500 Subject: [PATCH 039/123] USB: storage: set bounce limit for non-DMA-capable host controllers This patch (as1175) makes usb-storage set a SCSI device's request-queue bounce limit such that all buffers will be located in addressable memory (i.e., not in high memory) if the host controller's dma_mask is NULL. This is necessary when the host controller doesn't support DMA: If a buffer is in high memory then the both the virtual and DMA addresses produced by the scatter-gather library will be NULL, preventing the HCD from accessing the buffer's data. In particular, the isp1760 driver needs this when used on a system with more than 1 GB of memory. Signed-off-by: Alan Stern CC: Matthew Dharm Acked-by: Jens Axboe Tested-by: Thomas Hommel Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/scsiglue.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/drivers/usb/storage/scsiglue.c b/drivers/usb/storage/scsiglue.c index 1b35e011a34..e9d6c196a7a 100644 --- a/drivers/usb/storage/scsiglue.c +++ b/drivers/usb/storage/scsiglue.c @@ -129,6 +129,14 @@ static int slave_configure(struct scsi_device *sdev) max_sectors); } + /* Some USB host controllers can't do DMA; they have to use PIO. + * They indicate this by setting their dma_mask to NULL. For + * such controllers we need to make sure the block layer sets + * up bounce buffers in addressable memory. + */ + if (!us->pusb_dev->bus->controller->dma_mask) + blk_queue_bounce_limit(sdev->request_queue, BLK_BOUNCE_HIGH); + /* We can't put these settings in slave_alloc() because that gets * called before the device type is known. Consequently these * settings can't be overridden via the scsi devinfo mechanism. */ From 785895ff1fedddd09576d2c600d90404fef6506c Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Sat, 22 Nov 2008 13:01:06 +1030 Subject: [PATCH 040/123] USB: Don't use __module_param_call; use core_param. Impact: cleanup Found this when I changed args to __module_param_call. We now have core_param for exactly this, but Greg assures me "nousb" is used as a module parameter, so we need the #ifdef MODULE. Signed-off-by: Rusty Russell Cc: Pete Zaitcev Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/usb.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/usb/core/usb.c b/drivers/usb/core/usb.c index 44f2fc750b6..51854c2bc91 100644 --- a/drivers/usb/core/usb.c +++ b/drivers/usb/core/usb.c @@ -954,8 +954,12 @@ void usb_buffer_unmap_sg(const struct usb_device *dev, int is_in, } EXPORT_SYMBOL_GPL(usb_buffer_unmap_sg); -/* format to disable USB on kernel command line is: nousb */ -__module_param_call("", nousb, param_set_bool, param_get_bool, &nousb, 0444); +/* To disable USB, kernel command line is 'nousb' not 'usbcore.nousb' */ +#ifdef MODULE +module_param(nousb, bool, 0444); +#else +core_param(nousb, nousb, bool, 0444); +#endif /* * for external read access to From bb9496c6f7e853e5d4edd5397c9d45f1968d623c Mon Sep 17 00:00:00 2001 From: Julia Lawall Date: Tue, 25 Nov 2008 14:15:19 +0100 Subject: [PATCH 041/123] USB: gadget: change simple_strtol to simple_strtoul Since num is unsigned, it would seem better to use simple_strtoul that simple_strtol. A simplified version of the semantic patch that makes this change is as follows: (http://www.emn.fr/x-info/coccinelle/) // @r2@ long e; position p; @@ e = simple_strtol@p(...) @@ position p != r2.p; type T; T e; @@ e = - simple_strtol@p + simple_strtoul (...) // Signed-off-by: Julia Lawall Acked-by: David Brownell Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/epautoconf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/gadget/epautoconf.c b/drivers/usb/gadget/epautoconf.c index 9462e30192d..a36b1175b18 100644 --- a/drivers/usb/gadget/epautoconf.c +++ b/drivers/usb/gadget/epautoconf.c @@ -161,7 +161,7 @@ ep_matches ( /* report address */ desc->bEndpointAddress &= USB_DIR_IN; if (isdigit (ep->name [2])) { - u8 num = simple_strtol (&ep->name [2], NULL, 10); + u8 num = simple_strtoul (&ep->name [2], NULL, 10); desc->bEndpointAddress |= num; #ifdef MANY_ENDPOINTS } else if (desc->bEndpointAddress & USB_DIR_IN) { From aa69a8093ff985873cb44fe1157bd6db29a20fe4 Mon Sep 17 00:00:00 2001 From: David Lopo Date: Mon, 17 Nov 2008 14:14:51 -0800 Subject: [PATCH 042/123] USB: gadget: MIPS ci13xxx_udc MIPS USB IP core family device controller Currently it only supports IP part number CI13412. [dbrownell@users.sourceforge.net: minor comment tweaks] Signed-off-by: David Lopo Signed-off-by: David Brownell Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/Kconfig | 18 + drivers/usb/gadget/Makefile | 1 + drivers/usb/gadget/ci13xxx_udc.c | 2830 +++++++++++++++++++++++++++++ drivers/usb/gadget/ci13xxx_udc.h | 195 ++ drivers/usb/gadget/gadget_chips.h | 7 + 5 files changed, 3051 insertions(+) create mode 100644 drivers/usb/gadget/ci13xxx_udc.c create mode 100644 drivers/usb/gadget/ci13xxx_udc.h diff --git a/drivers/usb/gadget/Kconfig b/drivers/usb/gadget/Kconfig index dd4cd5a5137..cf3a662c333 100644 --- a/drivers/usb/gadget/Kconfig +++ b/drivers/usb/gadget/Kconfig @@ -377,6 +377,24 @@ config USB_FSL_QE default USB_GADGET select USB_GADGET_SELECTED +config USB_GADGET_CI13XXX + boolean "MIPS USB CI13xxx" + depends on PCI + select USB_GADGET_DUALSPEED + help + MIPS USB IP core family device controller + Currently it only supports IP part number CI13412 + + Say "y" to link the driver statically, or "m" to build a + dynamically linked module called "ci13xxx_udc" and force all + gadget drivers to also be dynamically linked. + +config USB_CI13XXX + tristate + depends on USB_GADGET_CI13XXX + default USB_GADGET + select USB_GADGET_SELECTED + config USB_GADGET_NET2280 boolean "NetChip 228x" depends on PCI diff --git a/drivers/usb/gadget/Makefile b/drivers/usb/gadget/Makefile index bd4041b47dc..5d64a2ab188 100644 --- a/drivers/usb/gadget/Makefile +++ b/drivers/usb/gadget/Makefile @@ -19,6 +19,7 @@ obj-$(CONFIG_USB_ATMEL_USBA) += atmel_usba_udc.o obj-$(CONFIG_USB_FSL_USB2) += fsl_usb2_udc.o obj-$(CONFIG_USB_M66592) += m66592-udc.o obj-$(CONFIG_USB_FSL_QE) += fsl_qe_udc.o +obj-$(CONFIG_USB_CI13XXX) += ci13xxx_udc.o # # USB gadget drivers diff --git a/drivers/usb/gadget/ci13xxx_udc.c b/drivers/usb/gadget/ci13xxx_udc.c new file mode 100644 index 00000000000..bebf911c7e5 --- /dev/null +++ b/drivers/usb/gadget/ci13xxx_udc.c @@ -0,0 +1,2830 @@ +/* + * ci13xxx_udc.c - MIPS USB IP core family device controller + * + * Copyright (C) 2008 Chipidea - MIPS Technologies, Inc. All rights reserved. + * + * Author: David Lopo + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* + * Description: MIPS USB IP core family device controller + * Currently it only supports IP part number CI13412 + * + * This driver is composed of several blocks: + * - HW: hardware interface + * - DBG: debug facilities (optional) + * - UTIL: utilities + * - ISR: interrupts handling + * - ENDPT: endpoint operations (Gadget API) + * - GADGET: gadget operations (Gadget API) + * - BUS: bus glue code, bus abstraction layer + * - PCI: PCI core interface and PCI resources (interrupts, memory...) + * + * Compile Options + * - CONFIG_USB_GADGET_DEBUG_FILES: enable debug facilities + * - STALL_IN: non-empty bulk-in pipes cannot be halted + * if defined mass storage compliance succeeds but with warnings + * => case 4: Hi > Dn + * => case 5: Hi > Di + * => case 8: Hi <> Do + * if undefined usbtest 13 fails + * - TRACE: enable function tracing (depends on DEBUG) + * + * Main Features + * - Chapter 9 & Mass Storage Compliance with Gadget File Storage + * - Chapter 9 Compliance with Gadget Zero (STALL_IN undefined) + * - Normal & LPM support + * + * USBTEST Report + * - OK: 0-12, 13 (STALL_IN defined) & 14 + * - Not Supported: 15 & 16 (ISO) + * + * TODO List + * - OTG + * - Isochronous & Interrupt Traffic + * - Handle requests which spawns into several TDs + * - GET_STATUS(device) - always reports 0 + * - Gadget API (majority of optional features) + * - Suspend & Remote Wakeup + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "ci13xxx_udc.h" + + +/****************************************************************************** + * DEFINE + *****************************************************************************/ +/* ctrl register bank access */ +static DEFINE_SPINLOCK(udc_lock); + +/* driver name */ +#define UDC_DRIVER_NAME "ci13xxx_udc" + +/* control endpoint description */ +static const struct usb_endpoint_descriptor +ctrl_endpt_desc = { + .bLength = USB_DT_ENDPOINT_SIZE, + .bDescriptorType = USB_DT_ENDPOINT, + + .bmAttributes = USB_ENDPOINT_XFER_CONTROL, + .wMaxPacketSize = cpu_to_le16(CTRL_PAYLOAD_MAX), +}; + +/* UDC descriptor */ +static struct ci13xxx *_udc; + +/* Interrupt statistics */ +#define ISR_MASK 0x1F +static struct { + u32 test; + u32 ui; + u32 uei; + u32 pci; + u32 uri; + u32 sli; + u32 none; + struct { + u32 cnt; + u32 buf[ISR_MASK+1]; + u32 idx; + } hndl; +} isr_statistics; + +/** + * ffs_nr: find first (least significant) bit set + * @x: the word to search + * + * This function returns bit number (instead of position) + */ +static int ffs_nr(u32 x) +{ + int n = ffs(x); + + return n ? n-1 : 32; +} + +/****************************************************************************** + * HW block + *****************************************************************************/ +/* register bank descriptor */ +static struct { + unsigned lpm; /* is LPM? */ + void __iomem *abs; /* bus map offset */ + void __iomem *cap; /* bus map offset + CAP offset + CAP data */ + size_t size; /* bank size */ +} hw_bank; + +/* UDC register map */ +#define ABS_CAPLENGTH (0x100UL) +#define ABS_HCCPARAMS (0x108UL) +#define ABS_DCCPARAMS (0x124UL) +#define ABS_TESTMODE (hw_bank.lpm ? 0x0FCUL : 0x138UL) +/* offset to CAPLENTGH (addr + data) */ +#define CAP_USBCMD (0x000UL) +#define CAP_USBSTS (0x004UL) +#define CAP_USBINTR (0x008UL) +#define CAP_DEVICEADDR (0x014UL) +#define CAP_ENDPTLISTADDR (0x018UL) +#define CAP_PORTSC (0x044UL) +#define CAP_DEVLC (0x0B4UL) +#define CAP_USBMODE (hw_bank.lpm ? 0x0C8UL : 0x068UL) +#define CAP_ENDPTSETUPSTAT (hw_bank.lpm ? 0x0D8UL : 0x06CUL) +#define CAP_ENDPTPRIME (hw_bank.lpm ? 0x0DCUL : 0x070UL) +#define CAP_ENDPTFLUSH (hw_bank.lpm ? 0x0E0UL : 0x074UL) +#define CAP_ENDPTSTAT (hw_bank.lpm ? 0x0E4UL : 0x078UL) +#define CAP_ENDPTCOMPLETE (hw_bank.lpm ? 0x0E8UL : 0x07CUL) +#define CAP_ENDPTCTRL (hw_bank.lpm ? 0x0ECUL : 0x080UL) +#define CAP_LAST (hw_bank.lpm ? 0x12CUL : 0x0C0UL) + +/* maximum number of enpoints: valid only after hw_device_reset() */ +static unsigned hw_ep_max; + +/** + * hw_ep_bit: calculates the bit number + * @num: endpoint number + * @dir: endpoint direction + * + * This function returns bit number + */ +static inline int hw_ep_bit(int num, int dir) +{ + return num + (dir ? 16 : 0); +} + +/** + * hw_aread: reads from register bitfield + * @addr: address relative to bus map + * @mask: bitfield mask + * + * This function returns register bitfield data + */ +static u32 hw_aread(u32 addr, u32 mask) +{ + return ioread32(addr + hw_bank.abs) & mask; +} + +/** + * hw_awrite: writes to register bitfield + * @addr: address relative to bus map + * @mask: bitfield mask + * @data: new data + */ +static void hw_awrite(u32 addr, u32 mask, u32 data) +{ + iowrite32(hw_aread(addr, ~mask) | (data & mask), + addr + hw_bank.abs); +} + +/** + * hw_cread: reads from register bitfield + * @addr: address relative to CAP offset plus content + * @mask: bitfield mask + * + * This function returns register bitfield data + */ +static u32 hw_cread(u32 addr, u32 mask) +{ + return ioread32(addr + hw_bank.cap) & mask; +} + +/** + * hw_cwrite: writes to register bitfield + * @addr: address relative to CAP offset plus content + * @mask: bitfield mask + * @data: new data + */ +static void hw_cwrite(u32 addr, u32 mask, u32 data) +{ + iowrite32(hw_cread(addr, ~mask) | (data & mask), + addr + hw_bank.cap); +} + +/** + * hw_ctest_and_clear: tests & clears register bitfield + * @addr: address relative to CAP offset plus content + * @mask: bitfield mask + * + * This function returns register bitfield data + */ +static u32 hw_ctest_and_clear(u32 addr, u32 mask) +{ + u32 reg = hw_cread(addr, mask); + + iowrite32(reg, addr + hw_bank.cap); + return reg; +} + +/** + * hw_ctest_and_write: tests & writes register bitfield + * @addr: address relative to CAP offset plus content + * @mask: bitfield mask + * @data: new data + * + * This function returns register bitfield data + */ +static u32 hw_ctest_and_write(u32 addr, u32 mask, u32 data) +{ + u32 reg = hw_cread(addr, ~0); + + iowrite32((reg & ~mask) | (data & mask), addr + hw_bank.cap); + return (reg & mask) >> ffs_nr(mask); +} + +/** + * hw_device_reset: resets chip (execute without interruption) + * @base: register base address + * + * This function returns an error code + */ +static int hw_device_reset(void __iomem *base) +{ + u32 reg; + + /* bank is a module variable */ + hw_bank.abs = base; + + hw_bank.cap = hw_bank.abs; + hw_bank.cap += ABS_CAPLENGTH; + hw_bank.cap += ioread8(hw_bank.cap); + + reg = hw_aread(ABS_HCCPARAMS, HCCPARAMS_LEN) >> ffs_nr(HCCPARAMS_LEN); + hw_bank.lpm = reg; + hw_bank.size = hw_bank.cap - hw_bank.abs; + hw_bank.size += CAP_LAST; + hw_bank.size /= sizeof(u32); + + /* should flush & stop before reset */ + hw_cwrite(CAP_ENDPTFLUSH, ~0, ~0); + hw_cwrite(CAP_USBCMD, USBCMD_RS, 0); + + hw_cwrite(CAP_USBCMD, USBCMD_RST, USBCMD_RST); + while (hw_cread(CAP_USBCMD, USBCMD_RST)) + udelay(10); /* not RTOS friendly */ + + /* USBMODE should be configured step by step */ + hw_cwrite(CAP_USBMODE, USBMODE_CM, USBMODE_CM_IDLE); + hw_cwrite(CAP_USBMODE, USBMODE_CM, USBMODE_CM_DEVICE); + hw_cwrite(CAP_USBMODE, USBMODE_SLOM, USBMODE_SLOM); /* HW >= 2.3 */ + + if (hw_cread(CAP_USBMODE, USBMODE_CM) != USBMODE_CM_DEVICE) { + pr_err("cannot enter in device mode"); + pr_err("lpm = %i", hw_bank.lpm); + return -ENODEV; + } + + reg = hw_aread(ABS_DCCPARAMS, DCCPARAMS_DEN) >> ffs_nr(DCCPARAMS_DEN); + if (reg == 0 || reg > ENDPT_MAX) + return -ENODEV; + + hw_ep_max = reg; /* cache hw ENDPT_MAX */ + + /* setup lock mode ? */ + + /* ENDPTSETUPSTAT is '0' by default */ + + /* HCSPARAMS.bf.ppc SHOULD BE zero for device */ + + return 0; +} + +/** + * hw_device_state: enables/disables interrupts & starts/stops device (execute + * without interruption) + * @dma: 0 => disable, !0 => enable and set dma engine + * + * This function returns an error code + */ +static int hw_device_state(u32 dma) +{ + if (dma) { + hw_cwrite(CAP_ENDPTLISTADDR, ~0, dma); + /* interrupt, error, port change, reset, sleep/suspend */ + hw_cwrite(CAP_USBINTR, ~0, + USBi_UI|USBi_UEI|USBi_PCI|USBi_URI|USBi_SLI); + hw_cwrite(CAP_USBCMD, USBCMD_RS, USBCMD_RS); + } else { + hw_cwrite(CAP_USBCMD, USBCMD_RS, 0); + hw_cwrite(CAP_USBINTR, ~0, 0); + } + return 0; +} + +/** + * hw_ep_flush: flush endpoint fifo (execute without interruption) + * @num: endpoint number + * @dir: endpoint direction + * + * This function returns an error code + */ +static int hw_ep_flush(int num, int dir) +{ + int n = hw_ep_bit(num, dir); + + do { + /* flush any pending transfer */ + hw_cwrite(CAP_ENDPTFLUSH, BIT(n), BIT(n)); + while (hw_cread(CAP_ENDPTFLUSH, BIT(n))) + cpu_relax(); + } while (hw_cread(CAP_ENDPTSTAT, BIT(n))); + + return 0; +} + +/** + * hw_ep_disable: disables endpoint (execute without interruption) + * @num: endpoint number + * @dir: endpoint direction + * + * This function returns an error code + */ +static int hw_ep_disable(int num, int dir) +{ + hw_ep_flush(num, dir); + hw_cwrite(CAP_ENDPTCTRL + num * sizeof(u32), + dir ? ENDPTCTRL_TXE : ENDPTCTRL_RXE, 0); + return 0; +} + +/** + * hw_ep_enable: enables endpoint (execute without interruption) + * @num: endpoint number + * @dir: endpoint direction + * @type: endpoint type + * + * This function returns an error code + */ +static int hw_ep_enable(int num, int dir, int type) +{ + u32 mask, data; + + if (dir) { + mask = ENDPTCTRL_TXT; /* type */ + data = type << ffs_nr(mask); + + mask |= ENDPTCTRL_TXS; /* unstall */ + mask |= ENDPTCTRL_TXR; /* reset data toggle */ + data |= ENDPTCTRL_TXR; + mask |= ENDPTCTRL_TXE; /* enable */ + data |= ENDPTCTRL_TXE; + } else { + mask = ENDPTCTRL_RXT; /* type */ + data = type << ffs_nr(mask); + + mask |= ENDPTCTRL_RXS; /* unstall */ + mask |= ENDPTCTRL_RXR; /* reset data toggle */ + data |= ENDPTCTRL_RXR; + mask |= ENDPTCTRL_RXE; /* enable */ + data |= ENDPTCTRL_RXE; + } + hw_cwrite(CAP_ENDPTCTRL + num * sizeof(u32), mask, data); + return 0; +} + +/** + * hw_ep_get_halt: return endpoint halt status + * @num: endpoint number + * @dir: endpoint direction + * + * This function returns 1 if endpoint halted + */ +static int hw_ep_get_halt(int num, int dir) +{ + u32 mask = dir ? ENDPTCTRL_TXS : ENDPTCTRL_RXS; + + return hw_cread(CAP_ENDPTCTRL + num * sizeof(u32), mask) ? 1 : 0; +} + +/** + * hw_ep_is_primed: test if endpoint is primed (execute without interruption) + * @num: endpoint number + * @dir: endpoint direction + * + * This function returns true if endpoint primed + */ +static int hw_ep_is_primed(int num, int dir) +{ + u32 reg = hw_cread(CAP_ENDPTPRIME, ~0) | hw_cread(CAP_ENDPTSTAT, ~0); + + return test_bit(hw_ep_bit(num, dir), (void *)®); +} + +/** + * hw_test_and_clear_setup_status: test & clear setup status (execute without + * interruption) + * @n: bit number (endpoint) + * + * This function returns setup status + */ +static int hw_test_and_clear_setup_status(int n) +{ + return hw_ctest_and_clear(CAP_ENDPTSETUPSTAT, BIT(n)); +} + +/** + * hw_ep_prime: primes endpoint (execute without interruption) + * @num: endpoint number + * @dir: endpoint direction + * @is_ctrl: true if control endpoint + * + * This function returns an error code + */ +static int hw_ep_prime(int num, int dir, int is_ctrl) +{ + int n = hw_ep_bit(num, dir); + + /* the caller should flush first */ + if (hw_ep_is_primed(num, dir)) + return -EBUSY; + + if (is_ctrl && dir == RX && hw_cread(CAP_ENDPTSETUPSTAT, BIT(num))) + return -EAGAIN; + + hw_cwrite(CAP_ENDPTPRIME, BIT(n), BIT(n)); + + while (hw_cread(CAP_ENDPTPRIME, BIT(n))) + cpu_relax(); + if (is_ctrl && dir == RX && hw_cread(CAP_ENDPTSETUPSTAT, BIT(num))) + return -EAGAIN; + + /* status shoult be tested according with manual but it doesn't work */ + return 0; +} + +/** + * hw_ep_set_halt: configures ep halt & resets data toggle after clear (execute + * without interruption) + * @num: endpoint number + * @dir: endpoint direction + * @value: true => stall, false => unstall + * + * This function returns an error code + */ +static int hw_ep_set_halt(int num, int dir, int value) +{ + if (value != 0 && value != 1) + return -EINVAL; + + do { + u32 addr = CAP_ENDPTCTRL + num * sizeof(u32); + u32 mask_xs = dir ? ENDPTCTRL_TXS : ENDPTCTRL_RXS; + u32 mask_xr = dir ? ENDPTCTRL_TXR : ENDPTCTRL_RXR; + + /* data toggle - reserved for EP0 but it's in ESS */ + hw_cwrite(addr, mask_xs|mask_xr, value ? mask_xs : mask_xr); + + } while (value != hw_ep_get_halt(num, dir)); + + return 0; +} + +/** + * hw_intr_clear: disables interrupt & clears interrupt status (execute without + * interruption) + * @n: interrupt bit + * + * This function returns an error code + */ +static int hw_intr_clear(int n) +{ + if (n >= REG_BITS) + return -EINVAL; + + hw_cwrite(CAP_USBINTR, BIT(n), 0); + hw_cwrite(CAP_USBSTS, BIT(n), BIT(n)); + return 0; +} + +/** + * hw_intr_force: enables interrupt & forces interrupt status (execute without + * interruption) + * @n: interrupt bit + * + * This function returns an error code + */ +static int hw_intr_force(int n) +{ + if (n >= REG_BITS) + return -EINVAL; + + hw_awrite(ABS_TESTMODE, TESTMODE_FORCE, TESTMODE_FORCE); + hw_cwrite(CAP_USBINTR, BIT(n), BIT(n)); + hw_cwrite(CAP_USBSTS, BIT(n), BIT(n)); + hw_awrite(ABS_TESTMODE, TESTMODE_FORCE, 0); + return 0; +} + +/** + * hw_is_port_high_speed: test if port is high speed + * + * This function returns true if high speed port + */ +static int hw_port_is_high_speed(void) +{ + return hw_bank.lpm ? hw_cread(CAP_DEVLC, DEVLC_PSPD) : + hw_cread(CAP_PORTSC, PORTSC_HSP); +} + +/** + * hw_port_test_get: reads port test mode value + * + * This function returns port test mode value + */ +static u8 hw_port_test_get(void) +{ + return hw_cread(CAP_PORTSC, PORTSC_PTC) >> ffs_nr(PORTSC_PTC); +} + +/** + * hw_port_test_set: writes port test mode (execute without interruption) + * @mode: new value + * + * This function returns an error code + */ +static int hw_port_test_set(u8 mode) +{ + const u8 TEST_MODE_MAX = 7; + + if (mode > TEST_MODE_MAX) + return -EINVAL; + + hw_cwrite(CAP_PORTSC, PORTSC_PTC, mode << ffs_nr(PORTSC_PTC)); + return 0; +} + +/** + * hw_read_intr_enable: returns interrupt enable register + * + * This function returns register data + */ +static u32 hw_read_intr_enable(void) +{ + return hw_cread(CAP_USBINTR, ~0); +} + +/** + * hw_read_intr_status: returns interrupt status register + * + * This function returns register data + */ +static u32 hw_read_intr_status(void) +{ + return hw_cread(CAP_USBSTS, ~0); +} + +/** + * hw_register_read: reads all device registers (execute without interruption) + * @buf: destination buffer + * @size: buffer size + * + * This function returns number of registers read + */ +static size_t hw_register_read(u32 *buf, size_t size) +{ + unsigned i; + + if (size > hw_bank.size) + size = hw_bank.size; + + for (i = 0; i < size; i++) + buf[i] = hw_aread(i * sizeof(u32), ~0); + + return size; +} + +/** + * hw_register_write: writes to register + * @addr: register address + * @data: register value + * + * This function returns an error code + */ +static int hw_register_write(u16 addr, u32 data) +{ + /* align */ + addr /= sizeof(u32); + + if (addr >= hw_bank.size) + return -EINVAL; + + /* align */ + addr *= sizeof(u32); + + hw_awrite(addr, ~0, data); + return 0; +} + +/** + * hw_test_and_clear_complete: test & clear complete status (execute without + * interruption) + * @n: bit number (endpoint) + * + * This function returns complete status + */ +static int hw_test_and_clear_complete(int n) +{ + return hw_ctest_and_clear(CAP_ENDPTCOMPLETE, BIT(n)); +} + +/** + * hw_test_and_clear_intr_active: test & clear active interrupts (execute + * without interruption) + * + * This function returns active interrutps + */ +static u32 hw_test_and_clear_intr_active(void) +{ + u32 reg = hw_read_intr_status() & hw_read_intr_enable(); + + hw_cwrite(CAP_USBSTS, ~0, reg); + return reg; +} + +/** + * hw_test_and_clear_setup_guard: test & clear setup guard (execute without + * interruption) + * + * This function returns guard value + */ +static int hw_test_and_clear_setup_guard(void) +{ + return hw_ctest_and_write(CAP_USBCMD, USBCMD_SUTW, 0); +} + +/** + * hw_test_and_set_setup_guard: test & set setup guard (execute without + * interruption) + * + * This function returns guard value + */ +static int hw_test_and_set_setup_guard(void) +{ + return hw_ctest_and_write(CAP_USBCMD, USBCMD_SUTW, USBCMD_SUTW); +} + +/** + * hw_usb_set_address: configures USB address (execute without interruption) + * @value: new USB address + * + * This function returns an error code + */ +static int hw_usb_set_address(u8 value) +{ + /* advance */ + hw_cwrite(CAP_DEVICEADDR, DEVICEADDR_USBADR | DEVICEADDR_USBADRA, + value << ffs_nr(DEVICEADDR_USBADR) | DEVICEADDR_USBADRA); + return 0; +} + +/** + * hw_usb_reset: restart device after a bus reset (execute without + * interruption) + * + * This function returns an error code + */ +static int hw_usb_reset(void) +{ + hw_usb_set_address(0); + + /* ESS flushes only at end?!? */ + hw_cwrite(CAP_ENDPTFLUSH, ~0, ~0); /* flush all EPs */ + + /* clear setup token semaphores */ + hw_cwrite(CAP_ENDPTSETUPSTAT, 0, 0); /* writes its content */ + + /* clear complete status */ + hw_cwrite(CAP_ENDPTCOMPLETE, 0, 0); /* writes its content */ + + /* wait until all bits cleared */ + while (hw_cread(CAP_ENDPTPRIME, ~0)) + udelay(10); /* not RTOS friendly */ + + /* reset all endpoints ? */ + + /* reset internal status and wait for further instructions + no need to verify the port reset status (ESS does it) */ + + return 0; +} + +/****************************************************************************** + * DBG block + *****************************************************************************/ +/** + * show_device: prints information about device capabilities and status + * + * Check "device.h" for details + */ +static ssize_t show_device(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + struct usb_gadget *gadget = &udc->gadget; + int n = 0; + + dbg_trace("[%s] %p\n", __func__, buf); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + return 0; + } + + n += scnprintf(buf + n, PAGE_SIZE - n, "speed = %d\n", + gadget->speed); + n += scnprintf(buf + n, PAGE_SIZE - n, "is_dualspeed = %d\n", + gadget->is_dualspeed); + n += scnprintf(buf + n, PAGE_SIZE - n, "is_otg = %d\n", + gadget->is_otg); + n += scnprintf(buf + n, PAGE_SIZE - n, "is_a_peripheral = %d\n", + gadget->is_a_peripheral); + n += scnprintf(buf + n, PAGE_SIZE - n, "b_hnp_enable = %d\n", + gadget->b_hnp_enable); + n += scnprintf(buf + n, PAGE_SIZE - n, "a_hnp_support = %d\n", + gadget->a_hnp_support); + n += scnprintf(buf + n, PAGE_SIZE - n, "a_alt_hnp_support = %d\n", + gadget->a_alt_hnp_support); + n += scnprintf(buf + n, PAGE_SIZE - n, "name = %s\n", + (gadget->name ? gadget->name : "")); + + return n; +} +static DEVICE_ATTR(device, S_IRUSR, show_device, NULL); + +/** + * show_driver: prints information about attached gadget (if any) + * + * Check "device.h" for details + */ +static ssize_t show_driver(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + struct usb_gadget_driver *driver = udc->driver; + int n = 0; + + dbg_trace("[%s] %p\n", __func__, buf); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + return 0; + } + + if (driver == NULL) + return scnprintf(buf, PAGE_SIZE, + "There is no gadget attached!\n"); + + n += scnprintf(buf + n, PAGE_SIZE - n, "function = %s\n", + (driver->function ? driver->function : "")); + n += scnprintf(buf + n, PAGE_SIZE - n, "max speed = %d\n", + driver->speed); + + return n; +} +static DEVICE_ATTR(driver, S_IRUSR, show_driver, NULL); + +/* Maximum event message length */ +#define DBG_DATA_MSG 64UL + +/* Maximum event messages */ +#define DBG_DATA_MAX 128UL + +/* Event buffer descriptor */ +static struct { + char (buf[DBG_DATA_MAX])[DBG_DATA_MSG]; /* buffer */ + unsigned idx; /* index */ + unsigned tty; /* print to console? */ + rwlock_t lck; /* lock */ +} dbg_data = { + .idx = 0, + .tty = 0, + .lck = __RW_LOCK_UNLOCKED(lck) +}; + +/** + * dbg_dec: decrements debug event index + * @idx: buffer index + */ +static void dbg_dec(unsigned *idx) +{ + *idx = (*idx - 1) & (DBG_DATA_MAX-1); +} + +/** + * dbg_inc: increments debug event index + * @idx: buffer index + */ +static void dbg_inc(unsigned *idx) +{ + *idx = (*idx + 1) & (DBG_DATA_MAX-1); +} + +/** + * dbg_print: prints the common part of the event + * @addr: endpoint address + * @name: event name + * @status: status + * @extra: extra information + */ +static void dbg_print(u8 addr, const char *name, int status, const char *extra) +{ + struct timeval tval; + unsigned int stamp; + unsigned long flags; + + write_lock_irqsave(&dbg_data.lck, flags); + + do_gettimeofday(&tval); + stamp = tval.tv_sec & 0xFFFF; /* 2^32 = 4294967296. Limit to 4096s */ + stamp = stamp * 1000000 + tval.tv_usec; + + scnprintf(dbg_data.buf[dbg_data.idx], DBG_DATA_MSG, + "%04X\t» %02X %-7.7s %4i «\t%s\n", + stamp, addr, name, status, extra); + + dbg_inc(&dbg_data.idx); + + write_unlock_irqrestore(&dbg_data.lck, flags); + + if (dbg_data.tty != 0) + pr_notice("%04X\t» %02X %-7.7s %4i «\t%s\n", + stamp, addr, name, status, extra); +} + +/** + * dbg_done: prints a DONE event + * @addr: endpoint address + * @td: transfer descriptor + * @status: status + */ +static void dbg_done(u8 addr, const u32 token, int status) +{ + char msg[DBG_DATA_MSG]; + + scnprintf(msg, sizeof(msg), "%d %02X", + (int)(token & TD_TOTAL_BYTES) >> ffs_nr(TD_TOTAL_BYTES), + (int)(token & TD_STATUS) >> ffs_nr(TD_STATUS)); + dbg_print(addr, "DONE", status, msg); +} + +/** + * dbg_event: prints a generic event + * @addr: endpoint address + * @name: event name + * @status: status + */ +static void dbg_event(u8 addr, const char *name, int status) +{ + if (name != NULL) + dbg_print(addr, name, status, ""); +} + +/* + * dbg_queue: prints a QUEUE event + * @addr: endpoint address + * @req: USB request + * @status: status + */ +static void dbg_queue(u8 addr, const struct usb_request *req, int status) +{ + char msg[DBG_DATA_MSG]; + + if (req != NULL) { + scnprintf(msg, sizeof(msg), + "%d %d", !req->no_interrupt, req->length); + dbg_print(addr, "QUEUE", status, msg); + } +} + +/** + * dbg_setup: prints a SETUP event + * @addr: endpoint address + * @req: setup request + */ +static void dbg_setup(u8 addr, const struct usb_ctrlrequest *req) +{ + char msg[DBG_DATA_MSG]; + + if (req != NULL) { + scnprintf(msg, sizeof(msg), + "%02X %02X %04X %04X %d", req->bRequestType, + req->bRequest, le16_to_cpu(req->wValue), + le16_to_cpu(req->wIndex), le16_to_cpu(req->wLength)); + dbg_print(addr, "SETUP", 0, msg); + } +} + +/** + * show_events: displays the event buffer + * + * Check "device.h" for details + */ +static ssize_t show_events(struct device *dev, struct device_attribute *attr, + char *buf) +{ + unsigned long flags; + unsigned i, j, n = 0; + + dbg_trace("[%s] %p\n", __func__, buf); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + return 0; + } + + read_lock_irqsave(&dbg_data.lck, flags); + + i = dbg_data.idx; + for (dbg_dec(&i); i != dbg_data.idx; dbg_dec(&i)) { + n += strlen(dbg_data.buf[i]); + if (n >= PAGE_SIZE) { + n -= strlen(dbg_data.buf[i]); + break; + } + } + for (j = 0, dbg_inc(&i); j < n; dbg_inc(&i)) + j += scnprintf(buf + j, PAGE_SIZE - j, + "%s", dbg_data.buf[i]); + + read_unlock_irqrestore(&dbg_data.lck, flags); + + return n; +} + +/** + * store_events: configure if events are going to be also printed to console + * + * Check "device.h" for details + */ +static ssize_t store_events(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + unsigned tty; + + dbg_trace("[%s] %p, %d\n", __func__, buf, count); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + goto done; + } + + if (sscanf(buf, "%u", &tty) != 1 || tty > 1) { + dev_err(dev, "<1|0>: enable|disable console log\n"); + goto done; + } + + dbg_data.tty = tty; + dev_info(dev, "tty = %u", dbg_data.tty); + + done: + return count; +} +static DEVICE_ATTR(events, S_IRUSR | S_IWUSR, show_events, store_events); + +/** + * show_inters: interrupt status, enable status and historic + * + * Check "device.h" for details + */ +static ssize_t show_inters(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + unsigned long flags; + u32 intr; + unsigned i, j, n = 0; + + dbg_trace("[%s] %p\n", __func__, buf); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + return 0; + } + + spin_lock_irqsave(udc->lock, flags); + + n += scnprintf(buf + n, PAGE_SIZE - n, + "status = %08x\n", hw_read_intr_status()); + n += scnprintf(buf + n, PAGE_SIZE - n, + "enable = %08x\n", hw_read_intr_enable()); + + n += scnprintf(buf + n, PAGE_SIZE - n, "*test = %d\n", + isr_statistics.test); + n += scnprintf(buf + n, PAGE_SIZE - n, "» ui = %d\n", + isr_statistics.ui); + n += scnprintf(buf + n, PAGE_SIZE - n, "» uei = %d\n", + isr_statistics.uei); + n += scnprintf(buf + n, PAGE_SIZE - n, "» pci = %d\n", + isr_statistics.pci); + n += scnprintf(buf + n, PAGE_SIZE - n, "» uri = %d\n", + isr_statistics.uri); + n += scnprintf(buf + n, PAGE_SIZE - n, "» sli = %d\n", + isr_statistics.sli); + n += scnprintf(buf + n, PAGE_SIZE - n, "*none = %d\n", + isr_statistics.none); + n += scnprintf(buf + n, PAGE_SIZE - n, "*hndl = %d\n", + isr_statistics.hndl.cnt); + + for (i = isr_statistics.hndl.idx, j = 0; j <= ISR_MASK; j++, i++) { + i &= ISR_MASK; + intr = isr_statistics.hndl.buf[i]; + + if (USBi_UI & intr) + n += scnprintf(buf + n, PAGE_SIZE - n, "ui "); + intr &= ~USBi_UI; + if (USBi_UEI & intr) + n += scnprintf(buf + n, PAGE_SIZE - n, "uei "); + intr &= ~USBi_UEI; + if (USBi_PCI & intr) + n += scnprintf(buf + n, PAGE_SIZE - n, "pci "); + intr &= ~USBi_PCI; + if (USBi_URI & intr) + n += scnprintf(buf + n, PAGE_SIZE - n, "uri "); + intr &= ~USBi_URI; + if (USBi_SLI & intr) + n += scnprintf(buf + n, PAGE_SIZE - n, "sli "); + intr &= ~USBi_SLI; + if (intr) + n += scnprintf(buf + n, PAGE_SIZE - n, "??? "); + if (isr_statistics.hndl.buf[i]) + n += scnprintf(buf + n, PAGE_SIZE - n, "\n"); + } + + spin_unlock_irqrestore(udc->lock, flags); + + return n; +} + +/** + * store_inters: enable & force or disable an individual interrutps + * (to be used for test purposes only) + * + * Check "device.h" for details + */ +static ssize_t store_inters(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + unsigned long flags; + unsigned en, bit; + + dbg_trace("[%s] %p, %d\n", __func__, buf, count); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + goto done; + } + + if (sscanf(buf, "%u %u", &en, &bit) != 2 || en > 1) { + dev_err(dev, "<1|0> : enable|disable interrupt"); + goto done; + } + + spin_lock_irqsave(udc->lock, flags); + if (en) { + if (hw_intr_force(bit)) + dev_err(dev, "invalid bit number\n"); + else + isr_statistics.test++; + } else { + if (hw_intr_clear(bit)) + dev_err(dev, "invalid bit number\n"); + } + spin_unlock_irqrestore(udc->lock, flags); + + done: + return count; +} +static DEVICE_ATTR(inters, S_IRUSR | S_IWUSR, show_inters, store_inters); + +/** + * show_port_test: reads port test mode + * + * Check "device.h" for details + */ +static ssize_t show_port_test(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + unsigned long flags; + unsigned mode; + + dbg_trace("[%s] %p\n", __func__, buf); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + return 0; + } + + spin_lock_irqsave(udc->lock, flags); + mode = hw_port_test_get(); + spin_unlock_irqrestore(udc->lock, flags); + + return scnprintf(buf, PAGE_SIZE, "mode = %u\n", mode); +} + +/** + * store_port_test: writes port test mode + * + * Check "device.h" for details + */ +static ssize_t store_port_test(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + unsigned long flags; + unsigned mode; + + dbg_trace("[%s] %p, %d\n", __func__, buf, count); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + goto done; + } + + if (sscanf(buf, "%u", &mode) != 1) { + dev_err(dev, ": set port test mode"); + goto done; + } + + spin_lock_irqsave(udc->lock, flags); + if (hw_port_test_set(mode)) + dev_err(dev, "invalid mode\n"); + spin_unlock_irqrestore(udc->lock, flags); + + done: + return count; +} +static DEVICE_ATTR(port_test, S_IRUSR | S_IWUSR, + show_port_test, store_port_test); + +/** + * show_qheads: DMA contents of all queue heads + * + * Check "device.h" for details + */ +static ssize_t show_qheads(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + unsigned long flags; + unsigned i, j, n = 0; + + dbg_trace("[%s] %p\n", __func__, buf); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + return 0; + } + + spin_lock_irqsave(udc->lock, flags); + for (i = 0; i < hw_ep_max; i++) { + struct ci13xxx_ep *mEp = &udc->ci13xxx_ep[i]; + n += scnprintf(buf + n, PAGE_SIZE - n, + "EP=%02i: RX=%08X TX=%08X\n", + i, (u32)mEp->qh[RX].dma, (u32)mEp->qh[TX].dma); + for (j = 0; j < (sizeof(struct ci13xxx_qh)/sizeof(u32)); j++) { + n += scnprintf(buf + n, PAGE_SIZE - n, + " %04X: %08X %08X\n", j, + *((u32 *)mEp->qh[RX].ptr + j), + *((u32 *)mEp->qh[TX].ptr + j)); + } + } + spin_unlock_irqrestore(udc->lock, flags); + + return n; +} +static DEVICE_ATTR(qheads, S_IRUSR, show_qheads, NULL); + +/** + * show_registers: dumps all registers + * + * Check "device.h" for details + */ +static ssize_t show_registers(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + unsigned long flags; + u32 dump[512]; + unsigned i, k, n = 0; + + dbg_trace("[%s] %p\n", __func__, buf); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + return 0; + } + + spin_lock_irqsave(udc->lock, flags); + k = hw_register_read(dump, sizeof(dump)/sizeof(u32)); + spin_unlock_irqrestore(udc->lock, flags); + + for (i = 0; i < k; i++) { + n += scnprintf(buf + n, PAGE_SIZE - n, + "reg[0x%04X] = 0x%08X\n", + i * (unsigned)sizeof(u32), dump[i]); + } + + return n; +} + +/** + * store_registers: writes value to register address + * + * Check "device.h" for details + */ +static ssize_t store_registers(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + unsigned long addr, data, flags; + + dbg_trace("[%s] %p, %d\n", __func__, buf, count); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + goto done; + } + + if (sscanf(buf, "%li %li", &addr, &data) != 2) { + dev_err(dev, " : write data to register address"); + goto done; + } + + spin_lock_irqsave(udc->lock, flags); + if (hw_register_write(addr, data)) + dev_err(dev, "invalid address range\n"); + spin_unlock_irqrestore(udc->lock, flags); + + done: + return count; +} +static DEVICE_ATTR(registers, S_IRUSR | S_IWUSR, + show_registers, store_registers); + +/** + * show_requests: DMA contents of all requests currently queued (all endpts) + * + * Check "device.h" for details + */ +static ssize_t show_requests(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct ci13xxx *udc = container_of(dev, struct ci13xxx, gadget.dev); + unsigned long flags; + struct list_head *ptr = NULL; + struct ci13xxx_req *req = NULL; + unsigned i, j, k, n = 0, qSize = sizeof(struct ci13xxx_td)/sizeof(u32); + + dbg_trace("[%s] %p\n", __func__, buf); + if (attr == NULL || buf == NULL) { + dev_err(dev, "[%s] EINVAL\n", __func__); + return 0; + } + + spin_lock_irqsave(udc->lock, flags); + for (i = 0; i < hw_ep_max; i++) + for (k = RX; k <= TX; k++) + list_for_each(ptr, &udc->ci13xxx_ep[i].qh[k].queue) + { + req = list_entry(ptr, + struct ci13xxx_req, queue); + + n += scnprintf(buf + n, PAGE_SIZE - n, + "EP=%02i: TD=%08X %s\n", + i, (u32)req->dma, + ((k == RX) ? "RX" : "TX")); + + for (j = 0; j < qSize; j++) + n += scnprintf(buf + n, PAGE_SIZE - n, + " %04X: %08X\n", j, + *((u32 *)req->ptr + j)); + } + spin_unlock_irqrestore(udc->lock, flags); + + return n; +} +static DEVICE_ATTR(requests, S_IRUSR, show_requests, NULL); + +/** + * dbg_create_files: initializes the attribute interface + * @dev: device + * + * This function returns an error code + */ +__maybe_unused static int dbg_create_files(struct device *dev) +{ + int retval = 0; + + if (dev == NULL) + return -EINVAL; + retval = device_create_file(dev, &dev_attr_device); + if (retval) + goto done; + retval = device_create_file(dev, &dev_attr_driver); + if (retval) + goto rm_device; + retval = device_create_file(dev, &dev_attr_events); + if (retval) + goto rm_driver; + retval = device_create_file(dev, &dev_attr_inters); + if (retval) + goto rm_events; + retval = device_create_file(dev, &dev_attr_port_test); + if (retval) + goto rm_inters; + retval = device_create_file(dev, &dev_attr_qheads); + if (retval) + goto rm_port_test; + retval = device_create_file(dev, &dev_attr_registers); + if (retval) + goto rm_qheads; + retval = device_create_file(dev, &dev_attr_requests); + if (retval) + goto rm_registers; + return 0; + + rm_registers: + device_remove_file(dev, &dev_attr_registers); + rm_qheads: + device_remove_file(dev, &dev_attr_qheads); + rm_port_test: + device_remove_file(dev, &dev_attr_port_test); + rm_inters: + device_remove_file(dev, &dev_attr_inters); + rm_events: + device_remove_file(dev, &dev_attr_events); + rm_driver: + device_remove_file(dev, &dev_attr_driver); + rm_device: + device_remove_file(dev, &dev_attr_device); + done: + return retval; +} + +/** + * dbg_remove_files: destroys the attribute interface + * @dev: device + * + * This function returns an error code + */ +__maybe_unused static int dbg_remove_files(struct device *dev) +{ + if (dev == NULL) + return -EINVAL; + device_remove_file(dev, &dev_attr_requests); + device_remove_file(dev, &dev_attr_registers); + device_remove_file(dev, &dev_attr_qheads); + device_remove_file(dev, &dev_attr_port_test); + device_remove_file(dev, &dev_attr_inters); + device_remove_file(dev, &dev_attr_events); + device_remove_file(dev, &dev_attr_driver); + device_remove_file(dev, &dev_attr_device); + return 0; +} + +/****************************************************************************** + * UTIL block + *****************************************************************************/ +/** + * _usb_addr: calculates endpoint address from direction & number + * @ep: endpoint + */ +static inline u8 _usb_addr(struct ci13xxx_ep *ep) +{ + return ((ep->dir == TX) ? USB_ENDPOINT_DIR_MASK : 0) | ep->num; +} + +/** + * _hardware_queue: configures a request at hardware level + * @gadget: gadget + * @mEp: endpoint + * + * This function returns an error code + */ +static int _hardware_enqueue(struct ci13xxx_ep *mEp, struct ci13xxx_req *mReq) +{ + unsigned i; + + trace("%p, %p", mEp, mReq); + + /* don't queue twice */ + if (mReq->req.status == -EALREADY) + return -EALREADY; + + if (hw_ep_is_primed(mEp->num, mEp->dir)) + return -EBUSY; + + mReq->req.status = -EALREADY; + + if (mReq->req.length && !mReq->req.dma) { + mReq->req.dma = \ + dma_map_single(mEp->device, mReq->req.buf, + mReq->req.length, mEp->dir ? + DMA_TO_DEVICE : DMA_FROM_DEVICE); + if (mReq->req.dma == 0) + return -ENOMEM; + + mReq->map = 1; + } + + /* + * TD configuration + * TODO - handle requests which spawns into several TDs + */ + memset(mReq->ptr, 0, sizeof(*mReq->ptr)); + mReq->ptr->next |= TD_TERMINATE; + mReq->ptr->token = mReq->req.length << ffs_nr(TD_TOTAL_BYTES); + mReq->ptr->token &= TD_TOTAL_BYTES; + mReq->ptr->token |= TD_IOC; + mReq->ptr->token |= TD_STATUS_ACTIVE; + mReq->ptr->page[0] = mReq->req.dma; + for (i = 1; i < 5; i++) + mReq->ptr->page[i] = + (mReq->req.dma + i * PAGE_SIZE) & ~TD_RESERVED_MASK; + + /* + * QH configuration + * At this point it's guaranteed exclusive access to qhead + * (endpt is not primed) so it's no need to use tripwire + */ + mEp->qh[mEp->dir].ptr->td.next = mReq->dma; /* TERMINATE = 0 */ + mEp->qh[mEp->dir].ptr->td.token &= ~TD_STATUS; /* clear status */ + if (mReq->req.zero == 0) + mEp->qh[mEp->dir].ptr->cap |= QH_ZLT; + else + mEp->qh[mEp->dir].ptr->cap &= ~QH_ZLT; + + wmb(); /* synchronize before ep prime */ + + return hw_ep_prime(mEp->num, mEp->dir, + mEp->type == USB_ENDPOINT_XFER_CONTROL); +} + +/** + * _hardware_dequeue: handles a request at hardware level + * @gadget: gadget + * @mEp: endpoint + * + * This function returns an error code + */ +static int _hardware_dequeue(struct ci13xxx_ep *mEp, struct ci13xxx_req *mReq) +{ + trace("%p, %p", mEp, mReq); + + if (mReq->req.status != -EALREADY) + return -EINVAL; + + if (hw_ep_is_primed(mEp->num, mEp->dir)) + hw_ep_flush(mEp->num, mEp->dir); + + mReq->req.status = 0; + + if (mReq->map) { + dma_unmap_single(mEp->device, mReq->req.dma, mReq->req.length, + mEp->dir ? DMA_TO_DEVICE : DMA_FROM_DEVICE); + mReq->req.dma = 0; + mReq->map = 0; + } + + mReq->req.status = mReq->ptr->token & TD_STATUS; + if ((TD_STATUS_ACTIVE & mReq->req.status) != 0) + mReq->req.status = -ECONNRESET; + else if ((TD_STATUS_HALTED & mReq->req.status) != 0) + mReq->req.status = -1; + else if ((TD_STATUS_DT_ERR & mReq->req.status) != 0) + mReq->req.status = -1; + else if ((TD_STATUS_TR_ERR & mReq->req.status) != 0) + mReq->req.status = -1; + + mReq->req.actual = mReq->ptr->token & TD_TOTAL_BYTES; + mReq->req.actual >>= ffs_nr(TD_TOTAL_BYTES); + mReq->req.actual = mReq->req.length - mReq->req.actual; + mReq->req.actual = mReq->req.status ? 0 : mReq->req.actual; + + return mReq->req.actual; +} + +/** + * _ep_nuke: dequeues all endpoint requests + * @mEp: endpoint + * + * This function returns an error code + * Caller must hold lock + */ +static int _ep_nuke(struct ci13xxx_ep *mEp) +__releases(mEp->lock) +__acquires(mEp->lock) +{ + trace("%p", mEp); + + if (mEp == NULL) + return -EINVAL; + + hw_ep_flush(mEp->num, mEp->dir); + + while (!list_empty(&mEp->qh[mEp->dir].queue)) { + + /* pop oldest request */ + struct ci13xxx_req *mReq = \ + list_entry(mEp->qh[mEp->dir].queue.next, + struct ci13xxx_req, queue); + list_del_init(&mReq->queue); + mReq->req.status = -ESHUTDOWN; + + if (!mReq->req.no_interrupt && mReq->req.complete != NULL) { + spin_unlock(mEp->lock); + mReq->req.complete(&mEp->ep, &mReq->req); + spin_lock(mEp->lock); + } + } + return 0; +} + +/** + * _gadget_stop_activity: stops all USB activity, flushes & disables all endpts + * @gadget: gadget + * + * This function returns an error code + * Caller must hold lock + */ +static int _gadget_stop_activity(struct usb_gadget *gadget) +__releases(udc->lock) +__acquires(udc->lock) +{ + struct usb_ep *ep; + struct ci13xxx *udc = container_of(gadget, struct ci13xxx, gadget); + struct ci13xxx_ep *mEp = container_of(gadget->ep0, + struct ci13xxx_ep, ep); + + trace("%p", gadget); + + if (gadget == NULL) + return -EINVAL; + + spin_unlock(udc->lock); + + /* flush all endpoints */ + gadget_for_each_ep(ep, gadget) { + usb_ep_fifo_flush(ep); + } + usb_ep_fifo_flush(gadget->ep0); + + udc->driver->disconnect(gadget); + + /* make sure to disable all endpoints */ + gadget_for_each_ep(ep, gadget) { + usb_ep_disable(ep); + } + usb_ep_disable(gadget->ep0); + + if (mEp->status != NULL) { + usb_ep_free_request(gadget->ep0, mEp->status); + mEp->status = NULL; + } + + spin_lock(udc->lock); + + return 0; +} + +/****************************************************************************** + * ISR block + *****************************************************************************/ +/** + * isr_reset_handler: USB reset interrupt handler + * @udc: UDC device + * + * This function resets USB engine after a bus reset occurred + */ +static void isr_reset_handler(struct ci13xxx *udc) +__releases(udc->lock) +__acquires(udc->lock) +{ + struct ci13xxx_ep *mEp = &udc->ci13xxx_ep[0]; + int retval; + + trace("%p", udc); + + if (udc == NULL) { + err("EINVAL"); + return; + } + + dbg_event(0xFF, "BUS RST", 0); + + retval = _gadget_stop_activity(&udc->gadget); + if (retval) + goto done; + + retval = hw_usb_reset(); + if (retval) + goto done; + + spin_unlock(udc->lock); + retval = usb_ep_enable(&mEp->ep, &ctrl_endpt_desc); + if (!retval) { + mEp->status = usb_ep_alloc_request(&mEp->ep, GFP_KERNEL); + if (mEp->status == NULL) { + usb_ep_disable(&mEp->ep); + retval = -ENOMEM; + } + } + spin_lock(udc->lock); + + done: + if (retval) + err("error: %i", retval); +} + +/** + * isr_get_status_complete: get_status request complete function + * @ep: endpoint + * @req: request handled + * + * Caller must release lock + */ +static void isr_get_status_complete(struct usb_ep *ep, struct usb_request *req) +{ + trace("%p, %p", ep, req); + + if (ep == NULL || req == NULL) { + err("EINVAL"); + return; + } + + kfree(req->buf); + usb_ep_free_request(ep, req); +} + +/** + * isr_get_status_response: get_status request response + * @ep: endpoint + * @setup: setup request packet + * + * This function returns an error code + */ +static int isr_get_status_response(struct ci13xxx_ep *mEp, + struct usb_ctrlrequest *setup) +__releases(mEp->lock) +__acquires(mEp->lock) +{ + struct usb_request *req = NULL; + gfp_t gfp_flags = GFP_ATOMIC; + int dir, num, retval; + + trace("%p, %p", mEp, setup); + + if (mEp == NULL || setup == NULL) + return -EINVAL; + + spin_unlock(mEp->lock); + req = usb_ep_alloc_request(&mEp->ep, gfp_flags); + spin_lock(mEp->lock); + if (req == NULL) + return -ENOMEM; + + req->complete = isr_get_status_complete; + req->length = 2; + req->buf = kzalloc(req->length, gfp_flags); + if (req->buf == NULL) { + retval = -ENOMEM; + goto err_free_req; + } + + if ((setup->bRequestType & USB_RECIP_MASK) == USB_RECIP_DEVICE) { + /* TODO: D1 - Remote Wakeup; D0 - Self Powered */ + retval = 0; + } else if ((setup->bRequestType & USB_RECIP_MASK) \ + == USB_RECIP_ENDPOINT) { + dir = (le16_to_cpu(setup->wIndex) & USB_ENDPOINT_DIR_MASK) ? + TX : RX; + num = le16_to_cpu(setup->wIndex) & USB_ENDPOINT_NUMBER_MASK; + *((u16 *)req->buf) = hw_ep_get_halt(num, dir); + } + /* else do nothing; reserved for future use */ + + spin_unlock(mEp->lock); + retval = usb_ep_queue(&mEp->ep, req, gfp_flags); + spin_lock(mEp->lock); + if (retval) + goto err_free_buf; + + return 0; + + err_free_buf: + kfree(req->buf); + err_free_req: + spin_unlock(mEp->lock); + usb_ep_free_request(&mEp->ep, req); + spin_lock(mEp->lock); + return retval; +} + +/** + * isr_setup_status_phase: queues the status phase of a setup transation + * @mEp: endpoint + * + * This function returns an error code + */ +static int isr_setup_status_phase(struct ci13xxx_ep *mEp) +__releases(mEp->lock) +__acquires(mEp->lock) +{ + int retval; + + trace("%p", mEp); + + /* mEp is always valid & configured */ + + if (mEp->type == USB_ENDPOINT_XFER_CONTROL) + mEp->dir = (mEp->dir == TX) ? RX : TX; + + mEp->status->no_interrupt = 1; + + spin_unlock(mEp->lock); + retval = usb_ep_queue(&mEp->ep, mEp->status, GFP_ATOMIC); + spin_lock(mEp->lock); + + return retval; +} + +/** + * isr_tr_complete_low: transaction complete low level handler + * @mEp: endpoint + * + * This function returns an error code + * Caller must hold lock + */ +static int isr_tr_complete_low(struct ci13xxx_ep *mEp) +__releases(mEp->lock) +__acquires(mEp->lock) +{ + struct ci13xxx_req *mReq; + int retval; + + trace("%p", mEp); + + if (list_empty(&mEp->qh[mEp->dir].queue)) + return -EINVAL; + + /* pop oldest request */ + mReq = list_entry(mEp->qh[mEp->dir].queue.next, + struct ci13xxx_req, queue); + list_del_init(&mReq->queue); + + retval = _hardware_dequeue(mEp, mReq); + if (retval < 0) { + dbg_event(_usb_addr(mEp), "DONE", retval); + goto done; + } + + dbg_done(_usb_addr(mEp), mReq->ptr->token, retval); + + if (!mReq->req.no_interrupt && mReq->req.complete != NULL) { + spin_unlock(mEp->lock); + mReq->req.complete(&mEp->ep, &mReq->req); + spin_lock(mEp->lock); + } + + if (!list_empty(&mEp->qh[mEp->dir].queue)) { + mReq = list_entry(mEp->qh[mEp->dir].queue.next, + struct ci13xxx_req, queue); + _hardware_enqueue(mEp, mReq); + } + + done: + return retval; +} + +/** + * isr_tr_complete_handler: transaction complete interrupt handler + * @udc: UDC descriptor + * + * This function handles traffic events + */ +static void isr_tr_complete_handler(struct ci13xxx *udc) +__releases(udc->lock) +__acquires(udc->lock) +{ + unsigned i; + + trace("%p", udc); + + if (udc == NULL) { + err("EINVAL"); + return; + } + + for (i = 0; i < hw_ep_max; i++) { + struct ci13xxx_ep *mEp = &udc->ci13xxx_ep[i]; + int type, num, err = -EINVAL; + struct usb_ctrlrequest req; + + + if (mEp->desc == NULL) + continue; /* not configured */ + + if ((mEp->dir == RX && hw_test_and_clear_complete(i)) || + (mEp->dir == TX && hw_test_and_clear_complete(i + 16))) { + err = isr_tr_complete_low(mEp); + if (mEp->type == USB_ENDPOINT_XFER_CONTROL) { + if (err > 0) /* needs status phase */ + err = isr_setup_status_phase(mEp); + if (err < 0) { + dbg_event(_usb_addr(mEp), + "ERROR", err); + spin_unlock(udc->lock); + if (usb_ep_set_halt(&mEp->ep)) + err("error: ep_set_halt"); + spin_lock(udc->lock); + } + } + } + + if (mEp->type != USB_ENDPOINT_XFER_CONTROL || + !hw_test_and_clear_setup_status(i)) + continue; + + if (i != 0) { + warn("ctrl traffic received at endpoint"); + continue; + } + + /* read_setup_packet */ + do { + hw_test_and_set_setup_guard(); + memcpy(&req, &mEp->qh[RX].ptr->setup, sizeof(req)); + } while (!hw_test_and_clear_setup_guard()); + + type = req.bRequestType; + + mEp->dir = (type & USB_DIR_IN) ? TX : RX; + + dbg_setup(_usb_addr(mEp), &req); + + switch (req.bRequest) { + case USB_REQ_CLEAR_FEATURE: + if (type != (USB_DIR_OUT|USB_RECIP_ENDPOINT) && + le16_to_cpu(req.wValue) != USB_ENDPOINT_HALT) + goto delegate; + if (req.wLength != 0) + break; + num = le16_to_cpu(req.wIndex); + num &= USB_ENDPOINT_NUMBER_MASK; + if (!udc->ci13xxx_ep[num].wedge) { + spin_unlock(udc->lock); + err = usb_ep_clear_halt( + &udc->ci13xxx_ep[num].ep); + spin_lock(udc->lock); + if (err) + break; + } + err = isr_setup_status_phase(mEp); + break; + case USB_REQ_GET_STATUS: + if (type != (USB_DIR_IN|USB_RECIP_DEVICE) && + type != (USB_DIR_IN|USB_RECIP_ENDPOINT) && + type != (USB_DIR_IN|USB_RECIP_INTERFACE)) + goto delegate; + if (le16_to_cpu(req.wLength) != 2 || + le16_to_cpu(req.wValue) != 0) + break; + err = isr_get_status_response(mEp, &req); + break; + case USB_REQ_SET_ADDRESS: + if (type != (USB_DIR_OUT|USB_RECIP_DEVICE)) + goto delegate; + if (le16_to_cpu(req.wLength) != 0 || + le16_to_cpu(req.wIndex) != 0) + break; + err = hw_usb_set_address((u8)le16_to_cpu(req.wValue)); + if (err) + break; + err = isr_setup_status_phase(mEp); + break; + case USB_REQ_SET_FEATURE: + if (type != (USB_DIR_OUT|USB_RECIP_ENDPOINT) && + le16_to_cpu(req.wValue) != USB_ENDPOINT_HALT) + goto delegate; + if (req.wLength != 0) + break; + num = le16_to_cpu(req.wIndex); + num &= USB_ENDPOINT_NUMBER_MASK; + + spin_unlock(udc->lock); + err = usb_ep_set_halt(&udc->ci13xxx_ep[num].ep); + spin_lock(udc->lock); + if (err) + break; + err = isr_setup_status_phase(mEp); + break; + default: +delegate: + if (req.wLength == 0) /* no data phase */ + mEp->dir = TX; + + spin_unlock(udc->lock); + err = udc->driver->setup(&udc->gadget, &req); + spin_lock(udc->lock); + break; + } + + if (err < 0) { + dbg_event(_usb_addr(mEp), "ERROR", err); + + spin_unlock(udc->lock); + if (usb_ep_set_halt(&mEp->ep)) + err("error: ep_set_halt"); + spin_lock(udc->lock); + } + } +} + +/****************************************************************************** + * ENDPT block + *****************************************************************************/ +/** + * ep_enable: configure endpoint, making it usable + * + * Check usb_ep_enable() at "usb_gadget.h" for details + */ +static int ep_enable(struct usb_ep *ep, + const struct usb_endpoint_descriptor *desc) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + int direction, retval = 0; + unsigned long flags; + + trace("%p, %p", ep, desc); + + if (ep == NULL || desc == NULL) + return -EINVAL; + + spin_lock_irqsave(mEp->lock, flags); + + /* only internal SW should enable ctrl endpts */ + + mEp->desc = desc; + + if (!list_empty(&mEp->qh[mEp->dir].queue)) + warn("enabling a non-empty endpoint!"); + + mEp->dir = (desc->bEndpointAddress & USB_ENDPOINT_DIR_MASK) ? TX : RX; + mEp->num = desc->bEndpointAddress & USB_ENDPOINT_NUMBER_MASK; + mEp->type = desc->bmAttributes & USB_ENDPOINT_XFERTYPE_MASK; + + mEp->ep.maxpacket = __constant_le16_to_cpu(desc->wMaxPacketSize); + + direction = mEp->dir; + do { + dbg_event(_usb_addr(mEp), "ENABLE", 0); + + if (mEp->type == USB_ENDPOINT_XFER_CONTROL) + mEp->qh[mEp->dir].ptr->cap |= QH_IOS; + else if (mEp->type == USB_ENDPOINT_XFER_ISOC) + mEp->qh[mEp->dir].ptr->cap &= ~QH_MULT; + else + mEp->qh[mEp->dir].ptr->cap &= ~QH_ZLT; + + mEp->qh[mEp->dir].ptr->cap |= + (mEp->ep.maxpacket << ffs_nr(QH_MAX_PKT)) & QH_MAX_PKT; + mEp->qh[mEp->dir].ptr->td.next |= TD_TERMINATE; /* needed? */ + + retval |= hw_ep_enable(mEp->num, mEp->dir, mEp->type); + + if (mEp->type == USB_ENDPOINT_XFER_CONTROL) + mEp->dir = (mEp->dir == TX) ? RX : TX; + + } while (mEp->dir != direction); + + spin_unlock_irqrestore(mEp->lock, flags); + return retval; +} + +/** + * ep_disable: endpoint is no longer usable + * + * Check usb_ep_disable() at "usb_gadget.h" for details + */ +static int ep_disable(struct usb_ep *ep) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + int direction, retval = 0; + unsigned long flags; + + trace("%p", ep); + + if (ep == NULL) + return -EINVAL; + else if (mEp->desc == NULL) + return -EBUSY; + + spin_lock_irqsave(mEp->lock, flags); + + /* only internal SW should disable ctrl endpts */ + + direction = mEp->dir; + do { + dbg_event(_usb_addr(mEp), "DISABLE", 0); + + retval |= _ep_nuke(mEp); + retval |= hw_ep_disable(mEp->num, mEp->dir); + + if (mEp->type == USB_ENDPOINT_XFER_CONTROL) + mEp->dir = (mEp->dir == TX) ? RX : TX; + + } while (mEp->dir != direction); + + mEp->desc = NULL; + + spin_unlock_irqrestore(mEp->lock, flags); + return retval; +} + +/** + * ep_alloc_request: allocate a request object to use with this endpoint + * + * Check usb_ep_alloc_request() at "usb_gadget.h" for details + */ +static struct usb_request *ep_alloc_request(struct usb_ep *ep, gfp_t gfp_flags) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + struct ci13xxx_req *mReq = NULL; + unsigned long flags; + + trace("%p, %i", ep, gfp_flags); + + if (ep == NULL) { + err("EINVAL"); + return NULL; + } + + spin_lock_irqsave(mEp->lock, flags); + + mReq = kzalloc(sizeof(struct ci13xxx_req), gfp_flags); + if (mReq != NULL) { + INIT_LIST_HEAD(&mReq->queue); + + mReq->ptr = dma_pool_alloc(mEp->td_pool, gfp_flags, + &mReq->dma); + if (mReq->ptr == NULL) { + kfree(mReq); + mReq = NULL; + } + } + + dbg_event(_usb_addr(mEp), "ALLOC", mReq == NULL); + + spin_unlock_irqrestore(mEp->lock, flags); + + return (mReq == NULL) ? NULL : &mReq->req; +} + +/** + * ep_free_request: frees a request object + * + * Check usb_ep_free_request() at "usb_gadget.h" for details + */ +static void ep_free_request(struct usb_ep *ep, struct usb_request *req) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + struct ci13xxx_req *mReq = container_of(req, struct ci13xxx_req, req); + unsigned long flags; + + trace("%p, %p", ep, req); + + if (ep == NULL || req == NULL) { + err("EINVAL"); + return; + } else if (!list_empty(&mReq->queue)) { + err("EBUSY"); + return; + } + + spin_lock_irqsave(mEp->lock, flags); + + if (mReq->ptr) + dma_pool_free(mEp->td_pool, mReq->ptr, mReq->dma); + kfree(mReq); + + dbg_event(_usb_addr(mEp), "FREE", 0); + + spin_unlock_irqrestore(mEp->lock, flags); +} + +/** + * ep_queue: queues (submits) an I/O request to an endpoint + * + * Check usb_ep_queue()* at usb_gadget.h" for details + */ +static int ep_queue(struct usb_ep *ep, struct usb_request *req, + gfp_t __maybe_unused gfp_flags) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + struct ci13xxx_req *mReq = container_of(req, struct ci13xxx_req, req); + int retval = 0; + unsigned long flags; + + trace("%p, %p, %X", ep, req, gfp_flags); + + if (ep == NULL || req == NULL || mEp->desc == NULL) + return -EINVAL; + + spin_lock_irqsave(mEp->lock, flags); + + if (mEp->type == USB_ENDPOINT_XFER_CONTROL && + !list_empty(&mEp->qh[mEp->dir].queue)) { + _ep_nuke(mEp); + retval = -EOVERFLOW; + warn("endpoint ctrl %X nuked", _usb_addr(mEp)); + } + + /* first nuke then test link, e.g. previous status has not sent */ + if (!list_empty(&mReq->queue)) { + retval = -EBUSY; + err("request already in queue"); + goto done; + } + + if (req->length > (4 * PAGE_SIZE)) { + req->length = (4 * PAGE_SIZE); + retval = -EMSGSIZE; + warn("request length truncated"); + } + + dbg_queue(_usb_addr(mEp), req, retval); + + /* push request */ + mReq->req.status = -EINPROGRESS; + mReq->req.actual = 0; + list_add_tail(&mReq->queue, &mEp->qh[mEp->dir].queue); + + retval = _hardware_enqueue(mEp, mReq); + if (retval == -EALREADY || retval == -EBUSY) { + dbg_event(_usb_addr(mEp), "QUEUE", retval); + retval = 0; + } + + done: + spin_unlock_irqrestore(mEp->lock, flags); + return retval; +} + +/** + * ep_dequeue: dequeues (cancels, unlinks) an I/O request from an endpoint + * + * Check usb_ep_dequeue() at "usb_gadget.h" for details + */ +static int ep_dequeue(struct usb_ep *ep, struct usb_request *req) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + struct ci13xxx_req *mReq = container_of(req, struct ci13xxx_req, req); + unsigned long flags; + + trace("%p, %p", ep, req); + + if (ep == NULL || req == NULL || mEp->desc == NULL || + list_empty(&mReq->queue) || list_empty(&mEp->qh[mEp->dir].queue)) + return -EINVAL; + + spin_lock_irqsave(mEp->lock, flags); + + dbg_event(_usb_addr(mEp), "DEQUEUE", 0); + + if (mReq->req.status == -EALREADY) + _hardware_dequeue(mEp, mReq); + + /* pop request */ + list_del_init(&mReq->queue); + req->status = -ECONNRESET; + + if (!mReq->req.no_interrupt && mReq->req.complete != NULL) { + spin_unlock(mEp->lock); + mReq->req.complete(&mEp->ep, &mReq->req); + spin_lock(mEp->lock); + } + + spin_unlock_irqrestore(mEp->lock, flags); + return 0; +} + +/** + * ep_set_halt: sets the endpoint halt feature + * + * Check usb_ep_set_halt() at "usb_gadget.h" for details + */ +static int ep_set_halt(struct usb_ep *ep, int value) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + int direction, retval = 0; + unsigned long flags; + + trace("%p, %i", ep, value); + + if (ep == NULL || mEp->desc == NULL) + return -EINVAL; + + spin_lock_irqsave(mEp->lock, flags); + +#ifndef STALL_IN + /* g_file_storage MS compliant but g_zero fails chapter 9 compliance */ + if (value && mEp->type == USB_ENDPOINT_XFER_BULK && mEp->dir == TX && + !list_empty(&mEp->qh[mEp->dir].queue)) { + spin_unlock_irqrestore(mEp->lock, flags); + return -EAGAIN; + } +#endif + + direction = mEp->dir; + do { + dbg_event(_usb_addr(mEp), "HALT", value); + retval |= hw_ep_set_halt(mEp->num, mEp->dir, value); + + if (!value) + mEp->wedge = 0; + + if (mEp->type == USB_ENDPOINT_XFER_CONTROL) + mEp->dir = (mEp->dir == TX) ? RX : TX; + + } while (mEp->dir != direction); + + spin_unlock_irqrestore(mEp->lock, flags); + return retval; +} + +/** + * ep_set_wedge: sets the halt feature and ignores clear requests + * + * Check usb_ep_set_wedge() at "usb_gadget.h" for details + */ +static int ep_set_wedge(struct usb_ep *ep) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + unsigned long flags; + + trace("%p", ep); + + if (ep == NULL || mEp->desc == NULL) + return -EINVAL; + + spin_lock_irqsave(mEp->lock, flags); + + dbg_event(_usb_addr(mEp), "WEDGE", 0); + mEp->wedge = 1; + + spin_unlock_irqrestore(mEp->lock, flags); + + return usb_ep_set_halt(ep); +} + +/** + * ep_fifo_flush: flushes contents of a fifo + * + * Check usb_ep_fifo_flush() at "usb_gadget.h" for details + */ +static void ep_fifo_flush(struct usb_ep *ep) +{ + struct ci13xxx_ep *mEp = container_of(ep, struct ci13xxx_ep, ep); + unsigned long flags; + + trace("%p", ep); + + if (ep == NULL) { + err("%02X: -EINVAL", _usb_addr(mEp)); + return; + } + + spin_lock_irqsave(mEp->lock, flags); + + dbg_event(_usb_addr(mEp), "FFLUSH", 0); + hw_ep_flush(mEp->num, mEp->dir); + + spin_unlock_irqrestore(mEp->lock, flags); +} + +/** + * Endpoint-specific part of the API to the USB controller hardware + * Check "usb_gadget.h" for details + */ +static const struct usb_ep_ops usb_ep_ops = { + .enable = ep_enable, + .disable = ep_disable, + .alloc_request = ep_alloc_request, + .free_request = ep_free_request, + .queue = ep_queue, + .dequeue = ep_dequeue, + .set_halt = ep_set_halt, + .set_wedge = ep_set_wedge, + .fifo_flush = ep_fifo_flush, +}; + +/****************************************************************************** + * GADGET block + *****************************************************************************/ +/** + * Device operations part of the API to the USB controller hardware, + * which don't involve endpoints (or i/o) + * Check "usb_gadget.h" for details + */ +static const struct usb_gadget_ops usb_gadget_ops; + +/** + * usb_gadget_register_driver: register a gadget driver + * + * Check usb_gadget_register_driver() at "usb_gadget.h" for details + * Interrupts are enabled here + */ +int usb_gadget_register_driver(struct usb_gadget_driver *driver) +{ + struct ci13xxx *udc = _udc; + unsigned long i, k, flags; + int retval = -ENOMEM; + + trace("%p", driver); + + if (driver == NULL || + driver->bind == NULL || + driver->unbind == NULL || + driver->setup == NULL || + driver->disconnect == NULL || + driver->suspend == NULL || + driver->resume == NULL) + return -EINVAL; + else if (udc == NULL) + return -ENODEV; + else if (udc->driver != NULL) + return -EBUSY; + + /* alloc resources */ + udc->qh_pool = dma_pool_create("ci13xxx_qh", &udc->gadget.dev, + sizeof(struct ci13xxx_qh), + 64, PAGE_SIZE); + if (udc->qh_pool == NULL) + return -ENOMEM; + + udc->td_pool = dma_pool_create("ci13xxx_td", &udc->gadget.dev, + sizeof(struct ci13xxx_td), + 64, PAGE_SIZE); + if (udc->td_pool == NULL) { + dma_pool_destroy(udc->qh_pool); + udc->qh_pool = NULL; + return -ENOMEM; + } + + spin_lock_irqsave(udc->lock, flags); + + info("hw_ep_max = %d", hw_ep_max); + + udc->driver = driver; + udc->gadget.ops = NULL; + udc->gadget.dev.driver = NULL; + + retval = 0; + for (i = 0; i < hw_ep_max; i++) { + struct ci13xxx_ep *mEp = &udc->ci13xxx_ep[i]; + + scnprintf(mEp->name, sizeof(mEp->name), "ep%i", (int)i); + + mEp->lock = udc->lock; + mEp->device = &udc->gadget.dev; + mEp->td_pool = udc->td_pool; + + mEp->ep.name = mEp->name; + mEp->ep.ops = &usb_ep_ops; + mEp->ep.maxpacket = CTRL_PAYLOAD_MAX; + + /* this allocation cannot be random */ + for (k = RX; k <= TX; k++) { + INIT_LIST_HEAD(&mEp->qh[k].queue); + mEp->qh[k].ptr = dma_pool_alloc(udc->qh_pool, + GFP_KERNEL, + &mEp->qh[k].dma); + if (mEp->qh[k].ptr == NULL) + retval = -ENOMEM; + else + memset(mEp->qh[k].ptr, 0, + sizeof(*mEp->qh[k].ptr)); + } + if (i == 0) + udc->gadget.ep0 = &mEp->ep; + else + list_add_tail(&mEp->ep.ep_list, &udc->gadget.ep_list); + } + if (retval) + goto done; + + /* bind gadget */ + driver->driver.bus = NULL; + udc->gadget.ops = &usb_gadget_ops; + udc->gadget.dev.driver = &driver->driver; + + spin_unlock_irqrestore(udc->lock, flags); + retval = driver->bind(&udc->gadget); /* MAY SLEEP */ + spin_lock_irqsave(udc->lock, flags); + + if (retval) { + udc->gadget.ops = NULL; + udc->gadget.dev.driver = NULL; + goto done; + } + + retval = hw_device_state(udc->ci13xxx_ep[0].qh[RX].dma); + + done: + spin_unlock_irqrestore(udc->lock, flags); + if (retval) + usb_gadget_unregister_driver(driver); + return retval; +} +EXPORT_SYMBOL(usb_gadget_register_driver); + +/** + * usb_gadget_unregister_driver: unregister a gadget driver + * + * Check usb_gadget_unregister_driver() at "usb_gadget.h" for details + */ +int usb_gadget_unregister_driver(struct usb_gadget_driver *driver) +{ + struct ci13xxx *udc = _udc; + unsigned long i, k, flags; + + trace("%p", driver); + + if (driver == NULL || + driver->bind == NULL || + driver->unbind == NULL || + driver->setup == NULL || + driver->disconnect == NULL || + driver->suspend == NULL || + driver->resume == NULL || + driver != udc->driver) + return -EINVAL; + + spin_lock_irqsave(udc->lock, flags); + + hw_device_state(0); + + /* unbind gadget */ + if (udc->gadget.ops != NULL) { + _gadget_stop_activity(&udc->gadget); + + spin_unlock_irqrestore(udc->lock, flags); + driver->unbind(&udc->gadget); /* MAY SLEEP */ + spin_lock_irqsave(udc->lock, flags); + + udc->gadget.ops = NULL; + udc->gadget.dev.driver = NULL; + } + + /* free resources */ + for (i = 0; i < hw_ep_max; i++) { + struct ci13xxx_ep *mEp = &udc->ci13xxx_ep[i]; + + if (i == 0) + udc->gadget.ep0 = NULL; + else if (!list_empty(&mEp->ep.ep_list)) + list_del_init(&mEp->ep.ep_list); + + for (k = RX; k <= TX; k++) + if (mEp->qh[k].ptr != NULL) + dma_pool_free(udc->qh_pool, + mEp->qh[k].ptr, mEp->qh[k].dma); + } + + udc->driver = NULL; + + spin_unlock_irqrestore(udc->lock, flags); + + if (udc->td_pool != NULL) { + dma_pool_destroy(udc->td_pool); + udc->td_pool = NULL; + } + if (udc->qh_pool != NULL) { + dma_pool_destroy(udc->qh_pool); + udc->qh_pool = NULL; + } + + return 0; +} +EXPORT_SYMBOL(usb_gadget_unregister_driver); + +/****************************************************************************** + * BUS block + *****************************************************************************/ +/** + * udc_irq: global interrupt handler + * + * This function returns IRQ_HANDLED if the IRQ has been handled + * It locks access to registers + */ +static irqreturn_t udc_irq(void) +{ + struct ci13xxx *udc = _udc; + irqreturn_t retval; + u32 intr; + + trace(); + + if (udc == NULL) { + err("ENODEV"); + return IRQ_HANDLED; + } + + spin_lock(udc->lock); + intr = hw_test_and_clear_intr_active(); + if (intr) { + isr_statistics.hndl.buf[isr_statistics.hndl.idx++] = intr; + isr_statistics.hndl.idx &= ISR_MASK; + isr_statistics.hndl.cnt++; + + /* order defines priority - do NOT change it */ + if (USBi_URI & intr) { + isr_statistics.uri++; + isr_reset_handler(udc); + } + if (USBi_PCI & intr) { + isr_statistics.pci++; + udc->gadget.speed = hw_port_is_high_speed() ? + USB_SPEED_HIGH : USB_SPEED_FULL; + } + if (USBi_UEI & intr) + isr_statistics.uei++; + if (USBi_UI & intr) { + isr_statistics.ui++; + isr_tr_complete_handler(udc); + } + if (USBi_SLI & intr) + isr_statistics.sli++; + retval = IRQ_HANDLED; + } else { + isr_statistics.none++; + retval = IRQ_NONE; + } + spin_unlock(udc->lock); + + return retval; +} + +/** + * udc_release: driver release function + * @dev: device + * + * Currently does nothing + */ +static void udc_release(struct device *dev) +{ + trace("%p", dev); + + if (dev == NULL) + err("EINVAL"); +} + +/** + * udc_probe: parent probe must call this to initialize UDC + * @dev: parent device + * @regs: registers base address + * @name: driver name + * + * This function returns an error code + * No interrupts active, the IRQ has not been requested yet + * Kernel assumes 32-bit DMA operations by default, no need to dma_set_mask + */ +static int udc_probe(struct device *dev, void __iomem *regs, const char *name) +{ + struct ci13xxx *udc; + int retval = 0; + + trace("%p, %p, %p", dev, regs, name); + + if (dev == NULL || regs == NULL || name == NULL) + return -EINVAL; + + udc = kzalloc(sizeof(struct ci13xxx), GFP_KERNEL); + if (udc == NULL) + return -ENOMEM; + + udc->lock = &udc_lock; + + retval = hw_device_reset(regs); + if (retval) + goto done; + + udc->gadget.ops = NULL; + udc->gadget.speed = USB_SPEED_UNKNOWN; + udc->gadget.is_dualspeed = 1; + udc->gadget.is_otg = 0; + udc->gadget.name = name; + + INIT_LIST_HEAD(&udc->gadget.ep_list); + udc->gadget.ep0 = NULL; + + strcpy(udc->gadget.dev.bus_id, "gadget"); + udc->gadget.dev.dma_mask = dev->dma_mask; + udc->gadget.dev.parent = dev; + udc->gadget.dev.release = udc_release; + + retval = device_register(&udc->gadget.dev); + if (retval) + goto done; + +#ifdef CONFIG_USB_GADGET_DEBUG_FILES + retval = dbg_create_files(&udc->gadget.dev); +#endif + if (retval) { + device_unregister(&udc->gadget.dev); + goto done; + } + + _udc = udc; + return retval; + + done: + err("error = %i", retval); + kfree(udc); + _udc = NULL; + return retval; +} + +/** + * udc_remove: parent remove must call this to remove UDC + * + * No interrupts active, the IRQ has been released + */ +static void udc_remove(void) +{ + struct ci13xxx *udc = _udc; + + if (udc == NULL) { + err("EINVAL"); + return; + } + +#ifdef CONFIG_USB_GADGET_DEBUG_FILES + dbg_remove_files(&udc->gadget.dev); +#endif + device_unregister(&udc->gadget.dev); + + kfree(udc); + _udc = NULL; +} + +/****************************************************************************** + * PCI block + *****************************************************************************/ +/** + * ci13xxx_pci_irq: interrut handler + * @irq: irq number + * @pdev: USB Device Controller interrupt source + * + * This function returns IRQ_HANDLED if the IRQ has been handled + * This is an ISR don't trace, use attribute interface instead + */ +static irqreturn_t ci13xxx_pci_irq(int irq, void *pdev) +{ + if (irq == 0) { + dev_err(&((struct pci_dev *)pdev)->dev, "Invalid IRQ0 usage!"); + return IRQ_HANDLED; + } + return udc_irq(); +} + +/** + * ci13xxx_pci_probe: PCI probe + * @pdev: USB device controller being probed + * @id: PCI hotplug ID connecting controller to UDC framework + * + * This function returns an error code + * Allocates basic PCI resources for this USB device controller, and then + * invokes the udc_probe() method to start the UDC associated with it + */ +static int __devinit ci13xxx_pci_probe(struct pci_dev *pdev, + const struct pci_device_id *id) +{ + void __iomem *regs = NULL; + int retval = 0; + + if (id == NULL) + return -EINVAL; + + retval = pci_enable_device(pdev); + if (retval) + goto done; + + if (!pdev->irq) { + dev_err(&pdev->dev, "No IRQ, check BIOS/PCI setup!"); + retval = -ENODEV; + goto disable_device; + } + + retval = pci_request_regions(pdev, UDC_DRIVER_NAME); + if (retval) + goto disable_device; + + /* BAR 0 holds all the registers */ + regs = pci_iomap(pdev, 0, 0); + if (!regs) { + dev_err(&pdev->dev, "Error mapping memory!"); + retval = -EFAULT; + goto release_regions; + } + pci_set_drvdata(pdev, (__force void *)regs); + + pci_set_master(pdev); + pci_try_set_mwi(pdev); + + retval = udc_probe(&pdev->dev, regs, UDC_DRIVER_NAME); + if (retval) + goto iounmap; + + /* our device does not have MSI capability */ + + retval = request_irq(pdev->irq, ci13xxx_pci_irq, IRQF_SHARED, + UDC_DRIVER_NAME, pdev); + if (retval) + goto gadget_remove; + + return 0; + + gadget_remove: + udc_remove(); + iounmap: + pci_iounmap(pdev, regs); + release_regions: + pci_release_regions(pdev); + disable_device: + pci_disable_device(pdev); + done: + return retval; +} + +/** + * ci13xxx_pci_remove: PCI remove + * @pdev: USB Device Controller being removed + * + * Reverses the effect of ci13xxx_pci_probe(), + * first invoking the udc_remove() and then releases + * all PCI resources allocated for this USB device controller + */ +static void __devexit ci13xxx_pci_remove(struct pci_dev *pdev) +{ + free_irq(pdev->irq, pdev); + udc_remove(); + pci_iounmap(pdev, (__force void __iomem *)pci_get_drvdata(pdev)); + pci_release_regions(pdev); + pci_disable_device(pdev); +} + +/** + * PCI device table + * PCI device structure + * + * Check "pci.h" for details + */ +static DEFINE_PCI_DEVICE_TABLE(ci13xxx_pci_id_table) = { + { PCI_DEVICE(0x153F, 0x1004) }, + { PCI_DEVICE(0x153F, 0x1006) }, + { 0, 0, 0, 0, 0, 0, 0 /* end: all zeroes */ } +}; +MODULE_DEVICE_TABLE(pci, ci13xxx_pci_id_table); + +static struct pci_driver ci13xxx_pci_driver = { + .name = UDC_DRIVER_NAME, + .id_table = ci13xxx_pci_id_table, + .probe = ci13xxx_pci_probe, + .remove = __devexit_p(ci13xxx_pci_remove), +}; + +/** + * ci13xxx_pci_init: module init + * + * Driver load + */ +static int __init ci13xxx_pci_init(void) +{ + return pci_register_driver(&ci13xxx_pci_driver); +} +module_init(ci13xxx_pci_init); + +/** + * ci13xxx_pci_exit: module exit + * + * Driver unload + */ +static void __exit ci13xxx_pci_exit(void) +{ + pci_unregister_driver(&ci13xxx_pci_driver); +} +module_exit(ci13xxx_pci_exit); + +MODULE_AUTHOR("MIPS - David Lopo "); +MODULE_DESCRIPTION("MIPS CI13XXX USB Peripheral Controller"); +MODULE_LICENSE("GPL"); +MODULE_VERSION("June 2008"); diff --git a/drivers/usb/gadget/ci13xxx_udc.h b/drivers/usb/gadget/ci13xxx_udc.h new file mode 100644 index 00000000000..4026e9cede3 --- /dev/null +++ b/drivers/usb/gadget/ci13xxx_udc.h @@ -0,0 +1,195 @@ +/* + * ci13xxx_udc.h - structures, registers, and macros MIPS USB IP core + * + * Copyright (C) 2008 Chipidea - MIPS Technologies, Inc. All rights reserved. + * + * Author: David Lopo + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + * + * Description: MIPS USB IP core family device controller + * Structures, registers and logging macros + */ + +#ifndef _CI13XXX_h_ +#define _CI13XXX_h_ + +/****************************************************************************** + * DEFINE + *****************************************************************************/ +#define ENDPT_MAX (16) +#define CTRL_PAYLOAD_MAX (64) +#define RX (0) /* similar to USB_DIR_OUT but can be used as an index */ +#define TX (1) /* similar to USB_DIR_IN but can be used as an index */ + +/****************************************************************************** + * STRUCTURES + *****************************************************************************/ +/* DMA layout of transfer descriptors */ +struct ci13xxx_td { + /* 0 */ + u32 next; +#define TD_TERMINATE BIT(0) + /* 1 */ + u32 token; +#define TD_STATUS (0x00FFUL << 0) +#define TD_STATUS_TR_ERR BIT(3) +#define TD_STATUS_DT_ERR BIT(5) +#define TD_STATUS_HALTED BIT(6) +#define TD_STATUS_ACTIVE BIT(7) +#define TD_MULTO (0x0003UL << 10) +#define TD_IOC BIT(15) +#define TD_TOTAL_BYTES (0x7FFFUL << 16) + /* 2 */ + u32 page[5]; +#define TD_CURR_OFFSET (0x0FFFUL << 0) +#define TD_FRAME_NUM (0x07FFUL << 0) +#define TD_RESERVED_MASK (0x0FFFUL << 0) +} __attribute__ ((packed)); + +/* DMA layout of queue heads */ +struct ci13xxx_qh { + /* 0 */ + u32 cap; +#define QH_IOS BIT(15) +#define QH_MAX_PKT (0x07FFUL << 16) +#define QH_ZLT BIT(29) +#define QH_MULT (0x0003UL << 30) + /* 1 */ + u32 curr; + /* 2 - 8 */ + struct ci13xxx_td td; + /* 9 */ + u32 RESERVED; + struct usb_ctrlrequest setup; +} __attribute__ ((packed)); + +/* Extension of usb_request */ +struct ci13xxx_req { + struct usb_request req; + unsigned map; + struct list_head queue; + struct ci13xxx_td *ptr; + dma_addr_t dma; +}; + +/* Extension of usb_ep */ +struct ci13xxx_ep { + struct usb_ep ep; + const struct usb_endpoint_descriptor *desc; + u8 dir; + u8 num; + u8 type; + char name[16]; + struct { + struct list_head queue; + struct ci13xxx_qh *ptr; + dma_addr_t dma; + } qh[2]; + struct usb_request *status; + int wedge; + + /* global resources */ + spinlock_t *lock; + struct device *device; + struct dma_pool *td_pool; +}; + +/* CI13XXX UDC descriptor & global resources */ +struct ci13xxx { + spinlock_t *lock; /* ctrl register bank access */ + + struct dma_pool *qh_pool; /* DMA pool for queue heads */ + struct dma_pool *td_pool; /* DMA pool for transfer descs */ + + struct usb_gadget gadget; /* USB slave device */ + struct ci13xxx_ep ci13xxx_ep[ENDPT_MAX]; /* extended endpts */ + + struct usb_gadget_driver *driver; /* 3rd party gadget driver */ +}; + +/****************************************************************************** + * REGISTERS + *****************************************************************************/ +/* register size */ +#define REG_BITS (32) + +/* HCCPARAMS */ +#define HCCPARAMS_LEN BIT(17) + +/* DCCPARAMS */ +#define DCCPARAMS_DEN (0x1F << 0) +#define DCCPARAMS_DC BIT(7) + +/* TESTMODE */ +#define TESTMODE_FORCE BIT(0) + +/* USBCMD */ +#define USBCMD_RS BIT(0) +#define USBCMD_RST BIT(1) +#define USBCMD_SUTW BIT(13) + +/* USBSTS & USBINTR */ +#define USBi_UI BIT(0) +#define USBi_UEI BIT(1) +#define USBi_PCI BIT(2) +#define USBi_URI BIT(6) +#define USBi_SLI BIT(8) + +/* DEVICEADDR */ +#define DEVICEADDR_USBADRA BIT(24) +#define DEVICEADDR_USBADR (0x7FUL << 25) + +/* PORTSC */ +#define PORTSC_SUSP BIT(7) +#define PORTSC_HSP BIT(9) +#define PORTSC_PTC (0x0FUL << 16) + +/* DEVLC */ +#define DEVLC_PSPD (0x03UL << 25) +#define DEVLC_PSPD_HS (0x02UL << 25) + +/* USBMODE */ +#define USBMODE_CM (0x03UL << 0) +#define USBMODE_CM_IDLE (0x00UL << 0) +#define USBMODE_CM_DEVICE (0x02UL << 0) +#define USBMODE_CM_HOST (0x03UL << 0) +#define USBMODE_SLOM BIT(3) + +/* ENDPTCTRL */ +#define ENDPTCTRL_RXS BIT(0) +#define ENDPTCTRL_RXT (0x03UL << 2) +#define ENDPTCTRL_RXR BIT(6) /* reserved for port 0 */ +#define ENDPTCTRL_RXE BIT(7) +#define ENDPTCTRL_TXS BIT(16) +#define ENDPTCTRL_TXT (0x03UL << 18) +#define ENDPTCTRL_TXR BIT(22) /* reserved for port 0 */ +#define ENDPTCTRL_TXE BIT(23) + +/****************************************************************************** + * LOGGING + *****************************************************************************/ +#define ci13xxx_printk(level, format, args...) \ +do { \ + if (_udc == NULL) \ + printk(level "[%s] " format "\n", __func__, ## args); \ + else \ + dev_printk(level, _udc->gadget.dev.parent, \ + "[%s] " format "\n", __func__, ## args); \ +} while (0) + +#define err(format, args...) ci13xxx_printk(KERN_ERR, format, ## args) +#define warn(format, args...) ci13xxx_printk(KERN_WARNING, format, ## args) +#define info(format, args...) ci13xxx_printk(KERN_INFO, format, ## args) + +#ifdef TRACE +#define trace(format, args...) ci13xxx_printk(KERN_DEBUG, format, ## args) +#define dbg_trace(format, args...) dev_dbg(dev, format, ##args) +#else +#define trace(format, args...) do {} while (0) +#define dbg_trace(format, args...) do {} while (0) +#endif + +#endif /* _CI13XXX_h_ */ diff --git a/drivers/usb/gadget/gadget_chips.h b/drivers/usb/gadget/gadget_chips.h index 4e3107dd2f3..c0679b17602 100644 --- a/drivers/usb/gadget/gadget_chips.h +++ b/drivers/usb/gadget/gadget_chips.h @@ -158,6 +158,11 @@ #define gadget_is_fsl_qe(g) 0 #endif +#ifdef CONFIG_USB_GADGET_CI13XXX +#define gadget_is_ci13xxx(g) (!strcmp("ci13xxx_udc", (g)->name)) +#else +#define gadget_is_ci13xxx(g) 0 +#endif // CONFIG_USB_GADGET_SX2 // CONFIG_USB_GADGET_AU1X00 @@ -225,6 +230,8 @@ static inline int usb_gadget_controller_number(struct usb_gadget *gadget) return 0x21; else if (gadget_is_fsl_qe(gadget)) return 0x22; + else if (gadget_is_ci13xxx(gadget)) + return 0x23; return -ENOENT; } From 96a274d1da950a96cb31ac1bed044e049d770980 Mon Sep 17 00:00:00 2001 From: David Brownell Date: Mon, 24 Nov 2008 13:06:47 +0200 Subject: [PATCH 043/123] USB: musb: sysfs mode updates Fix three omissions in the "mode" sysfs attribute support: (a) inability to report errors; (b) no DaVinci support ... just report an error; (c) for omap2430, accepting unsupportable values The 2430 stuff is still odd.... Signed-off-by: David Brownell Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/davinci.c | 6 ++++++ drivers/usb/musb/musb_core.c | 17 ++++++++++------- drivers/usb/musb/musb_core.h | 2 +- drivers/usb/musb/omap2430.c | 11 ++++++++++- drivers/usb/musb/tusb6010.c | 7 +++++-- 5 files changed, 32 insertions(+), 11 deletions(-) diff --git a/drivers/usb/musb/davinci.c b/drivers/usb/musb/davinci.c index dfb3bcbe00f..87aef82fbcd 100644 --- a/drivers/usb/musb/davinci.c +++ b/drivers/usb/musb/davinci.c @@ -364,6 +364,12 @@ static irqreturn_t davinci_interrupt(int irq, void *__hci) return IRQ_HANDLED; } +int musb_platform_set_mode(struct musb *musb, u8 mode) +{ + /* EVM can't do this (right?) */ + return -EIO; +} + int __init musb_platform_init(struct musb *musb) { void __iomem *tibase = musb->ctrl_base; diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c index 0b742ebe5c8..de7f8bec6bd 100644 --- a/drivers/usb/musb/musb_core.c +++ b/drivers/usb/musb/musb_core.c @@ -1671,17 +1671,20 @@ musb_mode_store(struct device *dev, struct device_attribute *attr, { struct musb *musb = dev_to_musb(dev); unsigned long flags; + int status; spin_lock_irqsave(&musb->lock, flags); - if (!strncmp(buf, "host", 4)) - musb_platform_set_mode(musb, MUSB_HOST); - if (!strncmp(buf, "peripheral", 10)) - musb_platform_set_mode(musb, MUSB_PERIPHERAL); - if (!strncmp(buf, "otg", 3)) - musb_platform_set_mode(musb, MUSB_OTG); + if (sysfs_streq(buf, "host")) + status = musb_platform_set_mode(musb, MUSB_HOST); + else if (sysfs_streq(buf, "peripheral")) + status = musb_platform_set_mode(musb, MUSB_PERIPHERAL); + else if (sysfs_streq(buf, "otg")) + status = musb_platform_set_mode(musb, MUSB_OTG); + else + status = -EINVAL; spin_unlock_irqrestore(&musb->lock, flags); - return n; + return (status == 0) ? n : status; } static DEVICE_ATTR(mode, 0644, musb_mode_show, musb_mode_store); diff --git a/drivers/usb/musb/musb_core.h b/drivers/usb/musb/musb_core.h index 82227251931..b3d60507307 100644 --- a/drivers/usb/musb/musb_core.h +++ b/drivers/usb/musb/musb_core.h @@ -467,7 +467,7 @@ extern void musb_platform_disable(struct musb *musb); extern void musb_hnp_stop(struct musb *musb); -extern void musb_platform_set_mode(struct musb *musb, u8 musb_mode); +extern int musb_platform_set_mode(struct musb *musb, u8 musb_mode); #if defined(CONFIG_USB_TUSB6010) || \ defined(CONFIG_ARCH_OMAP2430) || defined(CONFIG_ARCH_OMAP34XX) diff --git a/drivers/usb/musb/omap2430.c b/drivers/usb/musb/omap2430.c index ce6c162920f..52988a403b5 100644 --- a/drivers/usb/musb/omap2430.c +++ b/drivers/usb/musb/omap2430.c @@ -196,7 +196,7 @@ static int omap_set_power(struct otg_transceiver *x, unsigned mA) static int musb_platform_resume(struct musb *musb); -void musb_platform_set_mode(struct musb *musb, u8 musb_mode) +int musb_platform_set_mode(struct musb *musb, u8 musb_mode) { u8 devctl = musb_readb(musb->mregs, MUSB_DEVCTL); @@ -204,15 +204,24 @@ void musb_platform_set_mode(struct musb *musb, u8 musb_mode) musb_writeb(musb->mregs, MUSB_DEVCTL, devctl); switch (musb_mode) { +#ifdef CONFIG_USB_MUSB_HDRC_HCD case MUSB_HOST: otg_set_host(&musb->xceiv, musb->xceiv.host); break; +#endif +#ifdef CONFIG_USB_GADGET_MUSB_HDRC case MUSB_PERIPHERAL: otg_set_peripheral(&musb->xceiv, musb->xceiv.gadget); break; +#endif +#ifdef CONFIG_USB_MUSB_OTG case MUSB_OTG: break; +#endif + default: + return -EINVAL; } + return 0; } int __init musb_platform_init(struct musb *musb) diff --git a/drivers/usb/musb/tusb6010.c b/drivers/usb/musb/tusb6010.c index ee8fca92a4a..9e20fd070d7 100644 --- a/drivers/usb/musb/tusb6010.c +++ b/drivers/usb/musb/tusb6010.c @@ -598,7 +598,7 @@ static void tusb_source_power(struct musb *musb, int is_on) * and peripheral modes in non-OTG configurations by reconfiguring hardware * and then setting musb->board_mode. For now, only support OTG mode. */ -void musb_platform_set_mode(struct musb *musb, u8 musb_mode) +int musb_platform_set_mode(struct musb *musb, u8 musb_mode) { void __iomem *tbase = musb->ctrl_base; u32 otg_stat, phy_otg_ctrl, phy_otg_ena, dev_conf; @@ -641,7 +641,8 @@ void musb_platform_set_mode(struct musb *musb, u8 musb_mode) #endif default: - DBG(2, "Trying to set unknown mode %i\n", musb_mode); + DBG(2, "Trying to set mode %i\n", musb_mode); + return -EINVAL; } musb_writel(tbase, TUSB_PHY_OTG_CTRL, @@ -655,6 +656,8 @@ void musb_platform_set_mode(struct musb *musb, u8 musb_mode) !(otg_stat & TUSB_DEV_OTG_STAT_ID_STATUS)) INFO("Cannot be peripheral with mini-A cable " "otg_stat: %08x\n", otg_stat); + + return 0; } static inline unsigned long From 58e660266d92aaa186c3df607c0ee88f18c8f4da Mon Sep 17 00:00:00 2001 From: Kevin Hilman Date: Mon, 24 Nov 2008 13:06:48 +0200 Subject: [PATCH 044/123] USB: musb: build fixes for DaVinci - update includes after asm/arch --> mach headers move - adds musb_platform_set_mode() stub Signed-off-by: Kevin Hilman Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/davinci.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/drivers/usb/musb/davinci.c b/drivers/usb/musb/davinci.c index 87aef82fbcd..0d566dc5ce0 100644 --- a/drivers/usb/musb/davinci.c +++ b/drivers/usb/musb/davinci.c @@ -32,9 +32,9 @@ #include #include -#include -#include -#include +#include +#include +#include #include #include "musb_core.h" @@ -370,6 +370,12 @@ int musb_platform_set_mode(struct musb *musb, u8 mode) return -EIO; } +int musb_platform_set_mode(struct musb *musb, u8 mode) +{ + /* EVM can't do this (right?) */ + return -EIO; +} + int __init musb_platform_init(struct musb *musb) { void __iomem *tibase = musb->ctrl_base; From 71783e0defa16ca5abca7750df98ff8e7767bd2e Mon Sep 17 00:00:00 2001 From: David Brownell Date: Mon, 24 Nov 2008 13:06:49 +0200 Subject: [PATCH 045/123] USB: musb: minor locking fix Minor locking fix for musb_hdrc on OMAP3 and OMAP2430: don't read DEVCTL without holding the spinlock, since an IRQ could come in and corrupt things. Signed-off-by: David Brownell Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/omap2430.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/usb/musb/omap2430.c b/drivers/usb/musb/omap2430.c index 52988a403b5..901dffdf23b 100644 --- a/drivers/usb/musb/omap2430.c +++ b/drivers/usb/musb/omap2430.c @@ -58,10 +58,10 @@ static void musb_do_idle(unsigned long _musb) #endif u8 devctl; - devctl = musb_readb(musb->mregs, MUSB_DEVCTL); - spin_lock_irqsave(&musb->lock, flags); + devctl = musb_readb(musb->mregs, MUSB_DEVCTL); + switch (musb->xceiv.state) { case OTG_STATE_A_WAIT_BCON: devctl &= ~MUSB_DEVCTL_SESSION; From bb1c9ef1b4cd64f22e15e8447deac6043eeb151c Mon Sep 17 00:00:00 2001 From: David Brownell Date: Mon, 24 Nov 2008 13:06:50 +0200 Subject: [PATCH 046/123] USB: musb: host side diagnostics tweaks Random host-side MUSB updates, mostly relating to better diagnostics: + Improve diagnostics on host side: - tx flush fifo: * Avoid hundreds of duplicate TX FIFONOTEMPTY messages * make "Can't flush TX fifo" a warning, and say which endpoint - giveback: * use correct status code * show completion function name not just URB pointer - Fix annoying "1 bytes" (should be "1 byte") + Be more consistent about failing init of unusable fifo_mode It's not clear why that "can't flush TX fifo" message appears, though it might relate to disconnection; I see it not infrequently Signed-off-by: David Brownell Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/musb_core.c | 3 +-- drivers/usb/musb/musb_host.c | 23 ++++++++++++++--------- 2 files changed, 15 insertions(+), 11 deletions(-) diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c index de7f8bec6bd..8b14437331c 100644 --- a/drivers/usb/musb/musb_core.c +++ b/drivers/usb/musb/musb_core.c @@ -1212,7 +1212,7 @@ static int __init ep_config_from_table(struct musb *musb) if (epn >= musb->config->num_eps) { pr_debug("%s: invalid ep %d\n", musb_driver_name, epn); - continue; + return -EINVAL; } offset = fifo_setup(musb, hw_ep + epn, cfg++, offset); if (offset < 0) { @@ -1813,7 +1813,6 @@ allocate_instance(struct device *dev, for (epnum = 0, ep = musb->endpoints; epnum < musb->config->num_eps; epnum++, ep++) { - ep->musb = musb; ep->epnum = epnum; } diff --git a/drivers/usb/musb/musb_host.c b/drivers/usb/musb/musb_host.c index cc64462d4c4..bf25ef4cc15 100644 --- a/drivers/usb/musb/musb_host.c +++ b/drivers/usb/musb/musb_host.c @@ -112,18 +112,21 @@ static void musb_h_tx_flush_fifo(struct musb_hw_ep *ep) { void __iomem *epio = ep->regs; u16 csr; + u16 lastcsr = 0; int retries = 1000; csr = musb_readw(epio, MUSB_TXCSR); while (csr & MUSB_TXCSR_FIFONOTEMPTY) { - DBG(5, "Host TX FIFONOTEMPTY csr: %02x\n", csr); + if (csr != lastcsr) + DBG(3, "Host TX FIFONOTEMPTY csr: %02x\n", csr); + lastcsr = csr; csr |= MUSB_TXCSR_FLUSHFIFO; musb_writew(epio, MUSB_TXCSR, csr); csr = musb_readw(epio, MUSB_TXCSR); - if (retries-- < 1) { - ERR("Could not flush host TX fifo: csr: %04x\n", csr); + if (WARN(retries-- < 1, + "Could not flush host TX%d fifo: csr: %04x\n", + ep->epnum, csr)) return; - } mdelay(1); } } @@ -268,7 +271,7 @@ __musb_giveback(struct musb *musb, struct urb *urb, int status) __releases(musb->lock) __acquires(musb->lock) { - DBG(({ int level; switch (urb->status) { + DBG(({ int level; switch (status) { case 0: level = 4; break; @@ -283,8 +286,8 @@ __acquires(musb->lock) level = 2; break; }; level; }), - "complete %p (%d), dev%d ep%d%s, %d/%d\n", - urb, urb->status, + "complete %p %pF (%d), dev%d ep%d%s, %d/%d\n", + urb, urb->complete, status, usb_pipedevice(urb->pipe), usb_pipeendpoint(urb->pipe), usb_pipein(urb->pipe) ? "in" : "out", @@ -988,8 +991,10 @@ static bool musb_h_ep0_continue(struct musb *musb, u16 len, struct urb *urb) if (fifo_count) { fifo_dest = (u8 *) (urb->transfer_buffer + urb->actual_length); - DBG(3, "Sending %d bytes to %p\n", - fifo_count, fifo_dest); + DBG(3, "Sending %d byte%s to ep0 fifo %p\n", + fifo_count, + (fifo_count == 1) ? "" : "s", + fifo_dest); musb_write_fifo(hw_ep, fifo_count, fifo_dest); urb->actual_length += fifo_count; From b6a49b8490fe6f22f0027a3f05eb498918f09303 Mon Sep 17 00:00:00 2001 From: Felipe Balbi Date: Mon, 24 Nov 2008 13:06:51 +0200 Subject: [PATCH 047/123] USB: musb: Add musb git tree to maintainers entry Trivial patch adding musb's git tree to MAINTAINERS file. Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index 094dd52d730..f62fd6ac45a 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -2982,6 +2982,7 @@ MUSB MULTIPOINT HIGH SPEED DUAL-ROLE CONTROLLER P: Felipe Balbi M: felipe.balbi@nokia.com L: linux-usb@vger.kernel.org +T: git gitorious.org:/musb/mainline.git S: Maintained MYRICOM MYRI-10G 10GbE DRIVER (MYRI10GE) From c48a5155252fd9cba7bedc59e5f8a339a3454d58 Mon Sep 17 00:00:00 2001 From: Felipe Balbi Date: Mon, 24 Nov 2008 13:06:53 +0200 Subject: [PATCH 048/123] USB: musb: check if set_irq_wake succeded and remember it Without it, in platforms that don't provide irq_chip.set_wake(), like omap, musb will WARN() on driver removal. Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/musb_core.c | 8 ++++++-- drivers/usb/musb/musb_core.h | 1 + 2 files changed, 7 insertions(+), 2 deletions(-) diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c index 8b14437331c..83720f61fdb 100644 --- a/drivers/usb/musb/musb_core.c +++ b/drivers/usb/musb/musb_core.c @@ -1840,7 +1840,7 @@ static void musb_free(struct musb *musb) musb_gadget_cleanup(musb); #endif - if (musb->nIrq >= 0) { + if (musb->nIrq >= 0 && musb->irq_wake) { disable_irq_wake(musb->nIrq); free_irq(musb->nIrq, musb); } @@ -1993,8 +1993,12 @@ bad_config: } musb->nIrq = nIrq; /* FIXME this handles wakeup irqs wrong */ - if (enable_irq_wake(nIrq) == 0) + if (enable_irq_wake(nIrq) == 0) { + musb->irq_wake = 1; device_init_wakeup(dev, 1); + } else { + musb->irq_wake = 0; + } pr_info("%s: USB %s mode controller at %p using %s, IRQ %d\n", musb_driver_name, diff --git a/drivers/usb/musb/musb_core.h b/drivers/usb/musb/musb_core.h index b3d60507307..d45d258e63f 100644 --- a/drivers/usb/musb/musb_core.h +++ b/drivers/usb/musb/musb_core.h @@ -359,6 +359,7 @@ struct musb { struct otg_transceiver xceiv; int nIrq; + unsigned irq_wake:1; struct musb_hw_ep endpoints[MUSB_C_NUM_EPS]; #define control_ep endpoints From b40fc2a2563b03a7b925cd16d1657aa2c24fd9c0 Mon Sep 17 00:00:00 2001 From: David Brownell Date: Mon, 24 Nov 2008 11:43:30 -0800 Subject: [PATCH 049/123] USB: gadget: pxa25x_udc vbus sense initialization Some code in the pxa25x_udc driver wrongly expects the value of is_vbus_present() to be 0/1, not zero/nonzero ... cope. Issue noted by Jaya Kumar This bug has been around since July 2007, and has a simple workaround: unplug the Linux gadget, then re-plug it. Signed-off-by: David Brownell Cc: Jaya Kumar Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/pxa25x_udc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/gadget/pxa25x_udc.c b/drivers/usb/gadget/pxa25x_udc.c index 697a0ca349b..9b36205c575 100644 --- a/drivers/usb/gadget/pxa25x_udc.c +++ b/drivers/usb/gadget/pxa25x_udc.c @@ -2198,7 +2198,7 @@ static int __init pxa25x_udc_probe(struct platform_device *pdev) udc_disable(dev); udc_reinit(dev); - dev->vbus = is_vbus_present(); + dev->vbus = !!is_vbus_present(); /* irq setup after old hardware state is cleaned up */ retval = request_irq(irq, pxa25x_udc_irq, From ea99ecfdc81266e61f4445c1830315a65eb8175a Mon Sep 17 00:00:00 2001 From: Ben Dooks Date: Mon, 24 Nov 2008 11:45:03 -0800 Subject: [PATCH 050/123] USB: gadget: s3c2410_udc uses standard GPIO calls Change the gpio code in the s3c2410_udc to use the generic gpio calls instead of the s3c24xx specific gpio functions. Signed-off-by: Ben Dooks Signed-off-by: David Brownell Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/s3c2410_udc.c | 34 +++++++++++++++++++++----------- 1 file changed, 22 insertions(+), 12 deletions(-) diff --git a/drivers/usb/gadget/s3c2410_udc.c b/drivers/usb/gadget/s3c2410_udc.c index c7e25563680..9a2b8920532 100644 --- a/drivers/usb/gadget/s3c2410_udc.c +++ b/drivers/usb/gadget/s3c2410_udc.c @@ -36,6 +36,7 @@ #include #include #include +#include #include #include @@ -51,7 +52,6 @@ #include #include -#include #include #include @@ -1510,11 +1510,7 @@ static irqreturn_t s3c2410_udc_vbus_irq(int irq, void *_dev) dprintk(DEBUG_NORMAL, "%s()\n", __func__); - /* some cpus cannot read from an line configured to IRQ! */ - s3c2410_gpio_cfgpin(udc_info->vbus_pin, S3C2410_GPIO_INPUT); - value = s3c2410_gpio_getpin(udc_info->vbus_pin); - s3c2410_gpio_cfgpin(udc_info->vbus_pin, S3C2410_GPIO_SFN2); - + value = gpio_get_value(udc_info->vbus_pin) ? 1 : 0; if (udc_info->vbus_pin_inverted) value = !value; @@ -1802,7 +1798,7 @@ static int s3c2410_udc_probe(struct platform_device *pdev) struct s3c2410_udc *udc = &memory; struct device *dev = &pdev->dev; int retval; - unsigned int irq; + int irq; dev_dbg(dev, "%s()\n", __func__); @@ -1861,7 +1857,7 @@ static int s3c2410_udc_probe(struct platform_device *pdev) /* irq setup after old hardware state is cleaned up */ retval = request_irq(IRQ_USBD, s3c2410_udc_irq, - IRQF_DISABLED, gadget_name, udc); + IRQF_DISABLED, gadget_name, udc); if (retval != 0) { dev_err(dev, "cannot get irq %i, err %d\n", IRQ_USBD, retval); @@ -1872,17 +1868,28 @@ static int s3c2410_udc_probe(struct platform_device *pdev) dev_dbg(dev, "got irq %i\n", IRQ_USBD); if (udc_info && udc_info->vbus_pin > 0) { - irq = s3c2410_gpio_getirq(udc_info->vbus_pin); + retval = gpio_request(udc_info->vbus_pin, "udc vbus"); + if (retval < 0) { + dev_err(dev, "cannot claim vbus pin\n"); + goto err_int; + } + + irq = gpio_to_irq(udc_info->vbus_pin); + if (irq < 0) { + dev_err(dev, "no irq for gpio vbus pin\n"); + goto err_gpio_claim; + } + retval = request_irq(irq, s3c2410_udc_vbus_irq, IRQF_DISABLED | IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING | IRQF_SHARED, gadget_name, udc); if (retval != 0) { - dev_err(dev, "can't get vbus irq %i, err %d\n", + dev_err(dev, "can't get vbus irq %d, err %d\n", irq, retval); retval = -EBUSY; - goto err_int; + goto err_gpio_claim; } dev_dbg(dev, "got irq %i\n", irq); @@ -1902,6 +1909,9 @@ static int s3c2410_udc_probe(struct platform_device *pdev) return 0; +err_gpio_claim: + if (udc_info && udc_info->vbus_pin > 0) + gpio_free(udc_info->vbus_pin); err_int: free_irq(IRQ_USBD, udc); err_map: @@ -1927,7 +1937,7 @@ static int s3c2410_udc_remove(struct platform_device *pdev) debugfs_remove(udc->regs_info); if (udc_info && udc_info->vbus_pin > 0) { - irq = s3c2410_gpio_getirq(udc_info->vbus_pin); + irq = gpio_to_irq(udc_info->vbus_pin); free_irq(irq, udc); } From b8da8677d4f88db066c1cfe34529d970a060de46 Mon Sep 17 00:00:00 2001 From: David Brownell Date: Mon, 24 Nov 2008 11:53:35 -0800 Subject: [PATCH 051/123] USB: move isp1301_omap to drivers/usb/otg This moves the isp1301-omap driver from the drivers/i2c/chips directory (which will be shrinking) into a new drivers/usb/otg directory (which will grow, with more drivers and utilities). Note that OTG infrastructure needs to be initialized before either host or peripheral side USB support, and may be needed before for pure host or pure peripheral configurations. Signed-off-by: David Brownell Acked-by: Jean Delvare Signed-off-by: Greg Kroah-Hartman --- drivers/Makefile | 1 + drivers/i2c/chips/Kconfig | 12 ------- drivers/i2c/chips/Makefile | 1 - drivers/usb/Kconfig | 2 ++ drivers/usb/otg/Kconfig | 31 +++++++++++++++++++ drivers/usb/otg/Makefile | 8 +++++ drivers/{i2c/chips => usb/otg}/isp1301_omap.c | 0 7 files changed, 42 insertions(+), 13 deletions(-) create mode 100644 drivers/usb/otg/Kconfig create mode 100644 drivers/usb/otg/Makefile rename drivers/{i2c/chips => usb/otg}/isp1301_omap.c (100%) diff --git a/drivers/Makefile b/drivers/Makefile index fceb71a741c..e121b66ef08 100644 --- a/drivers/Makefile +++ b/drivers/Makefile @@ -57,6 +57,7 @@ obj-$(CONFIG_ATA_OVER_ETH) += block/aoe/ obj-$(CONFIG_PARIDE) += block/paride/ obj-$(CONFIG_TC) += tc/ obj-$(CONFIG_UWB) += uwb/ +obj-$(CONFIG_USB_OTG_UTILS) += usb/otg/ obj-$(CONFIG_USB) += usb/ obj-$(CONFIG_USB_MUSB_HDRC) += usb/musb/ obj-$(CONFIG_PCI) += usb/ diff --git a/drivers/i2c/chips/Kconfig b/drivers/i2c/chips/Kconfig index 864ac561fdb..59c3d23f5bd 100644 --- a/drivers/i2c/chips/Kconfig +++ b/drivers/i2c/chips/Kconfig @@ -114,18 +114,6 @@ config SENSORS_PCF8591 These devices are hard to detect and rarely found on mainstream hardware. If unsure, say N. -config ISP1301_OMAP - tristate "Philips ISP1301 with OMAP OTG" - depends on ARCH_OMAP_OTG - help - If you say yes here you get support for the Philips ISP1301 - USB-On-The-Go transceiver working with the OMAP OTG controller. - The ISP1301 is used in products including H2 and H3 development - boards for Texas Instruments OMAP processors. - - This driver can also be built as a module. If so, the module - will be called isp1301_omap. - config SENSORS_MAX6875 tristate "Maxim MAX6875 Power supply supervisor" depends on EXPERIMENTAL diff --git a/drivers/i2c/chips/Makefile b/drivers/i2c/chips/Makefile index 8b95f41a500..83accaaf816 100644 --- a/drivers/i2c/chips/Makefile +++ b/drivers/i2c/chips/Makefile @@ -18,7 +18,6 @@ obj-$(CONFIG_SENSORS_PCA9539) += pca9539.o obj-$(CONFIG_SENSORS_PCF8574) += pcf8574.o obj-$(CONFIG_PCF8575) += pcf8575.o obj-$(CONFIG_SENSORS_PCF8591) += pcf8591.o -obj-$(CONFIG_ISP1301_OMAP) += isp1301_omap.o obj-$(CONFIG_SENSORS_TSL2550) += tsl2550.o obj-$(CONFIG_MCU_MPC8349EMITX) += mcu_mpc8349emitx.o diff --git a/drivers/usb/Kconfig b/drivers/usb/Kconfig index 289d81adfb9..83babb0a1df 100644 --- a/drivers/usb/Kconfig +++ b/drivers/usb/Kconfig @@ -150,4 +150,6 @@ source "drivers/usb/atm/Kconfig" source "drivers/usb/gadget/Kconfig" +source "drivers/usb/otg/Kconfig" + endif # USB_SUPPORT diff --git a/drivers/usb/otg/Kconfig b/drivers/usb/otg/Kconfig new file mode 100644 index 00000000000..860342902bb --- /dev/null +++ b/drivers/usb/otg/Kconfig @@ -0,0 +1,31 @@ +# +# USB OTG infrastructure may be needed for peripheral-only, host-only, +# or OTG-capable configurations when OTG transceivers or controllers +# are used. +# + +comment "OTG and related infrastructure" + +if USB || USB_GADGET + +config USB_OTG_UTILS + bool + help + Select this to make sure the build includes objects from + the OTG infrastructure directory. + +config ISP1301_OMAP + tristate "Philips ISP1301 with OMAP OTG" + depends on I2C && ARCH_OMAP_OTG + select USB_OTG_UTILS + help + If you say yes here you get support for the Philips ISP1301 + USB-On-The-Go transceiver working with the OMAP OTG controller. + The ISP1301 is a full speed USB transceiver which is used in + products including H2, H3, and H4 development boards for Texas + Instruments OMAP processors. + + This driver can also be built as a module. If so, the module + will be called isp1301_omap. + +endif # USB || OTG diff --git a/drivers/usb/otg/Makefile b/drivers/usb/otg/Makefile new file mode 100644 index 00000000000..483816685d0 --- /dev/null +++ b/drivers/usb/otg/Makefile @@ -0,0 +1,8 @@ +obj-$(CONFIG_ISP1301_OMAP) += isp1301_omap.o + +ifeq ($(CONFIG_USB_DEBUG),y) +EXTRA_CFLAGS += -DDEBUG +else ifeq ($(CONFIG_USB_GADGET_DEBUG),y) +EXTRA_CFLAGS += -DDEBUG +endif + diff --git a/drivers/i2c/chips/isp1301_omap.c b/drivers/usb/otg/isp1301_omap.c similarity index 100% rename from drivers/i2c/chips/isp1301_omap.c rename to drivers/usb/otg/isp1301_omap.c From 6084f1bf0c51a99cbba612ee90a4607cffb8b042 Mon Sep 17 00:00:00 2001 From: Philipp Zabel Date: Mon, 24 Nov 2008 12:00:01 -0800 Subject: [PATCH 052/123] USB: otg: gpio_vbus transceiver stub gpio_vbus provides simple GPIO VBUS sensing for peripheral controllers with an internal transceiver. Optionally, a second GPIO can be used to control D+ pullup. It also interfaces with the regulator framework to limit charging currents when powered via USB. gpio_vbus requests the regulator supplying "vbus_draw" and can enable/disable it or limit its current depending on USB state. [dbrownell@users.sourceforge.net: use drivers/otg, cleanups ] Signed-off-by: Philipp Zabel Signed-off-by: David Brownell Signed-off-by: Greg Kroah-Hartman --- drivers/usb/otg/Kconfig | 13 ++ drivers/usb/otg/Makefile | 5 + drivers/usb/otg/gpio_vbus.c | 335 ++++++++++++++++++++++++++++++++++ include/linux/usb/gpio_vbus.h | 30 +++ 4 files changed, 383 insertions(+) create mode 100644 drivers/usb/otg/gpio_vbus.c create mode 100644 include/linux/usb/gpio_vbus.h diff --git a/drivers/usb/otg/Kconfig b/drivers/usb/otg/Kconfig index 860342902bb..afe91bfea7f 100644 --- a/drivers/usb/otg/Kconfig +++ b/drivers/usb/otg/Kconfig @@ -14,6 +14,19 @@ config USB_OTG_UTILS Select this to make sure the build includes objects from the OTG infrastructure directory. +# +# USB Transceiver Drivers +# +config USB_GPIO_VBUS + tristate "GPIO based peripheral-only VBUS sensing 'transceiver'" + depends on GENERIC_GPIO + select USB_OTG_UTILS + help + Provides simple GPIO VBUS sensing for controllers with an + internal transceiver via the otg_transceiver interface, and + optionally control of a D+ pullup GPIO as well as a VBUS + current limit regulator. + config ISP1301_OMAP tristate "Philips ISP1301 with OMAP OTG" depends on I2C && ARCH_OMAP_OTG diff --git a/drivers/usb/otg/Makefile b/drivers/usb/otg/Makefile index 483816685d0..6c58b36ca7c 100644 --- a/drivers/usb/otg/Makefile +++ b/drivers/usb/otg/Makefile @@ -1,3 +1,8 @@ +# +# OTG infrastructure and transceiver drivers +# + +obj-$(CONFIG_USB_GPIO_VBUS) += gpio_vbus.o obj-$(CONFIG_ISP1301_OMAP) += isp1301_omap.o ifeq ($(CONFIG_USB_DEBUG),y) diff --git a/drivers/usb/otg/gpio_vbus.c b/drivers/usb/otg/gpio_vbus.c new file mode 100644 index 00000000000..63a6036f04b --- /dev/null +++ b/drivers/usb/otg/gpio_vbus.c @@ -0,0 +1,335 @@ +/* + * gpio-vbus.c - simple GPIO VBUS sensing driver for B peripheral devices + * + * Copyright (c) 2008 Philipp Zabel + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include +#include +#include +#include +#include + +#include + +#include +#include +#include + + +/* + * A simple GPIO VBUS sensing driver for B peripheral only devices + * with internal transceivers. It can control a D+ pullup GPIO and + * a regulator to limit the current drawn from VBUS. + * + * Needs to be loaded before the UDC driver that will use it. + */ +struct gpio_vbus_data { + struct otg_transceiver otg; + struct device *dev; + struct regulator *vbus_draw; + int vbus_draw_enabled; + unsigned mA; +}; + + +/* + * This driver relies on "both edges" triggering. VBUS has 100 msec to + * stabilize, so the peripheral controller driver may need to cope with + * some bouncing due to current surges (e.g. charging local capacitance) + * and contact chatter. + * + * REVISIT in desperate straits, toggling between rising and falling + * edges might be workable. + */ +#define VBUS_IRQ_FLAGS \ + ( IRQF_SAMPLE_RANDOM | IRQF_SHARED \ + | IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING ) + + +/* interface to regulator framework */ +static void set_vbus_draw(struct gpio_vbus_data *gpio_vbus, unsigned mA) +{ + struct regulator *vbus_draw = gpio_vbus->vbus_draw; + int enabled; + + if (!vbus_draw) + return; + + enabled = gpio_vbus->vbus_draw_enabled; + if (mA) { + regulator_set_current_limit(vbus_draw, 0, 1000 * mA); + if (!enabled) { + regulator_enable(vbus_draw); + gpio_vbus->vbus_draw_enabled = 1; + } + } else { + if (enabled) { + regulator_disable(vbus_draw); + gpio_vbus->vbus_draw_enabled = 0; + } + } + gpio_vbus->mA = mA; +} + +/* VBUS change IRQ handler */ +static irqreturn_t gpio_vbus_irq(int irq, void *data) +{ + struct platform_device *pdev = data; + struct gpio_vbus_mach_info *pdata = pdev->dev.platform_data; + struct gpio_vbus_data *gpio_vbus = platform_get_drvdata(pdev); + int gpio, vbus; + + vbus = gpio_get_value(pdata->gpio_vbus); + if (pdata->gpio_vbus_inverted) + vbus = !vbus; + + dev_dbg(&pdev->dev, "VBUS %s (gadget: %s)\n", + vbus ? "supplied" : "inactive", + gpio_vbus->otg.gadget ? gpio_vbus->otg.gadget->name : "none"); + + if (!gpio_vbus->otg.gadget) + return IRQ_HANDLED; + + /* Peripheral controllers which manage the pullup themselves won't have + * gpio_pullup configured here. If it's configured here, we'll do what + * isp1301_omap::b_peripheral() does and enable the pullup here... although + * that may complicate usb_gadget_{,dis}connect() support. + */ + gpio = pdata->gpio_pullup; + if (vbus) { + gpio_vbus->otg.state = OTG_STATE_B_PERIPHERAL; + usb_gadget_vbus_connect(gpio_vbus->otg.gadget); + + /* drawing a "unit load" is *always* OK, except for OTG */ + set_vbus_draw(gpio_vbus, 100); + + /* optionally enable D+ pullup */ + if (gpio_is_valid(gpio)) + gpio_set_value(gpio, !pdata->gpio_pullup_inverted); + } else { + /* optionally disable D+ pullup */ + if (gpio_is_valid(gpio)) + gpio_set_value(gpio, pdata->gpio_pullup_inverted); + + set_vbus_draw(gpio_vbus, 0); + + usb_gadget_vbus_disconnect(gpio_vbus->otg.gadget); + gpio_vbus->otg.state = OTG_STATE_B_IDLE; + } + + return IRQ_HANDLED; +} + +/* OTG transceiver interface */ + +/* bind/unbind the peripheral controller */ +static int gpio_vbus_set_peripheral(struct otg_transceiver *otg, + struct usb_gadget *gadget) +{ + struct gpio_vbus_data *gpio_vbus; + struct gpio_vbus_mach_info *pdata; + struct platform_device *pdev; + int gpio, irq; + + gpio_vbus = container_of(otg, struct gpio_vbus_data, otg); + pdev = to_platform_device(gpio_vbus->dev); + pdata = gpio_vbus->dev->platform_data; + irq = gpio_to_irq(pdata->gpio_vbus); + gpio = pdata->gpio_pullup; + + if (!gadget) { + dev_dbg(&pdev->dev, "unregistering gadget '%s'\n", + otg->gadget->name); + + /* optionally disable D+ pullup */ + if (gpio_is_valid(gpio)) + gpio_set_value(gpio, pdata->gpio_pullup_inverted); + + set_vbus_draw(gpio_vbus, 0); + + usb_gadget_vbus_disconnect(otg->gadget); + otg->state = OTG_STATE_UNDEFINED; + + otg->gadget = NULL; + return 0; + } + + otg->gadget = gadget; + dev_dbg(&pdev->dev, "registered gadget '%s'\n", gadget->name); + + /* initialize connection state */ + gpio_vbus_irq(irq, pdev); + return 0; +} + +/* effective for B devices, ignored for A-peripheral */ +static int gpio_vbus_set_power(struct otg_transceiver *otg, unsigned mA) +{ + struct gpio_vbus_data *gpio_vbus; + + gpio_vbus = container_of(otg, struct gpio_vbus_data, otg); + + if (otg->state == OTG_STATE_B_PERIPHERAL) + set_vbus_draw(gpio_vbus, mA); + return 0; +} + +/* for non-OTG B devices: set/clear transceiver suspend mode */ +static int gpio_vbus_set_suspend(struct otg_transceiver *otg, int suspend) +{ + struct gpio_vbus_data *gpio_vbus; + + gpio_vbus = container_of(otg, struct gpio_vbus_data, otg); + + /* draw max 0 mA from vbus in suspend mode; or the previously + * recorded amount of current if not suspended + * + * NOTE: high powered configs (mA > 100) may draw up to 2.5 mA + * if they're wake-enabled ... we don't handle that yet. + */ + return gpio_vbus_set_power(otg, suspend ? 0 : gpio_vbus->mA); +} + +/* platform driver interface */ + +static int __init gpio_vbus_probe(struct platform_device *pdev) +{ + struct gpio_vbus_mach_info *pdata = pdev->dev.platform_data; + struct gpio_vbus_data *gpio_vbus; + struct resource *res; + int err, gpio, irq; + + if (!pdata || !gpio_is_valid(pdata->gpio_vbus)) + return -EINVAL; + gpio = pdata->gpio_vbus; + + gpio_vbus = kzalloc(sizeof(struct gpio_vbus_data), GFP_KERNEL); + if (!gpio_vbus) + return -ENOMEM; + + platform_set_drvdata(pdev, gpio_vbus); + gpio_vbus->dev = &pdev->dev; + gpio_vbus->otg.label = "gpio-vbus"; + gpio_vbus->otg.state = OTG_STATE_UNDEFINED; + gpio_vbus->otg.set_peripheral = gpio_vbus_set_peripheral; + gpio_vbus->otg.set_power = gpio_vbus_set_power; + gpio_vbus->otg.set_suspend = gpio_vbus_set_suspend; + + err = gpio_request(gpio, "vbus_detect"); + if (err) { + dev_err(&pdev->dev, "can't request vbus gpio %d, err: %d\n", + gpio, err); + goto err_gpio; + } + gpio_direction_input(gpio); + + res = platform_get_resource(pdev, IORESOURCE_IRQ, 0); + if (res) { + irq = res->start; + res->flags &= IRQF_TRIGGER_MASK; + res->flags |= IRQF_SAMPLE_RANDOM | IRQF_SHARED; + } else + irq = gpio_to_irq(gpio); + + /* if data line pullup is in use, initialize it to "not pulling up" */ + gpio = pdata->gpio_pullup; + if (gpio_is_valid(gpio)) { + err = gpio_request(gpio, "udc_pullup"); + if (err) { + dev_err(&pdev->dev, + "can't request pullup gpio %d, err: %d\n", + gpio, err); + gpio_free(pdata->gpio_vbus); + goto err_gpio; + } + gpio_direction_output(gpio, pdata->gpio_pullup_inverted); + } + + err = request_irq(irq, gpio_vbus_irq, VBUS_IRQ_FLAGS, + "vbus_detect", pdev); + if (err) { + dev_err(&pdev->dev, "can't request irq %i, err: %d\n", + irq, err); + goto err_irq; + } + + /* only active when a gadget is registered */ + err = otg_set_transceiver(&gpio_vbus->otg); + if (err) { + dev_err(&pdev->dev, "can't register transceiver, err: %d\n", + err); + goto err_otg; + } + + gpio_vbus->vbus_draw = regulator_get(&pdev->dev, "vbus_draw"); + if (IS_ERR(gpio_vbus->vbus_draw)) { + dev_dbg(&pdev->dev, "can't get vbus_draw regulator, err: %ld\n", + PTR_ERR(gpio_vbus->vbus_draw)); + gpio_vbus->vbus_draw = NULL; + } + + return 0; +err_otg: + free_irq(irq, &pdev->dev); +err_irq: + if (gpio_is_valid(pdata->gpio_pullup)) + gpio_free(pdata->gpio_pullup); + gpio_free(pdata->gpio_vbus); +err_gpio: + platform_set_drvdata(pdev, NULL); + kfree(gpio_vbus); + return err; +} + +static int __exit gpio_vbus_remove(struct platform_device *pdev) +{ + struct gpio_vbus_data *gpio_vbus = platform_get_drvdata(pdev); + struct gpio_vbus_mach_info *pdata = pdev->dev.platform_data; + int gpio = pdata->gpio_vbus; + + regulator_put(gpio_vbus->vbus_draw); + + otg_set_transceiver(NULL); + + free_irq(gpio_to_irq(gpio), &pdev->dev); + if (gpio_is_valid(pdata->gpio_pullup)) + gpio_free(pdata->gpio_pullup); + gpio_free(gpio); + platform_set_drvdata(pdev, NULL); + kfree(gpio_vbus); + + return 0; +} + +/* NOTE: the gpio-vbus device may *NOT* be hotplugged */ + +MODULE_ALIAS("platform:gpio-vbus"); + +static struct platform_driver gpio_vbus_driver = { + .driver = { + .name = "gpio-vbus", + .owner = THIS_MODULE, + }, + .remove = __exit_p(gpio_vbus_remove), +}; + +static int __init gpio_vbus_init(void) +{ + return platform_driver_probe(&gpio_vbus_driver, gpio_vbus_probe); +} +module_init(gpio_vbus_init); + +static void __exit gpio_vbus_exit(void) +{ + platform_driver_unregister(&gpio_vbus_driver); +} +module_exit(gpio_vbus_exit); + +MODULE_DESCRIPTION("simple GPIO controlled OTG transceiver driver"); +MODULE_AUTHOR("Philipp Zabel"); +MODULE_LICENSE("GPL"); diff --git a/include/linux/usb/gpio_vbus.h b/include/linux/usb/gpio_vbus.h new file mode 100644 index 00000000000..d9f03ccc2d6 --- /dev/null +++ b/include/linux/usb/gpio_vbus.h @@ -0,0 +1,30 @@ +/* + * A simple GPIO VBUS sensing driver for B peripheral only devices + * with internal transceivers. + * Optionally D+ pullup can be controlled by a second GPIO. + * + * Copyright (c) 2008 Philipp Zabel + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + * + */ + +/** + * struct gpio_vbus_mach_info - configuration for gpio_vbus + * @gpio_vbus: VBUS sensing GPIO + * @gpio_pullup: optional D+ or D- pullup GPIO (else negative/invalid) + * @gpio_vbus_inverted: true if gpio_vbus is active low + * @gpio_pullup_inverted: true if gpio_pullup is active low + * + * The VBUS sensing GPIO should have a pulldown, which will normally be + * part of a resistor ladder turning a 4.0V-5.25V level on VBUS into a + * value the GPIO detects as active. Some systems will use comparators. + */ +struct gpio_vbus_mach_info { + int gpio_vbus; + int gpio_pullup; + bool gpio_vbus_inverted; + bool gpio_pullup_inverted; +}; From 68144e0cc92125f41157ede7b060f83367bc4fe7 Mon Sep 17 00:00:00 2001 From: Philipp Zabel Date: Mon, 24 Nov 2008 12:01:17 -0800 Subject: [PATCH 053/123] USB: otg: add otg_put_transceiver() As Russell King points out, calling put_device(otg_transceiver->dev) directly in driver cleanup paths makes assumptions about otg_transceiver internals. Signed-off-by: Philipp Zabel Signed-off-by: David Brownell Signed-off-by: Greg Kroah-Hartman --- arch/arm/plat-omap/usb.c | 6 ++++++ drivers/usb/gadget/omap_udc.c | 4 ++-- include/linux/usb/otg.h | 1 + 3 files changed, 9 insertions(+), 2 deletions(-) diff --git a/arch/arm/plat-omap/usb.c b/arch/arm/plat-omap/usb.c index 67ca1e216df..2f88ca8b8f3 100644 --- a/arch/arm/plat-omap/usb.c +++ b/arch/arm/plat-omap/usb.c @@ -96,6 +96,12 @@ struct otg_transceiver *otg_get_transceiver(void) } EXPORT_SYMBOL(otg_get_transceiver); +void otg_put_transceiver(struct otg_transceiver *x) +{ + put_device(x->dev); +} +EXPORT_SYMBOL(otg_put_transceiver); + int otg_set_transceiver(struct otg_transceiver *x) { if (xceiv && x) diff --git a/drivers/usb/gadget/omap_udc.c b/drivers/usb/gadget/omap_udc.c index 34e9e393f92..57d9641c6bf 100644 --- a/drivers/usb/gadget/omap_udc.c +++ b/drivers/usb/gadget/omap_udc.c @@ -3006,7 +3006,7 @@ cleanup1: cleanup0: if (xceiv) - put_device(xceiv->dev); + otg_put_transceiver(xceiv); if (cpu_is_omap16xx() || cpu_is_omap24xx()) { clk_disable(hhc_clk); @@ -3034,7 +3034,7 @@ static int __exit omap_udc_remove(struct platform_device *pdev) pullup_disable(udc); if (udc->transceiver) { - put_device(udc->transceiver->dev); + otg_put_transceiver(udc->transceiver); udc->transceiver = NULL; } omap_writew(0, UDC_SYSCON1); diff --git a/include/linux/usb/otg.h b/include/linux/usb/otg.h index 1db25d152ad..94df4fe6c6c 100644 --- a/include/linux/usb/otg.h +++ b/include/linux/usb/otg.h @@ -84,6 +84,7 @@ extern int otg_set_transceiver(struct otg_transceiver *); /* for usb host and peripheral controller drivers */ extern struct otg_transceiver *otg_get_transceiver(void); +extern void otg_put_transceiver(struct otg_transceiver *); static inline int otg_start_hnp(struct otg_transceiver *otg) From 3cb22d658668234edbe6dcb165501e9ef0c0a059 Mon Sep 17 00:00:00 2001 From: Tony Lindgren Date: Mon, 24 Nov 2008 12:02:21 -0800 Subject: [PATCH 054/123] USB: otg: sharable otg transceiver ops Move otg_get/set/put_transceiver() from omap specific code to common otg.c so other upcoming drivers can share them. [ dbrownell@users.sourceforge.net: move to drivers/usb/otg, dox ] Signed-off-by: Tony Lindgren Signed-off-by: Felipe Balbi Signed-off-by: Philipp Zabel Signed-off-by: David Brownell Signed-off-by: Greg Kroah-Hartman --- arch/arm/plat-omap/usb.c | 38 ----------------------- drivers/usb/otg/Makefile | 11 +++---- drivers/usb/otg/otg.c | 65 ++++++++++++++++++++++++++++++++++++++++ 3 files changed, 71 insertions(+), 43 deletions(-) create mode 100644 drivers/usb/otg/otg.c diff --git a/arch/arm/plat-omap/usb.c b/arch/arm/plat-omap/usb.c index 2f88ca8b8f3..add0485703b 100644 --- a/arch/arm/plat-omap/usb.c +++ b/arch/arm/plat-omap/usb.c @@ -77,44 +77,6 @@ /*-------------------------------------------------------------------------*/ -#if defined(CONFIG_ARCH_OMAP_OTG) || defined(CONFIG_USB_MUSB_OTG) - -static struct otg_transceiver *xceiv; - -/** - * otg_get_transceiver - find the (single) OTG transceiver driver - * - * Returns the transceiver driver, after getting a refcount to it; or - * null if there is no such transceiver. The caller is responsible for - * releasing that count. - */ -struct otg_transceiver *otg_get_transceiver(void) -{ - if (xceiv) - get_device(xceiv->dev); - return xceiv; -} -EXPORT_SYMBOL(otg_get_transceiver); - -void otg_put_transceiver(struct otg_transceiver *x) -{ - put_device(x->dev); -} -EXPORT_SYMBOL(otg_put_transceiver); - -int otg_set_transceiver(struct otg_transceiver *x) -{ - if (xceiv && x) - return -EBUSY; - xceiv = x; - return 0; -} -EXPORT_SYMBOL(otg_set_transceiver); - -#endif - -/*-------------------------------------------------------------------------*/ - #if defined(CONFIG_ARCH_OMAP_OTG) || defined(CONFIG_ARCH_OMAP15XX) static void omap2_usb_devconf_clear(u8 port, u32 mask) diff --git a/drivers/usb/otg/Makefile b/drivers/usb/otg/Makefile index 6c58b36ca7c..7c80fc379e6 100644 --- a/drivers/usb/otg/Makefile +++ b/drivers/usb/otg/Makefile @@ -2,12 +2,13 @@ # OTG infrastructure and transceiver drivers # +# infrastructure +obj-$(CONFIG_USB_OTG_UTILS) += otg.o + +# transceiver drivers obj-$(CONFIG_USB_GPIO_VBUS) += gpio_vbus.o obj-$(CONFIG_ISP1301_OMAP) += isp1301_omap.o -ifeq ($(CONFIG_USB_DEBUG),y) -EXTRA_CFLAGS += -DDEBUG -else ifeq ($(CONFIG_USB_GADGET_DEBUG),y) -EXTRA_CFLAGS += -DDEBUG -endif +ccflags-$(CONFIG_USB_DEBUG) += -DDEBUG +ccflags-$(CONFIG_USB_GADGET_DEBUG) += -DDEBUG diff --git a/drivers/usb/otg/otg.c b/drivers/usb/otg/otg.c new file mode 100644 index 00000000000..ff318fae7d4 --- /dev/null +++ b/drivers/usb/otg/otg.c @@ -0,0 +1,65 @@ +/* + * otg.c -- USB OTG utility code + * + * Copyright (C) 2004 Texas Instruments + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + */ + +#include +#include + +#include + +static struct otg_transceiver *xceiv; + +/** + * otg_get_transceiver - find the (single) OTG transceiver + * + * Returns the transceiver driver, after getting a refcount to it; or + * null if there is no such transceiver. The caller is responsible for + * calling otg_put_transceiver() to release that count. + * + * For use by USB host and peripheral drivers. + */ +struct otg_transceiver *otg_get_transceiver(void) +{ + if (xceiv) + get_device(xceiv->dev); + return xceiv; +} +EXPORT_SYMBOL(otg_get_transceiver); + +/** + * otg_put_transceiver - release the (single) OTG transceiver + * @x: the transceiver returned by otg_get_transceiver() + * + * Releases a refcount the caller received from otg_get_transceiver(). + * + * For use by USB host and peripheral drivers. + */ +void otg_put_transceiver(struct otg_transceiver *x) +{ + put_device(x->dev); +} +EXPORT_SYMBOL(otg_put_transceiver); + +/** + * otg_set_transceiver - declare the (single) OTG transceiver + * @x: the USB OTG transceiver to be used; or NULL + * + * This call is exclusively for use by transceiver drivers, which + * coordinate the activities of drivers for host and peripheral + * controllers, and in some cases for VBUS current regulation. + */ +int otg_set_transceiver(struct otg_transceiver *x) +{ + if (xceiv && x) + return -EBUSY; + xceiv = x; + return 0; +} +EXPORT_SYMBOL(otg_set_transceiver); From 4ec06d629628b6e5c7ff50d349a26ef5c35696e3 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Tue, 25 Nov 2008 16:40:02 -0500 Subject: [PATCH 055/123] USB: utilize round_jiffies_up_relative() This patch (as1178) uses the new round_jiffies_up_relative() routine for setting the autosuspend delayed_work timer. It's appropriate since we don't care too much about the exact length of the delay, but we don't want it to be too short (rounded down). Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/driver.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/usb/core/driver.c b/drivers/usb/core/driver.c index 7e26fb3c275..0226e019326 100644 --- a/drivers/usb/core/driver.c +++ b/drivers/usb/core/driver.c @@ -1111,7 +1111,7 @@ static int autosuspend_check(struct usb_device *udev, int reschedule) if (reschedule) { if (!timer_pending(&udev->autosuspend.timer)) { queue_delayed_work(ksuspend_usb_wq, &udev->autosuspend, - round_jiffies_relative(suspend_time - j)); + round_jiffies_up_relative(suspend_time - j)); } return -EAGAIN; } @@ -1553,7 +1553,7 @@ void usb_autopm_put_interface_async(struct usb_interface *intf) else if (intf->pm_usage_cnt <= 0 && !timer_pending(&udev->autosuspend.timer)) { queue_delayed_work(ksuspend_usb_wq, &udev->autosuspend, - round_jiffies_relative( + round_jiffies_up_relative( udev->autosuspend_delay)); } } From 65bfd2967c906ca322a4bb69a285fe0de8916ac6 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Tue, 25 Nov 2008 16:39:18 -0500 Subject: [PATCH 056/123] USB: Enhance usage of pm_message_t This patch (as1177) modifies the USB core suspend and resume routines. The resume functions now will take a pm_message_t argument, so they will know what sort of resume is occurring. The new argument is also passed to the port suspend/resume and bus suspend/resume routines (although they don't use it for anything but debugging). In addition, special pm_message_t values are used for user-initiated, device-initiated (i.e., remote wakeup), and automatic suspend/resume. By testing these values, drivers can tell whether or not a particular suspend was an autosuspend. Unfortunately, they can't do the same for resumes -- not until the pm_message_t argument is also passed to the drivers' resume methods. That will require a bigger change. IMO, the whole Power Management framework should have been set up this way in the first place. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/class/cdc-acm.c | 2 +- drivers/usb/class/cdc-wdm.c | 3 +- drivers/usb/core/driver.c | 61 +++++++++++++++++++++---------------- drivers/usb/core/generic.c | 10 +++--- drivers/usb/core/hcd.c | 10 +++--- drivers/usb/core/hcd.h | 4 +-- drivers/usb/core/hub.c | 16 +++++----- drivers/usb/core/sysfs.c | 6 ++-- drivers/usb/core/usb.c | 8 ++--- drivers/usb/core/usb.h | 15 +++++---- include/linux/usb.h | 2 +- 11 files changed, 74 insertions(+), 63 deletions(-) diff --git a/drivers/usb/class/cdc-acm.c b/drivers/usb/class/cdc-acm.c index d50a99f70ae..00b47ea24f8 100644 --- a/drivers/usb/class/cdc-acm.c +++ b/drivers/usb/class/cdc-acm.c @@ -1275,7 +1275,7 @@ static int acm_suspend(struct usb_interface *intf, pm_message_t message) struct acm *acm = usb_get_intfdata(intf); int cnt; - if (acm->dev->auto_pm) { + if (message.event & PM_EVENT_AUTO) { int b; spin_lock_irq(&acm->read_lock); diff --git a/drivers/usb/class/cdc-wdm.c b/drivers/usb/class/cdc-wdm.c index 5a8ecc045e3..3771d6e6d0c 100644 --- a/drivers/usb/class/cdc-wdm.c +++ b/drivers/usb/class/cdc-wdm.c @@ -764,7 +764,8 @@ static int wdm_suspend(struct usb_interface *intf, pm_message_t message) mutex_lock(&desc->plock); #ifdef CONFIG_PM - if (interface_to_usbdev(desc->intf)->auto_pm && test_bit(WDM_IN_USE, &desc->flags)) { + if ((message.event & PM_EVENT_AUTO) && + test_bit(WDM_IN_USE, &desc->flags)) { rv = -EBUSY; } else { #endif diff --git a/drivers/usb/core/driver.c b/drivers/usb/core/driver.c index 0226e019326..41c06025506 100644 --- a/drivers/usb/core/driver.c +++ b/drivers/usb/core/driver.c @@ -922,7 +922,7 @@ static int usb_suspend_device(struct usb_device *udev, pm_message_t msg) } /* Caller has locked udev's pm_mutex */ -static int usb_resume_device(struct usb_device *udev) +static int usb_resume_device(struct usb_device *udev, pm_message_t msg) { struct usb_device_driver *udriver; int status = 0; @@ -940,7 +940,7 @@ static int usb_resume_device(struct usb_device *udev) udev->reset_resume = 1; udriver = to_usb_device_driver(udev->dev.driver); - status = udriver->resume(udev); + status = udriver->resume(udev, msg); done: dev_vdbg(&udev->dev, "%s: status %d\n", __func__, status); @@ -969,7 +969,7 @@ static int usb_suspend_interface(struct usb_device *udev, status = driver->suspend(intf, msg); if (status == 0) mark_quiesced(intf); - else if (!udev->auto_pm) + else if (!(msg.event & PM_EVENT_AUTO)) dev_err(&intf->dev, "%s error %d\n", "suspend", status); } else { @@ -987,7 +987,7 @@ static int usb_suspend_interface(struct usb_device *udev, /* Caller has locked intf's usb_device's pm_mutex */ static int usb_resume_interface(struct usb_device *udev, - struct usb_interface *intf, int reset_resume) + struct usb_interface *intf, pm_message_t msg, int reset_resume) { struct usb_driver *driver; int status = 0; @@ -1138,10 +1138,9 @@ static inline int autosuspend_check(struct usb_device *udev, int reschedule) * all the interfaces which were suspended are resumed so that they remain * in the same state as the device. * - * If an autosuspend is in progress (@udev->auto_pm is set), the routine - * checks first to make sure that neither the device itself or any of its - * active interfaces is in use (pm_usage_cnt is greater than 0). If they - * are, the autosuspend fails. + * If an autosuspend is in progress the routine checks first to make sure + * that neither the device itself or any of its active interfaces is in use + * (pm_usage_cnt is greater than 0). If they are, the autosuspend fails. * * If the suspend succeeds, the routine recursively queues an autosuspend * request for @udev's parent device, thereby propagating the change up @@ -1176,7 +1175,7 @@ static int usb_suspend_both(struct usb_device *udev, pm_message_t msg) udev->do_remote_wakeup = device_may_wakeup(&udev->dev); - if (udev->auto_pm) { + if (msg.event & PM_EVENT_AUTO) { status = autosuspend_check(udev, 0); if (status < 0) goto done; @@ -1196,13 +1195,16 @@ static int usb_suspend_both(struct usb_device *udev, pm_message_t msg) /* If the suspend failed, resume interfaces that did get suspended */ if (status != 0) { + pm_message_t msg2; + + msg2.event = msg.event ^ (PM_EVENT_SUSPEND | PM_EVENT_RESUME); while (--i >= 0) { intf = udev->actconfig->interface[i]; - usb_resume_interface(udev, intf, 0); + usb_resume_interface(udev, intf, msg2, 0); } /* Try another autosuspend when the interfaces aren't busy */ - if (udev->auto_pm) + if (msg.event & PM_EVENT_AUTO) autosuspend_check(udev, status == -EBUSY); /* If the suspend succeeded then prevent any more URB submissions, @@ -1232,6 +1234,7 @@ static int usb_suspend_both(struct usb_device *udev, pm_message_t msg) /** * usb_resume_both - resume a USB device and its interfaces * @udev: the usb_device to resume + * @msg: Power Management message describing this state transition * * This is the central routine for resuming USB devices. It calls the * the resume method for @udev and then calls the resume methods for all @@ -1257,7 +1260,7 @@ static int usb_suspend_both(struct usb_device *udev, pm_message_t msg) * * This routine can run only in process context. */ -static int usb_resume_both(struct usb_device *udev) +static int usb_resume_both(struct usb_device *udev, pm_message_t msg) { int status = 0; int i; @@ -1273,14 +1276,15 @@ static int usb_resume_both(struct usb_device *udev) /* Propagate the resume up the tree, if necessary */ if (udev->state == USB_STATE_SUSPENDED) { - if (udev->auto_pm && udev->autoresume_disabled) { + if ((msg.event & PM_EVENT_AUTO) && + udev->autoresume_disabled) { status = -EPERM; goto done; } if (parent) { status = usb_autoresume_device(parent); if (status == 0) { - status = usb_resume_device(udev); + status = usb_resume_device(udev, msg); if (status || udev->state == USB_STATE_NOTATTACHED) { usb_autosuspend_device(parent); @@ -1303,15 +1307,16 @@ static int usb_resume_both(struct usb_device *udev) /* We can't progagate beyond the USB subsystem, * so if a root hub's controller is suspended * then we're stuck. */ - status = usb_resume_device(udev); + status = usb_resume_device(udev, msg); } } else if (udev->reset_resume) - status = usb_resume_device(udev); + status = usb_resume_device(udev, msg); if (status == 0 && udev->actconfig) { for (i = 0; i < udev->actconfig->desc.bNumInterfaces; i++) { intf = udev->actconfig->interface[i]; - usb_resume_interface(udev, intf, udev->reset_resume); + usb_resume_interface(udev, intf, msg, + udev->reset_resume); } } @@ -1339,13 +1344,13 @@ static int usb_autopm_do_device(struct usb_device *udev, int inc_usage_cnt) udev->last_busy = jiffies; if (inc_usage_cnt >= 0 && udev->pm_usage_cnt > 0) { if (udev->state == USB_STATE_SUSPENDED) - status = usb_resume_both(udev); + status = usb_resume_both(udev, PMSG_AUTO_RESUME); if (status != 0) udev->pm_usage_cnt -= inc_usage_cnt; else if (inc_usage_cnt) udev->last_busy = jiffies; } else if (inc_usage_cnt <= 0 && udev->pm_usage_cnt <= 0) { - status = usb_suspend_both(udev, PMSG_SUSPEND); + status = usb_suspend_both(udev, PMSG_AUTO_SUSPEND); } usb_pm_unlock(udev); return status; @@ -1469,13 +1474,14 @@ static int usb_autopm_do_interface(struct usb_interface *intf, udev->last_busy = jiffies; if (inc_usage_cnt >= 0 && intf->pm_usage_cnt > 0) { if (udev->state == USB_STATE_SUSPENDED) - status = usb_resume_both(udev); + status = usb_resume_both(udev, + PMSG_AUTO_RESUME); if (status != 0) intf->pm_usage_cnt -= inc_usage_cnt; else udev->last_busy = jiffies; } else if (inc_usage_cnt <= 0 && intf->pm_usage_cnt <= 0) { - status = usb_suspend_both(udev, PMSG_SUSPEND); + status = usb_suspend_both(udev, PMSG_AUTO_SUSPEND); } } usb_pm_unlock(udev); @@ -1700,6 +1706,7 @@ int usb_external_suspend_device(struct usb_device *udev, pm_message_t msg) /** * usb_external_resume_device - external resume of a USB device and its interfaces * @udev: the usb_device to resume + * @msg: Power Management message describing this state transition * * This routine handles external resume requests: ones not generated * internally by a USB driver (autoresume) but rather coming from the user @@ -1708,13 +1715,13 @@ int usb_external_suspend_device(struct usb_device *udev, pm_message_t msg) * * The caller must hold @udev's device lock. */ -int usb_external_resume_device(struct usb_device *udev) +int usb_external_resume_device(struct usb_device *udev, pm_message_t msg) { int status; usb_pm_lock(udev); udev->auto_pm = 0; - status = usb_resume_both(udev); + status = usb_resume_both(udev, msg); udev->last_busy = jiffies; usb_pm_unlock(udev); if (status == 0) @@ -1727,7 +1734,7 @@ int usb_external_resume_device(struct usb_device *udev) return status; } -int usb_suspend(struct device *dev, pm_message_t message) +int usb_suspend(struct device *dev, pm_message_t msg) { struct usb_device *udev; @@ -1746,10 +1753,10 @@ int usb_suspend(struct device *dev, pm_message_t message) } udev->skip_sys_resume = 0; - return usb_external_suspend_device(udev, message); + return usb_external_suspend_device(udev, msg); } -int usb_resume(struct device *dev) +int usb_resume(struct device *dev, pm_message_t msg) { struct usb_device *udev; @@ -1761,7 +1768,7 @@ int usb_resume(struct device *dev) */ if (udev->skip_sys_resume) return 0; - return usb_external_resume_device(udev); + return usb_external_resume_device(udev, msg); } #endif /* CONFIG_PM */ diff --git a/drivers/usb/core/generic.c b/drivers/usb/core/generic.c index 7e912f21fd3..30ecac3af15 100644 --- a/drivers/usb/core/generic.c +++ b/drivers/usb/core/generic.c @@ -200,18 +200,18 @@ static int generic_suspend(struct usb_device *udev, pm_message_t msg) * interfaces manually by doing a bus (or "global") suspend. */ if (!udev->parent) - rc = hcd_bus_suspend(udev); + rc = hcd_bus_suspend(udev, msg); /* Non-root devices don't need to do anything for FREEZE or PRETHAW */ else if (msg.event == PM_EVENT_FREEZE || msg.event == PM_EVENT_PRETHAW) rc = 0; else - rc = usb_port_suspend(udev); + rc = usb_port_suspend(udev, msg); return rc; } -static int generic_resume(struct usb_device *udev) +static int generic_resume(struct usb_device *udev, pm_message_t msg) { int rc; @@ -221,9 +221,9 @@ static int generic_resume(struct usb_device *udev) * interfaces manually by doing a bus (or "global") resume. */ if (!udev->parent) - rc = hcd_bus_resume(udev); + rc = hcd_bus_resume(udev, msg); else - rc = usb_port_resume(udev); + rc = usb_port_resume(udev, msg); return rc; } diff --git a/drivers/usb/core/hcd.c b/drivers/usb/core/hcd.c index 7403ed871ab..a0079876d74 100644 --- a/drivers/usb/core/hcd.c +++ b/drivers/usb/core/hcd.c @@ -1573,14 +1573,14 @@ int usb_hcd_get_frame_number (struct usb_device *udev) #ifdef CONFIG_PM -int hcd_bus_suspend(struct usb_device *rhdev) +int hcd_bus_suspend(struct usb_device *rhdev, pm_message_t msg) { struct usb_hcd *hcd = container_of(rhdev->bus, struct usb_hcd, self); int status; int old_state = hcd->state; dev_dbg(&rhdev->dev, "bus %s%s\n", - rhdev->auto_pm ? "auto-" : "", "suspend"); + (msg.event & PM_EVENT_AUTO ? "auto-" : ""), "suspend"); if (!hcd->driver->bus_suspend) { status = -ENOENT; } else { @@ -1598,14 +1598,14 @@ int hcd_bus_suspend(struct usb_device *rhdev) return status; } -int hcd_bus_resume(struct usb_device *rhdev) +int hcd_bus_resume(struct usb_device *rhdev, pm_message_t msg) { struct usb_hcd *hcd = container_of(rhdev->bus, struct usb_hcd, self); int status; int old_state = hcd->state; dev_dbg(&rhdev->dev, "usb %s%s\n", - rhdev->auto_pm ? "auto-" : "", "resume"); + (msg.event & PM_EVENT_AUTO ? "auto-" : ""), "resume"); if (!hcd->driver->bus_resume) return -ENOENT; if (hcd->state == HC_STATE_RUNNING) @@ -1638,7 +1638,7 @@ static void hcd_resume_work(struct work_struct *work) usb_lock_device(udev); usb_mark_last_busy(udev); - usb_external_resume_device(udev); + usb_external_resume_device(udev, PMSG_REMOTE_RESUME); usb_unlock_device(udev); } diff --git a/drivers/usb/core/hcd.h b/drivers/usb/core/hcd.h index 0aaa9cea6b3..aa5da82d907 100644 --- a/drivers/usb/core/hcd.h +++ b/drivers/usb/core/hcd.h @@ -388,8 +388,8 @@ extern int usb_find_interface_driver(struct usb_device *dev, #ifdef CONFIG_PM extern void usb_hcd_resume_root_hub(struct usb_hcd *hcd); extern void usb_root_hub_lost_power(struct usb_device *rhdev); -extern int hcd_bus_suspend(struct usb_device *rhdev); -extern int hcd_bus_resume(struct usb_device *rhdev); +extern int hcd_bus_suspend(struct usb_device *rhdev, pm_message_t msg); +extern int hcd_bus_resume(struct usb_device *rhdev, pm_message_t msg); #else static inline void usb_hcd_resume_root_hub(struct usb_hcd *hcd) { diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c index ff066edf4dc..fc99ef67761 100644 --- a/drivers/usb/core/hub.c +++ b/drivers/usb/core/hub.c @@ -1984,7 +1984,7 @@ static int check_port_resume_type(struct usb_device *udev, * * Returns 0 on success, else negative errno. */ -int usb_port_suspend(struct usb_device *udev) +int usb_port_suspend(struct usb_device *udev, pm_message_t msg) { struct usb_hub *hub = hdev_to_hub(udev->parent); int port1 = udev->portnum; @@ -2023,7 +2023,7 @@ int usb_port_suspend(struct usb_device *udev) } else { /* device has up to 10 msec to fully suspend */ dev_dbg(&udev->dev, "usb %ssuspend\n", - udev->auto_pm ? "auto-" : ""); + (msg.event & PM_EVENT_AUTO ? "auto-" : "")); usb_set_device_state(udev, USB_STATE_SUSPENDED); msleep(10); } @@ -2142,7 +2142,7 @@ static int finish_port_resume(struct usb_device *udev) * * Returns 0 on success, else negative errno. */ -int usb_port_resume(struct usb_device *udev) +int usb_port_resume(struct usb_device *udev, pm_message_t msg) { struct usb_hub *hub = hdev_to_hub(udev->parent); int port1 = udev->portnum; @@ -2167,7 +2167,7 @@ int usb_port_resume(struct usb_device *udev) } else { /* drive resume for at least 20 msec */ dev_dbg(&udev->dev, "usb %sresume\n", - udev->auto_pm ? "auto-" : ""); + (msg.event & PM_EVENT_AUTO ? "auto-" : "")); msleep(25); /* Virtual root hubs can trigger on GET_PORT_STATUS to @@ -2208,7 +2208,7 @@ static int remote_wakeup(struct usb_device *udev) if (udev->state == USB_STATE_SUSPENDED) { dev_dbg(&udev->dev, "usb %sresume\n", "wakeup-"); usb_mark_last_busy(udev); - status = usb_external_resume_device(udev); + status = usb_external_resume_device(udev, PMSG_REMOTE_RESUME); } return status; } @@ -2217,14 +2217,14 @@ static int remote_wakeup(struct usb_device *udev) /* When CONFIG_USB_SUSPEND isn't set, we never suspend or resume any ports. */ -int usb_port_suspend(struct usb_device *udev) +int usb_port_suspend(struct usb_device *udev, pm_message_t msg) { return 0; } /* However we may need to do a reset-resume */ -int usb_port_resume(struct usb_device *udev) +int usb_port_resume(struct usb_device *udev, pm_message_t msg) { struct usb_hub *hub = hdev_to_hub(udev->parent); int port1 = udev->portnum; @@ -2264,7 +2264,7 @@ static int hub_suspend(struct usb_interface *intf, pm_message_t msg) udev = hdev->children [port1-1]; if (udev && udev->can_submit) { - if (!hdev->auto_pm) + if (!(msg.event & PM_EVENT_AUTO)) dev_dbg(&intf->dev, "port %d nyet suspended\n", port1); return -EBUSY; diff --git a/drivers/usb/core/sysfs.c b/drivers/usb/core/sysfs.c index 8c65fa75b5c..0f0ccf64011 100644 --- a/drivers/usb/core/sysfs.c +++ b/drivers/usb/core/sysfs.c @@ -359,19 +359,19 @@ set_level(struct device *dev, struct device_attribute *attr, strncmp(buf, on_string, len) == 0) { udev->autosuspend_disabled = 1; udev->autoresume_disabled = 0; - rc = usb_external_resume_device(udev); + rc = usb_external_resume_device(udev, PMSG_USER_RESUME); } else if (len == sizeof auto_string - 1 && strncmp(buf, auto_string, len) == 0) { udev->autosuspend_disabled = 0; udev->autoresume_disabled = 0; - rc = usb_external_resume_device(udev); + rc = usb_external_resume_device(udev, PMSG_USER_RESUME); } else if (len == sizeof suspend_string - 1 && strncmp(buf, suspend_string, len) == 0) { udev->autosuspend_disabled = 0; udev->autoresume_disabled = 1; - rc = usb_external_suspend_device(udev, PMSG_SUSPEND); + rc = usb_external_suspend_device(udev, PMSG_USER_SUSPEND); } else rc = -EINVAL; diff --git a/drivers/usb/core/usb.c b/drivers/usb/core/usb.c index 51854c2bc91..4c98f3975af 100644 --- a/drivers/usb/core/usb.c +++ b/drivers/usb/core/usb.c @@ -253,7 +253,7 @@ static int usb_dev_prepare(struct device *dev) static void usb_dev_complete(struct device *dev) { /* Currently used only for rebinding interfaces */ - usb_resume(dev); /* Implement eventually? */ + usb_resume(dev, PMSG_RESUME); /* Message event is meaningless */ } static int usb_dev_suspend(struct device *dev) @@ -263,7 +263,7 @@ static int usb_dev_suspend(struct device *dev) static int usb_dev_resume(struct device *dev) { - return usb_resume(dev); + return usb_resume(dev, PMSG_RESUME); } static int usb_dev_freeze(struct device *dev) @@ -273,7 +273,7 @@ static int usb_dev_freeze(struct device *dev) static int usb_dev_thaw(struct device *dev) { - return usb_resume(dev); + return usb_resume(dev, PMSG_THAW); } static int usb_dev_poweroff(struct device *dev) @@ -283,7 +283,7 @@ static int usb_dev_poweroff(struct device *dev) static int usb_dev_restore(struct device *dev) { - return usb_resume(dev); + return usb_resume(dev, PMSG_RESTORE); } static struct dev_pm_ops usb_device_pm_ops = { diff --git a/drivers/usb/core/usb.h b/drivers/usb/core/usb.h index b60ebb4de1a..9fb195665fa 100644 --- a/drivers/usb/core/usb.h +++ b/drivers/usb/core/usb.h @@ -1,3 +1,5 @@ +#include + /* Functions local to drivers/usb/core/ */ extern int usb_create_sysfs_dev_files(struct usb_device *dev); @@ -42,15 +44,16 @@ extern void usb_host_cleanup(void); #ifdef CONFIG_PM extern int usb_suspend(struct device *dev, pm_message_t msg); -extern int usb_resume(struct device *dev); +extern int usb_resume(struct device *dev, pm_message_t msg); extern void usb_autosuspend_work(struct work_struct *work); extern void usb_autoresume_work(struct work_struct *work); -extern int usb_port_suspend(struct usb_device *dev); -extern int usb_port_resume(struct usb_device *dev); +extern int usb_port_suspend(struct usb_device *dev, pm_message_t msg); +extern int usb_port_resume(struct usb_device *dev, pm_message_t msg); extern int usb_external_suspend_device(struct usb_device *udev, pm_message_t msg); -extern int usb_external_resume_device(struct usb_device *udev); +extern int usb_external_resume_device(struct usb_device *udev, + pm_message_t msg); static inline void usb_pm_lock(struct usb_device *udev) { @@ -64,12 +67,12 @@ static inline void usb_pm_unlock(struct usb_device *udev) #else -static inline int usb_port_suspend(struct usb_device *udev) +static inline int usb_port_suspend(struct usb_device *udev, pm_message_t msg) { return 0; } -static inline int usb_port_resume(struct usb_device *udev) +static inline int usb_port_resume(struct usb_device *udev, pm_message_t msg) { return 0; } diff --git a/include/linux/usb.h b/include/linux/usb.h index 8bc81bffc19..74d0b9990c7 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -1067,7 +1067,7 @@ struct usb_device_driver { void (*disconnect) (struct usb_device *udev); int (*suspend) (struct usb_device *udev, pm_message_t message); - int (*resume) (struct usb_device *udev); + int (*resume) (struct usb_device *udev, pm_message_t message); struct usbdrv_wrap drvwrap; unsigned int supports_autosuspend:1; }; From 619b572ae205cea26b147c2cec5ebf19974f4bb0 Mon Sep 17 00:00:00 2001 From: Roel Kluin Date: Tue, 9 Dec 2008 22:50:22 +0100 Subject: [PATCH 057/123] USB: serial: usb_debug: Make static Sparse asked whether these could be static. Signed-off-by: Roel Kluin Signed-off-by: Greg Kroah-Hartman --- drivers/usb/serial/usb_debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/serial/usb_debug.c b/drivers/usb/serial/usb_debug.c index fc5d9952b03..6c9cbb59552 100644 --- a/drivers/usb/serial/usb_debug.c +++ b/drivers/usb/serial/usb_debug.c @@ -31,7 +31,7 @@ static struct usb_driver debug_driver = { .no_dynamic_id = 1, }; -int usb_debug_open(struct tty_struct *tty, struct usb_serial_port *port, +static int usb_debug_open(struct tty_struct *tty, struct usb_serial_port *port, struct file *filp) { port->bulk_out_size = USB_DEBUG_MAX_PACKET_SIZE; From 889394b1eb95f040525d06dd4f2f1222b94023e2 Mon Sep 17 00:00:00 2001 From: Felipe Balbi Date: Mon, 8 Dec 2008 13:50:27 +0200 Subject: [PATCH 058/123] USB: gadget: don't wait for completion twice In some obscure scenarios e.g. passing a 0-byte backing file storage, wait_for_completion() would wait forever in fsg_cleanup(). Prevent it by completing the thread in fsg_bind() error path. Signed-off-by: Felipe Balbi Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/file_storage.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/usb/gadget/file_storage.c b/drivers/usb/gadget/file_storage.c index 93933155e81..b10fa31cc91 100644 --- a/drivers/usb/gadget/file_storage.c +++ b/drivers/usb/gadget/file_storage.c @@ -4182,6 +4182,7 @@ out: fsg->state = FSG_STATE_TERMINATED; // The thread is dead fsg_unbind(gadget); close_all_backing_files(fsg); + complete(&fsg->thread_notifier); return rc; } From 6c0735687d37e25a65866823881bcbf39a6a023f Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Sun, 30 Nov 2008 16:50:04 +0100 Subject: [PATCH 059/123] USB: isp1760: use a specific PLX bridge instead of any bdridge this driver can't handle (of course) any brdige class devices. So we now are just active on one specific bridge which should be only the isp1761 chip behind a PLX bridge. Signed-off-by: Sebastian Andrzej Siewior Tested-by: Karl Bongers Cc: stable Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/isp1760-if.c | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/drivers/usb/host/isp1760-if.c b/drivers/usb/host/isp1760-if.c index b87ca7cf4b3..a53114c367c 100644 --- a/drivers/usb/host/isp1760-if.c +++ b/drivers/usb/host/isp1760-if.c @@ -268,12 +268,16 @@ static void isp1761_pci_shutdown(struct pci_dev *dev) printk(KERN_ERR "ips1761_pci_shutdown\n"); } -static const struct pci_device_id isp1760_plx [] = { { - /* handle any USB 2.0 EHCI controller */ - PCI_DEVICE_CLASS(((PCI_CLASS_BRIDGE_OTHER << 8) | (0x06 << 16)), ~0), - .driver_data = 0, -}, -{ /* end: all zeroes */ } +static const struct pci_device_id isp1760_plx [] = { + { + .class = PCI_CLASS_BRIDGE_OTHER << 8, + .class_mask = ~0, + .vendor = PCI_VENDOR_ID_PLX, + .device = 0x5406, + .subvendor = PCI_VENDOR_ID_PLX, + .subdevice = 0x9054, + }, + { } }; MODULE_DEVICE_TABLE(pci, isp1760_plx); From 6013bbbab0dcbc43bcf9dd70beeab2a0b1ec5ea7 Mon Sep 17 00:00:00 2001 From: Karl Bongers Date: Mon, 1 Dec 2008 11:47:40 +0100 Subject: [PATCH 060/123] USB: isp1760: Fix probe in PCI glue code Contains fixes so probe on x86 PCI runs, apparently I'm first to try this. Several fixes to memory access to probe host scratch register. Previously would bug check on chip_addr var used uninitialized. Scratch reg write failed in one instance due to 16-bit initial access mode, so added "& 0x0000ffff" to the readl as fix. Includes some general cleanup - remove global vars, organize memory map resource use. Signed-off-by: Karl Bongers Signed-off-by: Sebastian Andrzej Siewior Cc: stable Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/isp1760-if.c | 97 ++++++++++++++++++++--------------- 1 file changed, 55 insertions(+), 42 deletions(-) diff --git a/drivers/usb/host/isp1760-if.c b/drivers/usb/host/isp1760-if.c index a53114c367c..25453066fda 100644 --- a/drivers/usb/host/isp1760-if.c +++ b/drivers/usb/host/isp1760-if.c @@ -129,23 +129,23 @@ static struct of_platform_driver isp1760_of_driver = { #endif #ifdef CONFIG_PCI -static u32 nxp_pci_io_base; -static u32 iolength; -static u32 pci_mem_phy0; -static u32 length; -static u8 __iomem *chip_addr; -static u8 __iomem *iobase; - static int __devinit isp1761_pci_probe(struct pci_dev *dev, const struct pci_device_id *id) { u8 latency, limit; __u32 reg_data; int retry_count; - int length; - int status = 1; struct usb_hcd *hcd; unsigned int devflags = 0; + int ret_status = 0; + + resource_size_t pci_mem_phy0; + resource_size_t memlength; + + u8 __iomem *chip_addr; + u8 __iomem *iobase; + resource_size_t nxp_pci_io_base; + resource_size_t iolength; if (usb_disabled()) return -ENODEV; @@ -168,26 +168,30 @@ static int __devinit isp1761_pci_probe(struct pci_dev *dev, iobase = ioremap_nocache(nxp_pci_io_base, iolength); if (!iobase) { printk(KERN_ERR "ioremap #1\n"); - release_mem_region(nxp_pci_io_base, iolength); - return -ENOMEM; + ret_status = -ENOMEM; + goto cleanup1; } /* Grab the PLX PCI shared memory of the ISP 1761 we need */ pci_mem_phy0 = pci_resource_start(dev, 3); - length = pci_resource_len(dev, 3); - - if (length < 0xffff) { - printk(KERN_ERR "memory length for this resource is less than " - "required\n"); - release_mem_region(nxp_pci_io_base, iolength); - iounmap(iobase); - return -ENOMEM; + memlength = pci_resource_len(dev, 3); + if (memlength < 0xffff) { + printk(KERN_ERR "memory length for this resource is wrong\n"); + ret_status = -ENOMEM; + goto cleanup2; } - if (!request_mem_region(pci_mem_phy0, length, "ISP-PCI")) { + if (!request_mem_region(pci_mem_phy0, memlength, "ISP-PCI")) { printk(KERN_ERR "host controller already in use\n"); - release_mem_region(nxp_pci_io_base, iolength); - iounmap(iobase); - return -EBUSY; + ret_status = -EBUSY; + goto cleanup2; + } + + /* map available memory */ + chip_addr = ioremap_nocache(pci_mem_phy0,memlength); + if (!chip_addr) { + printk(KERN_ERR "Error ioremap failed\n"); + ret_status = -ENOMEM; + goto cleanup3; } /* bad pci latencies can contribute to overruns */ @@ -210,39 +214,54 @@ static int __devinit isp1761_pci_probe(struct pci_dev *dev, * */ writel(0xface, chip_addr + HC_SCRATCH_REG); udelay(100); - reg_data = readl(chip_addr + HC_SCRATCH_REG); + reg_data = readl(chip_addr + HC_SCRATCH_REG) & 0x0000ffff; retry_count--; } + iounmap(chip_addr); + /* Host Controller presence is detected by writing to scratch register * and reading back and checking the contents are same or not */ if (reg_data != 0xFACE) { dev_err(&dev->dev, "scratch register mismatch %x\n", reg_data); - goto clean; + ret_status = -ENOMEM; + goto cleanup3; } pci_set_master(dev); - status = readl(iobase + 0x68); - status |= 0x900; - writel(status, iobase + 0x68); + /* configure PLX PCI chip to pass interrupts */ +#define PLX_INT_CSR_REG 0x68 + reg_data = readl(iobase + PLX_INT_CSR_REG); + reg_data |= 0x900; + writel(reg_data, iobase + PLX_INT_CSR_REG); dev->dev.dma_mask = NULL; - hcd = isp1760_register(pci_mem_phy0, length, dev->irq, + hcd = isp1760_register(pci_mem_phy0, memlength, dev->irq, IRQF_SHARED | IRQF_DISABLED, &dev->dev, dev_name(&dev->dev), devflags); - if (!IS_ERR(hcd)) { - pci_set_drvdata(dev, hcd); - return 0; + if (IS_ERR(hcd)) { + ret_status = -ENODEV; + goto cleanup3; } -clean: - status = -ENODEV; + + /* done with PLX IO access */ iounmap(iobase); - release_mem_region(pci_mem_phy0, length); release_mem_region(nxp_pci_io_base, iolength); - return status; + + pci_set_drvdata(dev, hcd); + return 0; + +cleanup3: + release_mem_region(pci_mem_phy0, memlength); +cleanup2: + iounmap(iobase); +cleanup1: + release_mem_region(nxp_pci_io_base, iolength); + return ret_status; } + static void isp1761_pci_remove(struct pci_dev *dev) { struct usb_hcd *hcd; @@ -255,12 +274,6 @@ static void isp1761_pci_remove(struct pci_dev *dev) usb_put_hcd(hcd); pci_disable_device(dev); - - iounmap(iobase); - iounmap(chip_addr); - - release_mem_region(nxp_pci_io_base, iolength); - release_mem_region(pci_mem_phy0, length); } static void isp1761_pci_shutdown(struct pci_dev *dev) From 50de36f7af3b1d791c402478210790c582126fe5 Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Wed, 10 Dec 2008 16:00:30 -0800 Subject: [PATCH 061/123] USB: serial: fix up urb->status usage Some of the usb-serial drivers are starting to use urb->status in ways they should not be doing. This fixes up some of them to prevent that. Signed-off-by: Greg Kroah-Hartman --- drivers/usb/serial/digi_acceleport.c | 28 ++++++++------------ drivers/usb/serial/garmin_gps.c | 2 +- drivers/usb/serial/iuu_phoenix.c | 38 +++++++++++++++------------- drivers/usb/serial/mos7840.c | 38 ++++++++++++++++++---------- drivers/usb/serial/option.c | 7 +---- drivers/usb/serial/spcp8x5.c | 20 +++++++-------- 6 files changed, 68 insertions(+), 65 deletions(-) diff --git a/drivers/usb/serial/digi_acceleport.c b/drivers/usb/serial/digi_acceleport.c index 69f84f0ea6f..38ba4ea8b6b 100644 --- a/drivers/usb/serial/digi_acceleport.c +++ b/drivers/usb/serial/digi_acceleport.c @@ -635,8 +635,7 @@ static int digi_write_oob_command(struct usb_serial_port *port, spin_lock_irqsave(&oob_priv->dp_port_lock, flags); while (count > 0) { - while (oob_port->write_urb->status == -EINPROGRESS - || oob_priv->dp_write_urb_in_use) { + while (oob_priv->dp_write_urb_in_use) { cond_wait_interruptible_timeout_irqrestore( &oob_port->write_wait, DIGI_RETRY_TIMEOUT, &oob_priv->dp_port_lock, flags); @@ -699,9 +698,8 @@ static int digi_write_inb_command(struct usb_serial_port *port, spin_lock_irqsave(&priv->dp_port_lock, flags); while (count > 0 && ret == 0) { - while ((port->write_urb->status == -EINPROGRESS - || priv->dp_write_urb_in_use) - && time_before(jiffies, timeout)) { + while (priv->dp_write_urb_in_use && + time_before(jiffies, timeout)) { cond_wait_interruptible_timeout_irqrestore( &port->write_wait, DIGI_RETRY_TIMEOUT, &priv->dp_port_lock, flags); @@ -779,8 +777,7 @@ static int digi_set_modem_signals(struct usb_serial_port *port, spin_lock_irqsave(&oob_priv->dp_port_lock, flags); spin_lock(&port_priv->dp_port_lock); - while (oob_port->write_urb->status == -EINPROGRESS || - oob_priv->dp_write_urb_in_use) { + while (oob_priv->dp_write_urb_in_use) { spin_unlock(&port_priv->dp_port_lock); cond_wait_interruptible_timeout_irqrestore( &oob_port->write_wait, DIGI_RETRY_TIMEOUT, @@ -1168,12 +1165,10 @@ static int digi_write(struct tty_struct *tty, struct usb_serial_port *port, /* be sure only one write proceeds at a time */ /* there are races on the port private buffer */ - /* and races to check write_urb->status */ spin_lock_irqsave(&priv->dp_port_lock, flags); /* wait for urb status clear to submit another urb */ - if (port->write_urb->status == -EINPROGRESS || - priv->dp_write_urb_in_use) { + if (priv->dp_write_urb_in_use) { /* buffer data if count is 1 (probably put_char) if possible */ if (count == 1 && priv->dp_out_buf_len < DIGI_OUT_BUF_SIZE) { priv->dp_out_buf[priv->dp_out_buf_len++] = *buf; @@ -1236,7 +1231,7 @@ static void digi_write_bulk_callback(struct urb *urb) int ret = 0; int status = urb->status; - dbg("digi_write_bulk_callback: TOP, urb->status=%d", status); + dbg("digi_write_bulk_callback: TOP, status=%d", status); /* port and serial sanity check */ if (port == NULL || (priv = usb_get_serial_port_data(port)) == NULL) { @@ -1266,8 +1261,7 @@ static void digi_write_bulk_callback(struct urb *urb) /* try to send any buffered data on this port, if it is open */ spin_lock(&priv->dp_port_lock); priv->dp_write_urb_in_use = 0; - if (port->port.count && port->write_urb->status != -EINPROGRESS - && priv->dp_out_buf_len > 0) { + if (port->port.count && priv->dp_out_buf_len > 0) { *((unsigned char *)(port->write_urb->transfer_buffer)) = (unsigned char)DIGI_CMD_SEND_DATA; *((unsigned char *)(port->write_urb->transfer_buffer) + 1) @@ -1305,8 +1299,7 @@ static int digi_write_room(struct tty_struct *tty) spin_lock_irqsave(&priv->dp_port_lock, flags); - if (port->write_urb->status == -EINPROGRESS || - priv->dp_write_urb_in_use) + if (priv->dp_write_urb_in_use) room = 0; else room = port->bulk_out_size - 2 - priv->dp_out_buf_len; @@ -1322,8 +1315,7 @@ static int digi_chars_in_buffer(struct tty_struct *tty) struct usb_serial_port *port = tty->driver_data; struct digi_port *priv = usb_get_serial_port_data(port); - if (port->write_urb->status == -EINPROGRESS - || priv->dp_write_urb_in_use) { + if (priv->dp_write_urb_in_use) { dbg("digi_chars_in_buffer: port=%d, chars=%d", priv->dp_port_num, port->bulk_out_size - 2); /* return(port->bulk_out_size - 2); */ @@ -1702,7 +1694,7 @@ static int digi_read_inb_callback(struct urb *urb) /* short/multiple packet check */ if (urb->actual_length != len + 2) { dev_err(&port->dev, "%s: INCOMPLETE OR MULTIPLE PACKET, " - "urb->status=%d, port=%d, opcode=%d, len=%d, " + "status=%d, port=%d, opcode=%d, len=%d, " "actual_length=%d, status=%d\n", __func__, status, priv->dp_port_num, opcode, len, urb->actual_length, port_status); diff --git a/drivers/usb/serial/garmin_gps.c b/drivers/usb/serial/garmin_gps.c index 8e6a66e38db..a26a0e2cdb4 100644 --- a/drivers/usb/serial/garmin_gps.c +++ b/drivers/usb/serial/garmin_gps.c @@ -1056,7 +1056,7 @@ static void garmin_write_bulk_callback(struct urb *urb) if (status) { dbg("%s - nonzero write bulk status received: %d", - __func__, urb->status); + __func__, status); spin_lock_irqsave(&garmin_data_p->lock, flags); garmin_data_p->flags |= CLEAR_HALT_REQUIRED; spin_unlock_irqrestore(&garmin_data_p->lock, flags); diff --git a/drivers/usb/serial/iuu_phoenix.c b/drivers/usb/serial/iuu_phoenix.c index e320972cb22..2314c6ae4fc 100644 --- a/drivers/usb/serial/iuu_phoenix.c +++ b/drivers/usb/serial/iuu_phoenix.c @@ -190,10 +190,12 @@ static void iuu_rxcmd(struct urb *urb) { struct usb_serial_port *port = urb->context; int result; + int status = urb->status; + dbg("%s - enter", __func__); - if (urb->status) { - dbg("%s - urb->status = %d", __func__, urb->status); + if (status) { + dbg("%s - status = %d", __func__, status); /* error stop all */ return; } @@ -245,10 +247,12 @@ static void iuu_update_status_callback(struct urb *urb) struct usb_serial_port *port = urb->context; struct iuu_private *priv = usb_get_serial_port_data(port); u8 *st; + int status = urb->status; + dbg("%s - enter", __func__); - if (urb->status) { - dbg("%s - urb->status = %d", __func__, urb->status); + if (status) { + dbg("%s - status = %d", __func__, status); /* error stop all */ return; } @@ -274,9 +278,9 @@ static void iuu_status_callback(struct urb *urb) { struct usb_serial_port *port = urb->context; int result; - dbg("%s - enter", __func__); + int status = urb->status; - dbg("%s - urb->status = %d", __func__, urb->status); + dbg("%s - status = %d", __func__, status); usb_fill_bulk_urb(port->read_urb, port->serial->dev, usb_rcvbulkpipe(port->serial->dev, port->bulk_in_endpointAddress), @@ -618,11 +622,12 @@ static void read_buf_callback(struct urb *urb) struct usb_serial_port *port = urb->context; unsigned char *data = urb->transfer_buffer; struct tty_struct *tty; - dbg("%s - urb->status = %d", __func__, urb->status); + int status = urb->status; - if (urb->status) { - dbg("%s - urb->status = %d", __func__, urb->status); - if (urb->status == -EPROTO) { + dbg("%s - status = %d", __func__, status); + + if (status) { + if (status == -EPROTO) { /* reschedule needed */ } return; @@ -695,7 +700,7 @@ static void iuu_uart_read_callback(struct urb *urb) struct usb_serial_port *port = urb->context; struct iuu_private *priv = usb_get_serial_port_data(port); unsigned long flags; - int status; + int status = urb->status; int error = 0; int len = 0; unsigned char *data = urb->transfer_buffer; @@ -703,8 +708,8 @@ static void iuu_uart_read_callback(struct urb *urb) dbg("%s - enter", __func__); - if (urb->status) { - dbg("%s - urb->status = %d", __func__, urb->status); + if (status) { + dbg("%s - status = %d", __func__, status); /* error stop all */ return; } @@ -782,12 +787,11 @@ static void read_rxcmd_callback(struct urb *urb) { struct usb_serial_port *port = urb->context; int result; - dbg("%s - enter", __func__); + int status = urb->status; - dbg("%s - urb->status = %d", __func__, urb->status); + dbg("%s - status = %d", __func__, status); - if (urb->status) { - dbg("%s - urb->status = %d", __func__, urb->status); + if (status) { /* error stop all */ return; } diff --git a/drivers/usb/serial/mos7840.c b/drivers/usb/serial/mos7840.c index 96a8c771321..2c20e88a91b 100644 --- a/drivers/usb/serial/mos7840.c +++ b/drivers/usb/serial/mos7840.c @@ -214,6 +214,7 @@ struct moschip_port { spinlock_t pool_lock; struct urb *write_urb_pool[NUM_URBS]; char busy[NUM_URBS]; + bool read_urb_busy; }; @@ -679,26 +680,30 @@ static void mos7840_bulk_in_callback(struct urb *urb) struct tty_struct *tty; int status = urb->status; - if (status) { - dbg("nonzero read bulk status received: %d", status); - return; - } - mos7840_port = urb->context; if (!mos7840_port) { dbg("%s", "NULL mos7840_port pointer \n"); + mos7840_port->read_urb_busy = false; + return; + } + + if (status) { + dbg("nonzero read bulk status received: %d", status); + mos7840_port->read_urb_busy = false; return; } port = (struct usb_serial_port *)mos7840_port->port; if (mos7840_port_paranoia_check(port, __func__)) { dbg("%s", "Port Paranoia failed \n"); + mos7840_port->read_urb_busy = false; return; } serial = mos7840_get_usb_serial(port, __func__); if (!serial) { dbg("%s\n", "Bad serial pointer "); + mos7840_port->read_urb_busy = false; return; } @@ -725,17 +730,19 @@ static void mos7840_bulk_in_callback(struct urb *urb) if (!mos7840_port->read_urb) { dbg("%s", "URB KILLED !!!\n"); + mos7840_port->read_urb_busy = false; return; } mos7840_port->read_urb->dev = serial->dev; + mos7840_port->read_urb_busy = true; retval = usb_submit_urb(mos7840_port->read_urb, GFP_ATOMIC); if (retval) { - dbg(" usb_submit_urb(read bulk) failed, retval = %d", - retval); + dbg("usb_submit_urb(read bulk) failed, retval = %d", retval); + mos7840_port->read_urb_busy = false; } } @@ -1055,10 +1062,12 @@ static int mos7840_open(struct tty_struct *tty, dbg("mos7840_open: bulkin endpoint is %d\n", port->bulk_in_endpointAddress); + mos7840_port->read_urb_busy = true; response = usb_submit_urb(mos7840_port->read_urb, GFP_KERNEL); if (response) { dev_err(&port->dev, "%s - Error %d submitting control urb\n", __func__, response); + mos7840_port->read_urb_busy = false; } /* initialize our wait queues */ @@ -1227,6 +1236,7 @@ static void mos7840_close(struct tty_struct *tty, if (mos7840_port->read_urb) { dbg("%s", "Shutdown bulk read\n"); usb_kill_urb(mos7840_port->read_urb); + mos7840_port->read_urb_busy = false; } if ((&mos7840_port->control_urb)) { dbg("%s", "Shutdown control read\n"); @@ -2043,14 +2053,14 @@ static void mos7840_change_port_settings(struct tty_struct *tty, Data = 0x0c; mos7840_set_uart_reg(port, INTERRUPT_ENABLE_REGISTER, Data); - if (mos7840_port->read_urb->status != -EINPROGRESS) { + if (mos7840_port->read_urb_busy == false) { mos7840_port->read_urb->dev = serial->dev; - + mos7840_port->read_urb_busy = true; status = usb_submit_urb(mos7840_port->read_urb, GFP_ATOMIC); - if (status) { - dbg(" usb_submit_urb(read bulk) failed, status = %d", + dbg("usb_submit_urb(read bulk) failed, status = %d", status); + mos7840_port->read_urb_busy = false; } } wake_up(&mos7840_port->delta_msr_wait); @@ -2117,12 +2127,14 @@ static void mos7840_set_termios(struct tty_struct *tty, return; } - if (mos7840_port->read_urb->status != -EINPROGRESS) { + if (mos7840_port->read_urb_busy == false) { mos7840_port->read_urb->dev = serial->dev; + mos7840_port->read_urb_busy = true; status = usb_submit_urb(mos7840_port->read_urb, GFP_ATOMIC); if (status) { - dbg(" usb_submit_urb(read bulk) failed, status = %d", + dbg("usb_submit_urb(read bulk) failed, status = %d", status); + mos7840_port->read_urb_busy = false; } } return; diff --git a/drivers/usb/serial/option.c b/drivers/usb/serial/option.c index 809697b3c7f..4c9497d05d1 100644 --- a/drivers/usb/serial/option.c +++ b/drivers/usb/serial/option.c @@ -654,10 +654,6 @@ static int option_write(struct tty_struct *tty, struct usb_serial_port *port, usb_unlink_urb(this_urb); continue; } - if (this_urb->status != 0) - dbg("usb_write %p failed (err=%d)", - this_urb, this_urb->status); - dbg("%s: endpoint %d buf %d", __func__, usb_pipeendpoint(this_urb->pipe), i); @@ -669,8 +665,7 @@ static int option_write(struct tty_struct *tty, struct usb_serial_port *port, err = usb_submit_urb(this_urb, GFP_ATOMIC); if (err) { dbg("usb_submit_urb %p (write bulk) failed " - "(%d, has %d)", this_urb, - err, this_urb->status); + "(%d)", this_urb, err); clear_bit(i, &portdata->out_busy); continue; } diff --git a/drivers/usb/serial/spcp8x5.c b/drivers/usb/serial/spcp8x5.c index a65bc2bd8e7..5e7528cc81a 100644 --- a/drivers/usb/serial/spcp8x5.c +++ b/drivers/usb/serial/spcp8x5.c @@ -709,21 +709,20 @@ static void spcp8x5_read_bulk_callback(struct urb *urb) unsigned char *data = urb->transfer_buffer; unsigned long flags; int i; - int result; - u8 status = 0; + int result = urb->status; + u8 status; char tty_flag; - dev_dbg(&port->dev, "start, urb->status = %d, " - "urb->actual_length = %d\n,", urb->status, urb->actual_length); + dev_dbg(&port->dev, "start, result = %d, urb->actual_length = %d\n,", + result, urb->actual_length); /* check the urb status */ - if (urb->status) { + if (result) { if (!port->port.count) return; - if (urb->status == -EPROTO) { + if (result == -EPROTO) { /* spcp8x5 mysteriously fails with -EPROTO */ /* reschedule the read */ - urb->status = 0; urb->dev = port->serial->dev; result = usb_submit_urb(urb , GFP_ATOMIC); if (result) @@ -833,8 +832,9 @@ static void spcp8x5_write_bulk_callback(struct urb *urb) struct usb_serial_port *port = urb->context; struct spcp8x5_private *priv = usb_get_serial_port_data(port); int result; + int status = urb->status; - switch (urb->status) { + switch (status) { case 0: /* success */ break; @@ -843,14 +843,14 @@ static void spcp8x5_write_bulk_callback(struct urb *urb) case -ESHUTDOWN: /* this urb is terminated, clean up */ dev_dbg(&port->dev, "urb shutting down with status: %d\n", - urb->status); + status); priv->write_urb_in_use = 0; return; default: /* error in the urb, so we have to resubmit it */ dbg("%s - Overflow in write", __func__); dbg("%s - nonzero write bulk status received: %d", - __func__, urb->status); + __func__, status); port->write_urb->transfer_buffer_length = 1; port->write_urb->dev = port->serial->dev; result = usb_submit_urb(port->write_urb, GFP_ATOMIC); From f1632df36b9467b75b7abfd2799aef67ec74a60a Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Tue, 16 Dec 2008 13:26:07 -0800 Subject: [PATCH 062/123] USB: option: increase outgoing buffer size and number This should speed up the option driver's upload speed quite a bit. It has been tested by a number of different people on different devices with success. Cc: Roland Wolters Signed-off-by: Greg Kroah-Hartman --- drivers/usb/serial/option.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/usb/serial/option.c b/drivers/usb/serial/option.c index 4c9497d05d1..5ed183477aa 100644 --- a/drivers/usb/serial/option.c +++ b/drivers/usb/serial/option.c @@ -522,9 +522,9 @@ static int debug; /* per port private data */ #define N_IN_URB 4 -#define N_OUT_URB 1 +#define N_OUT_URB 4 #define IN_BUFLEN 4096 -#define OUT_BUFLEN 128 +#define OUT_BUFLEN 4096 struct option_port_private { /* Input endpoints and buffer for this port */ From c20b15fde50c32174af4b48851e5ddadba36330e Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Mon, 1 Dec 2008 10:36:15 -0500 Subject: [PATCH 063/123] USB: usb-storage: merge DPCM support into SDDR09 The DPCM subdriver is a little peculiar, in that it's meant to support devices where LUN 0 is Compact Flash and uses the CB transport whereas LUN 1 is SmartMedia and uses the SDDR09 transport. Thus DPCM isn't really a transport in itself; it's more like a demultiplexer. Much of the DPCM code is part of the SDDR09 subdriver already, and the remaining part is fairly small. This patch (as1182) moves that extra piece into sddr09.c, thereby eliminating dpcm.c. Also eliminated is the Kconfig entry for DPCM support; it is now listed as part of the SDDR09 entry. In order to make sure that the semantics are the same as before, each unusual_devs entry for DPCM is now present twice: once with DPCM support if SDDR09 is configured (as before), and once with the SINGLE_LUN flag and CB support otherwise. Signed-off-by: Alan Stern CC: Matthew Dharm Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/Kconfig | 11 +--- drivers/usb/storage/Makefile | 1 - drivers/usb/storage/dpcm.c | 86 ------------------------------ drivers/usb/storage/dpcm.h | 32 ----------- drivers/usb/storage/sddr09.c | 43 +++++++++++++++ drivers/usb/storage/sddr09.h | 7 ++- drivers/usb/storage/unusual_devs.h | 22 +++++++- drivers/usb/storage/usb.c | 3 -- 8 files changed, 70 insertions(+), 135 deletions(-) delete mode 100644 drivers/usb/storage/dpcm.c delete mode 100644 drivers/usb/storage/dpcm.h diff --git a/drivers/usb/storage/Kconfig b/drivers/usb/storage/Kconfig index c68b738900b..9df6887b91f 100644 --- a/drivers/usb/storage/Kconfig +++ b/drivers/usb/storage/Kconfig @@ -61,13 +61,6 @@ config USB_STORAGE_ISD200 - CyQ've CQ8060A CDRW drive - Planex eXtreme Drive RX-25HU USB-IDE cable (not model RX-25U) -config USB_STORAGE_DPCM - bool "Microtech/ZiO! CompactFlash/SmartMedia support" - depends on USB_STORAGE - help - Say Y here to support the Microtech/ZiO! CompactFlash reader. - There is a web page at . - config USB_STORAGE_USBAT bool "USBAT/USBAT02-based storage support" depends on USB_STORAGE @@ -90,12 +83,12 @@ config USB_STORAGE_USBAT - Sandisk ImageMate SDDR-05b config USB_STORAGE_SDDR09 - bool "SanDisk SDDR-09 (and other SmartMedia) support" + bool "SanDisk SDDR-09 (and other SmartMedia, including DPCM) support" depends on USB_STORAGE help Say Y here to include additional code to support the Sandisk SDDR-09 SmartMedia reader in the USB Mass Storage driver. - Also works for the Microtech Zio! SmartMedia reader. + Also works for the Microtech Zio! CompactFlash/SmartMedia reader. config USB_STORAGE_SDDR55 bool "SanDisk SDDR-55 SmartMedia support" diff --git a/drivers/usb/storage/Makefile b/drivers/usb/storage/Makefile index 7f8beb5366a..facf610f168 100644 --- a/drivers/usb/storage/Makefile +++ b/drivers/usb/storage/Makefile @@ -14,7 +14,6 @@ usb-storage-obj-$(CONFIG_USB_STORAGE_USBAT) += shuttle_usbat.o usb-storage-obj-$(CONFIG_USB_STORAGE_SDDR09) += sddr09.o usb-storage-obj-$(CONFIG_USB_STORAGE_SDDR55) += sddr55.o usb-storage-obj-$(CONFIG_USB_STORAGE_FREECOM) += freecom.o -usb-storage-obj-$(CONFIG_USB_STORAGE_DPCM) += dpcm.o usb-storage-obj-$(CONFIG_USB_STORAGE_ISD200) += isd200.o usb-storage-obj-$(CONFIG_USB_STORAGE_DATAFAB) += datafab.o usb-storage-obj-$(CONFIG_USB_STORAGE_JUMPSHOT) += jumpshot.o diff --git a/drivers/usb/storage/dpcm.c b/drivers/usb/storage/dpcm.c deleted file mode 100644 index 939923471af..00000000000 --- a/drivers/usb/storage/dpcm.c +++ /dev/null @@ -1,86 +0,0 @@ -/* Driver for Microtech DPCM-USB CompactFlash/SmartMedia reader - * - * DPCM driver v0.1: - * - * First release - * - * Current development and maintenance by: - * (c) 2000 Brian Webb (webbb@earthlink.net) - * - * This device contains both a CompactFlash card reader, which - * uses the Control/Bulk w/o Interrupt protocol and - * a SmartMedia card reader that uses the same protocol - * as the SDDR09. - * - * This program is free software; you can redistribute it and/or modify it - * under the terms of the GNU General Public License as published by the - * Free Software Foundation; either version 2, or (at your option) any - * later version. - * - * This program is distributed in the hope that it will be useful, but - * WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public License along - * with this program; if not, write to the Free Software Foundation, Inc., - * 675 Mass Ave, Cambridge, MA 02139, USA. - */ - -#include -#include -#include - -#include "usb.h" -#include "transport.h" -#include "protocol.h" -#include "debug.h" -#include "dpcm.h" -#include "sddr09.h" - -/* - * Transport for the Microtech DPCM-USB - * - */ -int dpcm_transport(struct scsi_cmnd *srb, struct us_data *us) -{ - int ret; - - if (srb == NULL) - return USB_STOR_TRANSPORT_ERROR; - - US_DEBUGP("dpcm_transport: LUN=%d\n", srb->device->lun); - - switch (srb->device->lun) { - case 0: - - /* - * LUN 0 corresponds to the CompactFlash card reader. - */ - ret = usb_stor_CB_transport(srb, us); - break; - -#ifdef CONFIG_USB_STORAGE_SDDR09 - case 1: - - /* - * LUN 1 corresponds to the SmartMedia card reader. - */ - - /* - * Set the LUN to 0 (just in case). - */ - srb->device->lun = 0; us->srb->device->lun = 0; - ret = sddr09_transport(srb, us); - srb->device->lun = 1; us->srb->device->lun = 1; - break; - -#endif - - default: - US_DEBUGP("dpcm_transport: Invalid LUN %d\n", srb->device->lun); - ret = USB_STOR_TRANSPORT_ERROR; - break; - } - return ret; -} diff --git a/drivers/usb/storage/dpcm.h b/drivers/usb/storage/dpcm.h deleted file mode 100644 index e7b7b0f120d..00000000000 --- a/drivers/usb/storage/dpcm.h +++ /dev/null @@ -1,32 +0,0 @@ -/* Driver for Microtech DPCM-USB CompactFlash/SmartMedia reader - * - * DPCM driver v0.1: - * - * First release - * - * Current development and maintenance by: - * (c) 2000 Brian Webb (webbb@earthlink.net) - * - * See dpcm.c for more explanation - * - * This program is free software; you can redistribute it and/or modify it - * under the terms of the GNU General Public License as published by the - * Free Software Foundation; either version 2, or (at your option) any - * later version. - * - * This program is distributed in the hope that it will be useful, but - * WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public License along - * with this program; if not, write to the Free Software Foundation, Inc., - * 675 Mass Ave, Cambridge, MA 02139, USA. - */ - -#ifndef _MICROTECH_DPCM_USB_H -#define _MICROTECH_DPCM_USB_H - -extern int dpcm_transport(struct scsi_cmnd *srb, struct us_data *us); - -#endif diff --git a/drivers/usb/storage/sddr09.c b/drivers/usb/storage/sddr09.c index c5a54b872c2..531ae5c5abf 100644 --- a/drivers/usb/storage/sddr09.c +++ b/drivers/usb/storage/sddr09.c @@ -45,6 +45,7 @@ #include #include +#include #include "usb.h" #include "transport.h" @@ -1445,6 +1446,48 @@ usb_stor_sddr09_dpcm_init(struct us_data *us) { return 0; /* not result */ } +/* + * Transport for the Microtech DPCM-USB + */ +int dpcm_transport(struct scsi_cmnd *srb, struct us_data *us) +{ + int ret; + + US_DEBUGP("dpcm_transport: LUN=%d\n", srb->device->lun); + + switch (srb->device->lun) { + case 0: + + /* + * LUN 0 corresponds to the CompactFlash card reader. + */ + ret = usb_stor_CB_transport(srb, us); + break; + + case 1: + + /* + * LUN 1 corresponds to the SmartMedia card reader. + */ + + /* + * Set the LUN to 0 (just in case). + */ + srb->device->lun = 0; + ret = sddr09_transport(srb, us); + srb->device->lun = 1; + break; + + default: + US_DEBUGP("dpcm_transport: Invalid LUN %d\n", + srb->device->lun); + ret = USB_STOR_TRANSPORT_ERROR; + break; + } + return ret; +} + + /* * Transport for the Sandisk SDDR-09 */ diff --git a/drivers/usb/storage/sddr09.h b/drivers/usb/storage/sddr09.h index e50033ad7b1..b701172e12e 100644 --- a/drivers/usb/storage/sddr09.h +++ b/drivers/usb/storage/sddr09.h @@ -28,8 +28,11 @@ /* Sandisk SDDR-09 stuff */ extern int sddr09_transport(struct scsi_cmnd *srb, struct us_data *us); - -extern int usb_stor_sddr09_dpcm_init(struct us_data *us); extern int usb_stor_sddr09_init(struct us_data *us); +/* Microtech DPCM-USB stuff */ + +extern int dpcm_transport(struct scsi_cmnd *srb, struct us_data *us); +extern int usb_stor_sddr09_dpcm_init(struct us_data *us); + #endif diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index 74580977831..0fd42a0c794 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -296,11 +296,17 @@ UNUSUAL_DEV( 0x0424, 0x0fdc, 0x0210, 0x0210, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_SINGLE_LUN ), -#ifdef CONFIG_USB_STORAGE_DPCM +#ifdef CONFIG_USB_STORAGE_SDDR09 UNUSUAL_DEV( 0x0436, 0x0005, 0x0100, 0x0100, "Microtech", "CameraMate (DPCM_USB)", US_SC_SCSI, US_PR_DPCM_USB, NULL, 0 ), +#else +UNUSUAL_DEV( 0x0436, 0x0005, 0x0100, 0x0100, + "Microtech", + "CameraMate", + US_SC_SCSI, US_PR_CB, NULL, + US_FL_SINGLE_LUN ), #endif /* Patch submitted by Daniel Drake @@ -601,6 +607,12 @@ UNUSUAL_DEV( 0x04e6, 0x0005, 0x0100, 0x0208, "eUSB SmartMedia / CompactFlash Adapter", US_SC_SCSI, US_PR_DPCM_USB, usb_stor_sddr09_dpcm_init, 0), +#else +UNUSUAL_DEV( 0x04e6, 0x0005, 0x0100, 0x0208, + "SCM Microsystems", + "eUSB CompactFlash Adapter", + US_SC_SCSI, US_PR_CB, NULL, + US_FL_SINGLE_LUN), #endif /* Reported by Markus Demleitner */ @@ -1175,11 +1187,17 @@ UNUSUAL_DEV( 0x07af, 0x0005, 0x0100, 0x0100, US_SC_DEVICE, US_PR_DEVICE, usb_stor_euscsi_init, US_FL_SCM_MULT_TARG ), -#ifdef CONFIG_USB_STORAGE_DPCM +#ifdef CONFIG_USB_STORAGE_SDDR09 UNUSUAL_DEV( 0x07af, 0x0006, 0x0100, 0x0100, "Microtech", "CameraMate (DPCM_USB)", US_SC_SCSI, US_PR_DPCM_USB, NULL, 0 ), +#else +UNUSUAL_DEV( 0x07af, 0x0006, 0x0100, 0x0100, + "Microtech", + "CameraMate", + US_SC_SCSI, US_PR_CB, NULL, + US_FL_SINGLE_LUN ), #endif #ifdef CONFIG_USB_STORAGE_ALAUDA diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index 06c735703f4..b25c448d5eb 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -75,9 +75,6 @@ #ifdef CONFIG_USB_STORAGE_SDDR55 #include "sddr55.h" #endif -#ifdef CONFIG_USB_STORAGE_DPCM -#include "dpcm.h" -#endif #ifdef CONFIG_USB_STORAGE_FREECOM #include "freecom.h" #endif From 2ffcdb3bdadaf8260986e96384df26c94a6ad42c Mon Sep 17 00:00:00 2001 From: Bryan Wu Date: Tue, 2 Dec 2008 21:33:43 +0200 Subject: [PATCH 064/123] USB: musb: use new platform data interface of musb to replace old one Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- include/linux/usb/musb.h | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/include/linux/usb/musb.h b/include/linux/usb/musb.h index 630962c04ca..d6aad0ea603 100644 --- a/include/linux/usb/musb.h +++ b/include/linux/usb/musb.h @@ -47,6 +47,11 @@ struct musb_hdrc_config { u8 ram_bits; /* ram address size */ struct musb_hdrc_eps_bits *eps_bits; +#ifdef CONFIG_BLACKFIN + /* A GPIO controlling VRSEL in Blackfin */ + unsigned int gpio_vrsel; +#endif + }; struct musb_hdrc_platform_data { From 0c6a8818447d38f7bb0b0013448659113d37a3e1 Mon Sep 17 00:00:00 2001 From: Bryan Wu Date: Tue, 2 Dec 2008 21:33:44 +0200 Subject: [PATCH 065/123] USB: musb: add Blackfin driver to MUSB framework (v2) - replace MUSB_FIFOSIZE register to MUSB_TXCOUNT, cause no MUSB_FIFOSIZE register on Blackfin - use #ifdef to replace #if defined() Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/blackfin.c | 316 ++++++++++++++++++++++++++++++++++++ drivers/usb/musb/blackfin.h | 52 ++++++ 2 files changed, 368 insertions(+) create mode 100644 drivers/usb/musb/blackfin.c create mode 100644 drivers/usb/musb/blackfin.h diff --git a/drivers/usb/musb/blackfin.c b/drivers/usb/musb/blackfin.c new file mode 100644 index 00000000000..3fb719aecf0 --- /dev/null +++ b/drivers/usb/musb/blackfin.c @@ -0,0 +1,316 @@ +/* + * MUSB OTG controller driver for Blackfin Processors + * + * Copyright 2006-2008 Analog Devices Inc. + * + * Enter bugs at http://blackfin.uclinux.org/ + * + * Licensed under the GPL-2 or later. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include "musb_core.h" +#include "blackfin.h" + +/* + * Load an endpoint's FIFO + */ +void musb_write_fifo(struct musb_hw_ep *hw_ep, u16 len, const u8 *src) +{ + void __iomem *fifo = hw_ep->fifo; + void __iomem *epio = hw_ep->regs; + + prefetch((u8 *)src); + + musb_writew(epio, MUSB_TXCOUNT, len); + + DBG(4, "TX ep%d fifo %p count %d buf %p, epio %p\n", + hw_ep->epnum, fifo, len, src, epio); + + dump_fifo_data(src, len); + + if (unlikely((unsigned long)src & 0x01)) + outsw_8(fifo, src, len & 0x01 ? (len >> 1) + 1 : len >> 1); + else + outsw(fifo, src, len & 0x01 ? (len >> 1) + 1 : len >> 1); +} + +/* + * Unload an endpoint's FIFO + */ +void musb_read_fifo(struct musb_hw_ep *hw_ep, u16 len, u8 *dst) +{ + void __iomem *fifo = hw_ep->fifo; + u8 epnum = hw_ep->epnum; + u16 dma_reg = 0; + int i; + u16 *data; + + DBG(4, "%cX ep%d fifo %p count %d buf %p\n", + 'R', hw_ep->epnum, fifo, len, dst); + +#ifdef CONFIG_BF52x + invalidate_dcache_range((unsigned int)dst, + (unsigned int)(dst + len)); + + /* Setup DMA address register */ + dma_reg = (u16) ((u32) dst & 0xFFFF); + bfin_write16(USB_DMA_REG(epnum, USB_DMAx_ADDR_LOW), dma_reg); + SSYNC(); + + dma_reg = (u16) (((u32) dst >> 16) & 0xFFFF); + bfin_write16(USB_DMA_REG(epnum, USB_DMAx_ADDR_HIGH), dma_reg); + SSYNC(); + + /* Setup DMA count register */ + bfin_write16(USB_DMA_REG(epnum, USB_DMAx_COUNT_LOW), len); + bfin_write16(USB_DMA_REG(epnum, USB_DMAx_COUNT_HIGH), 0); + SSYNC(); + + /* Enable the DMA */ + dma_reg = (epnum << 4) | DMA_ENA | INT_ENA; + bfin_write16(USB_DMA_REG(epnum, USB_DMAx_CTRL), dma_reg); + SSYNC(); + + /* Wait for compelete */ + while (!(bfin_read_USB_DMA_INTERRUPT() & (1 << epnum))) + cpu_relax(); + + /* acknowledge dma interrupt */ + bfin_write_USB_DMA_INTERRUPT(1 << epnum); + SSYNC(); + + /* Reset DMA */ + bfin_write16(USB_DMA_REG(epnum, USB_DMAx_CTRL), 0); + SSYNC(); +#else + if (unlikely((unsigned long)dst & 0x01)) + insw_8(fifo, dst, len & 0x01 ? (len >> 1) + 1 : len >> 1); + else + insw(fifo, dst, len & 0x01 ? (len >> 1) + 1 : len >> 1); +#endif + + dump_fifo_data(dst, len); +} + +static irqreturn_t blackfin_interrupt(int irq, void *__hci) +{ + unsigned long flags; + irqreturn_t retval = IRQ_NONE; + struct musb *musb = __hci; + + spin_lock_irqsave(&musb->lock, flags); + + musb->int_usb = musb_readb(musb->mregs, MUSB_INTRUSB); + musb->int_tx = musb_readw(musb->mregs, MUSB_INTRTX); + musb->int_rx = musb_readw(musb->mregs, MUSB_INTRRX); + + if (musb->int_usb || musb->int_tx || musb->int_rx) { + musb_writeb(musb->mregs, MUSB_INTRUSB, musb->int_usb); + musb_writew(musb->mregs, MUSB_INTRTX, musb->int_tx); + musb_writew(musb->mregs, MUSB_INTRRX, musb->int_rx); + retval = musb_interrupt(musb); + } + + spin_unlock_irqrestore(&musb->lock, flags); + + /* REVISIT we sometimes get spurious IRQs on g_ep0 + * not clear why... fall in BF54x too. + */ + if (retval != IRQ_HANDLED) + DBG(5, "spurious?\n"); + + return IRQ_HANDLED; +} + +static void musb_conn_timer_handler(unsigned long _musb) +{ + struct musb *musb = (void *)_musb; + unsigned long flags; + u16 val; + + spin_lock_irqsave(&musb->lock, flags); + switch (musb->xceiv.state) { + case OTG_STATE_A_IDLE: + case OTG_STATE_A_WAIT_BCON: + /* Start a new session */ + val = musb_readw(musb->mregs, MUSB_DEVCTL); + val |= MUSB_DEVCTL_SESSION; + musb_writew(musb->mregs, MUSB_DEVCTL, val); + + val = musb_readw(musb->mregs, MUSB_DEVCTL); + if (!(val & MUSB_DEVCTL_BDEVICE)) { + gpio_set_value(musb->config->gpio_vrsel, 1); + musb->xceiv.state = OTG_STATE_A_WAIT_BCON; + } else { + gpio_set_value(musb->config->gpio_vrsel, 0); + + /* Ignore VBUSERROR and SUSPEND IRQ */ + val = musb_readb(musb->mregs, MUSB_INTRUSBE); + val &= ~MUSB_INTR_VBUSERROR; + musb_writeb(musb->mregs, MUSB_INTRUSBE, val); + + val = MUSB_INTR_SUSPEND | MUSB_INTR_VBUSERROR; + musb_writeb(musb->mregs, MUSB_INTRUSB, val); + + val = MUSB_POWER_HSENAB; + musb_writeb(musb->mregs, MUSB_POWER, val); + } + mod_timer(&musb_conn_timer, jiffies + TIMER_DELAY); + break; + + default: + DBG(1, "%s state not handled\n", otg_state_string(musb)); + break; + } + spin_unlock_irqrestore(&musb->lock, flags); + + DBG(4, "state is %s\n", otg_state_string(musb)); +} + +void musb_platform_enable(struct musb *musb) +{ + if (is_host_enabled(musb)) { + mod_timer(&musb_conn_timer, jiffies + TIMER_DELAY); + musb->a_wait_bcon = TIMER_DELAY; + } +} + +void musb_platform_disable(struct musb *musb) +{ +} + +static void bfin_vbus_power(struct musb *musb, int is_on, int sleeping) +{ +} + +static void bfin_set_vbus(struct musb *musb, int is_on) +{ + if (is_on) + gpio_set_value(musb->config->gpio_vrsel, 1); + else + gpio_set_value(musb->config->gpio_vrsel, 0); + + DBG(1, "VBUS %s, devctl %02x " + /* otg %3x conf %08x prcm %08x */ "\n", + otg_state_string(musb), + musb_readb(musb->mregs, MUSB_DEVCTL)); +} + +static int bfin_set_power(struct otg_transceiver *x, unsigned mA) +{ + return 0; +} + +void musb_platform_try_idle(struct musb *musb, unsigned long timeout) +{ + if (is_host_enabled(musb)) + mod_timer(&musb_conn_timer, jiffies + TIMER_DELAY); +} + +int musb_platform_get_vbus_status(struct musb *musb) +{ + return 0; +} + +void musb_platform_set_mode(struct musb *musb, u8 musb_mode) +{ +} + +int __init musb_platform_init(struct musb *musb) +{ + + /* + * Rev 1.0 BF549 EZ-KITs require PE7 to be high for both DEVICE + * and OTG HOST modes, while rev 1.1 and greater require PE7 to + * be low for DEVICE mode and high for HOST mode. We set it high + * here because we are in host mode + */ + + if (gpio_request(musb->config->gpio_vrsel, "USB_VRSEL")) { + printk(KERN_ERR "Failed ro request USB_VRSEL GPIO_%d \n", + musb->config->gpio_vrsel); + return -ENODEV; + } + gpio_direction_output(musb->config->gpio_vrsel, 0); + + /* Anomaly #05000346 */ + bfin_write_USB_APHY_CALIB(0x5411); + SSYNC(); + + /* Anomaly #05000347 */ + bfin_write_USB_APHY_CNTRL(0x0); + SSYNC(); + + /* TODO + * Set SIC-IVG register + */ + + /* Configure PLL oscillator register */ + bfin_write_USB_PLLOSC_CTRL(0x30a8); + SSYNC(); + + bfin_write_USB_SRP_CLKDIV((get_sclk()/1000) / 32 - 1); + SSYNC(); + + bfin_write_USB_EP_NI0_RXMAXP(64); + SSYNC(); + + bfin_write_USB_EP_NI0_TXMAXP(64); + SSYNC(); + + /* Route INTRUSB/INTR_RX/INTR_TX to USB_INT0*/ + bfin_write_USB_GLOBINTR(0x7); + SSYNC(); + + bfin_write_USB_GLOBAL_CTL(GLOBAL_ENA | EP1_TX_ENA | EP2_TX_ENA | + EP3_TX_ENA | EP4_TX_ENA | EP5_TX_ENA | + EP6_TX_ENA | EP7_TX_ENA | EP1_RX_ENA | + EP2_RX_ENA | EP3_RX_ENA | EP4_RX_ENA | + EP5_RX_ENA | EP6_RX_ENA | EP7_RX_ENA); + SSYNC(); + + if (is_host_enabled(musb)) { + musb->board_set_vbus = bfin_set_vbus; + setup_timer(&musb_conn_timer, + musb_conn_timer_handler, (unsigned long) musb); + } + if (is_peripheral_enabled(musb)) + musb->xceiv.set_power = bfin_set_power; + + musb->isr = blackfin_interrupt; + + return 0; +} + +int musb_platform_suspend(struct musb *musb) +{ + return 0; +} + +int musb_platform_resume(struct musb *musb) +{ + return 0; +} + + +int musb_platform_exit(struct musb *musb) +{ + + bfin_vbus_power(musb, 0 /*off*/, 1); + gpio_free(musb->config->gpio_vrsel); + musb_platform_suspend(musb); + + return 0; +} diff --git a/drivers/usb/musb/blackfin.h b/drivers/usb/musb/blackfin.h new file mode 100644 index 00000000000..a240c1e53d1 --- /dev/null +++ b/drivers/usb/musb/blackfin.h @@ -0,0 +1,52 @@ +/* + * Copyright (C) 2007 by Analog Devices, Inc. + * + * The Inventra Controller Driver for Linux is free software; you + * can redistribute it and/or modify it under the terms of the GNU + * General Public License version 2 as published by the Free Software + * Foundation. + */ + +#ifndef __MUSB_BLACKFIN_H__ +#define __MUSB_BLACKFIN_H__ + +/* + * Blackfin specific definitions + */ + +#undef DUMP_FIFO_DATA +#ifdef DUMP_FIFO_DATA +static void dump_fifo_data(u8 *buf, u16 len) +{ + u8 *tmp = buf; + int i; + + for (i = 0; i < len; i++) { + if (!(i % 16) && i) + pr_debug("\n"); + pr_debug("%02x ", *tmp++); + } + pr_debug("\n"); +} +#else +#define dump_fifo_data(buf, len) do {} while (0) +#endif + +#ifdef CONFIG_BF52x + +#define USB_DMA_BASE USB_DMA_INTERRUPT +#define USB_DMAx_CTRL 0x04 +#define USB_DMAx_ADDR_LOW 0x08 +#define USB_DMAx_ADDR_HIGH 0x0C +#define USB_DMAx_COUNT_LOW 0x10 +#define USB_DMAx_COUNT_HIGH 0x14 + +#define USB_DMA_REG(ep, reg) (USB_DMA_BASE + 0x20 * ep + reg) +#endif + +/* Almost 1 second */ +#define TIMER_DELAY (1 * HZ) + +static struct timer_list musb_conn_timer; + +#endif /* __MUSB_BLACKFIN_H__ */ From d426e60dbf6aa2c3199e37a59c6d134eb204d628 Mon Sep 17 00:00:00 2001 From: Robin Getz Date: Tue, 2 Dec 2008 21:33:45 +0200 Subject: [PATCH 066/123] USB: musb: Make sure we program the correct values in only when necessary. Make sure we program the correct values in only when necessary. Signed-off-by: Robin Getz Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/blackfin.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/drivers/usb/musb/blackfin.c b/drivers/usb/musb/blackfin.c index 3fb719aecf0..eed35c933bc 100644 --- a/drivers/usb/musb/blackfin.c +++ b/drivers/usb/musb/blackfin.c @@ -245,13 +245,15 @@ int __init musb_platform_init(struct musb *musb) } gpio_direction_output(musb->config->gpio_vrsel, 0); - /* Anomaly #05000346 */ - bfin_write_USB_APHY_CALIB(0x5411); - SSYNC(); + if (ANOMALY_05000346) { + bfin_write_USB_APHY_CALIB(ANOMALY_05000346_value); + SSYNC(); + } - /* Anomaly #05000347 */ - bfin_write_USB_APHY_CNTRL(0x0); - SSYNC(); + if (ANOMALY_05000347) { + bfin_write_USB_APHY_CNTRL(0x0); + SSYNC(); + } /* TODO * Set SIC-IVG register From 2c557a4a98be67ea54dfd0e8497050a24b7311c6 Mon Sep 17 00:00:00 2001 From: Bryan Wu Date: Tue, 2 Dec 2008 21:33:46 +0200 Subject: [PATCH 067/123] USB: musb: add Blackfin version low level register accessing helper functions add Blackfin version low level register accessing helper functions Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/musb_io.h | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/drivers/usb/musb/musb_io.h b/drivers/usb/musb/musb_io.h index 223f0a51409..15f057722b6 100644 --- a/drivers/usb/musb/musb_io.h +++ b/drivers/usb/musb/musb_io.h @@ -56,6 +56,8 @@ static inline void writesb(const void __iomem *addr, const void *buf, int len) #endif +#ifndef CONFIG_BLACKFIN + /* NOTE: these offsets are all in bytes */ static inline u16 musb_readw(const void __iomem *addr, unsigned offset) @@ -114,4 +116,26 @@ static inline void musb_writeb(void __iomem *addr, unsigned offset, u8 data) #endif /* CONFIG_USB_TUSB6010 */ +#else + +static inline u8 musb_readb(const void __iomem *addr, unsigned offset) + { return (u8) (bfin_read16(addr + offset)); } + +static inline u16 musb_readw(const void __iomem *addr, unsigned offset) + { return bfin_read16(addr + offset); } + +static inline u32 musb_readl(const void __iomem *addr, unsigned offset) + { return (u32) (bfin_read16(addr + offset)); } + +static inline void musb_writeb(void __iomem *addr, unsigned offset, u8 data) + { bfin_write16(addr + offset, (u16) data); } + +static inline void musb_writew(void __iomem *addr, unsigned offset, u16 data) + { bfin_write16(addr + offset, data); } + +static inline void musb_writel(void __iomem *addr, unsigned offset, u32 data) + { bfin_write16(addr + offset, (u16) data); } + +#endif /* CONFIG_BLACKFIN */ + #endif From 6995eb68aab70e79eedb710d7d6d1f22d8aea4a7 Mon Sep 17 00:00:00 2001 From: Bryan Wu Date: Tue, 2 Dec 2008 21:33:47 +0200 Subject: [PATCH 068/123] USB: musb: enable low level DMA operation for Blackfin - DMA registers in Blackfin have different layout - DMA interrupt flags need to be cleared by software Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/musbhsdma.c | 82 +++---------------- drivers/usb/musb/musbhsdma.h | 149 +++++++++++++++++++++++++++++++++++ 2 files changed, 162 insertions(+), 69 deletions(-) create mode 100644 drivers/usb/musb/musbhsdma.h diff --git a/drivers/usb/musb/musbhsdma.c b/drivers/usb/musb/musbhsdma.c index d193b784908..8662e9e159c 100644 --- a/drivers/usb/musb/musbhsdma.c +++ b/drivers/usb/musb/musbhsdma.c @@ -34,58 +34,7 @@ #include #include #include "musb_core.h" - -#if defined(CONFIG_ARCH_OMAP2430) || defined(CONFIG_ARCH_OMAP3430) -#include "omap2430.h" -#endif - -#define MUSB_HSDMA_BASE 0x200 -#define MUSB_HSDMA_INTR (MUSB_HSDMA_BASE + 0) -#define MUSB_HSDMA_CONTROL 0x4 -#define MUSB_HSDMA_ADDRESS 0x8 -#define MUSB_HSDMA_COUNT 0xc - -#define MUSB_HSDMA_CHANNEL_OFFSET(_bchannel, _offset) \ - (MUSB_HSDMA_BASE + (_bchannel << 4) + _offset) - -/* control register (16-bit): */ -#define MUSB_HSDMA_ENABLE_SHIFT 0 -#define MUSB_HSDMA_TRANSMIT_SHIFT 1 -#define MUSB_HSDMA_MODE1_SHIFT 2 -#define MUSB_HSDMA_IRQENABLE_SHIFT 3 -#define MUSB_HSDMA_ENDPOINT_SHIFT 4 -#define MUSB_HSDMA_BUSERROR_SHIFT 8 -#define MUSB_HSDMA_BURSTMODE_SHIFT 9 -#define MUSB_HSDMA_BURSTMODE (3 << MUSB_HSDMA_BURSTMODE_SHIFT) -#define MUSB_HSDMA_BURSTMODE_UNSPEC 0 -#define MUSB_HSDMA_BURSTMODE_INCR4 1 -#define MUSB_HSDMA_BURSTMODE_INCR8 2 -#define MUSB_HSDMA_BURSTMODE_INCR16 3 - -#define MUSB_HSDMA_CHANNELS 8 - -struct musb_dma_controller; - -struct musb_dma_channel { - struct dma_channel channel; - struct musb_dma_controller *controller; - u32 start_addr; - u32 len; - u16 max_packet_sz; - u8 idx; - u8 epnum; - u8 transmit; -}; - -struct musb_dma_controller { - struct dma_controller controller; - struct musb_dma_channel channel[MUSB_HSDMA_CHANNELS]; - void *private_data; - void __iomem *base; - u8 channel_count; - u8 used_channels; - u8 irq; -}; +#include "musbhsdma.h" static int dma_controller_start(struct dma_controller *c) { @@ -203,12 +152,8 @@ static void configure_channel(struct dma_channel *channel, : 0); /* address/count */ - musb_writel(mbase, - MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_ADDRESS), - dma_addr); - musb_writel(mbase, - MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_COUNT), - len); + musb_write_hsdma_addr(mbase, bchannel, dma_addr); + musb_write_hsdma_count(mbase, bchannel, len); /* control (this should start things) */ musb_writew(mbase, @@ -279,13 +224,8 @@ static int dma_channel_abort(struct dma_channel *channel) musb_writew(mbase, MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_CONTROL), 0); - musb_writel(mbase, - MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_ADDRESS), - 0); - musb_writel(mbase, - MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_COUNT), - 0); - + musb_write_hsdma_addr(mbase, bchannel, 0); + musb_write_hsdma_count(mbase, bchannel, 0); channel->status = MUSB_DMA_STATUS_FREE; } @@ -333,10 +273,8 @@ static irqreturn_t dma_controller_irq(int irq, void *private_data) } else { u8 devctl; - addr = musb_readl(mbase, - MUSB_HSDMA_CHANNEL_OFFSET( - bchannel, - MUSB_HSDMA_ADDRESS)); + addr = musb_read_hsdma_addr(mbase, + bchannel); channel->actual_len = addr - musb_channel->start_addr; @@ -375,6 +313,12 @@ static irqreturn_t dma_controller_irq(int irq, void *private_data) } } } + +#ifdef CONFIG_BLACKFIN + /* Clear DMA interrup flags */ + musb_writeb(mbase, MUSB_HSDMA_INTR, int_hsdma); +#endif + retval = IRQ_HANDLED; done: spin_unlock_irqrestore(&musb->lock, flags); diff --git a/drivers/usb/musb/musbhsdma.h b/drivers/usb/musb/musbhsdma.h new file mode 100644 index 00000000000..1299d92dc83 --- /dev/null +++ b/drivers/usb/musb/musbhsdma.h @@ -0,0 +1,149 @@ +/* + * MUSB OTG driver - support for Mentor's DMA controller + * + * Copyright 2005 Mentor Graphics Corporation + * Copyright (C) 2005-2007 by Texas Instruments + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * version 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA + * 02110-1301 USA + * + * THIS SOFTWARE IS PROVIDED "AS IS" AND ANY EXPRESS OR IMPLIED + * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN + * NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF + * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON + * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + */ + +#if defined(CONFIG_ARCH_OMAP2430) || defined(CONFIG_ARCH_OMAP3430) +#include "omap2430.h" +#endif + +#ifndef CONFIG_BLACKFIN + +#define MUSB_HSDMA_BASE 0x200 +#define MUSB_HSDMA_INTR (MUSB_HSDMA_BASE + 0) +#define MUSB_HSDMA_CONTROL 0x4 +#define MUSB_HSDMA_ADDRESS 0x8 +#define MUSB_HSDMA_COUNT 0xc + +#define MUSB_HSDMA_CHANNEL_OFFSET(_bchannel, _offset) \ + (MUSB_HSDMA_BASE + (_bchannel << 4) + _offset) + +#define musb_read_hsdma_addr(mbase, bchannel) \ + musb_readl(mbase, \ + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_ADDRESS)) + +#define musb_write_hsdma_addr(mbase, bchannel, addr) \ + musb_writel(mbase, \ + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_ADDRESS), \ + addr) + +#define musb_write_hsdma_count(mbase, bchannel, len) \ + musb_writel(mbase, \ + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_COUNT), \ + len) +#else + +#define MUSB_HSDMA_BASE 0x400 +#define MUSB_HSDMA_INTR (MUSB_HSDMA_BASE + 0) +#define MUSB_HSDMA_CONTROL 0x04 +#define MUSB_HSDMA_ADDR_LOW 0x08 +#define MUSB_HSDMA_ADDR_HIGH 0x0C +#define MUSB_HSDMA_COUNT_LOW 0x10 +#define MUSB_HSDMA_COUNT_HIGH 0x14 + +#define MUSB_HSDMA_CHANNEL_OFFSET(_bchannel, _offset) \ + (MUSB_HSDMA_BASE + (_bchannel * 0x20) + _offset) + +static inline u32 musb_read_hsdma_addr(void __iomem *mbase, u8 bchannel) +{ + u32 addr = musb_readw(mbase, + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_ADDR_HIGH)); + + addr = addr << 16; + + addr |= musb_readw(mbase, + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_ADDR_LOW)); + + return addr; +} + +static inline void musb_write_hsdma_addr(void __iomem *mbase, + u8 bchannel, dma_addr_t dma_addr) +{ + musb_writew(mbase, + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_ADDR_LOW), + ((u16)((u32) dma_addr & 0xFFFF))); + musb_writew(mbase, + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_ADDR_HIGH), + ((u16)(((u32) dma_addr >> 16) & 0xFFFF))); +} + +static inline void musb_write_hsdma_count(void __iomem *mbase, + u8 bchannel, u32 len) +{ + musb_writew(mbase, + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_COUNT_LOW), + ((u16)((u32) len & 0xFFFF))); + musb_writew(mbase, + MUSB_HSDMA_CHANNEL_OFFSET(bchannel, MUSB_HSDMA_COUNT_HIGH), + ((u16)(((u32) len >> 16) & 0xFFFF))); +} + +#endif /* CONFIG_BLACKFIN */ + +/* control register (16-bit): */ +#define MUSB_HSDMA_ENABLE_SHIFT 0 +#define MUSB_HSDMA_TRANSMIT_SHIFT 1 +#define MUSB_HSDMA_MODE1_SHIFT 2 +#define MUSB_HSDMA_IRQENABLE_SHIFT 3 +#define MUSB_HSDMA_ENDPOINT_SHIFT 4 +#define MUSB_HSDMA_BUSERROR_SHIFT 8 +#define MUSB_HSDMA_BURSTMODE_SHIFT 9 +#define MUSB_HSDMA_BURSTMODE (3 << MUSB_HSDMA_BURSTMODE_SHIFT) +#define MUSB_HSDMA_BURSTMODE_UNSPEC 0 +#define MUSB_HSDMA_BURSTMODE_INCR4 1 +#define MUSB_HSDMA_BURSTMODE_INCR8 2 +#define MUSB_HSDMA_BURSTMODE_INCR16 3 + +#define MUSB_HSDMA_CHANNELS 8 + +struct musb_dma_controller; + +struct musb_dma_channel { + struct dma_channel channel; + struct musb_dma_controller *controller; + u32 start_addr; + u32 len; + u16 max_packet_sz; + u8 idx; + u8 epnum; + u8 transmit; +}; + +struct musb_dma_controller { + struct dma_controller controller; + struct musb_dma_channel channel[MUSB_HSDMA_CHANNELS]; + void *private_data; + void __iomem *base; + u8 channel_count; + u8 used_channels; + u8 irq; +}; From c6cf8b003e5a37f8193c2883876c5942adcd7284 Mon Sep 17 00:00:00 2001 From: Bryan Wu Date: Tue, 2 Dec 2008 21:33:48 +0200 Subject: [PATCH 069/123] USB: musb: add Blackfin specific configuration to MUSB Some config registers are not avaiable in Blackfin, we have to comment them out. v1-v2: - remove Blackfin specific header file - add Blackfin register version to musb_regs.h header file Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/musb_core.c | 52 ++--- drivers/usb/musb/musb_core.h | 70 +++++- drivers/usb/musb/musb_host.c | 22 +- drivers/usb/musb/musb_regs.h | 397 +++++++++++++++++++++++++++-------- 4 files changed, 397 insertions(+), 144 deletions(-) diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c index 83720f61fdb..6c7faacfb53 100644 --- a/drivers/usb/musb/musb_core.c +++ b/drivers/usb/musb/musb_core.c @@ -148,7 +148,8 @@ static inline struct musb *dev_to_musb(struct device *dev) /*-------------------------------------------------------------------------*/ -#ifndef CONFIG_USB_TUSB6010 +#if !defined(CONFIG_USB_TUSB6010) && !defined(CONFIG_BLACKFIN) + /* * Load an endpoint's FIFO */ @@ -1124,25 +1125,25 @@ fifo_setup(struct musb *musb, struct musb_hw_ep *hw_ep, #endif switch (cfg->style) { case FIFO_TX: - musb_writeb(mbase, MUSB_TXFIFOSZ, c_size); - musb_writew(mbase, MUSB_TXFIFOADD, c_off); + musb_write_txfifosz(mbase, c_size); + musb_write_txfifoadd(mbase, c_off); hw_ep->tx_double_buffered = !!(c_size & MUSB_FIFOSZ_DPB); hw_ep->max_packet_sz_tx = maxpacket; break; case FIFO_RX: - musb_writeb(mbase, MUSB_RXFIFOSZ, c_size); - musb_writew(mbase, MUSB_RXFIFOADD, c_off); + musb_write_rxfifosz(mbase, c_size); + musb_write_rxfifoadd(mbase, c_off); hw_ep->rx_double_buffered = !!(c_size & MUSB_FIFOSZ_DPB); hw_ep->max_packet_sz_rx = maxpacket; break; case FIFO_RXTX: - musb_writeb(mbase, MUSB_TXFIFOSZ, c_size); - musb_writew(mbase, MUSB_TXFIFOADD, c_off); + musb_write_txfifosz(mbase, c_size); + musb_write_txfifoadd(mbase, c_off); hw_ep->rx_double_buffered = !!(c_size & MUSB_FIFOSZ_DPB); hw_ep->max_packet_sz_rx = maxpacket; - musb_writeb(mbase, MUSB_RXFIFOSZ, c_size); - musb_writew(mbase, MUSB_RXFIFOADD, c_off); + musb_write_rxfifosz(mbase, c_size); + musb_write_rxfifoadd(mbase, c_off); hw_ep->tx_double_buffered = hw_ep->rx_double_buffered; hw_ep->max_packet_sz_tx = maxpacket; @@ -1246,9 +1247,10 @@ static int __init ep_config_from_table(struct musb *musb) */ static int __init ep_config_from_hw(struct musb *musb) { - u8 epnum = 0, reg; + u8 epnum = 0; struct musb_hw_ep *hw_ep; void *mbase = musb->mregs; + int ret = 0; DBG(2, "<== static silicon ep config\n"); @@ -1258,26 +1260,9 @@ static int __init ep_config_from_hw(struct musb *musb) musb_ep_select(mbase, epnum); hw_ep = musb->endpoints + epnum; - /* read from core using indexed model */ - reg = musb_readb(hw_ep->regs, 0x10 + MUSB_FIFOSIZE); - if (!reg) { - /* 0's returned when no more endpoints */ + ret = musb_read_fifosize(musb, hw_ep, epnum); + if (ret < 0) break; - } - musb->nr_endpoints++; - musb->epmask |= (1 << epnum); - - hw_ep->max_packet_sz_tx = 1 << (reg & 0x0f); - - /* shared TX/RX FIFO? */ - if ((reg & 0xf0) == 0xf0) { - hw_ep->max_packet_sz_rx = hw_ep->max_packet_sz_tx; - hw_ep->is_shared_fifo = true; - continue; - } else { - hw_ep->max_packet_sz_rx = 1 << ((reg & 0xf0) >> 4); - hw_ep->is_shared_fifo = false; - } /* FIXME set up hw_ep->{rx,tx}_double_buffered */ @@ -1326,7 +1311,7 @@ static int __init musb_core_init(u16 musb_type, struct musb *musb) /* log core options (read using indexed model) */ musb_ep_select(mbase, 0); - reg = musb_readb(mbase, 0x10 + MUSB_CONFIGDATA); + reg = musb_read_configdata(mbase); strcpy(aInfo, (reg & MUSB_CONFIGDATA_UTMIDW) ? "UTMI-16" : "UTMI-8"); if (reg & MUSB_CONFIGDATA_DYNFIFO) @@ -1391,7 +1376,7 @@ static int __init musb_core_init(u16 musb_type, struct musb *musb) } /* log release info */ - hwvers = musb_readw(mbase, MUSB_HWVERS); + hwvers = musb_read_hwvers(mbase); rev_major = (hwvers >> 10) & 0x1f; rev_minor = hwvers & 0x3ff; snprintf(aRevision, 32, "%d.%d%s", rev_major, @@ -1400,8 +1385,7 @@ static int __init musb_core_init(u16 musb_type, struct musb *musb) musb_driver_name, type, aRevision, aDate); /* configure ep0 */ - musb->endpoints[0].max_packet_sz_tx = MUSB_EP0_FIFOSIZE; - musb->endpoints[0].max_packet_sz_rx = MUSB_EP0_FIFOSIZE; + musb_configure_ep0(musb); /* discover endpoint configuration */ musb->nr_endpoints = 1; @@ -1445,7 +1429,7 @@ static int __init musb_core_init(u16 musb_type, struct musb *musb) hw_ep->regs = MUSB_EP_OFFSET(i, 0) + mbase; #ifdef CONFIG_USB_MUSB_HDRC_HCD - hw_ep->target_regs = MUSB_BUSCTL_OFFSET(i, 0) + mbase; + hw_ep->target_regs = musb_read_target_reg_base(i, mbase); hw_ep->rx_reinit = 1; hw_ep->tx_reinit = 1; #endif diff --git a/drivers/usb/musb/musb_core.h b/drivers/usb/musb/musb_core.h index d45d258e63f..630946a2d9f 100644 --- a/drivers/usb/musb/musb_core.h +++ b/drivers/usb/musb/musb_core.h @@ -191,7 +191,7 @@ enum musb_g_ep0_state { */ #if defined(CONFIG_ARCH_DAVINCI) || defined(CONFIG_ARCH_OMAP2430) \ - || defined(CONFIG_ARCH_OMAP3430) + || defined(CONFIG_ARCH_OMAP3430) || defined(CONFIG_BLACKFIN) /* REVISIT indexed access seemed to * misbehave (on DaVinci) for at least peripheral IN ... */ @@ -448,6 +448,70 @@ static inline struct musb *gadget_to_musb(struct usb_gadget *g) } #endif +#ifdef CONFIG_BLACKFIN +static inline int musb_read_fifosize(struct musb *musb, + struct musb_hw_ep *hw_ep, u8 epnum) +{ + musb->nr_endpoints++; + musb->epmask |= (1 << epnum); + + if (epnum < 5) { + hw_ep->max_packet_sz_tx = 128; + hw_ep->max_packet_sz_rx = 128; + } else { + hw_ep->max_packet_sz_tx = 1024; + hw_ep->max_packet_sz_rx = 1024; + } + hw_ep->is_shared_fifo = false; + + return 0; +} + +static inline void musb_configure_ep0(struct musb *musb) +{ + musb->endpoints[0].max_packet_sz_tx = MUSB_EP0_FIFOSIZE; + musb->endpoints[0].max_packet_sz_rx = MUSB_EP0_FIFOSIZE; + musb->endpoints[0].is_shared_fifo = true; +} + +#else + +static inline int musb_read_fifosize(struct musb *musb, + struct musb_hw_ep *hw_ep, u8 epnum) +{ + u8 reg = 0; + + /* read from core using indexed model */ + reg = musb_readb(hw_ep->regs, 0x10 + MUSB_FIFOSIZE); + /* 0's returned when no more endpoints */ + if (!reg) + return -ENODEV; + + musb->nr_endpoints++; + musb->epmask |= (1 << epnum); + + hw_ep->max_packet_sz_tx = 1 << (reg & 0x0f); + + /* shared TX/RX FIFO? */ + if ((reg & 0xf0) == 0xf0) { + hw_ep->max_packet_sz_rx = hw_ep->max_packet_sz_tx; + hw_ep->is_shared_fifo = true; + return 0; + } else { + hw_ep->max_packet_sz_rx = 1 << ((reg & 0xf0) >> 4); + hw_ep->is_shared_fifo = false; + } + + return 0; +} + +static inline void musb_configure_ep0(struct musb *musb) +{ + musb->endpoints[0].max_packet_sz_tx = MUSB_EP0_FIFOSIZE; + musb->endpoints[0].max_packet_sz_rx = MUSB_EP0_FIFOSIZE; +} +#endif /* CONFIG_BLACKFIN */ + /***************************** Glue it together *****************************/ @@ -470,14 +534,14 @@ extern void musb_hnp_stop(struct musb *musb); extern int musb_platform_set_mode(struct musb *musb, u8 musb_mode); -#if defined(CONFIG_USB_TUSB6010) || \ +#if defined(CONFIG_USB_TUSB6010) || defined(CONFIG_BLACKFIN) || \ defined(CONFIG_ARCH_OMAP2430) || defined(CONFIG_ARCH_OMAP34XX) extern void musb_platform_try_idle(struct musb *musb, unsigned long timeout); #else #define musb_platform_try_idle(x, y) do {} while (0) #endif -#ifdef CONFIG_USB_TUSB6010 +#if defined(CONFIG_USB_TUSB6010) || defined(CONFIG_BLACKFIN) extern int musb_platform_get_vbus_status(struct musb *musb); #else #define musb_platform_get_vbus_status(x) 0 diff --git a/drivers/usb/musb/musb_host.c b/drivers/usb/musb/musb_host.c index bf25ef4cc15..99fa6123487 100644 --- a/drivers/usb/musb/musb_host.c +++ b/drivers/usb/musb/musb_host.c @@ -596,12 +596,10 @@ musb_rx_reinit(struct musb *musb, struct musb_qh *qh, struct musb_hw_ep *ep) /* target addr and (for multipoint) hub addr/port */ if (musb->is_multipoint) { - musb_writeb(ep->target_regs, MUSB_RXFUNCADDR, - qh->addr_reg); - musb_writeb(ep->target_regs, MUSB_RXHUBADDR, - qh->h_addr_reg); - musb_writeb(ep->target_regs, MUSB_RXHUBPORT, - qh->h_port_reg); + musb_write_rxfunaddr(ep->target_regs, qh->addr_reg); + musb_write_rxhubaddr(ep->target_regs, qh->h_addr_reg); + musb_write_rxhubport(ep->target_regs, qh->h_port_reg); + } else musb_writeb(musb->mregs, MUSB_FADDR, qh->addr_reg); @@ -715,15 +713,9 @@ static void musb_ep_program(struct musb *musb, u8 epnum, /* target addr and (for multipoint) hub addr/port */ if (musb->is_multipoint) { - musb_writeb(mbase, - MUSB_BUSCTL_OFFSET(epnum, MUSB_TXFUNCADDR), - qh->addr_reg); - musb_writeb(mbase, - MUSB_BUSCTL_OFFSET(epnum, MUSB_TXHUBADDR), - qh->h_addr_reg); - musb_writeb(mbase, - MUSB_BUSCTL_OFFSET(epnum, MUSB_TXHUBPORT), - qh->h_port_reg); + musb_write_txfunaddr(mbase, epnum, qh->addr_reg); + musb_write_txhubaddr(mbase, epnum, qh->h_addr_reg); + musb_write_txhubport(mbase, epnum, qh->h_port_reg); /* FIXME if !epnum, do the same for RX ... */ } else musb_writeb(mbase, MUSB_FADDR, qh->addr_reg); diff --git a/drivers/usb/musb/musb_regs.h b/drivers/usb/musb/musb_regs.h index 9c228661aa5..de3b2f18db4 100644 --- a/drivers/usb/musb/musb_regs.h +++ b/drivers/usb/musb/musb_regs.h @@ -37,97 +37,6 @@ #define MUSB_EP0_FIFOSIZE 64 /* This is non-configurable */ -/* - * Common USB registers - */ - -#define MUSB_FADDR 0x00 /* 8-bit */ -#define MUSB_POWER 0x01 /* 8-bit */ - -#define MUSB_INTRTX 0x02 /* 16-bit */ -#define MUSB_INTRRX 0x04 -#define MUSB_INTRTXE 0x06 -#define MUSB_INTRRXE 0x08 -#define MUSB_INTRUSB 0x0A /* 8 bit */ -#define MUSB_INTRUSBE 0x0B /* 8 bit */ -#define MUSB_FRAME 0x0C -#define MUSB_INDEX 0x0E /* 8 bit */ -#define MUSB_TESTMODE 0x0F /* 8 bit */ - -/* Get offset for a given FIFO from musb->mregs */ -#ifdef CONFIG_USB_TUSB6010 -#define MUSB_FIFO_OFFSET(epnum) (0x200 + ((epnum) * 0x20)) -#else -#define MUSB_FIFO_OFFSET(epnum) (0x20 + ((epnum) * 4)) -#endif - -/* - * Additional Control Registers - */ - -#define MUSB_DEVCTL 0x60 /* 8 bit */ - -/* These are always controlled through the INDEX register */ -#define MUSB_TXFIFOSZ 0x62 /* 8-bit (see masks) */ -#define MUSB_RXFIFOSZ 0x63 /* 8-bit (see masks) */ -#define MUSB_TXFIFOADD 0x64 /* 16-bit offset shifted right 3 */ -#define MUSB_RXFIFOADD 0x66 /* 16-bit offset shifted right 3 */ - -/* REVISIT: vctrl/vstatus: optional vendor utmi+phy register at 0x68 */ -#define MUSB_HWVERS 0x6C /* 8 bit */ - -#define MUSB_EPINFO 0x78 /* 8 bit */ -#define MUSB_RAMINFO 0x79 /* 8 bit */ -#define MUSB_LINKINFO 0x7a /* 8 bit */ -#define MUSB_VPLEN 0x7b /* 8 bit */ -#define MUSB_HS_EOF1 0x7c /* 8 bit */ -#define MUSB_FS_EOF1 0x7d /* 8 bit */ -#define MUSB_LS_EOF1 0x7e /* 8 bit */ - -/* Offsets to endpoint registers */ -#define MUSB_TXMAXP 0x00 -#define MUSB_TXCSR 0x02 -#define MUSB_CSR0 MUSB_TXCSR /* Re-used for EP0 */ -#define MUSB_RXMAXP 0x04 -#define MUSB_RXCSR 0x06 -#define MUSB_RXCOUNT 0x08 -#define MUSB_COUNT0 MUSB_RXCOUNT /* Re-used for EP0 */ -#define MUSB_TXTYPE 0x0A -#define MUSB_TYPE0 MUSB_TXTYPE /* Re-used for EP0 */ -#define MUSB_TXINTERVAL 0x0B -#define MUSB_NAKLIMIT0 MUSB_TXINTERVAL /* Re-used for EP0 */ -#define MUSB_RXTYPE 0x0C -#define MUSB_RXINTERVAL 0x0D -#define MUSB_FIFOSIZE 0x0F -#define MUSB_CONFIGDATA MUSB_FIFOSIZE /* Re-used for EP0 */ - -/* Offsets to endpoint registers in indexed model (using INDEX register) */ -#define MUSB_INDEXED_OFFSET(_epnum, _offset) \ - (0x10 + (_offset)) - -/* Offsets to endpoint registers in flat models */ -#define MUSB_FLAT_OFFSET(_epnum, _offset) \ - (0x100 + (0x10*(_epnum)) + (_offset)) - -#ifdef CONFIG_USB_TUSB6010 -/* TUSB6010 EP0 configuration register is special */ -#define MUSB_TUSB_OFFSET(_epnum, _offset) \ - (0x10 + _offset) -#include "tusb6010.h" /* Needed "only" for TUSB_EP0_CONF */ -#endif - -/* "bus control"/target registers, for host side multipoint (external hubs) */ -#define MUSB_TXFUNCADDR 0x00 -#define MUSB_TXHUBADDR 0x02 -#define MUSB_TXHUBPORT 0x03 - -#define MUSB_RXFUNCADDR 0x04 -#define MUSB_RXHUBADDR 0x06 -#define MUSB_RXHUBPORT 0x07 - -#define MUSB_BUSCTL_OFFSET(_epnum, _offset) \ - (0x80 + (8*(_epnum)) + (_offset)) - /* * MUSB Register bits */ @@ -228,7 +137,6 @@ /* TXCSR in Peripheral and Host mode */ #define MUSB_TXCSR_AUTOSET 0x8000 -#define MUSB_TXCSR_MODE 0x2000 #define MUSB_TXCSR_DMAENAB 0x1000 #define MUSB_TXCSR_FRCDATATOG 0x0800 #define MUSB_TXCSR_DMAMODE 0x0400 @@ -297,4 +205,309 @@ /* HUBADDR */ #define MUSB_HUBADDR_MULTI_TT 0x80 + +#ifndef CONFIG_BLACKFIN + +/* + * Common USB registers + */ + +#define MUSB_FADDR 0x00 /* 8-bit */ +#define MUSB_POWER 0x01 /* 8-bit */ + +#define MUSB_INTRTX 0x02 /* 16-bit */ +#define MUSB_INTRRX 0x04 +#define MUSB_INTRTXE 0x06 +#define MUSB_INTRRXE 0x08 +#define MUSB_INTRUSB 0x0A /* 8 bit */ +#define MUSB_INTRUSBE 0x0B /* 8 bit */ +#define MUSB_FRAME 0x0C +#define MUSB_INDEX 0x0E /* 8 bit */ +#define MUSB_TESTMODE 0x0F /* 8 bit */ + +/* Get offset for a given FIFO from musb->mregs */ +#ifdef CONFIG_USB_TUSB6010 +#define MUSB_FIFO_OFFSET(epnum) (0x200 + ((epnum) * 0x20)) +#else +#define MUSB_FIFO_OFFSET(epnum) (0x20 + ((epnum) * 4)) +#endif + +/* + * Additional Control Registers + */ + +#define MUSB_DEVCTL 0x60 /* 8 bit */ + +/* These are always controlled through the INDEX register */ +#define MUSB_TXFIFOSZ 0x62 /* 8-bit (see masks) */ +#define MUSB_RXFIFOSZ 0x63 /* 8-bit (see masks) */ +#define MUSB_TXFIFOADD 0x64 /* 16-bit offset shifted right 3 */ +#define MUSB_RXFIFOADD 0x66 /* 16-bit offset shifted right 3 */ + +/* REVISIT: vctrl/vstatus: optional vendor utmi+phy register at 0x68 */ +#define MUSB_HWVERS 0x6C /* 8 bit */ + +#define MUSB_EPINFO 0x78 /* 8 bit */ +#define MUSB_RAMINFO 0x79 /* 8 bit */ +#define MUSB_LINKINFO 0x7a /* 8 bit */ +#define MUSB_VPLEN 0x7b /* 8 bit */ +#define MUSB_HS_EOF1 0x7c /* 8 bit */ +#define MUSB_FS_EOF1 0x7d /* 8 bit */ +#define MUSB_LS_EOF1 0x7e /* 8 bit */ + +/* Offsets to endpoint registers */ +#define MUSB_TXMAXP 0x00 +#define MUSB_TXCSR 0x02 +#define MUSB_CSR0 MUSB_TXCSR /* Re-used for EP0 */ +#define MUSB_RXMAXP 0x04 +#define MUSB_RXCSR 0x06 +#define MUSB_RXCOUNT 0x08 +#define MUSB_COUNT0 MUSB_RXCOUNT /* Re-used for EP0 */ +#define MUSB_TXTYPE 0x0A +#define MUSB_TYPE0 MUSB_TXTYPE /* Re-used for EP0 */ +#define MUSB_TXINTERVAL 0x0B +#define MUSB_NAKLIMIT0 MUSB_TXINTERVAL /* Re-used for EP0 */ +#define MUSB_RXTYPE 0x0C +#define MUSB_RXINTERVAL 0x0D +#define MUSB_FIFOSIZE 0x0F +#define MUSB_CONFIGDATA MUSB_FIFOSIZE /* Re-used for EP0 */ + +/* Offsets to endpoint registers in indexed model (using INDEX register) */ +#define MUSB_INDEXED_OFFSET(_epnum, _offset) \ + (0x10 + (_offset)) + +/* Offsets to endpoint registers in flat models */ +#define MUSB_FLAT_OFFSET(_epnum, _offset) \ + (0x100 + (0x10*(_epnum)) + (_offset)) + +#ifdef CONFIG_USB_TUSB6010 +/* TUSB6010 EP0 configuration register is special */ +#define MUSB_TUSB_OFFSET(_epnum, _offset) \ + (0x10 + _offset) +#include "tusb6010.h" /* Needed "only" for TUSB_EP0_CONF */ +#endif + +#define MUSB_TXCSR_MODE 0x2000 + +/* "bus control"/target registers, for host side multipoint (external hubs) */ +#define MUSB_TXFUNCADDR 0x00 +#define MUSB_TXHUBADDR 0x02 +#define MUSB_TXHUBPORT 0x03 + +#define MUSB_RXFUNCADDR 0x04 +#define MUSB_RXHUBADDR 0x06 +#define MUSB_RXHUBPORT 0x07 + +#define MUSB_BUSCTL_OFFSET(_epnum, _offset) \ + (0x80 + (8*(_epnum)) + (_offset)) + +static inline void musb_write_txfifosz(void __iomem *mbase, u8 c_size) +{ + musb_writeb(mbase, MUSB_TXFIFOSZ, c_size); +} + +static inline void musb_write_txfifoadd(void __iomem *mbase, u16 c_off) +{ + musb_writew(mbase, MUSB_TXFIFOADD, c_off); +} + +static inline void musb_write_rxfifosz(void __iomem *mbase, u8 c_size) +{ + musb_writeb(mbase, MUSB_RXFIFOSZ, c_size); +} + +static inline void musb_write_rxfifoadd(void __iomem *mbase, u16 c_off) +{ + musb_writew(mbase, MUSB_RXFIFOADD, c_off); +} + +static inline u8 musb_read_configdata(void __iomem *mbase) +{ + return musb_readb(mbase, 0x10 + MUSB_CONFIGDATA); +} + +static inline u16 musb_read_hwvers(void __iomem *mbase) +{ + return musb_readw(mbase, MUSB_HWVERS); +} + +static inline void __iomem *musb_read_target_reg_base(u8 i, void __iomem *mbase) +{ + return (MUSB_BUSCTL_OFFSET(i, 0) + mbase); +} + +static inline void musb_write_rxfunaddr(void __iomem *ep_target_regs, + u8 qh_addr_reg) +{ + musb_writeb(ep_target_regs, MUSB_RXFUNCADDR, qh_addr_reg); +} + +static inline void musb_write_rxhubaddr(void __iomem *ep_target_regs, + u8 qh_h_addr_reg) +{ + musb_writeb(ep_target_regs, MUSB_RXHUBADDR, qh_h_addr_reg); +} + +static inline void musb_write_rxhubport(void __iomem *ep_target_regs, + u8 qh_h_port_reg) +{ + musb_writeb(ep_target_regs, MUSB_RXHUBPORT, qh_h_port_reg); +} + +static inline void musb_write_txfunaddr(void __iomem *mbase, u8 epnum, + u8 qh_addr_reg) +{ + musb_writeb(mbase, MUSB_BUSCTL_OFFSET(epnum, MUSB_TXFUNCADDR), + qh_addr_reg); +} + +static inline void musb_write_txhubaddr(void __iomem *mbase, u8 epnum, + u8 qh_addr_reg) +{ + musb_writeb(mbase, MUSB_BUSCTL_OFFSET(epnum, MUSB_TXHUBADDR), + qh_addr_reg); +} + +static inline void musb_write_txhubport(void __iomem *mbase, u8 epnum, + u8 qh_h_port_reg) +{ + musb_writeb(mbase, MUSB_BUSCTL_OFFSET(epnum, MUSB_TXHUBPORT), + qh_h_port_reg); +} + +#else /* CONFIG_BLACKFIN */ + +#define USB_BASE USB_FADDR +#define USB_OFFSET(reg) (reg - USB_BASE) + +/* + * Common USB registers + */ +#define MUSB_FADDR USB_OFFSET(USB_FADDR) /* 8-bit */ +#define MUSB_POWER USB_OFFSET(USB_POWER) /* 8-bit */ +#define MUSB_INTRTX USB_OFFSET(USB_INTRTX) /* 16-bit */ +#define MUSB_INTRRX USB_OFFSET(USB_INTRRX) +#define MUSB_INTRTXE USB_OFFSET(USB_INTRTXE) +#define MUSB_INTRRXE USB_OFFSET(USB_INTRRXE) +#define MUSB_INTRUSB USB_OFFSET(USB_INTRUSB) /* 8 bit */ +#define MUSB_INTRUSBE USB_OFFSET(USB_INTRUSBE)/* 8 bit */ +#define MUSB_FRAME USB_OFFSET(USB_FRAME) +#define MUSB_INDEX USB_OFFSET(USB_INDEX) /* 8 bit */ +#define MUSB_TESTMODE USB_OFFSET(USB_TESTMODE)/* 8 bit */ + +/* Get offset for a given FIFO from musb->mregs */ +#define MUSB_FIFO_OFFSET(epnum) \ + (USB_OFFSET(USB_EP0_FIFO) + ((epnum) * 8)) + +/* + * Additional Control Registers + */ + +#define MUSB_DEVCTL USB_OFFSET(USB_OTG_DEV_CTL) /* 8 bit */ + +#define MUSB_LINKINFO USB_OFFSET(USB_LINKINFO)/* 8 bit */ +#define MUSB_VPLEN USB_OFFSET(USB_VPLEN) /* 8 bit */ +#define MUSB_HS_EOF1 USB_OFFSET(USB_HS_EOF1) /* 8 bit */ +#define MUSB_FS_EOF1 USB_OFFSET(USB_FS_EOF1) /* 8 bit */ +#define MUSB_LS_EOF1 USB_OFFSET(USB_LS_EOF1) /* 8 bit */ + +/* Offsets to endpoint registers */ +#define MUSB_TXMAXP 0x00 +#define MUSB_TXCSR 0x04 +#define MUSB_CSR0 MUSB_TXCSR /* Re-used for EP0 */ +#define MUSB_RXMAXP 0x08 +#define MUSB_RXCSR 0x0C +#define MUSB_RXCOUNT 0x10 +#define MUSB_COUNT0 MUSB_RXCOUNT /* Re-used for EP0 */ +#define MUSB_TXTYPE 0x14 +#define MUSB_TYPE0 MUSB_TXTYPE /* Re-used for EP0 */ +#define MUSB_TXINTERVAL 0x18 +#define MUSB_NAKLIMIT0 MUSB_TXINTERVAL /* Re-used for EP0 */ +#define MUSB_RXTYPE 0x1C +#define MUSB_RXINTERVAL 0x20 +#define MUSB_TXCOUNT 0x28 + +/* Offsets to endpoint registers in indexed model (using INDEX register) */ +#define MUSB_INDEXED_OFFSET(_epnum, _offset) \ + (0x40 + (_offset)) + +/* Offsets to endpoint registers in flat models */ +#define MUSB_FLAT_OFFSET(_epnum, _offset) \ + (USB_OFFSET(USB_EP_NI0_TXMAXP) + (0x40 * (_epnum)) + (_offset)) + +/* Not implemented - HW has seperate Tx/Rx FIFO */ +#define MUSB_TXCSR_MODE 0x0000 + +/* + * Dummy stub for clk framework, it will be removed + * until Blackfin supports clk framework + */ +#define clk_get(dev, id) NULL +#define clk_put(clock) do {} while (0) +#define clk_enable(clock) do {} while (0) +#define clk_disable(clock) do {} while (0) + +static inline void musb_write_txfifosz(void __iomem *mbase, u8 c_size) +{ +} + +static inline void musb_write_txfifoadd(void __iomem *mbase, u16 c_off) +{ +} + +static inline void musb_write_rxfifosz(void __iomem *mbase, u8 c_size) +{ +} + +static inline void musb_write_rxfifoadd(void __iomem *mbase, u16 c_off) +{ +} + +static inline u8 musb_read_configdata(void __iomem *mbase) +{ + return 0; +} + +static inline u16 musb_read_hwvers(void __iomem *mbase) +{ + return 0; +} + +static inline u16 musb_read_target_reg_base(u8 i, void __iomem *mbase) +{ + return 0; +} + +static inline void musb_write_rxfunaddr(void __iomem *ep_target_regs, + u8 qh_addr_req) +{ +} + +static inline void musb_write_rxhubaddr(void __iomem *ep_target_regs, + u8 qh_h_addr_reg) +{ +} + +static inline void musb_write_rxhubport(void __iomem *ep_target_regs, + u8 qh_h_port_reg) +{ +} + +static inline void musb_write_txfunaddr(void __iomem *mbase, u8 epnum, + u8 qh_addr_reg) +{ +} + +static inline void musb_write_txhubaddr(void __iomem *mbase, u8 epnum, + u8 qh_addr_reg) +{ +} + +static inline void musb_write_txhubport(void __iomem *mbase, u8 epnum, + u8 qh_h_port_reg) +{ +} + +#endif /* CONFIG_BLACKFIN */ + #endif /* __MUSB_REGS_H__ */ From 085ad4067b5def12bb0e6f50ec65302053d9186d Mon Sep 17 00:00:00 2001 From: Bryan Wu Date: Tue, 2 Dec 2008 21:33:49 +0200 Subject: [PATCH 070/123] USB: musb: add Blackfin Kconfig options and Makefile Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/Kconfig | 4 ++-- drivers/usb/musb/Kconfig | 12 ++++++++++-- drivers/usb/musb/Makefile | 8 ++++++++ 3 files changed, 20 insertions(+), 4 deletions(-) diff --git a/drivers/usb/gadget/Kconfig b/drivers/usb/gadget/Kconfig index cf3a662c333..c322024bedf 100644 --- a/drivers/usb/gadget/Kconfig +++ b/drivers/usb/gadget/Kconfig @@ -297,13 +297,13 @@ config USB_S3C2410_DEBUG # musb builds in ../musb along with host support config USB_GADGET_MUSB_HDRC - boolean "Inventra HDRC USB Peripheral (TI, ...)" + boolean "Inventra HDRC USB Peripheral (TI, ADI, ...)" depends on USB_MUSB_HDRC && (USB_MUSB_PERIPHERAL || USB_MUSB_OTG) select USB_GADGET_DUALSPEED select USB_GADGET_SELECTED help This OTG-capable silicon IP is used in dual designs including - the TI DaVinci, OMAP 243x, OMAP 343x, and TUSB 6010. + the TI DaVinci, OMAP 243x, OMAP 343x, TUSB 6010, and ADI Blackfin config USB_GADGET_M66592 boolean "Renesas M66592 USB Peripheral Controller" diff --git a/drivers/usb/musb/Kconfig b/drivers/usb/musb/Kconfig index 4b9542bbb35..5af7379cd9a 100644 --- a/drivers/usb/musb/Kconfig +++ b/drivers/usb/musb/Kconfig @@ -11,7 +11,7 @@ config USB_MUSB_HDRC depends on (USB || USB_GADGET) && HAVE_CLK depends on !SUPERH select TWL4030_USB if MACH_OMAP_3430SDP - tristate 'Inventra Highspeed Dual Role Controller (TI, ...)' + tristate 'Inventra Highspeed Dual Role Controller (TI, ADI, ...)' help Say Y here if your system has a dual role high speed USB controller based on the Mentor Graphics silicon IP. Then @@ -22,6 +22,9 @@ config USB_MUSB_HDRC Texas Instruments parts using this IP include DaVinci 644x, OMAP 243x, OMAP 343x, and TUSB 6010. + Analog Devices parts using this IP include Blackfin BF54x, + BF525 and BF527. + If you do not know what this is, please say N. To compile this driver as a module, choose M here; the @@ -33,6 +36,8 @@ config USB_MUSB_SOC default y if ARCH_DAVINCI default y if ARCH_OMAP2430 default y if ARCH_OMAP34XX + default y if (BF54x && !BF544) + default y if (BF52x && !BF522 && !BF523) comment "DaVinci 644x USB support" depends on USB_MUSB_HDRC && ARCH_DAVINCI @@ -43,6 +48,9 @@ comment "OMAP 243x high speed USB support" comment "OMAP 343x high speed USB support" depends on USB_MUSB_HDRC && ARCH_OMAP34XX +comment "Blackfin high speed USB Support" + depends on USB_MUSB_HDRC && (BF54x && !BF544) || (BF52x && !BF522 && !BF523) + config USB_TUSB6010 boolean "TUSB 6010 support" depends on USB_MUSB_HDRC && !USB_MUSB_SOC @@ -142,7 +150,7 @@ config MUSB_PIO_ONLY config USB_INVENTRA_DMA bool depends on USB_MUSB_HDRC && !MUSB_PIO_ONLY - default ARCH_OMAP2430 || ARCH_OMAP34XX + default ARCH_OMAP2430 || ARCH_OMAP34XX || BLACKFIN help Enable DMA transfers using Mentor's engine. diff --git a/drivers/usb/musb/Makefile b/drivers/usb/musb/Makefile index b6af0d687a7..85710ccc188 100644 --- a/drivers/usb/musb/Makefile +++ b/drivers/usb/musb/Makefile @@ -22,6 +22,14 @@ ifeq ($(CONFIG_ARCH_OMAP3430),y) musb_hdrc-objs += omap2430.o endif +ifeq ($(CONFIG_BF54x),y) + musb_hdrc-objs += blackfin.o +endif + +ifeq ($(CONFIG_BF52x),y) + musb_hdrc-objs += blackfin.o +endif + ifeq ($(CONFIG_USB_GADGET_MUSB_HDRC),y) musb_hdrc-objs += musb_gadget_ep0.o musb_gadget.o endif From 7833414d4f39009d4d04e96648a099630adf3f62 Mon Sep 17 00:00:00 2001 From: Bryan Wu Date: Tue, 2 Dec 2008 21:33:50 +0200 Subject: [PATCH 071/123] USB: musb: Blackfin provides read/write I/O accessor in header files Don't redefine the functions in musb driver Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/musb_io.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/musb/musb_io.h b/drivers/usb/musb/musb_io.h index 15f057722b6..b06e9ef00cf 100644 --- a/drivers/usb/musb/musb_io.h +++ b/drivers/usb/musb/musb_io.h @@ -39,7 +39,7 @@ #if !defined(CONFIG_ARM) && !defined(CONFIG_SUPERH) \ && !defined(CONFIG_AVR32) && !defined(CONFIG_PPC32) \ - && !defined(CONFIG_PPC64) + && !defined(CONFIG_PPC64) && !defined(CONFIG_BLACKFIN) static inline void readsl(const void __iomem *addr, void *buf, int len) { insl((unsigned long)addr, buf, len); } static inline void readsw(const void __iomem *addr, void *buf, int len) From e22b582e9108b94a8bb59010da3e09709bd39746 Mon Sep 17 00:00:00 2001 From: Bryan Wu Date: Tue, 2 Dec 2008 21:33:51 +0200 Subject: [PATCH 072/123] USB: musb: Kill some compiling warning in musb Blackfin part Signed-off-by: Bryan Wu Signed-off-by: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/musb/blackfin.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/drivers/usb/musb/blackfin.c b/drivers/usb/musb/blackfin.c index eed35c933bc..78613485209 100644 --- a/drivers/usb/musb/blackfin.c +++ b/drivers/usb/musb/blackfin.c @@ -41,9 +41,11 @@ void musb_write_fifo(struct musb_hw_ep *hw_ep, u16 len, const u8 *src) dump_fifo_data(src, len); if (unlikely((unsigned long)src & 0x01)) - outsw_8(fifo, src, len & 0x01 ? (len >> 1) + 1 : len >> 1); + outsw_8((unsigned long)fifo, src, + len & 0x01 ? (len >> 1) + 1 : len >> 1); else - outsw(fifo, src, len & 0x01 ? (len >> 1) + 1 : len >> 1); + outsw((unsigned long)fifo, src, + len & 0x01 ? (len >> 1) + 1 : len >> 1); } /* @@ -54,8 +56,6 @@ void musb_read_fifo(struct musb_hw_ep *hw_ep, u16 len, u8 *dst) void __iomem *fifo = hw_ep->fifo; u8 epnum = hw_ep->epnum; u16 dma_reg = 0; - int i; - u16 *data; DBG(4, "%cX ep%d fifo %p count %d buf %p\n", 'R', hw_ep->epnum, fifo, len, dst); @@ -96,9 +96,11 @@ void musb_read_fifo(struct musb_hw_ep *hw_ep, u16 len, u8 *dst) SSYNC(); #else if (unlikely((unsigned long)dst & 0x01)) - insw_8(fifo, dst, len & 0x01 ? (len >> 1) + 1 : len >> 1); + insw_8((unsigned long)fifo, dst, + len & 0x01 ? (len >> 1) + 1 : len >> 1); else - insw(fifo, dst, len & 0x01 ? (len >> 1) + 1 : len >> 1); + insw((unsigned long)fifo, dst, + len & 0x01 ? (len >> 1) + 1 : len >> 1); #endif dump_fifo_data(dst, len); From b9cef6c31913c34fb1065b1d01e04c3b92c59016 Mon Sep 17 00:00:00 2001 From: Wu Fengguang Date: Mon, 15 Dec 2008 15:32:01 +0800 Subject: [PATCH 073/123] USB: make printk messages more searchable USB: make printk messages more searchable Make USB printk messages long and straightforward. One of these decorated USB error messages cost me non-trivial efforts to locate. Signed-off-by: Wu Fengguang Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/hub.c | 45 ++++++++++++++++++++++-------------------- 1 file changed, 24 insertions(+), 21 deletions(-) diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c index fc99ef67761..5abdc11be1e 100644 --- a/drivers/usb/core/hub.c +++ b/drivers/usb/core/hub.c @@ -107,7 +107,9 @@ MODULE_PARM_DESC (blinkenlights, "true to cycle leds on hubs"); /* define initial 64-byte descriptor request timeout in milliseconds */ static int initial_descriptor_timeout = USB_CTRL_GET_TIMEOUT; module_param(initial_descriptor_timeout, int, S_IRUGO|S_IWUSR); -MODULE_PARM_DESC(initial_descriptor_timeout, "initial 64-byte descriptor request timeout in milliseconds (default 5000 - 5.0 seconds)"); +MODULE_PARM_DESC(initial_descriptor_timeout, + "initial 64-byte descriptor request timeout in milliseconds " + "(default 5000 - 5.0 seconds)"); /* * As of 2.6.10 we introduce a new USB device initialization scheme which @@ -1136,8 +1138,8 @@ static int hub_probe(struct usb_interface *intf, const struct usb_device_id *id) hdev = interface_to_usbdev(intf); if (hdev->level == MAX_TOPO_LEVEL) { - dev_err(&intf->dev, "Unsupported bus topology: " - "hub nested too deep\n"); + dev_err(&intf->dev, + "Unsupported bus topology: hub nested too deep\n"); return -E2BIG; } @@ -1477,8 +1479,8 @@ static void announce_device(struct usb_device *udev) dev_info(&udev->dev, "New USB device found, idVendor=%04x, idProduct=%04x\n", le16_to_cpu(udev->descriptor.idVendor), le16_to_cpu(udev->descriptor.idProduct)); - dev_info(&udev->dev, "New USB device strings: Mfr=%d, Product=%d, " - "SerialNumber=%d\n", + dev_info(&udev->dev, + "New USB device strings: Mfr=%d, Product=%d, SerialNumber=%d\n", udev->descriptor.iManufacturer, udev->descriptor.iProduct, udev->descriptor.iSerialNumber); @@ -1543,7 +1545,7 @@ static int usb_configure_device_otg(struct usb_device *udev) * customize to match your product. */ dev_info(&udev->dev, - "can't set HNP mode; %d\n", + "can't set HNP mode: %d\n", err); bus->b_hnp_enable = 0; } @@ -2047,8 +2049,8 @@ static int finish_port_resume(struct usb_device *udev) u16 devstatus; /* caller owns the udev device lock */ - dev_dbg(&udev->dev, "finish %sresume\n", - udev->reset_resume ? "reset-" : ""); + dev_dbg(&udev->dev, "%s\n", + udev->reset_resume ? "finish reset-resume" : "finish resume"); /* usb ch9 identifies four variants of SUSPENDED, based on what * state the device resumes to. Linux currently won't see the @@ -2100,8 +2102,9 @@ static int finish_port_resume(struct usb_device *udev) NULL, 0, USB_CTRL_SET_TIMEOUT); if (status) - dev_dbg(&udev->dev, "disable remote " - "wakeup, status %d\n", status); + dev_dbg(&udev->dev, + "disable remote wakeup, status %d\n", + status); } status = 0; } @@ -2584,9 +2587,9 @@ hub_port_init (struct usb_hub *hub, struct usb_device *udev, int port1, goto fail; } if (r) { - dev_err(&udev->dev, "device descriptor " - "read/%s, error %d\n", - "64", r); + dev_err(&udev->dev, + "device descriptor read/64, error %d\n", + r); retval = -EMSGSIZE; continue; } @@ -2623,9 +2626,9 @@ hub_port_init (struct usb_hub *hub, struct usb_device *udev, int port1, retval = usb_get_device_descriptor(udev, 8); if (retval < 8) { - dev_err(&udev->dev, "device descriptor " - "read/%s, error %d\n", - "8", retval); + dev_err(&udev->dev, + "device descriptor read/8, error %d\n", + retval); if (retval >= 0) retval = -EMSGSIZE; } else { @@ -2652,8 +2655,8 @@ hub_port_init (struct usb_hub *hub, struct usb_device *udev, int port1, retval = usb_get_device_descriptor(udev, USB_DT_DEVICE_SIZE); if (retval < (signed)sizeof(udev->descriptor)) { - dev_err(&udev->dev, "device descriptor read/%s, error %d\n", - "all", retval); + dev_err(&udev->dev, "device descriptor read/all, error %d\n", + retval); if (retval >= 0) retval = -ENOMSG; goto fail; @@ -2721,9 +2724,9 @@ hub_power_remaining (struct usb_hub *hub) else delta = 8; if (delta > hub->mA_per_port) - dev_warn(&udev->dev, "%dmA is over %umA budget " - "for port %d!\n", - delta, hub->mA_per_port, port1); + dev_warn(&udev->dev, + "%dmA is over %umA budget for port %d!\n", + delta, hub->mA_per_port, port1); remaining -= delta; } if (remaining < 0) { From 3b23dd6f8a718e5339de4f7d86ce76a078b5f771 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Fri, 5 Dec 2008 14:10:34 -0500 Subject: [PATCH 074/123] USB: utilize the bus notifiers This patch (as1185) makes usbcore take advantage of the bus notifications sent out by the driver core. Now we can create all our device and interface attribute files before the device or interface uevent is broadcast. A side effect is that we no longer create the endpoint "pseudo" devices at the same time as a device or interface is registered -- it seems like a bad idea to try registering an endpoint before the registration of its parent is complete. So the routines for creating and removing endpoint devices have been split out and renamed, and they are called explicitly when needed. A new bitflag is used for keeping track of whether or not the interface's endpoint devices have been created, since (just as with the interface attributes) they vary with the altsetting and hence can be changed at random times. Signed-off-by: Alan Stern Cc: Kay Sievers Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/endpoint.c | 4 +-- drivers/usb/core/hub.c | 18 +++++-------- drivers/usb/core/message.c | 50 +++++++++++++++++++++++++++++++------ drivers/usb/core/sysfs.c | 22 +--------------- drivers/usb/core/usb.c | 37 +++++++++++++++++++++++++++ drivers/usb/core/usb.h | 4 +-- include/linux/usb.h | 2 ++ 7 files changed, 93 insertions(+), 44 deletions(-) diff --git a/drivers/usb/core/endpoint.c b/drivers/usb/core/endpoint.c index 946fae43d62..e1710f260b4 100644 --- a/drivers/usb/core/endpoint.c +++ b/drivers/usb/core/endpoint.c @@ -276,7 +276,7 @@ static void ep_device_release(struct device *dev) kfree(ep_dev); } -int usb_create_ep_files(struct device *parent, +int usb_create_ep_devs(struct device *parent, struct usb_host_endpoint *endpoint, struct usb_device *udev) { @@ -340,7 +340,7 @@ exit: return retval; } -void usb_remove_ep_files(struct usb_host_endpoint *endpoint) +void usb_remove_ep_devs(struct usb_host_endpoint *endpoint) { struct ep_device *ep_dev = endpoint->ep_dev; diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c index 5abdc11be1e..756b8d9993f 100644 --- a/drivers/usb/core/hub.c +++ b/drivers/usb/core/hub.c @@ -1437,17 +1437,12 @@ void usb_disconnect(struct usb_device **pdev) usb_disable_device(udev, 0); usb_hcd_synchronize_unlinks(udev); + usb_remove_ep_devs(&udev->ep0); usb_unlock_device(udev); - /* Remove the device-specific files from sysfs. This must be - * done with udev unlocked, because some of the attribute - * routines try to acquire the device lock. - */ - usb_remove_sysfs_dev_files(udev); - /* Unregister the device. The device driver is responsible - * for removing the device files from usbfs and sysfs and for - * de-configuring the device. + * for de-configuring the device and invoking the remove-device + * notifier chain (used by usbfs and possibly others). */ device_del(&udev->dev); @@ -1654,8 +1649,8 @@ int usb_new_device(struct usb_device *udev) announce_device(udev); /* Register the device. The device driver is responsible - * for adding the device files to sysfs and for configuring - * the device. + * for configuring the device and invoking the add-device + * notifier chain (used by usbfs and possibly others). */ err = device_add(&udev->dev); if (err) { @@ -1663,8 +1658,7 @@ int usb_new_device(struct usb_device *udev) goto fail; } - /* put device-specific files into sysfs */ - usb_create_sysfs_dev_files(udev); + (void) usb_create_ep_devs(&udev->dev, &udev->ep0, udev); return err; fail: diff --git a/drivers/usb/core/message.c b/drivers/usb/core/message.c index aadf29f09c4..7943901c641 100644 --- a/drivers/usb/core/message.c +++ b/drivers/usb/core/message.c @@ -1004,6 +1004,34 @@ int usb_clear_halt(struct usb_device *dev, int pipe) } EXPORT_SYMBOL_GPL(usb_clear_halt); +static int create_intf_ep_devs(struct usb_interface *intf) +{ + struct usb_device *udev = interface_to_usbdev(intf); + struct usb_host_interface *alt = intf->cur_altsetting; + int i; + + if (intf->ep_devs_created || intf->unregistering) + return 0; + + for (i = 0; i < alt->desc.bNumEndpoints; ++i) + (void) usb_create_ep_devs(&intf->dev, &alt->endpoint[i], udev); + intf->ep_devs_created = 1; + return 0; +} + +static void remove_intf_ep_devs(struct usb_interface *intf) +{ + struct usb_host_interface *alt = intf->cur_altsetting; + int i; + + if (!intf->ep_devs_created) + return; + + for (i = 0; i < alt->desc.bNumEndpoints; ++i) + usb_remove_ep_devs(&alt->endpoint[i]); + intf->ep_devs_created = 0; +} + /** * usb_disable_endpoint -- Disable an endpoint by address * @dev: the device whose endpoint is being disabled @@ -1092,7 +1120,7 @@ void usb_disable_device(struct usb_device *dev, int skip_ep0) dev_dbg(&dev->dev, "unregistering interface %s\n", dev_name(&interface->dev)); interface->unregistering = 1; - usb_remove_sysfs_intf_files(interface); + remove_intf_ep_devs(interface); device_del(&interface->dev); } @@ -1235,8 +1263,10 @@ int usb_set_interface(struct usb_device *dev, int interface, int alternate) */ /* prevent submissions using previous endpoint settings */ - if (iface->cur_altsetting != alt) + if (iface->cur_altsetting != alt) { + remove_intf_ep_devs(iface); usb_remove_sysfs_intf_files(iface); + } usb_disable_interface(dev, iface); iface->cur_altsetting = alt; @@ -1272,9 +1302,10 @@ int usb_set_interface(struct usb_device *dev, int interface, int alternate) * (Likewise, EP0 never "halts" on well designed devices.) */ usb_enable_interface(dev, iface); - if (device_is_registered(&iface->dev)) + if (device_is_registered(&iface->dev)) { usb_create_sysfs_intf_files(iface); - + create_intf_ep_devs(iface); + } return 0; } EXPORT_SYMBOL_GPL(usb_set_interface); @@ -1334,7 +1365,6 @@ int usb_reset_configuration(struct usb_device *dev) struct usb_interface *intf = config->interface[i]; struct usb_host_interface *alt; - usb_remove_sysfs_intf_files(intf); alt = usb_altnum_to_altsetting(intf, 0); /* No altsetting 0? We'll assume the first altsetting. @@ -1345,10 +1375,16 @@ int usb_reset_configuration(struct usb_device *dev) if (!alt) alt = &intf->altsetting[0]; + if (alt != intf->cur_altsetting) { + remove_intf_ep_devs(intf); + usb_remove_sysfs_intf_files(intf); + } intf->cur_altsetting = alt; usb_enable_interface(dev, intf); - if (device_is_registered(&intf->dev)) + if (device_is_registered(&intf->dev)) { usb_create_sysfs_intf_files(intf); + create_intf_ep_devs(intf); + } } return 0; } @@ -1682,7 +1718,7 @@ free_interfaces: dev_name(&intf->dev), ret); continue; } - usb_create_sysfs_intf_files(intf); + create_intf_ep_devs(intf); } usb_autosuspend_device(dev); diff --git a/drivers/usb/core/sysfs.c b/drivers/usb/core/sysfs.c index 0f0ccf64011..4cc2456ef3b 100644 --- a/drivers/usb/core/sysfs.c +++ b/drivers/usb/core/sysfs.c @@ -629,9 +629,6 @@ int usb_create_sysfs_dev_files(struct usb_device *udev) struct device *dev = &udev->dev; int retval; - /* Unforunately these attributes cannot be created before - * the uevent is broadcast. - */ retval = device_create_bin_file(dev, &dev_bin_attr_descriptors); if (retval) goto error; @@ -643,11 +640,7 @@ int usb_create_sysfs_dev_files(struct usb_device *udev) retval = add_power_attributes(dev); if (retval) goto error; - - retval = usb_create_ep_files(dev, &udev->ep0, udev); - if (retval) - goto error; - return 0; + return retval; error: usb_remove_sysfs_dev_files(udev); return retval; @@ -657,7 +650,6 @@ void usb_remove_sysfs_dev_files(struct usb_device *udev) { struct device *dev = &udev->dev; - usb_remove_ep_files(&udev->ep0); remove_power_attributes(dev); remove_persist_attributes(dev); device_remove_bin_file(dev, &dev_bin_attr_descriptors); @@ -816,36 +808,24 @@ int usb_create_sysfs_intf_files(struct usb_interface *intf) { struct usb_device *udev = interface_to_usbdev(intf); struct usb_host_interface *alt = intf->cur_altsetting; - int i; int retval; if (intf->sysfs_files_created || intf->unregistering) return 0; - /* The interface string may be present in some altsettings - * and missing in others. Hence its attribute cannot be created - * before the uevent is broadcast. - */ if (alt->string == NULL) alt->string = usb_cache_string(udev, alt->desc.iInterface); if (alt->string) retval = device_create_file(&intf->dev, &dev_attr_interface); - for (i = 0; i < alt->desc.bNumEndpoints; ++i) - usb_create_ep_files(&intf->dev, &alt->endpoint[i], udev); intf->sysfs_files_created = 1; return 0; } void usb_remove_sysfs_intf_files(struct usb_interface *intf) { - struct usb_host_interface *alt = intf->cur_altsetting; - int i; - if (!intf->sysfs_files_created) return; - for (i = 0; i < alt->desc.bNumEndpoints; ++i) - usb_remove_ep_files(&alt->endpoint[i]); device_remove_file(&intf->dev, &dev_attr_interface); intf->sysfs_files_created = 0; } diff --git a/drivers/usb/core/usb.c b/drivers/usb/core/usb.c index 4c98f3975af..c0821564a3f 100644 --- a/drivers/usb/core/usb.c +++ b/drivers/usb/core/usb.c @@ -970,6 +970,37 @@ int usb_disabled(void) } EXPORT_SYMBOL_GPL(usb_disabled); +/* + * Notifications of device and interface registration + */ +static int usb_bus_notify(struct notifier_block *nb, unsigned long action, + void *data) +{ + struct device *dev = data; + + switch (action) { + case BUS_NOTIFY_ADD_DEVICE: + if (dev->type == &usb_device_type) + (void) usb_create_sysfs_dev_files(to_usb_device(dev)); + else if (dev->type == &usb_if_device_type) + (void) usb_create_sysfs_intf_files( + to_usb_interface(dev)); + break; + + case BUS_NOTIFY_DEL_DEVICE: + if (dev->type == &usb_device_type) + usb_remove_sysfs_dev_files(to_usb_device(dev)); + else if (dev->type == &usb_if_device_type) + usb_remove_sysfs_intf_files(to_usb_interface(dev)); + break; + } + return 0; +} + +static struct notifier_block usb_bus_nb = { + .notifier_call = usb_bus_notify, +}; + /* * Init */ @@ -987,6 +1018,9 @@ static int __init usb_init(void) retval = bus_register(&usb_bus_type); if (retval) goto bus_register_failed; + retval = bus_register_notifier(&usb_bus_type, &usb_bus_nb); + if (retval) + goto bus_notifier_failed; retval = usb_host_init(); if (retval) goto host_init_failed; @@ -1021,6 +1055,8 @@ driver_register_failed: major_init_failed: usb_host_cleanup(); host_init_failed: + bus_unregister_notifier(&usb_bus_type, &usb_bus_nb); +bus_notifier_failed: bus_unregister(&usb_bus_type); bus_register_failed: ksuspend_usb_cleanup(); @@ -1044,6 +1080,7 @@ static void __exit usb_exit(void) usb_devio_cleanup(); usb_hub_cleanup(); usb_host_cleanup(); + bus_unregister_notifier(&usb_bus_type, &usb_bus_nb); bus_unregister(&usb_bus_type); ksuspend_usb_cleanup(); } diff --git a/drivers/usb/core/usb.h b/drivers/usb/core/usb.h index 9fb195665fa..381eae90c3b 100644 --- a/drivers/usb/core/usb.h +++ b/drivers/usb/core/usb.h @@ -6,10 +6,10 @@ extern int usb_create_sysfs_dev_files(struct usb_device *dev); extern void usb_remove_sysfs_dev_files(struct usb_device *dev); extern int usb_create_sysfs_intf_files(struct usb_interface *intf); extern void usb_remove_sysfs_intf_files(struct usb_interface *intf); -extern int usb_create_ep_files(struct device *parent, +extern int usb_create_ep_devs(struct device *parent, struct usb_host_endpoint *endpoint, struct usb_device *udev); -extern void usb_remove_ep_files(struct usb_host_endpoint *endpoint); +extern void usb_remove_ep_devs(struct usb_host_endpoint *endpoint); extern void usb_enable_endpoint(struct usb_device *dev, struct usb_host_endpoint *ep); diff --git a/include/linux/usb.h b/include/linux/usb.h index 74d0b9990c7..e9d63562325 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -108,6 +108,7 @@ enum usb_interface_condition { * (in probe()), bound to a driver, or unbinding (in disconnect()) * @is_active: flag set when the interface is bound and not suspended. * @sysfs_files_created: sysfs attributes exist + * @ep_devs_created: endpoint child pseudo-devices exist * @unregistering: flag set when the interface is being unregistered * @needs_remote_wakeup: flag set when the driver requires remote-wakeup * capability during autosuspend. @@ -169,6 +170,7 @@ struct usb_interface { enum usb_interface_condition condition; /* state of binding */ unsigned is_active:1; /* the interface is not suspended */ unsigned sysfs_files_created:1; /* the sysfs attributes exist */ + unsigned ep_devs_created:1; /* endpoint "devices" exist */ unsigned unregistering:1; /* unregistration is in progress */ unsigned needs_remote_wakeup:1; /* driver requires remote wakeup */ unsigned needs_altsetting0:1; /* switch to altsetting 0 is pending */ From 49367d8f1d9f26482cf7089489e90f0afd0a942c Mon Sep 17 00:00:00 2001 From: Ming Lei Date: Fri, 12 Dec 2008 21:38:45 +0800 Subject: [PATCH 075/123] USB: mark "reject" field of struct urb as atomic_t It is enough to protect accesses to reject field of urb by marking it as atomic_t,also it is the only reason of existence of usb_reject_lock,so remove the lock to make code more clean. Signed-off-by: Ming Lei Acked-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/hcd.c | 6 +++--- drivers/usb/core/urb.c | 23 +++++------------------ include/linux/usb.h | 2 +- 3 files changed, 9 insertions(+), 22 deletions(-) diff --git a/drivers/usb/core/hcd.c b/drivers/usb/core/hcd.c index a0079876d74..3c711db55d8 100644 --- a/drivers/usb/core/hcd.c +++ b/drivers/usb/core/hcd.c @@ -1010,7 +1010,7 @@ int usb_hcd_link_urb_to_ep(struct usb_hcd *hcd, struct urb *urb) spin_lock(&hcd_urb_list_lock); /* Check that the URB isn't being killed */ - if (unlikely(urb->reject)) { + if (unlikely(atomic_read(&urb->reject))) { rc = -EPERM; goto done; } @@ -1340,7 +1340,7 @@ int usb_hcd_submit_urb (struct urb *urb, gfp_t mem_flags) INIT_LIST_HEAD(&urb->urb_list); atomic_dec(&urb->use_count); atomic_dec(&urb->dev->urbnum); - if (urb->reject) + if (atomic_read(&urb->reject)) wake_up(&usb_kill_urb_queue); usb_put_urb(urb); } @@ -1444,7 +1444,7 @@ void usb_hcd_giveback_urb(struct usb_hcd *hcd, struct urb *urb, int status) urb->status = status; urb->complete (urb); atomic_dec (&urb->use_count); - if (unlikely (urb->reject)) + if (unlikely(atomic_read(&urb->reject))) wake_up (&usb_kill_urb_queue); usb_put_urb (urb); } diff --git a/drivers/usb/core/urb.c b/drivers/usb/core/urb.c index 1f68af9db3f..b5e9948698b 100644 --- a/drivers/usb/core/urb.c +++ b/drivers/usb/core/urb.c @@ -10,7 +10,6 @@ #define to_urb(d) container_of(d, struct urb, kref) -static DEFINE_SPINLOCK(usb_reject_lock); static void urb_destroy(struct kref *kref) { @@ -131,9 +130,7 @@ void usb_anchor_urb(struct urb *urb, struct usb_anchor *anchor) urb->anchor = anchor; if (unlikely(anchor->poisoned)) { - spin_lock(&usb_reject_lock); - urb->reject++; - spin_unlock(&usb_reject_lock); + atomic_inc(&urb->reject); } spin_unlock_irqrestore(&anchor->lock, flags); @@ -565,16 +562,12 @@ void usb_kill_urb(struct urb *urb) might_sleep(); if (!(urb && urb->dev && urb->ep)) return; - spin_lock_irq(&usb_reject_lock); - ++urb->reject; - spin_unlock_irq(&usb_reject_lock); + atomic_inc(&urb->reject); usb_hcd_unlink_urb(urb, -ENOENT); wait_event(usb_kill_urb_queue, atomic_read(&urb->use_count) == 0); - spin_lock_irq(&usb_reject_lock); - --urb->reject; - spin_unlock_irq(&usb_reject_lock); + atomic_dec(&urb->reject); } EXPORT_SYMBOL_GPL(usb_kill_urb); @@ -606,9 +599,7 @@ void usb_poison_urb(struct urb *urb) might_sleep(); if (!(urb && urb->dev && urb->ep)) return; - spin_lock_irq(&usb_reject_lock); - ++urb->reject; - spin_unlock_irq(&usb_reject_lock); + atomic_inc(&urb->reject); usb_hcd_unlink_urb(urb, -ENOENT); wait_event(usb_kill_urb_queue, atomic_read(&urb->use_count) == 0); @@ -617,14 +608,10 @@ EXPORT_SYMBOL_GPL(usb_poison_urb); void usb_unpoison_urb(struct urb *urb) { - unsigned long flags; - if (!urb) return; - spin_lock_irqsave(&usb_reject_lock, flags); - --urb->reject; - spin_unlock_irqrestore(&usb_reject_lock, flags); + atomic_dec(&urb->reject); } EXPORT_SYMBOL_GPL(usb_unpoison_urb); diff --git a/include/linux/usb.h b/include/linux/usb.h index e9d63562325..4e8654a1825 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -1340,7 +1340,7 @@ struct urb { struct kref kref; /* reference count of the URB */ void *hcpriv; /* private data for host controller */ atomic_t use_count; /* concurrent submissions counter */ - u8 reject; /* submissions will fail */ + atomic_t reject; /* submissions will fail */ int unlinked; /* unlink error code */ /* public: documented fields in the urb that can be used by drivers */ From 281b064f237205053ef1874ffc77b9211265af4c Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Sun, 14 Dec 2008 12:39:22 -0500 Subject: [PATCH 076/123] USB: unusual dev for Option N.V. ZeroCD modems Many newer Option mobile broadband devices initially provide a usb-storage "driver CD" device that's pretty useless on Linux since any software on it most likely wouldn't be compatible with your kernel or distro anyway. Thus, by default just kill the driver CD device by sending the SCSI 'rezero' command, but allow override of the default behavior via usb-storage module parameter so users can keep the ZeroCD device if they really want to. Inspired by the Sierra TruInstall patch. Signed-off-by: Dan Williams Acked-by: Marcel Holtmann Cc: Peter Henn Cc: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/Makefile | 2 +- drivers/usb/storage/option_ms.c | 147 +++++++++++++++++++++++++++++ drivers/usb/storage/option_ms.h | 4 + drivers/usb/storage/unusual_devs.h | 24 +++++ drivers/usb/storage/usb.c | 1 + 5 files changed, 177 insertions(+), 1 deletion(-) create mode 100644 drivers/usb/storage/option_ms.c create mode 100644 drivers/usb/storage/option_ms.h diff --git a/drivers/usb/storage/Makefile b/drivers/usb/storage/Makefile index facf610f168..b3206931339 100644 --- a/drivers/usb/storage/Makefile +++ b/drivers/usb/storage/Makefile @@ -23,7 +23,7 @@ usb-storage-obj-$(CONFIG_USB_STORAGE_KARMA) += karma.o usb-storage-obj-$(CONFIG_USB_STORAGE_CYPRESS_ATACB) += cypress_atacb.o usb-storage-objs := scsiglue.o protocol.o transport.o usb.o \ - initializers.o sierra_ms.o $(usb-storage-obj-y) + initializers.o sierra_ms.o option_ms.o $(usb-storage-obj-y) ifneq ($(CONFIG_USB_LIBUSUAL),) obj-$(CONFIG_USB) += libusual.o diff --git a/drivers/usb/storage/option_ms.c b/drivers/usb/storage/option_ms.c new file mode 100644 index 00000000000..353f922939a --- /dev/null +++ b/drivers/usb/storage/option_ms.c @@ -0,0 +1,147 @@ +/* + * Driver for Option High Speed Mobile Devices. + * + * (c) 2008 Dan Williams + * + * Inspiration taken from sierra_ms.c by Kevin Lloyd + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +#include + +#include "usb.h" +#include "transport.h" +#include "option_ms.h" +#include "debug.h" + +#define ZCD_FORCE_MODEM 0x01 +#define ZCD_ALLOW_MS 0x02 + +static unsigned int option_zero_cd = ZCD_FORCE_MODEM; +module_param(option_zero_cd, uint, S_IRUGO | S_IWUSR); +MODULE_PARM_DESC(option_zero_cd, "ZeroCD mode (1=Force Modem (default)," + " 2=Allow CD-Rom"); + +#define RESPONSE_LEN 1024 + +static int option_rezero(struct us_data *us, int ep_in, int ep_out) +{ + const unsigned char rezero_msg[] = { + 0x55, 0x53, 0x42, 0x43, 0x78, 0x56, 0x34, 0x12, + 0x01, 0x00, 0x00, 0x00, 0x80, 0x00, 0x06, 0x01, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 + }; + char *buffer; + int result; + + US_DEBUGP("Option MS: %s", "DEVICE MODE SWITCH\n"); + + buffer = kzalloc(RESPONSE_LEN, GFP_KERNEL); + if (buffer == NULL) + return USB_STOR_TRANSPORT_ERROR; + + memcpy(buffer, rezero_msg, sizeof (rezero_msg)); + result = usb_stor_bulk_transfer_buf(us, + usb_sndbulkpipe(us->pusb_dev, ep_out), + buffer, sizeof (rezero_msg), NULL); + if (result != USB_STOR_XFER_GOOD) { + result = USB_STOR_XFER_ERROR; + goto out; + } + + /* Some of the devices need to be asked for a response, but we don't + * care what that response is. + */ + result = usb_stor_bulk_transfer_buf(us, + usb_sndbulkpipe(us->pusb_dev, ep_out), + buffer, RESPONSE_LEN, NULL); + result = USB_STOR_XFER_GOOD; + +out: + kfree(buffer); + return result; +} + +int option_ms_init(struct us_data *us) +{ + struct usb_device *udev; + struct usb_interface *intf; + struct usb_host_interface *iface_desc; + struct usb_endpoint_descriptor *endpoint = NULL; + u8 ep_in = 0, ep_out = 0; + int ep_in_size = 0, ep_out_size = 0; + int i, result; + + udev = us->pusb_dev; + intf = us->pusb_intf; + + /* Ensure it's really a ZeroCD device; devices that are already + * in modem mode return 0xFF for class, subclass, and protocol. + */ + if (udev->descriptor.bDeviceClass != 0 || + udev->descriptor.bDeviceSubClass != 0 || + udev->descriptor.bDeviceProtocol != 0) + return USB_STOR_TRANSPORT_GOOD; + + US_DEBUGP("Option MS: option_ms_init called\n"); + + /* Find the right mass storage interface */ + iface_desc = intf->cur_altsetting; + if (iface_desc->desc.bInterfaceClass != 0x8 || + iface_desc->desc.bInterfaceSubClass != 0x6 || + iface_desc->desc.bInterfaceProtocol != 0x50) { + US_DEBUGP("Option MS: mass storage interface not found, no action " + "required\n"); + return USB_STOR_TRANSPORT_GOOD; + } + + /* Find the mass storage bulk endpoints */ + for (i = 0; i < iface_desc->desc.bNumEndpoints && (!ep_in_size || !ep_out_size); ++i) { + endpoint = &iface_desc->endpoint[i].desc; + + if (usb_endpoint_is_bulk_in(endpoint)) { + ep_in = usb_endpoint_num(endpoint); + ep_in_size = le16_to_cpu(endpoint->wMaxPacketSize); + } else if (usb_endpoint_is_bulk_out(endpoint)) { + ep_out = usb_endpoint_num(endpoint); + ep_out_size = le16_to_cpu(endpoint->wMaxPacketSize); + } + } + + /* Can't find the mass storage endpoints */ + if (!ep_in_size || !ep_out_size) { + US_DEBUGP("Option MS: mass storage endpoints not found, no action " + "required\n"); + return USB_STOR_TRANSPORT_GOOD; + } + + /* Force Modem mode */ + if (option_zero_cd == ZCD_FORCE_MODEM) { + US_DEBUGP("Option MS: %s", "Forcing Modem Mode\n"); + result = option_rezero(us, ep_in, ep_out); + if (result != USB_STOR_XFER_GOOD) + US_DEBUGP("Option MS: Failed to switch to modem mode.\n"); + return -EIO; + } else if (option_zero_cd == ZCD_ALLOW_MS) { + /* Allow Mass Storage mode (keep CD-Rom) */ + US_DEBUGP("Option MS: %s", "Allowing Mass Storage Mode if device" + " requests it\n"); + } + + return USB_STOR_TRANSPORT_GOOD; +} + diff --git a/drivers/usb/storage/option_ms.h b/drivers/usb/storage/option_ms.h new file mode 100644 index 00000000000..b6e448cab03 --- /dev/null +++ b/drivers/usb/storage/option_ms.h @@ -0,0 +1,4 @@ +#ifndef _OPTION_MS_H_ +#define _OPTION_MS_H_ +extern int option_ms_init(struct us_data *us); +#endif diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index 0fd42a0c794..0330ed53ec1 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -985,6 +985,18 @@ UNUSUAL_DEV( 0x05ac, 0x120a, 0x0000, 0x9999, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_CAPACITY ), +/* Reported by Dan Williams + * Option N.V. mobile broadband modems + * Ignore driver CD mode and force into modem mode by default. + */ + +/* Globetrotter HSDPA; mass storage shows up as Qualcomm for vendor */ +UNUSUAL_DEV( 0x05c6, 0x1000, 0x0000, 0x9999, + "Option N.V.", + "Mass Storage", + US_SC_DEVICE, US_PR_DEVICE, option_ms_init, + 0), + #ifdef CONFIG_USB_STORAGE_JUMPSHOT UNUSUAL_DEV( 0x05dc, 0x0001, 0x0000, 0x0001, "Lexar", @@ -1474,6 +1486,18 @@ UNUSUAL_DEV( 0x0ace, 0x20ff, 0x0101, 0x0101, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_IGNORE_DEVICE ), +/* Reported by Dan Williams + * Option N.V. mobile broadband modems + * Ignore driver CD mode and force into modem mode by default. + */ + +/* iCON 225 */ +UNUSUAL_DEV( 0x0af0, 0x6971, 0x0000, 0x9999, + "Option N.V.", + "Mass Storage", + US_SC_DEVICE, US_PR_DEVICE, option_ms_init, + 0), + /* Reported by F. Aben * This device (wrongly) has a vendor-specific device descriptor. * The entry is needed so usb-storage can bind to it's mass-storage diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index b25c448d5eb..ce0b580db5e 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -100,6 +100,7 @@ #include "cypress_atacb.h" #endif #include "sierra_ms.h" +#include "option_ms.h" /* Some informational data */ MODULE_AUTHOR("Matthew Dharm "); From 42c65396d4f10f25bdab13f8e2f33fe63fa94418 Mon Sep 17 00:00:00 2001 From: Thomas Hommel Date: Thu, 18 Dec 2008 10:31:40 +0100 Subject: [PATCH 077/123] USB: isp1760: don't auto disable Port1 on ISP1761 There is no need to disable port 1 on ISP1761. That port could be used as an OTG port which would require a different init sequence. However we don't have OTG support (yet) so we can use it as a normal USB port. This patch allows port 1 to be used a normal Port on the ISP1761. Signed-off-by: Thomas Hommel Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Greg Kroah-Hartman --- drivers/usb/host/isp1760-hcd.c | 13 ++++++------- drivers/usb/host/isp1760-hcd.h | 1 - drivers/usb/host/isp1760-if.c | 3 --- 3 files changed, 6 insertions(+), 11 deletions(-) diff --git a/drivers/usb/host/isp1760-hcd.c b/drivers/usb/host/isp1760-hcd.c index 8017f1cf78e..b899f1a59c2 100644 --- a/drivers/usb/host/isp1760-hcd.c +++ b/drivers/usb/host/isp1760-hcd.c @@ -435,14 +435,13 @@ static int isp1760_hc_setup(struct usb_hcd *hcd) /* * PORT 1 Control register of the ISP1760 is the OTG control - * register on ISP1761. + * register on ISP1761. Since there is no OTG or device controller + * support in this driver, we use port 1 as a "normal" USB host port on + * both chips. */ - if (!(priv->devflags & ISP1760_FLAG_ISP1761) && - !(priv->devflags & ISP1760_FLAG_PORT1_DIS)) { - isp1760_writel(PORT1_POWER | PORT1_INIT2, - hcd->regs + HC_PORT1_CTRL); - mdelay(10); - } + isp1760_writel(PORT1_POWER | PORT1_INIT2, + hcd->regs + HC_PORT1_CTRL); + mdelay(10); priv->hcs_params = isp1760_readl(hcd->regs + HC_HCSPARAMS); diff --git a/drivers/usb/host/isp1760-hcd.h b/drivers/usb/host/isp1760-hcd.h index 4377277667d..a9daea58796 100644 --- a/drivers/usb/host/isp1760-hcd.h +++ b/drivers/usb/host/isp1760-hcd.h @@ -135,7 +135,6 @@ typedef void (packet_enqueue)(struct usb_hcd *hcd, struct isp1760_qh *qh, * indicate the most "atypical" case, so that a devflags of 0 is * a sane default configuration. */ -#define ISP1760_FLAG_PORT1_DIS 0x00000001 /* Port 1 disabled */ #define ISP1760_FLAG_BUS_WIDTH_16 0x00000002 /* 16-bit data bus width */ #define ISP1760_FLAG_OTG_EN 0x00000004 /* Port 1 supports OTG */ #define ISP1760_FLAG_ANALOG_OC 0x00000008 /* Analog overcurrent */ diff --git a/drivers/usb/host/isp1760-if.c b/drivers/usb/host/isp1760-if.c index 25453066fda..4cf7ca428b3 100644 --- a/drivers/usb/host/isp1760-if.c +++ b/drivers/usb/host/isp1760-if.c @@ -60,9 +60,6 @@ static int of_isp1760_probe(struct of_device *dev, if (of_device_is_compatible(dp, "nxp,usb-isp1761")) devflags |= ISP1760_FLAG_ISP1761; - if (of_get_property(dp, "port1-disable", NULL) != NULL) - devflags |= ISP1760_FLAG_PORT1_DIS; - /* Some systems wire up only 16 of the 32 data lines */ prop = of_get_property(dp, "bus-width", NULL); if (prop && *prop == 16) From 856395d6e137b4e7194972cb7765f3de6a72ba61 Mon Sep 17 00:00:00 2001 From: Oliver Neukum Date: Thu, 18 Dec 2008 09:17:49 +0100 Subject: [PATCH 078/123] USB: extension of anchor API to unpoison an anchor This extension allows unpoisoning an anchor allowing drivers that resubmit URBs to reuse an anchor for methods like resume() Signed-off-by: Oliver Neukum Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/urb.c | 20 ++++++++++++++++++++ include/linux/usb.h | 1 + 2 files changed, 21 insertions(+) diff --git a/drivers/usb/core/urb.c b/drivers/usb/core/urb.c index b5e9948698b..58bc5e3c256 100644 --- a/drivers/usb/core/urb.c +++ b/drivers/usb/core/urb.c @@ -678,6 +678,26 @@ void usb_poison_anchored_urbs(struct usb_anchor *anchor) } EXPORT_SYMBOL_GPL(usb_poison_anchored_urbs); +/** + * usb_unpoison_anchored_urbs - let an anchor be used successfully again + * @anchor: anchor the requests are bound to + * + * Reverses the effect of usb_poison_anchored_urbs + * the anchor can be used normally after it returns + */ +void usb_unpoison_anchored_urbs(struct usb_anchor *anchor) +{ + unsigned long flags; + struct urb *lazarus; + + spin_lock_irqsave(&anchor->lock, flags); + list_for_each_entry(lazarus, &anchor->urb_list, anchor_list) { + usb_unpoison_urb(lazarus); + } + anchor->poisoned = 0; + spin_unlock_irqrestore(&anchor->lock, flags); +} +EXPORT_SYMBOL_GPL(usb_unpoison_anchored_urbs); /** * usb_unlink_anchored_urbs - asynchronously cancel transfer requests en masse * @anchor: anchor the requests are bound to diff --git a/include/linux/usb.h b/include/linux/usb.h index 4e8654a1825..e8963989650 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -1485,6 +1485,7 @@ extern void usb_poison_urb(struct urb *urb); extern void usb_unpoison_urb(struct urb *urb); extern void usb_kill_anchored_urbs(struct usb_anchor *anchor); extern void usb_poison_anchored_urbs(struct usb_anchor *anchor); +extern void usb_unpoison_anchored_urbs(struct usb_anchor *anchor); extern void usb_unlink_anchored_urbs(struct usb_anchor *anchor); extern void usb_anchor_urb(struct urb *urb, struct usb_anchor *anchor); extern void usb_unanchor_urb(struct urb *urb); From 2a4f136fbdcd89d44d83ed54df2c492a89f5ba9c Mon Sep 17 00:00:00 2001 From: Darius Augulis Date: Wed, 12 Nov 2008 13:38:31 -0800 Subject: [PATCH 079/123] USB: add imx udc gadget driver Implementation of USB device driver integrated in Freescale's i.MXL processor. Adds USB device driver for i.MXL. Signed-off-by: Darius Augulis Signed-off-by: Greg Kroah-Hartman --- arch/arm/plat-mxc/include/mach/usb.h | 23 + drivers/usb/gadget/Kconfig | 21 + drivers/usb/gadget/Makefile | 1 + drivers/usb/gadget/gadget_chips.h | 1 - drivers/usb/gadget/imx_udc.c | 1516 ++++++++++++++++++++++++++ drivers/usb/gadget/imx_udc.h | 344 ++++++ 6 files changed, 1905 insertions(+), 1 deletion(-) create mode 100644 arch/arm/plat-mxc/include/mach/usb.h create mode 100644 drivers/usb/gadget/imx_udc.c create mode 100644 drivers/usb/gadget/imx_udc.h diff --git a/arch/arm/plat-mxc/include/mach/usb.h b/arch/arm/plat-mxc/include/mach/usb.h new file mode 100644 index 00000000000..2dacb3086f1 --- /dev/null +++ b/arch/arm/plat-mxc/include/mach/usb.h @@ -0,0 +1,23 @@ +/* + * Copyright (C) 2008 Darius Augulis + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ + +#ifndef __ASM_ARCH_MXC_USB +#define __ASM_ARCH_MXC_USB + +struct imxusb_platform_data { + int (*init)(struct device *); + int (*exit)(struct device *); +}; + +#endif /* __ASM_ARCH_MXC_USB */ diff --git a/drivers/usb/gadget/Kconfig b/drivers/usb/gadget/Kconfig index c322024bedf..3219d137340 100644 --- a/drivers/usb/gadget/Kconfig +++ b/drivers/usb/gadget/Kconfig @@ -305,6 +305,27 @@ config USB_GADGET_MUSB_HDRC This OTG-capable silicon IP is used in dual designs including the TI DaVinci, OMAP 243x, OMAP 343x, TUSB 6010, and ADI Blackfin +config USB_GADGET_IMX + boolean "Freescale IMX USB Peripheral Controller" + depends on ARCH_MX1 + help + Freescale's IMX series include an integrated full speed + USB 1.1 device controller. The controller in the IMX series + is register-compatible. + + It has Six fixed-function endpoints, as well as endpoint + zero (for control transfers). + + Say "y" to link the driver statically, or "m" to build a + dynamically linked module called "imx_udc" and force all + gadget drivers to also be dynamically linked. + +config USB_IMX + tristate + depends on USB_GADGET_IMX + default USB_GADGET + select USB_GADGET_SELECTED + config USB_GADGET_M66592 boolean "Renesas M66592 USB Peripheral Controller" select USB_GADGET_DUALSPEED diff --git a/drivers/usb/gadget/Makefile b/drivers/usb/gadget/Makefile index 5d64a2ab188..39a51d746cb 100644 --- a/drivers/usb/gadget/Makefile +++ b/drivers/usb/gadget/Makefile @@ -10,6 +10,7 @@ obj-$(CONFIG_USB_NET2280) += net2280.o obj-$(CONFIG_USB_AMD5536UDC) += amd5536udc.o obj-$(CONFIG_USB_PXA25X) += pxa25x_udc.o obj-$(CONFIG_USB_PXA27X) += pxa27x_udc.o +obj-$(CONFIG_USB_IMX) += imx_udc.o obj-$(CONFIG_USB_GOKU) += goku_udc.o obj-$(CONFIG_USB_OMAP) += omap_udc.o obj-$(CONFIG_USB_LH7A40X) += lh7a40x_udc.o diff --git a/drivers/usb/gadget/gadget_chips.h b/drivers/usb/gadget/gadget_chips.h index c0679b17602..ec6d439a2aa 100644 --- a/drivers/usb/gadget/gadget_chips.h +++ b/drivers/usb/gadget/gadget_chips.h @@ -110,7 +110,6 @@ #define gadget_is_at91(g) 0 #endif -/* status unclear */ #ifdef CONFIG_USB_GADGET_IMX #define gadget_is_imx(g) !strcmp("imx_udc", (g)->name) #else diff --git a/drivers/usb/gadget/imx_udc.c b/drivers/usb/gadget/imx_udc.c new file mode 100644 index 00000000000..cde8fdf15d5 --- /dev/null +++ b/drivers/usb/gadget/imx_udc.c @@ -0,0 +1,1516 @@ +/* + * driver/usb/gadget/imx_udc.c + * + * Copyright (C) 2005 Mike Lee(eemike@gmail.com) + * Copyright (C) 2008 Darius Augulis + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include +#include + +#include "imx_udc.h" + +static const char driver_name[] = "imx_udc"; +static const char ep0name[] = "ep0"; + +void ep0_chg_stat(const char *label, struct imx_udc_struct *imx_usb, + enum ep0_state stat); + +/******************************************************************************* + * IMX UDC hardware related functions + ******************************************************************************* + */ + +void imx_udc_enable(struct imx_udc_struct *imx_usb) +{ + int temp = __raw_readl(imx_usb->base + USB_CTRL); + __raw_writel(temp | CTRL_FE_ENA | CTRL_AFE_ENA, imx_usb->base + USB_CTRL); + imx_usb->gadget.speed = USB_SPEED_FULL; +} + +void imx_udc_disable(struct imx_udc_struct *imx_usb) +{ + int temp = __raw_readl(imx_usb->base + USB_CTRL); + + __raw_writel(temp & ~(CTRL_FE_ENA | CTRL_AFE_ENA), + imx_usb->base + USB_CTRL); + + ep0_chg_stat(__func__, imx_usb, EP0_IDLE); + imx_usb->gadget.speed = USB_SPEED_UNKNOWN; +} + +void imx_udc_reset(struct imx_udc_struct *imx_usb) +{ + int temp = __raw_readl(imx_usb->base + USB_ENAB); + + /* set RST bit */ + __raw_writel(temp | ENAB_RST, imx_usb->base + USB_ENAB); + + /* wait RST bit to clear */ + do {} while (__raw_readl(imx_usb->base + USB_ENAB) & ENAB_RST); + + /* wait CFG bit to assert */ + do {} while (!(__raw_readl(imx_usb->base + USB_DADR) & DADR_CFG)); + + /* udc module is now ready */ +} + +void imx_udc_config(struct imx_udc_struct *imx_usb) +{ + u8 ep_conf[5]; + u8 i, j, cfg; + struct imx_ep_struct *imx_ep; + + /* wait CFG bit to assert */ + do {} while (!(__raw_readl(imx_usb->base + USB_DADR) & DADR_CFG)); + + /* Download the endpoint buffer for endpoint 0. */ + for (j = 0; j < 5; j++) { + i = (j == 2 ? imx_usb->imx_ep[0].fifosize : 0x00); + __raw_writeb(i, imx_usb->base + USB_DDAT); + do {} while (__raw_readl(imx_usb->base + USB_DADR) & DADR_BSY); + } + + /* Download the endpoint buffers for endpoints 1-5. + * We specify two configurations, one interface + */ + for (cfg = 1; cfg < 3; cfg++) { + for (i = 1; i < IMX_USB_NB_EP; i++) { + imx_ep = &imx_usb->imx_ep[i]; + /* EP no | Config no */ + ep_conf[0] = (i << 4) | (cfg << 2); + /* Type | Direction */ + ep_conf[1] = (imx_ep->bmAttributes << 3) | + (EP_DIR(imx_ep) << 2); + /* Max packet size */ + ep_conf[2] = imx_ep->fifosize; + /* TRXTYP */ + ep_conf[3] = 0xC0; + /* FIFO no */ + ep_conf[4] = i; + + D_INI(imx_usb->dev, + "<%s> ep%d_conf[%d]:" + "[%02x-%02x-%02x-%02x-%02x]\n", + __func__, i, cfg, + ep_conf[0], ep_conf[1], ep_conf[2], + ep_conf[3], ep_conf[4]); + + for (j = 0; j < 5; j++) { + __raw_writeb(ep_conf[j], + imx_usb->base + USB_DDAT); + do {} while (__raw_readl(imx_usb->base + USB_DADR) + & DADR_BSY); + } + } + } + + /* wait CFG bit to clear */ + do {} while (__raw_readl(imx_usb->base + USB_DADR) & DADR_CFG); +} + +void imx_udc_init_irq(struct imx_udc_struct *imx_usb) +{ + int i; + + /* Mask and clear all irqs */ + __raw_writel(0xFFFFFFFF, imx_usb->base + USB_MASK); + __raw_writel(0xFFFFFFFF, imx_usb->base + USB_INTR); + for (i = 0; i < IMX_USB_NB_EP; i++) { + __raw_writel(0x1FF, imx_usb->base + USB_EP_MASK(i)); + __raw_writel(0x1FF, imx_usb->base + USB_EP_INTR(i)); + } + + /* Enable USB irqs */ + __raw_writel(INTR_MSOF | INTR_FRAME_MATCH, imx_usb->base + USB_MASK); + + /* Enable EP0 irqs */ + __raw_writel(0x1FF & ~(EPINTR_DEVREQ | EPINTR_MDEVREQ | EPINTR_EOT + | EPINTR_EOF | EPINTR_FIFO_EMPTY | EPINTR_FIFO_FULL), + imx_usb->base + USB_EP_MASK(0)); +} + +void imx_udc_init_ep(struct imx_udc_struct *imx_usb) +{ + int i, max, temp; + struct imx_ep_struct *imx_ep; + for (i = 0; i < IMX_USB_NB_EP; i++) { + imx_ep = &imx_usb->imx_ep[i]; + switch (imx_ep->fifosize) { + case 8: + max = 0; + break; + case 16: + max = 1; + break; + case 32: + max = 2; + break; + case 64: + max = 3; + break; + default: + max = 1; + break; + } + temp = (EP_DIR(imx_ep) << 7) | (max << 5) + | (imx_ep->bmAttributes << 3); + __raw_writel(temp, imx_usb->base + USB_EP_STAT(i)); + __raw_writel(temp | EPSTAT_FLUSH, imx_usb->base + USB_EP_STAT(i)); + D_INI(imx_usb->dev, "<%s> ep%d_stat %08x\n", __func__, i, + __raw_readl(imx_usb->base + USB_EP_STAT(i))); + } +} + +void imx_udc_init_fifo(struct imx_udc_struct *imx_usb) +{ + int i, temp; + struct imx_ep_struct *imx_ep; + for (i = 0; i < IMX_USB_NB_EP; i++) { + imx_ep = &imx_usb->imx_ep[i]; + + /* Fifo control */ + temp = EP_DIR(imx_ep) ? 0x0B000000 : 0x0F000000; + __raw_writel(temp, imx_usb->base + USB_EP_FCTRL(i)); + D_INI(imx_usb->dev, "<%s> ep%d_fctrl %08x\n", __func__, i, + __raw_readl(imx_usb->base + USB_EP_FCTRL(i))); + + /* Fifo alarm */ + temp = (i ? imx_ep->fifosize / 2 : 0); + __raw_writel(temp, imx_usb->base + USB_EP_FALRM(i)); + D_INI(imx_usb->dev, "<%s> ep%d_falrm %08x\n", __func__, i, + __raw_readl(imx_usb->base + USB_EP_FALRM(i))); + } +} + +static void imx_udc_init(struct imx_udc_struct *imx_usb) +{ + /* Reset UDC */ + imx_udc_reset(imx_usb); + + /* Download config to enpoint buffer */ + imx_udc_config(imx_usb); + + /* Setup interrups */ + imx_udc_init_irq(imx_usb); + + /* Setup endpoints */ + imx_udc_init_ep(imx_usb); + + /* Setup fifos */ + imx_udc_init_fifo(imx_usb); +} + +void imx_ep_irq_enable(struct imx_ep_struct *imx_ep) +{ + + int i = EP_NO(imx_ep); + + __raw_writel(0x1FF, imx_ep->imx_usb->base + USB_EP_MASK(i)); + __raw_writel(0x1FF, imx_ep->imx_usb->base + USB_EP_INTR(i)); + __raw_writel(0x1FF & ~(EPINTR_EOT | EPINTR_EOF), + imx_ep->imx_usb->base + USB_EP_MASK(i)); +} + +void imx_ep_irq_disable(struct imx_ep_struct *imx_ep) +{ + + int i = EP_NO(imx_ep); + + __raw_writel(0x1FF, imx_ep->imx_usb->base + USB_EP_MASK(i)); + __raw_writel(0x1FF, imx_ep->imx_usb->base + USB_EP_INTR(i)); +} + +int imx_ep_empty(struct imx_ep_struct *imx_ep) +{ + struct imx_udc_struct *imx_usb = imx_ep->imx_usb; + + return __raw_readl(imx_usb->base + USB_EP_FSTAT(EP_NO(imx_ep))) + & FSTAT_EMPTY; +} + +unsigned imx_fifo_bcount(struct imx_ep_struct *imx_ep) +{ + struct imx_udc_struct *imx_usb = imx_ep->imx_usb; + + return (__raw_readl(imx_usb->base + USB_EP_STAT(EP_NO(imx_ep))) + & EPSTAT_BCOUNT) >> 16; +} + +void imx_flush(struct imx_ep_struct *imx_ep) +{ + struct imx_udc_struct *imx_usb = imx_ep->imx_usb; + + int temp = __raw_readl(imx_usb->base + USB_EP_STAT(EP_NO(imx_ep))); + __raw_writel(temp | EPSTAT_FLUSH, + imx_usb->base + USB_EP_STAT(EP_NO(imx_ep))); +} + +void imx_ep_stall(struct imx_ep_struct *imx_ep) +{ + struct imx_udc_struct *imx_usb = imx_ep->imx_usb; + int temp, i; + + D_ERR(imx_usb->dev, "<%s> Forced stall on %s\n", __func__, imx_ep->ep.name); + + imx_flush(imx_ep); + + /* Special care for ep0 */ + if (EP_NO(imx_ep)) { + temp = __raw_readl(imx_usb->base + USB_CTRL); + __raw_writel(temp | CTRL_CMDOVER | CTRL_CMDERROR, imx_usb->base + USB_CTRL); + do { } while (__raw_readl(imx_usb->base + USB_CTRL) & CTRL_CMDOVER); + temp = __raw_readl(imx_usb->base + USB_CTRL); + __raw_writel(temp & ~CTRL_CMDERROR, imx_usb->base + USB_CTRL); + } + else { + temp = __raw_readl(imx_usb->base + USB_EP_STAT(EP_NO(imx_ep))); + __raw_writel(temp | EPSTAT_STALL, + imx_usb->base + USB_EP_STAT(EP_NO(imx_ep))); + + for (i = 0; i < 100; i ++) { + temp = __raw_readl(imx_usb->base + USB_EP_STAT(EP_NO(imx_ep))); + if (!temp & EPSTAT_STALL) + break; + udelay(20); + } + if (i == 50) + D_ERR(imx_usb->dev, "<%s> Non finished stall on %s\n", + __func__, imx_ep->ep.name); + } +} + +static int imx_udc_get_frame(struct usb_gadget *_gadget) +{ + struct imx_udc_struct *imx_usb = container_of(_gadget, + struct imx_udc_struct, gadget); + + return __raw_readl(imx_usb->base + USB_FRAME) & 0x7FF; +} + +static int imx_udc_wakeup(struct usb_gadget *_gadget) +{ + return 0; +} + +/******************************************************************************* + * USB request control functions + ******************************************************************************* + */ + +static void ep_add_request(struct imx_ep_struct *imx_ep, struct imx_request *req) +{ + if (unlikely(!req)) + return; + + req->in_use = 1; + list_add_tail(&req->queue, &imx_ep->queue); +} + +static void ep_del_request(struct imx_ep_struct *imx_ep, struct imx_request *req) +{ + if (unlikely(!req)) + return; + + list_del_init(&req->queue); + req->in_use = 0; +} + +static void done(struct imx_ep_struct *imx_ep, struct imx_request *req, int status) +{ + ep_del_request(imx_ep, req); + + if (likely(req->req.status == -EINPROGRESS)) + req->req.status = status; + else + status = req->req.status; + + if (status && status != -ESHUTDOWN) + D_ERR(imx_ep->imx_usb->dev, + "<%s> complete %s req %p stat %d len %u/%u\n", __func__, + imx_ep->ep.name, &req->req, status, + req->req.actual, req->req.length); + + req->req.complete(&imx_ep->ep, &req->req); +} + +static void nuke(struct imx_ep_struct *imx_ep, int status) +{ + struct imx_request *req; + + while (!list_empty(&imx_ep->queue)) { + req = list_entry(imx_ep->queue.next, struct imx_request, queue); + done(imx_ep, req, status); + } +} + +/******************************************************************************* + * Data tansfer over USB functions + ******************************************************************************* + */ +static int read_packet(struct imx_ep_struct *imx_ep, struct imx_request *req) +{ + u8 *buf; + int bytes_ep, bufferspace, count, i; + + bytes_ep = imx_fifo_bcount(imx_ep); + bufferspace = req->req.length - req->req.actual; + + buf = req->req.buf + req->req.actual; + prefetchw(buf); + + if (unlikely(imx_ep_empty(imx_ep))) + count = 0; /* zlp */ + else + count = min(bytes_ep, bufferspace); + + for (i = count; i > 0; i--) + *buf++ = __raw_readb(imx_ep->imx_usb->base + + USB_EP_FDAT0(EP_NO(imx_ep))); + req->req.actual += count; + + return count; +} + +static int write_packet(struct imx_ep_struct *imx_ep, struct imx_request *req) +{ + u8 *buf; + int length, count, temp; + + buf = req->req.buf + req->req.actual; + prefetch(buf); + + length = min(req->req.length - req->req.actual, (u32)imx_ep->fifosize); + + if (imx_fifo_bcount(imx_ep) + length > imx_ep->fifosize) { + D_TRX(imx_ep->imx_usb->dev, "<%s> packet overfill %s fifo\n", + __func__, imx_ep->ep.name); + return -1; + } + + req->req.actual += length; + count = length; + + if (!count && req->req.zero) { /* zlp */ + temp = __raw_readl(imx_ep->imx_usb->base + + USB_EP_STAT(EP_NO(imx_ep))); + __raw_writel(temp | EPSTAT_ZLPS, imx_ep->imx_usb->base + + USB_EP_STAT(EP_NO(imx_ep))); + D_TRX(imx_ep->imx_usb->dev, "<%s> zero packet\n", __func__); + return 0; + } + + while (count--) { + if (count == 0) { /* last byte */ + temp = __raw_readl(imx_ep->imx_usb->base + + USB_EP_FCTRL(EP_NO(imx_ep))); + __raw_writel(temp | FCTRL_WFR, imx_ep->imx_usb->base + + USB_EP_FCTRL(EP_NO(imx_ep))); + } + __raw_writeb(*buf++, + imx_ep->imx_usb->base + USB_EP_FDAT0(EP_NO(imx_ep))); + } + + return length; +} + +static int read_fifo(struct imx_ep_struct *imx_ep, struct imx_request *req) +{ + int bytes = 0, + count, + completed = 0; + + while (__raw_readl(imx_ep->imx_usb->base + USB_EP_FSTAT(EP_NO(imx_ep))) + & FSTAT_FR) { + count = read_packet(imx_ep, req); + bytes += count; + + completed = (count != imx_ep->fifosize); + if (completed || req->req.actual == req->req.length) { + completed = 1; + break; + } + } + + if (completed || !req->req.length) { + done(imx_ep, req, 0); + D_REQ(imx_ep->imx_usb->dev, "<%s> %s req<%p> %s\n", + __func__, imx_ep->ep.name, req, + completed ? "completed" : "not completed"); + if (!EP_NO(imx_ep)) + ep0_chg_stat(__func__, imx_ep->imx_usb, EP0_IDLE); + } + + D_TRX(imx_ep->imx_usb->dev, "<%s> bytes read: %d\n", __func__, bytes); + + return completed; +} + +static int write_fifo(struct imx_ep_struct *imx_ep, struct imx_request *req) +{ + int bytes = 0, + count, + completed = 0; + + while (!completed) { + count = write_packet(imx_ep, req); + if (count < 0) + break; /* busy */ + bytes += count; + + /* last packet "must be" short (or a zlp) */ + completed = (count != imx_ep->fifosize); + + if (unlikely(completed)) { + done(imx_ep, req, 0); + D_REQ(imx_ep->imx_usb->dev, "<%s> %s req<%p> %s\n", + __func__, imx_ep->ep.name, req, + completed ? "completed" : "not completed"); + if (!EP_NO(imx_ep)) + ep0_chg_stat(__func__, imx_ep->imx_usb, EP0_IDLE); + } + } + + D_TRX(imx_ep->imx_usb->dev, "<%s> bytes sent: %d\n", __func__, bytes); + + return completed; +} + +/******************************************************************************* + * Endpoint handlers + ******************************************************************************* + */ +static int handle_ep(struct imx_ep_struct *imx_ep) +{ + struct imx_request *req; + int completed = 0; + + do { + if (!list_empty(&imx_ep->queue)) + req = list_entry(imx_ep->queue.next, + struct imx_request, queue); + else { + D_REQ(imx_ep->imx_usb->dev, "<%s> no request on %s\n", + __func__, imx_ep->ep.name); + return 0; + } + + if (EP_DIR(imx_ep)) /* to host */ + completed = write_fifo(imx_ep, req); + else /* to device */ + completed = read_fifo(imx_ep, req); + + dump_ep_stat(__func__, imx_ep); + + } while (completed); + + return 0; +} + +static int handle_ep0(struct imx_ep_struct *imx_ep) +{ + struct imx_request *req = NULL; + int ret = 0; + + if (!list_empty(&imx_ep->queue)) + req = list_entry(imx_ep->queue.next, struct imx_request, queue); + + if (req) { + switch (imx_ep->imx_usb->ep0state) { + + case EP0_IN_DATA_PHASE: /* GET_DESCRIPTOR */ + write_fifo(imx_ep, req); + break; + case EP0_OUT_DATA_PHASE: /* SET_DESCRIPTOR */ + read_fifo(imx_ep, req); + break; + default: + D_EP0(imx_ep->imx_usb->dev, + "<%s> ep0 i/o, odd state %d\n", + __func__, imx_ep->imx_usb->ep0state); + ep_del_request(imx_ep, req); + ret = -EL2HLT; + break; + } + } + + return ret; +} + +static void handle_ep0_devreq(struct imx_udc_struct *imx_usb) +{ + struct imx_ep_struct *imx_ep = &imx_usb->imx_ep[0]; + union { + struct usb_ctrlrequest r; + u8 raw[8]; + u32 word[2]; + } u; + int temp, i; + + nuke(imx_ep, -EPROTO); + + /* read SETUP packet */ + for (i = 0; i < 2; i++) { + if (imx_ep_empty(imx_ep)) { + D_ERR(imx_usb->dev, + "<%s> no setup packet received\n", __func__); + goto stall; + } + u.word[i] = __raw_readl(imx_usb->base + USB_EP_FDAT(EP_NO(imx_ep))); + } + + temp = imx_ep_empty(imx_ep); + while (!imx_ep_empty(imx_ep)) { + i = __raw_readl(imx_usb->base + USB_EP_FDAT(EP_NO(imx_ep))); + D_ERR(imx_usb->dev, + "<%s> wrong to have extra bytes for setup : 0x%08x\n", + __func__, i); + } + if (!temp) + goto stall; + + le16_to_cpus(&u.r.wValue); + le16_to_cpus(&u.r.wIndex); + le16_to_cpus(&u.r.wLength); + + D_REQ(imx_usb->dev, "<%s> SETUP %02x.%02x v%04x i%04x l%04x\n", + __func__, u.r.bRequestType, u.r.bRequest, + u.r.wValue, u.r.wIndex, u.r.wLength); + + if (imx_usb->set_config) { + /* NACK the host by using CMDOVER */ + temp = __raw_readl(imx_usb->base + USB_CTRL); + __raw_writel(temp | CTRL_CMDOVER, imx_usb->base + USB_CTRL); + + D_ERR(imx_usb->dev, + "<%s> set config req is pending, NACK the host\n", + __func__); + return; + } + + if (u.r.bRequestType & USB_DIR_IN) + ep0_chg_stat(__func__, imx_usb, EP0_IN_DATA_PHASE); + else + ep0_chg_stat(__func__, imx_usb, EP0_OUT_DATA_PHASE); + + i = imx_usb->driver->setup(&imx_usb->gadget, &u.r); + if (i < 0) { + D_ERR(imx_usb->dev, "<%s> device setup error %d\n", + __func__, i); + goto stall; + } + + return; +stall: + D_ERR(imx_usb->dev, "<%s> protocol STALL\n", __func__); + imx_ep_stall(imx_ep); + ep0_chg_stat(__func__, imx_usb, EP0_STALL); + return; +} + +/******************************************************************************* + * USB gadget callback functions + ******************************************************************************* + */ + +static int imx_ep_enable(struct usb_ep *usb_ep, + const struct usb_endpoint_descriptor *desc) +{ + struct imx_ep_struct *imx_ep = container_of(usb_ep, + struct imx_ep_struct, ep); + struct imx_udc_struct *imx_usb = imx_ep->imx_usb; + unsigned long flags; + + if (!usb_ep + || !desc + || !EP_NO(imx_ep) + || desc->bDescriptorType != USB_DT_ENDPOINT + || imx_ep->bEndpointAddress != desc->bEndpointAddress) { + D_ERR(imx_usb->dev, + "<%s> bad ep or descriptor\n", __func__); + return -EINVAL; + } + + if (imx_ep->bmAttributes != desc->bmAttributes) { + D_ERR(imx_usb->dev, + "<%s> %s type mismatch\n", __func__, usb_ep->name); + return -EINVAL; + } + + if (imx_ep->fifosize < le16_to_cpu(desc->wMaxPacketSize)) { + D_ERR(imx_usb->dev, + "<%s> bad %s maxpacket\n", __func__, usb_ep->name); + return -ERANGE; + } + + if (!imx_usb->driver || imx_usb->gadget.speed == USB_SPEED_UNKNOWN) { + D_ERR(imx_usb->dev, "<%s> bogus device state\n", __func__); + return -ESHUTDOWN; + } + + local_irq_save(flags); + + imx_ep->stopped = 0; + imx_flush(imx_ep); + imx_ep_irq_enable(imx_ep); + + local_irq_restore(flags); + + D_EPX(imx_usb->dev, "<%s> ENABLED %s\n", __func__, usb_ep->name); + return 0; +} + +static int imx_ep_disable(struct usb_ep *usb_ep) +{ + struct imx_ep_struct *imx_ep = container_of(usb_ep, + struct imx_ep_struct, ep); + unsigned long flags; + + if (!usb_ep || !EP_NO(imx_ep) || !list_empty(&imx_ep->queue)) { + D_ERR(imx_ep->imx_usb->dev, "<%s> %s can not be disabled\n", + __func__, usb_ep ? imx_ep->ep.name : NULL); + return -EINVAL; + } + + local_irq_save(flags); + + imx_ep->stopped = 1; + nuke(imx_ep, -ESHUTDOWN); + imx_flush(imx_ep); + imx_ep_irq_disable(imx_ep); + + local_irq_restore(flags); + + D_EPX(imx_ep->imx_usb->dev, + "<%s> DISABLED %s\n", __func__, usb_ep->name); + return 0; +} + +static struct usb_request *imx_ep_alloc_request + (struct usb_ep *usb_ep, gfp_t gfp_flags) +{ + struct imx_request *req; + + req = kzalloc(sizeof *req, gfp_flags); + if (!req || !usb_ep) + return 0; + + INIT_LIST_HEAD(&req->queue); + req->in_use = 0; + + return &req->req; +} + +static void imx_ep_free_request + (struct usb_ep *usb_ep, struct usb_request *usb_req) +{ + struct imx_request *req; + + req = container_of(usb_req, struct imx_request, req); + WARN_ON(!list_empty(&req->queue)); + kfree(req); +} + +static int imx_ep_queue + (struct usb_ep *usb_ep, struct usb_request *usb_req, gfp_t gfp_flags) +{ + struct imx_ep_struct *imx_ep; + struct imx_udc_struct *imx_usb; + struct imx_request *req; + unsigned long flags; + int ret = 0; + + imx_ep = container_of(usb_ep, struct imx_ep_struct, ep); + imx_usb = imx_ep->imx_usb; + req = container_of(usb_req, struct imx_request, req); + + /* + Special care on IMX udc. + Ignore enqueue when after set configuration from the + host. This assume all gadget drivers reply set + configuration with the next ep0 req enqueue. + */ + if (imx_usb->set_config && !EP_NO(imx_ep)) { + imx_usb->set_config = 0; + D_EPX(imx_usb->dev, + "<%s> gadget reply set config\n", __func__); + return 0; + } + + if (unlikely(!usb_req || !req || !usb_req->complete || !usb_req->buf)) { + D_ERR(imx_usb->dev, "<%s> bad params\n", __func__); + return -EINVAL; + } + + if (unlikely(!usb_ep || !imx_ep)) { + D_ERR(imx_usb->dev, "<%s> bad ep\n", __func__); + return -EINVAL; + } + + if (!imx_usb->driver || imx_usb->gadget.speed == USB_SPEED_UNKNOWN) { + D_ERR(imx_usb->dev, "<%s> bogus device state\n", __func__); + return -ESHUTDOWN; + } + + local_irq_save(flags); + + /* Debug */ + D_REQ(imx_usb->dev, "<%s> ep%d %s request for [%d] bytes\n", + __func__, EP_NO(imx_ep), + ((!EP_NO(imx_ep) && imx_ep->imx_usb->ep0state == EP0_IN_DATA_PHASE) + || (EP_NO(imx_ep) && EP_DIR(imx_ep))) ? "IN" : "OUT", usb_req->length); + dump_req(__func__, imx_ep, usb_req); + + if (imx_ep->stopped) { + usb_req->status = -ESHUTDOWN; + ret = -ESHUTDOWN; + goto out; + } + + if (req->in_use) { + D_ERR(imx_usb->dev, + "<%s> refusing to queue req %p (already queued)\n", + __func__, req); + goto out; + } + + usb_req->status = -EINPROGRESS; + usb_req->actual = 0; + + ep_add_request(imx_ep, req); + + if (!EP_NO(imx_ep)) + ret = handle_ep0(imx_ep); + else + ret = handle_ep(imx_ep); +out: + local_irq_restore(flags); + return ret; +} + +static int imx_ep_dequeue(struct usb_ep *usb_ep, struct usb_request *usb_req) +{ + + struct imx_ep_struct *imx_ep = container_of + (usb_ep, struct imx_ep_struct, ep); + struct imx_request *req; + unsigned long flags; + + if (unlikely(!usb_ep || !EP_NO(imx_ep))) { + D_ERR(imx_ep->imx_usb->dev, "<%s> bad ep\n", __func__); + return -EINVAL; + } + + local_irq_save(flags); + + /* make sure it's actually queued on this endpoint */ + list_for_each_entry(req, &imx_ep->queue, queue) { + if (&req->req == usb_req) + break; + } + if (&req->req != usb_req) { + local_irq_restore(flags); + return -EINVAL; + } + + done(imx_ep, req, -ECONNRESET); + + local_irq_restore(flags); + return 0; +} + +static int imx_ep_set_halt(struct usb_ep *usb_ep, int value) +{ + struct imx_ep_struct *imx_ep = container_of + (usb_ep, struct imx_ep_struct, ep); + unsigned long flags; + + if (unlikely(!usb_ep || !EP_NO(imx_ep))) { + D_ERR(imx_ep->imx_usb->dev, "<%s> bad ep\n", __func__); + return -EINVAL; + } + + local_irq_save(flags); + + if ((imx_ep->bEndpointAddress & USB_DIR_IN) + && !list_empty(&imx_ep->queue)) { + local_irq_restore(flags); + return -EAGAIN; + } + + imx_ep_stall(imx_ep); + + local_irq_restore(flags); + + D_EPX(imx_ep->imx_usb->dev, "<%s> %s halt\n", __func__, usb_ep->name); + return 0; +} + +static int imx_ep_fifo_status(struct usb_ep *usb_ep) +{ + struct imx_ep_struct *imx_ep = container_of + (usb_ep, struct imx_ep_struct, ep); + + if (!usb_ep) { + D_ERR(imx_ep->imx_usb->dev, "<%s> bad ep\n", __func__); + return -ENODEV; + } + + if (imx_ep->imx_usb->gadget.speed == USB_SPEED_UNKNOWN) + return 0; + else + return imx_fifo_bcount(imx_ep); +} + +static void imx_ep_fifo_flush(struct usb_ep *usb_ep) +{ + struct imx_ep_struct *imx_ep = container_of + (usb_ep, struct imx_ep_struct, ep); + unsigned long flags; + + local_irq_save(flags); + + if (!usb_ep || !EP_NO(imx_ep) || !list_empty(&imx_ep->queue)) { + D_ERR(imx_ep->imx_usb->dev, "<%s> bad ep\n", __func__); + local_irq_restore(flags); + return; + } + + /* toggle and halt bits stay unchanged */ + imx_flush(imx_ep); + + local_irq_restore(flags); +} + +static struct usb_ep_ops imx_ep_ops = { + .enable = imx_ep_enable, + .disable = imx_ep_disable, + + .alloc_request = imx_ep_alloc_request, + .free_request = imx_ep_free_request, + + .queue = imx_ep_queue, + .dequeue = imx_ep_dequeue, + + .set_halt = imx_ep_set_halt, + .fifo_status = imx_ep_fifo_status, + .fifo_flush = imx_ep_fifo_flush, +}; + +/******************************************************************************* + * USB endpoint control functions + ******************************************************************************* + */ + +void ep0_chg_stat(const char *label, + struct imx_udc_struct *imx_usb, enum ep0_state stat) +{ + D_EP0(imx_usb->dev, "<%s> from %15s to %15s\n", + label, state_name[imx_usb->ep0state], state_name[stat]); + + if (imx_usb->ep0state == stat) + return; + + imx_usb->ep0state = stat; +} + +static void usb_init_data(struct imx_udc_struct *imx_usb) +{ + struct imx_ep_struct *imx_ep; + u8 i; + + /* device/ep0 records init */ + INIT_LIST_HEAD(&imx_usb->gadget.ep_list); + INIT_LIST_HEAD(&imx_usb->gadget.ep0->ep_list); + ep0_chg_stat(__func__, imx_usb, EP0_IDLE); + + /* basic endpoint records init */ + for (i = 0; i < IMX_USB_NB_EP; i++) { + imx_ep = &imx_usb->imx_ep[i]; + + if (i) { + list_add_tail(&imx_ep->ep.ep_list, + &imx_usb->gadget.ep_list); + imx_ep->stopped = 1; + } else + imx_ep->stopped = 0; + + INIT_LIST_HEAD(&imx_ep->queue); + } +} + +static void udc_stop_activity(struct imx_udc_struct *imx_usb, + struct usb_gadget_driver *driver) +{ + struct imx_ep_struct *imx_ep; + int i; + + if (imx_usb->gadget.speed == USB_SPEED_UNKNOWN) + driver = NULL; + + /* prevent new request submissions, kill any outstanding requests */ + for (i = 1; i < IMX_USB_NB_EP; i++) { + imx_ep = &imx_usb->imx_ep[i]; + imx_flush(imx_ep); + imx_ep->stopped = 1; + imx_ep_irq_disable(imx_ep); + nuke(imx_ep, -ESHUTDOWN); + } + + imx_usb->cfg = 0; + imx_usb->intf = 0; + imx_usb->alt = 0; + + if (driver) + driver->disconnect(&imx_usb->gadget); +} + +/******************************************************************************* + * Interrupt handlers + ******************************************************************************* + */ + +static irqreturn_t imx_udc_irq(int irq, void *dev) +{ + struct imx_udc_struct *imx_usb = dev; + struct usb_ctrlrequest u; + int temp, cfg, intf, alt; + int intr = __raw_readl(imx_usb->base + USB_INTR); + + if (intr & (INTR_WAKEUP | INTR_SUSPEND | INTR_RESUME | INTR_RESET_START + | INTR_RESET_STOP | INTR_CFG_CHG)) { + dump_intr(__func__, intr, imx_usb->dev); + dump_usb_stat(__func__, imx_usb); + } + + if (!imx_usb->driver) { + /*imx_udc_disable(imx_usb);*/ + goto end_irq; + } + + if (intr & INTR_WAKEUP) { + if (imx_usb->gadget.speed == USB_SPEED_UNKNOWN + && imx_usb->driver && imx_usb->driver->resume) + imx_usb->driver->resume(&imx_usb->gadget); + imx_usb->set_config = 0; + imx_usb->gadget.speed = USB_SPEED_FULL; + } + + if (intr & INTR_SUSPEND) { + if (imx_usb->gadget.speed != USB_SPEED_UNKNOWN + && imx_usb->driver && imx_usb->driver->suspend) + imx_usb->driver->suspend(&imx_usb->gadget); + imx_usb->set_config = 0; + imx_usb->gadget.speed = USB_SPEED_UNKNOWN; + } + + if (intr & INTR_RESET_START) { + __raw_writel(intr, imx_usb->base + USB_INTR); + udc_stop_activity(imx_usb, imx_usb->driver); + imx_usb->set_config = 0; + imx_usb->gadget.speed = USB_SPEED_UNKNOWN; + } + + if (intr & INTR_RESET_STOP) + imx_usb->gadget.speed = USB_SPEED_FULL; + + if (intr & INTR_CFG_CHG) { + __raw_writel(INTR_CFG_CHG, imx_usb->base + USB_INTR); + temp = __raw_readl(imx_usb->base + USB_STAT); + cfg = (temp & STAT_CFG) >> 5; + intf = (temp & STAT_INTF) >> 3; + alt = temp & STAT_ALTSET; + + D_REQ(imx_usb->dev, + "<%s> orig config C=%d, I=%d, A=%d / " + "req config C=%d, I=%d, A=%d\n", + __func__, imx_usb->cfg, imx_usb->intf, imx_usb->alt, + cfg, intf, alt); + + if (cfg != 1 && cfg != 2) + goto end_irq; + + imx_usb->set_config = 0; + + /* Config setup */ + if (imx_usb->cfg != cfg) { + D_REQ(imx_usb->dev, "<%s> Change config start\n",__func__); + u.bRequest = USB_REQ_SET_CONFIGURATION; + u.bRequestType = USB_DIR_OUT | + USB_TYPE_STANDARD | + USB_RECIP_DEVICE; + u.wValue = cfg; + u.wIndex = 0; + u.wLength = 0; + imx_usb->cfg = cfg; + imx_usb->set_config = 1; + imx_usb->driver->setup(&imx_usb->gadget, &u); + imx_usb->set_config = 0; + D_REQ(imx_usb->dev, "<%s> Change config done\n",__func__); + + } + if (imx_usb->intf != intf || imx_usb->alt != alt) { + D_REQ(imx_usb->dev, "<%s> Change interface start\n",__func__); + u.bRequest = USB_REQ_SET_INTERFACE; + u.bRequestType = USB_DIR_OUT | + USB_TYPE_STANDARD | + USB_RECIP_INTERFACE; + u.wValue = alt; + u.wIndex = intf; + u.wLength = 0; + imx_usb->intf = intf; + imx_usb->alt = alt; + imx_usb->set_config = 1; + imx_usb->driver->setup(&imx_usb->gadget, &u); + imx_usb->set_config = 0; + D_REQ(imx_usb->dev, "<%s> Change interface done\n",__func__); + } + } + + if (intr & INTR_SOF) { + if (imx_usb->ep0state == EP0_IDLE) { + temp = __raw_readl(imx_usb->base + USB_CTRL); + __raw_writel(temp | CTRL_CMDOVER, imx_usb->base + USB_CTRL); + } + } + +end_irq: + __raw_writel(intr, imx_usb->base + USB_INTR); + return IRQ_HANDLED; +} + +static irqreturn_t imx_udc_ctrl_irq(int irq, void *dev) +{ + struct imx_udc_struct *imx_usb = dev; + int intr = __raw_readl(imx_usb->base + USB_EP_INTR(0)); + + dump_ep_intr(__func__, 0, intr, imx_usb->dev); + + if (!imx_usb->driver) { + __raw_writel(intr, imx_usb->base + USB_EP_INTR(0)); + return IRQ_HANDLED; + } + + /* DEVREQ IRQ has highest priority */ + if (intr & (EPINTR_DEVREQ | EPINTR_MDEVREQ)) + handle_ep0_devreq(imx_usb); + /* Seem i.MX is missing EOF interrupt sometimes. + * Therefore we monitor both EOF and FIFO_EMPTY interrups + * when transmiting, and both EOF and FIFO_FULL when + * receiving data. + */ + else if (intr & (EPINTR_EOF | EPINTR_FIFO_EMPTY | EPINTR_FIFO_FULL)) + handle_ep0(&imx_usb->imx_ep[0]); + + __raw_writel(intr, imx_usb->base + USB_EP_INTR(0)); + + return IRQ_HANDLED; +} + +static irqreturn_t imx_udc_bulk_irq(int irq, void *dev) +{ + struct imx_udc_struct *imx_usb = dev; + struct imx_ep_struct *imx_ep = &imx_usb->imx_ep[irq - USBD_INT0]; + int intr = __raw_readl(imx_usb->base + USB_EP_INTR(EP_NO(imx_ep))); + + dump_ep_intr(__func__, irq - USBD_INT0, intr, imx_usb->dev); + + if (!imx_usb->driver) { + __raw_writel(intr, imx_usb->base + USB_EP_INTR(EP_NO(imx_ep))); + return IRQ_HANDLED; + } + + handle_ep(imx_ep); + + __raw_writel(intr, imx_usb->base + USB_EP_INTR(EP_NO(imx_ep))); + + return IRQ_HANDLED; +} + +irq_handler_t intr_handler(int i) +{ + switch (i) { + case 0: + return imx_udc_ctrl_irq; + case 1: + case 2: + case 3: + case 4: + case 5: + return imx_udc_bulk_irq; + default: + return imx_udc_irq; + } +} + +/******************************************************************************* + * Static defined IMX UDC structure + ******************************************************************************* + */ + +static const struct usb_gadget_ops imx_udc_ops = { + .get_frame = imx_udc_get_frame, + .wakeup = imx_udc_wakeup, +}; + +static struct imx_udc_struct controller = { + .gadget = { + .ops = &imx_udc_ops, + .ep0 = &controller.imx_ep[0].ep, + .name = driver_name, + .dev = { + .bus_id = "gadget", + }, + }, + + .imx_ep[0] = { + .ep = { + .name = ep0name, + .ops = &imx_ep_ops, + .maxpacket = 32, + }, + .imx_usb = &controller, + .fifosize = 32, + .bEndpointAddress = 0, + .bmAttributes = USB_ENDPOINT_XFER_CONTROL, + }, + .imx_ep[1] = { + .ep = { + .name = "ep1in-bulk", + .ops = &imx_ep_ops, + .maxpacket = 64, + }, + .imx_usb = &controller, + .fifosize = 64, + .bEndpointAddress = USB_DIR_IN | 1, + .bmAttributes = USB_ENDPOINT_XFER_BULK, + }, + .imx_ep[2] = { + .ep = { + .name = "ep2out-bulk", + .ops = &imx_ep_ops, + .maxpacket = 64, + }, + .imx_usb = &controller, + .fifosize = 64, + .bEndpointAddress = USB_DIR_OUT | 2, + .bmAttributes = USB_ENDPOINT_XFER_BULK, + }, + .imx_ep[3] = { + .ep = { + .name = "ep3out-bulk", + .ops = &imx_ep_ops, + .maxpacket = 32, + }, + .imx_usb = &controller, + .fifosize = 32, + .bEndpointAddress = USB_DIR_OUT | 3, + .bmAttributes = USB_ENDPOINT_XFER_BULK, + }, + .imx_ep[4] = { + .ep = { + .name = "ep4in-int", + .ops = &imx_ep_ops, + .maxpacket = 32, + }, + .imx_usb = &controller, + .fifosize = 32, + .bEndpointAddress = USB_DIR_IN | 4, + .bmAttributes = USB_ENDPOINT_XFER_INT, + }, + .imx_ep[5] = { + .ep = { + .name = "ep5out-int", + .ops = &imx_ep_ops, + .maxpacket = 32, + }, + .imx_usb = &controller, + .fifosize = 32, + .bEndpointAddress = USB_DIR_OUT | 5, + .bmAttributes = USB_ENDPOINT_XFER_INT, + }, +}; + +/******************************************************************************* + * USB gadged driver functions + ******************************************************************************* + */ +int usb_gadget_register_driver(struct usb_gadget_driver *driver) +{ + struct imx_udc_struct *imx_usb = &controller; + int retval; + + if (!driver + || driver->speed < USB_SPEED_FULL + || !driver->bind + || !driver->disconnect + || !driver->setup) + return -EINVAL; + if (!imx_usb) + return -ENODEV; + if (imx_usb->driver) + return -EBUSY; + + /* first hook up the driver ... */ + imx_usb->driver = driver; + imx_usb->gadget.dev.driver = &driver->driver; + + retval = device_add(&imx_usb->gadget.dev); + if (retval) + goto fail; + retval = driver->bind(&imx_usb->gadget); + if (retval) { + D_ERR(imx_usb->dev, "<%s> bind to driver %s --> error %d\n", + __func__, driver->driver.name, retval); + device_del(&imx_usb->gadget.dev); + + goto fail; + } + + D_INI(imx_usb->dev, "<%s> registered gadget driver '%s'\n", + __func__, driver->driver.name); + + imx_udc_enable(imx_usb); + + return 0; +fail: + imx_usb->driver = NULL; + imx_usb->gadget.dev.driver = NULL; + return retval; +} +EXPORT_SYMBOL(usb_gadget_register_driver); + +int usb_gadget_unregister_driver(struct usb_gadget_driver *driver) +{ + struct imx_udc_struct *imx_usb = &controller; + + if (!imx_usb) + return -ENODEV; + if (!driver || driver != imx_usb->driver || !driver->unbind) + return -EINVAL; + + udc_stop_activity(imx_usb, driver); + imx_udc_disable(imx_usb); + + driver->unbind(&imx_usb->gadget); + imx_usb->gadget.dev.driver = NULL; + imx_usb->driver = NULL; + + device_del(&imx_usb->gadget.dev); + + D_INI(imx_usb->dev, "<%s> unregistered gadget driver '%s'\n", + __func__, driver->driver.name); + + return 0; +} +EXPORT_SYMBOL(usb_gadget_unregister_driver); + +/******************************************************************************* + * Module functions + ******************************************************************************* + */ + +static int __init imx_udc_probe(struct platform_device *pdev) +{ + struct imx_udc_struct *imx_usb = &controller; + struct resource *res; + struct imxusb_platform_data *pdata; + struct clk *clk; + void __iomem *base; + int ret = 0; + int i, res_size; + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!res) { + dev_err(&pdev->dev, "can't get device resources\n"); + return -ENODEV; + } + + pdata = pdev->dev.platform_data; + if (!pdata) { + dev_err(&pdev->dev, "driver needs platform data\n"); + return -ENODEV; + } + + res_size = res->end - res->start + 1; + if (!request_mem_region(res->start, res_size, res->name)) { + dev_err(&pdev->dev, "can't allocate %d bytes at %d address\n", + res_size, res->start); + return -ENOMEM; + } + + if (pdata->init) { + ret = pdata->init(&pdev->dev); + if (ret) + goto fail0; + } + + base = ioremap(res->start, res_size); + if (!base) { + dev_err(&pdev->dev, "ioremap failed\n"); + ret = -EIO; + goto fail1; + } + + clk = clk_get(NULL, "usbd_clk"); + if (IS_ERR(clk)) { + ret = PTR_ERR(clk); + dev_err(&pdev->dev, "can't get USB clock\n"); + goto fail2; + } + clk_enable(clk); + + if (clk_get_rate(clk) != 48000000) { + D_INI(&pdev->dev, + "Bad USB clock (%d Hz), changing to 48000000 Hz\n", + (int)clk_get_rate(clk)); + if (clk_set_rate(clk, 48000000)) { + dev_err(&pdev->dev, + "Unable to set correct USB clock (48MHz)\n"); + ret = -EIO; + goto fail3; + } + } + + for (i = 0; i < IMX_USB_NB_EP + 1; i++) { + imx_usb->usbd_int[i] = platform_get_irq(pdev, i); + if (imx_usb->usbd_int[i] < 0) { + dev_err(&pdev->dev, "can't get irq number\n"); + ret = -ENODEV; + goto fail3; + } + } + + for (i = 0; i < IMX_USB_NB_EP + 1; i++) { + ret = request_irq(imx_usb->usbd_int[i], intr_handler(i), + IRQF_DISABLED, driver_name, imx_usb); + if (ret) { + dev_err(&pdev->dev, "can't get irq %i, err %d\n", + imx_usb->usbd_int[i], ret); + for (--i; i >= 0; i--) + free_irq(imx_usb->usbd_int[i], imx_usb); + goto fail3; + } + } + + imx_usb->res = res; + imx_usb->base = base; + imx_usb->clk = clk; + imx_usb->dev = &pdev->dev; + + device_initialize(&imx_usb->gadget.dev); + + imx_usb->gadget.dev.parent = &pdev->dev; + imx_usb->gadget.dev.dma_mask = pdev->dev.dma_mask; + + platform_set_drvdata(pdev, imx_usb); + + usb_init_data(imx_usb); + imx_udc_init(imx_usb); + + return 0; + +fail3: + clk_put(clk); + clk_disable(clk); +fail2: + iounmap(base); +fail1: + if (pdata->exit) + pdata->exit(&pdev->dev); +fail0: + release_mem_region(res->start, res_size); + return ret; +} + +static int __exit imx_udc_remove(struct platform_device *pdev) +{ + struct imx_udc_struct *imx_usb = platform_get_drvdata(pdev); + struct imxusb_platform_data *pdata = pdev->dev.platform_data; + int i; + + imx_udc_disable(imx_usb); + + for (i = 0; i < IMX_USB_NB_EP + 1; i++) + free_irq(imx_usb->usbd_int[i], imx_usb); + + clk_put(imx_usb->clk); + clk_disable(imx_usb->clk); + iounmap(imx_usb->base); + + release_mem_region(imx_usb->res->start, + imx_usb->res->end - imx_usb->res->start + 1); + + if (pdata->exit) + pdata->exit(&pdev->dev); + + platform_set_drvdata(pdev, NULL); + + return 0; +} + +/*----------------------------------------------------------------------------*/ + +#ifdef CONFIG_PM +#define imx_udc_suspend NULL +#define imx_udc_resume NULL +#else +#define imx_udc_suspend NULL +#define imx_udc_resume NULL +#endif + +/*----------------------------------------------------------------------------*/ + +static struct platform_driver udc_driver = { + .driver = { + .name = driver_name, + .owner = THIS_MODULE, + }, + .remove = __exit_p(imx_udc_remove), + .suspend = imx_udc_suspend, + .resume = imx_udc_resume, +}; + +static int __init udc_init(void) +{ + return platform_driver_probe(&udc_driver, imx_udc_probe); +} +module_init(udc_init); + +static void __exit udc_exit(void) +{ + platform_driver_unregister(&udc_driver); +} +module_exit(udc_exit); + +MODULE_DESCRIPTION("IMX USB Device Controller driver"); +MODULE_AUTHOR("Darius Augulis "); +MODULE_LICENSE("GPL"); +MODULE_ALIAS("platform:imx_udc"); diff --git a/drivers/usb/gadget/imx_udc.h b/drivers/usb/gadget/imx_udc.h new file mode 100644 index 00000000000..850076937d8 --- /dev/null +++ b/drivers/usb/gadget/imx_udc.h @@ -0,0 +1,344 @@ +/* + * Copyright (C) 2005 Mike Lee(eemike@gmail.com) + * + * This udc driver is now under testing and code is based on pxa2xx_udc.h + * Please use it with your own risk! + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ + +#ifndef __LINUX_USB_GADGET_IMX_H +#define __LINUX_USB_GADGET_IMX_H + +#include + +/* Helper macros */ +#define EP_NO(ep) ((ep->bEndpointAddress) & ~USB_DIR_IN) /* IN:1, OUT:0 */ +#define EP_DIR(ep) ((ep->bEndpointAddress) & USB_DIR_IN ? 1 : 0) +#define irq_to_ep(irq) (((irq) >= USBD_INT0) || ((irq) <= USBD_INT6) ? ((irq) - USBD_INT0) : (USBD_INT6)) /*should not happen*/ +#define ep_to_irq(ep) (EP_NO((ep)) + USBD_INT0) +#define IMX_USB_NB_EP 6 + +/* Driver structures */ +struct imx_request { + struct usb_request req; + struct list_head queue; + unsigned int in_use; +}; + +enum ep0_state { + EP0_IDLE, + EP0_IN_DATA_PHASE, + EP0_OUT_DATA_PHASE, + EP0_CONFIG, + EP0_STALL, +}; + +struct imx_ep_struct { + struct usb_ep ep; + struct imx_udc_struct *imx_usb; + struct list_head queue; + unsigned char stopped; + unsigned char fifosize; + unsigned char bEndpointAddress; + unsigned char bmAttributes; +}; + +struct imx_udc_struct { + struct usb_gadget gadget; + struct usb_gadget_driver *driver; + struct device *dev; + struct imx_ep_struct imx_ep[IMX_USB_NB_EP]; + struct clk *clk; + enum ep0_state ep0state; + struct resource *res; + void __iomem *base; + unsigned char set_config; + int cfg, + intf, + alt, + usbd_int[7]; +}; + +/* USB registers */ +#define USB_FRAME (0x00) /* USB frame */ +#define USB_SPEC (0x04) /* USB Spec */ +#define USB_STAT (0x08) /* USB Status */ +#define USB_CTRL (0x0C) /* USB Control */ +#define USB_DADR (0x10) /* USB Desc RAM addr */ +#define USB_DDAT (0x14) /* USB Desc RAM/EP buffer data */ +#define USB_INTR (0x18) /* USB interrupt */ +#define USB_MASK (0x1C) /* USB Mask */ +#define USB_ENAB (0x24) /* USB Enable */ +#define USB_EP_STAT(x) (0x30 + (x*0x30)) /* USB status/control */ +#define USB_EP_INTR(x) (0x34 + (x*0x30)) /* USB interrupt */ +#define USB_EP_MASK(x) (0x38 + (x*0x30)) /* USB mask */ +#define USB_EP_FDAT(x) (0x3C + (x*0x30)) /* USB FIFO data */ +#define USB_EP_FDAT0(x) (0x3C + (x*0x30)) /* USB FIFO data */ +#define USB_EP_FDAT1(x) (0x3D + (x*0x30)) /* USB FIFO data */ +#define USB_EP_FDAT2(x) (0x3E + (x*0x30)) /* USB FIFO data */ +#define USB_EP_FDAT3(x) (0x3F + (x*0x30)) /* USB FIFO data */ +#define USB_EP_FSTAT(x) (0x40 + (x*0x30)) /* USB FIFO status */ +#define USB_EP_FCTRL(x) (0x44 + (x*0x30)) /* USB FIFO control */ +#define USB_EP_LRFP(x) (0x48 + (x*0x30)) /* USB last read frame pointer */ +#define USB_EP_LWFP(x) (0x4C + (x*0x30)) /* USB last write frame pointer */ +#define USB_EP_FALRM(x) (0x50 + (x*0x30)) /* USB FIFO alarm */ +#define USB_EP_FRDP(x) (0x54 + (x*0x30)) /* USB FIFO read pointer */ +#define USB_EP_FWRP(x) (0x58 + (x*0x30)) /* USB FIFO write pointer */ +/* USB Control Register Bit Fields.*/ +#define CTRL_CMDOVER (1<<6) /* UDC status */ +#define CTRL_CMDERROR (1<<5) /* UDC status */ +#define CTRL_FE_ENA (1<<3) /* Enable Font End logic */ +#define CTRL_UDC_RST (1<<2) /* UDC reset */ +#define CTRL_AFE_ENA (1<<1) /* Analog Font end enable */ +#define CTRL_RESUME (1<<0) /* UDC resume */ +/* USB Status Register Bit Fields.*/ +#define STAT_RST (1<<8) +#define STAT_SUSP (1<<7) +#define STAT_CFG (3<<5) +#define STAT_INTF (3<<3) +#define STAT_ALTSET (7<<0) +/* USB Interrupt Status/Mask Registers Bit fields */ +#define INTR_WAKEUP (1<<31) /* Wake up Interrupt */ +#define INTR_MSOF (1<<7) /* Missed Start of Frame */ +#define INTR_SOF (1<<6) /* Start of Frame */ +#define INTR_RESET_STOP (1<<5) /* Reset Signaling stop */ +#define INTR_RESET_START (1<<4) /* Reset Signaling start */ +#define INTR_RESUME (1<<3) /* Suspend to resume */ +#define INTR_SUSPEND (1<<2) /* Active to suspend */ +#define INTR_FRAME_MATCH (1<<1) /* Frame matched */ +#define INTR_CFG_CHG (1<<0) /* Configuration change occurred */ +/* USB Enable Register Bit Fields.*/ +#define ENAB_RST (1<<31) /* Reset USB modules */ +#define ENAB_ENAB (1<<30) /* Enable USB modules*/ +#define ENAB_SUSPEND (1<<29) /* Suspend USB modules */ +#define ENAB_ENDIAN (1<<28) /* Endian of USB modules */ +#define ENAB_PWRMD (1<<0) /* Power mode of USB modules */ +/* USB Descriptor Ram Address Register bit fields */ +#define DADR_CFG (1<<31) /* Configuration */ +#define DADR_BSY (1<<30) /* Busy status */ +#define DADR_DADR (0x1FF) /* Descriptor Ram Address */ +/* USB Descriptor RAM/Endpoint Buffer Data Register bit fields */ +#define DDAT_DDAT (0xFF) /* Descriptor Endpoint Buffer */ +/* USB Endpoint Status Register bit fields */ +#define EPSTAT_BCOUNT (0x7F<<16) /* Endpoint FIFO byte count */ +#define EPSTAT_SIP (1<<8) /* Endpoint setup in progress */ +#define EPSTAT_DIR (1<<7) /* Endpoint transfer direction */ +#define EPSTAT_MAX (3<<5) /* Endpoint Max packet size */ +#define EPSTAT_TYP (3<<3) /* Endpoint type */ +#define EPSTAT_ZLPS (1<<2) /* Send zero length packet */ +#define EPSTAT_FLUSH (1<<1) /* Endpoint FIFO Flush */ +#define EPSTAT_STALL (1<<0) /* Force stall */ +/* USB Endpoint FIFO Status Register bit fields */ +#define FSTAT_FRAME_STAT (0xF<<24) /* Frame status bit [0-3] */ +#define FSTAT_ERR (1<<22) /* FIFO error */ +#define FSTAT_UF (1<<21) /* FIFO underflow */ +#define FSTAT_OF (1<<20) /* FIFO overflow */ +#define FSTAT_FR (1<<19) /* FIFO frame ready */ +#define FSTAT_FULL (1<<18) /* FIFO full */ +#define FSTAT_ALRM (1<<17) /* FIFO alarm */ +#define FSTAT_EMPTY (1<<16) /* FIFO empty */ +/* USB Endpoint FIFO Control Register bit fields */ +#define FCTRL_WFR (1<<29) /* Write frame end */ +/* USB Endpoint Interrupt Status Regsiter bit fields */ +#define EPINTR_FIFO_FULL (1<<8) /* fifo full */ +#define EPINTR_FIFO_EMPTY (1<<7) /* fifo empty */ +#define EPINTR_FIFO_ERROR (1<<6) /* fifo error */ +#define EPINTR_FIFO_HIGH (1<<5) /* fifo high */ +#define EPINTR_FIFO_LOW (1<<4) /* fifo low */ +#define EPINTR_MDEVREQ (1<<3) /* multi Device request */ +#define EPINTR_EOT (1<<2) /* fifo end of transfer */ +#define EPINTR_DEVREQ (1<<1) /* Device request */ +#define EPINTR_EOF (1<<0) /* fifo end of frame */ + +/* Debug macros */ +#ifdef DEBUG + +/* #define DEBUG_REQ */ +/* #define DEBUG_TRX */ +/* #define DEBUG_INIT */ +/* #define DEBUG_EP0 */ +/* #define DEBUG_EPX */ +/* #define DEBUG_IRQ */ +/* #define DEBUG_EPIRQ */ +/* #define DEBUG_DUMP */ +#define DEBUG_ERR + +#ifdef DEBUG_REQ + #define D_REQ(dev, args...) dev_dbg(dev, ## args) +#else + #define D_REQ(dev, args...) do {} while (0) +#endif /* DEBUG_REQ */ + +#ifdef DEBUG_TRX + #define D_TRX(dev, args...) dev_dbg(dev, ## args) +#else + #define D_TRX(dev, args...) do {} while (0) +#endif /* DEBUG_TRX */ + +#ifdef DEBUG_INIT + #define D_INI(dev, args...) dev_dbg(dev, ## args) +#else + #define D_INI(dev, args...) do {} while (0) +#endif /* DEBUG_INIT */ + +#ifdef DEBUG_EP0 + static const char *state_name[] = { + "EP0_IDLE", + "EP0_IN_DATA_PHASE", + "EP0_OUT_DATA_PHASE", + "EP0_CONFIG", + "EP0_STALL" + }; + #define D_EP0(dev, args...) dev_dbg(dev, ## args) +#else + #define D_EP0(dev, args...) do {} while (0) +#endif /* DEBUG_EP0 */ + +#ifdef DEBUG_EPX + #define D_EPX(dev, args...) dev_dbg(dev, ## args) +#else + #define D_EPX(dev, args...) do {} while (0) +#endif /* DEBUG_EP0 */ + +#ifdef DEBUG_IRQ + static void dump_intr(const char *label, int irqreg, struct device *dev) + { + dev_dbg(dev, "<%s> USB_INTR=[%s%s%s%s%s%s%s%s%s]\n", label, + (irqreg & INTR_WAKEUP) ? " wake" : "", + (irqreg & INTR_MSOF) ? " msof" : "", + (irqreg & INTR_SOF) ? " sof" : "", + (irqreg & INTR_RESUME) ? " resume" : "", + (irqreg & INTR_SUSPEND) ? " suspend" : "", + (irqreg & INTR_RESET_STOP) ? " noreset" : "", + (irqreg & INTR_RESET_START) ? " reset" : "", + (irqreg & INTR_FRAME_MATCH) ? " fmatch" : "", + (irqreg & INTR_CFG_CHG) ? " config" : ""); + } +#else + #define dump_intr(x, y, z) do {} while (0) +#endif /* DEBUG_IRQ */ + +#ifdef DEBUG_EPIRQ + static void dump_ep_intr(const char *label, int nr, int irqreg, struct device *dev) + { + dev_dbg(dev, "<%s> EP%d_INTR=[%s%s%s%s%s%s%s%s%s]\n", label, nr, + (irqreg & EPINTR_FIFO_FULL) ? " full" : "", + (irqreg & EPINTR_FIFO_EMPTY) ? " fempty" : "", + (irqreg & EPINTR_FIFO_ERROR) ? " ferr" : "", + (irqreg & EPINTR_FIFO_HIGH) ? " fhigh" : "", + (irqreg & EPINTR_FIFO_LOW) ? " flow" : "", + (irqreg & EPINTR_MDEVREQ) ? " mreq" : "", + (irqreg & EPINTR_EOF) ? " eof" : "", + (irqreg & EPINTR_DEVREQ) ? " devreq" : "", + (irqreg & EPINTR_EOT) ? " eot" : ""); + } +#else + #define dump_ep_intr(x, y, z, i) do {} while (0) +#endif /* DEBUG_IRQ */ + +#ifdef DEBUG_DUMP + static void dump_usb_stat(const char *label, struct imx_udc_struct *imx_usb) + { + int temp = __raw_readl(imx_usb->base + USB_STAT); + + dev_dbg(imx_usb->dev, + "<%s> USB_STAT=[%s%s CFG=%d, INTF=%d, ALTR=%d]\n", label, + (temp & STAT_RST) ? " reset" : "", + (temp & STAT_SUSP) ? " suspend" : "", + (temp & STAT_CFG) >> 5, + (temp & STAT_INTF) >> 3, + (temp & STAT_ALTSET)); + } + + static void dump_ep_stat(const char *label, struct imx_ep_struct *imx_ep) + { + int temp = __raw_readl(imx_ep->imx_usb->base + USB_EP_INTR(EP_NO(imx_ep))); + + dev_dbg(imx_ep->imx_usb->dev, + "<%s> EP%d_INTR=[%s%s%s%s%s%s%s%s%s]\n", label, EP_NO(imx_ep), + (temp & EPINTR_FIFO_FULL) ? " full" : "", + (temp & EPINTR_FIFO_EMPTY) ? " fempty" : "", + (temp & EPINTR_FIFO_ERROR) ? " ferr" : "", + (temp & EPINTR_FIFO_HIGH) ? " fhigh" : "", + (temp & EPINTR_FIFO_LOW) ? " flow" : "", + (temp & EPINTR_MDEVREQ) ? " mreq" : "", + (temp & EPINTR_EOF) ? " eof" : "", + (temp & EPINTR_DEVREQ) ? " devreq" : "", + (temp & EPINTR_EOT) ? " eot" : ""); + + temp = __raw_readl(imx_ep->imx_usb->base + USB_EP_STAT(EP_NO(imx_ep))); + + dev_dbg(imx_ep->imx_usb->dev, + "<%s> EP%d_STAT=[%s%s bcount=%d]\n", label, EP_NO(imx_ep), + (temp & EPSTAT_SIP) ? " sip" : "", + (temp & EPSTAT_STALL) ? " stall" : "", + (temp & EPSTAT_BCOUNT) >> 16); + + temp = __raw_readl(imx_ep->imx_usb->base + USB_EP_FSTAT(EP_NO(imx_ep))); + + dev_dbg(imx_ep->imx_usb->dev, + "<%s> EP%d_FSTAT=[%s%s%s%s%s%s%s]\n", label, EP_NO(imx_ep), + (temp & FSTAT_ERR) ? " ferr" : "", + (temp & FSTAT_UF) ? " funder" : "", + (temp & FSTAT_OF) ? " fover" : "", + (temp & FSTAT_FR) ? " fready" : "", + (temp & FSTAT_FULL) ? " ffull" : "", + (temp & FSTAT_ALRM) ? " falarm" : "", + (temp & FSTAT_EMPTY) ? " fempty" : ""); + } + + static void dump_req(const char *label, struct imx_ep_struct *imx_ep, struct usb_request *req) + { + int i; + + if (!req || !req->buf) { + dev_dbg(imx_ep->imx_usb->dev, "<%s> req or req buf is free\n", label); + return; + } + + if ((!EP_NO(imx_ep) && imx_ep->imx_usb->ep0state == EP0_IN_DATA_PHASE) + || (EP_NO(imx_ep) && EP_DIR(imx_ep))) { + + dev_dbg(imx_ep->imx_usb->dev, "<%s> request dump <", label); + for (i = 0; i < req->length; i++) + printk("%02x-", *((u8 *)req->buf + i)); + printk(">\n"); + } + } + +#else + #define dump_ep_stat(x, y) do {} while (0) + #define dump_usb_stat(x, y) do {} while (0) + #define dump_req(x, y, z) do {} while (0) +#endif /* DEBUG_DUMP */ + +#ifdef DEBUG_ERR + #define D_ERR(dev, args...) dev_dbg(dev, ## args) +#else + #define D_ERR(dev, args...) do {} while (0) +#endif + +#else + #define D_REQ(dev, args...) do {} while (0) + #define D_TRX(dev, args...) do {} while (0) + #define D_INI(dev, args...) do {} while (0) + #define D_EP0(dev, args...) do {} while (0) + #define D_EPX(dev, args...) do {} while (0) + #define dump_ep_intr(x, y, z, i) do {} while (0) + #define dump_intr(x, y, z) do {} while (0) + #define dump_ep_stat(x, y) do {} while (0) + #define dump_usb_stat(x, y) do {} while (0) + #define dump_req(x, y, z) do {} while (0) + #define D_ERR(dev, args...) do {} while (0) +#endif /* DEBUG */ + +#endif /* __LINUX_USB_GADGET_IMX_H */ From 9ebd9616648bc0e47e7f8e1898c919305f1e6347 Mon Sep 17 00:00:00 2001 From: David Brownell Date: Thu, 18 Dec 2008 12:25:44 -0800 Subject: [PATCH 080/123] USB: otg: twl4030 transceiver driver Add driver for the high speed USB-OTG transceiver in TI's TWL4030 family of chips. Given this and various other pending patches, OMAP3 hardware like that from beagleboard.org, gumstix.com (Overo), and openpandora.org should now have basic USB host and peripheral connectivity with mainline kernels. Ditto for less widely-available boards. Signed-off-by: David Brownell Signed-off-by: Tony Lindgren Cc: Kevin Hilman Cc: Felipe Balbi Signed-off-by: Greg Kroah-Hartman --- drivers/usb/otg/Kconfig | 10 + drivers/usb/otg/Makefile | 1 + drivers/usb/otg/twl4030-usb.c | 721 ++++++++++++++++++++++++++++++++++ 3 files changed, 732 insertions(+) create mode 100644 drivers/usb/otg/twl4030-usb.c diff --git a/drivers/usb/otg/Kconfig b/drivers/usb/otg/Kconfig index afe91bfea7f..8e8dbdb9b39 100644 --- a/drivers/usb/otg/Kconfig +++ b/drivers/usb/otg/Kconfig @@ -41,4 +41,14 @@ config ISP1301_OMAP This driver can also be built as a module. If so, the module will be called isp1301_omap. +config TWL4030_USB + tristate "TWL4030 USB Transceiver Driver" + depends on TWL4030_CORE + select USB_OTG_UTILS + help + Enable this to support the USB OTG transceiver on TWL4030 + family chips (including the TWL5030 and TPS659x0 devices). + This transceiver supports high and full speed devices plus, + in host mode, low speed. + endif # USB || OTG diff --git a/drivers/usb/otg/Makefile b/drivers/usb/otg/Makefile index 7c80fc379e6..d73c7cf5e2f 100644 --- a/drivers/usb/otg/Makefile +++ b/drivers/usb/otg/Makefile @@ -8,6 +8,7 @@ obj-$(CONFIG_USB_OTG_UTILS) += otg.o # transceiver drivers obj-$(CONFIG_USB_GPIO_VBUS) += gpio_vbus.o obj-$(CONFIG_ISP1301_OMAP) += isp1301_omap.o +obj-$(CONFIG_TWL4030_USB) += twl4030-usb.o ccflags-$(CONFIG_USB_DEBUG) += -DDEBUG ccflags-$(CONFIG_USB_GADGET_DEBUG) += -DDEBUG diff --git a/drivers/usb/otg/twl4030-usb.c b/drivers/usb/otg/twl4030-usb.c new file mode 100644 index 00000000000..416e4410be0 --- /dev/null +++ b/drivers/usb/otg/twl4030-usb.c @@ -0,0 +1,721 @@ +/* + * twl4030_usb - TWL4030 USB transceiver, talking to OMAP OTG controller + * + * Copyright (C) 2004-2007 Texas Instruments + * Copyright (C) 2008 Nokia Corporation + * Contact: Felipe Balbi + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Current status: + * - HS USB ULPI mode works. + * - 3-pin mode support may be added in future. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + + +/* Register defines */ + +#define VENDOR_ID_LO 0x00 +#define VENDOR_ID_HI 0x01 +#define PRODUCT_ID_LO 0x02 +#define PRODUCT_ID_HI 0x03 + +#define FUNC_CTRL 0x04 +#define FUNC_CTRL_SET 0x05 +#define FUNC_CTRL_CLR 0x06 +#define FUNC_CTRL_SUSPENDM (1 << 6) +#define FUNC_CTRL_RESET (1 << 5) +#define FUNC_CTRL_OPMODE_MASK (3 << 3) /* bits 3 and 4 */ +#define FUNC_CTRL_OPMODE_NORMAL (0 << 3) +#define FUNC_CTRL_OPMODE_NONDRIVING (1 << 3) +#define FUNC_CTRL_OPMODE_DISABLE_BIT_NRZI (2 << 3) +#define FUNC_CTRL_TERMSELECT (1 << 2) +#define FUNC_CTRL_XCVRSELECT_MASK (3 << 0) /* bits 0 and 1 */ +#define FUNC_CTRL_XCVRSELECT_HS (0 << 0) +#define FUNC_CTRL_XCVRSELECT_FS (1 << 0) +#define FUNC_CTRL_XCVRSELECT_LS (2 << 0) +#define FUNC_CTRL_XCVRSELECT_FS4LS (3 << 0) + +#define IFC_CTRL 0x07 +#define IFC_CTRL_SET 0x08 +#define IFC_CTRL_CLR 0x09 +#define IFC_CTRL_INTERFACE_PROTECT_DISABLE (1 << 7) +#define IFC_CTRL_AUTORESUME (1 << 4) +#define IFC_CTRL_CLOCKSUSPENDM (1 << 3) +#define IFC_CTRL_CARKITMODE (1 << 2) +#define IFC_CTRL_FSLSSERIALMODE_3PIN (1 << 1) + +#define TWL4030_OTG_CTRL 0x0A +#define TWL4030_OTG_CTRL_SET 0x0B +#define TWL4030_OTG_CTRL_CLR 0x0C +#define TWL4030_OTG_CTRL_DRVVBUS (1 << 5) +#define TWL4030_OTG_CTRL_CHRGVBUS (1 << 4) +#define TWL4030_OTG_CTRL_DISCHRGVBUS (1 << 3) +#define TWL4030_OTG_CTRL_DMPULLDOWN (1 << 2) +#define TWL4030_OTG_CTRL_DPPULLDOWN (1 << 1) +#define TWL4030_OTG_CTRL_IDPULLUP (1 << 0) + +#define USB_INT_EN_RISE 0x0D +#define USB_INT_EN_RISE_SET 0x0E +#define USB_INT_EN_RISE_CLR 0x0F +#define USB_INT_EN_FALL 0x10 +#define USB_INT_EN_FALL_SET 0x11 +#define USB_INT_EN_FALL_CLR 0x12 +#define USB_INT_STS 0x13 +#define USB_INT_LATCH 0x14 +#define USB_INT_IDGND (1 << 4) +#define USB_INT_SESSEND (1 << 3) +#define USB_INT_SESSVALID (1 << 2) +#define USB_INT_VBUSVALID (1 << 1) +#define USB_INT_HOSTDISCONNECT (1 << 0) + +#define CARKIT_CTRL 0x19 +#define CARKIT_CTRL_SET 0x1A +#define CARKIT_CTRL_CLR 0x1B +#define CARKIT_CTRL_MICEN (1 << 6) +#define CARKIT_CTRL_SPKRIGHTEN (1 << 5) +#define CARKIT_CTRL_SPKLEFTEN (1 << 4) +#define CARKIT_CTRL_RXDEN (1 << 3) +#define CARKIT_CTRL_TXDEN (1 << 2) +#define CARKIT_CTRL_IDGNDDRV (1 << 1) +#define CARKIT_CTRL_CARKITPWR (1 << 0) +#define CARKIT_PLS_CTRL 0x22 +#define CARKIT_PLS_CTRL_SET 0x23 +#define CARKIT_PLS_CTRL_CLR 0x24 +#define CARKIT_PLS_CTRL_SPKRRIGHT_BIASEN (1 << 3) +#define CARKIT_PLS_CTRL_SPKRLEFT_BIASEN (1 << 2) +#define CARKIT_PLS_CTRL_RXPLSEN (1 << 1) +#define CARKIT_PLS_CTRL_TXPLSEN (1 << 0) + +#define MCPC_CTRL 0x30 +#define MCPC_CTRL_SET 0x31 +#define MCPC_CTRL_CLR 0x32 +#define MCPC_CTRL_RTSOL (1 << 7) +#define MCPC_CTRL_EXTSWR (1 << 6) +#define MCPC_CTRL_EXTSWC (1 << 5) +#define MCPC_CTRL_VOICESW (1 << 4) +#define MCPC_CTRL_OUT64K (1 << 3) +#define MCPC_CTRL_RTSCTSSW (1 << 2) +#define MCPC_CTRL_HS_UART (1 << 0) + +#define MCPC_IO_CTRL 0x33 +#define MCPC_IO_CTRL_SET 0x34 +#define MCPC_IO_CTRL_CLR 0x35 +#define MCPC_IO_CTRL_MICBIASEN (1 << 5) +#define MCPC_IO_CTRL_CTS_NPU (1 << 4) +#define MCPC_IO_CTRL_RXD_PU (1 << 3) +#define MCPC_IO_CTRL_TXDTYP (1 << 2) +#define MCPC_IO_CTRL_CTSTYP (1 << 1) +#define MCPC_IO_CTRL_RTSTYP (1 << 0) + +#define MCPC_CTRL2 0x36 +#define MCPC_CTRL2_SET 0x37 +#define MCPC_CTRL2_CLR 0x38 +#define MCPC_CTRL2_MCPC_CK_EN (1 << 0) + +#define OTHER_FUNC_CTRL 0x80 +#define OTHER_FUNC_CTRL_SET 0x81 +#define OTHER_FUNC_CTRL_CLR 0x82 +#define OTHER_FUNC_CTRL_BDIS_ACON_EN (1 << 4) +#define OTHER_FUNC_CTRL_FIVEWIRE_MODE (1 << 2) + +#define OTHER_IFC_CTRL 0x83 +#define OTHER_IFC_CTRL_SET 0x84 +#define OTHER_IFC_CTRL_CLR 0x85 +#define OTHER_IFC_CTRL_OE_INT_EN (1 << 6) +#define OTHER_IFC_CTRL_CEA2011_MODE (1 << 5) +#define OTHER_IFC_CTRL_FSLSSERIALMODE_4PIN (1 << 4) +#define OTHER_IFC_CTRL_HIZ_ULPI_60MHZ_OUT (1 << 3) +#define OTHER_IFC_CTRL_HIZ_ULPI (1 << 2) +#define OTHER_IFC_CTRL_ALT_INT_REROUTE (1 << 0) + +#define OTHER_INT_EN_RISE 0x86 +#define OTHER_INT_EN_RISE_SET 0x87 +#define OTHER_INT_EN_RISE_CLR 0x88 +#define OTHER_INT_EN_FALL 0x89 +#define OTHER_INT_EN_FALL_SET 0x8A +#define OTHER_INT_EN_FALL_CLR 0x8B +#define OTHER_INT_STS 0x8C +#define OTHER_INT_LATCH 0x8D +#define OTHER_INT_VB_SESS_VLD (1 << 7) +#define OTHER_INT_DM_HI (1 << 6) /* not valid for "latch" reg */ +#define OTHER_INT_DP_HI (1 << 5) /* not valid for "latch" reg */ +#define OTHER_INT_BDIS_ACON (1 << 3) /* not valid for "fall" regs */ +#define OTHER_INT_MANU (1 << 1) +#define OTHER_INT_ABNORMAL_STRESS (1 << 0) + +#define ID_STATUS 0x96 +#define ID_RES_FLOAT (1 << 4) +#define ID_RES_440K (1 << 3) +#define ID_RES_200K (1 << 2) +#define ID_RES_102K (1 << 1) +#define ID_RES_GND (1 << 0) + +#define POWER_CTRL 0xAC +#define POWER_CTRL_SET 0xAD +#define POWER_CTRL_CLR 0xAE +#define POWER_CTRL_OTG_ENAB (1 << 5) + +#define OTHER_IFC_CTRL2 0xAF +#define OTHER_IFC_CTRL2_SET 0xB0 +#define OTHER_IFC_CTRL2_CLR 0xB1 +#define OTHER_IFC_CTRL2_ULPI_STP_LOW (1 << 4) +#define OTHER_IFC_CTRL2_ULPI_TXEN_POL (1 << 3) +#define OTHER_IFC_CTRL2_ULPI_4PIN_2430 (1 << 2) +#define OTHER_IFC_CTRL2_USB_INT_OUTSEL_MASK (3 << 0) /* bits 0 and 1 */ +#define OTHER_IFC_CTRL2_USB_INT_OUTSEL_INT1N (0 << 0) +#define OTHER_IFC_CTRL2_USB_INT_OUTSEL_INT2N (1 << 0) + +#define REG_CTRL_EN 0xB2 +#define REG_CTRL_EN_SET 0xB3 +#define REG_CTRL_EN_CLR 0xB4 +#define REG_CTRL_ERROR 0xB5 +#define ULPI_I2C_CONFLICT_INTEN (1 << 0) + +#define OTHER_FUNC_CTRL2 0xB8 +#define OTHER_FUNC_CTRL2_SET 0xB9 +#define OTHER_FUNC_CTRL2_CLR 0xBA +#define OTHER_FUNC_CTRL2_VBAT_TIMER_EN (1 << 0) + +/* following registers do not have separate _clr and _set registers */ +#define VBUS_DEBOUNCE 0xC0 +#define ID_DEBOUNCE 0xC1 +#define VBAT_TIMER 0xD3 +#define PHY_PWR_CTRL 0xFD +#define PHY_PWR_PHYPWD (1 << 0) +#define PHY_CLK_CTRL 0xFE +#define PHY_CLK_CTRL_CLOCKGATING_EN (1 << 2) +#define PHY_CLK_CTRL_CLK32K_EN (1 << 1) +#define REQ_PHY_DPLL_CLK (1 << 0) +#define PHY_CLK_CTRL_STS 0xFF +#define PHY_DPLL_CLK (1 << 0) + +/* In module TWL4030_MODULE_PM_MASTER */ +#define PROTECT_KEY 0x0E + +/* In module TWL4030_MODULE_PM_RECEIVER */ +#define VUSB_DEDICATED1 0x7D +#define VUSB_DEDICATED2 0x7E +#define VUSB1V5_DEV_GRP 0x71 +#define VUSB1V5_TYPE 0x72 +#define VUSB1V5_REMAP 0x73 +#define VUSB1V8_DEV_GRP 0x74 +#define VUSB1V8_TYPE 0x75 +#define VUSB1V8_REMAP 0x76 +#define VUSB3V1_DEV_GRP 0x77 +#define VUSB3V1_TYPE 0x78 +#define VUSB3V1_REMAP 0x79 + +/* In module TWL4030_MODULE_INTBR */ +#define PMBR1 0x0D +#define GPIO_USB_4PIN_ULPI_2430C (3 << 0) + + + +enum linkstat { + USB_LINK_UNKNOWN = 0, + USB_LINK_NONE, + USB_LINK_VBUS, + USB_LINK_ID, +}; + +struct twl4030_usb { + struct otg_transceiver otg; + struct device *dev; + + /* for vbus reporting with irqs disabled */ + spinlock_t lock; + + /* pin configuration */ + enum twl4030_usb_mode usb_mode; + + int irq; + u8 linkstat; + u8 asleep; + bool irq_enabled; +}; + +/* internal define on top of container_of */ +#define xceiv_to_twl(x) container_of((x), struct twl4030_usb, otg); + +/*-------------------------------------------------------------------------*/ + +static int twl4030_i2c_write_u8_verify(struct twl4030_usb *twl, + u8 module, u8 data, u8 address) +{ + u8 check; + + if ((twl4030_i2c_write_u8(module, data, address) >= 0) && + (twl4030_i2c_read_u8(module, &check, address) >= 0) && + (check == data)) + return 0; + dev_dbg(twl->dev, "Write%d[%d,0x%x] wrote %02x but read %02x\n", + 1, module, address, check, data); + + /* Failed once: Try again */ + if ((twl4030_i2c_write_u8(module, data, address) >= 0) && + (twl4030_i2c_read_u8(module, &check, address) >= 0) && + (check == data)) + return 0; + dev_dbg(twl->dev, "Write%d[%d,0x%x] wrote %02x but read %02x\n", + 2, module, address, check, data); + + /* Failed again: Return error */ + return -EBUSY; +} + +#define twl4030_usb_write_verify(twl, address, data) \ + twl4030_i2c_write_u8_verify(twl, TWL4030_MODULE_USB, (data), (address)) + +static inline int twl4030_usb_write(struct twl4030_usb *twl, + u8 address, u8 data) +{ + int ret = 0; + + ret = twl4030_i2c_write_u8(TWL4030_MODULE_USB, data, address); + if (ret < 0) + dev_dbg(twl->dev, + "TWL4030:USB:Write[0x%x] Error %d\n", address, ret); + return ret; +} + +static inline int twl4030_readb(struct twl4030_usb *twl, u8 module, u8 address) +{ + u8 data; + int ret = 0; + + ret = twl4030_i2c_read_u8(module, &data, address); + if (ret >= 0) + ret = data; + else + dev_dbg(twl->dev, + "TWL4030:readb[0x%x,0x%x] Error %d\n", + module, address, ret); + + return ret; +} + +static inline int twl4030_usb_read(struct twl4030_usb *twl, u8 address) +{ + return twl4030_readb(twl, TWL4030_MODULE_USB, address); +} + +/*-------------------------------------------------------------------------*/ + +static inline int +twl4030_usb_set_bits(struct twl4030_usb *twl, u8 reg, u8 bits) +{ + return twl4030_usb_write(twl, reg + 1, bits); +} + +static inline int +twl4030_usb_clear_bits(struct twl4030_usb *twl, u8 reg, u8 bits) +{ + return twl4030_usb_write(twl, reg + 2, bits); +} + +/*-------------------------------------------------------------------------*/ + +static enum linkstat twl4030_usb_linkstat(struct twl4030_usb *twl) +{ + int status; + int linkstat = USB_LINK_UNKNOWN; + + /* STS_HW_CONDITIONS */ + status = twl4030_readb(twl, TWL4030_MODULE_PM_MASTER, 0x0f); + if (status < 0) + dev_err(twl->dev, "USB link status err %d\n", status); + else if (status & BIT(7)) + linkstat = USB_LINK_VBUS; + else if (status & BIT(2)) + linkstat = USB_LINK_ID; + else + linkstat = USB_LINK_NONE; + + dev_dbg(twl->dev, "HW_CONDITIONS 0x%02x/%d; link %d\n", + status, status, linkstat); + + /* REVISIT this assumes host and peripheral controllers + * are registered, and that both are active... + */ + + spin_lock_irq(&twl->lock); + twl->linkstat = linkstat; + if (linkstat == USB_LINK_ID) { + twl->otg.default_a = true; + twl->otg.state = OTG_STATE_A_IDLE; + } else { + twl->otg.default_a = false; + twl->otg.state = OTG_STATE_B_IDLE; + } + spin_unlock_irq(&twl->lock); + + return linkstat; +} + +static void twl4030_usb_set_mode(struct twl4030_usb *twl, int mode) +{ + twl->usb_mode = mode; + + switch (mode) { + case T2_USB_MODE_ULPI: + twl4030_usb_clear_bits(twl, IFC_CTRL, IFC_CTRL_CARKITMODE); + twl4030_usb_set_bits(twl, POWER_CTRL, POWER_CTRL_OTG_ENAB); + twl4030_usb_clear_bits(twl, FUNC_CTRL, + FUNC_CTRL_XCVRSELECT_MASK | + FUNC_CTRL_OPMODE_MASK); + break; + case -1: + /* FIXME: power on defaults */ + break; + default: + dev_err(twl->dev, "unsupported T2 transceiver mode %d\n", + mode); + break; + }; +} + +static void twl4030_i2c_access(struct twl4030_usb *twl, int on) +{ + unsigned long timeout; + int val = twl4030_usb_read(twl, PHY_CLK_CTRL); + + if (val >= 0) { + if (on) { + /* enable DPLL to access PHY registers over I2C */ + val |= REQ_PHY_DPLL_CLK; + WARN_ON(twl4030_usb_write_verify(twl, PHY_CLK_CTRL, + (u8)val) < 0); + + timeout = jiffies + HZ; + while (!(twl4030_usb_read(twl, PHY_CLK_CTRL_STS) & + PHY_DPLL_CLK) + && time_before(jiffies, timeout)) + udelay(10); + if (!(twl4030_usb_read(twl, PHY_CLK_CTRL_STS) & + PHY_DPLL_CLK)) + dev_err(twl->dev, "Timeout setting T2 HSUSB " + "PHY DPLL clock\n"); + } else { + /* let ULPI control the DPLL clock */ + val &= ~REQ_PHY_DPLL_CLK; + WARN_ON(twl4030_usb_write_verify(twl, PHY_CLK_CTRL, + (u8)val) < 0); + } + } +} + +static void twl4030_phy_power(struct twl4030_usb *twl, int on) +{ + u8 pwr; + + pwr = twl4030_usb_read(twl, PHY_PWR_CTRL); + if (on) { + pwr &= ~PHY_PWR_PHYPWD; + WARN_ON(twl4030_usb_write_verify(twl, PHY_PWR_CTRL, pwr) < 0); + twl4030_usb_write(twl, PHY_CLK_CTRL, + twl4030_usb_read(twl, PHY_CLK_CTRL) | + (PHY_CLK_CTRL_CLOCKGATING_EN | + PHY_CLK_CTRL_CLK32K_EN)); + } else { + pwr |= PHY_PWR_PHYPWD; + WARN_ON(twl4030_usb_write_verify(twl, PHY_PWR_CTRL, pwr) < 0); + } +} + +static void twl4030_phy_suspend(struct twl4030_usb *twl, int controller_off) +{ + if (twl->asleep) + return; + + twl4030_phy_power(twl, 0); + twl->asleep = 1; +} + +static void twl4030_phy_resume(struct twl4030_usb *twl) +{ + if (!twl->asleep) + return; + + twl4030_phy_power(twl, 1); + twl4030_i2c_access(twl, 1); + twl4030_usb_set_mode(twl, twl->usb_mode); + if (twl->usb_mode == T2_USB_MODE_ULPI) + twl4030_i2c_access(twl, 0); + twl->asleep = 0; +} + +static void twl4030_usb_ldo_init(struct twl4030_usb *twl) +{ + /* Enable writing to power configuration registers */ + twl4030_i2c_write_u8(TWL4030_MODULE_PM_MASTER, 0xC0, PROTECT_KEY); + twl4030_i2c_write_u8(TWL4030_MODULE_PM_MASTER, 0x0C, PROTECT_KEY); + + /* put VUSB3V1 LDO in active state */ + twl4030_i2c_write_u8(TWL4030_MODULE_PM_RECEIVER, 0, VUSB_DEDICATED2); + + /* input to VUSB3V1 LDO is from VBAT, not VBUS */ + twl4030_i2c_write_u8(TWL4030_MODULE_PM_RECEIVER, 0x14, VUSB_DEDICATED1); + + /* turn on 3.1V regulator */ + twl4030_i2c_write_u8(TWL4030_MODULE_PM_RECEIVER, 0x20, VUSB3V1_DEV_GRP); + twl4030_i2c_write_u8(TWL4030_MODULE_PM_RECEIVER, 0, VUSB3V1_TYPE); + + /* turn on 1.5V regulator */ + twl4030_i2c_write_u8(TWL4030_MODULE_PM_RECEIVER, 0x20, VUSB1V5_DEV_GRP); + twl4030_i2c_write_u8(TWL4030_MODULE_PM_RECEIVER, 0, VUSB1V5_TYPE); + + /* turn on 1.8V regulator */ + twl4030_i2c_write_u8(TWL4030_MODULE_PM_RECEIVER, 0x20, VUSB1V8_DEV_GRP); + twl4030_i2c_write_u8(TWL4030_MODULE_PM_RECEIVER, 0, VUSB1V8_TYPE); + + /* disable access to power configuration registers */ + twl4030_i2c_write_u8(TWL4030_MODULE_PM_MASTER, 0, PROTECT_KEY); +} + +static ssize_t twl4030_usb_vbus_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct twl4030_usb *twl = dev_get_drvdata(dev); + unsigned long flags; + int ret = -EINVAL; + + spin_lock_irqsave(&twl->lock, flags); + ret = sprintf(buf, "%s\n", + (twl->linkstat == USB_LINK_VBUS) ? "on" : "off"); + spin_unlock_irqrestore(&twl->lock, flags); + + return ret; +} +static DEVICE_ATTR(vbus, 0444, twl4030_usb_vbus_show, NULL); + +static irqreturn_t twl4030_usb_irq(int irq, void *_twl) +{ + struct twl4030_usb *twl = _twl; + int status; + +#ifdef CONFIG_LOCKDEP + /* WORKAROUND for lockdep forcing IRQF_DISABLED on us, which + * we don't want and can't tolerate. Although it might be + * friendlier not to borrow this thread context... + */ + local_irq_enable(); +#endif + + status = twl4030_usb_linkstat(twl); + if (status != USB_LINK_UNKNOWN) { + + /* FIXME add a set_power() method so that B-devices can + * configure the charger appropriately. It's not always + * correct to consume VBUS power, and how much current to + * consume is a function of the USB configuration chosen + * by the host. + * + * REVISIT usb_gadget_vbus_connect(...) as needed, ditto + * its disconnect() sibling, when changing to/from the + * USB_LINK_VBUS state. musb_hdrc won't care until it + * starts to handle softconnect right. + */ + twl4030charger_usb_en(status == USB_LINK_VBUS); + + if (status == USB_LINK_NONE) + twl4030_phy_suspend(twl, 0); + else + twl4030_phy_resume(twl); + } + sysfs_notify(&twl->dev->kobj, NULL, "vbus"); + + return IRQ_HANDLED; +} + +static int twl4030_set_suspend(struct otg_transceiver *x, int suspend) +{ + struct twl4030_usb *twl = xceiv_to_twl(x); + + if (suspend) + twl4030_phy_suspend(twl, 1); + else + twl4030_phy_resume(twl); + + return 0; +} + +static int twl4030_set_peripheral(struct otg_transceiver *x, + struct usb_gadget *gadget) +{ + struct twl4030_usb *twl; + + if (!x) + return -ENODEV; + + twl = xceiv_to_twl(x); + twl->otg.gadget = gadget; + if (!gadget) + twl->otg.state = OTG_STATE_UNDEFINED; + + return 0; +} + +static int twl4030_set_host(struct otg_transceiver *x, struct usb_bus *host) +{ + struct twl4030_usb *twl; + + if (!x) + return -ENODEV; + + twl = xceiv_to_twl(x); + twl->otg.host = host; + if (!host) + twl->otg.state = OTG_STATE_UNDEFINED; + + return 0; +} + +static int __init twl4030_usb_probe(struct platform_device *pdev) +{ + struct twl4030_usb_data *pdata = pdev->dev.platform_data; + struct twl4030_usb *twl; + int status; + + if (!pdata) { + dev_dbg(&pdev->dev, "platform_data not available\n"); + return -EINVAL; + } + + twl = kzalloc(sizeof *twl, GFP_KERNEL); + if (!twl) + return -ENOMEM; + + twl->dev = &pdev->dev; + twl->irq = platform_get_irq(pdev, 0); + twl->otg.dev = twl->dev; + twl->otg.label = "twl4030"; + twl->otg.set_host = twl4030_set_host; + twl->otg.set_peripheral = twl4030_set_peripheral; + twl->otg.set_suspend = twl4030_set_suspend; + twl->usb_mode = pdata->usb_mode; + twl->asleep = 1; + + /* init spinlock for workqueue */ + spin_lock_init(&twl->lock); + + twl4030_usb_ldo_init(twl); + otg_set_transceiver(&twl->otg); + + platform_set_drvdata(pdev, twl); + if (device_create_file(&pdev->dev, &dev_attr_vbus)) + dev_warn(&pdev->dev, "could not create sysfs file\n"); + + /* Our job is to use irqs and status from the power module + * to keep the transceiver disabled when nothing's connected. + * + * FIXME we actually shouldn't start enabling it until the + * USB controller drivers have said they're ready, by calling + * set_host() and/or set_peripheral() ... OTG_capable boards + * need both handles, otherwise just one suffices. + */ + twl->irq_enabled = true; + status = request_irq(twl->irq, twl4030_usb_irq, + IRQF_TRIGGER_FALLING | IRQF_TRIGGER_RISING, + "twl4030_usb", twl); + if (status < 0) { + dev_dbg(&pdev->dev, "can't get IRQ %d, err %d\n", + twl->irq, status); + kfree(twl); + return status; + } + + /* The IRQ handler just handles changes from the previous states + * of the ID and VBUS pins ... in probe() we must initialize that + * previous state. The easy way: fake an IRQ. + * + * REVISIT: a real IRQ might have happened already, if PREEMPT is + * enabled. Else the IRQ may not yet be configured or enabled, + * because of scheduling delays. + */ + twl4030_usb_irq(twl->irq, twl); + + dev_info(&pdev->dev, "Initialized TWL4030 USB module\n"); + return 0; +} + +static int __exit twl4030_usb_remove(struct platform_device *pdev) +{ + struct twl4030_usb *twl = platform_get_drvdata(pdev); + int val; + + free_irq(twl->irq, twl); + device_remove_file(twl->dev, &dev_attr_vbus); + + /* set transceiver mode to power on defaults */ + twl4030_usb_set_mode(twl, -1); + + /* autogate 60MHz ULPI clock, + * clear dpll clock request for i2c access, + * disable 32KHz + */ + val = twl4030_usb_read(twl, PHY_CLK_CTRL); + if (val >= 0) { + val |= PHY_CLK_CTRL_CLOCKGATING_EN; + val &= ~(PHY_CLK_CTRL_CLK32K_EN | REQ_PHY_DPLL_CLK); + twl4030_usb_write(twl, PHY_CLK_CTRL, (u8)val); + } + + /* disable complete OTG block */ + twl4030_usb_clear_bits(twl, POWER_CTRL, POWER_CTRL_OTG_ENAB); + + twl4030_phy_power(twl, 0); + + kfree(twl); + + return 0; +} + +static struct platform_driver twl4030_usb_driver = { + .probe = twl4030_usb_probe, + .remove = __exit_p(twl4030_usb_remove), + .driver = { + .name = "twl4030_usb", + .owner = THIS_MODULE, + }, +}; + +static int __init twl4030_usb_init(void) +{ + return platform_driver_register(&twl4030_usb_driver); +} +subsys_initcall(twl4030_usb_init); + +static void __exit twl4030_usb_exit(void) +{ + platform_driver_unregister(&twl4030_usb_driver); +} +module_exit(twl4030_usb_exit); + +MODULE_ALIAS("platform:twl4030_usb"); +MODULE_AUTHOR("Texas Instruments, Inc, Nokia Corporation"); +MODULE_DESCRIPTION("TWL4030 USB transceiver driver"); +MODULE_LICENSE("GPL"); From 25ff1c316f6a763f1eefe7f8984b2d8c03888432 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Mon, 15 Dec 2008 12:43:41 -0500 Subject: [PATCH 081/123] USB: storage: add last-sector hacks This patch (as1189b) adds some hacks to usb-storage for dealing with the growing problems involving bad capacity values and last-sector accesses: A new flag, US_FL_CAPACITY_OK, is created to indicate that the device is known to report its capacity correctly. An unusual_devs entry for Linux's own File-backed Storage Gadget is added with this flag set, since g_file_storage always reports the correct capacity and since the capacity need not be even (it is determined by the size of the backing file). An entry in unusual_devs.h which has only the CAPACITY_OK flag set shouldn't prejudice libusual, since the device will work perfectly well with either usb-storage or ub. So a new macro, COMPLIANT_DEV, is added to let libusual know about these entries. When a last-sector access succeeds and the total number of sectors is odd (the unexpected case, in which guessing that the number is even might cause trouble), a WARN is triggered. The kerneloops.org project will collect these warnings, allowing us to add CAPACITY_OK flags for the devices in question before implementing the default-to-even heuristic. If users want to prevent the stack dump produced by the WARN, they can disable the hack by adding an unusual_devs entry for their device with the CAPACITY_OK flag. When a last-sector access fails three times in a row and neither the FIX_CAPACITY nor the CAPACITY_OK flag is set, we assume the last-sector bug is present. We replace the existing status and sense data with values that will cause the SCSI core to fail the access immediately rather than retry indefinitely. This should fix the difficulties people have been having with Nokia phones. Signed-off-by: Alan Stern Cc: stable Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/libusual.c | 7 ++ drivers/usb/storage/scsiglue.c | 8 +++ drivers/usb/storage/transport.c | 110 +++++++++++++++++++++++++++++ drivers/usb/storage/unusual_devs.h | 16 ++++- drivers/usb/storage/usb.c | 6 ++ drivers/usb/storage/usb.h | 4 ++ include/linux/usb_usual.h | 6 +- 7 files changed, 154 insertions(+), 3 deletions(-) diff --git a/drivers/usb/storage/libusual.c b/drivers/usb/storage/libusual.c index d617e8ae6b0..f970b27ba30 100644 --- a/drivers/usb/storage/libusual.c +++ b/drivers/usb/storage/libusual.c @@ -46,6 +46,12 @@ static int usu_probe_thread(void *arg); { USB_DEVICE_VER(id_vendor, id_product, bcdDeviceMin,bcdDeviceMax), \ .driver_info = (flags)|(USB_US_TYPE_STOR<<24) } +#define COMPLIANT_DEV(id_vendor, id_product, bcdDeviceMin, bcdDeviceMax, \ + vendorName, productName, useProtocol, useTransport, \ + initFunction, flags) \ +{ USB_DEVICE_VER(id_vendor, id_product, bcdDeviceMin, bcdDeviceMax), \ + .driver_info = (flags) } + #define USUAL_DEV(useProto, useTrans, useType) \ { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, useProto, useTrans), \ .driver_info = ((useType)<<24) } @@ -57,6 +63,7 @@ struct usb_device_id storage_usb_ids [] = { #undef USUAL_DEV #undef UNUSUAL_DEV +#undef COMPLIANT_DEV MODULE_DEVICE_TABLE(usb, storage_usb_ids); EXPORT_SYMBOL_GPL(storage_usb_ids); diff --git a/drivers/usb/storage/scsiglue.c b/drivers/usb/storage/scsiglue.c index e9d6c196a7a..8d78084abf9 100644 --- a/drivers/usb/storage/scsiglue.c +++ b/drivers/usb/storage/scsiglue.c @@ -208,6 +208,14 @@ static int slave_configure(struct scsi_device *sdev) * sector in a larger then 1 sector read, since the performance * impact is negible we set this flag for all USB disks */ sdev->last_sector_bug = 1; + + /* Enable last-sector hacks for single-target devices using + * the Bulk-only transport, unless we already know the + * capacity will be decremented or is correct. */ + if (!(us->fflags & (US_FL_FIX_CAPACITY | US_FL_CAPACITY_OK | + US_FL_SCM_MULT_TARG)) && + us->protocol == US_PR_BULK) + us->use_last_sector_hacks = 1; } else { /* Non-disk-type devices don't need to blacklist any pages diff --git a/drivers/usb/storage/transport.c b/drivers/usb/storage/transport.c index 9cc30afd6d3..1d5438e6363 100644 --- a/drivers/usb/storage/transport.c +++ b/drivers/usb/storage/transport.c @@ -57,6 +57,9 @@ #include "scsiglue.h" #include "debug.h" +#include +#include "../../scsi/sd.h" + /*********************************************************************** * Data transfer routines @@ -511,6 +514,110 @@ int usb_stor_bulk_transfer_sg(struct us_data* us, unsigned int pipe, * Transport routines ***********************************************************************/ +/* There are so many devices that report the capacity incorrectly, + * this routine was written to counteract some of the resulting + * problems. + */ +static void last_sector_hacks(struct us_data *us, struct scsi_cmnd *srb) +{ + struct gendisk *disk; + struct scsi_disk *sdkp; + u32 sector; + + /* To Report "Medium Error: Record Not Found */ + static unsigned char record_not_found[18] = { + [0] = 0x70, /* current error */ + [2] = MEDIUM_ERROR, /* = 0x03 */ + [7] = 0x0a, /* additional length */ + [12] = 0x14 /* Record Not Found */ + }; + + /* If last-sector problems can't occur, whether because the + * capacity was already decremented or because the device is + * known to report the correct capacity, then we don't need + * to do anything. + */ + if (!us->use_last_sector_hacks) + return; + + /* Was this command a READ(10) or a WRITE(10)? */ + if (srb->cmnd[0] != READ_10 && srb->cmnd[0] != WRITE_10) + goto done; + + /* Did this command access the last sector? */ + sector = (srb->cmnd[2] << 24) | (srb->cmnd[3] << 16) | + (srb->cmnd[4] << 8) | (srb->cmnd[5]); + disk = srb->request->rq_disk; + if (!disk) + goto done; + sdkp = scsi_disk(disk); + if (!sdkp) + goto done; + if (sector + 1 != sdkp->capacity) + goto done; + + if (srb->result == SAM_STAT_GOOD && scsi_get_resid(srb) == 0) { + + /* The command succeeded. If the capacity is odd + * (i.e., if the sector number is even) then the + * "always-even" heuristic would be wrong for this + * device. Issue a WARN() so that the kerneloops.org + * project will be notified and we will then know to + * mark the device with a CAPACITY_OK flag. Hopefully + * this will occur for only a few devices. + * + * Use the sign of us->last_sector_hacks to tell whether + * the warning has already been issued; we don't need + * more than one warning per device. + */ + if (!(sector & 1) && us->use_last_sector_hacks > 0) { + unsigned vid = le16_to_cpu( + us->pusb_dev->descriptor.idVendor); + unsigned pid = le16_to_cpu( + us->pusb_dev->descriptor.idProduct); + unsigned rev = le16_to_cpu( + us->pusb_dev->descriptor.bcdDevice); + + WARN(1, "%s: Successful last sector success at %u, " + "device %04x:%04x:%04x\n", + sdkp->disk->disk_name, sector, + vid, pid, rev); + us->use_last_sector_hacks = -1; + } + + } else { + /* The command failed. Allow up to 3 retries in case this + * is some normal sort of failure. After that, assume the + * capacity is wrong and we're trying to access the sector + * beyond the end. Replace the result code and sense data + * with values that will cause the SCSI core to fail the + * command immediately, instead of going into an infinite + * (or even just a very long) retry loop. + */ + if (++us->last_sector_retries < 3) + return; + srb->result = SAM_STAT_CHECK_CONDITION; + memcpy(srb->sense_buffer, record_not_found, + sizeof(record_not_found)); + + /* In theory we might want to issue a WARN() here if the + * capacity is even, since it could indicate the device + * has the READ CAPACITY bug _and_ the real capacity is + * odd. But it could also indicate that the device + * simply can't access its last sector, a failure mode + * which is surprisingly common. So no warning. + */ + } + + done: + /* Don't reset the retry counter for TEST UNIT READY commands, + * because they get issued after device resets which might be + * caused by a failed last-sector access. + */ + if (srb->cmnd[0] != TEST_UNIT_READY) + us->last_sector_retries = 0; +} + /* Invoke the transport and basic error-handling/recovery methods * * This is used by the protocol layers to actually send the message to @@ -544,6 +651,7 @@ void usb_stor_invoke_transport(struct scsi_cmnd *srb, struct us_data *us) /* if the transport provided its own sense data, don't auto-sense */ if (result == USB_STOR_TRANSPORT_NO_SENSE) { srb->result = SAM_STAT_CHECK_CONDITION; + last_sector_hacks(us, srb); return; } @@ -705,6 +813,7 @@ void usb_stor_invoke_transport(struct scsi_cmnd *srb, struct us_data *us) scsi_bufflen(srb) - scsi_get_resid(srb) < srb->underflow) srb->result = (DID_ERROR << 16) | (SUGGEST_RETRY << 24); + last_sector_hacks(us, srb); return; /* Error and abort processing: try to resynchronize with the device @@ -732,6 +841,7 @@ void usb_stor_invoke_transport(struct scsi_cmnd *srb, struct us_data *us) us->transport_reset(us); } clear_bit(US_FLIDX_RESETTING, &us->dflags); + last_sector_hacks(us, srb); } /* Stop the current URB transfer */ diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index 0330ed53ec1..035bbc5d823 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -27,7 +27,8 @@ /* IMPORTANT NOTE: This file must be included in another file which does * the following thing for it to work: - * The macro UNUSUAL_DEV() must be defined before this file is included + * The UNUSUAL_DEV, COMPLIANT_DEV, and USUAL_DEV macros must be defined + * before this file is included. */ /* If you edit this file, please try to keep it sorted first by VendorID, @@ -46,6 +47,12 @@ * */ +/* Note: If you add an entry only in order to set the CAPACITY_OK flag, + * use the COMPLIANT_DEV macro instead of UNUSUAL_DEV. This is + * because such entries mark devices which actually work correctly, + * as opposed to devices that do something strangely or wrongly. + */ + /* patch submitted by Vivian Bregier */ UNUSUAL_DEV( 0x03eb, 0x2002, 0x0100, 0x0100, @@ -704,6 +711,13 @@ UNUSUAL_DEV( 0x0525, 0xa140, 0x0100, 0x0100, US_SC_8070, US_PR_DEVICE, NULL, US_FL_FIX_INQUIRY ), +/* Added by Alan Stern */ +COMPLIANT_DEV(0x0525, 0xa4a5, 0x0000, 0x9999, + "Linux", + "File-backed Storage Gadget", + US_SC_DEVICE, US_PR_DEVICE, NULL, + US_FL_CAPACITY_OK ), + /* Yakumo Mega Image 37 * Submitted by Stephan Fuhrmann */ UNUSUAL_DEV( 0x052b, 0x1801, 0x0100, 0x0100, diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index ce0b580db5e..80e234bf4e5 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -134,6 +134,8 @@ static struct quirks_entry *quirks_list, *quirks_end; { USB_DEVICE_VER(id_vendor, id_product, bcdDeviceMin,bcdDeviceMax), \ .driver_info = (flags)|(USB_US_TYPE_STOR<<24) } +#define COMPLIANT_DEV UNUSUAL_DEV + #define USUAL_DEV(useProto, useTrans, useType) \ { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, useProto, useTrans), \ .driver_info = (USB_US_TYPE_STOR<<24) } @@ -142,6 +144,7 @@ static struct usb_device_id storage_usb_ids [] = { # include "unusual_devs.h" #undef UNUSUAL_DEV +#undef COMPLIANT_DEV #undef USUAL_DEV /* Terminating entry */ { } @@ -172,6 +175,8 @@ MODULE_DEVICE_TABLE (usb, storage_usb_ids); .initFunction = init_function, \ } +#define COMPLIANT_DEV UNUSUAL_DEV + #define USUAL_DEV(use_protocol, use_transport, use_type) \ { \ .useProtocol = use_protocol, \ @@ -181,6 +186,7 @@ MODULE_DEVICE_TABLE (usb, storage_usb_ids); static struct us_unusual_dev us_unusual_dev_list[] = { # include "unusual_devs.h" # undef UNUSUAL_DEV +# undef COMPLIANT_DEV # undef USUAL_DEV /* Terminating entry */ diff --git a/drivers/usb/storage/usb.h b/drivers/usb/storage/usb.h index e4674fc715e..65e674e4be9 100644 --- a/drivers/usb/storage/usb.h +++ b/drivers/usb/storage/usb.h @@ -154,6 +154,10 @@ struct us_data { #ifdef CONFIG_PM pm_hook suspend_resume_hook; #endif + + /* hacks for READ CAPACITY bug handling */ + int use_last_sector_hacks; + int last_sector_retries; }; /* Convert between us_data and the corresponding Scsi_Host */ diff --git a/include/linux/usb_usual.h b/include/linux/usb_usual.h index 998e5cbbf29..1eea1ab68dc 100644 --- a/include/linux/usb_usual.h +++ b/include/linux/usb_usual.h @@ -53,8 +53,10 @@ /* Sets max_sectors to arch min */ \ US_FLAG(BULK_IGNORE_TAG,0x00004000) \ /* Ignore tag mismatch in bulk operations */ \ - US_FLAG(SANE_SENSE, 0x00008000) - /* Sane Sense (> 18 bytes) */ + US_FLAG(SANE_SENSE, 0x00008000) \ + /* Sane Sense (> 18 bytes) */ \ + US_FLAG(CAPACITY_OK, 0x00010000) \ + /* READ CAPACITY response is correct */ #define US_FLAG(name, value) US_FL_##name = value , enum { US_DO_ALL_FLAGS }; From c838ea4626d6e982489ff519f9ecf5e1649ca90b Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Mon, 15 Dec 2008 10:40:06 -0500 Subject: [PATCH 082/123] USB: storage: make the "quirks=" module parameter writable This patch (as1190) makes usb-storage's "quirks=" module parameter writable, so that users can add entries for their devices at runtime with no need to reboot or reload usb-storage. New codes are added for the SANE_SENSE, CAPACITY_HEURISTICS, and CAPACITY_OK flags. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- Documentation/kernel-parameters.txt | 7 ++ drivers/usb/storage/usb.c | 167 +++++++++++----------------- 2 files changed, 75 insertions(+), 99 deletions(-) diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index 8eb6e35405c..a58fc8b7339 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt @@ -2396,14 +2396,21 @@ and is between 256 and 4096 characters. It is defined in the file and Product ID values (4-digit hex numbers) and Flags is a set of characters, each corresponding to a common usb-storage quirk flag as follows: + a = SANE_SENSE (collect more than 18 bytes + of sense data); c = FIX_CAPACITY (decrease the reported device capacity by one sector); + h = CAPACITY_HEURISTICS (decrease the + reported device capacity by one + sector if the number is odd); i = IGNORE_DEVICE (don't bind to this device); l = NOT_LOCKABLE (don't try to lock and unlock ejectable media); m = MAX_SECTORS_64 (don't transfer more than 64 sectors = 32 KB at a time); + o = CAPACITY_OK (accept the capacity + reported by the device); r = IGNORE_RESIDUE (the device reports bogus residue values); s = SINGLE_LUN (the device has only one diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index 80e234bf4e5..4becf495ca2 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -111,16 +111,10 @@ static unsigned int delay_use = 5; module_param(delay_use, uint, S_IRUGO | S_IWUSR); MODULE_PARM_DESC(delay_use, "seconds to delay before using a new device"); -static char *quirks; -module_param(quirks, charp, S_IRUGO); +static char quirks[128]; +module_param_string(quirks, quirks, sizeof(quirks), S_IRUGO | S_IWUSR); MODULE_PARM_DESC(quirks, "supplemental list of device IDs and their quirks"); -struct quirks_entry { - u16 vid, pid; - u32 fflags; -}; -static struct quirks_entry *quirks_list, *quirks_end; - /* * The entries in this table correspond, line for line, @@ -481,28 +475,80 @@ static int associate_dev(struct us_data *us, struct usb_interface *intf) return 0; } +/* Works only for digits and letters, but small and fast */ +#define TOLOWER(x) ((x) | 0x20) + /* Adjust device flags based on the "quirks=" module parameter */ static void adjust_quirks(struct us_data *us) { - u16 vid, pid; - struct quirks_entry *q; - unsigned int mask = (US_FL_FIX_CAPACITY | US_FL_IGNORE_DEVICE | + char *p; + u16 vid = le16_to_cpu(us->pusb_dev->descriptor.idVendor); + u16 pid = le16_to_cpu(us->pusb_dev->descriptor.idProduct); + unsigned f = 0; + unsigned int mask = (US_FL_SANE_SENSE | US_FL_FIX_CAPACITY | + US_FL_CAPACITY_HEURISTICS | US_FL_IGNORE_DEVICE | US_FL_NOT_LOCKABLE | US_FL_MAX_SECTORS_64 | - US_FL_IGNORE_RESIDUE | US_FL_SINGLE_LUN | - US_FL_NO_WP_DETECT); + US_FL_CAPACITY_OK | US_FL_IGNORE_RESIDUE | + US_FL_SINGLE_LUN | US_FL_NO_WP_DETECT); - vid = le16_to_cpu(us->pusb_dev->descriptor.idVendor); - pid = le16_to_cpu(us->pusb_dev->descriptor.idProduct); - - for (q = quirks_list; q != quirks_end; ++q) { - if (q->vid == vid && q->pid == pid) { - us->fflags = (us->fflags & ~mask) | q->fflags; - dev_info(&us->pusb_intf->dev, "Quirks match for " - "vid %04x pid %04x: %x\n", - vid, pid, q->fflags); + p = quirks; + while (*p) { + /* Each entry consists of VID:PID:flags */ + if (vid == simple_strtoul(p, &p, 16) && + *p == ':' && + pid == simple_strtoul(p+1, &p, 16) && + *p == ':') break; + + /* Move forward to the next entry */ + while (*p) { + if (*p++ == ',') + break; } } + if (!*p) /* No match */ + return; + + /* Collect the flags */ + while (*++p && *p != ',') { + switch (TOLOWER(*p)) { + case 'a': + f |= US_FL_SANE_SENSE; + break; + case 'c': + f |= US_FL_FIX_CAPACITY; + break; + case 'h': + f |= US_FL_CAPACITY_HEURISTICS; + break; + case 'i': + f |= US_FL_IGNORE_DEVICE; + break; + case 'l': + f |= US_FL_NOT_LOCKABLE; + break; + case 'm': + f |= US_FL_MAX_SECTORS_64; + break; + case 'o': + f |= US_FL_CAPACITY_OK; + break; + case 'r': + f |= US_FL_IGNORE_RESIDUE; + break; + case 's': + f |= US_FL_SINGLE_LUN; + break; + case 'w': + f |= US_FL_NO_WP_DETECT; + break; + /* Ignore unrecognized flag characters */ + } + } + us->fflags = (us->fflags & ~mask) | f; + dev_info(&us->pusb_intf->dev, "Quirks match for " + "vid %04x pid %04x: %x\n", + vid, pid, f); } /* Find an unusual_dev descriptor (always succeeds in the current code) */ @@ -1092,88 +1138,11 @@ static struct usb_driver usb_storage_driver = { .soft_unbind = 1, }; -/* Works only for digits and letters, but small and fast */ -#define TOLOWER(x) ((x) | 0x20) - -static void __init parse_quirks(void) -{ - int n, i; - char *p; - - if (!quirks) - return; - - /* Count the ':' characters to get 2 * the number of entries */ - n = 0; - for (p = quirks; *p; ++p) { - if (*p == ':') - ++n; - } - n /= 2; - if (n == 0) - return; /* Don't allocate 0 bytes */ - - quirks_list = kmalloc(n * sizeof(*quirks_list), GFP_KERNEL); - if (!quirks_list) - return; - - p = quirks; - quirks_end = quirks_list; - for (i = 0; i < n && *p; ++i) { - unsigned f = 0; - - /* Each entry consists of VID:PID:flags */ - quirks_end->vid = simple_strtoul(p, &p, 16); - if (*p != ':') - goto skip_to_next; - quirks_end->pid = simple_strtoul(p+1, &p, 16); - if (*p != ':') - goto skip_to_next; - - while (*++p && *p != ',') { - switch (TOLOWER(*p)) { - case 'c': - f |= US_FL_FIX_CAPACITY; - break; - case 'i': - f |= US_FL_IGNORE_DEVICE; - break; - case 'l': - f |= US_FL_NOT_LOCKABLE; - break; - case 'm': - f |= US_FL_MAX_SECTORS_64; - break; - case 'r': - f |= US_FL_IGNORE_RESIDUE; - break; - case 's': - f |= US_FL_SINGLE_LUN; - break; - case 'w': - f |= US_FL_NO_WP_DETECT; - break; - /* Ignore unrecognized flag characters */ - } - } - quirks_end->fflags = f; - ++quirks_end; - - skip_to_next: - /* Entries are separated by commas */ - while (*p) { - if (*p++ == ',') - break; - } - } /* for (i = 0; ...) */ -} - static int __init usb_stor_init(void) { int retval; printk(KERN_INFO "Initializing USB Mass Storage driver...\n"); - parse_quirks(); /* register the driver, return usb_register return code if error */ retval = usb_register(&usb_storage_driver); From a81a81a25d3ecdab777abca87c5ddf484056103d Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Thu, 18 Dec 2008 16:41:49 -0500 Subject: [PATCH 083/123] USB: storage: set CAPACITY_HEURISTICS flag for bad vendors This patch (as1194) makes usb-storage set the CAPACITY_HEURISTICS flag for all devices made by Nokia, Nikon, or Motorola. These companies seem to include the READ CAPACITY bug in all of their devices. Since cell phones and digital cameras rely on flash storage, which always has an even number of sectors, setting CAPACITY_HEURISTICS shouldn't cause any problems. Not even if the companies wise up and start making devices without the bug. A large number of unusual_devs entries are now unnecessary, so the patch removes them. Signed-off-by: Alan Stern Cc: stable Signed-off-by: Greg Kroah-Hartman --- drivers/usb/storage/scsiglue.c | 23 +++++ drivers/usb/storage/unusual_devs.h | 155 ----------------------------- 2 files changed, 23 insertions(+), 155 deletions(-) diff --git a/drivers/usb/storage/scsiglue.c b/drivers/usb/storage/scsiglue.c index 8d78084abf9..2a42b862aa9 100644 --- a/drivers/usb/storage/scsiglue.c +++ b/drivers/usb/storage/scsiglue.c @@ -59,6 +59,13 @@ #include "transport.h" #include "protocol.h" +/* Vendor IDs for companies that seem to include the READ CAPACITY bug + * in all their devices + */ +#define VENDOR_ID_NOKIA 0x0421 +#define VENDOR_ID_NIKON 0x04b0 +#define VENDOR_ID_MOTOROLA 0x22b8 + /*********************************************************************** * Host functions ***********************************************************************/ @@ -142,6 +149,22 @@ static int slave_configure(struct scsi_device *sdev) * settings can't be overridden via the scsi devinfo mechanism. */ if (sdev->type == TYPE_DISK) { + /* Some vendors seem to put the READ CAPACITY bug into + * all their devices -- primarily makers of cell phones + * and digital cameras. Since these devices always use + * flash media and can be expected to have an even number + * of sectors, we will always enable the CAPACITY_HEURISTICS + * flag unless told otherwise. */ + switch (le16_to_cpu(us->pusb_dev->descriptor.idVendor)) { + case VENDOR_ID_NOKIA: + case VENDOR_ID_NIKON: + case VENDOR_ID_MOTOROLA: + if (!(us->fflags & (US_FL_FIX_CAPACITY | + US_FL_CAPACITY_OK))) + us->fflags |= US_FL_CAPACITY_HEURISTICS; + break; + } + /* Disk-type devices use MODE SENSE(6) if the protocol * (SubClass) is Transparent SCSI, otherwise they use * MODE SENSE(10). */ diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index 035bbc5d823..a7f9513fa19 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -174,42 +174,6 @@ UNUSUAL_DEV( 0x0421, 0x0019, 0x0592, 0x0592, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_MAX_SECTORS_64 ), -/* Reported by Filip Joelsson */ -UNUSUAL_DEV( 0x0421, 0x005d, 0x0001, 0x0600, - "Nokia", - "Nokia 3110c", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - -/* Reported by Ozan Sener */ -UNUSUAL_DEV( 0x0421, 0x0060, 0x0551, 0x0551, - "Nokia", - "3500c", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - -/* Reported by CSECSY Laszlo */ -UNUSUAL_DEV( 0x0421, 0x0063, 0x0001, 0x0601, - "Nokia", - "Nokia 3109c", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - -/* Patch for Nokia 5310 capacity */ -UNUSUAL_DEV( 0x0421, 0x006a, 0x0000, 0x0701, - "Nokia", - "5310", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - -/* Submitted by Ricky Wong Yung Fei */ -/* Nokia 7610 Supernova - Too many sectors reported in usb storage mode */ -UNUSUAL_DEV( 0x0421, 0x00f5, 0x0000, 0x0470, - "Nokia", - "7610 Supernova", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - /* Reported by Mario Rettig */ UNUSUAL_DEV( 0x0421, 0x042e, 0x0100, 0x0100, "Nokia", @@ -275,27 +239,6 @@ UNUSUAL_DEV( 0x0421, 0x0495, 0x0370, 0x0370, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_MAX_SECTORS_64 ), -/* Reported by Cedric Godin */ -UNUSUAL_DEV( 0x0421, 0x04b9, 0x0500, 0x0551, - "Nokia", - "5300", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - -/* Reported by Paulo Fessel */ -UNUSUAL_DEV( 0x0421, 0x04bd, 0x0000, 0x9999, - "Nokia", - "5200", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - -/* Reported by Richard Nauber */ -UNUSUAL_DEV( 0x0421, 0x04fa, 0x0550, 0x0660, - "Nokia", - "6300", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY ), - /* Reported by Olaf Hering from novell bug #105878 */ UNUSUAL_DEV( 0x0424, 0x0fdc, 0x0210, 0x0210, "SMSC", @@ -425,83 +368,6 @@ UNUSUAL_DEV( 0x04b0, 0x0301, 0x0010, 0x0010, US_SC_DEVICE, US_PR_DEVICE,NULL, US_FL_NOT_LOCKABLE ), -/* Reported by Stefan de Konink */ -UNUSUAL_DEV( 0x04b0, 0x0401, 0x0200, 0x0200, - "NIKON", - "NIKON DSC D100", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Tobias Kunze Briseno */ -UNUSUAL_DEV( 0x04b0, 0x0403, 0x0200, 0x0200, - "NIKON", - "NIKON DSC D2H", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Milinevsky Dmitry */ -UNUSUAL_DEV( 0x04b0, 0x0409, 0x0100, 0x0100, - "NIKON", - "NIKON DSC D50", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Andreas Bockhold */ -UNUSUAL_DEV( 0x04b0, 0x0405, 0x0100, 0x0100, - "NIKON", - "NIKON DSC D70", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Jamie Kitson */ -UNUSUAL_DEV( 0x04b0, 0x040d, 0x0100, 0x0100, - "NIKON", - "NIKON DSC D70s", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Graber and Mike Pagano */ -UNUSUAL_DEV( 0x04b0, 0x040f, 0x0100, 0x0200, - "NIKON", - "NIKON DSC D200", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Emil Larsson */ -UNUSUAL_DEV( 0x04b0, 0x0411, 0x0100, 0x0111, - "NIKON", - "NIKON DSC D80", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Ortwin Glueck */ -UNUSUAL_DEV( 0x04b0, 0x0413, 0x0110, 0x0111, - "NIKON", - "NIKON DSC D40", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Paul Check */ -UNUSUAL_DEV( 0x04b0, 0x0415, 0x0100, 0x0100, - "NIKON", - "NIKON DSC D2Xs", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by Shan Destromp (shansan@gmail.com) */ -UNUSUAL_DEV( 0x04b0, 0x0417, 0x0100, 0x0100, - "NIKON", - "NIKON DSC D40X", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* Reported by paul ready */ -UNUSUAL_DEV( 0x04b0, 0x0419, 0x0100, 0x0200, - "NIKON", - "NIKON DSC D300", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - /* Reported by Doug Maxey (dwm@austin.ibm.com) */ UNUSUAL_DEV( 0x04b3, 0x4001, 0x0110, 0x0110, "IBM", @@ -2184,27 +2050,6 @@ UNUSUAL_DEV( 0x22b8, 0x3010, 0x0001, 0x0001, US_SC_DEVICE, US_PR_DEVICE, NULL, US_FL_FIX_CAPACITY | US_FL_IGNORE_RESIDUE ), -/* - * Patch by Pete Zaitcev - * Report by Mark Patton. Red Hat bz#208928. - * Added support for rev 0x0002 (Motorola ROKR W5) - * by Javier Smaldone - */ -UNUSUAL_DEV( 0x22b8, 0x4810, 0x0001, 0x0002, - "Motorola", - "RAZR V3i/ROKR W5", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - -/* - * Patch by Jost Diederichs - */ -UNUSUAL_DEV(0x22b8, 0x6410, 0x0001, 0x9999, - "Motorola Inc.", - "Motorola Phone (RAZRV3xx)", - US_SC_DEVICE, US_PR_DEVICE, NULL, - US_FL_FIX_CAPACITY), - /* * Patch by Constantin Baranov * Report by Andreas Koenecke. From a0d4922da2e4ccb0973095d8d29f36f6b1b5f703 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Wed, 17 Dec 2008 15:06:03 -0500 Subject: [PATCH 084/123] USB: fix up suspend and resume for PCI host controllers This patch (as1192) rearranges the USB PCI host controller suspend and resume and resume routines: Use pci_wake_from_d3() for enabling and disabling wakeup, instead of pci_enable_wake(). Carry out the actual state change while interrupts are disabled. Change the order of the preparations to agree with the general recommendation for PCI devices, instead of messing around with the wakeup settings while the device is in D3. In .suspend: Call the underlying driver to disable IRQ generation; pci_wake_from_d3(device_may_wakeup()); pci_disable_device(); In .suspend_late: pci_save_state(); pci_set_power_state(D3hot); (for PPC_PMAC) Disable ASIC clocks In .resume_early: (for PPC_PMAC) Enable ASIC clocks pci_set_power_state(D0); pci_restore_state(); In .resume: pci_enable_device(); pci_set_master(); pci_wake_from_d3(0); Call the underlying driver to reenable IRQ generation Add the necessary .suspend_late and .resume_early method pointers to the PCI host controller drivers. Signed-off-by: Alan Stern CC: Rafael J. Wysocki Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/hcd-pci.c | 200 +++++++++++++++++++----------------- drivers/usb/core/hcd.h | 4 +- drivers/usb/host/ehci-pci.c | 2 + drivers/usb/host/ohci-pci.c | 2 + drivers/usb/host/uhci-hcd.c | 2 + 5 files changed, 115 insertions(+), 95 deletions(-) diff --git a/drivers/usb/core/hcd-pci.c b/drivers/usb/core/hcd-pci.c index 5b87ae7f0a6..99432785f43 100644 --- a/drivers/usb/core/hcd-pci.c +++ b/drivers/usb/core/hcd-pci.c @@ -191,17 +191,15 @@ EXPORT_SYMBOL_GPL(usb_hcd_pci_remove); /** * usb_hcd_pci_suspend - power management suspend of a PCI-based HCD * @dev: USB Host Controller being suspended - * @message: semantics in flux + * @message: Power Management message describing this state transition * - * Store this function in the HCD's struct pci_driver as suspend(). + * Store this function in the HCD's struct pci_driver as .suspend. */ int usb_hcd_pci_suspend(struct pci_dev *dev, pm_message_t message) { - struct usb_hcd *hcd; + struct usb_hcd *hcd = pci_get_drvdata(dev); int retval = 0; - int has_pci_pm; - - hcd = pci_get_drvdata(dev); + int wake, w; /* Root hub suspend should have stopped all downstream traffic, * and all bus master traffic. And done so for both the interface @@ -212,8 +210,15 @@ int usb_hcd_pci_suspend(struct pci_dev *dev, pm_message_t message) * otherwise the swsusp will save (and restore) garbage state. */ if (!(hcd->state == HC_STATE_SUSPENDED || - hcd->state == HC_STATE_HALT)) - return -EBUSY; + hcd->state == HC_STATE_HALT)) { + dev_warn(&dev->dev, "Root hub is not suspended\n"); + retval = -EBUSY; + goto done; + } + + /* We might already be suspended (runtime PM -- not yet written) */ + if (dev->current_state != PCI_D0) + goto done; if (hcd->driver->pci_suspend) { retval = hcd->driver->pci_suspend(hcd, message); @@ -221,49 +226,60 @@ int usb_hcd_pci_suspend(struct pci_dev *dev, pm_message_t message) if (retval) goto done; } + synchronize_irq(dev->irq); - /* FIXME until the generic PM interfaces change a lot more, this - * can't use PCI D1 and D2 states. For example, the confusion - * between messages and states will need to vanish, and messages - * will need to provide a target system state again. - * - * It'll be important to learn characteristics of the target state, - * especially on embedded hardware where the HCD will often be in - * charge of an external VBUS power supply and one or more clocks. - * Some target system states will leave them active; others won't. - * (With PCI, that's often handled by platform BIOS code.) + /* Don't fail on error to enable wakeup. We rely on pci code + * to reject requests the hardware can't implement, rather + * than coding the same thing. */ - - /* even when the PCI layer rejects some of the PCI calls - * below, HCs can try global suspend and reduce DMA traffic. - * PM-sensitive HCDs may already have done this. - */ - has_pci_pm = pci_find_capability(dev, PCI_CAP_ID_PM); + wake = (hcd->state == HC_STATE_SUSPENDED && + device_may_wakeup(&dev->dev)); + w = pci_wake_from_d3(dev, wake); + if (w < 0) + wake = w; + dev_dbg(&dev->dev, "wakeup: %d\n", wake); /* Downstream ports from this root hub should already be quiesced, so * there will be no DMA activity. Now we can shut down the upstream * link (except maybe for PME# resume signaling) and enter some PCI * low power state, if the hardware allows. */ - if (hcd->state == HC_STATE_SUSPENDED) { + pci_disable_device(dev); + done: + return retval; +} +EXPORT_SYMBOL_GPL(usb_hcd_pci_suspend); - /* no DMA or IRQs except when HC is active */ - if (dev->current_state == PCI_D0) { - pci_save_state(dev); - pci_disable_device(dev); - } +/** + * usb_hcd_pci_suspend_late - suspend a PCI-based HCD after IRQs are disabled + * @dev: USB Host Controller being suspended + * @message: Power Management message describing this state transition + * + * Store this function in the HCD's struct pci_driver as .suspend_late. + */ +int usb_hcd_pci_suspend_late(struct pci_dev *dev, pm_message_t message) +{ + int retval = 0; + int has_pci_pm; - if (message.event == PM_EVENT_FREEZE || - message.event == PM_EVENT_PRETHAW) { - dev_dbg(hcd->self.controller, "--> no state change\n"); - goto done; - } + /* We might already be suspended (runtime PM -- not yet written) */ + if (dev->current_state != PCI_D0) + goto done; - if (!has_pci_pm) { - dev_dbg(hcd->self.controller, "--> PCI D0/legacy\n"); - goto done; - } + pci_save_state(dev); + + /* Don't change state if we don't need to */ + if (message.event == PM_EVENT_FREEZE || + message.event == PM_EVENT_PRETHAW) { + dev_dbg(&dev->dev, "--> no state change\n"); + goto done; + } + + has_pci_pm = pci_find_capability(dev, PCI_CAP_ID_PM); + if (!has_pci_pm) { + dev_dbg(&dev->dev, "--> PCI D0 legacy\n"); + } else { /* NOTE: dev->current_state becomes nonzero only here, and * only for devices that support PCI PM. Also, exiting @@ -273,35 +289,16 @@ int usb_hcd_pci_suspend(struct pci_dev *dev, pm_message_t message) retval = pci_set_power_state(dev, PCI_D3hot); suspend_report_result(pci_set_power_state, retval); if (retval == 0) { - int wake = device_can_wakeup(&hcd->self.root_hub->dev); - - wake = wake && device_may_wakeup(hcd->self.controller); - - dev_dbg(hcd->self.controller, "--> PCI D3%s\n", - wake ? "/wakeup" : ""); - - /* Ignore these return values. We rely on pci code to - * reject requests the hardware can't implement, rather - * than coding the same thing. - */ - (void) pci_enable_wake(dev, PCI_D3hot, wake); - (void) pci_enable_wake(dev, PCI_D3cold, wake); + dev_dbg(&dev->dev, "--> PCI D3\n"); } else { dev_dbg(&dev->dev, "PCI D3 suspend fail, %d\n", retval); - (void) usb_hcd_pci_resume(dev); + pci_restore_state(dev); } - - } else if (hcd->state != HC_STATE_HALT) { - dev_dbg(hcd->self.controller, "hcd state %d; not suspended\n", - hcd->state); - WARN_ON(1); - retval = -EINVAL; } -done: - if (retval == 0) { #ifdef CONFIG_PPC_PMAC + if (retval == 0) { /* Disable ASIC clocks for USB */ if (machine_is(powermac)) { struct device_node *of_node; @@ -311,30 +308,24 @@ done: pmac_call_feature(PMAC_FTR_USB_ENABLE, of_node, 0, 0); } -#endif } +#endif + done: return retval; } -EXPORT_SYMBOL_GPL(usb_hcd_pci_suspend); +EXPORT_SYMBOL_GPL(usb_hcd_pci_suspend_late); /** - * usb_hcd_pci_resume - power management resume of a PCI-based HCD + * usb_hcd_pci_resume_early - resume a PCI-based HCD before IRQs are enabled * @dev: USB Host Controller being resumed * - * Store this function in the HCD's struct pci_driver as resume(). + * Store this function in the HCD's struct pci_driver as .resume_early. */ -int usb_hcd_pci_resume(struct pci_dev *dev) +int usb_hcd_pci_resume_early(struct pci_dev *dev) { - struct usb_hcd *hcd; - int retval; - - hcd = pci_get_drvdata(dev); - if (hcd->state != HC_STATE_SUSPENDED) { - dev_dbg(hcd->self.controller, - "can't resume, not suspended!\n"); - return 0; - } + int retval = 0; + pci_power_t state = dev->current_state; #ifdef CONFIG_PPC_PMAC /* Reenable ASIC clocks for USB */ @@ -352,7 +343,7 @@ int usb_hcd_pci_resume(struct pci_dev *dev) * calls "standby", "suspend to RAM", and so on). There are also * dirty cases when swsusp fakes a suspend in "shutdown" mode. */ - if (dev->current_state != PCI_D0) { + if (state != PCI_D0) { #ifdef DEBUG int pci_pm; u16 pmcr; @@ -364,8 +355,7 @@ int usb_hcd_pci_resume(struct pci_dev *dev) /* Clean case: power to USB and to HC registers was * maintained; remote wakeup is easy. */ - dev_dbg(hcd->self.controller, "resume from PCI D%d\n", - pmcr); + dev_dbg(&dev->dev, "resume from PCI D%d\n", pmcr); } else { /* Clean: HC lost Vcc power, D0 uninitialized * + Vaux may have preserved port and transceiver @@ -376,32 +366,55 @@ int usb_hcd_pci_resume(struct pci_dev *dev) * + after BIOS init * + after Linux init (HCD statically linked) */ - dev_dbg(hcd->self.controller, - "PCI D0, from previous PCI D%d\n", - dev->current_state); + dev_dbg(&dev->dev, "resume from previous PCI D%d\n", + state); } #endif - /* yes, ignore these results too... */ - (void) pci_enable_wake(dev, dev->current_state, 0); - (void) pci_enable_wake(dev, PCI_D3cold, 0); + + retval = pci_set_power_state(dev, PCI_D0); } else { /* Same basic cases: clean (powered/not), dirty */ - dev_dbg(hcd->self.controller, "PCI legacy resume\n"); + dev_dbg(&dev->dev, "PCI legacy resume\n"); + } + + if (retval < 0) + dev_err(&dev->dev, "can't resume: %d\n", retval); + else + pci_restore_state(dev); + + return retval; +} +EXPORT_SYMBOL_GPL(usb_hcd_pci_resume_early); + +/** + * usb_hcd_pci_resume - power management resume of a PCI-based HCD + * @dev: USB Host Controller being resumed + * + * Store this function in the HCD's struct pci_driver as .resume. + */ +int usb_hcd_pci_resume(struct pci_dev *dev) +{ + struct usb_hcd *hcd; + int retval; + + hcd = pci_get_drvdata(dev); + if (hcd->state != HC_STATE_SUSPENDED) { + dev_dbg(hcd->self.controller, + "can't resume, not suspended!\n"); + return 0; } - /* NOTE: the PCI API itself is asymmetric here. We don't need to - * pci_set_power_state(PCI_D0) since that's part of re-enabling; - * but that won't re-enable bus mastering. Yet pci_disable_device() - * explicitly disables bus mastering... - */ retval = pci_enable_device(dev); if (retval < 0) { - dev_err(hcd->self.controller, - "can't re-enable after resume, %d!\n", retval); + dev_err(&dev->dev, "can't re-enable after resume, %d!\n", + retval); return retval; } + pci_set_master(dev); - pci_restore_state(dev); + + /* yes, ignore this result too... */ + (void) pci_wake_from_d3(dev, 0); clear_bit(HCD_FLAG_SAW_IRQ, &hcd->flags); @@ -413,7 +426,6 @@ int usb_hcd_pci_resume(struct pci_dev *dev) usb_hc_died(hcd); } } - return retval; } EXPORT_SYMBOL_GPL(usb_hcd_pci_resume); diff --git a/drivers/usb/core/hcd.h b/drivers/usb/core/hcd.h index aa5da82d907..572d2cf46e8 100644 --- a/drivers/usb/core/hcd.h +++ b/drivers/usb/core/hcd.h @@ -256,7 +256,9 @@ extern int usb_hcd_pci_probe(struct pci_dev *dev, extern void usb_hcd_pci_remove(struct pci_dev *dev); #ifdef CONFIG_PM -extern int usb_hcd_pci_suspend(struct pci_dev *dev, pm_message_t state); +extern int usb_hcd_pci_suspend(struct pci_dev *dev, pm_message_t msg); +extern int usb_hcd_pci_suspend_late(struct pci_dev *dev, pm_message_t msg); +extern int usb_hcd_pci_resume_early(struct pci_dev *dev); extern int usb_hcd_pci_resume(struct pci_dev *dev); #endif /* CONFIG_PM */ diff --git a/drivers/usb/host/ehci-pci.c b/drivers/usb/host/ehci-pci.c index 36864f95844..6af47a0937b 100644 --- a/drivers/usb/host/ehci-pci.c +++ b/drivers/usb/host/ehci-pci.c @@ -428,6 +428,8 @@ static struct pci_driver ehci_pci_driver = { #ifdef CONFIG_PM .suspend = usb_hcd_pci_suspend, + .suspend_late = usb_hcd_pci_suspend_late, + .resume_early = usb_hcd_pci_resume_early, .resume = usb_hcd_pci_resume, #endif .shutdown = usb_hcd_pci_shutdown, diff --git a/drivers/usb/host/ohci-pci.c b/drivers/usb/host/ohci-pci.c index a9c2ae36c7a..8380cc2e961 100644 --- a/drivers/usb/host/ohci-pci.c +++ b/drivers/usb/host/ohci-pci.c @@ -487,6 +487,8 @@ static struct pci_driver ohci_pci_driver = { #ifdef CONFIG_PM .suspend = usb_hcd_pci_suspend, + .suspend_late = usb_hcd_pci_suspend_late, + .resume_early = usb_hcd_pci_resume_early, .resume = usb_hcd_pci_resume, #endif diff --git a/drivers/usb/host/uhci-hcd.c b/drivers/usb/host/uhci-hcd.c index cf5e4cf7ea4..4e221060f58 100644 --- a/drivers/usb/host/uhci-hcd.c +++ b/drivers/usb/host/uhci-hcd.c @@ -942,6 +942,8 @@ static struct pci_driver uhci_pci_driver = { #ifdef CONFIG_PM .suspend = usb_hcd_pci_suspend, + .suspend_late = usb_hcd_pci_suspend_late, + .resume_early = usb_hcd_pci_resume_early, .resume = usb_hcd_pci_resume, #endif /* PM */ }; From 6fd9086a518d4f14213a32fe6c9ac17fabebbc1e Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Wed, 17 Dec 2008 17:20:38 -0500 Subject: [PATCH 085/123] USB: automatically enable wakeup for PCI host controllers This patch (as1193b) enables wakeup during initialization for all PCI host controllers, and it removes some code (and comments!) that are no longer needed now that the PCI core automatically initializes wakeup settings for all new devices. The idea is that the bus should initialize wakeup, and the bus glue or controller driver should enable it. Signed-off-by: Alan Stern Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/hcd-pci.c | 1 + drivers/usb/host/ehci-pci.c | 10 +++++++--- drivers/usb/host/ohci-hcd.c | 12 +++++++----- drivers/usb/host/ohci-pci.c | 4 ++-- 4 files changed, 17 insertions(+), 10 deletions(-) diff --git a/drivers/usb/core/hcd-pci.c b/drivers/usb/core/hcd-pci.c index 99432785f43..507741ed448 100644 --- a/drivers/usb/core/hcd-pci.c +++ b/drivers/usb/core/hcd-pci.c @@ -128,6 +128,7 @@ int usb_hcd_pci_probe(struct pci_dev *dev, const struct pci_device_id *id) } pci_set_master(dev); + device_set_wakeup_enable(&dev->dev, 1); retval = usb_add_hcd(hcd, dev->irq, IRQF_DISABLED | IRQF_SHARED); if (retval != 0) diff --git a/drivers/usb/host/ehci-pci.c b/drivers/usb/host/ehci-pci.c index 6af47a0937b..bdc6e86e1f8 100644 --- a/drivers/usb/host/ehci-pci.c +++ b/drivers/usb/host/ehci-pci.c @@ -219,15 +219,19 @@ static int ehci_pci_setup(struct usb_hcd *hcd) /* Serial Bus Release Number is at PCI 0x60 offset */ pci_read_config_byte(pdev, 0x60, &ehci->sbrn); - /* Workaround current PCI init glitch: wakeup bits aren't - * being set from PCI PM capability. + /* Keep this around for a while just in case some EHCI + * implementation uses legacy PCI PM support. This test + * can be removed on 17 Dec 2009 if the dev_warn() hasn't + * been triggered by then. */ if (!device_can_wakeup(&pdev->dev)) { u16 port_wake; pci_read_config_word(pdev, 0x62, &port_wake); - if (port_wake & 0x0001) + if (port_wake & 0x0001) { + dev_warn(&pdev->dev, "Enabling legacy PCI PM\n"); device_init_wakeup(&pdev->dev, 1); + } } #ifdef CONFIG_USB_SUSPEND diff --git a/drivers/usb/host/ohci-hcd.c b/drivers/usb/host/ohci-hcd.c index 8aa3f4556a3..65a9609f4ad 100644 --- a/drivers/usb/host/ohci-hcd.c +++ b/drivers/usb/host/ohci-hcd.c @@ -589,13 +589,15 @@ static int ohci_run (struct ohci_hcd *ohci) /* also: power/overcurrent flags in roothub.a */ } - /* Reset USB nearly "by the book". RemoteWakeupConnected was - * saved if boot firmware (BIOS/SMM/...) told us it's connected, - * or if bus glue did the same (e.g. for PCI add-in cards with - * PCI PM support). + /* Reset USB nearly "by the book". RemoteWakeupConnected has + * to be checked in case boot firmware (BIOS/SMM/...) has set up + * wakeup in a way the bus isn't aware of (e.g., legacy PCI PM). + * If the bus glue detected wakeup capability then it should + * already be enabled. Either way, if wakeup should be enabled + * but isn't, we'll enable it now. */ if ((ohci->hc_control & OHCI_CTRL_RWC) != 0 - && !device_may_wakeup(hcd->self.controller)) + && !device_can_wakeup(hcd->self.controller)) device_init_wakeup(hcd->self.controller, 1); switch (ohci->hc_control & OHCI_CTRL_HCFS) { diff --git a/drivers/usb/host/ohci-pci.c b/drivers/usb/host/ohci-pci.c index 8380cc2e961..8b28ae7865b 100644 --- a/drivers/usb/host/ohci-pci.c +++ b/drivers/usb/host/ohci-pci.c @@ -355,9 +355,9 @@ static int __devinit ohci_pci_start (struct usb_hcd *hcd) /* RWC may not be set for add-in PCI cards, since boot * firmware probably ignored them. This transfers PCI - * PM wakeup capabilities (once the PCI layer is fixed). + * PM wakeup capabilities. */ - if (device_may_wakeup(&pdev->dev)) + if (device_can_wakeup(&pdev->dev)) ohci->hc_control |= OHCI_CTRL_RWC; } #endif /* CONFIG_PM */ From df718962bf91c7bd345060aadaa24b03f6140b07 Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Fri, 19 Dec 2008 10:27:56 -0500 Subject: [PATCH 086/123] USB: cancel pending Set-Config requests if userspace gets there first This patch (as1195) eliminates a potential problem identified by Oliver Neukum. When a driver queues an asynchronous Set-Config request using usb_driver_set_configuration(), the request should be cancelled if userspace changes the configuration first. The patch introduces a linked list of pending async Set-Config requests, and uses it to invalidate the requests for a particular device whenever that device's configuration is set. Signed-off-by: Alan Stern Cc: Oliver Neukum Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/message.c | 42 ++++++++++++++++++++++++++++++++++---- 1 file changed, 38 insertions(+), 4 deletions(-) diff --git a/drivers/usb/core/message.c b/drivers/usb/core/message.c index 7943901c641..5589686981f 100644 --- a/drivers/usb/core/message.c +++ b/drivers/usb/core/message.c @@ -18,6 +18,8 @@ #include "hcd.h" /* for usbcore internals */ #include "usb.h" +static void cancel_async_set_config(struct usb_device *udev); + struct api_context { struct completion done; int status; @@ -1636,6 +1638,9 @@ free_interfaces: if (dev->state != USB_STATE_ADDRESS) usb_disable_device(dev, 1); /* Skip ep0 */ + /* Get rid of pending async Set-Config requests for this device */ + cancel_async_set_config(dev); + ret = usb_control_msg(dev, usb_sndctrlpipe(dev, 0), USB_REQ_SET_CONFIGURATION, 0, configuration, 0, NULL, 0, USB_CTRL_SET_TIMEOUT); @@ -1725,10 +1730,14 @@ free_interfaces: return 0; } +static LIST_HEAD(set_config_list); +static DEFINE_SPINLOCK(set_config_lock); + struct set_config_request { struct usb_device *udev; int config; struct work_struct work; + struct list_head node; }; /* Worker routine for usb_driver_set_configuration() */ @@ -1736,14 +1745,35 @@ static void driver_set_config_work(struct work_struct *work) { struct set_config_request *req = container_of(work, struct set_config_request, work); + struct usb_device *udev = req->udev; - usb_lock_device(req->udev); - usb_set_configuration(req->udev, req->config); - usb_unlock_device(req->udev); - usb_put_dev(req->udev); + usb_lock_device(udev); + spin_lock(&set_config_lock); + list_del(&req->node); + spin_unlock(&set_config_lock); + + if (req->config >= -1) /* Is req still valid? */ + usb_set_configuration(udev, req->config); + usb_unlock_device(udev); + usb_put_dev(udev); kfree(req); } +/* Cancel pending Set-Config requests for a device whose configuration + * was just changed + */ +static void cancel_async_set_config(struct usb_device *udev) +{ + struct set_config_request *req; + + spin_lock(&set_config_lock); + list_for_each_entry(req, &set_config_list, node) { + if (req->udev == udev) + req->config = -999; /* Mark as cancelled */ + } + spin_unlock(&set_config_lock); +} + /** * usb_driver_set_configuration - Provide a way for drivers to change device configurations * @udev: the device whose configuration is being updated @@ -1775,6 +1805,10 @@ int usb_driver_set_configuration(struct usb_device *udev, int config) req->config = config; INIT_WORK(&req->work, driver_set_config_work); + spin_lock(&set_config_lock); + list_add(&req->node, &set_config_list); + spin_unlock(&set_config_lock); + usb_get_dev(udev); schedule_work(&req->work); return 0; From 2caf7fcdb8532045680f06b67b9e63f0c9613aaa Mon Sep 17 00:00:00 2001 From: Alan Stern Date: Wed, 31 Dec 2008 11:31:33 -0500 Subject: [PATCH 087/123] USB: re-enable interface after driver unbinds This patch (as1197) fixes an error introduced recently. Since a significant number of devices can't handle Set-Interface requests, we no longer call usb_set_interface() when a driver unbinds from an interface, provided the interface is already in altsetting 0. However the interface still does get disabled, and the call to usb_set_interface() was the only thing re-enabling it. Since the interface doesn't get re-enabled, further attempts to use it fail. So the patch adds a call to usb_enable_interface() when a driver unbinds and the interface is in altsetting 0. For this to work right, the interface's endpoints have to be re-enabled but their toggles have to be left alone. Therefore an additional argument is added to usb_enable_endpoint() and usb_enable_interface(), a flag indicating whether or not the endpoint toggles should be reset. This is a forward-ported version of a patch which fixes Bugzilla #12301. Signed-off-by: Alan Stern Reported-by: David Roka Reported-by: Erik Ekman Tested-by: Erik Ekman Tested-by: Alon Bar-Lev Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/driver.c | 9 ++++++--- drivers/usb/core/hub.c | 2 +- drivers/usb/core/message.c | 25 +++++++++++++++---------- drivers/usb/core/usb.c | 2 +- drivers/usb/core/usb.h | 4 +++- 5 files changed, 26 insertions(+), 16 deletions(-) diff --git a/drivers/usb/core/driver.c b/drivers/usb/core/driver.c index 41c06025506..98760553bc9 100644 --- a/drivers/usb/core/driver.c +++ b/drivers/usb/core/driver.c @@ -295,9 +295,12 @@ static int usb_unbind_interface(struct device *dev) * altsetting means creating new endpoint device entries). * When either of these happens, defer the Set-Interface. */ - if (intf->cur_altsetting->desc.bAlternateSetting == 0) - ; /* Already in altsetting 0 so skip Set-Interface */ - else if (!error && intf->dev.power.status == DPM_ON) + if (intf->cur_altsetting->desc.bAlternateSetting == 0) { + /* Already in altsetting 0 so skip Set-Interface. + * Just re-enable it without affecting the endpoint toggles. + */ + usb_enable_interface(udev, intf, false); + } else if (!error && intf->dev.power.status == DPM_ON) usb_set_interface(udev, intf->altsetting[0]. desc.bInterfaceNumber, 0); else diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c index 756b8d9993f..d5d0e40b1e2 100644 --- a/drivers/usb/core/hub.c +++ b/drivers/usb/core/hub.c @@ -2384,7 +2384,7 @@ void usb_ep0_reinit(struct usb_device *udev) { usb_disable_endpoint(udev, 0 + USB_DIR_IN); usb_disable_endpoint(udev, 0 + USB_DIR_OUT); - usb_enable_endpoint(udev, &udev->ep0); + usb_enable_endpoint(udev, &udev->ep0, true); } EXPORT_SYMBOL_GPL(usb_ep0_reinit); diff --git a/drivers/usb/core/message.c b/drivers/usb/core/message.c index 5589686981f..de51667dd64 100644 --- a/drivers/usb/core/message.c +++ b/drivers/usb/core/message.c @@ -1143,22 +1143,26 @@ void usb_disable_device(struct usb_device *dev, int skip_ep0) * usb_enable_endpoint - Enable an endpoint for USB communications * @dev: the device whose interface is being enabled * @ep: the endpoint + * @reset_toggle: flag to set the endpoint's toggle back to 0 * - * Resets the endpoint toggle, and sets dev->ep_{in,out} pointers. + * Resets the endpoint toggle if asked, and sets dev->ep_{in,out} pointers. * For control endpoints, both the input and output sides are handled. */ -void usb_enable_endpoint(struct usb_device *dev, struct usb_host_endpoint *ep) +void usb_enable_endpoint(struct usb_device *dev, struct usb_host_endpoint *ep, + bool reset_toggle) { int epnum = usb_endpoint_num(&ep->desc); int is_out = usb_endpoint_dir_out(&ep->desc); int is_control = usb_endpoint_xfer_control(&ep->desc); if (is_out || is_control) { - usb_settoggle(dev, epnum, 1, 0); + if (reset_toggle) + usb_settoggle(dev, epnum, 1, 0); dev->ep_out[epnum] = ep; } if (!is_out || is_control) { - usb_settoggle(dev, epnum, 0, 0); + if (reset_toggle) + usb_settoggle(dev, epnum, 0, 0); dev->ep_in[epnum] = ep; } ep->enabled = 1; @@ -1168,17 +1172,18 @@ void usb_enable_endpoint(struct usb_device *dev, struct usb_host_endpoint *ep) * usb_enable_interface - Enable all the endpoints for an interface * @dev: the device whose interface is being enabled * @intf: pointer to the interface descriptor + * @reset_toggles: flag to set the endpoints' toggles back to 0 * * Enables all the endpoints for the interface's current altsetting. */ -static void usb_enable_interface(struct usb_device *dev, - struct usb_interface *intf) +void usb_enable_interface(struct usb_device *dev, + struct usb_interface *intf, bool reset_toggles) { struct usb_host_interface *alt = intf->cur_altsetting; int i; for (i = 0; i < alt->desc.bNumEndpoints; ++i) - usb_enable_endpoint(dev, &alt->endpoint[i]); + usb_enable_endpoint(dev, &alt->endpoint[i], reset_toggles); } /** @@ -1303,7 +1308,7 @@ int usb_set_interface(struct usb_device *dev, int interface, int alternate) * during the SETUP stage - hence EP0 toggles are "don't care" here. * (Likewise, EP0 never "halts" on well designed devices.) */ - usb_enable_interface(dev, iface); + usb_enable_interface(dev, iface, true); if (device_is_registered(&iface->dev)) { usb_create_sysfs_intf_files(iface); create_intf_ep_devs(iface); @@ -1382,7 +1387,7 @@ int usb_reset_configuration(struct usb_device *dev) usb_remove_sysfs_intf_files(intf); } intf->cur_altsetting = alt; - usb_enable_interface(dev, intf); + usb_enable_interface(dev, intf, true); if (device_is_registered(&intf->dev)) { usb_create_sysfs_intf_files(intf); create_intf_ep_devs(intf); @@ -1685,7 +1690,7 @@ free_interfaces: alt = &intf->altsetting[0]; intf->cur_altsetting = alt; - usb_enable_interface(dev, intf); + usb_enable_interface(dev, intf, true); intf->dev.parent = &dev->dev; intf->dev.driver = NULL; intf->dev.bus = &usb_bus_type; diff --git a/drivers/usb/core/usb.c b/drivers/usb/core/usb.c index c0821564a3f..dcfc072630c 100644 --- a/drivers/usb/core/usb.c +++ b/drivers/usb/core/usb.c @@ -362,7 +362,7 @@ struct usb_device *usb_alloc_dev(struct usb_device *parent, dev->ep0.desc.bLength = USB_DT_ENDPOINT_SIZE; dev->ep0.desc.bDescriptorType = USB_DT_ENDPOINT; /* ep0 maxpacket comes later, from device descriptor */ - usb_enable_endpoint(dev, &dev->ep0); + usb_enable_endpoint(dev, &dev->ep0, true); dev->can_submit = 1; /* Save readable and stable topology id, distinguishing devices diff --git a/drivers/usb/core/usb.h b/drivers/usb/core/usb.h index 381eae90c3b..386177867a8 100644 --- a/drivers/usb/core/usb.h +++ b/drivers/usb/core/usb.h @@ -12,7 +12,9 @@ extern int usb_create_ep_devs(struct device *parent, extern void usb_remove_ep_devs(struct usb_host_endpoint *endpoint); extern void usb_enable_endpoint(struct usb_device *dev, - struct usb_host_endpoint *ep); + struct usb_host_endpoint *ep, bool reset_toggle); +extern void usb_enable_interface(struct usb_device *dev, + struct usb_interface *intf, bool reset_toggles); extern void usb_disable_endpoint(struct usb_device *dev, unsigned int epaddr); extern void usb_disable_interface(struct usb_device *dev, struct usb_interface *intf); From ed0c7720d23d5c82787e17cb02b28ca9eb11853d Mon Sep 17 00:00:00 2001 From: Mark Lord Date: Fri, 2 Jan 2009 02:48:24 -0500 Subject: [PATCH 088/123] USB: fix minor nit in usbfs checking One minor nit did show up, though. The patch below seems to make more sense than the code does without it. Signed-off-by: Mark Lord Signed-off-by: Greg Kroah-Hartman --- drivers/usb/core/devio.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/core/devio.c b/drivers/usb/core/devio.c index c85e29381a8..26fece124e0 100644 --- a/drivers/usb/core/devio.c +++ b/drivers/usb/core/devio.c @@ -1317,7 +1317,7 @@ static int get_urb32(struct usbdevfs_urb *kurb, if (__get_user(uptr, &uurb->buffer)) return -EFAULT; kurb->buffer = compat_ptr(uptr); - if (__get_user(uptr, &uurb->buffer)) + if (__get_user(uptr, &uurb->usercontext)) return -EFAULT; kurb->usercontext = compat_ptr(uptr); From 32e7fea282b3765c64b935fa54f20ace437550a0 Mon Sep 17 00:00:00 2001 From: SangSu Park Date: Fri, 2 Jan 2009 01:11:14 +0900 Subject: [PATCH 089/123] USB: Fix goku_udc usb speed handling The usb gadget framework revealed weakness in the godu_udc gadget driver register function. Instead of checking if speed asked for was USB_LOW_SPEED upon usb_gadget_register() to deny service, it checked only for USB_FULL_SPEED, thus denying service to usb high speed capable gadgets. Signed-off-by: SangSu Park Acked-by: David Brownell Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/goku_udc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/gadget/goku_udc.c b/drivers/usb/gadget/goku_udc.c index 60aa04847b1..63419c4d503 100644 --- a/drivers/usb/gadget/goku_udc.c +++ b/drivers/usb/gadget/goku_udc.c @@ -1349,7 +1349,7 @@ int usb_gadget_register_driver(struct usb_gadget_driver *driver) int retval; if (!driver - || driver->speed != USB_SPEED_FULL + || driver->speed < USB_SPEED_FULL || !driver->bind || !driver->disconnect || !driver->setup) From 327d74f6b65ddc8a042c43c11fdd4be0bb354668 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 4 Jan 2009 13:25:13 +0100 Subject: [PATCH 090/123] USB: emi26: fix oops on load Fix oops introduced by commit ae93a55bf948753de0bb8e43fa9c027f786abb05 (emi26: use request_firmware()): usb 1-1: new full speed USB device using uhci_hcd and address 2 usb 1-1: configuration #1 chosen from 1 choice emi26 - firmware loader 1-1:1.0: emi26_probe start usb 1-1: firmware: requesting emi26/loader.fw usb 1-1: firmware: requesting emi26/bitstream.fw usb 1-1: firmware: requesting emi26/firmware.fw usb 1-1: emi26_set_reset - 1 usb 1-1: emi26_set_reset - 0 BUG: unable to handle kernel NULL pointer dereference at 00000000 IP: [] emi26_probe+0x2f7/0x620 [emi26] *pde = 00000000 Oops: 0000 [#1] SMP last sysfs file: /sys/devices/pci0000:00/0000:00:1d.0/usb1/1-1/firmware/1-1/loading Modules linked in: emi26(+) ipv6 cpufreq_ondemand coretemp arc4 ecb iwl3945 irtty_sir sir_dev nsc_ircc ehci_hcd uhci_hcd mac80211 irda usbcore snd_hda_intel thinkpad_acpi rfkill hwmon led_class e1000e snd_pcm cfg80211 snd_timer crc_ccitt snd snd_page_alloc aes_generic Pid: 5082, comm: modprobe Not tainted (2.6.28 #2) 17023QG EIP: 0060:[] EFLAGS: 00010206 CPU: 0 EIP is at emi26_probe+0x2f7/0x620 [emi26] EAX: 0000015c EBX: 00000000 ECX: c1ffd9c0 EDX: 00000000 ESI: 0000015c EDI: f6bb215c EBP: f6bb0400 ESP: f00ebcfc DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 Process modprobe (pid: 5082, ti=f00ea000 task=f5c7c700 task.ti=f00ea000) Stack: 0000015c 000000a5 f6a67cb8 f80dc7e0 c01c6262 fbef2986 f6bb2000 00008fe0 0000015c f715f748 f715f740 f715f738 f715f748 f6a67c00 f80dd040 f80dcfc0 f6bb0400 fbacb290 f6a67c94 fbae0160 c01c70bf 00000000 f6a67c1c 00000000 Call Trace: [] sysfs_add_one+0x12/0x50 [] usb_probe_interface+0xa0/0x140 [usbcore] [] sysfs_create_link+0xf/0x20 [] driver_probe_device+0x82/0x180 [] usb_match_id+0x3b/0x50 [usbcore] [] __driver_attach+0x7e/0x80 [] bus_for_each_dev+0x3a/0x60 [] driver_attach+0x16/0x20 [] __driver_attach+0x0/0x80 [] bus_add_driver+0x1a1/0x220 [] driver_register+0x4d/0x120 [] idr_get_empty_slot+0xf2/0x290 [] usb_register_driver+0x81/0x100 [usbcore] [] emi26_init+0x0/0x14 [emi26] [] do_one_initcall+0x36/0x1b0 [] sysfs_ilookup_test+0x0/0x10 [] ifind+0x31/0x90 [] __sysfs_add_one+0x59/0x80 [] sysfs_addrm_finish+0x14/0x1c0 [] __vunmap+0xa3/0xd0 [] load_module+0x1544/0x1640 [] sys_init_module+0x87/0x1b0 [] sys_read+0x41/0x70 [] sysenter_do_call+0x12/0x21 [] wait_for_common+0x40/0x110 Code: 66 c1 e8 08 66 09 d0 75 a5 31 d2 89 e8 e8 72 fc ff ff 85 c0 0f 88 9a 02 00 00 b8 fa 00 00 00 e8 30 46 05 c8 8b 74 24 28 8b 5e 04 <8b> 03 89 44 24 1c 0f c8 89 44 24 1c 0f b7 4b 04 c7 44 24 20 00 EIP: [] emi26_probe+0x2f7/0x620 [emi26] SS:ESP 0068:f00ebcfc ---[ end trace 2eefa13825431230 ]--- After the last "package" of firmware data is sent to the device, we dereference NULL pointer (on access to rec->addr). Fix it. Reported--by: David Flatz Tested-by: David Flatz Signed-off-by: Marcin Slusarz Cc: David Woodhouse Cc: stable [2.6.27, 2.6.28] Signed-off-by: Greg Kroah-Hartman --- drivers/usb/misc/emi26.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/usb/misc/emi26.c b/drivers/usb/misc/emi26.c index e762beb5f3c..879a980ca8c 100644 --- a/drivers/usb/misc/emi26.c +++ b/drivers/usb/misc/emi26.c @@ -160,7 +160,7 @@ static int emi26_load_firmware (struct usb_device *dev) err("%s - error loading firmware: error = %d", __func__, err); goto wraperr; } - } while (i > 0); + } while (rec); /* Assert reset (stop the CPU in the EMI) */ err = emi26_set_reset(dev,1); From 57262b82d601c5ca8e3db219aebd332950f62ba1 Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Mon, 3 Nov 2008 13:27:03 -0800 Subject: [PATCH 091/123] USB: add new opticon serial driver This is for the serial mode of the Opticon barcode scanner. Cc: Kees Stoop Signed-off-by: Greg Kroah-Hartman --- drivers/usb/serial/Kconfig | 9 + drivers/usb/serial/Makefile | 1 + drivers/usb/serial/opticon.c | 358 +++++++++++++++++++++++++++++++++++ 3 files changed, 368 insertions(+) create mode 100644 drivers/usb/serial/opticon.c diff --git a/drivers/usb/serial/Kconfig b/drivers/usb/serial/Kconfig index 70338f4ec91..da30784a0ef 100644 --- a/drivers/usb/serial/Kconfig +++ b/drivers/usb/serial/Kconfig @@ -565,6 +565,15 @@ config USB_SERIAL_OMNINET To compile this driver as a module, choose M here: the module will be called omninet. +config USB_SERIAL_OPTICON + tristate "USB Opticon Barcode driver (serial mode)" + help + Say Y here if you want to use a Opticon USB Barcode device + in serial emulation mode. + + To compile this driver as a module, choose M here: the + module will be called opticon. + config USB_SERIAL_DEBUG tristate "USB Debugging Device" help diff --git a/drivers/usb/serial/Makefile b/drivers/usb/serial/Makefile index 6047f818adf..9a4f59e0ee5 100644 --- a/drivers/usb/serial/Makefile +++ b/drivers/usb/serial/Makefile @@ -41,6 +41,7 @@ obj-$(CONFIG_USB_SERIAL_MOS7840) += mos7840.o obj-$(CONFIG_USB_SERIAL_MOTOROLA) += moto_modem.o obj-$(CONFIG_USB_SERIAL_NAVMAN) += navman.o obj-$(CONFIG_USB_SERIAL_OMNINET) += omninet.o +obj-$(CONFIG_USB_SERIAL_OPTICON) += opticon.o obj-$(CONFIG_USB_SERIAL_OPTION) += option.o obj-$(CONFIG_USB_SERIAL_OTI6858) += oti6858.o obj-$(CONFIG_USB_SERIAL_PL2303) += pl2303.o diff --git a/drivers/usb/serial/opticon.c b/drivers/usb/serial/opticon.c new file mode 100644 index 00000000000..cea326f1f10 --- /dev/null +++ b/drivers/usb/serial/opticon.c @@ -0,0 +1,358 @@ +/* + * Opticon USB barcode to serial driver + * + * Copyright (C) 2008 Greg Kroah-Hartman + * Copyright (C) 2008 Novell Inc. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +static int debug; + +static struct usb_device_id id_table[] = { + { USB_DEVICE(0x065a, 0x0009) }, + { }, +}; +MODULE_DEVICE_TABLE(usb, id_table); + +/* This structure holds all of the individual device information */ +struct opticon_private { + struct usb_device *udev; + struct usb_serial *serial; + struct usb_serial_port *port; + unsigned char *bulk_in_buffer; + struct urb *bulk_read_urb; + int buffer_size; + u8 bulk_address; + spinlock_t lock; /* protects the following flags */ + bool throttled; + bool actually_throttled; + bool rts; +}; + +static void opticon_bulk_callback(struct urb *urb) +{ + struct opticon_private *priv = urb->context; + unsigned char *data = urb->transfer_buffer; + struct usb_serial_port *port = priv->port; + int status = urb->status; + struct tty_struct *tty; + int result; + int available_room = 0; + int data_length; + + dbg("%s - port %d", __func__, port->number); + + switch (status) { + case 0: + /* success */ + break; + case -ECONNRESET: + case -ENOENT: + case -ESHUTDOWN: + /* this urb is terminated, clean up */ + dbg("%s - urb shutting down with status: %d", + __func__, status); + return; + default: + dbg("%s - nonzero urb status received: %d", + __func__, status); + goto exit; + } + + usb_serial_debug_data(debug, &port->dev, __func__, urb->actual_length, + data); + + if (urb->actual_length > 2) { + data_length = urb->actual_length - 2; + + /* + * Data from the device comes with a 2 byte header: + * + * <0x00><0x00>data... + * This is real data to be sent to the tty layer + * <0x00><0x01)level + * This is a RTS level change, the third byte is the RTS + * value (0 for low, 1 for high). + */ + if ((data[0] == 0x00) && (data[1] == 0x00)) { + /* real data, send it to the tty layer */ + tty = tty_port_tty_get(&port->port); + if (tty) { + available_room = tty_buffer_request_room(tty, + data_length); + if (available_room) { + tty_insert_flip_string(tty, data, + available_room); + tty_flip_buffer_push(tty); + } + tty_kref_put(tty); + } + } else { + if ((data[0] == 0x00) && (data[1] == 0x01)) { + if (data[2] == 0x00) + priv->rts = false; + else + priv->rts = true; + /* FIXME change the RTS level */ + } else { + dev_dbg(&priv->udev->dev, + "Unknown data packet received from the device:" + " %2x %2x\n", + data[0], data[1]); + } + } + } else { + dev_dbg(&priv->udev->dev, + "Improper ammount of data received from the device, " + "%d bytes", urb->actual_length); + } + +exit: + spin_lock(&priv->lock); + + /* Continue trying to always read if we should */ + if (!priv->throttled) { + usb_fill_bulk_urb(priv->bulk_read_urb, priv->udev, + usb_rcvbulkpipe(priv->udev, + priv->bulk_address), + priv->bulk_in_buffer, priv->buffer_size, + opticon_bulk_callback, priv); + result = usb_submit_urb(port->read_urb, GFP_ATOMIC); + if (result) + dev_err(&port->dev, + "%s - failed resubmitting read urb, error %d\n", + __func__, result); + } else + priv->actually_throttled = true; + spin_unlock(&priv->lock); +} + +static int opticon_open(struct tty_struct *tty, struct usb_serial_port *port, + struct file *filp) +{ + struct opticon_private *priv = usb_get_serial_data(port->serial); + unsigned long flags; + int result = 0; + + dbg("%s - port %d", __func__, port->number); + + spin_lock_irqsave(&priv->lock, flags); + priv->throttled = false; + priv->actually_throttled = false; + priv->port = port; + spin_unlock_irqrestore(&priv->lock, flags); + + /* + * Force low_latency on so that our tty_push actually forces the data + * through, otherwise it is scheduled, and with high data rates (like + * with OHCI) data can get lost. + */ + if (tty) + tty->low_latency = 1; + + /* Start reading from the device */ + usb_fill_bulk_urb(priv->bulk_read_urb, priv->udev, + usb_rcvbulkpipe(priv->udev, + priv->bulk_address), + priv->bulk_in_buffer, priv->buffer_size, + opticon_bulk_callback, priv); + result = usb_submit_urb(priv->bulk_read_urb, GFP_KERNEL); + if (result) + dev_err(&port->dev, + "%s - failed resubmitting read urb, error %d\n", + __func__, result); + return result; +} + +static void opticon_close(struct tty_struct *tty, struct usb_serial_port *port, + struct file *filp) +{ + struct opticon_private *priv = usb_get_serial_data(port->serial); + + dbg("%s - port %d", __func__, port->number); + + /* shutdown our urbs */ + usb_kill_urb(priv->bulk_read_urb); +} + +static void opticon_throttle(struct tty_struct *tty) +{ + struct usb_serial_port *port = tty->driver_data; + struct opticon_private *priv = usb_get_serial_data(port->serial); + unsigned long flags; + + dbg("%s - port %d", __func__, port->number); + spin_lock_irqsave(&priv->lock, flags); + priv->throttled = true; + spin_unlock_irqrestore(&priv->lock, flags); +} + + +static void opticon_unthrottle(struct tty_struct *tty) +{ + struct usb_serial_port *port = tty->driver_data; + struct opticon_private *priv = usb_get_serial_data(port->serial); + unsigned long flags; + int result; + + dbg("%s - port %d", __func__, port->number); + + spin_lock_irqsave(&priv->lock, flags); + priv->throttled = false; + priv->actually_throttled = false; + spin_unlock_irqrestore(&priv->lock, flags); + + priv->bulk_read_urb->dev = port->serial->dev; + result = usb_submit_urb(priv->bulk_read_urb, GFP_ATOMIC); + if (result) + dev_err(&port->dev, + "%s - failed submitting read urb, error %d\n", + __func__, result); +} + +static int opticon_startup(struct usb_serial *serial) +{ + struct opticon_private *priv; + struct usb_host_interface *intf; + int i; + int retval = -ENOMEM; + bool bulk_in_found = false; + + /* create our private serial structure */ + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (priv == NULL) { + dev_err(&serial->dev->dev, "%s - Out of memory\n", __func__); + return -ENOMEM; + } + spin_lock_init(&priv->lock); + priv->serial = serial; + priv->port = serial->port[0]; + priv->udev = serial->dev; + + /* find our bulk endpoint */ + intf = serial->interface->altsetting; + for (i = 0; i < intf->desc.bNumEndpoints; ++i) { + struct usb_endpoint_descriptor *endpoint; + + endpoint = &intf->endpoint[i].desc; + if (!usb_endpoint_is_bulk_in(endpoint)) + continue; + + priv->bulk_read_urb = usb_alloc_urb(0, GFP_KERNEL); + if (!priv->bulk_read_urb) { + dev_err(&priv->udev->dev, "out of memory\n"); + goto error; + } + + priv->buffer_size = le16_to_cpu(endpoint->wMaxPacketSize) * 2; + priv->bulk_in_buffer = kmalloc(priv->buffer_size, GFP_KERNEL); + if (!priv->bulk_in_buffer) { + dev_err(&priv->udev->dev, "out of memory\n"); + goto error; + } + + priv->bulk_address = endpoint->bEndpointAddress; + + /* set up our bulk urb */ + usb_fill_bulk_urb(priv->bulk_read_urb, priv->udev, + usb_rcvbulkpipe(priv->udev, + endpoint->bEndpointAddress), + priv->bulk_in_buffer, priv->buffer_size, + opticon_bulk_callback, priv); + + bulk_in_found = true; + break; + } + + if (!bulk_in_found) { + dev_err(&priv->udev->dev, + "Error - the proper endpoints were not found!\n"); + goto error; + } + + usb_set_serial_data(serial, priv); + return 0; + +error: + usb_free_urb(priv->bulk_read_urb); + kfree(priv->bulk_in_buffer); + kfree(priv); + return retval; +} + +static void opticon_shutdown(struct usb_serial *serial) +{ + struct opticon_private *priv = usb_get_serial_data(serial); + + dbg("%s", __func__); + + usb_kill_urb(priv->bulk_read_urb); + usb_free_urb(priv->bulk_read_urb); + kfree(priv->bulk_in_buffer); + kfree(priv); + usb_set_serial_data(serial, NULL); +} + + +static struct usb_driver opticon_driver = { + .name = "opticon", + .probe = usb_serial_probe, + .disconnect = usb_serial_disconnect, + .id_table = id_table, + .no_dynamic_id = 1, +}; + +static struct usb_serial_driver opticon_device = { + .driver = { + .owner = THIS_MODULE, + .name = "opticon", + }, + .id_table = id_table, + .usb_driver = &opticon_driver, + .num_ports = 1, + .attach = opticon_startup, + .open = opticon_open, + .close = opticon_close, + .shutdown = opticon_shutdown, + .throttle = opticon_throttle, + .unthrottle = opticon_unthrottle, +}; + +static int __init opticon_init(void) +{ + int retval; + + retval = usb_serial_register(&opticon_device); + if (retval) + return retval; + retval = usb_register(&opticon_driver); + if (retval) + usb_serial_deregister(&opticon_device); + return retval; +} + +static void __exit opticon_exit(void) +{ + usb_deregister(&opticon_driver); + usb_serial_deregister(&opticon_device); +} + +module_init(opticon_init); +module_exit(opticon_exit); +MODULE_LICENSE("GPL"); + +module_param(debug, bool, S_IRUGO | S_IWUSR); +MODULE_PARM_DESC(debug, "Debug enabled or not"); From 5aa637505fb8610cd2724b09fa0ab03fd6cdca63 Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Wed, 3 Dec 2008 22:30:04 -0800 Subject: [PATCH 092/123] USB: add siemens_mpi usb-serial "stub" driver This driver got rescued from a few years ago and was requested to be added. So I cleaned it up, ported it to the latest kernel version and here it is. Cc: Thomas Hergenhahn Cc: Emmanuele Signed-off-by: Greg Kroah-Hartman --- drivers/usb/serial/Kconfig | 8 ++++ drivers/usb/serial/Makefile | 1 + drivers/usb/serial/siemens_mpi.c | 77 ++++++++++++++++++++++++++++++++ 3 files changed, 86 insertions(+) create mode 100644 drivers/usb/serial/siemens_mpi.c diff --git a/drivers/usb/serial/Kconfig b/drivers/usb/serial/Kconfig index da30784a0ef..b361f05cafa 100644 --- a/drivers/usb/serial/Kconfig +++ b/drivers/usb/serial/Kconfig @@ -496,6 +496,14 @@ config USB_SERIAL_SAFE_PADDED bool "USB Secure Encapsulated Driver - Padded" depends on USB_SERIAL_SAFE +config USB_SERIAL_SIEMENS_MPI + tristate "USB Siemens MPI driver" + help + Say M here if you want to use a Siemens USB/MPI adapter. + + To compile this driver as a module, choose M here: the + module will be called siemens_mpi. + config USB_SERIAL_SIERRAWIRELESS tristate "USB Sierra Wireless Driver" help diff --git a/drivers/usb/serial/Makefile b/drivers/usb/serial/Makefile index 9a4f59e0ee5..b75be91eb8f 100644 --- a/drivers/usb/serial/Makefile +++ b/drivers/usb/serial/Makefile @@ -46,6 +46,7 @@ obj-$(CONFIG_USB_SERIAL_OPTION) += option.o obj-$(CONFIG_USB_SERIAL_OTI6858) += oti6858.o obj-$(CONFIG_USB_SERIAL_PL2303) += pl2303.o obj-$(CONFIG_USB_SERIAL_SAFE) += safe_serial.o +obj-$(CONFIG_USB_SERIAL_SIEMENS_MPI) += siemens_mpi.o obj-$(CONFIG_USB_SERIAL_SIERRAWIRELESS) += sierra.o obj-$(CONFIG_USB_SERIAL_SPCP8X5) += spcp8x5.o obj-$(CONFIG_USB_SERIAL_TI) += ti_usb_3410_5052.o diff --git a/drivers/usb/serial/siemens_mpi.c b/drivers/usb/serial/siemens_mpi.c new file mode 100644 index 00000000000..951ea0c6ba7 --- /dev/null +++ b/drivers/usb/serial/siemens_mpi.c @@ -0,0 +1,77 @@ +/* + * Siemens USB-MPI Serial USB driver + * + * Copyright (C) 2005 Thomas Hergenhahn + * Copyright (C) 2005,2008 Greg Kroah-Hartman + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + */ + +#include +#include +#include +#include +#include +#include + +/* Version Information */ +#define DRIVER_VERSION "Version 0.1 09/26/2005" +#define DRIVER_AUTHOR "Thomas Hergenhahn@web.de http://libnodave.sf.net" +#define DRIVER_DESC "Driver for Siemens USB/MPI adapter" + + +static struct usb_device_id id_table[] = { + /* Vendor and product id for 6ES7-972-0CB20-0XA0 */ + { USB_DEVICE(0x908, 0x0004) }, + { }, +}; +MODULE_DEVICE_TABLE(usb, id_table); + +static struct usb_driver siemens_usb_mpi_driver = { + .name = "siemens_mpi", + .probe = usb_serial_probe, + .disconnect = usb_serial_disconnect, + .id_table = id_table, +}; + +static struct usb_serial_driver siemens_usb_mpi_device = { + .driver = { + .owner = THIS_MODULE, + .name = "siemens_mpi", + }, + .id_table = id_table, + .num_ports = 1, +}; + +static int __init siemens_usb_mpi_init(void) +{ + int retval; + + retval = usb_serial_register(&siemens_usb_mpi_device); + if (retval) + goto failed_usb_serial_register; + retval = usb_register(&siemens_usb_mpi_driver); + if (retval) + goto failed_usb_register; + printk(KERN_INFO DRIVER_DESC "\n"); + printk(KERN_INFO DRIVER_VERSION " " DRIVER_AUTHOR "\n"); + return retval; +failed_usb_register: + usb_serial_deregister(&siemens_usb_mpi_device); +failed_usb_serial_register: + return retval; +} + +static void __exit siemens_usb_mpi_exit(void) +{ + usb_deregister(&siemens_usb_mpi_driver); + usb_serial_deregister(&siemens_usb_mpi_device); +} + +module_init(siemens_usb_mpi_init); +module_exit(siemens_usb_mpi_exit); +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL"); From 338b67b0c1a97ca705023a8189cf41aa0828d294 Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Thu, 14 Aug 2008 09:37:34 -0700 Subject: [PATCH 093/123] USB: remove warn() macro from usb.h USB should not be having it's own printk macros, so remove warn() and use the system-wide standard of dev_warn() wherever possible. In the few places that will not work out, use a basic printk(). Now that all in-tree users are gone, remove the macro. Signed-off-by: Greg Kroah-Hartman --- include/linux/usb.h | 2 -- 1 file changed, 2 deletions(-) diff --git a/include/linux/usb.h b/include/linux/usb.h index e8963989650..28f68f58779 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -1744,8 +1744,6 @@ extern void usb_unregister_notify(struct notifier_block *nb); format "\n" , ## arg) #define info(format, arg...) printk(KERN_INFO KBUILD_MODNAME ": " \ format "\n" , ## arg) -#define warn(format, arg...) printk(KERN_WARNING KBUILD_MODNAME ": " \ - format "\n" , ## arg) #endif /* __KERNEL__ */ From 34c65d82e02147331701c7795e3144d511adf4e9 Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Mon, 18 Aug 2008 13:21:04 -0700 Subject: [PATCH 094/123] USB: remove info() macro from usb.h USB should not be having it's own printk macros, so remove info() and use the system-wide standard of dev_info() wherever possible. No one in the tree is using the macro, so it can now be removed. Signed-off-by: Greg Kroah-Hartman --- include/linux/usb.h | 2 -- 1 file changed, 2 deletions(-) diff --git a/include/linux/usb.h b/include/linux/usb.h index 28f68f58779..85ee9be9361 100644 --- a/include/linux/usb.h +++ b/include/linux/usb.h @@ -1742,8 +1742,6 @@ extern void usb_unregister_notify(struct notifier_block *nb); #define err(format, arg...) printk(KERN_ERR KBUILD_MODNAME ": " \ format "\n" , ## arg) -#define info(format, arg...) printk(KERN_INFO KBUILD_MODNAME ": " \ - format "\n" , ## arg) #endif /* __KERNEL__ */ From 5e07878787ad07361571150230cc3a8d522ae046 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:39 -0800 Subject: [PATCH 095/123] debugfs: add helpers for exporting a size_t simple value MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit In the same spirit as debugfs_create_*(), introduce helpers for exporting size_t values over debugfs. The only trick done is that the format verifier is kept at %llu instead of %zu; otherwise type warnings would pop up: format ‘%zu’ expects type ‘size_t’, but argument 2 has type ‘long long unsigned int’ There is no real way to fix this one--however, we can consider %llu and %zu to be compatible if we consider that we are using the same for validating in debugfs_create_{x,u}{8,16,32}(). Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- fs/debugfs/file.c | 32 ++++++++++++++++++++++++++++++++ include/linux/debugfs.h | 2 ++ 2 files changed, 34 insertions(+) diff --git a/fs/debugfs/file.c b/fs/debugfs/file.c index 159a5efd6a8..33a90120f6a 100644 --- a/fs/debugfs/file.c +++ b/fs/debugfs/file.c @@ -294,6 +294,38 @@ struct dentry *debugfs_create_x32(const char *name, mode_t mode, } EXPORT_SYMBOL_GPL(debugfs_create_x32); + +static int debugfs_size_t_set(void *data, u64 val) +{ + *(size_t *)data = val; + return 0; +} +static int debugfs_size_t_get(void *data, u64 *val) +{ + *val = *(size_t *)data; + return 0; +} +DEFINE_SIMPLE_ATTRIBUTE(fops_size_t, debugfs_size_t_get, debugfs_size_t_set, + "%llu\n"); /* %llu and %zu are more or less the same */ + +/** + * debugfs_create_size_t - create a debugfs file that is used to read and write an size_t value + * @name: a pointer to a string containing the name of the file to create. + * @mode: the permission that the file should have + * @parent: a pointer to the parent dentry for this file. This should be a + * directory dentry if set. If this parameter is %NULL, then the + * file will be created in the root of the debugfs filesystem. + * @value: a pointer to the variable that the file should read to and write + * from. + */ +struct dentry *debugfs_create_size_t(const char *name, mode_t mode, + struct dentry *parent, size_t *value) +{ + return debugfs_create_file(name, mode, parent, value, &fops_size_t); +} +EXPORT_SYMBOL_GPL(debugfs_create_size_t); + + static ssize_t read_file_bool(struct file *file, char __user *user_buf, size_t count, loff_t *ppos) { diff --git a/include/linux/debugfs.h b/include/linux/debugfs.h index e1a6c046cea..23936b16426 100644 --- a/include/linux/debugfs.h +++ b/include/linux/debugfs.h @@ -63,6 +63,8 @@ struct dentry *debugfs_create_x16(const char *name, mode_t mode, struct dentry *parent, u16 *value); struct dentry *debugfs_create_x32(const char *name, mode_t mode, struct dentry *parent, u32 *value); +struct dentry *debugfs_create_size_t(const char *name, mode_t mode, + struct dentry *parent, size_t *value); struct dentry *debugfs_create_bool(const char *name, mode_t mode, struct dentry *parent, u32 *value); From 0d695913b0016b362a84a8bb6d6e28f8d90a70e2 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:32 -0800 Subject: [PATCH 096/123] wimax: documentation for the stack wimax documentation Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- Documentation/wimax/README.wimax | 81 ++++++++++++++++++++++++++++++++ 1 file changed, 81 insertions(+) create mode 100644 Documentation/wimax/README.wimax diff --git a/Documentation/wimax/README.wimax b/Documentation/wimax/README.wimax new file mode 100644 index 00000000000..b78c4378084 --- /dev/null +++ b/Documentation/wimax/README.wimax @@ -0,0 +1,81 @@ + + Linux kernel WiMAX stack + + (C) 2008 Intel Corporation < linux-wimax@intel.com > + + This provides a basic Linux kernel WiMAX stack to provide a common + control API for WiMAX devices, usable from kernel and user space. + +1. Design + + The WiMAX stack is designed to provide for common WiMAX control + services to current and future WiMAX devices from any vendor. + + Because currently there is only one and we don't know what would be the + common services, the APIs it currently provides are very minimal. + However, it is done in such a way that it is easily extensible to + accommodate future requirements. + + The stack works by embedding a struct wimax_dev in your device's + control structures. This provides a set of callbacks that the WiMAX + stack will call in order to implement control operations requested by + the user. As well, the stack provides API functions that the driver + calls to notify about changes of state in the device. + + The stack exports the API calls needed to control the device to user + space using generic netlink as a marshalling mechanism. You can access + them using your own code or use the wrappers provided for your + convenience in libwimax (in the wimax-tools package). + + For detailed information on the stack, please see + include/linux/wimax.h. + +2. Usage + + For usage in a driver (registration, API, etc) please refer to the + instructions in the header file include/linux/wimax.h. + + When a device is registered with the WiMAX stack, a set of debugfs + files will appear in /sys/kernel/debug/wimax:wmxX can tweak for + control. + +2.1. Obtaining debug information: debugfs entries + + The WiMAX stack is compiled, by default, with debug messages that can + be used to diagnose issues. By default, said messages are disabled. + + The drivers will register debugfs entries that allow the user to tweak + debug settings. + + Each driver, when registering with the stack, will cause a debugfs + directory named wimax:DEVICENAME to be created; optionally, it might + create more subentries below it. + +2.1.1. Increasing debug output + + The files named *dl_* indicate knobs for controlling the debug output + of different submodules of the WiMAX stack: + * +# find /sys/kernel/debug/wimax\:wmx0 -name \*dl_\* +/sys/kernel/debug/wimax:wmx0/wimax_dl_stack +/sys/kernel/debug/wimax:wmx0/wimax_dl_op_rfkill +/sys/kernel/debug/wimax:wmx0/wimax_dl_op_reset +/sys/kernel/debug/wimax:wmx0/wimax_dl_op_msg +/sys/kernel/debug/wimax:wmx0/wimax_dl_id_table +/sys/kernel/debug/wimax:wmx0/wimax_dl_debugfs +/sys/kernel/debug/wimax:wmx0/.... # other driver specific files + + NOTE: Of course, if debugfs is mounted in a directory other than + /sys/kernel/debug, those paths will change. + + By reading the file you can obtain the current value of said debug + level; by writing to it, you can set it. + + To increase the debug level of, for example, the id-table submodule, + just write: + +$ echo 3 > /sys/kernel/debug/wimax:wmx0/wimax_dl_id_table + + Increasing numbers yield increasing debug information; for details of + what is printed and the available levels, check the source. The code + uses 0 for disabled and increasing values until 8. From ace22f0881e1333d0c55ddf484e5352fe03a806a Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:33 -0800 Subject: [PATCH 097/123] wimax: headers for kernel API and user space interaction Definitions for the user/kernel API protocol through generic netlink. User space can copy it verbatim and use it. Kernel API definition declares the main data types and calls for the drivers to integrate into the WiMAX stack. Provides usage documentation. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- include/linux/wimax.h | 234 +++++++++++++++++++ include/net/wimax.h | 520 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 754 insertions(+) create mode 100644 include/linux/wimax.h create mode 100644 include/net/wimax.h diff --git a/include/linux/wimax.h b/include/linux/wimax.h new file mode 100644 index 00000000000..c89de7f4e5b --- /dev/null +++ b/include/linux/wimax.h @@ -0,0 +1,234 @@ +/* + * Linux WiMax + * API for user space + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Inaky Perez-Gonzalez + * - Initial implementation + * + * + * This file declares the user/kernel protocol that is spoken over + * Generic Netlink, as well as any type declaration that is to be used + * by kernel and user space. + * + * It is intended for user space to clone it verbatim to use it as a + * primary reference for definitions. + * + * Stuff intended for kernel usage as well as full protocol and stack + * documentation is rooted in include/net/wimax.h. + */ + +#ifndef __LINUX__WIMAX_H__ +#define __LINUX__WIMAX_H__ + +#include + +enum { + /** + * Version of the interface (unsigned decimal, MMm, max 25.5) + * M - Major: change if removing or modifying an existing call. + * m - minor: change when adding a new call + */ + WIMAX_GNL_VERSION = 00, + /* Generic NetLink attributes */ + WIMAX_GNL_ATTR_INVALID = 0x00, + WIMAX_GNL_ATTR_MAX = 10, +}; + + +/* + * Generic NetLink operations + * + * Most of these map to an API call; _OP_ stands for operation, _RP_ + * for reply and _RE_ for report (aka: signal). + */ +enum { + WIMAX_GNL_OP_MSG_FROM_USER, /* User to kernel message */ + WIMAX_GNL_OP_MSG_TO_USER, /* Kernel to user message */ + WIMAX_GNL_OP_RFKILL, /* Run wimax_rfkill() */ + WIMAX_GNL_OP_RESET, /* Run wimax_rfkill() */ + WIMAX_GNL_RE_STATE_CHANGE, /* Report: status change */ +}; + + +/* Message from user / to user */ +enum { + WIMAX_GNL_MSG_IFIDX = 1, + WIMAX_GNL_MSG_PIPE_NAME, + WIMAX_GNL_MSG_DATA, +}; + + +/* + * wimax_rfkill() + * + * The state of the radio (ON/OFF) is mapped to the rfkill subsystem's + * switch state (DISABLED/ENABLED). + */ +enum wimax_rf_state { + WIMAX_RF_OFF = 0, /* Radio is off, rfkill on/enabled */ + WIMAX_RF_ON = 1, /* Radio is on, rfkill off/disabled */ + WIMAX_RF_QUERY = 2, +}; + +/* Attributes */ +enum { + WIMAX_GNL_RFKILL_IFIDX = 1, + WIMAX_GNL_RFKILL_STATE, +}; + + +/* Attributes for wimax_reset() */ +enum { + WIMAX_GNL_RESET_IFIDX = 1, +}; + + +/* + * Attributes for the Report State Change + * + * For now we just have the old and new states; new attributes might + * be added later on. + */ +enum { + WIMAX_GNL_STCH_IFIDX = 1, + WIMAX_GNL_STCH_STATE_OLD, + WIMAX_GNL_STCH_STATE_NEW, +}; + + +/** + * enum wimax_st - The different states of a WiMAX device + * @__WIMAX_ST_NULL: The device structure has been allocated and zeroed, + * but still wimax_dev_add() hasn't been called. There is no state. + * + * @WIMAX_ST_DOWN: The device has been registered with the WiMAX and + * networking stacks, but it is not initialized (normally that is + * done with 'ifconfig DEV up' [or equivalent], which can upload + * firmware and enable communications with the device). + * In this state, the device is powered down and using as less + * power as possible. + * This state is the default after a call to wimax_dev_add(). It + * is ok to have drivers move directly to %WIMAX_ST_UNINITIALIZED + * or %WIMAX_ST_RADIO_OFF in _probe() after the call to + * wimax_dev_add(). + * It is recommended that the driver leaves this state when + * calling 'ifconfig DEV up' and enters it back on 'ifconfig DEV + * down'. + * + * @__WIMAX_ST_QUIESCING: The device is being torn down, so no API + * operations are allowed to proceed except the ones needed to + * complete the device clean up process. + * + * @WIMAX_ST_UNINITIALIZED: [optional] Communication with the device + * is setup, but the device still requires some configuration + * before being operational. + * Some WiMAX API calls might work. + * + * @WIMAX_ST_RADIO_OFF: The device is fully up; radio is off (wether + * by hardware or software switches). + * It is recommended to always leave the device in this state + * after initialization. + * + * @WIMAX_ST_READY: The device is fully up and radio is on. + * + * @WIMAX_ST_SCANNING: [optional] The device has been instructed to + * scan. In this state, the device cannot be actively connected to + * a network. + * + * @WIMAX_ST_CONNECTING: The device is connecting to a network. This + * state exists because in some devices, the connect process can + * include a number of negotiations between user space, kernel + * space and the device. User space needs to know what the device + * is doing. If the connect sequence in a device is atomic and + * fast, the device can transition directly to CONNECTED + * + * @WIMAX_ST_CONNECTED: The device is connected to a network. + * + * @__WIMAX_ST_INVALID: This is an invalid state used to mark the + * maximum numeric value of states. + * + * Description: + * + * Transitions from one state to another one are atomic and can only + * be caused in kernel space with wimax_state_change(). To read the + * state, use wimax_state_get(). + * + * States starting with __ are internal and shall not be used or + * referred to by drivers or userspace. They look ugly, but that's the + * point -- if any use is made non-internal to the stack, it is easier + * to catch on review. + * + * All API operations [with well defined exceptions] will take the + * device mutex before starting and then check the state. If the state + * is %__WIMAX_ST_NULL, %WIMAX_ST_DOWN, %WIMAX_ST_UNINITIALIZED or + * %__WIMAX_ST_QUIESCING, it will drop the lock and quit with + * -%EINVAL, -%ENOMEDIUM, -%ENOTCONN or -%ESHUTDOWN. + * + * The order of the definitions is important, so we can do numerical + * comparisons (eg: < %WIMAX_ST_RADIO_OFF means the device is not ready + * to operate). + */ +/* + * The allowed state transitions are described in the table below + * (states in rows can go to states in columns where there is an X): + * + * UNINI RADIO READY SCAN CONNEC CONNEC + * NULL DOWN QUIESCING TIALIZED OFF NING TING TED + * NULL - x + * DOWN - x x x + * QUIESCING x - + * UNINITIALIZED x - x + * RADIO_OFF x - x + * READY x x - x x x + * SCANNING x x x - x x + * CONNECTING x x x x - x + * CONNECTED x x x - + * + * This table not available in kernel-doc because the formatting messes it up. + */ + enum wimax_st { + __WIMAX_ST_NULL = 0, + WIMAX_ST_DOWN, + __WIMAX_ST_QUIESCING, + WIMAX_ST_UNINITIALIZED, + WIMAX_ST_RADIO_OFF, + WIMAX_ST_READY, + WIMAX_ST_SCANNING, + WIMAX_ST_CONNECTING, + WIMAX_ST_CONNECTED, + __WIMAX_ST_INVALID /* Always keep last */ +}; + + +#endif /* #ifndef __LINUX__WIMAX_H__ */ diff --git a/include/net/wimax.h b/include/net/wimax.h new file mode 100644 index 00000000000..1602614fdaf --- /dev/null +++ b/include/net/wimax.h @@ -0,0 +1,520 @@ +/* + * Linux WiMAX + * Kernel space API for accessing WiMAX devices + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * The WiMAX stack provides an API for controlling and managing the + * system's WiMAX devices. This API affects the control plane; the + * data plane is accessed via the network stack (netdev). + * + * Parts of the WiMAX stack API and notifications are exported to + * user space via Generic Netlink. In user space, libwimax (part of + * the wimax-tools package) provides a shim layer for accessing those + * calls. + * + * The API is standarized for all WiMAX devices and different drivers + * implement the backend support for it. However, device-specific + * messaging pipes are provided that can be used to issue commands and + * receive notifications in free form. + * + * Currently the messaging pipes are the only means of control as it + * is not known (due to the lack of more devices in the market) what + * will be a good abstraction layer. Expect this to change as more + * devices show in the market. This API is designed to be growable in + * order to address this problem. + * + * USAGE + * + * Embed a `struct wimax_dev` at the beginning of the the device's + * private structure, initialize and register it. For details, see + * `struct wimax_dev`s documentation. + * + * Once this is done, wimax-tools's libwimaxll can be used to + * communicate with the driver from user space. You user space + * application does not have to forcibily use libwimaxll and can talk + * the generic netlink protocol directly if desired. + * + * Remember this is a very low level API that will to provide all of + * WiMAX features. Other daemons and services running in user space + * are the expected clients of it. They offer a higher level API that + * applications should use (an example of this is the Intel's WiMAX + * Network Service for the i2400m). + * + * DESIGN + * + * Although not set on final stone, this very basic interface is + * mostly completed. Remember this is meant to grow as new common + * operations are decided upon. New operations will be added to the + * interface, intent being on keeping backwards compatibility as much + * as possible. + * + * This layer implements a set of calls to control a WiMAX device, + * exposing a frontend to the rest of the kernel and user space (via + * generic netlink) and a backend implementation in the driver through + * function pointers. + * + * WiMAX devices have a state, and a kernel-only API allows the + * drivers to manipulate that state. State transitions are atomic, and + * only some of them are allowed (see `enum wimax_st`). + * + * Most API calls will set the state automatically; in most cases + * drivers have to only report state changes due to external + * conditions. + * + * All API operations are 'atomic', serialized thorough a mutex in the + * `struct wimax_dev`. + * + * EXPORTING TO USER SPACE THROUGH GENERIC NETLINK + * + * The API is exported to user space using generic netlink (other + * methods can be added as needed). + * + * There is a Generic Netlink Family named "WiMAX", where interfaces + * supporting the WiMAX interface receive commands and broadcast their + * signals over a multicast group named "msg". + * + * Mapping to the source/destination interface is done by an interface + * index attribute. + * + * For user-to-kernel traffic (commands) we use a function call + * marshalling mechanism, where a message X with attributes A, B, C + * sent from user space to kernel space means executing the WiMAX API + * call wimax_X(A, B, C), sending the results back as a message. + * + * Kernel-to-user (notifications or signals) communication is sent + * over multicast groups. This allows to have multiple applications + * monitoring them. + * + * Each command/signal gets assigned it's own attribute policy. This + * way the validator will verify that all the attributes in there are + * only the ones that should be for each command/signal. Thing of an + * attribute mapping to a type+argumentname for each command/signal. + * + * If we had a single policy for *all* commands/signals, after running + * the validator we'd have to check "does this attribute belong in + * here"? for each one. It can be done manually, but it's just easier + * to have the validator do that job with multiple policies. As well, + * it makes it easier to later expand each command/signal signature + * without affecting others and keeping the namespace more or less + * sane. Not that it is too complicated, but it makes it even easier. + * + * No state information is maintained in the kernel for each user + * space connection (the connection is stateless). + * + * TESTING FOR THE INTERFACE AND VERSIONING + * + * If network interface X is a WiMAX device, there will be a Generic + * Netlink family named "WiMAX X" and the device will present a + * "wimax" directory in it's network sysfs directory + * (/sys/class/net/DEVICE/wimax) [used by HAL]. + * + * The inexistence of any of these means the device does not support + * this WiMAX API. + * + * By querying the generic netlink controller, versioning information + * and the multicast groups available can be found. Applications using + * the interface can either rely on that or use the generic netlink + * controller to figure out which generic netlink commands/signals are + * supported. + * + * NOTE: this versioning is a last resort to avoid hard + * incompatibilities. It is the intention of the design of this + * stack not to introduce backward incompatible changes. + * + * The version code has to fit in one byte (restrictions imposed by + * generic netlink); we use `version / 10` for the major version and + * `version % 10` for the minor. This gives 9 minors for each major + * and 25 majors. + * + * The version change protocol is as follow: + * + * - Major versions: needs to be increased if an existing message/API + * call is changed or removed. Doesn't need to be changed if a new + * message is added. + * + * - Minor version: needs to be increased if new messages/API calls are + * being added or some other consideration that doesn't impact the + * user-kernel interface too much (like some kind of bug fix) and + * that is kind of left up in the air to common sense. + * + * User space code should not try to work if the major version it was + * compiled for differs from what the kernel offers. As well, if the + * minor version of the kernel interface is lower than the one user + * space is expecting (the one it was compiled for), the kernel + * might be missing API calls; user space shall be ready to handle + * said condition. Use the generic netlink controller operations to + * find which ones are supported and which not. + * + * libwimaxll:wimaxll_open() takes care of checking versions. + * + * THE OPERATIONS: + * + * Each operation is defined in its on file (drivers/net/wimax/op-*.c) + * for clarity. The parts needed for an operation are: + * + * - a function pointer in `struct wimax_dev`: optional, as the + * operation might be implemented by the stack and not by the + * driver. + * + * All function pointers are named wimax_dev->op_*(), and drivers + * must implement them except where noted otherwise. + * + * - When exported to user space, a `struct nla_policy` to define the + * attributes of the generic netlink command and a `struct genl_ops` + * to define the operation. + * + * All the declarations for the operation codes (WIMAX_GNL_OP_) + * and generic netlink attributes (WIMAX_GNL__*) are declared in + * include/linux/wimax.h; this file is intended to be cloned by user + * space to gain access to those declarations. + * + * A few caveats to remember: + * + * - Need to define attribute numbers starting in 1; otherwise it + * fails. + * + * - the `struct genl_family` requires a maximum attribute id; when + * defining the `struct nla_policy` for each message, it has to have + * an array size of WIMAX_GNL_ATTR_MAX+1. + * + * THE PIPE INTERFACE: + * + * This interface is kept intentionally simple. The driver can send + * and receive free-form messages to/from user space through a + * pipe. See drivers/net/wimax/op-msg.c for details. + * + * The kernel-to-user messages are sent with + * wimax_msg(). user-to-kernel messages are delivered via + * wimax_dev->op_msg_from_user(). + * + * RFKILL: + * + * RFKILL support is built into the wimax_dev layer; the driver just + * needs to call wimax_report_rfkill_{hw,sw}() to inform of changes in + * the hardware or software RF kill switches. When the stack wants to + * turn the radio off, it will call wimax_dev->op_rfkill_sw_toggle(), + * which the driver implements. + * + * User space can set the software RF Kill switch by calling + * wimax_rfkill(). + * + * The code for now only supports devices that don't require polling; + * If the device needs to be polled, create a self-rearming delayed + * work struct for polling or look into adding polled support to the + * WiMAX stack. + * + * When initializing the hardware (_probe), after calling + * wimax_dev_add(), query the device for it's RF Kill switches status + * and feed it back to the WiMAX stack using + * wimax_report_rfkill_{hw,sw}(). If any switch is missing, always + * report it as ON. + * + * NOTE: the wimax stack uses an inverted terminology to that of the + * RFKILL subsystem: + * + * - ON: radio is ON, RFKILL is DISABLED or OFF. + * - OFF: radio is OFF, RFKILL is ENABLED or ON. + * + * MISCELLANEOUS OPS: + * + * wimax_reset() can be used to reset the device to power on state; by + * default it issues a warm reset that maintains the same device + * node. If that is not possible, it falls back to a cold reset + * (device reconnect). The driver implements the backend to this + * through wimax_dev->op_reset(). + */ + +#ifndef __NET__WIMAX_H__ +#define __NET__WIMAX_H__ +#ifdef __KERNEL__ + +#include +#include +#include + +struct net_device; +struct genl_info; +struct wimax_dev; +struct input_dev; + +/** + * struct wimax_dev - Generic WiMAX device + * + * @net_dev: [fill] Pointer to the &struct net_device this WiMAX + * device implements. + * + * @op_msg_from_user: [fill] Driver-specific operation to + * handle a raw message from user space to the driver. The + * driver can send messages to user space using with + * wimax_msg_to_user(). + * + * @op_rfkill_sw_toggle: [fill] Driver-specific operation to act on + * userspace (or any other agent) requesting the WiMAX device to + * change the RF Kill software switch (WIMAX_RF_ON or + * WIMAX_RF_OFF). + * If such hardware support is not present, it is assumed the + * radio cannot be switched off and it is always on (and the stack + * will error out when trying to switch it off). In such case, + * this function pointer can be left as NULL. + * + * @op_reset: [fill] Driver specific operation to reset the + * device. + * This operation should always attempt first a warm reset that + * does not disconnect the device from the bus and return 0. + * If that fails, it should resort to some sort of cold or bus + * reset (even if it implies a bus disconnection and device + * dissapearance). In that case, -ENODEV should be returned to + * indicate the device is gone. + * This operation has to be synchronous, and return only when the + * reset is complete. In case of having had to resort to bus/cold + * reset implying a device disconnection, the call is allowed to + * return inmediately. + * NOTE: wimax_dev->mutex is NOT locked when this op is being + * called; however, wimax_dev->mutex_reset IS locked to ensure + * serialization of calls to wimax_reset(). + * See wimax_reset()'s documentation. + * + * @name: [fill] A way to identify this device. We need to register a + * name with many subsystems (input for RFKILL, workqueue + * creation, etc). We can't use the network device name as that + * might change and in some instances we don't know it yet (until + * we don't call register_netdev()). So we generate an unique one + * using the driver name and device bus id, place it here and use + * it across the board. Recommended naming: + * DRIVERNAME-BUSNAME:BUSID (dev->bus->name, dev->bus_id). + * + * @id_table_node: [private] link to the list of wimax devices kept by + * id-table.c. Protected by it's own spinlock. + * + * @mutex: [private] Serializes all concurrent access and execution of + * operations. + * + * @mutex_reset: [private] Serializes reset operations. Needs to be a + * different mutex because as part of the reset operation, the + * driver has to call back into the stack to do things such as + * state change, that require wimax_dev->mutex. + * + * @state: [private] Current state of the WiMAX device. + * + * @rfkill: [private] integration into the RF-Kill infrastructure. + * + * @rfkill_input: [private] virtual input device to process the + * hardware RF Kill switches. + * + * @rf_sw: [private] State of the software radio switch (OFF/ON) + * + * @rf_hw: [private] State of the hardware radio switch (OFF/ON) + * + * Description: + * This structure defines a common interface to access all WiMAX + * devices from different vendors and provides a common API as well as + * a free-form device-specific messaging channel. + * + * Usage: + * 1. Embed a &struct wimax_dev at *the beginning* the network + * device structure so that netdev_priv() points to it. + * + * 2. memset() it to zero + * + * 3. Initialize with wimax_dev_init(). This will leave the WiMAX + * device in the %__WIMAX_ST_NULL state. + * + * 4. Fill all the fields marked with [fill]; once called + * wimax_dev_add(), those fields CANNOT be modified. + * + * 5. Call wimax_dev_add() *after* registering the network + * device. This will leave the WiMAX device in the %WIMAX_ST_DOWN + * state. + * Protect the driver's net_device->open() against succeeding if + * the wimax device state is lower than %WIMAX_ST_DOWN. + * + * 6. Select when the device is going to be turned on/initialized; + * for example, it could be initialized on 'ifconfig up' (when the + * netdev op 'open()' is called on the driver). + * + * When the device is initialized (at `ifconfig up` time, or right + * after calling wimax_dev_add() from _probe(), make sure the + * following steps are taken + * + * a. Move the device to %WIMAX_ST_UNINITIALIZED. This is needed so + * some API calls that shouldn't work until the device is ready + * can be blocked. + * + * b. Initialize the device. Make sure to turn the SW radio switch + * off and move the device to state %WIMAX_ST_RADIO_OFF when + * done. When just initialized, a device should be left in RADIO + * OFF state until user space devices to turn it on. + * + * c. Query the device for the state of the hardware rfkill switch + * and call wimax_rfkill_report_hw() and wimax_rfkill_report_sw() + * as needed. See below. + * + * wimax_dev_rm() undoes before unregistering the network device. Once + * wimax_dev_add() is called, the driver can get called on the + * wimax_dev->op_* function pointers + * + * CONCURRENCY: + * + * The stack provides a mutex for each device that will disallow API + * calls happening concurrently; thus, op calls into the driver + * through the wimax_dev->op*() function pointers will always be + * serialized and *never* concurrent. + * + * For locking, take wimax_dev->mutex is taken; (most) operations in + * the API have to check for wimax_dev_is_ready() to return 0 before + * continuing (this is done internally). + * + * REFERENCE COUNTING: + * + * The WiMAX device is reference counted by the associated network + * device. The only operation that can be used to reference the device + * is wimax_dev_get_by_genl_info(), and the reference it acquires has + * to be released with dev_put(wimax_dev->net_dev). + * + * RFKILL: + * + * At startup, both HW and SW radio switchess are assumed to be off. + * + * At initialization time [after calling wimax_dev_add()], have the + * driver query the device for the status of the software and hardware + * RF kill switches and call wimax_report_rfkill_hw() and + * wimax_rfkill_report_sw() to indicate their state. If any is + * missing, just call it to indicate it is ON (radio always on). + * + * Whenever the driver detects a change in the state of the RF kill + * switches, it should call wimax_report_rfkill_hw() or + * wimax_report_rfkill_sw() to report it to the stack. + */ +struct wimax_dev { + struct net_device *net_dev; + struct list_head id_table_node; + struct mutex mutex; /* Protects all members and API calls */ + struct mutex mutex_reset; + enum wimax_st state; + + int (*op_msg_from_user)(struct wimax_dev *wimax_dev, + const char *, + const void *, size_t, + const struct genl_info *info); + int (*op_rfkill_sw_toggle)(struct wimax_dev *wimax_dev, + enum wimax_rf_state); + int (*op_reset)(struct wimax_dev *wimax_dev); + + struct rfkill *rfkill; + struct input_dev *rfkill_input; + unsigned rf_hw; + unsigned rf_sw; + char name[32]; + + struct dentry *debugfs_dentry; +}; + + + +/* + * WiMAX stack public API for device drivers + * ----------------------------------------- + * + * These functions are not exported to user space. + */ +extern void wimax_dev_init(struct wimax_dev *); +extern int wimax_dev_add(struct wimax_dev *, struct net_device *); +extern void wimax_dev_rm(struct wimax_dev *); + +static inline +struct wimax_dev *net_dev_to_wimax(struct net_device *net_dev) +{ + return netdev_priv(net_dev); +} + +static inline +struct device *wimax_dev_to_dev(struct wimax_dev *wimax_dev) +{ + return wimax_dev->net_dev->dev.parent; +} + +extern void wimax_state_change(struct wimax_dev *, enum wimax_st); +extern enum wimax_st wimax_state_get(struct wimax_dev *); + +/* + * Radio Switch state reporting. + * + * enum wimax_rf_state is declared in linux/wimax.h so the exports + * to user space can use it. + */ +extern void wimax_report_rfkill_hw(struct wimax_dev *, enum wimax_rf_state); +extern void wimax_report_rfkill_sw(struct wimax_dev *, enum wimax_rf_state); + + +/* + * Free-form messaging to/from user space + * + * Sending a message: + * + * wimax_msg(wimax_dev, pipe_name, buf, buf_size, GFP_KERNEL); + * + * Broken up: + * + * skb = wimax_msg_alloc(wimax_dev, pipe_name, buf_size, GFP_KERNEL); + * ...fill up skb... + * wimax_msg_send(wimax_dev, pipe_name, skb); + * + * Be sure not to modify skb->data in the middle (ie: don't use + * skb_push()/skb_pull()/skb_reserve() on the skb). + * + * "pipe_name" is any string, than can be interpreted as the name of + * the pipe or destinatary; the interpretation of it is driver + * specific, so the recipient can multiplex it as wished. It can be + * NULL, it won't be used - an example is using a "diagnostics" tag to + * send diagnostics information that a device-specific diagnostics + * tool would be interested in. + */ +extern struct sk_buff *wimax_msg_alloc(struct wimax_dev *, const char *, + const void *, size_t, gfp_t); +extern int wimax_msg_send(struct wimax_dev *, struct sk_buff *); +extern int wimax_msg(struct wimax_dev *, const char *, + const void *, size_t, gfp_t); + +extern const void *wimax_msg_data_len(struct sk_buff *, size_t *); +extern const void *wimax_msg_data(struct sk_buff *); +extern ssize_t wimax_msg_len(struct sk_buff *); + + +/* + * WiMAX stack user space API + * -------------------------- + * + * This API is what gets exported to user space for general + * operations. As well, they can be called from within the kernel, + * (with a properly referenced `struct wimax_dev`). + * + * Properly referenced means: the 'struct net_device' that embeds the + * device's control structure and (as such) the 'struct wimax_dev' is + * referenced by the caller. + */ +extern int wimax_rfkill(struct wimax_dev *, enum wimax_rf_state); +extern int wimax_reset(struct wimax_dev *); + +#else +/* You might be looking for linux/wimax.h */ +#error This file should not be included from user space. +#endif /* #ifdef __KERNEL__ */ +#endif /* #ifndef __NET__WIMAX_H__ */ From 60fa9ca6cfff2be4132ea93b7dd632801ec0c817 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:34 -0800 Subject: [PATCH 098/123] wimax: internal API for the kernel space WiMAX stack This file contains declarations and definitions used by the different submodules of the stack. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- net/wimax/wimax-internal.h | 91 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 91 insertions(+) create mode 100644 net/wimax/wimax-internal.h diff --git a/net/wimax/wimax-internal.h b/net/wimax/wimax-internal.h new file mode 100644 index 00000000000..1e743d21485 --- /dev/null +++ b/net/wimax/wimax-internal.h @@ -0,0 +1,91 @@ +/* + * Linux WiMAX + * Internal API for kernel space WiMAX stack + * + * + * Copyright (C) 2007 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * This header file is for declarations and definitions internal to + * the WiMAX stack. For public APIs and documentation, see + * include/net/wimax.h and include/linux/wimax.h. + */ + +#ifndef __WIMAX_INTERNAL_H__ +#define __WIMAX_INTERNAL_H__ +#ifdef __KERNEL__ + +#include +#include + + +/* + * Decide if a (locked) device is ready for use + * + * Before using the device structure, it must be locked + * (wimax_dev->mutex). As well, most operations need to call this + * function to check if the state is the right one. + * + * An error value will be returned if the state is not the right + * one. In that case, the caller should not attempt to use the device + * and just unlock it. + */ +static inline __must_check +int wimax_dev_is_ready(struct wimax_dev *wimax_dev) +{ + if (wimax_dev->state == __WIMAX_ST_NULL) + return -EINVAL; /* Device is not even registered! */ + if (wimax_dev->state == WIMAX_ST_DOWN) + return -ENOMEDIUM; + if (wimax_dev->state == __WIMAX_ST_QUIESCING) + return -ESHUTDOWN; + return 0; +} + + +static inline +void __wimax_state_set(struct wimax_dev *wimax_dev, enum wimax_st state) +{ + wimax_dev->state = state; +} +extern void __wimax_state_change(struct wimax_dev *, enum wimax_st); + +#ifdef CONFIG_DEBUG_FS +extern int wimax_debugfs_add(struct wimax_dev *); +extern void wimax_debugfs_rm(struct wimax_dev *); +#else +static inline int wimax_debugfs_add(struct wimax_dev *wimax_dev) +{ + return 0; +} +static inline void wimax_debugfs_rm(struct wimax_dev *wimax_dev) {} +#endif + +extern void wimax_id_table_add(struct wimax_dev *); +extern struct wimax_dev *wimax_dev_get_by_genl_info(struct genl_info *, int); +extern void wimax_id_table_rm(struct wimax_dev *); +extern void wimax_id_table_release(void); + +extern int wimax_rfkill_add(struct wimax_dev *); +extern void wimax_rfkill_rm(struct wimax_dev *); + +extern struct genl_family wimax_gnl_family; +extern struct genl_multicast_group wimax_gnl_mcg; + +#endif /* #ifdef __KERNEL__ */ +#endif /* #ifndef __WIMAX_INTERNAL_H__ */ From ea912f4e7f264981faf8665cfb63d46d7f948117 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:35 -0800 Subject: [PATCH 099/123] wimax: debug macros and debug settings for the WiMAX stack This file contains a simple debug framework that is used in the stack; it allows the debug level to be controlled at compile-time (so the debug code is optimized out) and at run-time (for what wasn't compiled out). This is eventually going to be moved to use dynamic_printk(). Just need to find time to do it. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- include/linux/wimax/debug.h | 453 ++++++++++++++++++++++++++++++++++++ net/wimax/debug-levels.h | 42 ++++ 2 files changed, 495 insertions(+) create mode 100644 include/linux/wimax/debug.h create mode 100644 net/wimax/debug-levels.h diff --git a/include/linux/wimax/debug.h b/include/linux/wimax/debug.h new file mode 100644 index 00000000000..ba0c49399a8 --- /dev/null +++ b/include/linux/wimax/debug.h @@ -0,0 +1,453 @@ +/* + * Linux WiMAX + * Collection of tools to manage debug operations. + * + * + * Copyright (C) 2005-2007 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * Don't #include this file directly, read on! + * + * + * EXECUTING DEBUGGING ACTIONS OR NOT + * + * The main thing this framework provides is decission power to take a + * debug action (like printing a message) if the current debug level + * allows it. + * + * The decission power is at two levels: at compile-time (what does + * not make it is compiled out) and at run-time. The run-time + * selection is done per-submodule (as they are declared by the user + * of the framework). + * + * A call to d_test(L) (L being the target debug level) returns true + * if the action should be taken because the current debug levels + * allow it (both compile and run time). + * + * It follows that a call to d_test() that can be determined to be + * always false at compile time will get the code depending on it + * compiled out by optimization. + * + * + * DEBUG LEVELS + * + * It is up to the caller to define how much a debugging level is. + * + * Convention sets 0 as "no debug" (so an action marked as debug level 0 + * will always be taken). The increasing debug levels are used for + * increased verbosity. + * + * + * USAGE + * + * Group the code in modules and submodules inside each module [which + * in most cases maps to Linux modules and .c files that compose + * those]. + * + * + * For each module, there is: + * + * - a MODULENAME (single word, legal C identifier) + * + * - a debug-levels.h header file that declares the list of + * submodules and that is included by all .c files that use + * the debugging tools. The file name can be anything. + * + * - some (optional) .c code to manipulate the runtime debug levels + * through debugfs. + * + * The debug-levels.h file would look like: + * + * #ifndef __debug_levels__h__ + * #define __debug_levels__h__ + * + * #define D_MODULENAME modulename + * #define D_MASTER 10 + * + * #include + * + * enum d_module { + * D_SUBMODULE_DECLARE(submodule_1), + * D_SUBMODULE_DECLARE(submodule_2), + * ... + * D_SUBMODULE_DECLARE(submodule_N) + * }; + * + * #endif + * + * D_MASTER is the maximum compile-time debug level; any debug actions + * above this will be out. D_MODULENAME is the module name (legal C + * identifier), which has to be unique for each module (to avoid + * namespace collisions during linkage). Note those #defines need to + * be done before #including debug.h + * + * We declare N different submodules whose debug level can be + * independently controlled during runtime. + * + * In a .c file of the module (and only in one of them), define the + * following code: + * + * struct d_level D_LEVEL[] = { + * D_SUBMODULE_DEFINE(submodule_1), + * D_SUBMODULE_DEFINE(submodule_2), + * ... + * D_SUBMODULE_DEFINE(submodule_N), + * }; + * size_t D_LEVEL_SIZE = ARRAY_SIZE(D_LEVEL); + * + * Externs for d_level_MODULENAME and d_level_size_MODULENAME are used + * and declared in this file using the D_LEVEL and D_LEVEL_SIZE macros + * #defined also in this file. + * + * To manipulate from user space the levels, create a debugfs dentry + * and then register each submodule with: + * + * result = d_level_register_debugfs("PREFIX_", submodule_X, parent); + * if (result < 0) + * goto error; + * + * Where PREFIX_ is a name of your chosing. This will create debugfs + * file with a single numeric value that can be use to tweak it. To + * remove the entires, just use debugfs_remove_recursive() on 'parent'. + * + * NOTE: remember that even if this will show attached to some + * particular instance of a device, the settings are *global*. + * + * + * On each submodule (for example, .c files), the debug infrastructure + * should be included like this: + * + * #define D_SUBMODULE submodule_x // matches one in debug-levels.h + * #include "debug-levels.h" + * + * after #including all your include files. + * + * + * Now you can use the d_*() macros below [d_test(), d_fnstart(), + * d_fnend(), d_printf(), d_dump()]. + * + * If their debug level is greater than D_MASTER, they will be + * compiled out. + * + * If their debug level is lower or equal than D_MASTER but greater + * than the current debug level of their submodule, they'll be + * ignored. + * + * Otherwise, the action will be performed. + */ +#ifndef __debug__h__ +#define __debug__h__ + +#include +#include + + +/* Backend stuff */ + +/* + * Debug backend: generate a message header from a 'struct device' + * + * @head: buffer where to place the header + * @head_size: length of @head + * @dev: pointer to device used to generate a header from. If NULL, + * an empty ("") header is generated. + */ +static inline +void __d_head(char *head, size_t head_size, + struct device *dev) +{ + if (dev == NULL) + head[0] = 0; + else if ((unsigned long)dev < 4096) { + printk(KERN_ERR "E: Corrupt dev %p\n", dev); + WARN_ON(1); + } else + snprintf(head, head_size, "%s %s: ", + dev_driver_string(dev), dev->bus_id); +} + + +/* + * Debug backend: log some message if debugging is enabled + * + * @l: intended debug level + * @tag: tag to prefix the message with + * @dev: 'struct device' associated to this message + * @f: printf-like format and arguments + * + * Note this is optimized out if it doesn't pass the compile-time + * check; however, it is *always* compiled. This is useful to make + * sure the printf-like formats and variables are always checked and + * they don't get bit rot if you have all the debugging disabled. + */ +#define _d_printf(l, tag, dev, f, a...) \ +do { \ + char head[64]; \ + if (!d_test(l)) \ + break; \ + __d_head(head, sizeof(head), dev); \ + printk(KERN_ERR "%s%s%s: " f, head, __func__, tag, ##a); \ +} while (0) + + +/* + * CPP sintatic sugar to generate A_B like symbol names when one of + * the arguments is a a preprocessor #define. + */ +#define __D_PASTE__(varname, modulename) varname##_##modulename +#define __D_PASTE(varname, modulename) (__D_PASTE__(varname, modulename)) +#define _D_SUBMODULE_INDEX(_name) (D_SUBMODULE_DECLARE(_name)) + + +/* + * Store a submodule's runtime debug level and name + */ +struct d_level { + u8 level; + const char *name; +}; + + +/* + * List of available submodules and their debug levels + * + * We call them d_level_MODULENAME and d_level_size_MODULENAME; the + * macros D_LEVEL and D_LEVEL_SIZE contain the name already for + * convenience. + * + * This array and the size are defined on some .c file that is part of + * the current module. + */ +#define D_LEVEL __D_PASTE(d_level, D_MODULENAME) +#define D_LEVEL_SIZE __D_PASTE(d_level_size, D_MODULENAME) + +extern struct d_level D_LEVEL[]; +extern size_t D_LEVEL_SIZE; + + +/* + * Frontend stuff + * + * + * Stuff you need to declare prior to using the actual "debug" actions + * (defined below). + */ + +#ifndef D_MODULENAME +#error D_MODULENAME is not defined in your debug-levels.h file +/** + * D_MODULE - Name of the current module + * + * #define in your module's debug-levels.h, making sure it is + * unique. This has to be a legal C identifier. + */ +#define D_MODULENAME undefined_modulename +#endif + + +#ifndef D_MASTER +#warning D_MASTER not defined, but debug.h included! [see docs] +/** + * D_MASTER - Compile time maximum debug level + * + * #define in your debug-levels.h file to the maximum debug level the + * runtime code will be allowed to have. This allows you to provide a + * main knob. + * + * Anything above that level will be optimized out of the compile. + * + * Defaults to zero (no debug code compiled in). + * + * Maximum one definition per module (at the debug-levels.h file). + */ +#define D_MASTER 0 +#endif + +#ifndef D_SUBMODULE +#error D_SUBMODULE not defined, but debug.h included! [see docs] +/** + * D_SUBMODULE - Name of the current submodule + * + * #define in your submodule .c file before #including debug-levels.h + * to the name of the current submodule as previously declared and + * defined with D_SUBMODULE_DECLARE() (in your module's + * debug-levels.h) and D_SUBMODULE_DEFINE(). + * + * This is used to provide runtime-control over the debug levels. + * + * Maximum one per .c file! Can be shared among different .c files + * (meaning they belong to the same submodule categorization). + */ +#define D_SUBMODULE undefined_module +#endif + + +/** + * D_SUBMODULE_DECLARE - Declare a submodule for runtime debug level control + * + * @_name: name of the submodule, restricted to the chars that make up a + * valid C identifier ([a-zA-Z0-9_]). + * + * Declare in the module's debug-levels.h header file as: + * + * enum d_module { + * D_SUBMODULE_DECLARE(submodule_1), + * D_SUBMODULE_DECLARE(submodule_2), + * D_SUBMODULE_DECLARE(submodule_3), + * }; + * + * Some corresponding .c file needs to have a matching + * D_SUBMODULE_DEFINE(). + */ +#define D_SUBMODULE_DECLARE(_name) __D_SUBMODULE_##_name + + +/** + * D_SUBMODULE_DEFINE - Define a submodule for runtime debug level control + * + * @_name: name of the submodule, restricted to the chars that make up a + * valid C identifier ([a-zA-Z0-9_]). + * + * Use once per module (in some .c file) as: + * + * static + * struct d_level d_level_SUBMODULENAME[] = { + * D_SUBMODULE_DEFINE(submodule_1), + * D_SUBMODULE_DEFINE(submodule_2), + * D_SUBMODULE_DEFINE(submodule_3), + * }; + * size_t d_level_size_SUBDMODULENAME = ARRAY_SIZE(d_level_SUBDMODULENAME); + * + * Matching D_SUBMODULE_DECLARE()s have to be present in a + * debug-levels.h header file. + */ +#define D_SUBMODULE_DEFINE(_name) \ +[__D_SUBMODULE_##_name] = { \ + .level = 0, \ + .name = #_name \ +} + + + +/* The actual "debug" operations */ + + +/** + * d_test - Returns true if debugging should be enabled + * + * @l: intended debug level (unsigned) + * + * If the master debug switch is enabled and the current settings are + * higher or equal to the requested level, then debugging + * output/actions should be enabled. + * + * NOTE: + * + * This needs to be coded so that it can be evaluated in compile + * time; this is why the ugly BUG_ON() is placed in there, so the + * D_MASTER evaluation compiles all out if it is compile-time false. + */ +#define d_test(l) \ +({ \ + unsigned __l = l; /* type enforcer */ \ + (D_MASTER) >= __l \ + && ({ \ + BUG_ON(_D_SUBMODULE_INDEX(D_SUBMODULE) >= D_LEVEL_SIZE);\ + D_LEVEL[_D_SUBMODULE_INDEX(D_SUBMODULE)].level >= __l; \ + }); \ +}) + + +/** + * d_fnstart - log message at function start if debugging enabled + * + * @l: intended debug level + * @_dev: 'struct device' pointer, NULL if none (for context) + * @f: printf-like format and arguments + */ +#define d_fnstart(l, _dev, f, a...) _d_printf(l, " FNSTART", _dev, f, ## a) + + +/** + * d_fnend - log message at function end if debugging enabled + * + * @l: intended debug level + * @_dev: 'struct device' pointer, NULL if none (for context) + * @f: printf-like format and arguments + */ +#define d_fnend(l, _dev, f, a...) _d_printf(l, " FNEND", _dev, f, ## a) + + +/** + * d_printf - log message if debugging enabled + * + * @l: intended debug level + * @_dev: 'struct device' pointer, NULL if none (for context) + * @f: printf-like format and arguments + */ +#define d_printf(l, _dev, f, a...) _d_printf(l, "", _dev, f, ## a) + + +/** + * d_dump - log buffer hex dump if debugging enabled + * + * @l: intended debug level + * @_dev: 'struct device' pointer, NULL if none (for context) + * @f: printf-like format and arguments + */ +#define d_dump(l, dev, ptr, size) \ +do { \ + char head[64]; \ + if (!d_test(l)) \ + break; \ + __d_head(head, sizeof(head), dev); \ + print_hex_dump(KERN_ERR, head, 0, 16, 1, \ + ((void *) ptr), (size), 0); \ +} while (0) + + +/** + * Export a submodule's debug level over debugfs as PREFIXSUBMODULE + * + * @prefix: string to prefix the name with + * @submodule: name of submodule (not a string, just the name) + * @dentry: debugfs parent dentry + * + * Returns: 0 if ok, < 0 errno on error. + * + * For removing, just use debugfs_remove_recursive() on the parent. + */ +#define d_level_register_debugfs(prefix, name, parent) \ +({ \ + int rc; \ + struct dentry *fd; \ + struct dentry *verify_parent_type = parent; \ + fd = debugfs_create_u8( \ + prefix #name, 0600, verify_parent_type, \ + &(D_LEVEL[__D_SUBMODULE_ ## name].level)); \ + rc = PTR_ERR(fd); \ + if (IS_ERR(fd) && rc != -ENODEV) \ + printk(KERN_ERR "%s: Can't create debugfs entry %s: " \ + "%d\n", __func__, prefix #name, rc); \ + else \ + rc = 0; \ + rc; \ +}) + + +#endif /* #ifndef __debug__h__ */ diff --git a/net/wimax/debug-levels.h b/net/wimax/debug-levels.h new file mode 100644 index 00000000000..1c29123a3aa --- /dev/null +++ b/net/wimax/debug-levels.h @@ -0,0 +1,42 @@ +/* + * Linux WiMAX Stack + * Debug levels control file for the wimax module + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + */ +#ifndef __debug_levels__h__ +#define __debug_levels__h__ + +/* Maximum compile and run time debug level for all submodules */ +#define D_MODULENAME wimax +#define D_MASTER CONFIG_WIMAX_DEBUG_LEVEL + +#include + +/* List of all the enabled modules */ +enum d_module { + D_SUBMODULE_DECLARE(debugfs), + D_SUBMODULE_DECLARE(id_table), + D_SUBMODULE_DECLARE(op_msg), + D_SUBMODULE_DECLARE(op_reset), + D_SUBMODULE_DECLARE(op_rfkill), + D_SUBMODULE_DECLARE(stack), +}; + +#endif /* #ifndef __debug_levels__h__ */ From 15530dfd330bd19d14e096f88c70355a61fda3f2 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:36 -0800 Subject: [PATCH 100/123] wimax: generic device management (registration, deregistration, lookup) Implements the basic life cycles of a 'struct wimax_dev', some common generic netlink functionality for marshalling calls to user space, and the device state machine. For looking up net devices based on their generic netlink family IDs, use a low overhead method that optimizes for the case where most systems have a single WiMAX device, or at most, a very low number of WiMAX adaptors. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- net/wimax/id-table.c | 142 ++++++++++ net/wimax/stack.c | 599 +++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 741 insertions(+) create mode 100644 net/wimax/id-table.c create mode 100644 net/wimax/stack.c diff --git a/net/wimax/id-table.c b/net/wimax/id-table.c new file mode 100644 index 00000000000..d3b88558682 --- /dev/null +++ b/net/wimax/id-table.c @@ -0,0 +1,142 @@ +/* + * Linux WiMAX + * Mappping of generic netlink family IDs to net devices + * + * + * Copyright (C) 2005-2006 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * We assign a single generic netlink family ID to each device (to + * simplify lookup). + * + * We need a way to map family ID to a wimax_dev pointer. + * + * The idea is to use a very simple lookup. Using a netlink attribute + * with (for example) the interface name implies a heavier search over + * all the network devices; seemed kind of a waste given that we know + * we are looking for a WiMAX device and that most systems will have + * just a single WiMAX adapter. + * + * We put all the WiMAX devices in the system in a linked list and + * match the generic link family ID against the list. + * + * By using a linked list, the case of a single adapter in the system + * becomes (almost) no overhead, while still working for many more. If + * it ever goes beyond two, I'll be surprised. + */ +#include +#include +#include +#include +#include +#include "wimax-internal.h" + + +#define D_SUBMODULE id_table +#include "debug-levels.h" + + +static DEFINE_SPINLOCK(wimax_id_table_lock); +static struct list_head wimax_id_table = LIST_HEAD_INIT(wimax_id_table); + + +/* + * wimax_id_table_add - add a gennetlink familiy ID / wimax_dev mapping + * + * @wimax_dev: WiMAX device descriptor to associate to the Generic + * Netlink family ID. + * + * Look for an empty spot in the ID table; if none found, double the + * table's size and get the first spot. + */ +void wimax_id_table_add(struct wimax_dev *wimax_dev) +{ + d_fnstart(3, NULL, "(wimax_dev %p)\n", wimax_dev); + spin_lock(&wimax_id_table_lock); + list_add(&wimax_dev->id_table_node, &wimax_id_table); + spin_unlock(&wimax_id_table_lock); + d_fnend(3, NULL, "(wimax_dev %p)\n", wimax_dev); +} + + +/* + * wimax_get_netdev_by_info - lookup a wimax_dev from the gennetlink info + * + * The generic netlink family ID has been filled out in the + * nlmsghdr->nlmsg_type field, so we pull it from there, look it up in + * the mapping table and reference the wimax_dev. + * + * When done, the reference should be dropped with + * 'dev_put(wimax_dev->net_dev)'. + */ +struct wimax_dev *wimax_dev_get_by_genl_info( + struct genl_info *info, int ifindex) +{ + struct wimax_dev *wimax_dev = NULL; + + d_fnstart(3, NULL, "(info %p ifindex %d)\n", info, ifindex); + spin_lock(&wimax_id_table_lock); + list_for_each_entry(wimax_dev, &wimax_id_table, id_table_node) { + if (wimax_dev->net_dev->ifindex == ifindex) { + dev_hold(wimax_dev->net_dev); + break; + } + } + if (wimax_dev == NULL) + d_printf(1, NULL, "wimax: no devices found with ifindex %d\n", + ifindex); + spin_unlock(&wimax_id_table_lock); + d_fnend(3, NULL, "(info %p ifindex %d) = %p\n", + info, ifindex, wimax_dev); + return wimax_dev; +} + + +/* + * wimax_id_table_rm - Remove a gennetlink familiy ID / wimax_dev mapping + * + * @id: family ID to remove from the table + */ +void wimax_id_table_rm(struct wimax_dev *wimax_dev) +{ + spin_lock(&wimax_id_table_lock); + list_del_init(&wimax_dev->id_table_node); + spin_unlock(&wimax_id_table_lock); +} + + +/* + * Release the gennetlink family id / mapping table + * + * On debug, verify that the table is empty upon removal. + */ +void wimax_id_table_release(void) +{ +#ifndef CONFIG_BUG + return; +#endif + struct wimax_dev *wimax_dev; + + spin_lock(&wimax_id_table_lock); + list_for_each_entry(wimax_dev, &wimax_id_table, id_table_node) { + printk(KERN_ERR "BUG: %s wimax_dev %p ifindex %d not cleared\n", + __func__, wimax_dev, wimax_dev->net_dev->ifindex); + WARN_ON(1); + } + spin_unlock(&wimax_id_table_lock); +} diff --git a/net/wimax/stack.c b/net/wimax/stack.c new file mode 100644 index 00000000000..d4da92f8981 --- /dev/null +++ b/net/wimax/stack.c @@ -0,0 +1,599 @@ +/* + * Linux WiMAX + * Initialization, addition and removal of wimax devices + * + * + * Copyright (C) 2005-2006 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * This implements: + * + * - basic life cycle of 'struct wimax_dev' [wimax_dev_*()]; on + * addition/registration initialize all subfields and allocate + * generic netlink resources for user space communication. On + * removal/unregistration, undo all that. + * + * - device state machine [wimax_state_change()] and support to send + * reports to user space when the state changes + * [wimax_gnl_re_state_change*()]. + * + * See include/net/wimax.h for rationales and design. + * + * ROADMAP + * + * [__]wimax_state_change() Called by drivers to update device's state + * wimax_gnl_re_state_change_alloc() + * wimax_gnl_re_state_change_send() + * + * wimax_dev_init() Init a device + * wimax_dev_add() Register + * wimax_rfkill_add() + * wimax_gnl_add() Register all the generic netlink resources. + * wimax_id_table_add() + * wimax_dev_rm() Unregister + * wimax_id_table_rm() + * wimax_gnl_rm() + * wimax_rfkill_rm() + */ +#include +#include +#include +#include +#include "wimax-internal.h" + + +#define D_SUBMODULE stack +#include "debug-levels.h" + +/* + * Authoritative source for the RE_STATE_CHANGE attribute policy + * + * We don't really use it here, but /me likes to keep the definition + * close to where the data is generated. + */ +/* +static const +struct nla_policy wimax_gnl_re_status_change[WIMAX_GNL_ATTR_MAX + 1] = { + [WIMAX_GNL_STCH_STATE_OLD] = { .type = NLA_U8 }, + [WIMAX_GNL_STCH_STATE_NEW] = { .type = NLA_U8 }, +}; +*/ + + +/* + * Allocate a Report State Change message + * + * @header: save it, you need it for _send() + * + * Creates and fills a basic state change message; different code + * paths can then add more attributes to the message as needed. + * + * Use wimax_gnl_re_state_change_send() to send the returned skb. + * + * Returns: skb with the genl message if ok, IS_ERR() ptr on error + * with an errno code. + */ +static +struct sk_buff *wimax_gnl_re_state_change_alloc( + struct wimax_dev *wimax_dev, + enum wimax_st new_state, enum wimax_st old_state, + void **header) +{ + int result; + struct device *dev = wimax_dev_to_dev(wimax_dev); + void *data; + struct sk_buff *report_skb; + + d_fnstart(3, dev, "(wimax_dev %p new_state %u old_state %u)\n", + wimax_dev, new_state, old_state); + result = -ENOMEM; + report_skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL); + if (report_skb == NULL) { + dev_err(dev, "RE_STCH: can't create message\n"); + goto error_new; + } + data = genlmsg_put(report_skb, 0, wimax_gnl_mcg.id, &wimax_gnl_family, + 0, WIMAX_GNL_RE_STATE_CHANGE); + if (data == NULL) { + dev_err(dev, "RE_STCH: can't put data into message\n"); + goto error_put; + } + *header = data; + + result = nla_put_u8(report_skb, WIMAX_GNL_STCH_STATE_OLD, old_state); + if (result < 0) { + dev_err(dev, "RE_STCH: Error adding OLD attr: %d\n", result); + goto error_put; + } + result = nla_put_u8(report_skb, WIMAX_GNL_STCH_STATE_NEW, new_state); + if (result < 0) { + dev_err(dev, "RE_STCH: Error adding NEW attr: %d\n", result); + goto error_put; + } + result = nla_put_u32(report_skb, WIMAX_GNL_STCH_IFIDX, + wimax_dev->net_dev->ifindex); + if (result < 0) { + dev_err(dev, "RE_STCH: Error adding IFINDEX attribute\n"); + goto error_put; + } + d_fnend(3, dev, "(wimax_dev %p new_state %u old_state %u) = %p\n", + wimax_dev, new_state, old_state, report_skb); + return report_skb; + +error_put: + nlmsg_free(report_skb); +error_new: + d_fnend(3, dev, "(wimax_dev %p new_state %u old_state %u) = %d\n", + wimax_dev, new_state, old_state, result); + return ERR_PTR(result); +} + + +/* + * Send a Report State Change message (as created with _alloc). + * + * @report_skb: as returned by wimax_gnl_re_state_change_alloc() + * @header: as returned by wimax_gnl_re_state_change_alloc() + * + * Returns: 0 if ok, < 0 errno code on error. + * + * If the message is NULL, pretend it didn't happen. + */ +static +int wimax_gnl_re_state_change_send( + struct wimax_dev *wimax_dev, struct sk_buff *report_skb, + void *header) +{ + int result = 0; + struct device *dev = wimax_dev_to_dev(wimax_dev); + d_fnstart(3, dev, "(wimax_dev %p report_skb %p)\n", + wimax_dev, report_skb); + if (report_skb == NULL) + goto out; + genlmsg_end(report_skb, header); + result = genlmsg_multicast(report_skb, 0, wimax_gnl_mcg.id, GFP_KERNEL); + if (result == -ESRCH) /* Nobody connected, ignore it */ + result = 0; /* btw, the skb is freed already */ + if (result < 0) { + dev_err(dev, "RE_STCH: Error sending: %d\n", result); + nlmsg_free(report_skb); + } +out: + d_fnend(3, dev, "(wimax_dev %p report_skb %p) = %d\n", + wimax_dev, report_skb, result); + return result; +} + + +static +void __check_new_state(enum wimax_st old_state, enum wimax_st new_state, + unsigned allowed_states_bm) +{ + if (WARN_ON(((1 << new_state) & allowed_states_bm) == 0)) { + printk(KERN_ERR "SW BUG! Forbidden state change %u -> %u\n", + old_state, new_state); + } +} + + +/* + * Set the current state of a WiMAX device [unlocking version of + * wimax_state_change(). + */ +void __wimax_state_change(struct wimax_dev *wimax_dev, enum wimax_st new_state) +{ + struct device *dev = wimax_dev_to_dev(wimax_dev); + enum wimax_st old_state = wimax_dev->state; + struct sk_buff *stch_skb; + void *header; + + d_fnstart(3, dev, "(wimax_dev %p new_state %u [old %u])\n", + wimax_dev, new_state, old_state); + + if (WARN_ON(new_state >= __WIMAX_ST_INVALID)) { + dev_err(dev, "SW BUG: requesting invalid state %u\n", + new_state); + goto out; + } + if (old_state == new_state) + goto out; + header = NULL; /* gcc complains? can't grok why */ + stch_skb = wimax_gnl_re_state_change_alloc( + wimax_dev, new_state, old_state, &header); + + /* Verify the state transition and do exit-from-state actions */ + switch (old_state) { + case __WIMAX_ST_NULL: + __check_new_state(old_state, new_state, + 1 << WIMAX_ST_DOWN); + break; + case WIMAX_ST_DOWN: + __check_new_state(old_state, new_state, + 1 << __WIMAX_ST_QUIESCING + | 1 << WIMAX_ST_UNINITIALIZED + | 1 << WIMAX_ST_RADIO_OFF); + break; + case __WIMAX_ST_QUIESCING: + __check_new_state(old_state, new_state, 1 << WIMAX_ST_DOWN); + break; + case WIMAX_ST_UNINITIALIZED: + __check_new_state(old_state, new_state, + 1 << __WIMAX_ST_QUIESCING + | 1 << WIMAX_ST_RADIO_OFF); + break; + case WIMAX_ST_RADIO_OFF: + __check_new_state(old_state, new_state, + 1 << __WIMAX_ST_QUIESCING + | 1 << WIMAX_ST_READY); + break; + case WIMAX_ST_READY: + __check_new_state(old_state, new_state, + 1 << __WIMAX_ST_QUIESCING + | 1 << WIMAX_ST_RADIO_OFF + | 1 << WIMAX_ST_SCANNING + | 1 << WIMAX_ST_CONNECTING + | 1 << WIMAX_ST_CONNECTED); + break; + case WIMAX_ST_SCANNING: + __check_new_state(old_state, new_state, + 1 << __WIMAX_ST_QUIESCING + | 1 << WIMAX_ST_RADIO_OFF + | 1 << WIMAX_ST_READY + | 1 << WIMAX_ST_CONNECTING + | 1 << WIMAX_ST_CONNECTED); + break; + case WIMAX_ST_CONNECTING: + __check_new_state(old_state, new_state, + 1 << __WIMAX_ST_QUIESCING + | 1 << WIMAX_ST_RADIO_OFF + | 1 << WIMAX_ST_READY + | 1 << WIMAX_ST_SCANNING + | 1 << WIMAX_ST_CONNECTED); + break; + case WIMAX_ST_CONNECTED: + __check_new_state(old_state, new_state, + 1 << __WIMAX_ST_QUIESCING + | 1 << WIMAX_ST_RADIO_OFF + | 1 << WIMAX_ST_READY); + netif_tx_disable(wimax_dev->net_dev); + netif_carrier_off(wimax_dev->net_dev); + break; + case __WIMAX_ST_INVALID: + default: + dev_err(dev, "SW BUG: wimax_dev %p is in unknown state %u\n", + wimax_dev, wimax_dev->state); + WARN_ON(1); + goto out; + } + + /* Execute the actions of entry to the new state */ + switch (new_state) { + case __WIMAX_ST_NULL: + dev_err(dev, "SW BUG: wimax_dev %p entering NULL state " + "from %u\n", wimax_dev, wimax_dev->state); + WARN_ON(1); /* Nobody can enter this state */ + break; + case WIMAX_ST_DOWN: + break; + case __WIMAX_ST_QUIESCING: + break; + case WIMAX_ST_UNINITIALIZED: + break; + case WIMAX_ST_RADIO_OFF: + break; + case WIMAX_ST_READY: + break; + case WIMAX_ST_SCANNING: + break; + case WIMAX_ST_CONNECTING: + break; + case WIMAX_ST_CONNECTED: + netif_carrier_on(wimax_dev->net_dev); + netif_wake_queue(wimax_dev->net_dev); + break; + case __WIMAX_ST_INVALID: + default: + BUG(); + } + __wimax_state_set(wimax_dev, new_state); + if (stch_skb) + wimax_gnl_re_state_change_send(wimax_dev, stch_skb, header); +out: + d_fnend(3, dev, "(wimax_dev %p new_state %u [old %u]) = void\n", + wimax_dev, new_state, old_state); + return; +} + + +/** + * wimax_state_change - Set the current state of a WiMAX device + * + * @wimax_dev: WiMAX device descriptor (properly referenced) + * @new_state: New state to switch to + * + * This implements the state changes for the wimax devices. It will + * + * - verify that the state transition is legal (for now it'll just + * print a warning if not) according to the table in + * linux/wimax.h's documentation for 'enum wimax_st'. + * + * - perform the actions needed for leaving the current state and + * whichever are needed for entering the new state. + * + * - issue a report to user space indicating the new state (and an + * optional payload with information about the new state). + * + * NOTE: @wimax_dev must be locked + */ +void wimax_state_change(struct wimax_dev *wimax_dev, enum wimax_st new_state) +{ + mutex_lock(&wimax_dev->mutex); + __wimax_state_change(wimax_dev, new_state); + mutex_unlock(&wimax_dev->mutex); + return; +} +EXPORT_SYMBOL_GPL(wimax_state_change); + + +/** + * wimax_state_get() - Return the current state of a WiMAX device + * + * @wimax_dev: WiMAX device descriptor + * + * Returns: Current state of the device according to its driver. + */ +enum wimax_st wimax_state_get(struct wimax_dev *wimax_dev) +{ + enum wimax_st state; + mutex_lock(&wimax_dev->mutex); + state = wimax_dev->state; + mutex_unlock(&wimax_dev->mutex); + return state; +} +EXPORT_SYMBOL_GPL(wimax_state_get); + + +/** + * wimax_dev_init - initialize a newly allocated instance + * + * @wimax_dev: WiMAX device descriptor to initialize. + * + * Initializes fields of a freshly allocated @wimax_dev instance. This + * function assumes that after allocation, the memory occupied by + * @wimax_dev was zeroed. + */ +void wimax_dev_init(struct wimax_dev *wimax_dev) +{ + INIT_LIST_HEAD(&wimax_dev->id_table_node); + __wimax_state_set(wimax_dev, WIMAX_ST_UNINITIALIZED); + mutex_init(&wimax_dev->mutex); + mutex_init(&wimax_dev->mutex_reset); +} +EXPORT_SYMBOL_GPL(wimax_dev_init); + +/* + * This extern is declared here because it's easier to keep track -- + * both declarations are a list of the same + */ +extern struct genl_ops + wimax_gnl_msg_from_user, + wimax_gnl_reset, + wimax_gnl_rfkill; + +static +struct genl_ops *wimax_gnl_ops[] = { + &wimax_gnl_msg_from_user, + &wimax_gnl_reset, + &wimax_gnl_rfkill, +}; + + +static +size_t wimax_addr_scnprint(char *addr_str, size_t addr_str_size, + unsigned char *addr, size_t addr_len) +{ + unsigned cnt, total; + for (total = cnt = 0; cnt < addr_len; cnt++) + total += scnprintf(addr_str + total, addr_str_size - total, + "%02x%c", addr[cnt], + cnt == addr_len - 1 ? '\0' : ':'); + return total; +} + + +/** + * wimax_dev_add - Register a new WiMAX device + * + * @wimax_dev: WiMAX device descriptor (as embedded in your @net_dev's + * priv data). You must have called wimax_dev_init() on it before. + * + * @net_dev: net device the @wimax_dev is associated with. The + * function expects SET_NETDEV_DEV() and register_netdev() were + * already called on it. + * + * Registers the new WiMAX device, sets up the user-kernel control + * interface (generic netlink) and common WiMAX infrastructure. + * + * Note that the parts that will allow interaction with user space are + * setup at the very end, when the rest is in place, as once that + * happens, the driver might get user space control requests via + * netlink or from debugfs that might translate into calls into + * wimax_dev->op_*(). + */ +int wimax_dev_add(struct wimax_dev *wimax_dev, struct net_device *net_dev) +{ + int result; + struct device *dev = net_dev->dev.parent; + char addr_str[32]; + + d_fnstart(3, dev, "(wimax_dev %p net_dev %p)\n", wimax_dev, net_dev); + + /* Do the RFKILL setup before locking, as RFKILL will call + * into our functions. */ + wimax_dev->net_dev = net_dev; + result = wimax_rfkill_add(wimax_dev); + if (result < 0) + goto error_rfkill_add; + + /* Set up user-space interaction */ + mutex_lock(&wimax_dev->mutex); + wimax_id_table_add(wimax_dev); + result = wimax_debugfs_add(wimax_dev); + if (result < 0) { + dev_err(dev, "cannot initialize debugfs: %d\n", + result); + goto error_debugfs_add; + } + + __wimax_state_set(wimax_dev, WIMAX_ST_DOWN); + mutex_unlock(&wimax_dev->mutex); + + wimax_addr_scnprint(addr_str, sizeof(addr_str), + net_dev->dev_addr, net_dev->addr_len); + dev_err(dev, "WiMAX interface %s (%s) ready\n", + net_dev->name, addr_str); + d_fnend(3, dev, "(wimax_dev %p net_dev %p) = 0\n", wimax_dev, net_dev); + return 0; + +error_debugfs_add: + wimax_id_table_rm(wimax_dev); + mutex_unlock(&wimax_dev->mutex); + wimax_rfkill_rm(wimax_dev); +error_rfkill_add: + d_fnend(3, dev, "(wimax_dev %p net_dev %p) = %d\n", + wimax_dev, net_dev, result); + return result; +} +EXPORT_SYMBOL_GPL(wimax_dev_add); + + +/** + * wimax_dev_rm - Unregister an existing WiMAX device + * + * @wimax_dev: WiMAX device descriptor + * + * Unregisters a WiMAX device previously registered for use with + * wimax_add_rm(). + * + * IMPORTANT! Must call before calling unregister_netdev(). + * + * After this function returns, you will not get any more user space + * control requests (via netlink or debugfs) and thus to wimax_dev->ops. + * + * Reentrancy control is ensured by setting the state to + * %__WIMAX_ST_QUIESCING. rfkill operations coming through + * wimax_*rfkill*() will be stopped by the quiescing state; ops coming + * from the rfkill subsystem will be stopped by the support being + * removed by wimax_rfkill_rm(). + */ +void wimax_dev_rm(struct wimax_dev *wimax_dev) +{ + d_fnstart(3, NULL, "(wimax_dev %p)\n", wimax_dev); + + mutex_lock(&wimax_dev->mutex); + __wimax_state_change(wimax_dev, __WIMAX_ST_QUIESCING); + wimax_debugfs_rm(wimax_dev); + wimax_id_table_rm(wimax_dev); + __wimax_state_change(wimax_dev, WIMAX_ST_DOWN); + mutex_unlock(&wimax_dev->mutex); + wimax_rfkill_rm(wimax_dev); + d_fnend(3, NULL, "(wimax_dev %p) = void\n", wimax_dev); +} +EXPORT_SYMBOL_GPL(wimax_dev_rm); + +struct genl_family wimax_gnl_family = { + .id = GENL_ID_GENERATE, + .name = "WiMAX", + .version = WIMAX_GNL_VERSION, + .hdrsize = 0, + .maxattr = WIMAX_GNL_ATTR_MAX, +}; + +struct genl_multicast_group wimax_gnl_mcg = { + .name = "msg", +}; + + + +/* Shutdown the wimax stack */ +static +int __init wimax_subsys_init(void) +{ + int result, cnt; + + d_fnstart(4, NULL, "()\n"); + snprintf(wimax_gnl_family.name, sizeof(wimax_gnl_family.name), + "WiMAX"); + result = genl_register_family(&wimax_gnl_family); + if (unlikely(result < 0)) { + printk(KERN_ERR "cannot register generic netlink family: %d\n", + result); + goto error_register_family; + } + + for (cnt = 0; cnt < ARRAY_SIZE(wimax_gnl_ops); cnt++) { + result = genl_register_ops(&wimax_gnl_family, + wimax_gnl_ops[cnt]); + d_printf(4, NULL, "registering generic netlink op code " + "%u: %d\n", wimax_gnl_ops[cnt]->cmd, result); + if (unlikely(result < 0)) { + printk(KERN_ERR "cannot register generic netlink op " + "code %u: %d\n", + wimax_gnl_ops[cnt]->cmd, result); + goto error_register_ops; + } + } + + result = genl_register_mc_group(&wimax_gnl_family, &wimax_gnl_mcg); + if (result < 0) + goto error_mc_group; + d_fnend(4, NULL, "() = 0\n"); + return 0; + +error_mc_group: +error_register_ops: + for (cnt--; cnt >= 0; cnt--) + genl_unregister_ops(&wimax_gnl_family, + wimax_gnl_ops[cnt]); + genl_unregister_family(&wimax_gnl_family); +error_register_family: + d_fnend(4, NULL, "() = %d\n", result); + return result; + +} +module_init(wimax_subsys_init); + + +/* Shutdown the wimax stack */ +static +void __exit wimax_subsys_exit(void) +{ + int cnt; + wimax_id_table_release(); + genl_unregister_mc_group(&wimax_gnl_family, &wimax_gnl_mcg); + for (cnt = ARRAY_SIZE(wimax_gnl_ops) - 1; cnt >= 0; cnt--) + genl_unregister_ops(&wimax_gnl_family, + wimax_gnl_ops[cnt]); + genl_unregister_family(&wimax_gnl_family); +} +module_exit(wimax_subsys_exit); + +MODULE_AUTHOR("Intel Corporation "); +MODULE_DESCRIPTION("Linux WiMAX stack"); +MODULE_LICENSE("GPL"); + From 3efb40c2c6eea315abcf19239c15d088e2498299 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:37 -0800 Subject: [PATCH 101/123] genetlink: export genl_unregister_mc_group() Add an EXPORT_SYMBOL() to genl_unregister_mc_group(), to allow unregistering groups on the run. EXPORT_SYMBOL_GPL() is not used as the rest of the functions exported by this module (eg: genl_register_mc_group) are also not _GPL(). Cleanup is currently done when unregistering a family, but there is no way to unregister a single multicast group due to that function not being exported. Seems to be a mistake as it is documented as for external consumption. This is needed by the WiMAX stack to be able to cleanup unused mc groups. Signed-off-by: Inaky Perez-Gonzalez Acked-by: Johannes Berg Signed-off-by: Greg Kroah-Hartman --- net/netlink/genetlink.c | 1 + 1 file changed, 1 insertion(+) diff --git a/net/netlink/genetlink.c b/net/netlink/genetlink.c index 3e1191cecaf..1d3dd30099d 100644 --- a/net/netlink/genetlink.c +++ b/net/netlink/genetlink.c @@ -225,6 +225,7 @@ void genl_unregister_mc_group(struct genl_family *family, __genl_unregister_mc_group(family, grp); genl_unlock(); } +EXPORT_SYMBOL(genl_unregister_mc_group); static void genl_unregister_mc_groups(struct genl_family *family) { From 3e65646bb12be03b14dff53907391095a52d5f49 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:38 -0800 Subject: [PATCH 102/123] wimax: basic API: kernel/user messaging, rfkill and reset Implements the three basic operations provided by the stack's control interface to WiMAX devices: - Messaging channel between user space and driver/device This implements a direct communication channel between user space and the driver/device, by which free form messages can be sent back and forth. This is intended for device-specific features, vendor quirks, etc. - RF-kill framework integration Provide most of the RF-Kill integration for WiMAX drivers so that all device drivers have to do is after wimax_dev_add() is call wimax_report_rfkill_{hw,sw}() to update initial state and then every time it changes. Provides wimax_rfkill() for the kernel to call to set software RF-Kill status and/or query current hardware and software switch status. Exports wimax_rfkill() over generic netlink to user space. - Reset a WiMAX device Provides wimax_reset() for the kernel to reset a wimax device as needed and exports it over generic netlink to user space. This API is clearly limited, as it still provides no way to do the basic scan, connect and disconnect in a hardware independent way. The WiMAX case is more complex than WiFi due to the way networks are discovered and provisioned. The next developments are to add the basic operations so they can be offerent by different drivers. However, we'd like to get more vendors to jump in and provide feedback of how the user/kernel API/abstraction layer should be. The user space code for the i2400m, as of now, uses the messaging channel, but that will change as the API evolves. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- net/wimax/op-msg.c | 421 +++++++++++++++++++++++++++++++++ net/wimax/op-reset.c | 143 ++++++++++++ net/wimax/op-rfkill.c | 532 ++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 1096 insertions(+) create mode 100644 net/wimax/op-msg.c create mode 100644 net/wimax/op-reset.c create mode 100644 net/wimax/op-rfkill.c diff --git a/net/wimax/op-msg.c b/net/wimax/op-msg.c new file mode 100644 index 00000000000..cb3b4ad5368 --- /dev/null +++ b/net/wimax/op-msg.c @@ -0,0 +1,421 @@ +/* + * Linux WiMAX + * Generic messaging interface between userspace and driver/device + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * This implements a direct communication channel between user space and + * the driver/device, by which free form messages can be sent back and + * forth. + * + * This is intended for device-specific features, vendor quirks, etc. + * + * See include/net/wimax.h + * + * GENERIC NETLINK ENCODING AND CAPACITY + * + * A destination "pipe name" is added to each message; it is up to the + * drivers to assign or use those names (if using them at all). + * + * Messages are encoded as a binary netlink attribute using nla_put() + * using type NLA_UNSPEC (as some versions of libnl still in + * deployment don't yet understand NLA_BINARY). + * + * The maximum capacity of this transport is PAGESIZE per message (so + * the actual payload will be bit smaller depending on the + * netlink/generic netlink attributes and headers). + * + * RECEPTION OF MESSAGES + * + * When a message is received from user space, it is passed verbatim + * to the driver calling wimax_dev->op_msg_from_user(). The return + * value from this function is passed back to user space as an ack + * over the generic netlink protocol. + * + * The stack doesn't do any processing or interpretation of these + * messages. + * + * SENDING MESSAGES + * + * Messages can be sent with wimax_msg(). + * + * If the message delivery needs to happen on a different context to + * that of its creation, wimax_msg_alloc() can be used to get a + * pointer to the message that can be delivered later on with + * wimax_msg_send(). + * + * ROADMAP + * + * wimax_gnl_doit_msg_from_user() Process a message from user space + * wimax_dev_get_by_genl_info() + * wimax_dev->op_msg_from_user() Delivery of message to the driver + * + * wimax_msg() Send a message to user space + * wimax_msg_alloc() + * wimax_msg_send() + */ +#include +#include +#include +#include +#include +#include "wimax-internal.h" + + +#define D_SUBMODULE op_msg +#include "debug-levels.h" + + +/** + * wimax_msg_alloc - Create a new skb for sending a message to userspace + * + * @wimax_dev: WiMAX device descriptor + * @pipe_name: "named pipe" the message will be sent to + * @msg: pointer to the message data to send + * @size: size of the message to send (in bytes), including the header. + * @gfp_flags: flags for memory allocation. + * + * Returns: %0 if ok, negative errno code on error + * + * Description: + * + * Allocates an skb that will contain the message to send to user + * space over the messaging pipe and initializes it, copying the + * payload. + * + * Once this call is done, you can deliver it with + * wimax_msg_send(). + * + * IMPORTANT: + * + * Don't use skb_push()/skb_pull()/skb_reserve() on the skb, as + * wimax_msg_send() depends on skb->data being placed at the + * beginning of the user message. + */ +struct sk_buff *wimax_msg_alloc(struct wimax_dev *wimax_dev, + const char *pipe_name, + const void *msg, size_t size, + gfp_t gfp_flags) +{ + int result; + struct device *dev = wimax_dev->net_dev->dev.parent; + size_t msg_size; + void *genl_msg; + struct sk_buff *skb; + + msg_size = nla_total_size(size) + + nla_total_size(sizeof(u32)) + + (pipe_name ? nla_total_size(strlen(pipe_name)) : 0); + result = -ENOMEM; + skb = genlmsg_new(msg_size, gfp_flags); + if (skb == NULL) + goto error_new; + genl_msg = genlmsg_put(skb, 0, 0, &wimax_gnl_family, + 0, WIMAX_GNL_OP_MSG_TO_USER); + if (genl_msg == NULL) { + dev_err(dev, "no memory to create generic netlink message\n"); + goto error_genlmsg_put; + } + result = nla_put_u32(skb, WIMAX_GNL_MSG_IFIDX, + wimax_dev->net_dev->ifindex); + if (result < 0) { + dev_err(dev, "no memory to add ifindex attribute\n"); + goto error_nla_put; + } + if (pipe_name) { + result = nla_put_string(skb, WIMAX_GNL_MSG_PIPE_NAME, + pipe_name); + if (result < 0) { + dev_err(dev, "no memory to add pipe_name attribute\n"); + goto error_nla_put; + } + } + result = nla_put(skb, WIMAX_GNL_MSG_DATA, size, msg); + if (result < 0) { + dev_err(dev, "no memory to add payload in attribute\n"); + goto error_nla_put; + } + genlmsg_end(skb, genl_msg); + return skb; + +error_nla_put: +error_genlmsg_put: +error_new: + nlmsg_free(skb); + return ERR_PTR(result); + +} +EXPORT_SYMBOL_GPL(wimax_msg_alloc); + + +/** + * wimax_msg_data_len - Return a pointer and size of a message's payload + * + * @msg: Pointer to a message created with wimax_msg_alloc() + * @size: Pointer to where to store the message's size + * + * Returns the pointer to the message data. + */ +const void *wimax_msg_data_len(struct sk_buff *msg, size_t *size) +{ + struct nlmsghdr *nlh = (void *) msg->head; + struct nlattr *nla; + + nla = nlmsg_find_attr(nlh, sizeof(struct genlmsghdr), + WIMAX_GNL_MSG_DATA); + if (nla == NULL) { + printk(KERN_ERR "Cannot find attribute WIMAX_GNL_MSG_DATA\n"); + return NULL; + } + *size = nla_len(nla); + return nla_data(nla); +} +EXPORT_SYMBOL_GPL(wimax_msg_data_len); + + +/** + * wimax_msg_data - Return a pointer to a message's payload + * + * @msg: Pointer to a message created with wimax_msg_alloc() + */ +const void *wimax_msg_data(struct sk_buff *msg) +{ + struct nlmsghdr *nlh = (void *) msg->head; + struct nlattr *nla; + + nla = nlmsg_find_attr(nlh, sizeof(struct genlmsghdr), + WIMAX_GNL_MSG_DATA); + if (nla == NULL) { + printk(KERN_ERR "Cannot find attribute WIMAX_GNL_MSG_DATA\n"); + return NULL; + } + return nla_data(nla); +} +EXPORT_SYMBOL_GPL(wimax_msg_data); + + +/** + * wimax_msg_len - Return a message's payload length + * + * @msg: Pointer to a message created with wimax_msg_alloc() + */ +ssize_t wimax_msg_len(struct sk_buff *msg) +{ + struct nlmsghdr *nlh = (void *) msg->head; + struct nlattr *nla; + + nla = nlmsg_find_attr(nlh, sizeof(struct genlmsghdr), + WIMAX_GNL_MSG_DATA); + if (nla == NULL) { + printk(KERN_ERR "Cannot find attribute WIMAX_GNL_MSG_DATA\n"); + return -EINVAL; + } + return nla_len(nla); +} +EXPORT_SYMBOL_GPL(wimax_msg_len); + + +/** + * wimax_msg_send - Send a pre-allocated message to user space + * + * @wimax_dev: WiMAX device descriptor + * + * @skb: &struct sk_buff returned by wimax_msg_alloc(). Note the + * ownership of @skb is transferred to this function. + * + * Returns: 0 if ok, < 0 errno code on error + * + * Description: + * + * Sends a free-form message that was preallocated with + * wimax_msg_alloc() and filled up. + * + * Assumes that once you pass an skb to this function for sending, it + * owns it and will release it when done (on success). + * + * IMPORTANT: + * + * Don't use skb_push()/skb_pull()/skb_reserve() on the skb, as + * wimax_msg_send() depends on skb->data being placed at the + * beginning of the user message. + */ +int wimax_msg_send(struct wimax_dev *wimax_dev, struct sk_buff *skb) +{ + int result; + struct device *dev = wimax_dev->net_dev->dev.parent; + void *msg = skb->data; + size_t size = skb->len; + might_sleep(); + + d_printf(1, dev, "CTX: wimax msg, %zu bytes\n", size); + d_dump(2, dev, msg, size); + result = genlmsg_multicast(skb, 0, wimax_gnl_mcg.id, GFP_KERNEL); + d_printf(1, dev, "CTX: genl multicast result %d\n", result); + if (result == -ESRCH) /* Nobody connected, ignore it */ + result = 0; /* btw, the skb is freed already */ + return result; +} +EXPORT_SYMBOL_GPL(wimax_msg_send); + + +/** + * wimax_msg - Send a message to user space + * + * @wimax_dev: WiMAX device descriptor (properly referenced) + * @pipe_name: "named pipe" the message will be sent to + * @buf: pointer to the message to send. + * @size: size of the buffer pointed to by @buf (in bytes). + * @gfp_flags: flags for memory allocation. + * + * Returns: %0 if ok, negative errno code on error. + * + * Description: + * + * Sends a free-form message to user space on the device @wimax_dev. + * + * NOTES: + * + * Once the @skb is given to this function, who will own it and will + * release it when done (unless it returns error). + */ +int wimax_msg(struct wimax_dev *wimax_dev, const char *pipe_name, + const void *buf, size_t size, gfp_t gfp_flags) +{ + int result = -ENOMEM; + struct sk_buff *skb; + + skb = wimax_msg_alloc(wimax_dev, pipe_name, buf, size, gfp_flags); + if (skb == NULL) + goto error_msg_new; + result = wimax_msg_send(wimax_dev, skb); +error_msg_new: + return result; +} +EXPORT_SYMBOL_GPL(wimax_msg); + + +static const +struct nla_policy wimax_gnl_msg_policy[WIMAX_GNL_ATTR_MAX + 1] = { + [WIMAX_GNL_MSG_IFIDX] = { + .type = NLA_U32, + }, + [WIMAX_GNL_MSG_DATA] = { + .type = NLA_UNSPEC, /* libnl doesn't grok BINARY yet */ + }, +}; + + +/* + * Relays a message from user space to the driver + * + * The skb is passed to the driver-specific function with the netlink + * and generic netlink headers already stripped. + * + * This call will block while handling/relaying the message. + */ +static +int wimax_gnl_doit_msg_from_user(struct sk_buff *skb, struct genl_info *info) +{ + int result, ifindex; + struct wimax_dev *wimax_dev; + struct device *dev; + struct nlmsghdr *nlh = info->nlhdr; + char *pipe_name; + void *msg_buf; + size_t msg_len; + + might_sleep(); + d_fnstart(3, NULL, "(skb %p info %p)\n", skb, info); + result = -ENODEV; + if (info->attrs[WIMAX_GNL_MSG_IFIDX] == NULL) { + printk(KERN_ERR "WIMAX_GNL_MSG_FROM_USER: can't find IFIDX " + "attribute\n"); + goto error_no_wimax_dev; + } + ifindex = nla_get_u32(info->attrs[WIMAX_GNL_MSG_IFIDX]); + wimax_dev = wimax_dev_get_by_genl_info(info, ifindex); + if (wimax_dev == NULL) + goto error_no_wimax_dev; + dev = wimax_dev_to_dev(wimax_dev); + + /* Unpack arguments */ + result = -EINVAL; + if (info->attrs[WIMAX_GNL_MSG_DATA] == NULL) { + dev_err(dev, "WIMAX_GNL_MSG_FROM_USER: can't find MSG_DATA " + "attribute\n"); + goto error_no_data; + } + msg_buf = nla_data(info->attrs[WIMAX_GNL_MSG_DATA]); + msg_len = nla_len(info->attrs[WIMAX_GNL_MSG_DATA]); + + if (info->attrs[WIMAX_GNL_MSG_PIPE_NAME] == NULL) + pipe_name = NULL; + else { + struct nlattr *attr = info->attrs[WIMAX_GNL_MSG_PIPE_NAME]; + size_t attr_len = nla_len(attr); + /* libnl-1.1 does not yet support NLA_NUL_STRING */ + result = -ENOMEM; + pipe_name = kstrndup(nla_data(attr), attr_len + 1, GFP_KERNEL); + if (pipe_name == NULL) + goto error_alloc; + pipe_name[attr_len] = 0; + } + mutex_lock(&wimax_dev->mutex); + result = wimax_dev_is_ready(wimax_dev); + if (result < 0) + goto error_not_ready; + result = -ENOSYS; + if (wimax_dev->op_msg_from_user == NULL) + goto error_noop; + + d_printf(1, dev, + "CRX: nlmsghdr len %u type %u flags 0x%04x seq 0x%x pid %u\n", + nlh->nlmsg_len, nlh->nlmsg_type, nlh->nlmsg_flags, + nlh->nlmsg_seq, nlh->nlmsg_pid); + d_printf(1, dev, "CRX: wimax message %zu bytes\n", msg_len); + d_dump(2, dev, msg_buf, msg_len); + + result = wimax_dev->op_msg_from_user(wimax_dev, pipe_name, + msg_buf, msg_len, info); +error_noop: +error_not_ready: + mutex_unlock(&wimax_dev->mutex); +error_alloc: + kfree(pipe_name); +error_no_data: + dev_put(wimax_dev->net_dev); +error_no_wimax_dev: + d_fnend(3, NULL, "(skb %p info %p) = %d\n", skb, info, result); + return result; +} + + +/* + * Generic Netlink glue + */ + +struct genl_ops wimax_gnl_msg_from_user = { + .cmd = WIMAX_GNL_OP_MSG_FROM_USER, + .flags = GENL_ADMIN_PERM, + .policy = wimax_gnl_msg_policy, + .doit = wimax_gnl_doit_msg_from_user, + .dumpit = NULL, +}; + diff --git a/net/wimax/op-reset.c b/net/wimax/op-reset.c new file mode 100644 index 00000000000..ca269178c4d --- /dev/null +++ b/net/wimax/op-reset.c @@ -0,0 +1,143 @@ +/* + * Linux WiMAX + * Implement and export a method for resetting a WiMAX device + * + * + * Copyright (C) 2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * This implements a simple synchronous call to reset a WiMAX device. + * + * Resets aim at being warm, keeping the device handles active; + * however, when that fails, it falls back to a cold reset (that will + * disconnect and reconnect the device). + */ + +#include +#include +#include +#include +#include "wimax-internal.h" + +#define D_SUBMODULE op_reset +#include "debug-levels.h" + + +/** + * wimax_reset - Reset a WiMAX device + * + * @wimax_dev: WiMAX device descriptor + * + * Returns: + * + * %0 if ok and a warm reset was done (the device still exists in + * the system). + * + * -%ENODEV if a cold/bus reset had to be done (device has + * disconnected and reconnected, so current handle is not valid + * any more). + * + * -%EINVAL if the device is not even registered. + * + * Any other negative error code shall be considered as + * non-recoverable. + * + * Description: + * + * Called when wanting to reset the device for any reason. Device is + * taken back to power on status. + * + * This call blocks; on succesful return, the device has completed the + * reset process and is ready to operate. + */ +int wimax_reset(struct wimax_dev *wimax_dev) +{ + int result = -EINVAL; + struct device *dev = wimax_dev_to_dev(wimax_dev); + enum wimax_st state; + + might_sleep(); + d_fnstart(3, dev, "(wimax_dev %p)\n", wimax_dev); + mutex_lock(&wimax_dev->mutex); + dev_hold(wimax_dev->net_dev); + state = wimax_dev->state; + mutex_unlock(&wimax_dev->mutex); + + if (state >= WIMAX_ST_DOWN) { + mutex_lock(&wimax_dev->mutex_reset); + result = wimax_dev->op_reset(wimax_dev); + mutex_unlock(&wimax_dev->mutex_reset); + } + dev_put(wimax_dev->net_dev); + + d_fnend(3, dev, "(wimax_dev %p) = %d\n", wimax_dev, result); + return result; +} +EXPORT_SYMBOL(wimax_reset); + + +static const +struct nla_policy wimax_gnl_reset_policy[WIMAX_GNL_ATTR_MAX + 1] = { + [WIMAX_GNL_RESET_IFIDX] = { + .type = NLA_U32, + }, +}; + + +/* + * Exporting to user space over generic netlink + * + * Parse the reset command from user space, return error code. + * + * No attributes. + */ +static +int wimax_gnl_doit_reset(struct sk_buff *skb, struct genl_info *info) +{ + int result, ifindex; + struct wimax_dev *wimax_dev; + struct device *dev; + + d_fnstart(3, NULL, "(skb %p info %p)\n", skb, info); + result = -ENODEV; + if (info->attrs[WIMAX_GNL_RESET_IFIDX] == NULL) { + printk(KERN_ERR "WIMAX_GNL_OP_RFKILL: can't find IFIDX " + "attribute\n"); + goto error_no_wimax_dev; + } + ifindex = nla_get_u32(info->attrs[WIMAX_GNL_RESET_IFIDX]); + wimax_dev = wimax_dev_get_by_genl_info(info, ifindex); + if (wimax_dev == NULL) + goto error_no_wimax_dev; + dev = wimax_dev_to_dev(wimax_dev); + /* Execute the operation and send the result back to user space */ + result = wimax_reset(wimax_dev); + dev_put(wimax_dev->net_dev); +error_no_wimax_dev: + d_fnend(3, NULL, "(skb %p info %p) = %d\n", skb, info, result); + return result; +} + + +struct genl_ops wimax_gnl_reset = { + .cmd = WIMAX_GNL_OP_RESET, + .flags = GENL_ADMIN_PERM, + .policy = wimax_gnl_reset_policy, + .doit = wimax_gnl_doit_reset, + .dumpit = NULL, +}; diff --git a/net/wimax/op-rfkill.c b/net/wimax/op-rfkill.c new file mode 100644 index 00000000000..8745bac173f --- /dev/null +++ b/net/wimax/op-rfkill.c @@ -0,0 +1,532 @@ +/* + * Linux WiMAX + * RF-kill framework integration + * + * + * Copyright (C) 2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * This integrates into the Linux Kernel rfkill susbystem so that the + * drivers just have to do the bare minimal work, which is providing a + * method to set the software RF-Kill switch and to report changes in + * the software and hardware switch status. + * + * A non-polled generic rfkill device is embedded into the WiMAX + * subsystem's representation of a device. + * + * FIXME: Need polled support? use a timer or add the implementation + * to the stack. + * + * All device drivers have to do is after wimax_dev_init(), call + * wimax_report_rfkill_hw() and wimax_report_rfkill_sw() to update + * initial state and then every time it changes. See wimax.h:struct + * wimax_dev for more information. + * + * ROADMAP + * + * wimax_gnl_doit_rfkill() User space calling wimax_rfkill() + * wimax_rfkill() Kernel calling wimax_rfkill() + * __wimax_rf_toggle_radio() + * + * wimax_rfkill_toggle_radio() RF-Kill subsytem calling + * __wimax_rf_toggle_radio() + * + * __wimax_rf_toggle_radio() + * wimax_dev->op_rfkill_sw_toggle() Driver backend + * __wimax_state_change() + * + * wimax_report_rfkill_sw() Driver reports state change + * __wimax_state_change() + * + * wimax_report_rfkill_hw() Driver reports state change + * __wimax_state_change() + * + * wimax_rfkill_add() Initialize/shutdown rfkill support + * wimax_rfkill_rm() [called by wimax_dev_add/rm()] + */ + +#include +#include +#include +#include +#include +#include +#include "wimax-internal.h" + +#define D_SUBMODULE op_rfkill +#include "debug-levels.h" + +#ifdef CONFIG_RFKILL + + +/** + * wimax_report_rfkill_hw - Reports changes in the hardware RF switch + * + * @wimax_dev: WiMAX device descriptor + * + * @state: New state of the RF Kill switch. %WIMAX_RF_ON radio on, + * %WIMAX_RF_OFF radio off. + * + * When the device detects a change in the state of thehardware RF + * switch, it must call this function to let the WiMAX kernel stack + * know that the state has changed so it can be properly propagated. + * + * The WiMAX stack caches the state (the driver doesn't need to). As + * well, as the change is propagated it will come back as a request to + * change the software state to mirror the hardware state. + * + * If the device doesn't have a hardware kill switch, just report + * it on initialization as always on (%WIMAX_RF_ON, radio on). + */ +void wimax_report_rfkill_hw(struct wimax_dev *wimax_dev, + enum wimax_rf_state state) +{ + int result; + struct device *dev = wimax_dev_to_dev(wimax_dev); + enum wimax_st wimax_state; + enum rfkill_state rfkill_state; + + d_fnstart(3, dev, "(wimax_dev %p state %u)\n", wimax_dev, state); + BUG_ON(state == WIMAX_RF_QUERY); + BUG_ON(state != WIMAX_RF_ON && state != WIMAX_RF_OFF); + + mutex_lock(&wimax_dev->mutex); + result = wimax_dev_is_ready(wimax_dev); + if (result < 0) + goto error_not_ready; + + if (state != wimax_dev->rf_hw) { + wimax_dev->rf_hw = state; + rfkill_state = state == WIMAX_RF_ON ? + RFKILL_STATE_OFF : RFKILL_STATE_ON; + if (wimax_dev->rf_hw == WIMAX_RF_ON + && wimax_dev->rf_sw == WIMAX_RF_ON) + wimax_state = WIMAX_ST_READY; + else + wimax_state = WIMAX_ST_RADIO_OFF; + __wimax_state_change(wimax_dev, wimax_state); + input_report_key(wimax_dev->rfkill_input, KEY_WIMAX, + rfkill_state); + } +error_not_ready: + mutex_unlock(&wimax_dev->mutex); + d_fnend(3, dev, "(wimax_dev %p state %u) = void [%d]\n", + wimax_dev, state, result); +} +EXPORT_SYMBOL_GPL(wimax_report_rfkill_hw); + + +/** + * wimax_report_rfkill_sw - Reports changes in the software RF switch + * + * @wimax_dev: WiMAX device descriptor + * + * @state: New state of the RF kill switch. %WIMAX_RF_ON radio on, + * %WIMAX_RF_OFF radio off. + * + * Reports changes in the software RF switch state to the the WiMAX + * stack. + * + * The main use is during initialization, so the driver can query the + * device for its current software radio kill switch state and feed it + * to the system. + * + * On the side, the device does not change the software state by + * itself. In practice, this can happen, as the device might decide to + * switch (in software) the radio off for different reasons. + */ +void wimax_report_rfkill_sw(struct wimax_dev *wimax_dev, + enum wimax_rf_state state) +{ + int result; + struct device *dev = wimax_dev_to_dev(wimax_dev); + enum wimax_st wimax_state; + + d_fnstart(3, dev, "(wimax_dev %p state %u)\n", wimax_dev, state); + BUG_ON(state == WIMAX_RF_QUERY); + BUG_ON(state != WIMAX_RF_ON && state != WIMAX_RF_OFF); + + mutex_lock(&wimax_dev->mutex); + result = wimax_dev_is_ready(wimax_dev); + if (result < 0) + goto error_not_ready; + + if (state != wimax_dev->rf_sw) { + wimax_dev->rf_sw = state; + if (wimax_dev->rf_hw == WIMAX_RF_ON + && wimax_dev->rf_sw == WIMAX_RF_ON) + wimax_state = WIMAX_ST_READY; + else + wimax_state = WIMAX_ST_RADIO_OFF; + __wimax_state_change(wimax_dev, wimax_state); + } +error_not_ready: + mutex_unlock(&wimax_dev->mutex); + d_fnend(3, dev, "(wimax_dev %p state %u) = void [%d]\n", + wimax_dev, state, result); +} +EXPORT_SYMBOL_GPL(wimax_report_rfkill_sw); + + +/* + * Callback for the RF Kill toggle operation + * + * This function is called by: + * + * - The rfkill subsystem when the RF-Kill key is pressed in the + * hardware and the driver notifies through + * wimax_report_rfkill_hw(). The rfkill subsystem ends up calling back + * here so the software RF Kill switch state is changed to reflect + * the hardware switch state. + * + * - When the user sets the state through sysfs' rfkill/state file + * + * - When the user calls wimax_rfkill(). + * + * This call blocks! + * + * WARNING! When we call rfkill_unregister(), this will be called with + * state 0! + * + * WARNING: wimax_dev must be locked + */ +static +int __wimax_rf_toggle_radio(struct wimax_dev *wimax_dev, + enum wimax_rf_state state) +{ + int result = 0; + struct device *dev = wimax_dev_to_dev(wimax_dev); + enum wimax_st wimax_state; + + might_sleep(); + d_fnstart(3, dev, "(wimax_dev %p state %u)\n", wimax_dev, state); + if (wimax_dev->rf_sw == state) + goto out_no_change; + if (wimax_dev->op_rfkill_sw_toggle != NULL) + result = wimax_dev->op_rfkill_sw_toggle(wimax_dev, state); + else if (state == WIMAX_RF_OFF) /* No op? can't turn off */ + result = -ENXIO; + else /* No op? can turn on */ + result = 0; /* should never happen tho */ + if (result >= 0) { + result = 0; + wimax_dev->rf_sw = state; + wimax_state = state == WIMAX_RF_ON ? + WIMAX_ST_READY : WIMAX_ST_RADIO_OFF; + __wimax_state_change(wimax_dev, wimax_state); + } +out_no_change: + d_fnend(3, dev, "(wimax_dev %p state %u) = %d\n", + wimax_dev, state, result); + return result; +} + + +/* + * Translate from rfkill state to wimax state + * + * NOTE: Special state handling rules here + * + * Just pretend the call didn't happen if we are in a state where + * we know for sure it cannot be handled (WIMAX_ST_DOWN or + * __WIMAX_ST_QUIESCING). rfkill() needs it to register and + * unregister, as it will run this path. + * + * NOTE: This call will block until the operation is completed. + */ +static +int wimax_rfkill_toggle_radio(void *data, enum rfkill_state state) +{ + int result; + struct wimax_dev *wimax_dev = data; + struct device *dev = wimax_dev_to_dev(wimax_dev); + enum wimax_rf_state rf_state; + + d_fnstart(3, dev, "(wimax_dev %p state %u)\n", wimax_dev, state); + switch (state) { + case RFKILL_STATE_ON: + rf_state = WIMAX_RF_OFF; + break; + case RFKILL_STATE_OFF: + rf_state = WIMAX_RF_ON; + break; + default: + BUG(); + } + mutex_lock(&wimax_dev->mutex); + if (wimax_dev->state <= __WIMAX_ST_QUIESCING) + result = 0; /* just pretend it didn't happen */ + else + result = __wimax_rf_toggle_radio(wimax_dev, rf_state); + mutex_unlock(&wimax_dev->mutex); + d_fnend(3, dev, "(wimax_dev %p state %u) = %d\n", + wimax_dev, state, result); + return result; +} + + +/** + * wimax_rfkill - Set the software RF switch state for a WiMAX device + * + * @wimax_dev: WiMAX device descriptor + * + * @state: New RF state. + * + * Returns: + * + * >= 0 toggle state if ok, < 0 errno code on error. The toggle state + * is returned as a bitmap, bit 0 being the hardware RF state, bit 1 + * the software RF state. + * + * 0 means disabled (%WIMAX_RF_ON, radio on), 1 means enabled radio + * off (%WIMAX_RF_OFF). + * + * Description: + * + * Called by the user when he wants to request the WiMAX radio to be + * switched on (%WIMAX_RF_ON) or off (%WIMAX_RF_OFF). With + * %WIMAX_RF_QUERY, just the current state is returned. + * + * NOTE: + * + * This call will block until the operation is complete. + */ +int wimax_rfkill(struct wimax_dev *wimax_dev, enum wimax_rf_state state) +{ + int result; + struct device *dev = wimax_dev_to_dev(wimax_dev); + + d_fnstart(3, dev, "(wimax_dev %p state %u)\n", wimax_dev, state); + mutex_lock(&wimax_dev->mutex); + result = wimax_dev_is_ready(wimax_dev); + if (result < 0) + goto error_not_ready; + switch (state) { + case WIMAX_RF_ON: + case WIMAX_RF_OFF: + result = __wimax_rf_toggle_radio(wimax_dev, state); + if (result < 0) + goto error; + break; + case WIMAX_RF_QUERY: + break; + default: + result = -EINVAL; + goto error; + } + result = wimax_dev->rf_sw << 1 | wimax_dev->rf_hw; +error: +error_not_ready: + mutex_unlock(&wimax_dev->mutex); + d_fnend(3, dev, "(wimax_dev %p state %u) = %d\n", + wimax_dev, state, result); + return result; +} +EXPORT_SYMBOL(wimax_rfkill); + + +/* + * Register a new WiMAX device's RF Kill support + * + * WARNING: wimax_dev->mutex must be unlocked + */ +int wimax_rfkill_add(struct wimax_dev *wimax_dev) +{ + int result; + struct rfkill *rfkill; + struct input_dev *input_dev; + struct device *dev = wimax_dev_to_dev(wimax_dev); + + d_fnstart(3, dev, "(wimax_dev %p)\n", wimax_dev); + /* Initialize RF Kill */ + result = -ENOMEM; + rfkill = rfkill_allocate(dev, RFKILL_TYPE_WIMAX); + if (rfkill == NULL) + goto error_rfkill_allocate; + wimax_dev->rfkill = rfkill; + + rfkill->name = wimax_dev->name; + rfkill->state = RFKILL_STATE_OFF; + rfkill->data = wimax_dev; + rfkill->toggle_radio = wimax_rfkill_toggle_radio; + rfkill->user_claim_unsupported = 1; + + /* Initialize the input device for the hw key */ + input_dev = input_allocate_device(); + if (input_dev == NULL) + goto error_input_allocate; + wimax_dev->rfkill_input = input_dev; + d_printf(1, dev, "rfkill %p input %p\n", rfkill, input_dev); + + input_dev->name = wimax_dev->name; + /* FIXME: get a real device bus ID and stuff? do we care? */ + input_dev->id.bustype = BUS_HOST; + input_dev->id.vendor = 0xffff; + input_dev->evbit[0] = BIT(EV_KEY); + set_bit(KEY_WIMAX, input_dev->keybit); + + /* Register both */ + result = input_register_device(wimax_dev->rfkill_input); + if (result < 0) + goto error_input_register; + result = rfkill_register(wimax_dev->rfkill); + if (result < 0) + goto error_rfkill_register; + + /* If there is no SW toggle op, SW RFKill is always on */ + if (wimax_dev->op_rfkill_sw_toggle == NULL) + wimax_dev->rf_sw = WIMAX_RF_ON; + + d_fnend(3, dev, "(wimax_dev %p) = 0\n", wimax_dev); + return 0; + + /* if rfkill_register() suceeds, can't use rfkill_free() any + * more, only rfkill_unregister() [it owns the refcount]; with + * the input device we have the same issue--hence the if. */ +error_rfkill_register: + input_unregister_device(wimax_dev->rfkill_input); + wimax_dev->rfkill_input = NULL; +error_input_register: + if (wimax_dev->rfkill_input) + input_free_device(wimax_dev->rfkill_input); +error_input_allocate: + rfkill_free(wimax_dev->rfkill); +error_rfkill_allocate: + d_fnend(3, dev, "(wimax_dev %p) = %d\n", wimax_dev, result); + return result; +} + + +/* + * Deregister a WiMAX device's RF Kill support + * + * Ick, we can't call rfkill_free() after rfkill_unregister()...oh + * well. + * + * WARNING: wimax_dev->mutex must be unlocked + */ +void wimax_rfkill_rm(struct wimax_dev *wimax_dev) +{ + struct device *dev = wimax_dev_to_dev(wimax_dev); + d_fnstart(3, dev, "(wimax_dev %p)\n", wimax_dev); + rfkill_unregister(wimax_dev->rfkill); /* frees */ + input_unregister_device(wimax_dev->rfkill_input); + d_fnend(3, dev, "(wimax_dev %p)\n", wimax_dev); +} + + +#else /* #ifdef CONFIG_RFKILL */ + +void wimax_report_rfkill_hw(struct wimax_dev *wimax_dev, + enum wimax_rf_state state) +{ +} +EXPORT_SYMBOL_GPL(wimax_report_rfkill_hw); + +void wimax_report_rfkill_sw(struct wimax_dev *wimax_dev, + enum wimax_rf_state state) +{ +} +EXPORT_SYMBOL_GPL(wimax_report_rfkill_sw); + +int wimax_rfkill(struct wimax_dev *wimax_dev, + enum wimax_rf_state state) +{ + return WIMAX_RF_ON << 1 | WIMAX_RF_ON; +} +EXPORT_SYMBOL_GPL(wimax_rfkill); + +int wimax_rfkill_add(struct wimax_dev *wimax_dev) +{ + return 0; +} + +void wimax_rfkill_rm(struct wimax_dev *wimax_dev) +{ +} + +#endif /* #ifdef CONFIG_RFKILL */ + + +/* + * Exporting to user space over generic netlink + * + * Parse the rfkill command from user space, return a combination + * value that describe the states of the different toggles. + * + * Only one attribute: the new state requested (on, off or no change, + * just query). + */ + +static const +struct nla_policy wimax_gnl_rfkill_policy[WIMAX_GNL_ATTR_MAX + 1] = { + [WIMAX_GNL_RFKILL_IFIDX] = { + .type = NLA_U32, + }, + [WIMAX_GNL_RFKILL_STATE] = { + .type = NLA_U32 /* enum wimax_rf_state */ + }, +}; + + +static +int wimax_gnl_doit_rfkill(struct sk_buff *skb, struct genl_info *info) +{ + int result, ifindex; + struct wimax_dev *wimax_dev; + struct device *dev; + enum wimax_rf_state new_state; + + d_fnstart(3, NULL, "(skb %p info %p)\n", skb, info); + result = -ENODEV; + if (info->attrs[WIMAX_GNL_RFKILL_IFIDX] == NULL) { + printk(KERN_ERR "WIMAX_GNL_OP_RFKILL: can't find IFIDX " + "attribute\n"); + goto error_no_wimax_dev; + } + ifindex = nla_get_u32(info->attrs[WIMAX_GNL_RFKILL_IFIDX]); + wimax_dev = wimax_dev_get_by_genl_info(info, ifindex); + if (wimax_dev == NULL) + goto error_no_wimax_dev; + dev = wimax_dev_to_dev(wimax_dev); + result = -EINVAL; + if (info->attrs[WIMAX_GNL_RFKILL_STATE] == NULL) { + dev_err(dev, "WIMAX_GNL_RFKILL: can't find RFKILL_STATE " + "attribute\n"); + goto error_no_pid; + } + new_state = nla_get_u32(info->attrs[WIMAX_GNL_RFKILL_STATE]); + + /* Execute the operation and send the result back to user space */ + result = wimax_rfkill(wimax_dev, new_state); +error_no_pid: + dev_put(wimax_dev->net_dev); +error_no_wimax_dev: + d_fnend(3, NULL, "(skb %p info %p) = %d\n", skb, info, result); + return result; +} + + +struct genl_ops wimax_gnl_rfkill = { + .cmd = WIMAX_GNL_OP_RFKILL, + .flags = GENL_ADMIN_PERM, + .policy = wimax_gnl_rfkill_policy, + .doit = wimax_gnl_doit_rfkill, + .dumpit = NULL, +}; + From 617209ccf73b571953cf4c76519c65a5e52d15fd Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:40 -0800 Subject: [PATCH 103/123] wimax: debugfs controls Expose knobs to control the stack's debug output. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- net/wimax/debugfs.c | 90 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 90 insertions(+) create mode 100644 net/wimax/debugfs.c diff --git a/net/wimax/debugfs.c b/net/wimax/debugfs.c new file mode 100644 index 00000000000..87cf4430079 --- /dev/null +++ b/net/wimax/debugfs.c @@ -0,0 +1,90 @@ +/* + * Linux WiMAX + * Debugfs support + * + * + * Copyright (C) 2005-2006 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + */ +#include +#include +#include "wimax-internal.h" + +#define D_SUBMODULE debugfs +#include "debug-levels.h" + + +/* Debug framework control of debug levels */ +struct d_level D_LEVEL[] = { + D_SUBMODULE_DEFINE(debugfs), + D_SUBMODULE_DEFINE(id_table), + D_SUBMODULE_DEFINE(op_msg), + D_SUBMODULE_DEFINE(op_reset), + D_SUBMODULE_DEFINE(op_rfkill), + D_SUBMODULE_DEFINE(stack), +}; +size_t D_LEVEL_SIZE = ARRAY_SIZE(D_LEVEL); + +#define __debugfs_register(prefix, name, parent) \ +do { \ + result = d_level_register_debugfs(prefix, name, parent); \ + if (result < 0) \ + goto error; \ +} while (0) + + +int wimax_debugfs_add(struct wimax_dev *wimax_dev) +{ + int result; + struct net_device *net_dev = wimax_dev->net_dev; + struct device *dev = net_dev->dev.parent; + struct dentry *dentry; + char buf[128]; + + snprintf(buf, sizeof(buf), "wimax:%s", net_dev->name); + dentry = debugfs_create_dir(buf, NULL); + result = PTR_ERR(dentry); + if (IS_ERR(dentry)) { + if (result == -ENODEV) + result = 0; /* No debugfs support */ + else + dev_err(dev, "Can't create debugfs dentry: %d\n", + result); + goto out; + } + wimax_dev->debugfs_dentry = dentry; + __debugfs_register("wimax_dl_", debugfs, dentry); + __debugfs_register("wimax_dl_", id_table, dentry); + __debugfs_register("wimax_dl_", op_msg, dentry); + __debugfs_register("wimax_dl_", op_reset, dentry); + __debugfs_register("wimax_dl_", op_rfkill, dentry); + __debugfs_register("wimax_dl_", stack, dentry); + result = 0; +out: + return result; + +error: + debugfs_remove_recursive(wimax_dev->debugfs_dentry); + return result; +} + +void wimax_debugfs_rm(struct wimax_dev *wimax_dev) +{ + debugfs_remove_recursive(wimax_dev->debugfs_dentry); +} + + From b0c83ae1de01880075955c7224e751440688ec74 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Tue, 23 Dec 2008 16:18:24 -0800 Subject: [PATCH 104/123] wimax: Makefile, Kconfig and docbook linkage for the stack This patch provides Makefile and KConfig for the WiMAX stack, integrating them into the networking stack's Makefile, Kconfig and doc-book templates. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- Documentation/DocBook/networking.tmpl | 8 ++++++ net/Kconfig | 2 ++ net/Makefile | 1 + net/wimax/Kconfig | 38 +++++++++++++++++++++++++++ net/wimax/Makefile | 13 +++++++++ 5 files changed, 62 insertions(+) create mode 100644 net/wimax/Kconfig create mode 100644 net/wimax/Makefile diff --git a/Documentation/DocBook/networking.tmpl b/Documentation/DocBook/networking.tmpl index 627707a3cb9..59ad69a9d77 100644 --- a/Documentation/DocBook/networking.tmpl +++ b/Documentation/DocBook/networking.tmpl @@ -74,6 +74,14 @@ !Enet/sunrpc/rpcb_clnt.c !Enet/sunrpc/clnt.c + WiMAX +!Enet/wimax/op-msg.c +!Enet/wimax/op-reset.c +!Enet/wimax/op-rfkill.c +!Enet/wimax/stack.c +!Iinclude/net/wimax.h +!Iinclude/linux/wimax.h + diff --git a/net/Kconfig b/net/Kconfig index 6ec2cce7c16..bf2776018f7 100644 --- a/net/Kconfig +++ b/net/Kconfig @@ -254,6 +254,8 @@ source "net/mac80211/Kconfig" endif # WIRELESS +source "net/wimax/Kconfig" + source "net/rfkill/Kconfig" source "net/9p/Kconfig" diff --git a/net/Makefile b/net/Makefile index ba4460432b7..0fcce89d716 100644 --- a/net/Makefile +++ b/net/Makefile @@ -63,3 +63,4 @@ endif ifeq ($(CONFIG_NET),y) obj-$(CONFIG_SYSCTL) += sysctl_net.o endif +obj-$(CONFIG_WIMAX) += wimax/ diff --git a/net/wimax/Kconfig b/net/wimax/Kconfig new file mode 100644 index 00000000000..0bdbb692820 --- /dev/null +++ b/net/wimax/Kconfig @@ -0,0 +1,38 @@ +# +# WiMAX LAN device configuration +# + +menuconfig WIMAX + tristate "WiMAX Wireless Broadband support" + help + + Select to configure support for devices that provide + wireless broadband connectivity using the WiMAX protocol + (IEEE 802.16). + + Please note that most of these devices require signing up + for a service plan with a provider. + + The different WiMAX drivers can be enabled in the menu entry + + Device Drivers > Network device support > WiMAX Wireless + Broadband devices + + If unsure, it is safe to select M (module). + +config WIMAX_DEBUG_LEVEL + int "WiMAX debug level" + depends on WIMAX + default 8 + help + + Select the maximum debug verbosity level to be compiled into + the WiMAX stack code. + + By default, debug messages are disabled at runtime and can + be selectively enabled for different parts of the code using + the sysfs debug-levels file. + + If set at zero, this will compile out all the debug code. + + It is recommended that it is left at 8. diff --git a/net/wimax/Makefile b/net/wimax/Makefile new file mode 100644 index 00000000000..5b80b941c2c --- /dev/null +++ b/net/wimax/Makefile @@ -0,0 +1,13 @@ + +obj-$(CONFIG_WIMAX) += wimax.o + +wimax-y := \ + id-table.o \ + op-msg.o \ + op-reset.o \ + op-rfkill.o \ + stack.o + +wimax-$(CONFIG_DEBUG_FS) += debugfs.o + + From 3e91029ae049852c153da3fc355ab255ea7e2e0a Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:42 -0800 Subject: [PATCH 105/123] i2400m: documentation and instructions for usage The driver for the i2400m is a stacked driver. There is a core driver, the bus-generic driver that has no knowledge or dependencies on how the device is connected to the system; it only knows how to speak the device protocol. Then there are the bus-specific drivers (for USB and SDIO) that provide backends for the generic driver to communicate with the device. The bus generic driver connects to the network and WiMAX stacks on the top side, and on the bottom to the bus-specific drivers. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- Documentation/wimax/README.i2400m | 260 ++++++++++++++++++++++++++++++ 1 file changed, 260 insertions(+) create mode 100644 Documentation/wimax/README.i2400m diff --git a/Documentation/wimax/README.i2400m b/Documentation/wimax/README.i2400m new file mode 100644 index 00000000000..7dffd8919cb --- /dev/null +++ b/Documentation/wimax/README.i2400m @@ -0,0 +1,260 @@ + + Driver for the Intel Wireless Wimax Connection 2400m + + (C) 2008 Intel Corporation < linux-wimax@intel.com > + + This provides a driver for the Intel Wireless WiMAX Connection 2400m + and a basic Linux kernel WiMAX stack. + +1. Requirements + + * Linux installation with Linux kernel 2.6.22 or newer (if building + from a separate tree) + * Intel i2400m Echo Peak or Baxter Peak; this includes the Intel + Wireless WiMAX/WiFi Link 5x50 series. + * build tools: + + Linux kernel development package for the target kernel; to + build against your currently running kernel, you need to have + the kernel development package corresponding to the running + image installed (usually if your kernel is named + linux-VERSION, the development package is called + linux-dev-VERSION or linux-headers-VERSION). + + GNU C Compiler, make + +2. Compilation and installation + +2.1. Compilation of the drivers included in the kernel + + Configure the kernel; to enable the WiMAX drivers select Drivers > + Networking Drivers > WiMAX device support. Enable all of them as + modules (easier). + + If USB or SDIO are not enabled in the kernel configuration, the options + to build the i2400m USB or SDIO drivers will not show. Enable said + subsystems and go back to the WiMAX menu to enable the drivers. + + Compile and install your kernel as usual. + +2.2. Compilation of the drivers distributed as an standalone module + + To compile + +$ cd source/directory +$ make + + Once built you can load and unload using the provided load.sh script; + load.sh will load the modules, load.sh u will unload them. + + To install in the default kernel directories (and enable auto loading + when the device is plugged): + +$ make install +$ depmod -a + + If your kernel development files are located in a non standard + directory or if you want to build for a kernel that is not the + currently running one, set KDIR to the right location: + +$ make KDIR=/path/to/kernel/dev/tree + + For more information, please contact linux-wimax@intel.com. + +3. Installing the firmware + + The firmware can be obtained from http://linuxwimax.org or might have + been supplied with your hardware. + + It has to be installed in the target system: + * +$ cp FIRMWAREFILE.sbcf /lib/firmware/i2400m-fw-BUSTYPE-1.3.sbcf + + * NOTE: if your firmware came in an .rpm or .deb file, just install + it as normal, with the rpm (rpm -i FIRMWARE.rpm) or dpkg + (dpkg -i FIRMWARE.deb) commands. No further action is needed. + * BUSTYPE will be usb or sdio, depending on the hardware you have. + Each hardware type comes with its own firmware and will not work + with other types. + +4. Design + + This package contains two major parts: a WiMAX kernel stack and a + driver for the Intel i2400m. + + The WiMAX stack is designed to provide for common WiMAX control + services to current and future WiMAX devices from any vendor; please + see README.wimax for details. + + The i2400m kernel driver is broken up in two main parts: the bus + generic driver and the bus-specific drivers. The bus generic driver + forms the drivercore and contain no knowledge of the actual method we + use to connect to the device. The bus specific drivers are just the + glue to connect the bus-generic driver and the device. Currently only + USB and SDIO are supported. See drivers/net/wimax/i2400m/i2400m.h for + more information. + + The bus generic driver is logically broken up in two parts: OS-glue and + hardware-glue. The OS-glue interfaces with Linux. The hardware-glue + interfaces with the device on using an interface provided by the + bus-specific driver. The reason for this breakup is to be able to + easily reuse the hardware-glue to write drivers for other OSes; note + the hardware glue part is written as a native Linux driver; no + abstraction layers are used, so to port to another OS, the Linux kernel + API calls should be replaced with the target OS's. + +5. Usage + + To load the driver, follow the instructions in the install section; + once the driver is loaded, plug in the device (unless it is permanently + plugged in). The driver will enumerate the device, upload the firmware + and output messages in the kernel log (dmesg, /var/log/messages or + /var/log/kern.log) such as: + +... +i2400m_usb 5-4:1.0: firmware interface version 8.0.0 +i2400m_usb 5-4:1.0: WiMAX interface wmx0 (00:1d:e1:01:94:2c) ready + + At this point the device is ready to work. + + Current versions require the Intel WiMAX Network Service in userspace + to make things work. See the network service's README for instructions + on how to scan, connect and disconnect. + +5.1. Module parameters + + Module parameters can be set at kernel or module load time or by + echoing values: + +$ echo VALUE > /sys/module/MODULENAME/parameters/PARAMETERNAME + + To make changes permanent, for example, for the i2400m module, you can + also create a file named /etc/modprobe.d/i2400m containing: + +options i2400m idle_mode_disabled=1 + + To find which parameters are supported by a module, run: + +$ modinfo path/to/module.ko + + During kernel bootup (if the driver is linked in the kernel), specify + the following to the kernel command line: + +i2400m.PARAMETER=VALUE + +5.1.1. i2400m: idle_mode_disabled + + The i2400m module supports a parameter to disable idle mode. This + parameter, once set, will take effect only when the device is + reinitialized by the driver (eg: following a reset or a reconnect). + +5.2. Debug operations: debugfs entries + + The driver will register debugfs entries that allow the user to tweak + debug settings. There are three main container directories where + entries are placed, which correspond to the three blocks a i2400m WiMAX + driver has: + * /sys/kernel/debug/wimax:DEVNAME/ for the generic WiMAX stack + controls + * /sys/kernel/debug/wimax:DEVNAME/i2400m for the i2400m generic + driver controls + * /sys/kernel/debug/wimax:DEVNAME/i2400m-usb (or -sdio) for the + bus-specific i2400m-usb or i2400m-sdio controls). + + Of course, if debugfs is mounted in a directory other than + /sys/kernel/debug, those paths will change. + +5.2.1. Increasing debug output + + The files named *dl_* indicate knobs for controlling the debug output + of different submodules: + * +# find /sys/kernel/debug/wimax\:wmx0 -name \*dl_\* +/sys/kernel/debug/wimax:wmx0/i2400m-usb/dl_tx +/sys/kernel/debug/wimax:wmx0/i2400m-usb/dl_rx +/sys/kernel/debug/wimax:wmx0/i2400m-usb/dl_notif +/sys/kernel/debug/wimax:wmx0/i2400m-usb/dl_fw +/sys/kernel/debug/wimax:wmx0/i2400m-usb/dl_usb +/sys/kernel/debug/wimax:wmx0/i2400m/dl_tx +/sys/kernel/debug/wimax:wmx0/i2400m/dl_rx +/sys/kernel/debug/wimax:wmx0/i2400m/dl_rfkill +/sys/kernel/debug/wimax:wmx0/i2400m/dl_netdev +/sys/kernel/debug/wimax:wmx0/i2400m/dl_fw +/sys/kernel/debug/wimax:wmx0/i2400m/dl_debugfs +/sys/kernel/debug/wimax:wmx0/i2400m/dl_driver +/sys/kernel/debug/wimax:wmx0/i2400m/dl_control +/sys/kernel/debug/wimax:wmx0/wimax_dl_stack +/sys/kernel/debug/wimax:wmx0/wimax_dl_op_rfkill +/sys/kernel/debug/wimax:wmx0/wimax_dl_op_reset +/sys/kernel/debug/wimax:wmx0/wimax_dl_op_msg +/sys/kernel/debug/wimax:wmx0/wimax_dl_id_table +/sys/kernel/debug/wimax:wmx0/wimax_dl_debugfs + + By reading the file you can obtain the current value of said debug + level; by writing to it, you can set it. + + To increase the debug level of, for example, the i2400m's generic TX + engine, just write: + +$ echo 3 > /sys/kernel/debug/wimax:wmx0/i2400m/dl_tx + + Increasing numbers yield increasing debug information; for details of + what is printed and the available levels, check the source. The code + uses 0 for disabled and increasing values until 8. + +5.2.2. RX and TX statistics + + The i2400m/rx_stats and i2400m/tx_stats provide statistics about the + data reception/delivery from the device: + +$ cat /sys/kernel/debug/wimax:wmx0/i2400m/rx_stats +45 1 3 34 3104 48 480 + + The numbers reported are + * packets/RX-buffer: total, min, max + * RX-buffers: total RX buffers received, accumulated RX buffer size + in bytes, min size received, max size received + + Thus, to find the average buffer size received, divide accumulated + RX-buffer / total RX-buffers. + + To clear the statistics back to 0, write anything to the rx_stats file: + +$ echo 1 > /sys/kernel/debug/wimax:wmx0/i2400m_rx_stats + + Likewise for TX. + + Note the packets this debug file refers to are not network packet, but + packets in the sense of the device-specific protocol for communication + to the host. See drivers/net/wimax/i2400m/tx.c. + +5.2.3. Tracing messages received from user space + + To echo messages received from user space into the trace pipe that the + i2400m driver creates, set the debug file i2400m/trace_msg_from_user to + 1: + * +$ echo 1 > /sys/kernel/debug/wimax:wmx0/i2400m/trace_msg_from_user + +5.2.4. Performing a device reset + + By writing a 0, a 1 or a 2 to the file + /sys/kernel/debug/wimax:wmx0/reset, the driver performs a warm (without + disconnecting from the bus), cold (disconnecting from the bus) or bus + (bus specific) reset on the device. + +5.2.5. Asking the device to enter power saving mode + + By writing any value to the /sys/kernel/debug/wimax:wmx0 file, the + device will attempt to enter power saving mode. + +6. Troubleshooting + +6.1. Driver complains about 'i2400m-fw-usb-1.2.sbcf: request failed' + + If upon connecting the device, the following is output in the kernel + log: + +i2400m_usb 5-4:1.0: fw i2400m-fw-usb-1.3.sbcf: request failed: -2 + + This means that the driver cannot locate the firmware file named + /lib/firmware/i2400m-fw-usb-1.2.sbcf. Check that the file is present in + the right location. From ea24652d253eabfb83e955e55ce032228d9d99b9 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:43 -0800 Subject: [PATCH 106/123] i2400m: host/device procotol and core driver definitions The wimax/i2400m.h defines the structures and constants for the host-device protocols: - boot / firmware upload protocol - general data transport protocol - control protocol It is done in such a way that can also be used verbatim by user space. drivers/net/wimax/i2400m.h defines all the APIs used by the core, bus-generic driver (i2400m) and the bus specific drivers (i2400m-BUSNAME). It also gives a roadmap to the driver implementation. debug-levels.h adds the core driver's debug settings. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/debug-levels.h | 45 ++ drivers/net/wimax/i2400m/i2400m.h | 755 ++++++++++++++++++++++++ include/linux/wimax/i2400m.h | 512 ++++++++++++++++ 3 files changed, 1312 insertions(+) create mode 100644 drivers/net/wimax/i2400m/debug-levels.h create mode 100644 drivers/net/wimax/i2400m/i2400m.h create mode 100644 include/linux/wimax/i2400m.h diff --git a/drivers/net/wimax/i2400m/debug-levels.h b/drivers/net/wimax/i2400m/debug-levels.h new file mode 100644 index 00000000000..3183baa16a5 --- /dev/null +++ b/drivers/net/wimax/i2400m/debug-levels.h @@ -0,0 +1,45 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Debug levels control file for the i2400m module + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + */ +#ifndef __debug_levels__h__ +#define __debug_levels__h__ + +/* Maximum compile and run time debug level for all submodules */ +#define D_MODULENAME i2400m +#define D_MASTER CONFIG_WIMAX_I2400M_DEBUG_LEVEL + +#include + +/* List of all the enabled modules */ +enum d_module { + D_SUBMODULE_DECLARE(control), + D_SUBMODULE_DECLARE(driver), + D_SUBMODULE_DECLARE(debugfs), + D_SUBMODULE_DECLARE(fw), + D_SUBMODULE_DECLARE(netdev), + D_SUBMODULE_DECLARE(rfkill), + D_SUBMODULE_DECLARE(rx), + D_SUBMODULE_DECLARE(tx), +}; + + +#endif /* #ifndef __debug_levels__h__ */ diff --git a/drivers/net/wimax/i2400m/i2400m.h b/drivers/net/wimax/i2400m/i2400m.h new file mode 100644 index 00000000000..067c871cc22 --- /dev/null +++ b/drivers/net/wimax/i2400m/i2400m.h @@ -0,0 +1,755 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Declarations for bus-generic internal APIs + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Inaky Perez-Gonzalez + * Yanir Lubetkin + * - Initial implementation + * + * + * GENERAL DRIVER ARCHITECTURE + * + * The i2400m driver is split in the following two major parts: + * + * - bus specific driver + * - bus generic driver (this part) + * + * The bus specific driver sets up stuff specific to the bus the + * device is connected to (USB, SDIO, PCI, tam-tam...non-authoritative + * nor binding list) which is basically the device-model management + * (probe/disconnect, etc), moving data from device to kernel and + * back, doing the power saving details and reseting the device. + * + * For details on each bus-specific driver, see it's include file, + * i2400m-BUSNAME.h + * + * The bus-generic functionality break up is: + * + * - Firmware upload: fw.c - takes care of uploading firmware to the + * device. bus-specific driver just needs to provides a way to + * execute boot-mode commands and to reset the device. + * + * - RX handling: rx.c - receives data from the bus-specific code and + * feeds it to the network or WiMAX stack or uses it to modify + * the driver state. bus-specific driver only has to receive + * frames and pass them to this module. + * + * - TX handling: tx.c - manages the TX FIFO queue and provides means + * for the bus-specific TX code to pull data from the FIFO + * queue. bus-specific code just pulls frames from this module + * to sends them to the device. + * + * - netdev glue: netdev.c - interface with Linux networking + * stack. Pass around data frames, and configure when the + * device is up and running or shutdown (through ifconfig up / + * down). Bus-generic only. + * + * - control ops: control.c - implements various commmands for + * controlling the device. bus-generic only. + * + * - device model glue: driver.c - implements helpers for the + * device-model glue done by the bus-specific layer + * (setup/release the driver resources), turning the device on + * and off, handling the device reboots/resets and a few simple + * WiMAX stack ops. + * + * Code is also broken up in linux-glue / device-glue. + * + * Linux glue contains functions that deal mostly with gluing with the + * rest of the Linux kernel. + * + * Device-glue are functions that deal mostly with the way the device + * does things and talk the device's language. + * + * device-glue code is licensed BSD so other open source OSes can take + * it to implement their drivers. + * + * + * APIs AND HEADER FILES + * + * This bus generic code exports three APIs: + * + * - HDI (host-device interface) definitions common to all busses + * (include/linux/wimax/i2400m.h); these can be also used by user + * space code. + * - internal API for the bus-generic code + * - external API for the bus-specific drivers + * + * + * LIFE CYCLE: + * + * When the bus-specific driver probes, it allocates a network device + * with enough space for it's data structue, that must contain a + * &struct i2400m at the top. + * + * On probe, it needs to fill the i2400m members marked as [fill], as + * well as i2400m->wimax_dev.net_dev and call i2400m_setup(). The + * i2400m driver will only register with the WiMAX and network stacks; + * the only access done to the device is to read the MAC address so we + * can register a network device. This calls i2400m_dev_start() to + * load firmware, setup communication with the device and configure it + * for operation. + * + * At this point, control and data communications are possible. + * + * On disconnect/driver unload, the bus-specific disconnect function + * calls i2400m_release() to undo i2400m_setup(). i2400m_dev_stop() + * shuts the firmware down and releases resources uses to communicate + * with the device. + * + * While the device is up, it might reset. The bus-specific driver has + * to catch that situation and call i2400m_dev_reset_handle() to deal + * with it (reset the internal driver structures and go back to square + * one). + */ + +#ifndef __I2400M_H__ +#define __I2400M_H__ + +#include +#include +#include +#include +#include +#include +#include +#include + +/* Misc constants */ +enum { + /* Firmware uploading */ + I2400M_BOOT_RETRIES = 3, + /* Size of the Boot Mode Command buffer */ + I2400M_BM_CMD_BUF_SIZE = 16 * 1024, + I2400M_BM_ACK_BUF_SIZE = 256, +}; + + +/* Firmware version we request when pulling the fw image file */ +#define I2400M_FW_VERSION "1.3" + + +/** + * i2400m_reset_type - methods to reset a device + * + * @I2400M_RT_WARM: Reset without device disconnection, device handles + * are kept valid but state is back to power on, with firmware + * re-uploaded. + * @I2400M_RT_COLD: Tell the device to disconnect itself from the bus + * and reconnect. Renders all device handles invalid. + * @I2400M_RT_BUS: Tells the bus to reset the device; last measure + * used when both types above don't work. + */ +enum i2400m_reset_type { + I2400M_RT_WARM, /* first measure */ + I2400M_RT_COLD, /* second measure */ + I2400M_RT_BUS, /* call in artillery */ +}; + +struct i2400m_reset_ctx; + +/** + * struct i2400m - descriptor for an Intel 2400m + * + * Members marked with [fill] must be filled out/initialized before + * calling i2400m_setup(). + * + * @bus_tx_block_size: [fill] SDIO imposes a 256 block size, USB 16, + * so we have a tx_blk_size variable that the bus layer sets to + * tell the engine how much of that we need. + * + * @bus_pl_size_max: [fill] Maximum payload size. + * + * @bus_dev_start: [fill] Function called by the bus-generic code + * [i2400m_dev_start()] to setup the bus-specific communications + * to the the device. See LIFE CYCLE above. + * + * NOTE: Doesn't need to upload the firmware, as that is taken + * care of by the bus-generic code. + * + * @bus_dev_stop: [fill] Function called by the bus-generic code + * [i2400m_dev_stop()] to shutdown the bus-specific communications + * to the the device. See LIFE CYCLE above. + * + * This function does not need to reset the device, just tear down + * all the host resources created to handle communication with + * the device. + * + * @bus_tx_kick: [fill] Function called by the bus-generic code to let + * the bus-specific code know that there is data available in the + * TX FIFO for transmission to the device. + * + * This function cannot sleep. + * + * @bus_reset: [fill] Function called by the bus-generic code to reset + * the device in in various ways. Doesn't need to wait for the + * reset to finish. + * + * If warm or cold reset fail, this function is expected to do a + * bus-specific reset (eg: USB reset) to get the device to a + * working state (even if it implies device disconecction). + * + * Note the warm reset is used by the firmware uploader to + * reinitialize the device. + * + * IMPORTANT: this is called very early in the device setup + * process, so it cannot rely on common infrastructure being laid + * out. + * + * @bus_bm_cmd_send: [fill] Function called to send a boot-mode + * command. Flags are defined in 'enum i2400m_bm_cmd_flags'. This + * is synchronous and has to return 0 if ok or < 0 errno code in + * any error condition. + * + * @bus_bm_wait_for_ack: [fill] Function called to wait for a + * boot-mode notification (that can be a response to a previously + * issued command or an asynchronous one). Will read until all the + * indicated size is read or timeout. Reading more or less data + * than asked for is an error condition. Return 0 if ok, < 0 errno + * code on error. + * + * The caller to this function will check if the response is a + * barker that indicates the device going into reset mode. + * + * @bus_fw_name: [fill] name of the firmware image (in most cases, + * they are all the same for a single release, except that they + * have the type of the bus embedded in the name (eg: + * i2400m-fw-X-VERSION.sbcf, where X is the bus name). + * + * @bus_bm_mac_addr_impaired: [fill] Set to true if the device's MAC + * address provided in boot mode is kind of broken and needs to + * be re-read later on. + * + * + * @wimax_dev: WiMAX generic device for linkage into the kernel WiMAX + * stack. Due to the way a net_device is allocated, we need to + * force this to be the first field so that we can get from + * netdev_priv() the right pointer. + * + * @state: device's state (as reported by it) + * + * @state_wq: waitqueue that is woken up whenever the state changes + * + * @tx_lock: spinlock to protect TX members + * + * @tx_buf: FIFO buffer for TX; we queue data here + * + * @tx_in: FIFO index for incoming data. Note this doesn't wrap around + * and it is always greater than @tx_out. + * + * @tx_out: FIFO index for outgoing data + * + * @tx_msg: current TX message that is active in the FIFO for + * appending payloads. + * + * @tx_sequence: current sequence number for TX messages from the + * device to the host. + * + * @tx_msg_size: size of the current message being transmitted by the + * bus-specific code. + * + * @tx_pl_num: total number of payloads sent + * + * @tx_pl_max: maximum number of payloads sent in a TX message + * + * @tx_pl_min: minimum number of payloads sent in a TX message + * + * @tx_num: number of TX messages sent + * + * @tx_size_acc: number of bytes in all TX messages sent + * (this is different to net_dev's statistics as it also counts + * control messages). + * + * @tx_size_min: smallest TX message sent. + * + * @tx_size_max: biggest TX message sent. + * + * @rx_lock: spinlock to protect RX members + * + * @rx_pl_num: total number of payloads received + * + * @rx_pl_max: maximum number of payloads received in a RX message + * + * @rx_pl_min: minimum number of payloads received in a RX message + * + * @rx_num: number of RX messages received + * + * @rx_size_acc: number of bytes in all RX messages received + * (this is different to net_dev's statistics as it also counts + * control messages). + * + * @rx_size_min: smallest RX message received. + * + * @rx_size_max: buggest RX message received. + * + * @init_mutex: Mutex used for serializing the device bringup + * sequence; this way if the device reboots in the middle, we + * don't try to do a bringup again while we are tearing down the + * one that failed. + * + * Can't reuse @msg_mutex because from within the bringup sequence + * we need to send messages to the device and thus use @msg_mutex. + * + * @msg_mutex: mutex used to send control commands to the device (we + * only allow one at a time, per host-device interface design). + * + * @msg_completion: used to wait for an ack to a control command sent + * to the device. + * + * @ack_skb: used to store the actual ack to a control command if the + * reception of the command was successful. Otherwise, a ERR_PTR() + * errno code that indicates what failed with the ack reception. + * + * Only valid after @msg_completion is woken up. Only updateable + * if @msg_completion is armed. Only touched by + * i2400m_msg_to_dev(). + * + * Protected by @rx_lock. In theory the command execution flow is + * sequential, but in case the device sends an out-of-phase or + * very delayed response, we need to avoid it trampling current + * execution. + * + * @bm_cmd_buf: boot mode command buffer for composing firmware upload + * commands. + * + * USB can't r/w to stack, vmalloc, etc...as well, we end up + * having to alloc/free a lot to compose commands, so we use these + * for stagging and not having to realloc all the time. + * + * This assumes the code always runs serialized. Only one thread + * can call i2400m_bm_cmd() at the same time. + * + * @bm_ack_buf: boot mode acknoledge buffer for staging reception of + * responses to commands. + * + * See @bm_cmd_buf. + * + * @work_queue: work queue for processing device reports. This + * workqueue cannot be used for processing TX or RX to the device, + * as from it we'll process device reports, which might require + * further communication with the device. + * + * @debugfs_dentry: hookup for debugfs files. + * These have to be in a separate directory, a child of + * (wimax_dev->debugfs_dentry) so they can be removed when the + * module unloads, as we don't keep each dentry. + */ +struct i2400m { + struct wimax_dev wimax_dev; /* FIRST! See doc */ + + unsigned updown:1; /* Network device is up or down */ + unsigned boot_mode:1; /* is the device in boot mode? */ + unsigned sboot:1; /* signed or unsigned fw boot */ + unsigned ready:1; /* all probing steps done */ + u8 trace_msg_from_user; /* echo rx msgs to 'trace' pipe */ + /* typed u8 so debugfs/u8 can tweak */ + enum i2400m_system_state state; + wait_queue_head_t state_wq; /* Woken up when on state updates */ + + size_t bus_tx_block_size; + size_t bus_pl_size_max; + int (*bus_dev_start)(struct i2400m *); + void (*bus_dev_stop)(struct i2400m *); + void (*bus_tx_kick)(struct i2400m *); + int (*bus_reset)(struct i2400m *, enum i2400m_reset_type); + ssize_t (*bus_bm_cmd_send)(struct i2400m *, + const struct i2400m_bootrom_header *, + size_t, int flags); + ssize_t (*bus_bm_wait_for_ack)(struct i2400m *, + struct i2400m_bootrom_header *, size_t); + const char *bus_fw_name; + unsigned bus_bm_mac_addr_impaired:1; + + spinlock_t tx_lock; /* protect TX state */ + void *tx_buf; + size_t tx_in, tx_out; + struct i2400m_msg_hdr *tx_msg; + size_t tx_sequence, tx_msg_size; + /* TX stats */ + unsigned tx_pl_num, tx_pl_max, tx_pl_min, + tx_num, tx_size_acc, tx_size_min, tx_size_max; + + /* RX stats */ + spinlock_t rx_lock; /* protect RX state */ + unsigned rx_pl_num, rx_pl_max, rx_pl_min, + rx_num, rx_size_acc, rx_size_min, rx_size_max; + + struct mutex msg_mutex; /* serialize command execution */ + struct completion msg_completion; + struct sk_buff *ack_skb; /* protected by rx_lock */ + + void *bm_ack_buf; /* for receiving acks over USB */ + void *bm_cmd_buf; /* for issuing commands over USB */ + + struct workqueue_struct *work_queue; + + struct mutex init_mutex; /* protect bringup seq */ + struct i2400m_reset_ctx *reset_ctx; /* protected by init_mutex */ + + struct work_struct wake_tx_ws; + struct sk_buff *wake_tx_skb; + + struct dentry *debugfs_dentry; +}; + + +/* + * Initialize a 'struct i2400m' from all zeroes + * + * This is a bus-generic API call. + */ +static inline +void i2400m_init(struct i2400m *i2400m) +{ + wimax_dev_init(&i2400m->wimax_dev); + + i2400m->boot_mode = 1; + init_waitqueue_head(&i2400m->state_wq); + + spin_lock_init(&i2400m->tx_lock); + i2400m->tx_pl_min = UINT_MAX; + i2400m->tx_size_min = UINT_MAX; + + spin_lock_init(&i2400m->rx_lock); + i2400m->rx_pl_min = UINT_MAX; + i2400m->rx_size_min = UINT_MAX; + + mutex_init(&i2400m->msg_mutex); + init_completion(&i2400m->msg_completion); + + mutex_init(&i2400m->init_mutex); + /* wake_tx_ws is initialized in i2400m_tx_setup() */ +} + + +/* + * Bus-generic internal APIs + * ------------------------- + */ + +static inline +struct i2400m *wimax_dev_to_i2400m(struct wimax_dev *wimax_dev) +{ + return container_of(wimax_dev, struct i2400m, wimax_dev); +} + +static inline +struct i2400m *net_dev_to_i2400m(struct net_device *net_dev) +{ + return wimax_dev_to_i2400m(netdev_priv(net_dev)); +} + +/* + * Boot mode support + */ + +/** + * i2400m_bm_cmd_flags - flags to i2400m_bm_cmd() + * + * @I2400M_BM_CMD_RAW: send the command block as-is, without doing any + * extra processing for adding CRC. + */ +enum i2400m_bm_cmd_flags { + I2400M_BM_CMD_RAW = 1 << 2, +}; + +/** + * i2400m_bri - Boot-ROM indicators + * + * Flags for i2400m_bootrom_init() and i2400m_dev_bootstrap() [which + * are passed from things like i2400m_setup()]. Can be combined with + * |. + * + * @I2400M_BRI_SOFT: The device rebooted already and a reboot + * barker received, proceed directly to ack the boot sequence. + * @I2400M_BRI_NO_REBOOT: Do not reboot the device and proceed + * directly to wait for a reboot barker from the device. + * @I2400M_BRI_MAC_REINIT: We need to reinitialize the boot + * rom after reading the MAC adress. This is quite a dirty hack, + * if you ask me -- the device requires the bootrom to be + * intialized after reading the MAC address. + */ +enum i2400m_bri { + I2400M_BRI_SOFT = 1 << 1, + I2400M_BRI_NO_REBOOT = 1 << 2, + I2400M_BRI_MAC_REINIT = 1 << 3, +}; + +extern void i2400m_bm_cmd_prepare(struct i2400m_bootrom_header *); +extern int i2400m_dev_bootstrap(struct i2400m *, enum i2400m_bri); +extern int i2400m_read_mac_addr(struct i2400m *); +extern int i2400m_bootrom_init(struct i2400m *, enum i2400m_bri); + +/* Make/grok boot-rom header commands */ + +static inline +__le32 i2400m_brh_command(enum i2400m_brh_opcode opcode, unsigned use_checksum, + unsigned direct_access) +{ + return cpu_to_le32( + I2400M_BRH_SIGNATURE + | (direct_access ? I2400M_BRH_DIRECT_ACCESS : 0) + | I2400M_BRH_RESPONSE_REQUIRED /* response always required */ + | (use_checksum ? I2400M_BRH_USE_CHECKSUM : 0) + | (opcode & I2400M_BRH_OPCODE_MASK)); +} + +static inline +void i2400m_brh_set_opcode(struct i2400m_bootrom_header *hdr, + enum i2400m_brh_opcode opcode) +{ + hdr->command = cpu_to_le32( + (le32_to_cpu(hdr->command) & ~I2400M_BRH_OPCODE_MASK) + | (opcode & I2400M_BRH_OPCODE_MASK)); +} + +static inline +unsigned i2400m_brh_get_opcode(const struct i2400m_bootrom_header *hdr) +{ + return le32_to_cpu(hdr->command) & I2400M_BRH_OPCODE_MASK; +} + +static inline +unsigned i2400m_brh_get_response(const struct i2400m_bootrom_header *hdr) +{ + return (le32_to_cpu(hdr->command) & I2400M_BRH_RESPONSE_MASK) + >> I2400M_BRH_RESPONSE_SHIFT; +} + +static inline +unsigned i2400m_brh_get_use_checksum(const struct i2400m_bootrom_header *hdr) +{ + return le32_to_cpu(hdr->command) & I2400M_BRH_USE_CHECKSUM; +} + +static inline +unsigned i2400m_brh_get_response_required( + const struct i2400m_bootrom_header *hdr) +{ + return le32_to_cpu(hdr->command) & I2400M_BRH_RESPONSE_REQUIRED; +} + +static inline +unsigned i2400m_brh_get_direct_access(const struct i2400m_bootrom_header *hdr) +{ + return le32_to_cpu(hdr->command) & I2400M_BRH_DIRECT_ACCESS; +} + +static inline +unsigned i2400m_brh_get_signature(const struct i2400m_bootrom_header *hdr) +{ + return (le32_to_cpu(hdr->command) & I2400M_BRH_SIGNATURE_MASK) + >> I2400M_BRH_SIGNATURE_SHIFT; +} + + +/* + * Driver / device setup and internal functions + */ +extern void i2400m_netdev_setup(struct net_device *net_dev); +extern int i2400m_tx_setup(struct i2400m *); +extern void i2400m_wake_tx_work(struct work_struct *); +extern void i2400m_tx_release(struct i2400m *); + +extern void i2400m_net_rx(struct i2400m *, struct sk_buff *, unsigned, + const void *, int); +enum i2400m_pt; +extern int i2400m_tx(struct i2400m *, const void *, size_t, enum i2400m_pt); + +#ifdef CONFIG_DEBUG_FS +extern int i2400m_debugfs_add(struct i2400m *); +extern void i2400m_debugfs_rm(struct i2400m *); +#else +static inline int i2400m_debugfs_add(struct i2400m *i2400m) +{ + return 0; +} +static inline void i2400m_debugfs_rm(struct i2400m *i2400m) {} +#endif + +/* Called by _dev_start()/_dev_stop() to initialize the device itself */ +extern int i2400m_dev_initialize(struct i2400m *); +extern void i2400m_dev_shutdown(struct i2400m *); + +extern struct attribute_group i2400m_dev_attr_group; + +extern int i2400m_schedule_work(struct i2400m *, + void (*)(struct work_struct *), gfp_t); + +/* HDI message's payload description handling */ + +static inline +size_t i2400m_pld_size(const struct i2400m_pld *pld) +{ + return I2400M_PLD_SIZE_MASK & le32_to_cpu(pld->val); +} + +static inline +enum i2400m_pt i2400m_pld_type(const struct i2400m_pld *pld) +{ + return (I2400M_PLD_TYPE_MASK & le32_to_cpu(pld->val)) + >> I2400M_PLD_TYPE_SHIFT; +} + +static inline +void i2400m_pld_set(struct i2400m_pld *pld, size_t size, + enum i2400m_pt type) +{ + pld->val = cpu_to_le32( + ((type << I2400M_PLD_TYPE_SHIFT) & I2400M_PLD_TYPE_MASK) + | (size & I2400M_PLD_SIZE_MASK)); +} + + +/* + * API for the bus-specific drivers + * -------------------------------- + */ + +static inline +struct i2400m *i2400m_get(struct i2400m *i2400m) +{ + dev_hold(i2400m->wimax_dev.net_dev); + return i2400m; +} + +static inline +void i2400m_put(struct i2400m *i2400m) +{ + dev_put(i2400m->wimax_dev.net_dev); +} + +extern int i2400m_dev_reset_handle(struct i2400m *); + +/* + * _setup()/_release() are called by the probe/disconnect functions of + * the bus-specific drivers. + */ +extern int i2400m_setup(struct i2400m *, enum i2400m_bri bm_flags); +extern void i2400m_release(struct i2400m *); + +extern int i2400m_rx(struct i2400m *, struct sk_buff *); +extern struct i2400m_msg_hdr *i2400m_tx_msg_get(struct i2400m *, size_t *); +extern void i2400m_tx_msg_sent(struct i2400m *); + +static const __le32 i2400m_NBOOT_BARKER[4] = { + __constant_cpu_to_le32(I2400M_NBOOT_BARKER), + __constant_cpu_to_le32(I2400M_NBOOT_BARKER), + __constant_cpu_to_le32(I2400M_NBOOT_BARKER), + __constant_cpu_to_le32(I2400M_NBOOT_BARKER) +}; + +static const __le32 i2400m_SBOOT_BARKER[4] = { + __constant_cpu_to_le32(I2400M_SBOOT_BARKER), + __constant_cpu_to_le32(I2400M_SBOOT_BARKER), + __constant_cpu_to_le32(I2400M_SBOOT_BARKER), + __constant_cpu_to_le32(I2400M_SBOOT_BARKER) +}; + + +/* + * Utility functions + */ + +static inline +struct device *i2400m_dev(struct i2400m *i2400m) +{ + return i2400m->wimax_dev.net_dev->dev.parent; +} + +/* + * Helper for scheduling simple work functions + * + * This struct can get any kind of payload attached (normally in the + * form of a struct where you pack the stuff you want to pass to the + * _work function). + */ +struct i2400m_work { + struct work_struct ws; + struct i2400m *i2400m; + u8 pl[0]; +}; +extern int i2400m_queue_work(struct i2400m *, + void (*)(struct work_struct *), gfp_t, + const void *, size_t); + +extern int i2400m_msg_check_status(const struct i2400m_l3l4_hdr *, + char *, size_t); +extern int i2400m_msg_size_check(struct i2400m *, + const struct i2400m_l3l4_hdr *, size_t); +extern struct sk_buff *i2400m_msg_to_dev(struct i2400m *, const void *, size_t); +extern void i2400m_msg_to_dev_cancel_wait(struct i2400m *, int); +extern void i2400m_msg_ack_hook(struct i2400m *, + const struct i2400m_l3l4_hdr *, size_t); +extern void i2400m_report_hook(struct i2400m *, + const struct i2400m_l3l4_hdr *, size_t); +extern int i2400m_cmd_enter_powersave(struct i2400m *); +extern int i2400m_cmd_get_state(struct i2400m *); +extern int i2400m_cmd_exit_idle(struct i2400m *); +extern struct sk_buff *i2400m_get_device_info(struct i2400m *); +extern int i2400m_firmware_check(struct i2400m *); +extern int i2400m_set_init_config(struct i2400m *, + const struct i2400m_tlv_hdr **, size_t); + +static inline +struct usb_endpoint_descriptor *usb_get_epd(struct usb_interface *iface, int ep) +{ + return &iface->cur_altsetting->endpoint[ep].desc; +} + +extern int i2400m_op_rfkill_sw_toggle(struct wimax_dev *, + enum wimax_rf_state); +extern void i2400m_report_tlv_rf_switches_status( + struct i2400m *, const struct i2400m_tlv_rf_switches_status *); + + +/* + * Do a millisecond-sleep for allowing wireshark to dump all the data + * packets. Used only for debugging. + */ +static inline +void __i2400m_msleep(unsigned ms) +{ +#if 1 +#else + msleep(ms); +#endif +} + +/* Module parameters */ + +extern int i2400m_idle_mode_disabled; + + +#endif /* #ifndef __I2400M_H__ */ diff --git a/include/linux/wimax/i2400m.h b/include/linux/wimax/i2400m.h new file mode 100644 index 00000000000..74198f5bb4d --- /dev/null +++ b/include/linux/wimax/i2400m.h @@ -0,0 +1,512 @@ +/* + * Intel Wireless WiMax Connection 2400m + * Host-Device protocol interface definitions + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Inaky Perez-Gonzalez + * - Initial implementation + * + * + * This header defines the data structures and constants used to + * communicate with the device. + * + * BOOTMODE/BOOTROM/FIRMWARE UPLOAD PROTOCOL + * + * The firmware upload protocol is quite simple and only requires a + * handful of commands. See drivers/net/wimax/i2400m/fw.c for more + * details. + * + * The BCF data structure is for the firmware file header. + * + * + * THE DATA / CONTROL PROTOCOL + * + * This is the normal protocol spoken with the device once the + * firmware is uploaded. It transports data payloads and control + * messages back and forth. + * + * It consists 'messages' that pack one or more payloads each. The + * format is described in detail in drivers/net/wimax/i2400m/rx.c and + * tx.c. + * + * + * THE L3L4 PROTOCOL + * + * The term L3L4 refers to Layer 3 (the device), Layer 4 (the + * driver/host software). + * + * This is the control protocol used by the host to control the i2400m + * device (scan, connect, disconnect...). This is sent to / received + * as control frames. These frames consist of a header and zero or + * more TLVs with information. We call each control frame a "message". + * + * Each message is composed of: + * + * HEADER + * [TLV0 + PAYLOAD0] + * [TLV1 + PAYLOAD1] + * [...] + * [TLVN + PAYLOADN] + * + * The HEADER is defined by 'struct i2400m_l3l4_hdr'. The payloads are + * defined by a TLV structure (Type Length Value) which is a 'header' + * (struct i2400m_tlv_hdr) and then the payload. + * + * All integers are represented as Little Endian. + * + * - REQUESTS AND EVENTS + * + * The requests can be clasified as follows: + * + * COMMAND: implies a request from the host to the device requesting + * an action being performed. The device will reply with a + * message (with the same type as the command), status and + * no (TLV) payload. Execution of a command might cause + * events (of different type) to be sent later on as + * device's state changes. + * + * GET/SET: similar to COMMAND, but will not cause other + * EVENTs. The reply, in the case of GET, will contain + * TLVs with the requested information. + * + * EVENT: asynchronous messages sent from the device, maybe as a + * consequence of previous COMMANDs but disassociated from + * them. + * + * Only one request might be pending at the same time (ie: don't + * parallelize nor post another GET request before the previous + * COMMAND has been acknowledged with it's corresponding reply by the + * device). + * + * The different requests and their formats are described below: + * + * I2400M_MT_* Message types + * I2400M_MS_* Message status (for replies, events) + * i2400m_tlv_* TLVs + * + * data types are named 'struct i2400m_msg_OPNAME', OPNAME matching the + * operation. + */ + +#ifndef __LINUX__WIMAX__I2400M_H__ +#define __LINUX__WIMAX__I2400M_H__ + +#include + + +/* + * Host Device Interface (HDI) common to all busses + */ + +/* Boot-mode (firmware upload mode) commands */ + +/* Header for the firmware file */ +struct i2400m_bcf_hdr { + __le32 module_type; + __le32 header_len; + __le32 header_version; + __le32 module_id; + __le32 module_vendor; + __le32 date; /* BCD YYYMMDD */ + __le32 size; + __le32 key_size; /* in dwords */ + __le32 modulus_size; /* in dwords */ + __le32 exponent_size; /* in dwords */ + __u8 reserved[88]; +} __attribute__ ((packed)); + +/* Boot mode opcodes */ +enum i2400m_brh_opcode { + I2400M_BRH_READ = 1, + I2400M_BRH_WRITE = 2, + I2400M_BRH_JUMP = 3, + I2400M_BRH_SIGNED_JUMP = 8, + I2400M_BRH_HASH_PAYLOAD_ONLY = 9, +}; + +/* Boot mode command masks and stuff */ +enum i2400m_brh { + I2400M_BRH_SIGNATURE = 0xcbbc0000, + I2400M_BRH_SIGNATURE_MASK = 0xffff0000, + I2400M_BRH_SIGNATURE_SHIFT = 16, + I2400M_BRH_OPCODE_MASK = 0x0000000f, + I2400M_BRH_RESPONSE_MASK = 0x000000f0, + I2400M_BRH_RESPONSE_SHIFT = 4, + I2400M_BRH_DIRECT_ACCESS = 0x00000400, + I2400M_BRH_RESPONSE_REQUIRED = 0x00000200, + I2400M_BRH_USE_CHECKSUM = 0x00000100, +}; + + +/* Constants for bcf->module_id */ +enum i2400m_bcf_mod_id { + /* Firmware file carries its own pokes -- pokes are a set of + * magical values that have to be written in certain memory + * addresses to get the device up and ready for firmware + * download when it is in non-signed boot mode. */ + I2400M_BCF_MOD_ID_POKES = 0x000000001, +}; + + +/** + * i2400m_bootrom_header - Header for a boot-mode command + * + * @cmd: the above command descriptor + * @target_addr: where on the device memory should the action be performed. + * @data_size: for read/write, amount of data to be read/written + * @block_checksum: checksum value (if applicable) + * @payload: the beginning of data attached to this header + */ +struct i2400m_bootrom_header { + __le32 command; /* Compose with enum i2400_brh */ + __le32 target_addr; + __le32 data_size; + __le32 block_checksum; + char payload[0]; +} __attribute__ ((packed)); + + +/* + * Data / control protocol + */ + +/* Packet types for the host-device interface */ +enum i2400m_pt { + I2400M_PT_DATA = 0, + I2400M_PT_CTRL, + I2400M_PT_TRACE, /* For device debug */ + I2400M_PT_RESET_WARM, /* device reset */ + I2400M_PT_RESET_COLD, /* USB[transport] reset, like reconnect */ + I2400M_PT_ILLEGAL +}; + + +/* + * Payload for a data packet + * + * This is prefixed to each and every outgoing DATA type. + */ +struct i2400m_pl_data_hdr { + __le32 reserved; +} __attribute__((packed)); + + +/* Misc constants */ +enum { + I2400M_PL_PAD = 16, /* Payload data size alignment */ + I2400M_PL_SIZE_MAX = 0x3EFF, + I2400M_MAX_PLS_IN_MSG = 60, + /* protocol barkers: sync sequences; for notifications they + * are sent in groups of four. */ + I2400M_H2D_PREVIEW_BARKER = 0xcafe900d, + I2400M_COLD_RESET_BARKER = 0xc01dc01d, + I2400M_WARM_RESET_BARKER = 0x50f750f7, + I2400M_NBOOT_BARKER = 0xdeadbeef, + I2400M_SBOOT_BARKER = 0x0ff1c1a1, + I2400M_ACK_BARKER = 0xfeedbabe, + I2400M_D2H_MSG_BARKER = 0xbeefbabe, +}; + + +/* + * Hardware payload descriptor + * + * Bitfields encoded in a struct to enforce typing semantics. + * + * Look in rx.c and tx.c for a full description of the format. + */ +struct i2400m_pld { + __le32 val; +} __attribute__ ((packed)); + +#define I2400M_PLD_SIZE_MASK 0x00003fff +#define I2400M_PLD_TYPE_SHIFT 16 +#define I2400M_PLD_TYPE_MASK 0x000f0000 + +/* + * Header for a TX message or RX message + * + * @barker: preamble + * @size: used for management of the FIFO queue buffer; before + * sending, this is converted to be a real preamble. This + * indicates the real size of the TX message that starts at this + * point. If the highest bit is set, then this message is to be + * skipped. + * @sequence: sequence number of this message + * @offset: offset where the message itself starts -- see the comments + * in the file header about message header and payload descriptor + * alignment. + * @num_pls: number of payloads in this message + * @padding: amount of padding bytes at the end of the message to make + * it be of block-size aligned + * + * Look in rx.c and tx.c for a full description of the format. + */ +struct i2400m_msg_hdr { + union { + __le32 barker; + __u32 size; /* same size type as barker!! */ + }; + union { + __le32 sequence; + __u32 offset; /* same size type as barker!! */ + }; + __le16 num_pls; + __le16 rsv1; + __le16 padding; + __le16 rsv2; + struct i2400m_pld pld[0]; +} __attribute__ ((packed)); + + + +/* + * L3/L4 control protocol + */ + +enum { + /* Interface version */ + I2400M_L3L4_VERSION = 0x0100, +}; + +/* Message types */ +enum i2400m_mt { + I2400M_MT_RESERVED = 0x0000, + I2400M_MT_INVALID = 0xffff, + I2400M_MT_REPORT_MASK = 0x8000, + + I2400M_MT_GET_SCAN_RESULT = 0x4202, + I2400M_MT_SET_SCAN_PARAM = 0x4402, + I2400M_MT_CMD_RF_CONTROL = 0x4602, + I2400M_MT_CMD_SCAN = 0x4603, + I2400M_MT_CMD_CONNECT = 0x4604, + I2400M_MT_CMD_DISCONNECT = 0x4605, + I2400M_MT_CMD_EXIT_IDLE = 0x4606, + I2400M_MT_GET_LM_VERSION = 0x5201, + I2400M_MT_GET_DEVICE_INFO = 0x5202, + I2400M_MT_GET_LINK_STATUS = 0x5203, + I2400M_MT_GET_STATISTICS = 0x5204, + I2400M_MT_GET_STATE = 0x5205, + I2400M_MT_GET_MEDIA_STATUS = 0x5206, + I2400M_MT_SET_INIT_CONFIG = 0x5404, + I2400M_MT_CMD_INIT = 0x5601, + I2400M_MT_CMD_TERMINATE = 0x5602, + I2400M_MT_CMD_MODE_OF_OP = 0x5603, + I2400M_MT_CMD_RESET_DEVICE = 0x5604, + I2400M_MT_CMD_MONITOR_CONTROL = 0x5605, + I2400M_MT_CMD_ENTER_POWERSAVE = 0x5606, + I2400M_MT_GET_TLS_OPERATION_RESULT = 0x6201, + I2400M_MT_SET_EAP_SUCCESS = 0x6402, + I2400M_MT_SET_EAP_FAIL = 0x6403, + I2400M_MT_SET_EAP_KEY = 0x6404, + I2400M_MT_CMD_SEND_EAP_RESPONSE = 0x6602, + I2400M_MT_REPORT_SCAN_RESULT = 0xc002, + I2400M_MT_REPORT_STATE = 0xd002, + I2400M_MT_REPORT_POWERSAVE_READY = 0xd005, + I2400M_MT_REPORT_EAP_REQUEST = 0xe002, + I2400M_MT_REPORT_EAP_RESTART = 0xe003, + I2400M_MT_REPORT_ALT_ACCEPT = 0xe004, + I2400M_MT_REPORT_KEY_REQUEST = 0xe005, +}; + + +/* + * Message Ack Status codes + * + * When a message is replied-to, this status is reported. + */ +enum i2400m_ms { + I2400M_MS_DONE_OK = 0, + I2400M_MS_DONE_IN_PROGRESS = 1, + I2400M_MS_INVALID_OP = 2, + I2400M_MS_BAD_STATE = 3, + I2400M_MS_ILLEGAL_VALUE = 4, + I2400M_MS_MISSING_PARAMS = 5, + I2400M_MS_VERSION_ERROR = 6, + I2400M_MS_ACCESSIBILITY_ERROR = 7, + I2400M_MS_BUSY = 8, + I2400M_MS_CORRUPTED_TLV = 9, + I2400M_MS_UNINITIALIZED = 10, + I2400M_MS_UNKNOWN_ERROR = 11, + I2400M_MS_PRODUCTION_ERROR = 12, + I2400M_MS_NO_RF = 13, + I2400M_MS_NOT_READY_FOR_POWERSAVE = 14, + I2400M_MS_THERMAL_CRITICAL = 15, + I2400M_MS_MAX +}; + + +/** + * i2400m_tlv - enumeration of the different types of TLVs + * + * TLVs stand for type-length-value and are the header for a payload + * composed of almost anything. Each payload has a type assigned + * and a length. + */ +enum i2400m_tlv { + I2400M_TLV_L4_MESSAGE_VERSIONS = 129, + I2400M_TLV_SYSTEM_STATE = 141, + I2400M_TLV_MEDIA_STATUS = 161, + I2400M_TLV_RF_OPERATION = 162, + I2400M_TLV_RF_STATUS = 163, + I2400M_TLV_DEVICE_RESET_TYPE = 132, + I2400M_TLV_CONFIG_IDLE_PARAMETERS = 601, +}; + + +struct i2400m_tlv_hdr { + __le16 type; + __le16 length; /* payload's */ + __u8 pl[0]; +} __attribute__((packed)); + + +struct i2400m_l3l4_hdr { + __le16 type; + __le16 length; /* payload's */ + __le16 version; + __le16 resv1; + __le16 status; + __le16 resv2; + struct i2400m_tlv_hdr pl[0]; +} __attribute__((packed)); + + +/** + * i2400m_system_state - different states of the device + */ +enum i2400m_system_state { + I2400M_SS_UNINITIALIZED = 1, + I2400M_SS_INIT, + I2400M_SS_READY, + I2400M_SS_SCAN, + I2400M_SS_STANDBY, + I2400M_SS_CONNECTING, + I2400M_SS_WIMAX_CONNECTED, + I2400M_SS_DATA_PATH_CONNECTED, + I2400M_SS_IDLE, + I2400M_SS_DISCONNECTING, + I2400M_SS_OUT_OF_ZONE, + I2400M_SS_SLEEPACTIVE, + I2400M_SS_PRODUCTION, + I2400M_SS_CONFIG, + I2400M_SS_RF_OFF, + I2400M_SS_RF_SHUTDOWN, + I2400M_SS_DEVICE_DISCONNECT, + I2400M_SS_MAX, +}; + + +/** + * i2400m_tlv_system_state - report on the state of the system + * + * @state: see enum i2400m_system_state + */ +struct i2400m_tlv_system_state { + struct i2400m_tlv_hdr hdr; + __le32 state; +} __attribute__((packed)); + + +struct i2400m_tlv_l4_message_versions { + struct i2400m_tlv_hdr hdr; + __le16 major; + __le16 minor; + __le16 branch; + __le16 reserved; +} __attribute__((packed)); + + +struct i2400m_tlv_detailed_device_info { + struct i2400m_tlv_hdr hdr; + __u8 reserved1[400]; + __u8 mac_address[6]; + __u8 reserved2[2]; +} __attribute__((packed)); + + +enum i2400m_rf_switch_status { + I2400M_RF_SWITCH_ON = 1, + I2400M_RF_SWITCH_OFF = 2, +}; + +struct i2400m_tlv_rf_switches_status { + struct i2400m_tlv_hdr hdr; + __u8 sw_rf_switch; /* 1 ON, 2 OFF */ + __u8 hw_rf_switch; /* 1 ON, 2 OFF */ + __u8 reserved[2]; +} __attribute__((packed)); + + +enum { + i2400m_rf_operation_on = 1, + i2400m_rf_operation_off = 2 +}; + +struct i2400m_tlv_rf_operation { + struct i2400m_tlv_hdr hdr; + __le32 status; /* 1 ON, 2 OFF */ +} __attribute__((packed)); + + +enum i2400m_tlv_reset_type { + I2400M_RESET_TYPE_COLD = 1, + I2400M_RESET_TYPE_WARM +}; + +struct i2400m_tlv_device_reset_type { + struct i2400m_tlv_hdr hdr; + __le32 reset_type; +} __attribute__((packed)); + + +struct i2400m_tlv_config_idle_parameters { + struct i2400m_tlv_hdr hdr; + __le32 idle_timeout; /* 100 to 300000 ms [5min], 100 increments + * 0 disabled */ + __le32 idle_paging_interval; /* frames */ +} __attribute__((packed)); + + +enum i2400m_media_status { + I2400M_MEDIA_STATUS_LINK_UP = 1, + I2400M_MEDIA_STATUS_LINK_DOWN, + I2400M_MEDIA_STATUS_LINK_RENEW, +}; + +struct i2400m_tlv_media_status { + struct i2400m_tlv_hdr hdr; + __le32 media_status; +} __attribute__((packed)); + +#endif /* #ifndef __LINUX__WIMAX__I2400M_H__ */ From 024f7f31ed15c471f80408d8b5045497e27e1135 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:44 -0800 Subject: [PATCH 107/123] i2400m: Generic probe/disconnect, reset and message passing Implements the generic probe and disconnect functions that will be called by the USB and SDIO driver's probe/disconnect functions. Implements the backends for the WiMAX stack's basic operations: message passing, rfkill control and reset. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/driver.c | 728 +++++++++++++++++++++++++++ drivers/net/wimax/i2400m/op-rfkill.c | 207 ++++++++ 2 files changed, 935 insertions(+) create mode 100644 drivers/net/wimax/i2400m/driver.c create mode 100644 drivers/net/wimax/i2400m/op-rfkill.c diff --git a/drivers/net/wimax/i2400m/driver.c b/drivers/net/wimax/i2400m/driver.c new file mode 100644 index 00000000000..5f98047e18c --- /dev/null +++ b/drivers/net/wimax/i2400m/driver.c @@ -0,0 +1,728 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Generic probe/disconnect, reset and message passing + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * See i2400m.h for driver documentation. This contains helpers for + * the driver model glue [_setup()/_release()], handling device resets + * [_dev_reset_handle()], and the backends for the WiMAX stack ops + * reset [_op_reset()] and message from user [_op_msg_from_user()]. + * + * ROADMAP: + * + * i2400m_op_msg_from_user() + * i2400m_msg_to_dev() + * wimax_msg_to_user_send() + * + * i2400m_op_reset() + * i240m->bus_reset() + * + * i2400m_dev_reset_handle() + * __i2400m_dev_reset_handle() + * __i2400m_dev_stop() + * __i2400m_dev_start() + * + * i2400m_setup() + * i2400m_bootrom_init() + * register_netdev() + * i2400m_dev_start() + * __i2400m_dev_start() + * i2400m_dev_bootstrap() + * i2400m_tx_setup() + * i2400m->bus_dev_start() + * i2400m_check_mac_addr() + * wimax_dev_add() + * + * i2400m_release() + * wimax_dev_rm() + * i2400m_dev_stop() + * __i2400m_dev_stop() + * i2400m_dev_shutdown() + * i2400m->bus_dev_stop() + * i2400m_tx_release() + * unregister_netdev() + */ +#include "i2400m.h" +#include +#include +#include + +#define D_SUBMODULE driver +#include "debug-levels.h" + + +int i2400m_idle_mode_disabled; /* 0 (idle mode enabled) by default */ +module_param_named(idle_mode_disabled, i2400m_idle_mode_disabled, int, 0644); +MODULE_PARM_DESC(idle_mode_disabled, + "If true, the device will not enable idle mode negotiation " + "with the base station (when connected) to save power."); + +/** + * i2400m_queue_work - schedule work on a i2400m's queue + * + * @i2400m: device descriptor + * + * @fn: function to run to execute work. It gets passed a 'struct + * work_struct' that is wrapped in a 'struct i2400m_work'. Once + * done, you have to (1) i2400m_put(i2400m_work->i2400m) and then + * (2) kfree(i2400m_work). + * + * @gfp_flags: GFP flags for memory allocation. + * + * @pl: pointer to a payload buffer that you want to pass to the _work + * function. Use this to pack (for example) a struct with extra + * arguments. + * + * @pl_size: size of the payload buffer. + * + * We do this quite often, so this just saves typing; allocate a + * wrapper for a i2400m, get a ref to it, pack arguments and launch + * the work. + * + * A usual workflow is: + * + * struct my_work_args { + * void *something; + * int whatever; + * }; + * ... + * + * struct my_work_args my_args = { + * .something = FOO, + * .whaetever = BLAH + * }; + * i2400m_queue_work(i2400m, 1, my_work_function, GFP_KERNEL, + * &args, sizeof(args)) + * + * And now the work function can unpack the arguments and call the + * real function (or do the job itself): + * + * static + * void my_work_fn((struct work_struct *ws) + * { + * struct i2400m_work *iw = + * container_of(ws, struct i2400m_work, ws); + * struct my_work_args *my_args = (void *) iw->pl; + * + * my_work(iw->i2400m, my_args->something, my_args->whatevert); + * } + */ +int i2400m_queue_work(struct i2400m *i2400m, + void (*fn)(struct work_struct *), gfp_t gfp_flags, + const void *pl, size_t pl_size) +{ + int result; + struct i2400m_work *iw; + + BUG_ON(i2400m->work_queue == NULL); + result = -ENOMEM; + iw = kzalloc(sizeof(*iw) + pl_size, gfp_flags); + if (iw == NULL) + goto error_kzalloc; + iw->i2400m = i2400m_get(i2400m); + memcpy(iw->pl, pl, pl_size); + INIT_WORK(&iw->ws, fn); + result = queue_work(i2400m->work_queue, &iw->ws); +error_kzalloc: + return result; +} +EXPORT_SYMBOL_GPL(i2400m_queue_work); + + +/* + * Schedule i2400m's specific work on the system's queue. + * + * Used for a few cases where we really need it; otherwise, identical + * to i2400m_queue_work(). + * + * Returns < 0 errno code on error, 1 if ok. + * + * If it returns zero, something really bad happened, as it means the + * works struct was already queued, but we have just allocated it, so + * it should not happen. + */ +int i2400m_schedule_work(struct i2400m *i2400m, + void (*fn)(struct work_struct *), gfp_t gfp_flags) +{ + int result; + struct i2400m_work *iw; + + BUG_ON(i2400m->work_queue == NULL); + result = -ENOMEM; + iw = kzalloc(sizeof(*iw), gfp_flags); + if (iw == NULL) + goto error_kzalloc; + iw->i2400m = i2400m_get(i2400m); + INIT_WORK(&iw->ws, fn); + result = schedule_work(&iw->ws); + if (result == 0) + result = -ENXIO; +error_kzalloc: + return result; +} + + +/* + * WiMAX stack operation: relay a message from user space + * + * @wimax_dev: device descriptor + * @pipe_name: named pipe the message is for + * @msg_buf: pointer to the message bytes + * @msg_len: length of the buffer + * @genl_info: passed by the generic netlink layer + * + * The WiMAX stack will call this function when a message was received + * from user space. + * + * For the i2400m, this is an L3L4 message, as specified in + * include/linux/wimax/i2400m.h, and thus prefixed with a 'struct + * i2400m_l3l4_hdr'. Driver (and device) expect the messages to be + * coded in Little Endian. + * + * This function just verifies that the header declaration and the + * payload are consistent and then deals with it, either forwarding it + * to the device or procesing it locally. + * + * In the i2400m, messages are basically commands that will carry an + * ack, so we use i2400m_msg_to_dev() and then deliver the ack back to + * user space. The rx.c code might intercept the response and use it + * to update the driver's state, but then it will pass it on so it can + * be relayed back to user space. + * + * Note that asynchronous events from the device are processed and + * sent to user space in rx.c. + */ +static +int i2400m_op_msg_from_user(struct wimax_dev *wimax_dev, + const char *pipe_name, + const void *msg_buf, size_t msg_len, + const struct genl_info *genl_info) +{ + int result; + struct i2400m *i2400m = wimax_dev_to_i2400m(wimax_dev); + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *ack_skb; + + d_fnstart(4, dev, "(wimax_dev %p [i2400m %p] msg_buf %p " + "msg_len %zu genl_info %p)\n", wimax_dev, i2400m, + msg_buf, msg_len, genl_info); + ack_skb = i2400m_msg_to_dev(i2400m, msg_buf, msg_len); + result = PTR_ERR(ack_skb); + if (IS_ERR(ack_skb)) + goto error_msg_to_dev; + if (unlikely(i2400m->trace_msg_from_user)) + wimax_msg(&i2400m->wimax_dev, "trace", + msg_buf, msg_len, GFP_KERNEL); + result = wimax_msg_send(&i2400m->wimax_dev, ack_skb); +error_msg_to_dev: + d_fnend(4, dev, "(wimax_dev %p [i2400m %p] msg_buf %p msg_len %zu " + "genl_info %p) = %d\n", wimax_dev, i2400m, msg_buf, msg_len, + genl_info, result); + return result; +} + + +/* + * Context to wait for a reset to finalize + */ +struct i2400m_reset_ctx { + struct completion completion; + int result; +}; + + +/* + * WiMAX stack operation: reset a device + * + * @wimax_dev: device descriptor + * + * See the documentation for wimax_reset() and wimax_dev->op_reset for + * the requirements of this function. The WiMAX stack guarantees + * serialization on calls to this function. + * + * Do a warm reset on the device; if it fails, resort to a cold reset + * and return -ENODEV. On successful warm reset, we need to block + * until it is complete. + * + * The bus-driver implementation of reset takes care of falling back + * to cold reset if warm fails. + */ +static +int i2400m_op_reset(struct wimax_dev *wimax_dev) +{ + int result; + struct i2400m *i2400m = wimax_dev_to_i2400m(wimax_dev); + struct device *dev = i2400m_dev(i2400m); + struct i2400m_reset_ctx ctx = { + .completion = COMPLETION_INITIALIZER_ONSTACK(ctx.completion), + .result = 0, + }; + + d_fnstart(4, dev, "(wimax_dev %p)\n", wimax_dev); + mutex_lock(&i2400m->init_mutex); + i2400m->reset_ctx = &ctx; + mutex_unlock(&i2400m->init_mutex); + result = i2400m->bus_reset(i2400m, I2400M_RT_WARM); + if (result < 0) + goto out; + result = wait_for_completion_timeout(&ctx.completion, 4*HZ); + if (result == 0) + result = -ETIMEDOUT; + else if (result > 0) + result = ctx.result; + /* if result < 0, pass it on */ + mutex_lock(&i2400m->init_mutex); + i2400m->reset_ctx = NULL; + mutex_unlock(&i2400m->init_mutex); +out: + d_fnend(4, dev, "(wimax_dev %p) = %d\n", wimax_dev, result); + return result; +} + + +/* + * Check the MAC address we got from boot mode is ok + * + * @i2400m: device descriptor + * + * Returns: 0 if ok, < 0 errno code on error. + */ +static +int i2400m_check_mac_addr(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *skb; + const struct i2400m_tlv_detailed_device_info *ddi; + struct net_device *net_dev = i2400m->wimax_dev.net_dev; + const unsigned char zeromac[ETH_ALEN] = { 0 }; + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + skb = i2400m_get_device_info(i2400m); + if (IS_ERR(skb)) { + result = PTR_ERR(skb); + dev_err(dev, "Cannot verify MAC address, error reading: %d\n", + result); + goto error; + } + /* Extract MAC addresss */ + ddi = (void *) skb->data; + BUILD_BUG_ON(ETH_ALEN != sizeof(ddi->mac_address)); + d_printf(2, dev, "GET DEVICE INFO: mac addr " + "%02x:%02x:%02x:%02x:%02x:%02x\n", + ddi->mac_address[0], ddi->mac_address[1], + ddi->mac_address[2], ddi->mac_address[3], + ddi->mac_address[4], ddi->mac_address[5]); + if (!memcmp(net_dev->perm_addr, ddi->mac_address, + sizeof(ddi->mac_address))) + goto ok; + dev_warn(dev, "warning: device reports a different MAC address " + "to that of boot mode's\n"); + dev_warn(dev, "device reports %02x:%02x:%02x:%02x:%02x:%02x\n", + ddi->mac_address[0], ddi->mac_address[1], + ddi->mac_address[2], ddi->mac_address[3], + ddi->mac_address[4], ddi->mac_address[5]); + dev_warn(dev, "boot mode reported %02x:%02x:%02x:%02x:%02x:%02x\n", + net_dev->perm_addr[0], net_dev->perm_addr[1], + net_dev->perm_addr[2], net_dev->perm_addr[3], + net_dev->perm_addr[4], net_dev->perm_addr[5]); + if (!memcmp(zeromac, ddi->mac_address, sizeof(zeromac))) + dev_err(dev, "device reports an invalid MAC address, " + "not updating\n"); + else { + dev_warn(dev, "updating MAC address\n"); + net_dev->addr_len = ETH_ALEN; + memcpy(net_dev->perm_addr, ddi->mac_address, ETH_ALEN); + memcpy(net_dev->dev_addr, ddi->mac_address, ETH_ALEN); + } +ok: + result = 0; + kfree_skb(skb); +error: + d_fnend(3, dev, "(i2400m %p) = %d\n", i2400m, result); + return result; +} + + +/** + * __i2400m_dev_start - Bring up driver communication with the device + * + * @i2400m: device descriptor + * @flags: boot mode flags + * + * Returns: 0 if ok, < 0 errno code on error. + * + * Uploads firmware and brings up all the resources needed to be able + * to communicate with the device. + * + * TX needs to be setup before the bus-specific code (otherwise on + * shutdown, the bus-tx code could try to access it). + */ +static +int __i2400m_dev_start(struct i2400m *i2400m, enum i2400m_bri flags) +{ + int result; + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + struct net_device *net_dev = wimax_dev->net_dev; + struct device *dev = i2400m_dev(i2400m); + int times = 3; + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); +retry: + result = i2400m_dev_bootstrap(i2400m, flags); + if (result < 0) { + dev_err(dev, "cannot bootstrap device: %d\n", result); + goto error_bootstrap; + } + result = i2400m_tx_setup(i2400m); + if (result < 0) + goto error_tx_setup; + result = i2400m->bus_dev_start(i2400m); + if (result < 0) + goto error_bus_dev_start; + i2400m->work_queue = create_singlethread_workqueue(wimax_dev->name); + if (i2400m->work_queue == NULL) { + result = -ENOMEM; + dev_err(dev, "cannot create workqueue\n"); + goto error_create_workqueue; + } + /* At this point is ok to send commands to the device */ + result = i2400m_check_mac_addr(i2400m); + if (result < 0) + goto error_check_mac_addr; + i2400m->ready = 1; + wimax_state_change(wimax_dev, WIMAX_ST_UNINITIALIZED); + result = i2400m_dev_initialize(i2400m); + if (result < 0) + goto error_dev_initialize; + /* At this point, reports will come for the device and set it + * to the right state if it is different than UNINITIALIZED */ + d_fnend(3, dev, "(net_dev %p [i2400m %p]) = %d\n", + net_dev, i2400m, result); + return result; + +error_dev_initialize: +error_check_mac_addr: + destroy_workqueue(i2400m->work_queue); +error_create_workqueue: + i2400m->bus_dev_stop(i2400m); +error_bus_dev_start: + i2400m_tx_release(i2400m); +error_tx_setup: +error_bootstrap: + if (result == -ERESTARTSYS && times-- > 0) { + flags = I2400M_BRI_SOFT; + goto retry; + } + d_fnend(3, dev, "(net_dev %p [i2400m %p]) = %d\n", + net_dev, i2400m, result); + return result; +} + + +static +int i2400m_dev_start(struct i2400m *i2400m, enum i2400m_bri bm_flags) +{ + int result; + mutex_lock(&i2400m->init_mutex); /* Well, start the device */ + result = __i2400m_dev_start(i2400m, bm_flags); + if (result >= 0) + i2400m->updown = 1; + mutex_unlock(&i2400m->init_mutex); + return result; +} + + +/** + * i2400m_dev_stop - Tear down driver communication with the device + * + * @i2400m: device descriptor + * + * Returns: 0 if ok, < 0 errno code on error. + * + * Releases all the resources allocated to communicate with the device. + */ +static +void __i2400m_dev_stop(struct i2400m *i2400m) +{ + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + wimax_state_change(wimax_dev, __WIMAX_ST_QUIESCING); + i2400m_dev_shutdown(i2400m); + i2400m->ready = 0; + destroy_workqueue(i2400m->work_queue); + i2400m->bus_dev_stop(i2400m); + i2400m_tx_release(i2400m); + wimax_state_change(wimax_dev, WIMAX_ST_DOWN); + d_fnend(3, dev, "(i2400m %p) = 0\n", i2400m); +} + + +/* + * Watch out -- we only need to stop if there is a need for it. The + * device could have reset itself and failed to come up again (see + * _i2400m_dev_reset_handle()). + */ +static +void i2400m_dev_stop(struct i2400m *i2400m) +{ + mutex_lock(&i2400m->init_mutex); + if (i2400m->updown) { + __i2400m_dev_stop(i2400m); + i2400m->updown = 0; + } + mutex_unlock(&i2400m->init_mutex); +} + + +/* + * The device has rebooted; fix up the device and the driver + * + * Tear down the driver communication with the device, reload the + * firmware and reinitialize the communication with the device. + * + * If someone calls a reset when the device's firmware is down, in + * theory we won't see it because we are not listening. However, just + * in case, leave the code to handle it. + * + * If there is a reset context, use it; this means someone is waiting + * for us to tell him when the reset operation is complete and the + * device is ready to rock again. + * + * NOTE: if we are in the process of bringing up or down the + * communication with the device [running i2400m_dev_start() or + * _stop()], don't do anything, let it fail and handle it. + * + * This function is ran always in a thread context + */ +static +void __i2400m_dev_reset_handle(struct work_struct *ws) +{ + int result; + struct i2400m_work *iw = container_of(ws, struct i2400m_work, ws); + struct i2400m *i2400m = iw->i2400m; + struct device *dev = i2400m_dev(i2400m); + enum wimax_st wimax_state; + struct i2400m_reset_ctx *ctx = i2400m->reset_ctx; + + d_fnstart(3, dev, "(ws %p i2400m %p)\n", ws, i2400m); + result = 0; + if (mutex_trylock(&i2400m->init_mutex) == 0) { + /* We are still in i2400m_dev_start() [let it fail] or + * i2400m_dev_stop() [we are shutting down anyway, so + * ignore it] or we are resetting somewhere else. */ + dev_err(dev, "device rebooted\n"); + i2400m_msg_to_dev_cancel_wait(i2400m, -ERESTARTSYS); + complete(&i2400m->msg_completion); + goto out; + } + wimax_state = wimax_state_get(&i2400m->wimax_dev); + if (wimax_state < WIMAX_ST_UNINITIALIZED) { + dev_info(dev, "device rebooted: it is down, ignoring\n"); + goto out_unlock; /* ifconfig up/down wasn't called */ + } + dev_err(dev, "device rebooted: reinitializing driver\n"); + __i2400m_dev_stop(i2400m); + i2400m->updown = 0; + result = __i2400m_dev_start(i2400m, + I2400M_BRI_SOFT | I2400M_BRI_MAC_REINIT); + if (result < 0) { + dev_err(dev, "device reboot: cannot start the device: %d\n", + result); + result = i2400m->bus_reset(i2400m, I2400M_RT_BUS); + if (result >= 0) + result = -ENODEV; + } else + i2400m->updown = 1; +out_unlock: + if (i2400m->reset_ctx) { + ctx->result = result; + complete(&ctx->completion); + } + mutex_unlock(&i2400m->init_mutex); +out: + i2400m_put(i2400m); + kfree(iw); + d_fnend(3, dev, "(ws %p i2400m %p) = void\n", ws, i2400m); + return; +} + + +/** + * i2400m_dev_reset_handle - Handle a device's reset in a thread context + * + * Schedule a device reset handling out on a thread context, so it + * is safe to call from atomic context. We can't use the i2400m's + * queue as we are going to destroy it and reinitialize it as part of + * the driver bringup/bringup process. + * + * See __i2400m_dev_reset_handle() for details; that takes care of + * reinitializing the driver to handle the reset, calling into the + * bus-specific functions ops as needed. + */ +int i2400m_dev_reset_handle(struct i2400m *i2400m) +{ + return i2400m_schedule_work(i2400m, __i2400m_dev_reset_handle, + GFP_ATOMIC); +} +EXPORT_SYMBOL_GPL(i2400m_dev_reset_handle); + + +/** + * i2400m_setup - bus-generic setup function for the i2400m device + * + * @i2400m: device descriptor (bus-specific parts have been initialized) + * + * Returns: 0 if ok, < 0 errno code on error. + * + * Initializes the bus-generic parts of the i2400m driver; the + * bus-specific parts have been initialized, function pointers filled + * out by the bus-specific probe function. + * + * As well, this registers the WiMAX and net device nodes. Once this + * function returns, the device is operative and has to be ready to + * receive and send network traffic and WiMAX control operations. + */ +int i2400m_setup(struct i2400m *i2400m, enum i2400m_bri bm_flags) +{ + int result = -ENODEV; + struct device *dev = i2400m_dev(i2400m); + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + struct net_device *net_dev = i2400m->wimax_dev.net_dev; + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + + snprintf(wimax_dev->name, sizeof(wimax_dev->name), + "i2400m-%s:%s", dev->bus->name, dev->bus_id); + + i2400m->bm_cmd_buf = kzalloc(I2400M_BM_CMD_BUF_SIZE, GFP_KERNEL); + if (i2400m->bm_cmd_buf == NULL) { + dev_err(dev, "cannot allocate USB command buffer\n"); + goto error_bm_cmd_kzalloc; + } + i2400m->bm_ack_buf = kzalloc(I2400M_BM_ACK_BUF_SIZE, GFP_KERNEL); + if (i2400m->bm_ack_buf == NULL) { + dev_err(dev, "cannot allocate USB ack buffer\n"); + goto error_bm_ack_buf_kzalloc; + } + result = i2400m_bootrom_init(i2400m, bm_flags); + if (result < 0) { + dev_err(dev, "read mac addr: bootrom init " + "failed: %d\n", result); + goto error_bootrom_init; + } + result = i2400m_read_mac_addr(i2400m); + if (result < 0) + goto error_read_mac_addr; + + result = register_netdev(net_dev); /* Okey dokey, bring it up */ + if (result < 0) { + dev_err(dev, "cannot register i2400m network device: %d\n", + result); + goto error_register_netdev; + } + netif_carrier_off(net_dev); + + result = i2400m_dev_start(i2400m, bm_flags); + if (result < 0) + goto error_dev_start; + + i2400m->wimax_dev.op_msg_from_user = i2400m_op_msg_from_user; + i2400m->wimax_dev.op_rfkill_sw_toggle = i2400m_op_rfkill_sw_toggle; + i2400m->wimax_dev.op_reset = i2400m_op_reset; + result = wimax_dev_add(&i2400m->wimax_dev, net_dev); + if (result < 0) + goto error_wimax_dev_add; + /* User space needs to do some init stuff */ + wimax_state_change(wimax_dev, WIMAX_ST_UNINITIALIZED); + + /* Now setup all that requires a registered net and wimax device. */ + result = i2400m_debugfs_add(i2400m); + if (result < 0) { + dev_err(dev, "cannot setup i2400m's debugfs: %d\n", result); + goto error_debugfs_setup; + } + d_fnend(3, dev, "(i2400m %p) = %d\n", i2400m, result); + return result; + +error_debugfs_setup: + wimax_dev_rm(&i2400m->wimax_dev); +error_wimax_dev_add: + i2400m_dev_stop(i2400m); +error_dev_start: + unregister_netdev(net_dev); +error_register_netdev: +error_read_mac_addr: +error_bootrom_init: + kfree(i2400m->bm_ack_buf); +error_bm_ack_buf_kzalloc: + kfree(i2400m->bm_cmd_buf); +error_bm_cmd_kzalloc: + d_fnend(3, dev, "(i2400m %p) = %d\n", i2400m, result); + return result; +} +EXPORT_SYMBOL_GPL(i2400m_setup); + + +/** + * i2400m_release - release the bus-generic driver resources + * + * Sends a disconnect message and undoes any setup done by i2400m_setup() + */ +void i2400m_release(struct i2400m *i2400m) +{ + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + netif_stop_queue(i2400m->wimax_dev.net_dev); + + i2400m_debugfs_rm(i2400m); + wimax_dev_rm(&i2400m->wimax_dev); + i2400m_dev_stop(i2400m); + unregister_netdev(i2400m->wimax_dev.net_dev); + kfree(i2400m->bm_ack_buf); + kfree(i2400m->bm_cmd_buf); + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); +} +EXPORT_SYMBOL_GPL(i2400m_release); + + +static +int __init i2400m_driver_init(void) +{ + return 0; +} +module_init(i2400m_driver_init); + +static +void __exit i2400m_driver_exit(void) +{ + /* for scheds i2400m_dev_reset_handle() */ + flush_scheduled_work(); + return; +} +module_exit(i2400m_driver_exit); + +MODULE_AUTHOR("Intel Corporation "); +MODULE_DESCRIPTION("Intel 2400M WiMAX networking bus-generic driver"); +MODULE_LICENSE("GPL"); diff --git a/drivers/net/wimax/i2400m/op-rfkill.c b/drivers/net/wimax/i2400m/op-rfkill.c new file mode 100644 index 00000000000..487ec58cea4 --- /dev/null +++ b/drivers/net/wimax/i2400m/op-rfkill.c @@ -0,0 +1,207 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Implement backend for the WiMAX stack rfkill support + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * The WiMAX kernel stack integrates into RF-Kill and keeps the + * switches's status. We just need to: + * + * - report changes in the HW RF Kill switch [with + * wimax_rfkill_{sw,hw}_report(), which happens when we detect those + * indications coming through hardware reports]. We also do it on + * initialization to let the stack know the intial HW state. + * + * - implement indications from the stack to change the SW RF Kill + * switch (coming from sysfs, the wimax stack or user space). + */ +#include "i2400m.h" +#include + + + +#define D_SUBMODULE rfkill +#include "debug-levels.h" + +/* + * Return true if the i2400m radio is in the requested wimax_rf_state state + * + */ +static +int i2400m_radio_is(struct i2400m *i2400m, enum wimax_rf_state state) +{ + if (state == WIMAX_RF_OFF) + return i2400m->state == I2400M_SS_RF_OFF + || i2400m->state == I2400M_SS_RF_SHUTDOWN; + else if (state == WIMAX_RF_ON) + /* state == WIMAX_RF_ON */ + return i2400m->state != I2400M_SS_RF_OFF + && i2400m->state != I2400M_SS_RF_SHUTDOWN; + else + BUG(); +} + + +/* + * WiMAX stack operation: implement SW RFKill toggling + * + * @wimax_dev: device descriptor + * @skb: skb where the message has been received; skb->data is + * expected to point to the message payload. + * @genl_info: passed by the generic netlink layer + * + * Generic Netlink will call this function when a message is sent from + * userspace to change the software RF-Kill switch status. + * + * This function will set the device's sofware RF-Kill switch state to + * match what is requested. + * + * NOTE: the i2400m has a strict state machine; we can only set the + * RF-Kill switch when it is on, the HW RF-Kill is on and the + * device is initialized. So we ignore errors steaming from not + * being in the right state (-EILSEQ). + */ +int i2400m_op_rfkill_sw_toggle(struct wimax_dev *wimax_dev, + enum wimax_rf_state state) +{ + int result; + struct i2400m *i2400m = wimax_dev_to_i2400m(wimax_dev); + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *ack_skb; + struct { + struct i2400m_l3l4_hdr hdr; + struct i2400m_tlv_rf_operation sw_rf; + } __attribute__((packed)) *cmd; + char strerr[32]; + + d_fnstart(4, dev, "(wimax_dev %p state %d)\n", wimax_dev, state); + + result = -ENOMEM; + cmd = kzalloc(sizeof(*cmd), GFP_KERNEL); + if (cmd == NULL) + goto error_alloc; + cmd->hdr.type = cpu_to_le16(I2400M_MT_CMD_RF_CONTROL); + cmd->hdr.length = sizeof(cmd->sw_rf); + cmd->hdr.version = cpu_to_le16(I2400M_L3L4_VERSION); + cmd->sw_rf.hdr.type = cpu_to_le16(I2400M_TLV_RF_OPERATION); + cmd->sw_rf.hdr.length = cpu_to_le16(sizeof(cmd->sw_rf.status)); + switch (state) { + case WIMAX_RF_OFF: /* RFKILL ON, radio OFF */ + cmd->sw_rf.status = cpu_to_le32(2); + break; + case WIMAX_RF_ON: /* RFKILL OFF, radio ON */ + cmd->sw_rf.status = cpu_to_le32(1); + break; + default: + BUG(); + } + + ack_skb = i2400m_msg_to_dev(i2400m, cmd, sizeof(*cmd)); + result = PTR_ERR(ack_skb); + if (IS_ERR(ack_skb)) { + dev_err(dev, "Failed to issue 'RF Control' command: %d\n", + result); + goto error_msg_to_dev; + } + result = i2400m_msg_check_status(wimax_msg_data(ack_skb), + strerr, sizeof(strerr)); + if (result < 0) { + dev_err(dev, "'RF Control' (0x%04x) command failed: %d - %s\n", + I2400M_MT_CMD_RF_CONTROL, result, strerr); + goto error_cmd; + } + + /* Now we wait for the state to change to RADIO_OFF or RADIO_ON */ + result = wait_event_timeout( + i2400m->state_wq, i2400m_radio_is(i2400m, state), + 5 * HZ); + if (result == 0) + result = -ETIMEDOUT; + if (result < 0) + dev_err(dev, "Error waiting for device to toggle RF state: " + "%d\n", result); + result = 0; +error_cmd: + kfree_skb(ack_skb); +error_msg_to_dev: +error_alloc: + d_fnend(4, dev, "(wimax_dev %p state %d) = %d\n", + wimax_dev, state, result); + return result; +} + + +/* + * Inform the WiMAX stack of changes in the RF Kill switches reported + * by the device + * + * @i2400m: device descriptor + * @rfss: TLV for RF Switches status; already validated + * + * NOTE: the reports on RF switch status cannot be trusted + * or used until the device is in a state of RADIO_OFF + * or greater. + */ +void i2400m_report_tlv_rf_switches_status( + struct i2400m *i2400m, + const struct i2400m_tlv_rf_switches_status *rfss) +{ + struct device *dev = i2400m_dev(i2400m); + enum i2400m_rf_switch_status hw, sw; + enum wimax_st wimax_state; + + sw = le32_to_cpu(rfss->sw_rf_switch); + hw = le32_to_cpu(rfss->hw_rf_switch); + + d_fnstart(3, dev, "(i2400m %p rfss %p [hw %u sw %u])\n", + i2400m, rfss, hw, sw); + /* We only process rw switch evens when the device has been + * fully initialized */ + wimax_state = wimax_state_get(&i2400m->wimax_dev); + if (wimax_state < WIMAX_ST_RADIO_OFF) { + d_printf(3, dev, "ignoring RF switches report, state %u\n", + wimax_state); + goto out; + } + switch (sw) { + case I2400M_RF_SWITCH_ON: /* RF Kill disabled (radio on) */ + wimax_report_rfkill_sw(&i2400m->wimax_dev, WIMAX_RF_ON); + break; + case I2400M_RF_SWITCH_OFF: /* RF Kill enabled (radio off) */ + wimax_report_rfkill_sw(&i2400m->wimax_dev, WIMAX_RF_OFF); + break; + default: + dev_err(dev, "HW BUG? Unknown RF SW state 0x%x\n", sw); + } + + switch (hw) { + case I2400M_RF_SWITCH_ON: /* RF Kill disabled (radio on) */ + wimax_report_rfkill_hw(&i2400m->wimax_dev, WIMAX_RF_ON); + break; + case I2400M_RF_SWITCH_OFF: /* RF Kill enabled (radio off) */ + wimax_report_rfkill_hw(&i2400m->wimax_dev, WIMAX_RF_OFF); + break; + default: + dev_err(dev, "HW BUG? Unknown RF HW state 0x%x\n", hw); + } +out: + d_fnend(3, dev, "(i2400m %p rfss %p [hw %u sw %u]) = void\n", + i2400m, rfss, hw, sw); +} From ce6cde92803e961d95ddacdf74bd8b067f82f7d4 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:45 -0800 Subject: [PATCH 108/123] i2400m: linkage to the networking stack Implementation of the glue to the network stack so the WiMAX device shows up as an Ethernet device. Initially we shot for implementing a Pure IP device -- however, the world seems to turn around Ethernet devices. Main issues were with the ISC DHCP client and servers (as they don't understand types other than Ethernet and Token Ring). We proceeded to register with IANA the PureIP hw type, so that DHCP requests could declare such. We also created patches to the main ISC DHCP versions to support it. However, until all that permeates into deployments, there is going to be a long time. So we moved back to wrap Ethernet frames around the PureIP device. At the time being this has overhead; we need to reallocate with space for an Ethernet header. The reason is the device-to-host protocol coalesces many network packets into a single message, so we can't introduce Ethernet headers without overwriting valid data from other packets. Coming-soon versions of the firmware have this issue solved. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/netdev.c | 524 ++++++++++++++++++++++++++++++ 1 file changed, 524 insertions(+) create mode 100644 drivers/net/wimax/i2400m/netdev.c diff --git a/drivers/net/wimax/i2400m/netdev.c b/drivers/net/wimax/i2400m/netdev.c new file mode 100644 index 00000000000..63fe708e8a3 --- /dev/null +++ b/drivers/net/wimax/i2400m/netdev.c @@ -0,0 +1,524 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Glue with the networking stack + * + * + * Copyright (C) 2007 Intel Corporation + * Yanir Lubetkin + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * This implements an ethernet device for the i2400m. + * + * We fake being an ethernet device to simplify the support from user + * space and from the other side. The world is (sadly) configured to + * take in only Ethernet devices... + * + * Because of this, currently there is an copy-each-rxed-packet + * overhead on the RX path. Each IP packet has to be reallocated to + * add an ethernet header (as there is no space in what we get from + * the device). This is a known drawback and coming versions of the + * device's firmware are being changed to add header space that can be + * used to insert the ethernet header without having to reallocate and + * copy. + * + * TX error handling is tricky; because we have to FIFO/queue the + * buffers for transmission (as the hardware likes it aggregated), we + * just give the skb to the TX subsystem and by the time it is + * transmitted, we have long forgotten about it. So we just don't care + * too much about it. + * + * Note that when the device is in idle mode with the basestation, we + * need to negotiate coming back up online. That involves negotiation + * and possible user space interaction. Thus, we defer to a workqueue + * to do all that. By default, we only queue a single packet and drop + * the rest, as potentially the time to go back from idle to normal is + * long. + * + * ROADMAP + * + * i2400m_open Called on ifconfig up + * i2400m_stop Called on ifconfig down + * + * i2400m_hard_start_xmit Called by the network stack to send a packet + * i2400m_net_wake_tx Wake up device from basestation-IDLE & TX + * i2400m_wake_tx_work + * i2400m_cmd_exit_idle + * i2400m_tx + * i2400m_net_tx TX a data frame + * i2400m_tx + * + * i2400m_change_mtu Called on ifconfig mtu XXX + * + * i2400m_tx_timeout Called when the device times out + * + * i2400m_net_rx Called by the RX code when a data frame is + * available. + * i2400m_netdev_setup Called to setup all the netdev stuff from + * alloc_netdev. + */ +#include +#include +#include "i2400m.h" + + +#define D_SUBMODULE netdev +#include "debug-levels.h" + +enum { +/* netdev interface */ + /* + * Out of NWG spec (R1_v1.2.2), 3.3.3 ASN Bearer Plane MTU Size + * + * The MTU is 1400 or less + */ + I2400M_MAX_MTU = 1400, + I2400M_TX_TIMEOUT = HZ, + I2400M_TX_QLEN = 5, +}; + + +static +int i2400m_open(struct net_device *net_dev) +{ + int result; + struct i2400m *i2400m = net_dev_to_i2400m(net_dev); + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(3, dev, "(net_dev %p [i2400m %p])\n", net_dev, i2400m); + if (i2400m->ready == 0) { + dev_err(dev, "Device is still initializing\n"); + result = -EBUSY; + } else + result = 0; + d_fnend(3, dev, "(net_dev %p [i2400m %p]) = %d\n", + net_dev, i2400m, result); + return result; +} + + +/* + * + * On kernel versions where cancel_work_sync() didn't return anything, + * we rely on wake_tx_skb() being non-NULL. + */ +static +int i2400m_stop(struct net_device *net_dev) +{ + struct i2400m *i2400m = net_dev_to_i2400m(net_dev); + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(3, dev, "(net_dev %p [i2400m %p])\n", net_dev, i2400m); + /* See i2400m_hard_start_xmit(), references are taken there + * and here we release them if the work was still + * pending. Note we can't differentiate work not pending vs + * never scheduled, so the NULL check does that. */ + if (cancel_work_sync(&i2400m->wake_tx_ws) == 0 + && i2400m->wake_tx_skb != NULL) { + unsigned long flags; + struct sk_buff *wake_tx_skb; + spin_lock_irqsave(&i2400m->tx_lock, flags); + wake_tx_skb = i2400m->wake_tx_skb; /* compat help */ + i2400m->wake_tx_skb = NULL; /* compat help */ + spin_unlock_irqrestore(&i2400m->tx_lock, flags); + i2400m_put(i2400m); + kfree_skb(wake_tx_skb); + } + d_fnend(3, dev, "(net_dev %p [i2400m %p]) = 0\n", net_dev, i2400m); + return 0; +} + + +/* + * Wake up the device and transmit a held SKB, then restart the net queue + * + * When the device goes into basestation-idle mode, we need to tell it + * to exit that mode; it will negotiate with the base station, user + * space may have to intervene to rehandshake crypto and then tell us + * when it is ready to transmit the packet we have "queued". Still we + * need to give it sometime after it reports being ok. + * + * On error, there is not much we can do. If the error was on TX, we + * still wake the queue up to see if the next packet will be luckier. + * + * If _cmd_exit_idle() fails...well, it could be many things; most + * commonly it is that something else took the device out of IDLE mode + * (for example, the base station). In that case we get an -EILSEQ and + * we are just going to ignore that one. If the device is back to + * connected, then fine -- if it is someother state, the packet will + * be dropped anyway. + */ +void i2400m_wake_tx_work(struct work_struct *ws) +{ + int result; + struct i2400m *i2400m = container_of(ws, struct i2400m, wake_tx_ws); + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *skb = i2400m->wake_tx_skb; + unsigned long flags; + + spin_lock_irqsave(&i2400m->tx_lock, flags); + skb = i2400m->wake_tx_skb; + i2400m->wake_tx_skb = NULL; + spin_unlock_irqrestore(&i2400m->tx_lock, flags); + + d_fnstart(3, dev, "(ws %p i2400m %p skb %p)\n", ws, i2400m, skb); + result = -EINVAL; + if (skb == NULL) { + dev_err(dev, "WAKE&TX: skb dissapeared!\n"); + goto out_put; + } + result = i2400m_cmd_exit_idle(i2400m); + if (result == -EILSEQ) + result = 0; + if (result < 0) { + dev_err(dev, "WAKE&TX: device didn't get out of idle: " + "%d\n", result); + goto error; + } + result = wait_event_timeout(i2400m->state_wq, + i2400m->state != I2400M_SS_IDLE, 5 * HZ); + if (result == 0) + result = -ETIMEDOUT; + if (result < 0) { + dev_err(dev, "WAKE&TX: error waiting for device to exit IDLE: " + "%d\n", result); + goto error; + } + msleep(20); /* device still needs some time or it drops it */ + result = i2400m_tx(i2400m, skb->data, skb->len, I2400M_PT_DATA); + netif_wake_queue(i2400m->wimax_dev.net_dev); +error: + kfree_skb(skb); /* refcount transferred by _hard_start_xmit() */ +out_put: + i2400m_put(i2400m); + d_fnend(3, dev, "(ws %p i2400m %p skb %p) = void [%d]\n", + ws, i2400m, skb, result); +} + + +/* + * Prepare the data payload TX header + * + * The i2400m expects a 4 byte header in front of a data packet. + * + * Because we pretend to be an ethernet device, this packet comes with + * an ethernet header. Pull it and push our header. + */ +static +void i2400m_tx_prep_header(struct sk_buff *skb) +{ + struct i2400m_pl_data_hdr *pl_hdr; + skb_pull(skb, ETH_HLEN); + pl_hdr = (struct i2400m_pl_data_hdr *) skb_push(skb, sizeof(*pl_hdr)); + pl_hdr->reserved = 0; +} + + +/* + * TX an skb to an idle device + * + * When the device is in basestation-idle mode, we need to wake it up + * and then TX. So we queue a work_struct for doing so. + * + * We need to get an extra ref for the skb (so it is not dropped), as + * well as be careful not to queue more than one request (won't help + * at all). If more than one request comes or there are errors, we + * just drop the packets (see i2400m_hard_start_xmit()). + */ +static +int i2400m_net_wake_tx(struct i2400m *i2400m, struct net_device *net_dev, + struct sk_buff *skb) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + unsigned long flags; + + d_fnstart(3, dev, "(skb %p net_dev %p)\n", skb, net_dev); + if (net_ratelimit()) { + d_printf(3, dev, "WAKE&NETTX: " + "skb %p sending %d bytes to radio\n", + skb, skb->len); + d_dump(4, dev, skb->data, skb->len); + } + /* We hold a ref count for i2400m and skb, so when + * stopping() the device, we need to cancel that work + * and if pending, release those resources. */ + result = 0; + spin_lock_irqsave(&i2400m->tx_lock, flags); + if (!work_pending(&i2400m->wake_tx_ws)) { + netif_stop_queue(net_dev); + i2400m_get(i2400m); + i2400m->wake_tx_skb = skb_get(skb); /* transfer ref count */ + i2400m_tx_prep_header(skb); + result = schedule_work(&i2400m->wake_tx_ws); + WARN_ON(result == 0); + } + spin_unlock_irqrestore(&i2400m->tx_lock, flags); + if (result == 0) { + /* Yes, this happens even if we stopped the + * queue -- blame the queue disciplines that + * queue without looking -- I guess there is a reason + * for that. */ + if (net_ratelimit()) + d_printf(1, dev, "NETTX: device exiting idle, " + "dropping skb %p, queue running %d\n", + skb, netif_queue_stopped(net_dev)); + result = -EBUSY; + } + d_fnend(3, dev, "(skb %p net_dev %p) = %d\n", skb, net_dev, result); + return result; +} + + +/* + * Transmit a packet to the base station on behalf of the network stack. + * + * Returns: 0 if ok, < 0 errno code on error. + * + * We need to pull the ethernet header and add the hardware header, + * which is currently set to all zeroes and reserved. + */ +static +int i2400m_net_tx(struct i2400m *i2400m, struct net_device *net_dev, + struct sk_buff *skb) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(3, dev, "(i2400m %p net_dev %p skb %p)\n", + i2400m, net_dev, skb); + /* FIXME: check eth hdr, only IPv4 is routed by the device as of now */ + net_dev->trans_start = jiffies; + i2400m_tx_prep_header(skb); + d_printf(3, dev, "NETTX: skb %p sending %d bytes to radio\n", + skb, skb->len); + d_dump(4, dev, skb->data, skb->len); + result = i2400m_tx(i2400m, skb->data, skb->len, I2400M_PT_DATA); + d_fnend(3, dev, "(i2400m %p net_dev %p skb %p) = %d\n", + i2400m, net_dev, skb, result); + return result; +} + + +/* + * Transmit a packet to the base station on behalf of the network stack + * + * + * Returns: NETDEV_TX_OK (always, even in case of error) + * + * In case of error, we just drop it. Reasons: + * + * - we add a hw header to each skb, and if the network stack + * retries, we have no way to know if that skb has it or not. + * + * - network protocols have their own drop-recovery mechanisms + * + * - there is not much else we can do + * + * If the device is idle, we need to wake it up; that is an operation + * that will sleep. See i2400m_net_wake_tx() for details. + */ +static +int i2400m_hard_start_xmit(struct sk_buff *skb, + struct net_device *net_dev) +{ + int result; + struct i2400m *i2400m = net_dev_to_i2400m(net_dev); + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(3, dev, "(skb %p net_dev %p)\n", skb, net_dev); + if (i2400m->state == I2400M_SS_IDLE) + result = i2400m_net_wake_tx(i2400m, net_dev, skb); + else + result = i2400m_net_tx(i2400m, net_dev, skb); + if (result < 0) + net_dev->stats.tx_dropped++; + else { + net_dev->stats.tx_packets++; + net_dev->stats.tx_bytes += skb->len; + } + kfree_skb(skb); + result = NETDEV_TX_OK; + d_fnend(3, dev, "(skb %p net_dev %p) = %d\n", skb, net_dev, result); + return result; +} + + +static +int i2400m_change_mtu(struct net_device *net_dev, int new_mtu) +{ + int result; + struct i2400m *i2400m = net_dev_to_i2400m(net_dev); + struct device *dev = i2400m_dev(i2400m); + + if (new_mtu >= I2400M_MAX_MTU) { + dev_err(dev, "Cannot change MTU to %d (max is %d)\n", + new_mtu, I2400M_MAX_MTU); + result = -EINVAL; + } else { + net_dev->mtu = new_mtu; + result = 0; + } + return result; +} + + +static +void i2400m_tx_timeout(struct net_device *net_dev) +{ + /* + * We might want to kick the device + * + * There is not much we can do though, as the device requires + * that we send the data aggregated. By the time we receive + * this, there might be data pending to be sent or not... + */ + net_dev->stats.tx_errors++; + return; +} + + +/* + * Create a fake ethernet header + * + * For emulating an ethernet device, every received IP header has to + * be prefixed with an ethernet header. + * + * What we receive has (potentially) many IP packets concatenated with + * no ETH_HLEN bytes prefixed. Thus there is no space for an eth + * header. + * + * We would have to reallocate or do ugly fragment tricks in order to + * add it. + * + * But what we do is use the header space of the RX transaction + * (*msg_hdr) as we don't need it anymore; then we'll point all the + * data skbs there, as they share the same backing store. + * + * We only support IPv4 for v3 firmware. + */ +static +void i2400m_rx_fake_eth_header(struct net_device *net_dev, + void *_eth_hdr) +{ + struct ethhdr *eth_hdr = _eth_hdr; + + memcpy(eth_hdr->h_dest, net_dev->dev_addr, sizeof(eth_hdr->h_dest)); + memset(eth_hdr->h_source, 0, sizeof(eth_hdr->h_dest)); + eth_hdr->h_proto = __constant_cpu_to_be16(ETH_P_IP); +} + + +/* + * i2400m_net_rx - pass a network packet to the stack + * + * @i2400m: device instance + * @skb_rx: the skb where the buffer pointed to by @buf is + * @i: 1 if payload is the only one + * @buf: pointer to the buffer containing the data + * @len: buffer's length + * + * We just clone the skb and set it up so that it's skb->data pointer + * points to "buf" and it's length. + * + * Note that if the payload is the last (or the only one) in a + * multi-payload message, we don't clone the SKB but just reuse it. + * + * This function is normally run from a thread context. However, we + * still use netif_rx() instead of netif_receive_skb() as was + * recommended in the mailing list. Reason is in some stress tests + * when sending/receiving a lot of data we seem to hit a softlock in + * the kernel's TCP implementation [aroudn tcp_delay_timer()]. Using + * netif_rx() took care of the issue. + * + * This is, of course, still open to do more research on why running + * with netif_receive_skb() hits this softlock. FIXME. + * + * FIXME: currently we don't do any efforts at distinguishing if what + * we got was an IPv4 or IPv6 header, to setup the protocol field + * correctly. + */ +void i2400m_net_rx(struct i2400m *i2400m, struct sk_buff *skb_rx, + unsigned i, const void *buf, int buf_len) +{ + struct net_device *net_dev = i2400m->wimax_dev.net_dev; + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *skb; + + d_fnstart(2, dev, "(i2400m %p buf %p buf_len %d)\n", + i2400m, buf, buf_len); + if (i) { + skb = skb_get(skb_rx); + d_printf(2, dev, "RX: reusing first payload skb %p\n", skb); + skb_pull(skb, buf - (void *) skb->data); + skb_trim(skb, (void *) skb_end_pointer(skb) - buf); + } else { + /* Yes, this is bad -- a lot of overhead -- see + * comments at the top of the file */ + skb = __netdev_alloc_skb(net_dev, buf_len, GFP_KERNEL); + if (skb == NULL) { + dev_err(dev, "NETRX: no memory to realloc skb\n"); + net_dev->stats.rx_dropped++; + goto error_skb_realloc; + } + memcpy(skb_put(skb, buf_len), buf, buf_len); + } + i2400m_rx_fake_eth_header(i2400m->wimax_dev.net_dev, + skb->data - ETH_HLEN); + skb_set_mac_header(skb, -ETH_HLEN); + skb->dev = i2400m->wimax_dev.net_dev; + skb->protocol = htons(ETH_P_IP); + net_dev->stats.rx_packets++; + net_dev->stats.rx_bytes += buf_len; + d_printf(3, dev, "NETRX: receiving %d bytes to network stack\n", + buf_len); + d_dump(4, dev, buf, buf_len); + netif_rx_ni(skb); /* see notes in function header */ +error_skb_realloc: + d_fnend(2, dev, "(i2400m %p buf %p buf_len %d) = void\n", + i2400m, buf, buf_len); +} + + +/** + * i2400m_netdev_setup - Setup setup @net_dev's i2400m private data + * + * Called by alloc_netdev() + */ +void i2400m_netdev_setup(struct net_device *net_dev) +{ + d_fnstart(3, NULL, "(net_dev %p)\n", net_dev); + ether_setup(net_dev); + net_dev->mtu = I2400M_MAX_MTU; + net_dev->tx_queue_len = I2400M_TX_QLEN; + net_dev->features = + NETIF_F_VLAN_CHALLENGED + | NETIF_F_HIGHDMA; + net_dev->flags = + IFF_NOARP /* i2400m is apure IP device */ + & (~IFF_BROADCAST /* i2400m is P2P */ + & ~IFF_MULTICAST); + net_dev->watchdog_timeo = I2400M_TX_TIMEOUT; + net_dev->open = i2400m_open; + net_dev->stop = i2400m_stop; + net_dev->hard_start_xmit = i2400m_hard_start_xmit; + net_dev->change_mtu = i2400m_change_mtu; + net_dev->tx_timeout = i2400m_tx_timeout; + d_fnend(3, NULL, "(net_dev %p) = void\n", net_dev); +} +EXPORT_SYMBOL_GPL(i2400m_netdev_setup); + From 467cc396fb4665957bc7d182c96e45a4d7c575e4 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:46 -0800 Subject: [PATCH 109/123] i2400m: firmware loading and bootrom initialization Implements the firmware loader (using the bus-specific driver's backends for the actual upload). The most critical thing in here is the piece that puts the device in boot-mode from any other (undetermined) state, otherwise, it is just pushing the bytes from the firmware file to the device. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/fw.c | 1095 +++++++++++++++++++++++++++++++++ 1 file changed, 1095 insertions(+) create mode 100644 drivers/net/wimax/i2400m/fw.c diff --git a/drivers/net/wimax/i2400m/fw.c b/drivers/net/wimax/i2400m/fw.c new file mode 100644 index 00000000000..1d8271f34c3 --- /dev/null +++ b/drivers/net/wimax/i2400m/fw.c @@ -0,0 +1,1095 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Firmware uploader + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Yanir Lubetkin + * Inaky Perez-Gonzalez + * - Initial implementation + * + * + * THE PROCEDURE + * + * (this is decribed for USB, but for SDIO is similar) + * + * The 2400m works in two modes: boot-mode or normal mode. In boot + * mode we can execute only a handful of commands targeted at + * uploading the firmware and launching it. + * + * The 2400m enters boot mode when it is first connected to the + * system, when it crashes and when you ask it to reboot. There are + * two submodes of the boot mode: signed and non-signed. Signed takes + * firmwares signed with a certain private key, non-signed takes any + * firmware. Normal hardware takes only signed firmware. + * + * Upon entrance to boot mode, the device sends a few zero length + * packets (ZLPs) on the notification endpoint, then a reboot barker + * (4 le32 words with value I2400M_{S,N}BOOT_BARKER). We ack it by + * sending the same barker on the bulk out endpoint. The device acks + * with a reboot ack barker (4 le32 words with value 0xfeedbabe) and + * then the device is fully rebooted. At this point we can upload the + * firmware. + * + * This process is accomplished by the i2400m_bootrom_init() + * function. All the device interaction happens through the + * i2400m_bm_cmd() [boot mode command]. Special return values will + * indicate if the device resets. + * + * After this, we read the MAC address and then (if needed) + * reinitialize the device. We need to read it ahead of time because + * in the future, we might not upload the firmware until userspace + * 'ifconfig up's the device. + * + * We can then upload the firmware file. The file is composed of a BCF + * header (basic data, keys and signatures) and a list of write + * commands and payloads. We first upload the header + * [i2400m_dnload_init()] and then pass the commands and payloads + * verbatim to the i2400m_bm_cmd() function + * [i2400m_dnload_bcf()]. Then we tell the device to jump to the new + * firmware [i2400m_dnload_finalize()]. + * + * Once firmware is uploaded, we are good to go :) + * + * When we don't know in which mode we are, we first try by sending a + * warm reset request that will take us to boot-mode. If we time out + * waiting for a reboot barker, that means maybe we are already in + * boot mode, so we send a reboot barker. + * + * COMMAND EXECUTION + * + * This code (and process) is single threaded; for executing commands, + * we post a URB to the notification endpoint, post the command, wait + * for data on the notification buffer. We don't need to worry about + * others as we know we are the only ones in there. + * + * BACKEND IMPLEMENTATION + * + * This code is bus-generic; the bus-specific driver provides back end + * implementations to send a boot mode command to the device and to + * read an acknolwedgement from it (or an asynchronous notification) + * from it. + * + * ROADMAP + * + * i2400m_dev_bootstrap Called by __i2400m_dev_start() + * request_firmware + * i2400m_fw_check + * i2400m_fw_dnload + * release_firmware + * + * i2400m_fw_dnload + * i2400m_bootrom_init + * i2400m_bm_cmd + * i2400m->bus_reset + * i2400m_dnload_init + * i2400m_dnload_init_signed + * i2400m_dnload_init_nonsigned + * i2400m_download_chunk + * i2400m_bm_cmd + * i2400m_dnload_bcf + * i2400m_bm_cmd + * i2400m_dnload_finalize + * i2400m_bm_cmd + * + * i2400m_bm_cmd + * i2400m->bus_bm_cmd_send() + * i2400m->bus_bm_wait_for_ack + * __i2400m_bm_ack_verify + * + * i2400m_bm_cmd_prepare Used by bus-drivers to prep + * commands before sending + */ +#include +#include +#include +#include "i2400m.h" + + +#define D_SUBMODULE fw +#include "debug-levels.h" + + +static const __le32 i2400m_ACK_BARKER[4] = { + __constant_cpu_to_le32(I2400M_ACK_BARKER), + __constant_cpu_to_le32(I2400M_ACK_BARKER), + __constant_cpu_to_le32(I2400M_ACK_BARKER), + __constant_cpu_to_le32(I2400M_ACK_BARKER) +}; + + +/** + * Prepare a boot-mode command for delivery + * + * @cmd: pointer to bootrom header to prepare + * + * Computes checksum if so needed. After calling this function, DO NOT + * modify the command or header as the checksum won't work anymore. + * + * We do it from here because some times we cannot do it in the + * original context the command was sent (it is a const), so when we + * copy it to our staging buffer, we add the checksum there. + */ +void i2400m_bm_cmd_prepare(struct i2400m_bootrom_header *cmd) +{ + if (i2400m_brh_get_use_checksum(cmd)) { + int i; + u32 checksum = 0; + const u32 *checksum_ptr = (void *) cmd->payload; + for (i = 0; i < cmd->data_size / 4; i++) + checksum += cpu_to_le32(*checksum_ptr++); + checksum += cmd->command + cmd->target_addr + cmd->data_size; + cmd->block_checksum = cpu_to_le32(checksum); + } +} +EXPORT_SYMBOL_GPL(i2400m_bm_cmd_prepare); + + +/* + * Verify the ack data received + * + * Given a reply to a boot mode command, chew it and verify everything + * is ok. + * + * @opcode: opcode which generated this ack. For error messages. + * @ack: pointer to ack data we received + * @ack_size: size of that data buffer + * @flags: I2400M_BM_CMD_* flags we called the command with. + * + * Way too long function -- maybe it should be further split + */ +static +ssize_t __i2400m_bm_ack_verify(struct i2400m *i2400m, int opcode, + struct i2400m_bootrom_header *ack, + size_t ack_size, int flags) +{ + ssize_t result = -ENOMEM; + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(8, dev, "(i2400m %p opcode %d ack %p size %zu)\n", + i2400m, opcode, ack, ack_size); + if (ack_size < sizeof(*ack)) { + result = -EIO; + dev_err(dev, "boot-mode cmd %d: HW BUG? notification didn't " + "return enough data (%zu bytes vs %zu expected)\n", + opcode, ack_size, sizeof(*ack)); + goto error_ack_short; + } + if (ack_size == sizeof(i2400m_NBOOT_BARKER) + && memcmp(ack, i2400m_NBOOT_BARKER, sizeof(*ack)) == 0) { + result = -ERESTARTSYS; + i2400m->sboot = 0; + d_printf(6, dev, "boot-mode cmd %d: " + "HW non-signed boot barker\n", opcode); + goto error_reboot; + } + if (ack_size == sizeof(i2400m_SBOOT_BARKER) + && memcmp(ack, i2400m_SBOOT_BARKER, sizeof(*ack)) == 0) { + result = -ERESTARTSYS; + i2400m->sboot = 1; + d_printf(6, dev, "boot-mode cmd %d: HW signed reboot barker\n", + opcode); + goto error_reboot; + } + if (ack_size == sizeof(i2400m_ACK_BARKER) + && memcmp(ack, i2400m_ACK_BARKER, sizeof(*ack)) == 0) { + result = -EISCONN; + d_printf(3, dev, "boot-mode cmd %d: HW reboot ack barker\n", + opcode); + goto error_reboot_ack; + } + result = 0; + if (flags & I2400M_BM_CMD_RAW) + goto out_raw; + ack->data_size = le32_to_cpu(ack->data_size); + ack->target_addr = le32_to_cpu(ack->target_addr); + ack->block_checksum = le32_to_cpu(ack->block_checksum); + d_printf(5, dev, "boot-mode cmd %d: notification for opcode %u " + "response %u csum %u rr %u da %u\n", + opcode, i2400m_brh_get_opcode(ack), + i2400m_brh_get_response(ack), + i2400m_brh_get_use_checksum(ack), + i2400m_brh_get_response_required(ack), + i2400m_brh_get_direct_access(ack)); + result = -EIO; + if (i2400m_brh_get_signature(ack) != 0xcbbc) { + dev_err(dev, "boot-mode cmd %d: HW BUG? wrong signature " + "0x%04x\n", opcode, i2400m_brh_get_signature(ack)); + goto error_ack_signature; + } + if (opcode != -1 && opcode != i2400m_brh_get_opcode(ack)) { + dev_err(dev, "boot-mode cmd %d: HW BUG? " + "received response for opcode %u, expected %u\n", + opcode, i2400m_brh_get_opcode(ack), opcode); + goto error_ack_opcode; + } + if (i2400m_brh_get_response(ack) != 0) { /* failed? */ + dev_err(dev, "boot-mode cmd %d: error; hw response %u\n", + opcode, i2400m_brh_get_response(ack)); + goto error_ack_failed; + } + if (ack_size < ack->data_size + sizeof(*ack)) { + dev_err(dev, "boot-mode cmd %d: SW BUG " + "driver provided only %zu bytes for %zu bytes " + "of data\n", opcode, ack_size, + (size_t) le32_to_cpu(ack->data_size) + sizeof(*ack)); + goto error_ack_short_buffer; + } + result = ack_size; + /* Don't you love this stack of empty targets? Well, I don't + * either, but it helps track exactly who comes in here and + * why :) */ +error_ack_short_buffer: +error_ack_failed: +error_ack_opcode: +error_ack_signature: +out_raw: +error_reboot_ack: +error_reboot: +error_ack_short: + d_fnend(8, dev, "(i2400m %p opcode %d ack %p size %zu) = %d\n", + i2400m, opcode, ack, ack_size, (int) result); + return result; +} + + +/** + * i2400m_bm_cmd - Execute a boot mode command + * + * @cmd: buffer containing the command data (pointing at the header). + * This data can be ANYWHERE (for USB, we will copy it to an + * specific buffer). Make sure everything is in proper little + * endian. + * + * A raw buffer can be also sent, just cast it and set flags to + * I2400M_BM_CMD_RAW. + * + * This function will generate a checksum for you if the + * checksum bit in the command is set (unless I2400M_BM_CMD_RAW + * is set). + * + * You can use the i2400m->bm_cmd_buf to stage your commands and + * send them. + * + * If NULL, no command is sent (we just wait for an ack). + * + * @cmd_size: size of the command. Will be auto padded to the + * bus-specific drivers padding requirements. + * + * @ack: buffer where to place the acknowledgement. If it is a regular + * command response, all fields will be returned with the right, + * native endianess. + * + * You *cannot* use i2400m->bm_ack_buf for this buffer. + * + * @ack_size: size of @ack, 16 aligned; you need to provide at least + * sizeof(*ack) bytes and then enough to contain the return data + * from the command + * + * @flags: see I2400M_BM_CMD_* above. + * + * @returns: bytes received by the notification; if < 0, an errno code + * denoting an error or: + * + * -ERESTARTSYS The device has rebooted + * + * Executes a boot-mode command and waits for a response, doing basic + * validation on it; if a zero length response is received, it retries + * waiting for a response until a non-zero one is received (timing out + * after %I2400M_BOOT_RETRIES retries). + */ +static +ssize_t i2400m_bm_cmd(struct i2400m *i2400m, + const struct i2400m_bootrom_header *cmd, size_t cmd_size, + struct i2400m_bootrom_header *ack, size_t ack_size, + int flags) +{ + ssize_t result = -ENOMEM, rx_bytes; + struct device *dev = i2400m_dev(i2400m); + int opcode = cmd == NULL ? -1 : i2400m_brh_get_opcode(cmd); + + d_fnstart(6, dev, "(i2400m %p cmd %p size %zu ack %p size %zu)\n", + i2400m, cmd, cmd_size, ack, ack_size); + BUG_ON(ack_size < sizeof(*ack)); + BUG_ON(i2400m->boot_mode == 0); + + if (cmd != NULL) { /* send the command */ + memcpy(i2400m->bm_cmd_buf, cmd, cmd_size); + result = i2400m->bus_bm_cmd_send(i2400m, cmd, cmd_size, flags); + if (result < 0) + goto error_cmd_send; + if ((flags & I2400M_BM_CMD_RAW) == 0) + d_printf(5, dev, + "boot-mode cmd %d csum %u rr %u da %u: " + "addr 0x%04x size %u block csum 0x%04x\n", + opcode, i2400m_brh_get_use_checksum(cmd), + i2400m_brh_get_response_required(cmd), + i2400m_brh_get_direct_access(cmd), + cmd->target_addr, cmd->data_size, + cmd->block_checksum); + } + result = i2400m->bus_bm_wait_for_ack(i2400m, ack, ack_size); + if (result < 0) { + dev_err(dev, "boot-mode cmd %d: error waiting for an ack: %d\n", + opcode, (int) result); /* bah, %zd doesn't work */ + goto error_wait_for_ack; + } + rx_bytes = result; + /* verify the ack and read more if neccessary [result is the + * final amount of bytes we get in the ack] */ + result = __i2400m_bm_ack_verify(i2400m, opcode, ack, ack_size, flags); + if (result < 0) + goto error_bad_ack; + /* Don't you love this stack of empty targets? Well, I don't + * either, but it helps track exactly who comes in here and + * why :) */ + result = rx_bytes; +error_bad_ack: +error_wait_for_ack: +error_cmd_send: + d_fnend(6, dev, "(i2400m %p cmd %p size %zu ack %p size %zu) = %d\n", + i2400m, cmd, cmd_size, ack, ack_size, (int) result); + return result; +} + + +/** + * i2400m_download_chunk - write a single chunk of data to the device's memory + * + * @i2400m: device descriptor + * @buf: the buffer to write + * @buf_len: length of the buffer to write + * @addr: address in the device memory space + * @direct: bootrom write mode + * @do_csum: should a checksum validation be performed + */ +static int i2400m_download_chunk(struct i2400m *i2400m, const void *chunk, + size_t __chunk_len, unsigned long addr, + unsigned int direct, unsigned int do_csum) +{ + int ret; + size_t chunk_len = ALIGN(__chunk_len, I2400M_PL_PAD); + struct device *dev = i2400m_dev(i2400m); + struct { + struct i2400m_bootrom_header cmd; + u8 cmd_payload[chunk_len]; + } __attribute__((packed)) *buf; + struct i2400m_bootrom_header ack; + + d_fnstart(5, dev, "(i2400m %p chunk %p __chunk_len %zu addr 0x%08lx " + "direct %u do_csum %u)\n", i2400m, chunk, __chunk_len, + addr, direct, do_csum); + buf = i2400m->bm_cmd_buf; + memcpy(buf->cmd_payload, chunk, __chunk_len); + memset(buf->cmd_payload + __chunk_len, 0xad, chunk_len - __chunk_len); + + buf->cmd.command = i2400m_brh_command(I2400M_BRH_WRITE, + __chunk_len & 0x3 ? 0 : do_csum, + __chunk_len & 0xf ? 0 : direct); + buf->cmd.target_addr = cpu_to_le32(addr); + buf->cmd.data_size = cpu_to_le32(__chunk_len); + ret = i2400m_bm_cmd(i2400m, &buf->cmd, sizeof(buf->cmd) + chunk_len, + &ack, sizeof(ack), 0); + if (ret >= 0) + ret = 0; + d_fnend(5, dev, "(i2400m %p chunk %p __chunk_len %zu addr 0x%08lx " + "direct %u do_csum %u) = %d\n", i2400m, chunk, __chunk_len, + addr, direct, do_csum, ret); + return ret; +} + + +/* + * Download a BCF file's sections to the device + * + * @i2400m: device descriptor + * @bcf: pointer to firmware data (followed by the payloads). Assumed + * verified and consistent. + * @bcf_len: length (in bytes) of the @bcf buffer. + * + * Returns: < 0 errno code on error or the offset to the jump instruction. + * + * Given a BCF file, downloads each section (a command and a payload) + * to the device's address space. Actually, it just executes each + * command i the BCF file. + * + * The section size has to be aligned to 4 bytes AND the padding has + * to be taken from the firmware file, as the signature takes it into + * account. + */ +static +ssize_t i2400m_dnload_bcf(struct i2400m *i2400m, + const struct i2400m_bcf_hdr *bcf, size_t bcf_len) +{ + ssize_t ret; + struct device *dev = i2400m_dev(i2400m); + size_t offset, /* iterator offset */ + data_size, /* Size of the data payload */ + section_size, /* Size of the whole section (cmd + payload) */ + section = 1; + const struct i2400m_bootrom_header *bh; + struct i2400m_bootrom_header ack; + + d_fnstart(3, dev, "(i2400m %p bcf %p bcf_len %zu)\n", + i2400m, bcf, bcf_len); + /* Iterate over the command blocks in the BCF file that start + * after the header */ + offset = le32_to_cpu(bcf->header_len) * sizeof(u32); + while (1) { /* start sending the file */ + bh = (void *) bcf + offset; + data_size = le32_to_cpu(bh->data_size); + section_size = ALIGN(sizeof(*bh) + data_size, 4); + d_printf(7, dev, + "downloading section #%zu (@%zu %zu B) to 0x%08x\n", + section, offset, sizeof(*bh) + data_size, + le32_to_cpu(bh->target_addr)); + if (i2400m_brh_get_opcode(bh) == I2400M_BRH_SIGNED_JUMP) { + /* Secure boot needs to stop here */ + d_printf(5, dev, "signed jump found @%zu\n", offset); + break; + } + if (offset + section_size == bcf_len) + /* Non-secure boot stops here */ + break; + if (offset + section_size > bcf_len) { + dev_err(dev, "fw %s: bad section #%zu, " + "end (@%zu) beyond EOF (@%zu)\n", + i2400m->bus_fw_name, section, + offset + section_size, bcf_len); + ret = -EINVAL; + goto error_section_beyond_eof; + } + __i2400m_msleep(20); + ret = i2400m_bm_cmd(i2400m, bh, section_size, + &ack, sizeof(ack), I2400M_BM_CMD_RAW); + if (ret < 0) { + dev_err(dev, "fw %s: section #%zu (@%zu %zu B) " + "failed %d\n", i2400m->bus_fw_name, section, + offset, sizeof(*bh) + data_size, (int) ret); + goto error_send; + } + offset += section_size; + section++; + } + ret = offset; +error_section_beyond_eof: +error_send: + d_fnend(3, dev, "(i2400m %p bcf %p bcf_len %zu) = %d\n", + i2400m, bcf, bcf_len, (int) ret); + return ret; +} + + +/* + * Do the final steps of uploading firmware + * + * Depending on the boot mode (signed vs non-signed), different + * actions need to be taken. + */ +static +int i2400m_dnload_finalize(struct i2400m *i2400m, + const struct i2400m_bcf_hdr *bcf, size_t offset) +{ + int ret = 0; + struct device *dev = i2400m_dev(i2400m); + struct i2400m_bootrom_header *cmd, ack; + struct { + struct i2400m_bootrom_header cmd; + u8 cmd_pl[0]; + } __attribute__((packed)) *cmd_buf; + size_t signature_block_offset, signature_block_size; + + d_fnstart(3, dev, "offset %zu\n", offset); + cmd = (void *) bcf + offset; + if (i2400m->sboot == 0) { + struct i2400m_bootrom_header jump_ack; + d_printf(3, dev, "unsecure boot, jumping to 0x%08x\n", + le32_to_cpu(cmd->target_addr)); + i2400m_brh_set_opcode(cmd, I2400M_BRH_JUMP); + cmd->data_size = 0; + ret = i2400m_bm_cmd(i2400m, cmd, sizeof(*cmd), + &jump_ack, sizeof(jump_ack), 0); + } else { + d_printf(3, dev, "secure boot, jumping to 0x%08x\n", + le32_to_cpu(cmd->target_addr)); + cmd_buf = i2400m->bm_cmd_buf; + memcpy(&cmd_buf->cmd, cmd, sizeof(*cmd)); + signature_block_offset = + sizeof(*bcf) + + le32_to_cpu(bcf->key_size) * sizeof(u32) + + le32_to_cpu(bcf->exponent_size) * sizeof(u32); + signature_block_size = + le32_to_cpu(bcf->modulus_size) * sizeof(u32); + memcpy(cmd_buf->cmd_pl, (void *) bcf + signature_block_offset, + signature_block_size); + ret = i2400m_bm_cmd(i2400m, &cmd_buf->cmd, + sizeof(cmd_buf->cmd) + signature_block_size, + &ack, sizeof(ack), I2400M_BM_CMD_RAW); + } + d_fnend(3, dev, "returning %d\n", ret); + return ret; +} + + +/** + * i2400m_bootrom_init - Reboots a powered device into boot mode + * + * @i2400m: device descriptor + * @flags: + * I2400M_BRI_SOFT: a reboot notification has been seen + * already, so don't wait for it. + * + * I2400M_BRI_NO_REBOOT: Don't send a reboot command, but wait + * for a reboot barker notification. This is a one shot; if + * the state machine needs to send a reboot command it will. + * + * Returns: + * + * < 0 errno code on error, 0 if ok. + * + * i2400m->sboot set to 0 for unsecure boot process, 1 for secure + * boot process. + * + * Description: + * + * Tries hard enough to put the device in boot-mode. There are two + * main phases to this: + * + * a. (1) send a reboot command and (2) get a reboot barker + * b. (1) ack the reboot sending a reboot barker and (2) getting an + * ack barker in return + * + * We want to skip (a) in some cases [soft]. The state machine is + * horrible, but it is basically: on each phase, send what has to be + * sent (if any), wait for the answer and act on the answer. We might + * have to backtrack and retry, so we keep a max tries counter for + * that. + * + * If we get a timeout after sending a warm reset, we do it again. + */ +int i2400m_bootrom_init(struct i2400m *i2400m, enum i2400m_bri flags) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct i2400m_bootrom_header *cmd; + struct i2400m_bootrom_header ack; + int count = I2400M_BOOT_RETRIES; + int ack_timeout_cnt = 1; + + BUILD_BUG_ON(sizeof(*cmd) != sizeof(i2400m_NBOOT_BARKER)); + BUILD_BUG_ON(sizeof(ack) != sizeof(i2400m_ACK_BARKER)); + + d_fnstart(4, dev, "(i2400m %p flags 0x%08x)\n", i2400m, flags); + result = -ENOMEM; + cmd = i2400m->bm_cmd_buf; + if (flags & I2400M_BRI_SOFT) + goto do_reboot_ack; +do_reboot: + if (--count < 0) + goto error_timeout; + d_printf(4, dev, "device reboot: reboot command [%d # left]\n", + count); + if ((flags & I2400M_BRI_NO_REBOOT) == 0) + i2400m->bus_reset(i2400m, I2400M_RT_WARM); + result = i2400m_bm_cmd(i2400m, NULL, 0, &ack, sizeof(ack), + I2400M_BM_CMD_RAW); + flags &= ~I2400M_BRI_NO_REBOOT; + switch (result) { + case -ERESTARTSYS: + d_printf(4, dev, "device reboot: got reboot barker\n"); + break; + case -EISCONN: /* we don't know how it got here...but we follow it */ + d_printf(4, dev, "device reboot: got ack barker - whatever\n"); + goto do_reboot; + case -ETIMEDOUT: /* device has timed out, we might be in boot + * mode already and expecting an ack, let's try + * that */ + dev_info(dev, "warm reset timed out, trying an ack\n"); + goto do_reboot_ack; + case -EPROTO: + case -ESHUTDOWN: /* dev is gone */ + case -EINTR: /* user cancelled */ + goto error_dev_gone; + default: + dev_err(dev, "device reboot: error %d while waiting " + "for reboot barker - rebooting\n", result); + goto do_reboot; + } + /* At this point we ack back with 4 REBOOT barkers and expect + * 4 ACK barkers. This is ugly, as we send a raw command -- + * hence the cast. _bm_cmd() will catch the reboot ack + * notification and report it as -EISCONN. */ +do_reboot_ack: + d_printf(4, dev, "device reboot ack: sending ack [%d # left]\n", count); + if (i2400m->sboot == 0) + memcpy(cmd, i2400m_NBOOT_BARKER, + sizeof(i2400m_NBOOT_BARKER)); + else + memcpy(cmd, i2400m_SBOOT_BARKER, + sizeof(i2400m_SBOOT_BARKER)); + result = i2400m_bm_cmd(i2400m, cmd, sizeof(*cmd), + &ack, sizeof(ack), I2400M_BM_CMD_RAW); + switch (result) { + case -ERESTARTSYS: + d_printf(4, dev, "reboot ack: got reboot barker - retrying\n"); + if (--count < 0) + goto error_timeout; + goto do_reboot_ack; + case -EISCONN: + d_printf(4, dev, "reboot ack: got ack barker - good\n"); + break; + case -ETIMEDOUT: /* no response, maybe it is the other type? */ + if (ack_timeout_cnt-- >= 0) { + d_printf(4, dev, "reboot ack timedout: " + "trying the other type?\n"); + i2400m->sboot = !i2400m->sboot; + goto do_reboot_ack; + } else { + dev_err(dev, "reboot ack timedout too long: " + "trying reboot\n"); + goto do_reboot; + } + break; + case -EPROTO: + case -ESHUTDOWN: /* dev is gone */ + goto error_dev_gone; + default: + dev_err(dev, "device reboot ack: error %d while waiting for " + "reboot ack barker - rebooting\n", result); + goto do_reboot; + } + d_printf(2, dev, "device reboot ack: got ack barker - boot done\n"); + result = 0; +exit_timeout: +error_dev_gone: + d_fnend(4, dev, "(i2400m %p flags 0x%08x) = %d\n", + i2400m, flags, result); + return result; + +error_timeout: + dev_err(dev, "Timed out waiting for reboot ack, resetting\n"); + i2400m->bus_reset(i2400m, I2400M_RT_BUS); + result = -ETIMEDOUT; + goto exit_timeout; +} + + +/* + * Read the MAC addr + * + * The position this function reads is fixed in device memory and + * always available, even without firmware. + * + * Note we specify we want to read only six bytes, but provide space + * for 16, as we always get it rounded up. + */ +int i2400m_read_mac_addr(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct net_device *net_dev = i2400m->wimax_dev.net_dev; + struct i2400m_bootrom_header *cmd; + struct { + struct i2400m_bootrom_header ack; + u8 ack_pl[16]; + } __attribute__((packed)) ack_buf; + + d_fnstart(5, dev, "(i2400m %p)\n", i2400m); + cmd = i2400m->bm_cmd_buf; + cmd->command = i2400m_brh_command(I2400M_BRH_READ, 0, 1); + cmd->target_addr = cpu_to_le32(0x00203fe8); + cmd->data_size = cpu_to_le32(6); + result = i2400m_bm_cmd(i2400m, cmd, sizeof(*cmd), + &ack_buf.ack, sizeof(ack_buf), 0); + if (result < 0) { + dev_err(dev, "BM: read mac addr failed: %d\n", result); + goto error_read_mac; + } + d_printf(2, dev, + "mac addr is %02x:%02x:%02x:%02x:%02x:%02x\n", + ack_buf.ack_pl[0], ack_buf.ack_pl[1], + ack_buf.ack_pl[2], ack_buf.ack_pl[3], + ack_buf.ack_pl[4], ack_buf.ack_pl[5]); + if (i2400m->bus_bm_mac_addr_impaired == 1) { + ack_buf.ack_pl[0] = 0x00; + ack_buf.ack_pl[1] = 0x16; + ack_buf.ack_pl[2] = 0xd3; + get_random_bytes(&ack_buf.ack_pl[3], 3); + dev_err(dev, "BM is MAC addr impaired, faking MAC addr to " + "mac addr is %02x:%02x:%02x:%02x:%02x:%02x\n", + ack_buf.ack_pl[0], ack_buf.ack_pl[1], + ack_buf.ack_pl[2], ack_buf.ack_pl[3], + ack_buf.ack_pl[4], ack_buf.ack_pl[5]); + result = 0; + } + net_dev->addr_len = ETH_ALEN; + memcpy(net_dev->perm_addr, ack_buf.ack_pl, ETH_ALEN); + memcpy(net_dev->dev_addr, ack_buf.ack_pl, ETH_ALEN); +error_read_mac: + d_fnend(5, dev, "(i2400m %p) = %d\n", i2400m, result); + return result; +} + + +/* + * Initialize a non signed boot + * + * This implies sending some magic values to the device's memory. Note + * we convert the values to little endian in the same array + * declaration. + */ +static +int i2400m_dnload_init_nonsigned(struct i2400m *i2400m) +{ +#define POKE(a, d) { \ + .address = __constant_cpu_to_le32(a), \ + .data = __constant_cpu_to_le32(d) \ +} + static const struct { + __le32 address; + __le32 data; + } i2400m_pokes[] = { + POKE(0x081A58, 0xA7810230), + POKE(0x080040, 0x00000000), + POKE(0x080048, 0x00000082), + POKE(0x08004C, 0x0000081F), + POKE(0x080054, 0x00000085), + POKE(0x080058, 0x00000180), + POKE(0x08005C, 0x00000018), + POKE(0x080060, 0x00000010), + POKE(0x080574, 0x00000001), + POKE(0x080550, 0x00000005), + POKE(0xAE0000, 0x00000000), + }; +#undef POKE + unsigned i; + int ret; + struct device *dev = i2400m_dev(i2400m); + + dev_warn(dev, "WARNING!!! non-signed boot UNTESTED PATH!\n"); + + d_fnstart(5, dev, "(i2400m %p)\n", i2400m); + for (i = 0; i < ARRAY_SIZE(i2400m_pokes); i++) { + ret = i2400m_download_chunk(i2400m, &i2400m_pokes[i].data, + sizeof(i2400m_pokes[i].data), + i2400m_pokes[i].address, 1, 1); + if (ret < 0) + break; + } + d_fnend(5, dev, "(i2400m %p) = %d\n", i2400m, ret); + return ret; +} + + +/* + * Initialize the signed boot process + * + * @i2400m: device descriptor + * + * @bcf_hdr: pointer to the firmware header; assumes it is fully in + * memory (it has gone through basic validation). + * + * Returns: 0 if ok, < 0 errno code on error, -ERESTARTSYS if the hw + * rebooted. + * + * This writes the firmware BCF header to the device using the + * HASH_PAYLOAD_ONLY command. + */ +static +int i2400m_dnload_init_signed(struct i2400m *i2400m, + const struct i2400m_bcf_hdr *bcf_hdr) +{ + int ret; + struct device *dev = i2400m_dev(i2400m); + struct { + struct i2400m_bootrom_header cmd; + struct i2400m_bcf_hdr cmd_pl; + } __attribute__((packed)) *cmd_buf; + struct i2400m_bootrom_header ack; + + d_fnstart(5, dev, "(i2400m %p bcf_hdr %p)\n", i2400m, bcf_hdr); + cmd_buf = i2400m->bm_cmd_buf; + cmd_buf->cmd.command = + i2400m_brh_command(I2400M_BRH_HASH_PAYLOAD_ONLY, 0, 0); + cmd_buf->cmd.target_addr = 0; + cmd_buf->cmd.data_size = cpu_to_le32(sizeof(cmd_buf->cmd_pl)); + memcpy(&cmd_buf->cmd_pl, bcf_hdr, sizeof(*bcf_hdr)); + ret = i2400m_bm_cmd(i2400m, &cmd_buf->cmd, sizeof(*cmd_buf), + &ack, sizeof(ack), 0); + if (ret >= 0) + ret = 0; + d_fnend(5, dev, "(i2400m %p bcf_hdr %p) = %d\n", i2400m, bcf_hdr, ret); + return ret; +} + + +/* + * Initialize the firmware download at the device size + * + * Multiplex to the one that matters based on the device's mode + * (signed or non-signed). + */ +static +int i2400m_dnload_init(struct i2400m *i2400m, const struct i2400m_bcf_hdr *bcf) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + u32 module_id = le32_to_cpu(bcf->module_id); + + if (i2400m->sboot == 0 + && (module_id & I2400M_BCF_MOD_ID_POKES) == 0) { + /* non-signed boot process without pokes */ + result = i2400m_dnload_init_nonsigned(i2400m); + if (result == -ERESTARTSYS) + return result; + if (result < 0) + dev_err(dev, "fw %s: non-signed download " + "initialization failed: %d\n", + i2400m->bus_fw_name, result); + } else if (i2400m->sboot == 0 + && (module_id & I2400M_BCF_MOD_ID_POKES)) { + /* non-signed boot process with pokes, nothing to do */ + result = 0; + } else { /* signed boot process */ + result = i2400m_dnload_init_signed(i2400m, bcf); + if (result == -ERESTARTSYS) + return result; + if (result < 0) + dev_err(dev, "fw %s: signed boot download " + "initialization failed: %d\n", + i2400m->bus_fw_name, result); + } + return result; +} + + +/* + * Run quick consistency tests on the firmware file + * + * Check for the firmware being made for the i2400m device, + * etc...These checks are mostly informative, as the device will make + * them too; but the driver's response is more informative on what + * went wrong. + */ +static +int i2400m_fw_check(struct i2400m *i2400m, + const struct i2400m_bcf_hdr *bcf, + size_t bcf_size) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + unsigned module_type, header_len, major_version, minor_version, + module_id, module_vendor, date, size; + + /* Check hard errors */ + result = -EINVAL; + if (bcf_size < sizeof(*bcf)) { /* big enough header? */ + dev_err(dev, "firmware %s too short: " + "%zu B vs %zu (at least) expected\n", + i2400m->bus_fw_name, bcf_size, sizeof(*bcf)); + goto error; + } + + module_type = bcf->module_type; + header_len = sizeof(u32) * le32_to_cpu(bcf->header_len); + major_version = le32_to_cpu(bcf->header_version) & 0xffff0000 >> 16; + minor_version = le32_to_cpu(bcf->header_version) & 0x0000ffff; + module_id = le32_to_cpu(bcf->module_id); + module_vendor = le32_to_cpu(bcf->module_vendor); + date = le32_to_cpu(bcf->date); + size = sizeof(u32) * le32_to_cpu(bcf->size); + + if (bcf_size != size) { /* annoyingly paranoid */ + dev_err(dev, "firmware %s: bad size, got " + "%zu B vs %u expected\n", + i2400m->bus_fw_name, bcf_size, size); + goto error; + } + + d_printf(2, dev, "type 0x%x id 0x%x vendor 0x%x; header v%u.%u (%zu B) " + "date %08x (%zu B)\n", + module_type, module_id, module_vendor, + major_version, minor_version, (size_t) header_len, + date, (size_t) size); + + if (module_type != 6) { /* built for the right hardware? */ + dev_err(dev, "bad fw %s: unexpected module type 0x%x; " + "aborting\n", i2400m->bus_fw_name, module_type); + goto error; + } + + /* Check soft-er errors */ + result = 0; + if (module_vendor != 0x8086) + dev_err(dev, "bad fw %s? unexpected vendor 0x%04x\n", + i2400m->bus_fw_name, module_vendor); + if (date < 0x20080300) + dev_err(dev, "bad fw %s? build date too old %08x\n", + i2400m->bus_fw_name, date); +error: + return result; +} + + +/* + * Download the firmware to the device + * + * @i2400m: device descriptor + * @bcf: pointer to loaded (and minimally verified for consistency) + * firmware + * @bcf_size: size of the @bcf buffer (header plus payloads) + * + * The process for doing this is described in this file's header. + * + * Note we only reinitialize boot-mode if the flags say so. Some hw + * iterations need it, some don't. In any case, if we loop, we always + * need to reinitialize the boot room, hence the flags modification. + */ +static +int i2400m_fw_dnload(struct i2400m *i2400m, const struct i2400m_bcf_hdr *bcf, + size_t bcf_size, enum i2400m_bri flags) +{ + int ret = 0; + struct device *dev = i2400m_dev(i2400m); + int count = I2400M_BOOT_RETRIES; + + d_fnstart(5, dev, "(i2400m %p bcf %p size %zu)\n", + i2400m, bcf, bcf_size); + i2400m->boot_mode = 1; +hw_reboot: + if (count-- == 0) { + ret = -ERESTARTSYS; + dev_err(dev, "device rebooted too many times, aborting\n"); + goto error_too_many_reboots; + } + if (flags & I2400M_BRI_MAC_REINIT) { + ret = i2400m_bootrom_init(i2400m, flags); + if (ret < 0) { + dev_err(dev, "bootrom init failed: %d\n", ret); + goto error_bootrom_init; + } + } + flags |= I2400M_BRI_MAC_REINIT; + + /* + * Initialize the download, push the bytes to the device and + * then jump to the new firmware. Note @ret is passed with the + * offset of the jump instruction to _dnload_finalize() + */ + ret = i2400m_dnload_init(i2400m, bcf); /* Init device's dnload */ + if (ret == -ERESTARTSYS) + goto error_dev_rebooted; + if (ret < 0) + goto error_dnload_init; + + ret = i2400m_dnload_bcf(i2400m, bcf, bcf_size); + if (ret == -ERESTARTSYS) + goto error_dev_rebooted; + if (ret < 0) { + dev_err(dev, "fw %s: download failed: %d\n", + i2400m->bus_fw_name, ret); + goto error_dnload_bcf; + } + + ret = i2400m_dnload_finalize(i2400m, bcf, ret); + if (ret == -ERESTARTSYS) + goto error_dev_rebooted; + if (ret < 0) { + dev_err(dev, "fw %s: " + "download finalization failed: %d\n", + i2400m->bus_fw_name, ret); + goto error_dnload_finalize; + } + + d_printf(2, dev, "fw %s successfully uploaded\n", + i2400m->bus_fw_name); + i2400m->boot_mode = 0; +error_dnload_finalize: +error_dnload_bcf: +error_dnload_init: +error_bootrom_init: +error_too_many_reboots: + d_fnend(5, dev, "(i2400m %p bcf %p size %zu) = %d\n", + i2400m, bcf, bcf_size, ret); + return ret; + +error_dev_rebooted: + dev_err(dev, "device rebooted, %d tries left\n", count); + /* we got the notification already, no need to wait for it again */ + flags |= I2400M_BRI_SOFT; + goto hw_reboot; +} + + +/** + * i2400m_dev_bootstrap - Bring the device to a known state and upload firmware + * + * @i2400m: device descriptor + * + * Returns: >= 0 if ok, < 0 errno code on error. + * + * This sets up the firmware upload environment, loads the firmware + * file from disk, verifies and then calls the firmware upload process + * per se. + * + * Can be called either from probe, or after a warm reset. Can not be + * called from within an interrupt. All the flow in this code is + * single-threade; all I/Os are synchronous. + */ +int i2400m_dev_bootstrap(struct i2400m *i2400m, enum i2400m_bri flags) +{ + int ret = 0; + struct device *dev = i2400m_dev(i2400m); + const struct firmware *fw; + const struct i2400m_bcf_hdr *bcf; /* Firmware data */ + + d_fnstart(5, dev, "(i2400m %p)\n", i2400m); + /* Load firmware files to memory. */ + ret = request_firmware(&fw, i2400m->bus_fw_name, dev); + if (ret) { + dev_err(dev, "fw %s: request failed: %d\n", + i2400m->bus_fw_name, ret); + goto error_fw_req; + } + bcf = (void *) fw->data; + + ret = i2400m_fw_check(i2400m, bcf, fw->size); + if (ret < 0) + goto error_fw_bad; + ret = i2400m_fw_dnload(i2400m, bcf, fw->size, flags); +error_fw_bad: + release_firmware(fw); +error_fw_req: + d_fnend(5, dev, "(i2400m %p) = %d\n", i2400m, ret); + return ret; +} +EXPORT_SYMBOL_GPL(i2400m_dev_bootstrap); From aa5a7acabe31ec27a212cbd25cad9f72aa476591 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:47 -0800 Subject: [PATCH 110/123] i2400m: RX and TX data/control paths Handling of TX/RX data to/from the i2400m device (IP packets, control and diagnostics). On RX, this parses the received read transaction from the device, breaks it in chunks and passes it to the corresponding subsystems (network and control). Transmission to the device is done through a software FIFO, as data/control frames can be coalesced (while the device is reading the previous tx transaction, others accumulate). A FIFO is used because at the end it is resource-cheaper that scatter/gather over USB. As well, most traffic is going to be download (vs upload). Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/rx.c | 534 ++++++++++++++++++++++ drivers/net/wimax/i2400m/tx.c | 817 ++++++++++++++++++++++++++++++++++ 2 files changed, 1351 insertions(+) create mode 100644 drivers/net/wimax/i2400m/rx.c create mode 100644 drivers/net/wimax/i2400m/tx.c diff --git a/drivers/net/wimax/i2400m/rx.c b/drivers/net/wimax/i2400m/rx.c new file mode 100644 index 00000000000..6922022710a --- /dev/null +++ b/drivers/net/wimax/i2400m/rx.c @@ -0,0 +1,534 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Handle incoming traffic and deliver it to the control or data planes + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Yanir Lubetkin + * - Initial implementation + * Inaky Perez-Gonzalez + * - Use skb_clone(), break up processing in chunks + * - Split transport/device specific + * - Make buffer size dynamic to exert less memory pressure + * + * + * This handles the RX path. + * + * We receive an RX message from the bus-specific driver, which + * contains one or more payloads that have potentially different + * destinataries (data or control paths). + * + * So we just take that payload from the transport specific code in + * the form of an skb, break it up in chunks (a cloned skb each in the + * case of network packets) and pass it to netdev or to the + * command/ack handler (and from there to the WiMAX stack). + * + * PROTOCOL FORMAT + * + * The format of the buffer is: + * + * HEADER (struct i2400m_msg_hdr) + * PAYLOAD DESCRIPTOR 0 (struct i2400m_pld) + * PAYLOAD DESCRIPTOR 1 + * ... + * PAYLOAD DESCRIPTOR N + * PAYLOAD 0 (raw bytes) + * PAYLOAD 1 + * ... + * PAYLOAD N + * + * See tx.c for a deeper description on alignment requirements and + * other fun facts of it. + * + * ROADMAP + * + * i2400m_rx + * i2400m_rx_msg_hdr_check + * i2400m_rx_pl_descr_check + * i2400m_rx_payload + * i2400m_net_rx + * i2400m_rx_ctl + * i2400m_msg_size_check + * i2400m_report_hook_work [in a workqueue] + * i2400m_report_hook + * wimax_msg_to_user + * i2400m_rx_ctl_ack + * wimax_msg_to_user_alloc + * i2400m_rx_trace + * i2400m_msg_size_check + * wimax_msg + */ +#include +#include +#include +#include +#include "i2400m.h" + + +#define D_SUBMODULE rx +#include "debug-levels.h" + +struct i2400m_report_hook_args { + struct sk_buff *skb_rx; + const struct i2400m_l3l4_hdr *l3l4_hdr; + size_t size; +}; + + +/* + * Execute i2400m_report_hook in a workqueue + * + * Unpacks arguments from the deferred call, executes it and then + * drops the references. + * + * Obvious NOTE: References are needed because we are a separate + * thread; otherwise the buffer changes under us because it is + * released by the original caller. + */ +static +void i2400m_report_hook_work(struct work_struct *ws) +{ + struct i2400m_work *iw = + container_of(ws, struct i2400m_work, ws); + struct i2400m_report_hook_args *args = (void *) iw->pl; + i2400m_report_hook(iw->i2400m, args->l3l4_hdr, args->size); + kfree_skb(args->skb_rx); + i2400m_put(iw->i2400m); + kfree(iw); +} + + +/* + * Process an ack to a command + * + * @i2400m: device descriptor + * @payload: pointer to message + * @size: size of the message + * + * Pass the acknodledgment (in an skb) to the thread that is waiting + * for it in i2400m->msg_completion. + * + * We need to coordinate properly with the thread waiting for the + * ack. Check if it is waiting or if it is gone. We loose the spinlock + * to avoid allocating on atomic contexts (yeah, could use GFP_ATOMIC, + * but this is not so speed critical). + */ +static +void i2400m_rx_ctl_ack(struct i2400m *i2400m, + const void *payload, size_t size) +{ + struct device *dev = i2400m_dev(i2400m); + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + unsigned long flags; + struct sk_buff *ack_skb; + + /* Anyone waiting for an answer? */ + spin_lock_irqsave(&i2400m->rx_lock, flags); + if (i2400m->ack_skb != ERR_PTR(-EINPROGRESS)) { + dev_err(dev, "Huh? reply to command with no waiters\n"); + goto error_no_waiter; + } + spin_unlock_irqrestore(&i2400m->rx_lock, flags); + + ack_skb = wimax_msg_alloc(wimax_dev, NULL, payload, size, GFP_KERNEL); + + /* Check waiter didn't time out waiting for the answer... */ + spin_lock_irqsave(&i2400m->rx_lock, flags); + if (i2400m->ack_skb != ERR_PTR(-EINPROGRESS)) { + d_printf(1, dev, "Huh? waiter for command reply cancelled\n"); + goto error_waiter_cancelled; + } + if (ack_skb == NULL) { + dev_err(dev, "CMD/GET/SET ack: cannot allocate SKB\n"); + i2400m->ack_skb = ERR_PTR(-ENOMEM); + } else + i2400m->ack_skb = ack_skb; + spin_unlock_irqrestore(&i2400m->rx_lock, flags); + complete(&i2400m->msg_completion); + return; + +error_waiter_cancelled: + if (ack_skb) + kfree_skb(ack_skb); +error_no_waiter: + spin_unlock_irqrestore(&i2400m->rx_lock, flags); + return; +} + + +/* + * Receive and process a control payload + * + * @i2400m: device descriptor + * @skb_rx: skb that contains the payload (for reference counting) + * @payload: pointer to message + * @size: size of the message + * + * There are two types of control RX messages: reports (asynchronous, + * like your every day interrupts) and 'acks' (reponses to a command, + * get or set request). + * + * If it is a report, we run hooks on it (to extract information for + * things we need to do in the driver) and then pass it over to the + * WiMAX stack to send it to user space. + * + * NOTE: report processing is done in a workqueue specific to the + * generic driver, to avoid deadlocks in the system. + * + * If it is not a report, it is an ack to a previously executed + * command, set or get, so wake up whoever is waiting for it from + * i2400m_msg_to_dev(). i2400m_rx_ctl_ack() takes care of that. + * + * Note that the sizes we pass to other functions from here are the + * sizes of the _l3l4_hdr + payload, not full buffer sizes, as we have + * verified in _msg_size_check() that they are congruent. + * + * For reports: We can't clone the original skb where the data is + * because we need to send this up via netlink; netlink has to add + * headers and we can't overwrite what's preceeding the payload...as + * it is another message. So we just dup them. + */ +static +void i2400m_rx_ctl(struct i2400m *i2400m, struct sk_buff *skb_rx, + const void *payload, size_t size) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + const struct i2400m_l3l4_hdr *l3l4_hdr = payload; + unsigned msg_type; + + result = i2400m_msg_size_check(i2400m, l3l4_hdr, size); + if (result < 0) { + dev_err(dev, "HW BUG? device sent a bad message: %d\n", + result); + goto error_check; + } + msg_type = le16_to_cpu(l3l4_hdr->type); + d_printf(1, dev, "%s 0x%04x: %zu bytes\n", + msg_type & I2400M_MT_REPORT_MASK ? "REPORT" : "CMD/SET/GET", + msg_type, size); + d_dump(2, dev, l3l4_hdr, size); + if (msg_type & I2400M_MT_REPORT_MASK) { + /* These hooks have to be ran serialized; as well, the + * handling might force the execution of commands, and + * that might cause reentrancy issues with + * bus-specific subdrivers and workqueues. So we run + * it in a separate workqueue. */ + struct i2400m_report_hook_args args = { + .skb_rx = skb_rx, + .l3l4_hdr = l3l4_hdr, + .size = size + }; + if (unlikely(i2400m->ready == 0)) /* only send if up */ + return; + skb_get(skb_rx); + i2400m_queue_work(i2400m, i2400m_report_hook_work, + GFP_KERNEL, &args, sizeof(args)); + result = wimax_msg(&i2400m->wimax_dev, NULL, l3l4_hdr, size, + GFP_KERNEL); + if (result < 0) + dev_err(dev, "error sending report to userspace: %d\n", + result); + } else /* an ack to a CMD, GET or SET */ + i2400m_rx_ctl_ack(i2400m, payload, size); +error_check: + return; +} + + + + +/* + * Receive and send up a trace + * + * @i2400m: device descriptor + * @skb_rx: skb that contains the trace (for reference counting) + * @payload: pointer to trace message inside the skb + * @size: size of the message + * + * THe i2400m might produce trace information (diagnostics) and we + * send them through a different kernel-to-user pipe (to avoid + * clogging it). + * + * As in i2400m_rx_ctl(), we can't clone the original skb where the + * data is because we need to send this up via netlink; netlink has to + * add headers and we can't overwrite what's preceeding the + * payload...as it is another message. So we just dup them. + */ +static +void i2400m_rx_trace(struct i2400m *i2400m, + const void *payload, size_t size) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + const struct i2400m_l3l4_hdr *l3l4_hdr = payload; + unsigned msg_type; + + result = i2400m_msg_size_check(i2400m, l3l4_hdr, size); + if (result < 0) { + dev_err(dev, "HW BUG? device sent a bad trace message: %d\n", + result); + goto error_check; + } + msg_type = le16_to_cpu(l3l4_hdr->type); + d_printf(1, dev, "Trace %s 0x%04x: %zu bytes\n", + msg_type & I2400M_MT_REPORT_MASK ? "REPORT" : "CMD/SET/GET", + msg_type, size); + d_dump(2, dev, l3l4_hdr, size); + if (unlikely(i2400m->ready == 0)) /* only send if up */ + return; + result = wimax_msg(wimax_dev, "trace", l3l4_hdr, size, GFP_KERNEL); + if (result < 0) + dev_err(dev, "error sending trace to userspace: %d\n", + result); +error_check: + return; +} + + +/* + * Act on a received payload + * + * @i2400m: device instance + * @skb_rx: skb where the transaction was received + * @single: 1 if there is only one payload, 0 otherwise + * @pld: payload descriptor + * @payload: payload data + * + * Upon reception of a payload, look at its guts in the payload + * descriptor and decide what to do with it. + */ +static +void i2400m_rx_payload(struct i2400m *i2400m, struct sk_buff *skb_rx, + unsigned single, const struct i2400m_pld *pld, + const void *payload) +{ + struct device *dev = i2400m_dev(i2400m); + size_t pl_size = i2400m_pld_size(pld); + enum i2400m_pt pl_type = i2400m_pld_type(pld); + + switch (pl_type) { + case I2400M_PT_DATA: + d_printf(3, dev, "RX: data payload %zu bytes\n", pl_size); + i2400m_net_rx(i2400m, skb_rx, single, payload, pl_size); + break; + case I2400M_PT_CTRL: + i2400m_rx_ctl(i2400m, skb_rx, payload, pl_size); + break; + case I2400M_PT_TRACE: + i2400m_rx_trace(i2400m, payload, pl_size); + break; + default: /* Anything else shouldn't come to the host */ + if (printk_ratelimit()) + dev_err(dev, "RX: HW BUG? unexpected payload type %u\n", + pl_type); + } +} + + +/* + * Check a received transaction's message header + * + * @i2400m: device descriptor + * @msg_hdr: message header + * @buf_size: size of the received buffer + * + * Check that the declarations done by a RX buffer message header are + * sane and consistent with the amount of data that was received. + */ +static +int i2400m_rx_msg_hdr_check(struct i2400m *i2400m, + const struct i2400m_msg_hdr *msg_hdr, + size_t buf_size) +{ + int result = -EIO; + struct device *dev = i2400m_dev(i2400m); + if (buf_size < sizeof(*msg_hdr)) { + dev_err(dev, "RX: HW BUG? message with short header (%zu " + "vs %zu bytes expected)\n", buf_size, sizeof(*msg_hdr)); + goto error; + } + if (msg_hdr->barker != cpu_to_le32(I2400M_D2H_MSG_BARKER)) { + dev_err(dev, "RX: HW BUG? message received with unknown " + "barker 0x%08x (buf_size %zu bytes)\n", + le32_to_cpu(msg_hdr->barker), buf_size); + goto error; + } + if (msg_hdr->num_pls == 0) { + dev_err(dev, "RX: HW BUG? zero payload packets in message\n"); + goto error; + } + if (le16_to_cpu(msg_hdr->num_pls) > I2400M_MAX_PLS_IN_MSG) { + dev_err(dev, "RX: HW BUG? message contains more payload " + "than maximum; ignoring.\n"); + goto error; + } + result = 0; +error: + return result; +} + + +/* + * Check a payload descriptor against the received data + * + * @i2400m: device descriptor + * @pld: payload descriptor + * @pl_itr: offset (in bytes) in the received buffer the payload is + * located + * @buf_size: size of the received buffer + * + * Given a payload descriptor (part of a RX buffer), check it is sane + * and that the data it declares fits in the buffer. + */ +static +int i2400m_rx_pl_descr_check(struct i2400m *i2400m, + const struct i2400m_pld *pld, + size_t pl_itr, size_t buf_size) +{ + int result = -EIO; + struct device *dev = i2400m_dev(i2400m); + size_t pl_size = i2400m_pld_size(pld); + enum i2400m_pt pl_type = i2400m_pld_type(pld); + + if (pl_size > i2400m->bus_pl_size_max) { + dev_err(dev, "RX: HW BUG? payload @%zu: size %zu is " + "bigger than maximum %zu; ignoring message\n", + pl_itr, pl_size, i2400m->bus_pl_size_max); + goto error; + } + if (pl_itr + pl_size > buf_size) { /* enough? */ + dev_err(dev, "RX: HW BUG? payload @%zu: size %zu " + "goes beyond the received buffer " + "size (%zu bytes); ignoring message\n", + pl_itr, pl_size, buf_size); + goto error; + } + if (pl_type >= I2400M_PT_ILLEGAL) { + dev_err(dev, "RX: HW BUG? illegal payload type %u; " + "ignoring message\n", pl_type); + goto error; + } + result = 0; +error: + return result; +} + + +/** + * i2400m_rx - Receive a buffer of data from the device + * + * @i2400m: device descriptor + * @skb: skbuff where the data has been received + * + * Parse in a buffer of data that contains an RX message sent from the + * device. See the file header for the format. Run all checks on the + * buffer header, then run over each payload's descriptors, verify + * their consistency and act on each payload's contents. If + * everything is succesful, update the device's statistics. + * + * Note: You need to set the skb to contain only the length of the + * received buffer; for that, use skb_trim(skb, RECEIVED_SIZE). + * + * Returns: + * + * 0 if ok, < 0 errno on error + * + * If ok, this function owns now the skb and the caller DOESN'T have + * to run kfree_skb() on it. However, on error, the caller still owns + * the skb and it is responsible for releasing it. + */ +int i2400m_rx(struct i2400m *i2400m, struct sk_buff *skb) +{ + int i, result; + struct device *dev = i2400m_dev(i2400m); + const struct i2400m_msg_hdr *msg_hdr; + size_t pl_itr, pl_size, skb_len; + unsigned long flags; + unsigned num_pls; + + skb_len = skb->len; + d_fnstart(4, dev, "(i2400m %p skb %p [size %zu])\n", + i2400m, skb, skb_len); + result = -EIO; + msg_hdr = (void *) skb->data; + result = i2400m_rx_msg_hdr_check(i2400m, msg_hdr, skb->len); + if (result < 0) + goto error_msg_hdr_check; + result = -EIO; + num_pls = le16_to_cpu(msg_hdr->num_pls); + pl_itr = sizeof(*msg_hdr) + /* Check payload descriptor(s) */ + num_pls * sizeof(msg_hdr->pld[0]); + pl_itr = ALIGN(pl_itr, I2400M_PL_PAD); + if (pl_itr > skb->len) { /* got all the payload descriptors? */ + dev_err(dev, "RX: HW BUG? message too short (%u bytes) for " + "%u payload descriptors (%zu each, total %zu)\n", + skb->len, num_pls, sizeof(msg_hdr->pld[0]), pl_itr); + goto error_pl_descr_short; + } + /* Walk each payload payload--check we really got it */ + for (i = 0; i < num_pls; i++) { + /* work around old gcc warnings */ + pl_size = i2400m_pld_size(&msg_hdr->pld[i]); + result = i2400m_rx_pl_descr_check(i2400m, &msg_hdr->pld[i], + pl_itr, skb->len); + if (result < 0) + goto error_pl_descr_check; + i2400m_rx_payload(i2400m, skb, num_pls == 1, &msg_hdr->pld[i], + skb->data + pl_itr); + pl_itr += ALIGN(pl_size, I2400M_PL_PAD); + cond_resched(); /* Don't monopolize */ + } + kfree_skb(skb); + /* Update device statistics */ + spin_lock_irqsave(&i2400m->rx_lock, flags); + i2400m->rx_pl_num += i; + if (i > i2400m->rx_pl_max) + i2400m->rx_pl_max = i; + if (i < i2400m->rx_pl_min) + i2400m->rx_pl_min = i; + i2400m->rx_num++; + i2400m->rx_size_acc += skb->len; + if (skb->len < i2400m->rx_size_min) + i2400m->rx_size_min = skb->len; + if (skb->len > i2400m->rx_size_max) + i2400m->rx_size_max = skb->len; + spin_unlock_irqrestore(&i2400m->rx_lock, flags); +error_pl_descr_check: +error_pl_descr_short: +error_msg_hdr_check: + d_fnend(4, dev, "(i2400m %p skb %p [size %zu]) = %d\n", + i2400m, skb, skb_len, result); + return result; +} +EXPORT_SYMBOL_GPL(i2400m_rx); diff --git a/drivers/net/wimax/i2400m/tx.c b/drivers/net/wimax/i2400m/tx.c new file mode 100644 index 00000000000..613a88ffd65 --- /dev/null +++ b/drivers/net/wimax/i2400m/tx.c @@ -0,0 +1,817 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Generic (non-bus specific) TX handling + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Yanir Lubetkin + * - Initial implementation + * + * Intel Corporation + * Inaky Perez-Gonzalez + * - Rewritten to use a single FIFO to lower the memory allocation + * pressure and optimize cache hits when copying to the queue, as + * well as splitting out bus-specific code. + * + * + * Implements data transmission to the device; this is done through a + * software FIFO, as data/control frames can be coalesced (while the + * device is reading the previous tx transaction, others accumulate). + * + * A FIFO is used because at the end it is resource-cheaper that trying + * to implement scatter/gather over USB. As well, most traffic is going + * to be download (vs upload). + * + * The format for sending/receiving data to/from the i2400m is + * described in detail in rx.c:PROTOCOL FORMAT. In here we implement + * the transmission of that. This is split between a bus-independent + * part that just prepares everything and a bus-specific part that + * does the actual transmission over the bus to the device (in the + * bus-specific driver). + * + * + * The general format of a device-host transaction is MSG-HDR, PLD1, + * PLD2...PLDN, PL1, PL2,...PLN, PADDING. + * + * Because we need the send payload descriptors and then payloads and + * because it is kind of expensive to do scatterlists in USB (one URB + * per node), it becomes cheaper to append all the data to a FIFO + * (copying to a FIFO potentially in cache is cheaper). + * + * Then the bus-specific code takes the parts of that FIFO that are + * written and passes them to the device. + * + * So the concepts to keep in mind there are: + * + * We use a FIFO to queue the data in a linear buffer. We first append + * a MSG-HDR, space for I2400M_TX_PLD_MAX payload descriptors and then + * go appending payloads until we run out of space or of payload + * descriptors. Then we append padding to make the whole transaction a + * multiple of i2400m->bus_tx_block_size (as defined by the bus layer). + * + * - A TX message: a combination of a message header, payload + * descriptors and payloads. + * + * Open: it is marked as active (i2400m->tx_msg is valid) and we + * can keep adding payloads to it. + * + * Closed: we are not appending more payloads to this TX message + * (exahusted space in the queue, too many payloads or + * whichever). We have appended padding so the whole message + * length is aligned to i2400m->bus_tx_block_size (as set by the + * bus/transport layer). + * + * - Most of the time we keep a TX message open to which we append + * payloads. + * + * - If we are going to append and there is no more space (we are at + * the end of the FIFO), we close the message, mark the rest of the + * FIFO space unusable (skip_tail), create a new message at the + * beginning of the FIFO (if there is space) and append the message + * there. + * + * This is because we need to give linear TX messages to the bus + * engine. So we don't write a message to the remaining FIFO space + * until the tail and continue at the head of it. + * + * - We overload one of the fields in the message header to use it as + * 'size' of the TX message, so we can iterate over them. It also + * contains a flag that indicates if we have to skip it or not. + * When we send the buffer, we update that to its real on-the-wire + * value. + * + * - The MSG-HDR PLD1...PLD2 stuff has to be a size multiple of 16. + * + * It follows that if MSG-HDR says we have N messages, the whole + * header + descriptors is 16 + 4*N; for those to be a multiple of + * 16, it follows that N can be 4, 8, 12, ... (32, 48, 64, 80... + * bytes). + * + * So if we have only 1 payload, we have to submit a header that in + * all truth has space for 4. + * + * The implication is that we reserve space for 12 (64 bytes); but + * if we fill up only (eg) 2, our header becomes 32 bytes only. So + * the TX engine has to shift those 32 bytes of msg header and 2 + * payloads and padding so that right after it the payloads start + * and the TX engine has to know about that. + * + * It is cheaper to move the header up than the whole payloads down. + * + * We do this in i2400m_tx_close(). See 'i2400m_msg_hdr->offset'. + * + * - Each payload has to be size-padded to 16 bytes; before appending + * it, we just do it. + * + * - The whole message has to be padded to i2400m->bus_tx_block_size; + * we do this at close time. Thus, when reserving space for the + * payload, we always make sure there is also free space for this + * padding that sooner or later will happen. + * + * When we append a message, we tell the bus specific code to kick in + * TXs. It will TX (in parallel) until the buffer is exhausted--hence + * the lockin we do. The TX code will only send a TX message at the + * time (which remember, might contain more than one payload). Of + * course, when the bus-specific driver attempts to TX a message that + * is still open, it gets closed first. + * + * Gee, this is messy; well a picture. In the example below we have a + * partially full FIFO, with a closed message ready to be delivered + * (with a moved message header to make sure it is size-aligned to + * 16), TAIL room that was unusable (and thus is marked with a message + * header that says 'skip this') and at the head of the buffer, an + * imcomplete message with a couple of payloads. + * + * N ___________________________________________________ + * | | + * | TAIL room | + * | | + * | msg_hdr to skip (size |= 0x80000) | + * |---------------------------------------------------|------- + * | | /|\ + * | | | + * | TX message padding | | + * | | | + * | | | + * |- - - - - - - - - - - - - - - - - - - - - - - - - -| | + * | | | + * | payload 1 | | + * | | N * tx_block_size + * | | | + * |- - - - - - - - - - - - - - - - - - - - - - - - - -| | + * | | | + * | payload 1 | | + * | | | + * | | | + * |- - - - - - - - - - - - - - - - - - - - - - - - - -|- -|- - - - + * | padding 3 /|\ | | /|\ + * | padding 2 | | | | + * | pld 1 32 bytes (2 * 16) | | | + * | pld 0 | | | | + * | moved msg_hdr \|/ | \|/ | + * |- - - - - - - - - - - - - - - - - - - - - - - - - -|- - - | + * | | _PLD_SIZE + * | unused | | + * | | | + * |- - - - - - - - - - - - - - - - - - - - - - - - - -| | + * | msg_hdr (size X) [this message is closed] | \|/ + * |===================================================|========== <=== OUT + * | | + * | | + * | | + * | Free rooom | + * | | + * | | + * | | + * | | + * | | + * | | + * | | + * | | + * | | + * |===================================================|========== <=== IN + * | | + * | | + * | | + * | | + * | payload 1 | + * | | + * | | + * |- - - - - - - - - - - - - - - - - - - - - - - - - -| + * | | + * | payload 0 | + * | | + * | | + * |- - - - - - - - - - - - - - - - - - - - - - - - - -| + * | pld 11 /|\ | + * | ... | | + * | pld 1 64 bytes (2 * 16) | + * | pld 0 | | + * | msg_hdr (size X) \|/ [message is open] | + * 0 --------------------------------------------------- + * + * + * ROADMAP + * + * i2400m_tx_setup() Called by i2400m_setup + * i2400m_tx_release() Called by i2400m_release() + * + * i2400m_tx() Called to send data or control frames + * i2400m_tx_fifo_push() Allocates append-space in the FIFO + * i2400m_tx_new() Opens a new message in the FIFO + * i2400m_tx_fits() Checks if a new payload fits in the message + * i2400m_tx_close() Closes an open message in the FIFO + * i2400m_tx_skip_tail() Marks unusable FIFO tail space + * i2400m->bus_tx_kick() + * + * Now i2400m->bus_tx_kick() is the the bus-specific driver backend + * implementation; that would do: + * + * i2400m->bus_tx_kick() + * i2400m_tx_msg_get() Gets first message ready to go + * ...sends it... + * i2400m_tx_msg_sent() Ack the message is sent; repeat from + * _tx_msg_get() until it returns NULL + * (FIFO empty). + */ +#include +#include "i2400m.h" + + +#define D_SUBMODULE tx +#include "debug-levels.h" + +enum { + /** + * TX Buffer size + * + * Doc says maximum transaction is 16KiB. If we had 16KiB en + * route and 16KiB being queued, it boils down to needing + * 32KiB. + */ + I2400M_TX_BUF_SIZE = 32768, + /** + * Message header and payload descriptors have to be 16 + * aligned (16 + 4 * N = 16 * M). If we take that average sent + * packets are MTU size (~1400-~1500) it follows that we could + * fit at most 10-11 payloads in one transaction. To meet the + * alignment requirement, that means we need to leave space + * for 12 (64 bytes). To simplify, we leave space for that. If + * at the end there are less, we pad up to the nearest + * multiple of 16. + */ + I2400M_TX_PLD_MAX = 12, + I2400M_TX_PLD_SIZE = sizeof(struct i2400m_msg_hdr) + + I2400M_TX_PLD_MAX * sizeof(struct i2400m_pld), + I2400M_TX_SKIP = 0x80000000, +}; + +#define TAIL_FULL ((void *)~(unsigned long)NULL) + +/* + * Allocate @size bytes in the TX fifo, return a pointer to it + * + * @i2400m: device descriptor + * @size: size of the buffer we need to allocate + * @padding: ensure that there is at least this many bytes of free + * contiguous space in the fifo. This is needed because later on + * we might need to add padding. + * + * Returns: + * + * Pointer to the allocated space. NULL if there is no + * space. TAIL_FULL if there is no space at the tail but there is at + * the head (Case B below). + * + * These are the two basic cases we need to keep an eye for -- it is + * much better explained in linux/kernel/kfifo.c, but this code + * basically does the same. No rocket science here. + * + * Case A Case B + * N ___________ ___________ + * | tail room | | data | + * | | | | + * |<- IN ->| |<- OUT ->| + * | | | | + * | data | | room | + * | | | | + * |<- OUT ->| |<- IN ->| + * | | | | + * | head room | | data | + * 0 ----------- ----------- + * + * We allocate only *contiguous* space. + * + * We can allocate only from 'room'. In Case B, it is simple; in case + * A, we only try from the tail room; if it is not enough, we just + * fail and return TAIL_FULL and let the caller figure out if we wants to + * skip the tail room and try to allocate from the head. + * + * Note: + * + * Assumes i2400m->tx_lock is taken, and we use that as a barrier + * + * The indexes keep increasing and we reset them to zero when we + * pop data off the queue + */ +static +void *i2400m_tx_fifo_push(struct i2400m *i2400m, size_t size, size_t padding) +{ + struct device *dev = i2400m_dev(i2400m); + size_t room, tail_room, needed_size; + void *ptr; + + needed_size = size + padding; + room = I2400M_TX_BUF_SIZE - (i2400m->tx_in - i2400m->tx_out); + if (room < needed_size) { /* this takes care of Case B */ + d_printf(2, dev, "fifo push %zu/%zu: no space\n", + size, padding); + return NULL; + } + /* Is there space at the tail? */ + tail_room = I2400M_TX_BUF_SIZE - i2400m->tx_in % I2400M_TX_BUF_SIZE; + if (tail_room < needed_size) { + if (i2400m->tx_out % I2400M_TX_BUF_SIZE + < i2400m->tx_in % I2400M_TX_BUF_SIZE) { + d_printf(2, dev, "fifo push %zu/%zu: tail full\n", + size, padding); + return TAIL_FULL; /* There might be head space */ + } else { + d_printf(2, dev, "fifo push %zu/%zu: no head space\n", + size, padding); + return NULL; /* There is no space */ + } + } + ptr = i2400m->tx_buf + i2400m->tx_in % I2400M_TX_BUF_SIZE; + d_printf(2, dev, "fifo push %zu/%zu: at @%zu\n", size, padding, + i2400m->tx_in % I2400M_TX_BUF_SIZE); + i2400m->tx_in += size; + return ptr; +} + + +/* + * Mark the tail of the FIFO buffer as 'to-skip' + * + * We should never hit the BUG_ON() because all the sizes we push to + * the FIFO are padded to be a multiple of 16 -- the size of *msg + * (I2400M_PL_PAD for the payloads, I2400M_TX_PLD_SIZE for the + * header). + * + * Note: + * + * Assumes i2400m->tx_lock is taken, and we use that as a barrier + */ +static +void i2400m_tx_skip_tail(struct i2400m *i2400m) +{ + struct device *dev = i2400m_dev(i2400m); + size_t tx_in = i2400m->tx_in % I2400M_TX_BUF_SIZE; + size_t tail_room = I2400M_TX_BUF_SIZE - tx_in; + struct i2400m_msg_hdr *msg = i2400m->tx_buf + tx_in; + BUG_ON(tail_room < sizeof(*msg)); + msg->size = tail_room | I2400M_TX_SKIP; + d_printf(2, dev, "skip tail: skipping %zu bytes @%zu\n", + tail_room, tx_in); + i2400m->tx_in += tail_room; +} + + +/* + * Check if a skb will fit in the TX queue's current active TX + * message (if there are still descriptors left unused). + * + * Returns: + * 0 if the message won't fit, 1 if it will. + * + * Note: + * + * Assumes a TX message is active (i2400m->tx_msg). + * + * Assumes i2400m->tx_lock is taken, and we use that as a barrier + */ +static +unsigned i2400m_tx_fits(struct i2400m *i2400m) +{ + struct i2400m_msg_hdr *msg_hdr = i2400m->tx_msg; + return le16_to_cpu(msg_hdr->num_pls) < I2400M_TX_PLD_MAX; + +} + + +/* + * Start a new TX message header in the queue. + * + * Reserve memory from the base FIFO engine and then just initialize + * the message header. + * + * We allocate the biggest TX message header we might need (one that'd + * fit I2400M_TX_PLD_MAX payloads) -- when it is closed it will be + * 'ironed it out' and the unneeded parts removed. + * + * NOTE: + * + * Assumes that the previous message is CLOSED (eg: either + * there was none or 'i2400m_tx_close()' was called on it). + * + * Assumes i2400m->tx_lock is taken, and we use that as a barrier + */ +static +void i2400m_tx_new(struct i2400m *i2400m) +{ + struct device *dev = i2400m_dev(i2400m); + struct i2400m_msg_hdr *tx_msg; + BUG_ON(i2400m->tx_msg != NULL); +try_head: + tx_msg = i2400m_tx_fifo_push(i2400m, I2400M_TX_PLD_SIZE, 0); + if (tx_msg == NULL) + goto out; + else if (tx_msg == TAIL_FULL) { + i2400m_tx_skip_tail(i2400m); + d_printf(2, dev, "new TX message: tail full, trying head\n"); + goto try_head; + } + memset(tx_msg, 0, I2400M_TX_PLD_SIZE); + tx_msg->size = I2400M_TX_PLD_SIZE; +out: + i2400m->tx_msg = tx_msg; + d_printf(2, dev, "new TX message: %p @%zu\n", + tx_msg, (void *) tx_msg - i2400m->tx_buf); +} + + +/* + * Finalize the current TX message header + * + * Sets the message header to be at the proper location depending on + * how many descriptors we have (check documentation at the file's + * header for more info on that). + * + * Appends padding bytes to make sure the whole TX message (counting + * from the 'relocated' message header) is aligned to + * tx_block_size. We assume the _append() code has left enough space + * in the FIFO for that. If there are no payloads, just pass, as it + * won't be transferred. + * + * The amount of padding bytes depends on how many payloads are in the + * TX message, as the "msg header and payload descriptors" will be + * shifted up in the buffer. + */ +static +void i2400m_tx_close(struct i2400m *i2400m) +{ + struct device *dev = i2400m_dev(i2400m); + struct i2400m_msg_hdr *tx_msg = i2400m->tx_msg; + struct i2400m_msg_hdr *tx_msg_moved; + size_t aligned_size, padding, hdr_size; + void *pad_buf; + + if (tx_msg->size & I2400M_TX_SKIP) /* a skipper? nothing to do */ + goto out; + + /* Relocate the message header + * + * Find the current header size, align it to 16 and if we need + * to move it so the tail is next to the payloads, move it and + * set the offset. + * + * If it moved, this header is good only for transmission; the + * original one (it is kept if we moved) is still used to + * figure out where the next TX message starts (and where the + * offset to the moved header is). + */ + hdr_size = sizeof(*tx_msg) + + le16_to_cpu(tx_msg->num_pls) * sizeof(tx_msg->pld[0]); + hdr_size = ALIGN(hdr_size, I2400M_PL_PAD); + tx_msg->offset = I2400M_TX_PLD_SIZE - hdr_size; + tx_msg_moved = (void *) tx_msg + tx_msg->offset; + memmove(tx_msg_moved, tx_msg, hdr_size); + tx_msg_moved->size -= tx_msg->offset; + /* + * Now figure out how much we have to add to the (moved!) + * message so the size is a multiple of i2400m->bus_tx_block_size. + */ + aligned_size = ALIGN(tx_msg_moved->size, i2400m->bus_tx_block_size); + padding = aligned_size - tx_msg_moved->size; + if (padding > 0) { + pad_buf = i2400m_tx_fifo_push(i2400m, padding, 0); + if (unlikely(WARN_ON(pad_buf == NULL + || pad_buf == TAIL_FULL))) { + /* This should not happen -- append should verify + * there is always space left at least to append + * tx_block_size */ + dev_err(dev, + "SW BUG! Possible data leakage from memory the " + "device should not read for padding - " + "size %lu aligned_size %zu tx_buf %p in " + "%zu out %zu\n", + (unsigned long) tx_msg_moved->size, + aligned_size, i2400m->tx_buf, i2400m->tx_in, + i2400m->tx_out); + } else + memset(pad_buf, 0xad, padding); + } + tx_msg_moved->padding = cpu_to_le16(padding); + tx_msg_moved->size += padding; + if (tx_msg != tx_msg_moved) + tx_msg->size += padding; +out: + i2400m->tx_msg = NULL; +} + + +/** + * i2400m_tx - send the data in a buffer to the device + * + * @buf: pointer to the buffer to transmit + * + * @buf_len: buffer size + * + * @pl_type: type of the payload we are sending. + * + * Returns: + * 0 if ok, < 0 errno code on error (-ENOSPC, if there is no more + * room for the message in the queue). + * + * Appends the buffer to the TX FIFO and notifies the bus-specific + * part of the driver that there is new data ready to transmit. + * Once this function returns, the buffer has been copied, so it can + * be reused. + * + * The steps followed to append are explained in detail in the file + * header. + * + * Whenever we write to a message, we increase msg->size, so it + * reflects exactly how big the message is. This is needed so that if + * we concatenate two messages before they can be sent, the code that + * sends the messages can find the boundaries (and it will replace the + * size with the real barker before sending). + * + * Note: + * + * Cold and warm reset payloads need to be sent as a single + * payload, so we handle that. + */ +int i2400m_tx(struct i2400m *i2400m, const void *buf, size_t buf_len, + enum i2400m_pt pl_type) +{ + int result = -ENOSPC; + struct device *dev = i2400m_dev(i2400m); + unsigned long flags; + size_t padded_len; + void *ptr; + unsigned is_singleton = pl_type == I2400M_PT_RESET_WARM + || pl_type == I2400M_PT_RESET_COLD; + + d_fnstart(3, dev, "(i2400m %p skb %p [%zu bytes] pt %u)\n", + i2400m, buf, buf_len, pl_type); + padded_len = ALIGN(buf_len, I2400M_PL_PAD); + d_printf(5, dev, "padded_len %zd buf_len %zd\n", padded_len, buf_len); + /* If there is no current TX message, create one; if the + * current one is out of payload slots or we have a singleton, + * close it and start a new one */ + spin_lock_irqsave(&i2400m->tx_lock, flags); +try_new: + if (unlikely(i2400m->tx_msg == NULL)) + i2400m_tx_new(i2400m); + else if (unlikely(!i2400m_tx_fits(i2400m) + || (is_singleton && i2400m->tx_msg->num_pls != 0))) { + d_printf(2, dev, "closing TX message (fits %u singleton " + "%u num_pls %u)\n", i2400m_tx_fits(i2400m), + is_singleton, i2400m->tx_msg->num_pls); + i2400m_tx_close(i2400m); + i2400m_tx_new(i2400m); + } + if (i2400m->tx_msg->size + padded_len > I2400M_TX_BUF_SIZE / 2) { + d_printf(2, dev, "TX: message too big, going new\n"); + i2400m_tx_close(i2400m); + i2400m_tx_new(i2400m); + } + if (i2400m->tx_msg == NULL) + goto error_tx_new; + /* So we have a current message header; now append space for + * the message -- if there is not enough, try the head */ + ptr = i2400m_tx_fifo_push(i2400m, padded_len, + i2400m->bus_tx_block_size); + if (ptr == TAIL_FULL) { /* Tail is full, try head */ + d_printf(2, dev, "pl append: tail full\n"); + i2400m_tx_close(i2400m); + i2400m_tx_skip_tail(i2400m); + goto try_new; + } else if (ptr == NULL) { /* All full */ + result = -ENOSPC; + d_printf(2, dev, "pl append: all full\n"); + } else { /* Got space, copy it, set padding */ + struct i2400m_msg_hdr *tx_msg = i2400m->tx_msg; + unsigned num_pls = le16_to_cpu(tx_msg->num_pls); + memcpy(ptr, buf, buf_len); + memset(ptr + buf_len, 0xad, padded_len - buf_len); + i2400m_pld_set(&tx_msg->pld[num_pls], buf_len, pl_type); + d_printf(3, dev, "pld 0x%08x (type 0x%1x len 0x%04zx\n", + le32_to_cpu(tx_msg->pld[num_pls].val), + pl_type, buf_len); + tx_msg->num_pls = le16_to_cpu(num_pls+1); + tx_msg->size += padded_len; + d_printf(2, dev, "TX: appended %zu b (up to %u b) pl #%u \n", + padded_len, tx_msg->size, num_pls+1); + d_printf(2, dev, + "TX: appended hdr @%zu %zu b pl #%u @%zu %zu/%zu b\n", + (void *)tx_msg - i2400m->tx_buf, (size_t)tx_msg->size, + num_pls+1, ptr - i2400m->tx_buf, buf_len, padded_len); + result = 0; + if (is_singleton) + i2400m_tx_close(i2400m); + } +error_tx_new: + spin_unlock_irqrestore(&i2400m->tx_lock, flags); + i2400m->bus_tx_kick(i2400m); /* always kick, might free up space */ + d_fnend(3, dev, "(i2400m %p skb %p [%zu bytes] pt %u) = %d\n", + i2400m, buf, buf_len, pl_type, result); + return result; +} +EXPORT_SYMBOL_GPL(i2400m_tx); + + +/** + * i2400m_tx_msg_get - Get the first TX message in the FIFO to start sending it + * + * @i2400m: device descriptors + * @bus_size: where to place the size of the TX message + * + * Called by the bus-specific driver to get the first TX message at + * the FIF that is ready for transmission. + * + * It sets the state in @i2400m to indicate the bus-specific driver is + * transfering that message (i2400m->tx_msg_size). + * + * Once the transfer is completed, call i2400m_tx_msg_sent(). + * + * Notes: + * + * The size of the TX message to be transmitted might be smaller than + * that of the TX message in the FIFO (in case the header was + * shorter). Hence, we copy it in @bus_size, for the bus layer to + * use. We keep the message's size in i2400m->tx_msg_size so that + * when the bus later is done transferring we know how much to + * advance the fifo. + * + * We collect statistics here as all the data is available and we + * assume it is going to work [see i2400m_tx_msg_sent()]. + */ +struct i2400m_msg_hdr *i2400m_tx_msg_get(struct i2400m *i2400m, + size_t *bus_size) +{ + struct device *dev = i2400m_dev(i2400m); + struct i2400m_msg_hdr *tx_msg, *tx_msg_moved; + unsigned long flags, pls; + + d_fnstart(3, dev, "(i2400m %p bus_size %p)\n", i2400m, bus_size); + spin_lock_irqsave(&i2400m->tx_lock, flags); +skip: + tx_msg_moved = NULL; + if (i2400m->tx_in == i2400m->tx_out) { /* Empty FIFO? */ + i2400m->tx_in = 0; + i2400m->tx_out = 0; + d_printf(2, dev, "TX: FIFO empty: resetting\n"); + goto out_unlock; + } + tx_msg = i2400m->tx_buf + i2400m->tx_out % I2400M_TX_BUF_SIZE; + if (tx_msg->size & I2400M_TX_SKIP) { /* skip? */ + d_printf(2, dev, "TX: skip: msg @%zu (%zu b)\n", + i2400m->tx_out % I2400M_TX_BUF_SIZE, + (size_t) tx_msg->size & ~I2400M_TX_SKIP); + i2400m->tx_out += tx_msg->size & ~I2400M_TX_SKIP; + goto skip; + } + + if (tx_msg->num_pls == 0) { /* No payloads? */ + if (tx_msg == i2400m->tx_msg) { /* open, we are done */ + d_printf(2, dev, + "TX: FIFO empty: open msg w/o payloads @%zu\n", + (void *) tx_msg - i2400m->tx_buf); + tx_msg = NULL; + goto out_unlock; + } else { /* closed, skip it */ + d_printf(2, dev, + "TX: skip msg w/o payloads @%zu (%zu b)\n", + (void *) tx_msg - i2400m->tx_buf, + (size_t) tx_msg->size); + i2400m->tx_out += tx_msg->size & ~I2400M_TX_SKIP; + goto skip; + } + } + if (tx_msg == i2400m->tx_msg) /* open msg? */ + i2400m_tx_close(i2400m); + + /* Now we have a valid TX message (with payloads) to TX */ + tx_msg_moved = (void *) tx_msg + tx_msg->offset; + i2400m->tx_msg_size = tx_msg->size; + *bus_size = tx_msg_moved->size; + d_printf(2, dev, "TX: pid %d msg hdr at @%zu offset +@%zu " + "size %zu bus_size %zu\n", + current->pid, (void *) tx_msg - i2400m->tx_buf, + (size_t) tx_msg->offset, (size_t) tx_msg->size, + (size_t) tx_msg_moved->size); + tx_msg_moved->barker = le32_to_cpu(I2400M_H2D_PREVIEW_BARKER); + tx_msg_moved->sequence = le32_to_cpu(i2400m->tx_sequence++); + + pls = le32_to_cpu(tx_msg_moved->num_pls); + i2400m->tx_pl_num += pls; /* Update stats */ + if (pls > i2400m->tx_pl_max) + i2400m->tx_pl_max = pls; + if (pls < i2400m->tx_pl_min) + i2400m->tx_pl_min = pls; + i2400m->tx_num++; + i2400m->tx_size_acc += *bus_size; + if (*bus_size < i2400m->tx_size_min) + i2400m->tx_size_min = *bus_size; + if (*bus_size > i2400m->tx_size_max) + i2400m->tx_size_max = *bus_size; +out_unlock: + spin_unlock_irqrestore(&i2400m->tx_lock, flags); + d_fnstart(3, dev, "(i2400m %p bus_size %p [%zu]) = %p\n", + i2400m, bus_size, *bus_size, tx_msg_moved); + return tx_msg_moved; +} +EXPORT_SYMBOL_GPL(i2400m_tx_msg_get); + + +/** + * i2400m_tx_msg_sent - indicate the transmission of a TX message + * + * @i2400m: device descriptor + * + * Called by the bus-specific driver when a message has been sent; + * this pops it from the FIFO; and as there is space, start the queue + * in case it was stopped. + * + * Should be called even if the message send failed and we are + * dropping this TX message. + */ +void i2400m_tx_msg_sent(struct i2400m *i2400m) +{ + unsigned n; + unsigned long flags; + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + spin_lock_irqsave(&i2400m->tx_lock, flags); + i2400m->tx_out += i2400m->tx_msg_size; + d_printf(2, dev, "TX: sent %zu b\n", (size_t) i2400m->tx_msg_size); + i2400m->tx_msg_size = 0; + BUG_ON(i2400m->tx_out > i2400m->tx_in); + /* level them FIFO markers off */ + n = i2400m->tx_out / I2400M_TX_BUF_SIZE; + i2400m->tx_out %= I2400M_TX_BUF_SIZE; + i2400m->tx_in -= n * I2400M_TX_BUF_SIZE; + netif_start_queue(i2400m->wimax_dev.net_dev); + spin_unlock_irqrestore(&i2400m->tx_lock, flags); + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); +} +EXPORT_SYMBOL_GPL(i2400m_tx_msg_sent); + + +/** + * i2400m_tx_setup - Initialize the TX queue and infrastructure + * + * Make sure we reset the TX sequence to zero, as when this function + * is called, the firmware has been just restarted. + */ +int i2400m_tx_setup(struct i2400m *i2400m) +{ + int result; + + /* Do this here only once -- can't do on + * i2400m_hard_start_xmit() as we'll cause race conditions if + * the WS was scheduled on another CPU */ + INIT_WORK(&i2400m->wake_tx_ws, i2400m_wake_tx_work); + + i2400m->tx_sequence = 0; + i2400m->tx_buf = kmalloc(I2400M_TX_BUF_SIZE, GFP_KERNEL); + if (i2400m->tx_buf == NULL) + result = -ENOMEM; + else + result = 0; + /* Huh? the bus layer has to define this... */ + BUG_ON(i2400m->bus_tx_block_size == 0); + return result; + +} + + +/** + * i2400m_tx_release - Tear down the TX queue and infrastructure + */ +void i2400m_tx_release(struct i2400m *i2400m) +{ + kfree(i2400m->tx_buf); +} From 3a35a1d0bdf7cc32cddc234b956605e6d4db4673 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:48 -0800 Subject: [PATCH 111/123] i2400m: various functions for device management This is a collection of functions used to control the device (plus a few helpers). There are utilities for handling TLV buffers, hooks on the device's reports to act on device changes of state [i2400m_report_hook()], on acks to commands [i2400m_msg_ack_hook()], a helper for sending commands to the device and blocking until a reply arrives [i2400m_msg_to_dev()], a few high level commands for manipulating the device state, powersaving mode and configuration plus the routines to setup the device once communication is established with it [i2400m_dev_initialize()]. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/control.c | 1291 ++++++++++++++++++++++++++++ 1 file changed, 1291 insertions(+) create mode 100644 drivers/net/wimax/i2400m/control.c diff --git a/drivers/net/wimax/i2400m/control.c b/drivers/net/wimax/i2400m/control.c new file mode 100644 index 00000000000..d3d37fed689 --- /dev/null +++ b/drivers/net/wimax/i2400m/control.c @@ -0,0 +1,1291 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Miscellaneous control functions for managing the device + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Inaky Perez-Gonzalez + * - Initial implementation + * + * This is a collection of functions used to control the device (plus + * a few helpers). + * + * There are utilities for handling TLV buffers, hooks on the device's + * reports to act on device changes of state [i2400m_report_hook()], + * on acks to commands [i2400m_msg_ack_hook()], a helper for sending + * commands to the device and blocking until a reply arrives + * [i2400m_msg_to_dev()], a few high level commands for manipulating + * the device state, powersving mode and configuration plus the + * routines to setup the device once communication is stablished with + * it [i2400m_dev_initialize()]. + * + * ROADMAP + * + * i2400m_dev_initalize() Called by i2400m_dev_start() + * i2400m_set_init_config() + * i2400m_firmware_check() + * i2400m_cmd_get_state() + * i2400m_dev_shutdown() Called by i2400m_dev_stop() + * i2400m->bus_reset() + * + * i2400m_{cmd,get,set}_*() + * i2400m_msg_to_dev() + * i2400m_msg_check_status() + * + * i2400m_report_hook() Called on reception of an event + * i2400m_report_state_hook() + * i2400m_tlv_buffer_walk() + * i2400m_tlv_match() + * i2400m_report_tlv_system_state() + * i2400m_report_tlv_rf_switches_status() + * i2400m_report_tlv_media_status() + * i2400m_cmd_enter_powersave() + * + * i2400m_msg_ack_hook() Called on reception of a reply to a + * command, get or set + */ + +#include +#include "i2400m.h" +#include +#include + + +#define D_SUBMODULE control +#include "debug-levels.h" + + +/* + * Return if a TLV is of a give type and size + * + * @tlv_hdr: pointer to the TLV + * @tlv_type: type of the TLV we are looking for + * @tlv_size: expected size of the TLV we are looking for (if -1, + * don't check the size). This includes the header + * Returns: 0 if the TLV matches + * < 0 if it doesn't match at all + * > 0 total TLV + payload size, if the type matches, but not + * the size + */ +static +ssize_t i2400m_tlv_match(const struct i2400m_tlv_hdr *tlv, + enum i2400m_tlv tlv_type, ssize_t tlv_size) +{ + if (le16_to_cpu(tlv->type) != tlv_type) /* Not our type? skip */ + return -1; + if (tlv_size != -1 + && le16_to_cpu(tlv->length) + sizeof(*tlv) != tlv_size) { + size_t size = le16_to_cpu(tlv->length) + sizeof(*tlv); + printk(KERN_WARNING "W: tlv type 0x%x mismatched because of " + "size (got %zu vs %zu expected)\n", + tlv_type, size, tlv_size); + return size; + } + return 0; +} + + +/* + * Given a buffer of TLVs, iterate over them + * + * @i2400m: device instance + * @tlv_buf: pointer to the beginning of the TLV buffer + * @buf_size: buffer size in bytes + * @tlv_pos: seek position; this is assumed to be a pointer returned + * by i2400m_tlv_buffer_walk() [and thus, validated]. The + * TLV returned will be the one following this one. + * + * Usage: + * + * tlv_itr = NULL; + * while (tlv_itr = i2400m_tlv_buffer_walk(i2400m, buf, size, tlv_itr)) { + * ... + * // Do stuff with tlv_itr, DON'T MODIFY IT + * ... + * } + */ +static +const struct i2400m_tlv_hdr *i2400m_tlv_buffer_walk( + struct i2400m *i2400m, + const void *tlv_buf, size_t buf_size, + const struct i2400m_tlv_hdr *tlv_pos) +{ + struct device *dev = i2400m_dev(i2400m); + const struct i2400m_tlv_hdr *tlv_top = tlv_buf + buf_size; + size_t offset, length, avail_size; + unsigned type; + + if (tlv_pos == NULL) /* Take the first one? */ + tlv_pos = tlv_buf; + else /* Nope, the next one */ + tlv_pos = (void *) tlv_pos + + le16_to_cpu(tlv_pos->length) + sizeof(*tlv_pos); + if (tlv_pos == tlv_top) { /* buffer done */ + tlv_pos = NULL; + goto error_beyond_end; + } + if (tlv_pos > tlv_top) { + tlv_pos = NULL; + WARN_ON(1); + goto error_beyond_end; + } + offset = (void *) tlv_pos - (void *) tlv_buf; + avail_size = buf_size - offset; + if (avail_size < sizeof(*tlv_pos)) { + dev_err(dev, "HW BUG? tlv_buf %p [%zu bytes], tlv @%zu: " + "short header\n", tlv_buf, buf_size, offset); + goto error_short_header; + } + type = le16_to_cpu(tlv_pos->type); + length = le16_to_cpu(tlv_pos->length); + if (avail_size < sizeof(*tlv_pos) + length) { + dev_err(dev, "HW BUG? tlv_buf %p [%zu bytes], " + "tlv type 0x%04x @%zu: " + "short data (%zu bytes vs %zu needed)\n", + tlv_buf, buf_size, type, offset, avail_size, + sizeof(*tlv_pos) + length); + goto error_short_header; + } +error_short_header: +error_beyond_end: + return tlv_pos; +} + + +/* + * Find a TLV in a buffer of sequential TLVs + * + * @i2400m: device descriptor + * @tlv_hdr: pointer to the first TLV in the sequence + * @size: size of the buffer in bytes; all TLVs are assumed to fit + * fully in the buffer (otherwise we'll complain). + * @tlv_type: type of the TLV we are looking for + * @tlv_size: expected size of the TLV we are looking for (if -1, + * don't check the size). This includes the header + * + * Returns: NULL if the TLV is not found, otherwise a pointer to + * it. If the sizes don't match, an error is printed and NULL + * returned. + */ +static +const struct i2400m_tlv_hdr *i2400m_tlv_find( + struct i2400m *i2400m, + const struct i2400m_tlv_hdr *tlv_hdr, size_t size, + enum i2400m_tlv tlv_type, ssize_t tlv_size) +{ + ssize_t match; + struct device *dev = i2400m_dev(i2400m); + const struct i2400m_tlv_hdr *tlv = NULL; + while ((tlv = i2400m_tlv_buffer_walk(i2400m, tlv_hdr, size, tlv))) { + match = i2400m_tlv_match(tlv, tlv_type, tlv_size); + if (match == 0) /* found it :) */ + break; + if (match > 0) + dev_warn(dev, "TLV type 0x%04x found with size " + "mismatch (%zu vs %zu needed)\n", + tlv_type, match, tlv_size); + } + return tlv; +} + + +static const struct +{ + char *msg; + int errno; +} ms_to_errno[I2400M_MS_MAX] = { + [I2400M_MS_DONE_OK] = { "", 0 }, + [I2400M_MS_DONE_IN_PROGRESS] = { "", 0 }, + [I2400M_MS_INVALID_OP] = { "invalid opcode", -ENOSYS }, + [I2400M_MS_BAD_STATE] = { "invalid state", -EILSEQ }, + [I2400M_MS_ILLEGAL_VALUE] = { "illegal value", -EINVAL }, + [I2400M_MS_MISSING_PARAMS] = { "missing parameters", -ENOMSG }, + [I2400M_MS_VERSION_ERROR] = { "bad version", -EIO }, + [I2400M_MS_ACCESSIBILITY_ERROR] = { "accesibility error", -EIO }, + [I2400M_MS_BUSY] = { "busy", -EBUSY }, + [I2400M_MS_CORRUPTED_TLV] = { "corrupted TLV", -EILSEQ }, + [I2400M_MS_UNINITIALIZED] = { "not unitialized", -EILSEQ }, + [I2400M_MS_UNKNOWN_ERROR] = { "unknown error", -EIO }, + [I2400M_MS_PRODUCTION_ERROR] = { "production error", -EIO }, + [I2400M_MS_NO_RF] = { "no RF", -EIO }, + [I2400M_MS_NOT_READY_FOR_POWERSAVE] = + { "not ready for powersave", -EACCES }, + [I2400M_MS_THERMAL_CRITICAL] = { "thermal critical", -EL3HLT }, +}; + + +/* + * i2400m_msg_check_status - translate a message's status code + * + * @i2400m: device descriptor + * @l3l4_hdr: message header + * @strbuf: buffer to place a formatted error message (unless NULL). + * @strbuf_size: max amount of available space; larger messages will + * be truncated. + * + * Returns: errno code corresponding to the status code in @l3l4_hdr + * and a message in @strbuf describing the error. + */ +int i2400m_msg_check_status(const struct i2400m_l3l4_hdr *l3l4_hdr, + char *strbuf, size_t strbuf_size) +{ + int result; + enum i2400m_ms status = le16_to_cpu(l3l4_hdr->status); + const char *str; + + if (status == 0) + return 0; + if (status > ARRAY_SIZE(ms_to_errno)) { + str = "unknown status code"; + result = -EBADR; + } else { + str = ms_to_errno[status].msg; + result = ms_to_errno[status].errno; + } + if (strbuf) + snprintf(strbuf, strbuf_size, "%s (%d)", str, status); + return result; +} + + +/* + * Act on a TLV System State reported by the device + * + * @i2400m: device descriptor + * @ss: validated System State TLV + */ +static +void i2400m_report_tlv_system_state(struct i2400m *i2400m, + const struct i2400m_tlv_system_state *ss) +{ + struct device *dev = i2400m_dev(i2400m); + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + enum i2400m_system_state i2400m_state = le32_to_cpu(ss->state); + + d_fnstart(3, dev, "(i2400m %p ss %p [%u])\n", i2400m, ss, i2400m_state); + + if (unlikely(i2400m->ready == 0)) /* act if up */ + goto out; + if (i2400m->state != i2400m_state) { + i2400m->state = i2400m_state; + wake_up_all(&i2400m->state_wq); + } + switch (i2400m_state) { + case I2400M_SS_UNINITIALIZED: + case I2400M_SS_INIT: + case I2400M_SS_CONFIG: + case I2400M_SS_PRODUCTION: + wimax_state_change(wimax_dev, WIMAX_ST_UNINITIALIZED); + break; + + case I2400M_SS_RF_OFF: + case I2400M_SS_RF_SHUTDOWN: + wimax_state_change(wimax_dev, WIMAX_ST_RADIO_OFF); + break; + + case I2400M_SS_READY: + case I2400M_SS_STANDBY: + case I2400M_SS_SLEEPACTIVE: + wimax_state_change(wimax_dev, WIMAX_ST_READY); + break; + + case I2400M_SS_CONNECTING: + case I2400M_SS_WIMAX_CONNECTED: + wimax_state_change(wimax_dev, WIMAX_ST_READY); + break; + + case I2400M_SS_SCAN: + case I2400M_SS_OUT_OF_ZONE: + wimax_state_change(wimax_dev, WIMAX_ST_SCANNING); + break; + + case I2400M_SS_IDLE: + d_printf(1, dev, "entering BS-negotiated idle mode\n"); + case I2400M_SS_DISCONNECTING: + case I2400M_SS_DATA_PATH_CONNECTED: + wimax_state_change(wimax_dev, WIMAX_ST_CONNECTED); + break; + + default: + /* Huh? just in case, shut it down */ + dev_err(dev, "HW BUG? unknown state %u: shutting down\n", + i2400m_state); + i2400m->bus_reset(i2400m, I2400M_RT_WARM); + break; + }; +out: + d_fnend(3, dev, "(i2400m %p ss %p [%u]) = void\n", + i2400m, ss, i2400m_state); +} + + +/* + * Parse and act on a TLV Media Status sent by the device + * + * @i2400m: device descriptor + * @ms: validated Media Status TLV + * + * This will set the carrier up on down based on the device's link + * report. This is done asides of what the WiMAX stack does based on + * the device's state as sometimes we need to do a link-renew (the BS + * wants us to renew a DHCP lease, for example). + * + * In fact, doc says that everytime we get a link-up, we should do a + * DHCP negotiation... + */ +static +void i2400m_report_tlv_media_status(struct i2400m *i2400m, + const struct i2400m_tlv_media_status *ms) +{ + struct device *dev = i2400m_dev(i2400m); + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + struct net_device *net_dev = wimax_dev->net_dev; + enum i2400m_media_status status = le32_to_cpu(ms->media_status); + + d_fnstart(3, dev, "(i2400m %p ms %p [%u])\n", i2400m, ms, status); + + if (unlikely(i2400m->ready == 0)) /* act if up */ + goto out; + switch (status) { + case I2400M_MEDIA_STATUS_LINK_UP: + netif_carrier_on(net_dev); + break; + case I2400M_MEDIA_STATUS_LINK_DOWN: + netif_carrier_off(net_dev); + break; + /* + * This is the network telling us we need to retrain the DHCP + * lease -- so far, we are trusting the WiMAX Network Service + * in user space to pick this up and poke the DHCP client. + */ + case I2400M_MEDIA_STATUS_LINK_RENEW: + netif_carrier_on(net_dev); + break; + default: + dev_err(dev, "HW BUG? unknown media status %u\n", + status); + }; +out: + d_fnend(3, dev, "(i2400m %p ms %p [%u]) = void\n", + i2400m, ms, status); +} + + +/* + * Parse a 'state report' and extract carrier on/off information + * + * @i2400m: device descriptor + * @l3l4_hdr: pointer to message; it has been already validated for + * consistent size. + * @size: size of the message (header + payload). The header length + * declaration is assumed to be congruent with @size (as in + * sizeof(*l3l4_hdr) + l3l4_hdr->length == size) + * + * Extract from the report state the system state TLV and infer from + * there if we have a carrier or not. Update our local state and tell + * netdev. + * + * When setting the carrier, it's fine to set OFF twice (for example), + * as netif_carrier_off() will not generate two OFF events (just on + * the transitions). + */ +static +void i2400m_report_state_hook(struct i2400m *i2400m, + const struct i2400m_l3l4_hdr *l3l4_hdr, + size_t size, const char *tag) +{ + struct device *dev = i2400m_dev(i2400m); + const struct i2400m_tlv_hdr *tlv; + const struct i2400m_tlv_system_state *ss; + const struct i2400m_tlv_rf_switches_status *rfss; + const struct i2400m_tlv_media_status *ms; + size_t tlv_size = le16_to_cpu(l3l4_hdr->length); + + d_fnstart(4, dev, "(i2400m %p, l3l4_hdr %p, size %zu, %s)\n", + i2400m, l3l4_hdr, size, tag); + tlv = NULL; + + while ((tlv = i2400m_tlv_buffer_walk(i2400m, &l3l4_hdr->pl, + tlv_size, tlv))) { + if (0 == i2400m_tlv_match(tlv, I2400M_TLV_SYSTEM_STATE, + sizeof(*ss))) { + ss = container_of(tlv, typeof(*ss), hdr); + d_printf(2, dev, "%s: system state TLV " + "found (0x%04x), state 0x%08x\n", + tag, I2400M_TLV_SYSTEM_STATE, + le32_to_cpu(ss->state)); + i2400m_report_tlv_system_state(i2400m, ss); + } + if (0 == i2400m_tlv_match(tlv, I2400M_TLV_RF_STATUS, + sizeof(*rfss))) { + rfss = container_of(tlv, typeof(*rfss), hdr); + d_printf(2, dev, "%s: RF status TLV " + "found (0x%04x), sw 0x%02x hw 0x%02x\n", + tag, I2400M_TLV_RF_STATUS, + le32_to_cpu(rfss->sw_rf_switch), + le32_to_cpu(rfss->hw_rf_switch)); + i2400m_report_tlv_rf_switches_status(i2400m, rfss); + } + if (0 == i2400m_tlv_match(tlv, I2400M_TLV_MEDIA_STATUS, + sizeof(*ms))) { + ms = container_of(tlv, typeof(*ms), hdr); + d_printf(2, dev, "%s: Media Status TLV: %u\n", + tag, le32_to_cpu(ms->media_status)); + i2400m_report_tlv_media_status(i2400m, ms); + } + } + d_fnend(4, dev, "(i2400m %p, l3l4_hdr %p, size %zu, %s) = void\n", + i2400m, l3l4_hdr, size, tag); +} + + +/* + * i2400m_report_hook - (maybe) act on a report + * + * @i2400m: device descriptor + * @l3l4_hdr: pointer to message; it has been already validated for + * consistent size. + * @size: size of the message (header + payload). The header length + * declaration is assumed to be congruent with @size (as in + * sizeof(*l3l4_hdr) + l3l4_hdr->length == size) + * + * Extract information we might need (like carrien on/off) from a + * device report. + */ +void i2400m_report_hook(struct i2400m *i2400m, + const struct i2400m_l3l4_hdr *l3l4_hdr, size_t size) +{ + struct device *dev = i2400m_dev(i2400m); + unsigned msg_type; + + d_fnstart(3, dev, "(i2400m %p l3l4_hdr %p size %zu)\n", + i2400m, l3l4_hdr, size); + /* Chew on the message, we might need some information from + * here */ + msg_type = le16_to_cpu(l3l4_hdr->type); + switch (msg_type) { + case I2400M_MT_REPORT_STATE: /* carrier detection... */ + i2400m_report_state_hook(i2400m, + l3l4_hdr, size, "REPORT STATE"); + break; + /* If the device is ready for power save, then ask it to do + * it. */ + case I2400M_MT_REPORT_POWERSAVE_READY: /* zzzzz */ + if (l3l4_hdr->status == cpu_to_le16(I2400M_MS_DONE_OK)) { + d_printf(1, dev, "ready for powersave, requesting\n"); + i2400m_cmd_enter_powersave(i2400m); + } + break; + }; + d_fnend(3, dev, "(i2400m %p l3l4_hdr %p size %zu) = void\n", + i2400m, l3l4_hdr, size); +} + + +/* + * i2400m_msg_ack_hook - process cmd/set/get ack for internal status + * + * @i2400m: device descriptor + * @l3l4_hdr: pointer to message; it has been already validated for + * consistent size. + * @size: size of the message + * + * Extract information we might need from acks to commands and act on + * it. This is akin to i2400m_report_hook(). Note most of this + * processing should be done in the function that calls the + * command. This is here for some cases where it can't happen... + */ +void i2400m_msg_ack_hook(struct i2400m *i2400m, + const struct i2400m_l3l4_hdr *l3l4_hdr, size_t size) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + unsigned ack_type, ack_status; + char strerr[32]; + + /* Chew on the message, we might need some information from + * here */ + ack_type = le16_to_cpu(l3l4_hdr->type); + ack_status = le16_to_cpu(l3l4_hdr->status); + switch (ack_type) { + case I2400M_MT_CMD_ENTER_POWERSAVE: + /* This is just left here for the sake of example, as + * the processing is done somewhere else. */ + if (0) { + result = i2400m_msg_check_status( + l3l4_hdr, strerr, sizeof(strerr)); + if (result >= 0) + d_printf(1, dev, "ready for power save: %zd\n", + size); + } + break; + }; + return; +} + + +/* + * i2400m_msg_size_check() - verify message size and header are congruent + * + * It is ok if the total message size is larger than the expected + * size, as there can be padding. + */ +int i2400m_msg_size_check(struct i2400m *i2400m, + const struct i2400m_l3l4_hdr *l3l4_hdr, + size_t msg_size) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + size_t expected_size; + d_fnstart(4, dev, "(i2400m %p l3l4_hdr %p msg_size %zu)\n", + i2400m, l3l4_hdr, msg_size); + if (msg_size < sizeof(*l3l4_hdr)) { + dev_err(dev, "bad size for message header " + "(expected at least %zu, got %zu)\n", + (size_t) sizeof(*l3l4_hdr), msg_size); + result = -EIO; + goto error_hdr_size; + } + expected_size = le16_to_cpu(l3l4_hdr->length) + sizeof(*l3l4_hdr); + if (msg_size < expected_size) { + dev_err(dev, "bad size for message code 0x%04x (expected %zu, " + "got %zu)\n", le16_to_cpu(l3l4_hdr->type), + expected_size, msg_size); + result = -EIO; + } else + result = 0; +error_hdr_size: + d_fnend(4, dev, + "(i2400m %p l3l4_hdr %p msg_size %zu) = %d\n", + i2400m, l3l4_hdr, msg_size, result); + return result; +} + + + +/* + * Cancel a wait for a command ACK + * + * @i2400m: device descriptor + * @code: [negative] errno code to cancel with (don't use + * -EINPROGRESS) + * + * If there is an ack already filled out, free it. + */ +void i2400m_msg_to_dev_cancel_wait(struct i2400m *i2400m, int code) +{ + struct sk_buff *ack_skb; + unsigned long flags; + + spin_lock_irqsave(&i2400m->rx_lock, flags); + ack_skb = i2400m->ack_skb; + if (ack_skb && !IS_ERR(ack_skb)) + kfree(ack_skb); + i2400m->ack_skb = ERR_PTR(code); + spin_unlock_irqrestore(&i2400m->rx_lock, flags); +} + + +/** + * i2400m_msg_to_dev - Send a control message to the device and get a response + * + * @i2400m: device descriptor + * + * @msg_skb: an skb * + * + * @buf: pointer to the buffer containing the message to be sent; it + * has to start with a &struct i2400M_l3l4_hdr and then + * followed by the payload. Once this function returns, the + * buffer can be reused. + * + * @buf_len: buffer size + * + * Returns: + * + * Pointer to skb containing the ack message. You need to check the + * pointer with IS_ERR(), as it might be an error code. Error codes + * could happen because: + * + * - the message wasn't formatted correctly + * - couldn't send the message + * - failed waiting for a response + * - the ack message wasn't formatted correctly + * + * The returned skb has been allocated with wimax_msg_to_user_alloc(), + * it contains the reponse in a netlink attribute and is ready to be + * passed up to user space with wimax_msg_to_user_send(). To access + * the payload and its length, use wimax_msg_{data,len}() on the skb. + * + * The skb has to be freed with kfree_skb() once done. + * + * Description: + * + * This function delivers a message/command to the device and waits + * for an ack to be received. The format is described in + * linux/wimax/i2400m.h. In summary, a command/get/set is followed by an + * ack. + * + * This function will not check the ack status, that's left up to the + * caller. Once done with the ack skb, it has to be kfree_skb()ed. + * + * The i2400m handles only one message at the same time, thus we need + * the mutex to exclude other players. + * + * We write the message and then wait for an answer to come back. The + * RX path intercepts control messages and handles them in + * i2400m_rx_ctl(). Reports (notifications) are (maybe) processed + * locally and then forwarded (as needed) to user space on the WiMAX + * stack message pipe. Acks are saved and passed back to us through an + * skb in i2400m->ack_skb which is ready to be given to generic + * netlink if need be. + */ +struct sk_buff *i2400m_msg_to_dev(struct i2400m *i2400m, + const void *buf, size_t buf_len) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + const struct i2400m_l3l4_hdr *msg_l3l4_hdr; + struct sk_buff *ack_skb; + const struct i2400m_l3l4_hdr *ack_l3l4_hdr; + size_t ack_len; + int ack_timeout; + unsigned msg_type; + unsigned long flags; + + d_fnstart(3, dev, "(i2400m %p buf %p len %zu)\n", + i2400m, buf, buf_len); + + if (i2400m->boot_mode) + return ERR_PTR(-ENODEV); + + msg_l3l4_hdr = buf; + /* Check msg & payload consistency */ + result = i2400m_msg_size_check(i2400m, msg_l3l4_hdr, buf_len); + if (result < 0) + goto error_bad_msg; + msg_type = le16_to_cpu(msg_l3l4_hdr->type); + d_printf(1, dev, "CMD/GET/SET 0x%04x %zu bytes\n", + msg_type, buf_len); + d_dump(2, dev, buf, buf_len); + + /* Setup the completion, ack_skb ("we are waiting") and send + * the message to the device */ + mutex_lock(&i2400m->msg_mutex); + spin_lock_irqsave(&i2400m->rx_lock, flags); + i2400m->ack_skb = ERR_PTR(-EINPROGRESS); + spin_unlock_irqrestore(&i2400m->rx_lock, flags); + init_completion(&i2400m->msg_completion); + result = i2400m_tx(i2400m, buf, buf_len, I2400M_PT_CTRL); + if (result < 0) { + dev_err(dev, "can't send message 0x%04x: %d\n", + le16_to_cpu(msg_l3l4_hdr->type), result); + goto error_tx; + } + + /* Some commands take longer to execute because of crypto ops, + * so we give them some more leeway on timeout */ + switch (msg_type) { + case I2400M_MT_GET_TLS_OPERATION_RESULT: + case I2400M_MT_CMD_SEND_EAP_RESPONSE: + ack_timeout = 5 * HZ; + break; + default: + ack_timeout = HZ; + }; + + /* The RX path in rx.c will put any response for this message + * in i2400m->ack_skb and wake us up. If we cancel the wait, + * we need to change the value of i2400m->ack_skb to something + * not -EINPROGRESS so RX knows there is no one waiting. */ + result = wait_for_completion_interruptible_timeout( + &i2400m->msg_completion, ack_timeout); + if (result == 0) { + dev_err(dev, "timeout waiting for reply to message 0x%04x\n", + msg_type); + result = -ETIMEDOUT; + i2400m_msg_to_dev_cancel_wait(i2400m, result); + goto error_wait_for_completion; + } else if (result < 0) { + dev_err(dev, "error waiting for reply to message 0x%04x: %d\n", + msg_type, result); + i2400m_msg_to_dev_cancel_wait(i2400m, result); + goto error_wait_for_completion; + } + + /* Pull out the ack data from i2400m->ack_skb -- see if it is + * an error and act accordingly */ + spin_lock_irqsave(&i2400m->rx_lock, flags); + ack_skb = i2400m->ack_skb; + if (IS_ERR(ack_skb)) + result = PTR_ERR(ack_skb); + else + result = 0; + i2400m->ack_skb = NULL; + spin_unlock_irqrestore(&i2400m->rx_lock, flags); + if (result < 0) + goto error_ack_status; + ack_l3l4_hdr = wimax_msg_data_len(ack_skb, &ack_len); + + /* Check the ack and deliver it if it is ok */ + result = i2400m_msg_size_check(i2400m, ack_l3l4_hdr, ack_len); + if (result < 0) { + dev_err(dev, "HW BUG? reply to message 0x%04x: %d\n", + msg_type, result); + goto error_bad_ack_len; + } + if (msg_type != le16_to_cpu(ack_l3l4_hdr->type)) { + dev_err(dev, "HW BUG? bad reply 0x%04x to message 0x%04x\n", + le16_to_cpu(ack_l3l4_hdr->type), msg_type); + result = -EIO; + goto error_bad_ack_type; + } + i2400m_msg_ack_hook(i2400m, ack_l3l4_hdr, ack_len); + mutex_unlock(&i2400m->msg_mutex); + d_fnend(3, dev, "(i2400m %p buf %p len %zu) = %p\n", + i2400m, buf, buf_len, ack_skb); + return ack_skb; + +error_bad_ack_type: +error_bad_ack_len: + kfree_skb(ack_skb); +error_ack_status: +error_wait_for_completion: +error_tx: + mutex_unlock(&i2400m->msg_mutex); +error_bad_msg: + d_fnend(3, dev, "(i2400m %p buf %p len %zu) = %d\n", + i2400m, buf, buf_len, result); + return ERR_PTR(result); +} + + +/* + * Definitions for the Enter Power Save command + * + * The Enter Power Save command requests the device to go into power + * saving mode. The device will ack or nak the command depending on it + * being ready for it. If it acks, we tell the USB subsystem to + * + * As well, the device might request to go into power saving mode by + * sending a report (REPORT_POWERSAVE_READY), in which case, we issue + * this command. The hookups in the RX coder allow + */ +enum { + I2400M_WAKEUP_ENABLED = 0x01, + I2400M_WAKEUP_DISABLED = 0x02, + I2400M_TLV_TYPE_WAKEUP_MODE = 144, +}; + +struct i2400m_cmd_enter_power_save { + struct i2400m_l3l4_hdr hdr; + struct i2400m_tlv_hdr tlv; + __le32 val; +} __attribute__((packed)); + + +/* + * Request entering power save + * + * This command is (mainly) executed when the device indicates that it + * is ready to go into powersave mode via a REPORT_POWERSAVE_READY. + */ +int i2400m_cmd_enter_powersave(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *ack_skb; + struct i2400m_cmd_enter_power_save *cmd; + char strerr[32]; + + result = -ENOMEM; + cmd = kzalloc(sizeof(*cmd), GFP_KERNEL); + if (cmd == NULL) + goto error_alloc; + cmd->hdr.type = cpu_to_le16(I2400M_MT_CMD_ENTER_POWERSAVE); + cmd->hdr.length = cpu_to_le16(sizeof(*cmd) - sizeof(cmd->hdr)); + cmd->hdr.version = cpu_to_le16(I2400M_L3L4_VERSION); + cmd->tlv.type = cpu_to_le16(I2400M_TLV_TYPE_WAKEUP_MODE); + cmd->tlv.length = cpu_to_le16(sizeof(cmd->val)); + cmd->val = cpu_to_le32(I2400M_WAKEUP_ENABLED); + + ack_skb = i2400m_msg_to_dev(i2400m, cmd, sizeof(*cmd)); + result = PTR_ERR(ack_skb); + if (IS_ERR(ack_skb)) { + dev_err(dev, "Failed to issue 'Enter power save' command: %d\n", + result); + goto error_msg_to_dev; + } + result = i2400m_msg_check_status(wimax_msg_data(ack_skb), + strerr, sizeof(strerr)); + if (result == -EACCES) + d_printf(1, dev, "Cannot enter power save mode\n"); + else if (result < 0) + dev_err(dev, "'Enter power save' (0x%04x) command failed: " + "%d - %s\n", I2400M_MT_CMD_ENTER_POWERSAVE, + result, strerr); + else + d_printf(1, dev, "device ready to power save\n"); + kfree_skb(ack_skb); +error_msg_to_dev: + kfree(cmd); +error_alloc: + return result; +} +EXPORT_SYMBOL_GPL(i2400m_cmd_enter_powersave); + + +/* + * Definitions for getting device information + */ +enum { + I2400M_TLV_DETAILED_DEVICE_INFO = 140 +}; + +/** + * i2400m_get_device_info - Query the device for detailed device information + * + * @i2400m: device descriptor + * + * Returns: an skb whose skb->data points to a 'struct + * i2400m_tlv_detailed_device_info'. When done, kfree_skb() it. The + * skb is *guaranteed* to contain the whole TLV data structure. + * + * On error, IS_ERR(skb) is true and ERR_PTR(skb) is the error + * code. + */ +struct sk_buff *i2400m_get_device_info(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *ack_skb; + struct i2400m_l3l4_hdr *cmd; + const struct i2400m_l3l4_hdr *ack; + size_t ack_len; + const struct i2400m_tlv_hdr *tlv; + const struct i2400m_tlv_detailed_device_info *ddi; + char strerr[32]; + + ack_skb = ERR_PTR(-ENOMEM); + cmd = kzalloc(sizeof(*cmd), GFP_KERNEL); + if (cmd == NULL) + goto error_alloc; + cmd->type = cpu_to_le16(I2400M_MT_GET_DEVICE_INFO); + cmd->length = 0; + cmd->version = cpu_to_le16(I2400M_L3L4_VERSION); + + ack_skb = i2400m_msg_to_dev(i2400m, cmd, sizeof(*cmd)); + if (IS_ERR(ack_skb)) { + dev_err(dev, "Failed to issue 'get device info' command: %ld\n", + PTR_ERR(ack_skb)); + goto error_msg_to_dev; + } + ack = wimax_msg_data_len(ack_skb, &ack_len); + result = i2400m_msg_check_status(ack, strerr, sizeof(strerr)); + if (result < 0) { + dev_err(dev, "'get device info' (0x%04x) command failed: " + "%d - %s\n", I2400M_MT_GET_DEVICE_INFO, result, + strerr); + goto error_cmd_failed; + } + tlv = i2400m_tlv_find(i2400m, ack->pl, ack_len - sizeof(*ack), + I2400M_TLV_DETAILED_DEVICE_INFO, sizeof(*ddi)); + if (tlv == NULL) { + dev_err(dev, "GET DEVICE INFO: " + "detailed device info TLV not found (0x%04x)\n", + I2400M_TLV_DETAILED_DEVICE_INFO); + result = -EIO; + goto error_no_tlv; + } + skb_pull(ack_skb, (void *) tlv - (void *) ack_skb->data); +error_msg_to_dev: + kfree(cmd); +error_alloc: + return ack_skb; + +error_no_tlv: +error_cmd_failed: + kfree_skb(ack_skb); + kfree(cmd); + return ERR_PTR(result); +} + + +/* Firmware interface versions we support */ +enum { + I2400M_HDIv_MAJOR = 9, + I2400M_HDIv_MAJOR_2 = 8, + I2400M_HDIv_MINOR = 1, +}; + + +/** + * i2400m_firmware_check - check firmware versions are compatible with + * the driver + * + * @i2400m: device descriptor + * + * Returns: 0 if ok, < 0 errno code an error and a message in the + * kernel log. + * + * Long function, but quite simple; first chunk launches the command + * and double checks the reply for the right TLV. Then we process the + * TLV (where the meat is). + */ +int i2400m_firmware_check(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *ack_skb; + struct i2400m_l3l4_hdr *cmd; + const struct i2400m_l3l4_hdr *ack; + size_t ack_len; + const struct i2400m_tlv_hdr *tlv; + const struct i2400m_tlv_l4_message_versions *l4mv; + char strerr[32]; + unsigned major, minor, branch; + + result = -ENOMEM; + cmd = kzalloc(sizeof(*cmd), GFP_KERNEL); + if (cmd == NULL) + goto error_alloc; + cmd->type = cpu_to_le16(I2400M_MT_GET_LM_VERSION); + cmd->length = 0; + cmd->version = cpu_to_le16(I2400M_L3L4_VERSION); + + ack_skb = i2400m_msg_to_dev(i2400m, cmd, sizeof(*cmd)); + if (IS_ERR(ack_skb)) { + result = PTR_ERR(ack_skb); + dev_err(dev, "Failed to issue 'get lm version' command: %-d\n", + result); + goto error_msg_to_dev; + } + ack = wimax_msg_data_len(ack_skb, &ack_len); + result = i2400m_msg_check_status(ack, strerr, sizeof(strerr)); + if (result < 0) { + dev_err(dev, "'get lm version' (0x%04x) command failed: " + "%d - %s\n", I2400M_MT_GET_LM_VERSION, result, + strerr); + goto error_cmd_failed; + } + tlv = i2400m_tlv_find(i2400m, ack->pl, ack_len - sizeof(*ack), + I2400M_TLV_L4_MESSAGE_VERSIONS, sizeof(*l4mv)); + if (tlv == NULL) { + dev_err(dev, "get lm version: TLV not found (0x%04x)\n", + I2400M_TLV_L4_MESSAGE_VERSIONS); + result = -EIO; + goto error_no_tlv; + } + l4mv = container_of(tlv, typeof(*l4mv), hdr); + major = le16_to_cpu(l4mv->major); + minor = le16_to_cpu(l4mv->minor); + branch = le16_to_cpu(l4mv->branch); + result = -EINVAL; + if (major != I2400M_HDIv_MAJOR + && major != I2400M_HDIv_MAJOR_2) { + dev_err(dev, "unsupported major fw interface version " + "%u.%u.%u\n", major, minor, branch); + goto error_bad_major; + } + if (major == I2400M_HDIv_MAJOR_2) + dev_err(dev, "deprecated major fw interface version " + "%u.%u.%u\n", major, minor, branch); + result = 0; + if (minor != I2400M_HDIv_MINOR) + dev_warn(dev, "untested minor fw firmware version %u.%u.%u\n", + major, minor, branch); +error_bad_major: + dev_info(dev, "firmware interface version %u.%u.%u\n", + major, minor, branch); +error_no_tlv: +error_cmd_failed: + kfree_skb(ack_skb); +error_msg_to_dev: + kfree(cmd); +error_alloc: + return result; +} + + +/* + * Send an DoExitIdle command to the device to ask it to go out of + * basestation-idle mode. + * + * @i2400m: device descriptor + * + * This starts a renegotiation with the basestation that might involve + * another crypto handshake with user space. + * + * Returns: 0 if ok, < 0 errno code on error. + */ +int i2400m_cmd_exit_idle(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *ack_skb; + struct i2400m_l3l4_hdr *cmd; + char strerr[32]; + + result = -ENOMEM; + cmd = kzalloc(sizeof(*cmd), GFP_KERNEL); + if (cmd == NULL) + goto error_alloc; + cmd->type = cpu_to_le16(I2400M_MT_CMD_EXIT_IDLE); + cmd->length = 0; + cmd->version = cpu_to_le16(I2400M_L3L4_VERSION); + + ack_skb = i2400m_msg_to_dev(i2400m, cmd, sizeof(*cmd)); + result = PTR_ERR(ack_skb); + if (IS_ERR(ack_skb)) { + dev_err(dev, "Failed to issue 'exit idle' command: %d\n", + result); + goto error_msg_to_dev; + } + result = i2400m_msg_check_status(wimax_msg_data(ack_skb), + strerr, sizeof(strerr)); + kfree_skb(ack_skb); +error_msg_to_dev: + kfree(cmd); +error_alloc: + return result; + +} + + +/* + * Query the device for its state, update the WiMAX stack's idea of it + * + * @i2400m: device descriptor + * + * Returns: 0 if ok, < 0 errno code on error. + * + * Executes a 'Get State' command and parses the returned + * TLVs. + * + * Because this is almost identical to a 'Report State', we use + * i2400m_report_state_hook() to parse the answer. This will set the + * carrier state, as well as the RF Kill switches state. + */ +int i2400m_cmd_get_state(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *ack_skb; + struct i2400m_l3l4_hdr *cmd; + const struct i2400m_l3l4_hdr *ack; + size_t ack_len; + char strerr[32]; + + result = -ENOMEM; + cmd = kzalloc(sizeof(*cmd), GFP_KERNEL); + if (cmd == NULL) + goto error_alloc; + cmd->type = cpu_to_le16(I2400M_MT_GET_STATE); + cmd->length = 0; + cmd->version = cpu_to_le16(I2400M_L3L4_VERSION); + + ack_skb = i2400m_msg_to_dev(i2400m, cmd, sizeof(*cmd)); + if (IS_ERR(ack_skb)) { + dev_err(dev, "Failed to issue 'get state' command: %ld\n", + PTR_ERR(ack_skb)); + result = PTR_ERR(ack_skb); + goto error_msg_to_dev; + } + ack = wimax_msg_data_len(ack_skb, &ack_len); + result = i2400m_msg_check_status(ack, strerr, sizeof(strerr)); + if (result < 0) { + dev_err(dev, "'get state' (0x%04x) command failed: " + "%d - %s\n", I2400M_MT_GET_STATE, result, strerr); + goto error_cmd_failed; + } + i2400m_report_state_hook(i2400m, ack, ack_len - sizeof(*ack), + "GET STATE"); + result = 0; + kfree_skb(ack_skb); +error_cmd_failed: +error_msg_to_dev: + kfree(cmd); +error_alloc: + return result; +} +EXPORT_SYMBOL_GPL(i2400m_cmd_get_state); + + +/** + * Set basic configuration settings + * + * @i2400m: device descriptor + * @args: array of pointers to the TLV headers to send for + * configuration (each followed by its payload). + * TLV headers and payloads must be properly initialized, with the + * right endianess (LE). + * @arg_size: number of pointers in the @args array + */ +int i2400m_set_init_config(struct i2400m *i2400m, + const struct i2400m_tlv_hdr **arg, size_t args) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct sk_buff *ack_skb; + struct i2400m_l3l4_hdr *cmd; + char strerr[32]; + unsigned argc, argsize, tlv_size; + const struct i2400m_tlv_hdr *tlv_hdr; + void *buf, *itr; + + d_fnstart(3, dev, "(i2400m %p arg %p args %zu)\n", i2400m, arg, args); + result = 0; + if (args == 0) + goto none; + /* Compute the size of all the TLVs, so we can alloc a + * contiguous command block to copy them. */ + argsize = 0; + for (argc = 0; argc < args; argc++) { + tlv_hdr = arg[argc]; + argsize += sizeof(*tlv_hdr) + le16_to_cpu(tlv_hdr->length); + } + WARN_ON(argc >= 9); /* As per hw spec */ + + /* Alloc the space for the command and TLVs*/ + result = -ENOMEM; + buf = kzalloc(sizeof(*cmd) + argsize, GFP_KERNEL); + if (buf == NULL) + goto error_alloc; + cmd = buf; + cmd->type = cpu_to_le16(I2400M_MT_SET_INIT_CONFIG); + cmd->length = cpu_to_le16(argsize); + cmd->version = cpu_to_le16(I2400M_L3L4_VERSION); + + /* Copy the TLVs */ + itr = buf + sizeof(*cmd); + for (argc = 0; argc < args; argc++) { + tlv_hdr = arg[argc]; + tlv_size = sizeof(*tlv_hdr) + le16_to_cpu(tlv_hdr->length); + memcpy(itr, tlv_hdr, tlv_size); + itr += tlv_size; + } + + /* Send the message! */ + ack_skb = i2400m_msg_to_dev(i2400m, buf, sizeof(*cmd) + argsize); + result = PTR_ERR(ack_skb); + if (IS_ERR(ack_skb)) { + dev_err(dev, "Failed to issue 'init config' command: %d\n", + result); + + goto error_msg_to_dev; + } + result = i2400m_msg_check_status(wimax_msg_data(ack_skb), + strerr, sizeof(strerr)); + if (result < 0) + dev_err(dev, "'init config' (0x%04x) command failed: %d - %s\n", + I2400M_MT_SET_INIT_CONFIG, result, strerr); + kfree_skb(ack_skb); +error_msg_to_dev: + kfree(buf); +error_alloc: +none: + d_fnend(3, dev, "(i2400m %p arg %p args %zu) = %d\n", + i2400m, arg, args, result); + return result; + +} +EXPORT_SYMBOL_GPL(i2400m_set_init_config); + + +/** + * i2400m_dev_initialize - Initialize the device once communications are ready + * + * @i2400m: device descriptor + * + * Returns: 0 if ok, < 0 errno code on error. + * + * Configures the device to work the way we like it. + * + * At the point of this call, the device is registered with the WiMAX + * and netdev stacks, firmware is uploaded and we can talk to the + * device normally. + */ +int i2400m_dev_initialize(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct i2400m_tlv_config_idle_parameters idle_params; + const struct i2400m_tlv_hdr *args[9]; + unsigned argc = 0; + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + /* Useless for now...might change */ + if (i2400m_idle_mode_disabled) { + idle_params.hdr.type = + cpu_to_le16(I2400M_TLV_CONFIG_IDLE_PARAMETERS); + idle_params.hdr.length = cpu_to_le16( + sizeof(idle_params) - sizeof(idle_params.hdr)); + idle_params.idle_timeout = 0; + idle_params.idle_paging_interval = 0; + args[argc++] = &idle_params.hdr; + } + result = i2400m_set_init_config(i2400m, args, argc); + if (result < 0) + goto error; + result = i2400m_firmware_check(i2400m); /* fw versions ok? */ + if (result < 0) + goto error; + /* + * Update state: Here it just calls a get state; parsing the + * result (System State TLV and RF Status TLV [done in the rx + * path hooks]) will set the hardware and software RF-Kill + * status. + */ + result = i2400m_cmd_get_state(i2400m); +error: + d_fnend(3, dev, "(i2400m %p) = %d\n", i2400m, result); + return result; +} + + +/** + * i2400m_dev_shutdown - Shutdown a running device + * + * @i2400m: device descriptor + * + * Gracefully stops the device, moving it to the lowest power + * consumption state possible. + */ +void i2400m_dev_shutdown(struct i2400m *i2400m) +{ + int result = -ENODEV; + struct device *dev = i2400m_dev(i2400m); + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + result = i2400m->bus_reset(i2400m, I2400M_RT_WARM); + d_fnend(3, dev, "(i2400m %p) = void [%d]\n", i2400m, result); + return; +} From c71228caf91ec6320b489dec5cd0087b64da9fb5 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:49 -0800 Subject: [PATCH 112/123] i2400m: debugfs controls Expose knobs to control the device (induce reset, power saving, querying tx or rx stats, internal debug information and debug level manipulation). Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/debugfs.c | 392 +++++++++++++++++++++++++++++ 1 file changed, 392 insertions(+) create mode 100644 drivers/net/wimax/i2400m/debugfs.c diff --git a/drivers/net/wimax/i2400m/debugfs.c b/drivers/net/wimax/i2400m/debugfs.c new file mode 100644 index 00000000000..62663298597 --- /dev/null +++ b/drivers/net/wimax/i2400m/debugfs.c @@ -0,0 +1,392 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Debugfs interfaces to manipulate driver and device information + * + * + * Copyright (C) 2007 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + */ + +#include +#include +#include +#include +#include +#include "i2400m.h" + + +#define D_SUBMODULE debugfs +#include "debug-levels.h" + +static +int debugfs_netdev_queue_stopped_get(void *data, u64 *val) +{ + struct i2400m *i2400m = data; + *val = netif_queue_stopped(i2400m->wimax_dev.net_dev); + return 0; +} +DEFINE_SIMPLE_ATTRIBUTE(fops_netdev_queue_stopped, + debugfs_netdev_queue_stopped_get, + NULL, "%llu\n"); + + +static +struct dentry *debugfs_create_netdev_queue_stopped( + const char *name, struct dentry *parent, struct i2400m *i2400m) +{ + return debugfs_create_file(name, 0400, parent, i2400m, + &fops_netdev_queue_stopped); +} + + +/* + * inode->i_private has the @data argument to debugfs_create_file() + */ +static +int i2400m_stats_open(struct inode *inode, struct file *filp) +{ + filp->private_data = inode->i_private; + return 0; +} + +/* + * We don't allow partial reads of this file, as then the reader would + * get weirdly confused data as it is updated. + * + * So or you read it all or nothing; if you try to read with an offset + * != 0, we consider you are done reading. + */ +static +ssize_t i2400m_rx_stats_read(struct file *filp, char __user *buffer, + size_t count, loff_t *ppos) +{ + struct i2400m *i2400m = filp->private_data; + char buf[128]; + unsigned long flags; + + if (*ppos != 0) + return 0; + if (count < sizeof(buf)) + return -ENOSPC; + spin_lock_irqsave(&i2400m->rx_lock, flags); + snprintf(buf, sizeof(buf), "%u %u %u %u %u %u %u\n", + i2400m->rx_pl_num, i2400m->rx_pl_min, + i2400m->rx_pl_max, i2400m->rx_num, + i2400m->rx_size_acc, + i2400m->rx_size_min, i2400m->rx_size_max); + spin_unlock_irqrestore(&i2400m->rx_lock, flags); + return simple_read_from_buffer(buffer, count, ppos, buf, strlen(buf)); +} + + +/* Any write clears the stats */ +static +ssize_t i2400m_rx_stats_write(struct file *filp, const char __user *buffer, + size_t count, loff_t *ppos) +{ + struct i2400m *i2400m = filp->private_data; + unsigned long flags; + + spin_lock_irqsave(&i2400m->rx_lock, flags); + i2400m->rx_pl_num = 0; + i2400m->rx_pl_max = 0; + i2400m->rx_pl_min = UINT_MAX; + i2400m->rx_num = 0; + i2400m->rx_size_acc = 0; + i2400m->rx_size_min = UINT_MAX; + i2400m->rx_size_max = 0; + spin_unlock_irqrestore(&i2400m->rx_lock, flags); + return count; +} + +static +const struct file_operations i2400m_rx_stats_fops = { + .owner = THIS_MODULE, + .open = i2400m_stats_open, + .read = i2400m_rx_stats_read, + .write = i2400m_rx_stats_write, +}; + + +/* See i2400m_rx_stats_read() */ +static +ssize_t i2400m_tx_stats_read(struct file *filp, char __user *buffer, + size_t count, loff_t *ppos) +{ + struct i2400m *i2400m = filp->private_data; + char buf[128]; + unsigned long flags; + + if (*ppos != 0) + return 0; + if (count < sizeof(buf)) + return -ENOSPC; + spin_lock_irqsave(&i2400m->tx_lock, flags); + snprintf(buf, sizeof(buf), "%u %u %u %u %u %u %u\n", + i2400m->tx_pl_num, i2400m->tx_pl_min, + i2400m->tx_pl_max, i2400m->tx_num, + i2400m->tx_size_acc, + i2400m->tx_size_min, i2400m->tx_size_max); + spin_unlock_irqrestore(&i2400m->tx_lock, flags); + return simple_read_from_buffer(buffer, count, ppos, buf, strlen(buf)); +} + +/* Any write clears the stats */ +static +ssize_t i2400m_tx_stats_write(struct file *filp, const char __user *buffer, + size_t count, loff_t *ppos) +{ + struct i2400m *i2400m = filp->private_data; + unsigned long flags; + + spin_lock_irqsave(&i2400m->tx_lock, flags); + i2400m->tx_pl_num = 0; + i2400m->tx_pl_max = 0; + i2400m->tx_pl_min = UINT_MAX; + i2400m->tx_num = 0; + i2400m->tx_size_acc = 0; + i2400m->tx_size_min = UINT_MAX; + i2400m->tx_size_max = 0; + spin_unlock_irqrestore(&i2400m->tx_lock, flags); + return count; +} + +static +const struct file_operations i2400m_tx_stats_fops = { + .owner = THIS_MODULE, + .open = i2400m_stats_open, + .read = i2400m_tx_stats_read, + .write = i2400m_tx_stats_write, +}; + + +/* Write 1 to ask the device to go into suspend */ +static +int debugfs_i2400m_suspend_set(void *data, u64 val) +{ + int result; + struct i2400m *i2400m = data; + result = i2400m_cmd_enter_powersave(i2400m); + if (result >= 0) + result = 0; + return result; +} +DEFINE_SIMPLE_ATTRIBUTE(fops_i2400m_suspend, + NULL, debugfs_i2400m_suspend_set, + "%llu\n"); + +static +struct dentry *debugfs_create_i2400m_suspend( + const char *name, struct dentry *parent, struct i2400m *i2400m) +{ + return debugfs_create_file(name, 0200, parent, i2400m, + &fops_i2400m_suspend); +} + + +/* + * Reset the device + * + * Write 0 to ask the device to soft reset, 1 to cold reset, 2 to bus + * reset (as defined by enum i2400m_reset_type). + */ +static +int debugfs_i2400m_reset_set(void *data, u64 val) +{ + int result; + struct i2400m *i2400m = data; + enum i2400m_reset_type rt = val; + switch(rt) { + case I2400M_RT_WARM: + case I2400M_RT_COLD: + case I2400M_RT_BUS: + result = i2400m->bus_reset(i2400m, rt); + if (result >= 0) + result = 0; + default: + result = -EINVAL; + } + return result; +} +DEFINE_SIMPLE_ATTRIBUTE(fops_i2400m_reset, + NULL, debugfs_i2400m_reset_set, + "%llu\n"); + +static +struct dentry *debugfs_create_i2400m_reset( + const char *name, struct dentry *parent, struct i2400m *i2400m) +{ + return debugfs_create_file(name, 0200, parent, i2400m, + &fops_i2400m_reset); +} + +/* + * Debug levels control; see debug.h + */ +struct d_level D_LEVEL[] = { + D_SUBMODULE_DEFINE(control), + D_SUBMODULE_DEFINE(driver), + D_SUBMODULE_DEFINE(debugfs), + D_SUBMODULE_DEFINE(fw), + D_SUBMODULE_DEFINE(netdev), + D_SUBMODULE_DEFINE(rfkill), + D_SUBMODULE_DEFINE(rx), + D_SUBMODULE_DEFINE(tx), +}; +size_t D_LEVEL_SIZE = ARRAY_SIZE(D_LEVEL); + +#define __debugfs_register(prefix, name, parent) \ +do { \ + result = d_level_register_debugfs(prefix, name, parent); \ + if (result < 0) \ + goto error; \ +} while (0) + + +int i2400m_debugfs_add(struct i2400m *i2400m) +{ + int result; + struct device *dev = i2400m_dev(i2400m); + struct dentry *dentry = i2400m->wimax_dev.debugfs_dentry; + struct dentry *fd; + + dentry = debugfs_create_dir("i2400m", dentry); + result = PTR_ERR(dentry); + if (IS_ERR(dentry)) { + if (result == -ENODEV) + result = 0; /* No debugfs support */ + goto error; + } + i2400m->debugfs_dentry = dentry; + __debugfs_register("dl_", control, dentry); + __debugfs_register("dl_", driver, dentry); + __debugfs_register("dl_", debugfs, dentry); + __debugfs_register("dl_", fw, dentry); + __debugfs_register("dl_", netdev, dentry); + __debugfs_register("dl_", rfkill, dentry); + __debugfs_register("dl_", rx, dentry); + __debugfs_register("dl_", tx, dentry); + + fd = debugfs_create_size_t("tx_in", 0400, dentry, + &i2400m->tx_in); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "tx_in: %d\n", result); + goto error; + } + + fd = debugfs_create_size_t("tx_out", 0400, dentry, + &i2400m->tx_out); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "tx_out: %d\n", result); + goto error; + } + + fd = debugfs_create_u32("state", 0600, dentry, + &i2400m->state); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "state: %d\n", result); + goto error; + } + + /* + * Trace received messages from user space + * + * In order to tap the bidirectional message stream in the + * 'msg' pipe, user space can read from the 'msg' pipe; + * however, due to limitations in libnl, we can't know what + * the different applications are sending down to the kernel. + * + * So we have this hack where the driver will echo any message + * received on the msg pipe from user space [through a call to + * wimax_dev->op_msg_from_user() into + * i2400m_op_msg_from_user()] into the 'trace' pipe that this + * driver creates. + * + * So then, reading from both the 'trace' and 'msg' pipes in + * user space will provide a full dump of the traffic. + * + * Write 1 to activate, 0 to clear. + * + * It is not really very atomic, but it is also not too + * critical. + */ + fd = debugfs_create_u8("trace_msg_from_user", 0600, dentry, + &i2400m->trace_msg_from_user); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "trace_msg_from_user: %d\n", result); + goto error; + } + + fd = debugfs_create_netdev_queue_stopped("netdev_queue_stopped", + dentry, i2400m); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "netdev_queue_stopped: %d\n", result); + goto error; + } + + fd = debugfs_create_file("rx_stats", 0600, dentry, i2400m, + &i2400m_rx_stats_fops); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "rx_stats: %d\n", result); + goto error; + } + + fd = debugfs_create_file("tx_stats", 0600, dentry, i2400m, + &i2400m_tx_stats_fops); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "tx_stats: %d\n", result); + goto error; + } + + fd = debugfs_create_i2400m_suspend("suspend", dentry, i2400m); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry suspend: %d\n", + result); + goto error; + } + + fd = debugfs_create_i2400m_reset("reset", dentry, i2400m); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry reset: %d\n", result); + goto error; + } + + result = 0; +error: + return result; +} + +void i2400m_debugfs_rm(struct i2400m *i2400m) +{ + debugfs_remove_recursive(i2400m->debugfs_dentry); +} From 11a7d0e3140d2f3e8052a856e8582ce9b021972c Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:50 -0800 Subject: [PATCH 113/123] i2400m/USB: header for the USB bus driver This contains the common function declaration and constants for the USB driver for the 2400m Wireless WiMAX Connection, as well as it's debug level settings. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/i2400m-usb.h | 264 ++++++++++++++++++++ drivers/net/wimax/i2400m/usb-debug-levels.h | 42 ++++ 2 files changed, 306 insertions(+) create mode 100644 drivers/net/wimax/i2400m/i2400m-usb.h create mode 100644 drivers/net/wimax/i2400m/usb-debug-levels.h diff --git a/drivers/net/wimax/i2400m/i2400m-usb.h b/drivers/net/wimax/i2400m/i2400m-usb.h new file mode 100644 index 00000000000..6f76558b170 --- /dev/null +++ b/drivers/net/wimax/i2400m/i2400m-usb.h @@ -0,0 +1,264 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * USB-specific i2400m driver definitions + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Inaky Perez-Gonzalez + * Yanir Lubetkin + * - Initial implementation + * + * + * This driver implements the bus-specific part of the i2400m for + * USB. Check i2400m.h for a generic driver description. + * + * ARCHITECTURE + * + * This driver listens to notifications sent from the notification + * endpoint (in usb-notif.c); when data is ready to read, the code in + * there schedules a read from the device (usb-rx.c) and then passes + * the data to the generic RX code (rx.c). + * + * When the generic driver needs to send data (network or control), it + * queues up in the TX FIFO (tx.c) and that will notify the driver + * through the i2400m->bus_tx_kick() callback + * (usb-tx.c:i2400mu_bus_tx_kick) which will send the items in the + * FIFO queue. + * + * This driver, as well, implements the USB-specific ops for the generic + * driver to be able to setup/teardown communication with the device + * [i2400m_bus_dev_start() and i2400m_bus_dev_stop()], reseting the + * device [i2400m_bus_reset()] and performing firmware upload + * [i2400m_bus_bm_cmd() and i2400_bus_bm_wait_for_ack()]. + */ + +#ifndef __I2400M_USB_H__ +#define __I2400M_USB_H__ + +#include "i2400m.h" +#include + + +/* + * Error Density Count: cheapo error density (over time) counter + * + * Originally by Reinette Chatre + * + * Embed an 'struct edc' somewhere. Each time there is a soft or + * retryable error, call edc_inc() and check if the error top + * watermark has been reached. + */ +enum { + EDC_MAX_ERRORS = 10, + EDC_ERROR_TIMEFRAME = HZ, +}; + +/* error density counter */ +struct edc { + unsigned long timestart; + u16 errorcount; +}; + +static inline void edc_init(struct edc *edc) +{ + edc->timestart = jiffies; +} + +/** + * edc_inc - report a soft error and check if we are over the watermark + * + * @edc: pointer to error density counter. + * @max_err: maximum number of errors we can accept over the timeframe + * @timeframe: lenght of the timeframe (in jiffies). + * + * Returns: !0 1 if maximum acceptable errors per timeframe has been + * exceeded. 0 otherwise. + * + * This is way to determine if the number of acceptable errors per time + * period has been exceeded. It is not accurate as there are cases in which + * this scheme will not work, for example if there are periodic occurences + * of errors that straddle updates to the start time. This scheme is + * sufficient for our usage. + * + * To use, embed a 'struct edc' somewhere, initialize it with + * edc_init() and when an error hits: + * + * if (do_something_fails_with_a_soft_error) { + * if (edc_inc(&my->edc, MAX_ERRORS, MAX_TIMEFRAME)) + * Ops, hard error, do something about it + * else + * Retry or ignore, depending on whatever + * } + */ +static inline int edc_inc(struct edc *edc, u16 max_err, u16 timeframe) +{ + unsigned long now; + + now = jiffies; + if (now - edc->timestart > timeframe) { + edc->errorcount = 1; + edc->timestart = now; + } else if (++edc->errorcount > max_err) { + edc->errorcount = 0; + edc->timestart = now; + return 1; + } + return 0; +} + +/* Host-Device interface for USB */ +enum { + I2400MU_MAX_NOTIFICATION_LEN = 256, + I2400MU_BLK_SIZE = 16, + I2400MU_PL_SIZE_MAX = 0x3EFF, + + /* Endpoints */ + I2400MU_EP_BULK_OUT = 0, + I2400MU_EP_NOTIFICATION, + I2400MU_EP_RESET_COLD, + I2400MU_EP_BULK_IN, +}; + + +/** + * struct i2400mu - descriptor for a USB connected i2400m + * + * @i2400m: bus-generic i2400m implementation; has to be first (see + * it's documentation in i2400m.h). + * + * @usb_dev: pointer to our USB device + * + * @usb_iface: pointer to our USB interface + * + * @urb_edc: error density counter; used to keep a density-on-time tab + * on how many soft (retryable or ignorable) errors we get. If we + * go over the threshold, we consider the bus transport is failing + * too much and reset. + * + * @notif_urb: URB for receiving notifications from the device. + * + * @tx_kthread: thread we use for data TX. We use a thread because in + * order to do deep power saving and put the device to sleep, we + * need to call usb_autopm_*() [blocking functions]. + * + * @tx_wq: waitqueue for the TX kthread to sleep when there is no data + * to be sent; when more data is available, it is woken up by + * i2400mu_bus_tx_kick(). + * + * @rx_kthread: thread we use for data RX. We use a thread because in + * order to do deep power saving and put the device to sleep, we + * need to call usb_autopm_*() [blocking functions]. + * + * @rx_wq: waitqueue for the RX kthread to sleep when there is no data + * to receive. When data is available, it is woken up by + * usb-notif.c:i2400mu_notification_grok(). + * + * @rx_pending_count: number of rx-data-ready notifications that were + * still not handled by the RX kthread. + * + * @rx_size: current RX buffer size that is being used. + * + * @rx_size_acc: accumulator of the sizes of the previous read + * transactions. + * + * @rx_size_cnt: number of read transactions accumulated in + * @rx_size_acc. + * + * @do_autopm: disable(0)/enable(>0) calling the + * usb_autopm_get/put_interface() barriers when executing + * commands. See doc in i2400mu_suspend() for more information. + * + * @rx_size_auto_shrink: if true, the rx_size is shrinked + * automatically based on the average size of the received + * transactions. This allows the receive code to allocate smaller + * chunks of memory and thus reduce pressure on the memory + * allocator by not wasting so much space. By default it is + * enabled. + * + * @debugfs_dentry: hookup for debugfs files. + * These have to be in a separate directory, a child of + * (wimax_dev->debugfs_dentry) so they can be removed when the + * module unloads, as we don't keep each dentry. + */ +struct i2400mu { + struct i2400m i2400m; /* FIRST! See doc */ + + struct usb_device *usb_dev; + struct usb_interface *usb_iface; + struct edc urb_edc; /* Error density counter */ + + struct urb *notif_urb; + struct task_struct *tx_kthread; + wait_queue_head_t tx_wq; + + struct task_struct *rx_kthread; + wait_queue_head_t rx_wq; + atomic_t rx_pending_count; + size_t rx_size, rx_size_acc, rx_size_cnt; + atomic_t do_autopm; + u8 rx_size_auto_shrink; + + struct dentry *debugfs_dentry; +}; + + +static inline +void i2400mu_init(struct i2400mu *i2400mu) +{ + i2400m_init(&i2400mu->i2400m); + edc_init(&i2400mu->urb_edc); + init_waitqueue_head(&i2400mu->tx_wq); + atomic_set(&i2400mu->rx_pending_count, 0); + init_waitqueue_head(&i2400mu->rx_wq); + i2400mu->rx_size = PAGE_SIZE - sizeof(struct skb_shared_info); + atomic_set(&i2400mu->do_autopm, 1); + i2400mu->rx_size_auto_shrink = 1; +} + +extern int i2400mu_notification_setup(struct i2400mu *); +extern void i2400mu_notification_release(struct i2400mu *); + +extern int i2400mu_rx_setup(struct i2400mu *); +extern void i2400mu_rx_release(struct i2400mu *); +extern void i2400mu_rx_kick(struct i2400mu *); + +extern int i2400mu_tx_setup(struct i2400mu *); +extern void i2400mu_tx_release(struct i2400mu *); +extern void i2400mu_bus_tx_kick(struct i2400m *); + +extern ssize_t i2400mu_bus_bm_cmd_send(struct i2400m *, + const struct i2400m_bootrom_header *, + size_t, int); +extern ssize_t i2400mu_bus_bm_wait_for_ack(struct i2400m *, + struct i2400m_bootrom_header *, + size_t); +#endif /* #ifndef __I2400M_USB_H__ */ diff --git a/drivers/net/wimax/i2400m/usb-debug-levels.h b/drivers/net/wimax/i2400m/usb-debug-levels.h new file mode 100644 index 00000000000..e4358bd880b --- /dev/null +++ b/drivers/net/wimax/i2400m/usb-debug-levels.h @@ -0,0 +1,42 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Debug levels control file for the i2400m-usb module + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Inaky Perez-Gonzalez + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + */ +#ifndef __debug_levels__h__ +#define __debug_levels__h__ + +/* Maximum compile and run time debug level for all submodules */ +#define D_MODULENAME i2400m_usb +#define D_MASTER CONFIG_WIMAX_I2400M_DEBUG_LEVEL + +#include + +/* List of all the enabled modules */ +enum d_module { + D_SUBMODULE_DECLARE(usb), + D_SUBMODULE_DECLARE(fw), + D_SUBMODULE_DECLARE(notif), + D_SUBMODULE_DECLARE(rx), + D_SUBMODULE_DECLARE(tx), +}; + + +#endif /* #ifndef __debug_levels__h__ */ From f398e4240fce962d0dd74dc11c59fe20860e7a71 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:51 -0800 Subject: [PATCH 114/123] i2400m/USB: probe/disconnect, dev init/shutdown and reset backends Implements probe/disconnect for the USB device, as well as main backends for the generic driver to control the USB device (bus_dev_start(), bus_dev_stop() and bus_reset()). Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/usb.c | 591 +++++++++++++++++++++++++++++++++ 1 file changed, 591 insertions(+) create mode 100644 drivers/net/wimax/i2400m/usb.c diff --git a/drivers/net/wimax/i2400m/usb.c b/drivers/net/wimax/i2400m/usb.c new file mode 100644 index 00000000000..6d4b65fd9c1 --- /dev/null +++ b/drivers/net/wimax/i2400m/usb.c @@ -0,0 +1,591 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Linux driver model glue for USB device, reset & fw upload + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Inaky Perez-Gonzalez + * Yanir Lubetkin + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * See i2400m-usb.h for a general description of this driver. + * + * This file implements driver model glue, and hook ups for the + * generic driver to implement the bus-specific functions (device + * communication setup/tear down, firmware upload and resetting). + * + * ROADMAP + * + * i2400mu_probe() + * alloc_netdev()... + * i2400mu_netdev_setup() + * i2400mu_init() + * i2400m_netdev_setup() + * i2400m_setup()... + * + * i2400mu_disconnect + * i2400m_release() + * free_netdev() + * + * i2400mu_suspend() + * i2400m_cmd_enter_powersave() + * i2400mu_notification_release() + * + * i2400mu_resume() + * i2400mu_notification_setup() + * + * i2400mu_bus_dev_start() Called by i2400m_dev_start() [who is + * i2400mu_tx_setup() called by i2400m_setup()] + * i2400mu_rx_setup() + * i2400mu_notification_setup() + * + * i2400mu_bus_dev_stop() Called by i2400m_dev_stop() [who is + * i2400mu_notification_release() called by i2400m_release()] + * i2400mu_rx_release() + * i2400mu_tx_release() + * + * i2400mu_bus_reset() Called by i2400m->bus_reset + * __i2400mu_reset() + * __i2400mu_send_barker() + * usb_reset_device() + */ +#include "i2400m-usb.h" +#include +#include + + +#define D_SUBMODULE usb +#include "usb-debug-levels.h" + + +/* Our firmware file name */ +#define I2400MU_FW_FILE_NAME "i2400m-fw-usb-" I2400M_FW_VERSION ".sbcf" + +static +int i2400mu_bus_dev_start(struct i2400m *i2400m) +{ + int result; + struct i2400mu *i2400mu = container_of(i2400m, struct i2400mu, i2400m); + struct device *dev = &i2400mu->usb_iface->dev; + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + result = i2400mu_tx_setup(i2400mu); + if (result < 0) + goto error_usb_tx_setup; + result = i2400mu_rx_setup(i2400mu); + if (result < 0) + goto error_usb_rx_setup; + result = i2400mu_notification_setup(i2400mu); + if (result < 0) + goto error_notif_setup; + d_fnend(3, dev, "(i2400m %p) = %d\n", i2400m, result); + return result; + +error_notif_setup: + i2400mu_rx_release(i2400mu); +error_usb_rx_setup: + i2400mu_tx_release(i2400mu); +error_usb_tx_setup: + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); + return result; +} + + +static +void i2400mu_bus_dev_stop(struct i2400m *i2400m) +{ + struct i2400mu *i2400mu = container_of(i2400m, struct i2400mu, i2400m); + struct device *dev = &i2400mu->usb_iface->dev; + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + i2400mu_notification_release(i2400mu); + i2400mu_rx_release(i2400mu); + i2400mu_tx_release(i2400mu); + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); +} + + +/* + * Sends a barker buffer to the device + * + * This helper will allocate a kmalloced buffer and use it to transmit + * (then free it). Reason for this is that other arches cannot use + * stack/vmalloc/text areas for DMA transfers. + * + * Error recovery here is simpler: anything is considered a hard error + * and will move the reset code to use a last-resort bus-based reset. + */ +static +int __i2400mu_send_barker(struct i2400mu *i2400mu, + const __le32 *barker, + size_t barker_size, + unsigned endpoint) +{ + struct usb_endpoint_descriptor *epd = NULL; + int pipe, actual_len, ret; + struct device *dev = &i2400mu->usb_iface->dev; + void *buffer; + int do_autopm = 1; + + ret = usb_autopm_get_interface(i2400mu->usb_iface); + if (ret < 0) { + dev_err(dev, "RESET: can't get autopm: %d\n", ret); + do_autopm = 0; + } + ret = -ENOMEM; + buffer = kmalloc(barker_size, GFP_KERNEL); + if (buffer == NULL) + goto error_kzalloc; + epd = usb_get_epd(i2400mu->usb_iface, endpoint); + pipe = usb_sndbulkpipe(i2400mu->usb_dev, epd->bEndpointAddress); + memcpy(buffer, barker, barker_size); + ret = usb_bulk_msg(i2400mu->usb_dev, pipe, buffer, barker_size, + &actual_len, HZ); + if (ret < 0) { + if (ret != -EINVAL) + dev_err(dev, "E: barker error: %d\n", ret); + } else if (actual_len != barker_size) { + dev_err(dev, "E: only %d bytes transmitted\n", actual_len); + ret = -EIO; + } + kfree(buffer); +error_kzalloc: + if (do_autopm) + usb_autopm_put_interface(i2400mu->usb_iface); + return ret; +} + + +/* + * Reset a device at different levels (warm, cold or bus) + * + * @i2400m: device descriptor + * @reset_type: soft, warm or bus reset (I2400M_RT_WARM/SOFT/BUS) + * + * Warm and cold resets get a USB reset if they fail. + * + * Warm reset: + * + * The device will be fully reset internally, but won't be + * disconnected from the USB bus (so no reenumeration will + * happen). Firmware upload will be neccessary. + * + * The device will send a reboot barker in the notification endpoint + * that will trigger the driver to reinitialize the state + * automatically from notif.c:i2400m_notification_grok() into + * i2400m_dev_bootstrap_delayed(). + * + * Cold and bus (USB) reset: + * + * The device will be fully reset internally, disconnected from the + * USB bus an a reenumeration will happen. Firmware upload will be + * neccessary. Thus, we don't do any locking or struct + * reinitialization, as we are going to be fully disconnected and + * reenumerated. + * + * Note we need to return -ENODEV if a warm reset was requested and we + * had to resort to a bus reset. See i2400m_op_reset(), wimax_reset() + * and wimax_dev->op_reset. + * + * WARNING: no driver state saved/fixed + */ +static +int i2400mu_bus_reset(struct i2400m *i2400m, enum i2400m_reset_type rt) +{ + int result; + struct i2400mu *i2400mu = + container_of(i2400m, struct i2400mu, i2400m); + struct device *dev = i2400m_dev(i2400m); + static const __le32 i2400m_WARM_BOOT_BARKER[4] = { + __constant_cpu_to_le32(I2400M_WARM_RESET_BARKER), + __constant_cpu_to_le32(I2400M_WARM_RESET_BARKER), + __constant_cpu_to_le32(I2400M_WARM_RESET_BARKER), + __constant_cpu_to_le32(I2400M_WARM_RESET_BARKER), + }; + static const __le32 i2400m_COLD_BOOT_BARKER[4] = { + __constant_cpu_to_le32(I2400M_COLD_RESET_BARKER), + __constant_cpu_to_le32(I2400M_COLD_RESET_BARKER), + __constant_cpu_to_le32(I2400M_COLD_RESET_BARKER), + __constant_cpu_to_le32(I2400M_COLD_RESET_BARKER), + }; + + d_fnstart(3, dev, "(i2400m %p rt %u)\n", i2400m, rt); + if (rt == I2400M_RT_WARM) + result = __i2400mu_send_barker(i2400mu, i2400m_WARM_BOOT_BARKER, + sizeof(i2400m_WARM_BOOT_BARKER), + I2400MU_EP_BULK_OUT); + else if (rt == I2400M_RT_COLD) + result = __i2400mu_send_barker(i2400mu, i2400m_COLD_BOOT_BARKER, + sizeof(i2400m_COLD_BOOT_BARKER), + I2400MU_EP_RESET_COLD); + else if (rt == I2400M_RT_BUS) { +do_bus_reset: + result = usb_reset_device(i2400mu->usb_dev); + switch (result) { + case 0: + case -EINVAL: /* device is gone */ + case -ENODEV: + case -ENOENT: + case -ESHUTDOWN: + result = rt == I2400M_RT_WARM ? -ENODEV : 0; + break; /* We assume the device is disconnected */ + default: + dev_err(dev, "USB reset failed (%d), giving up!\n", + result); + } + } else + BUG(); + if (result < 0 + && result != -EINVAL /* device is gone */ + && rt != I2400M_RT_BUS) { + dev_err(dev, "%s reset failed (%d); trying USB reset\n", + rt == I2400M_RT_WARM ? "warm" : "cold", result); + rt = I2400M_RT_BUS; + goto do_bus_reset; + } + d_fnend(3, dev, "(i2400m %p rt %u) = %d\n", i2400m, rt, result); + return result; +} + + +static +void i2400mu_netdev_setup(struct net_device *net_dev) +{ + struct i2400m *i2400m = net_dev_to_i2400m(net_dev); + struct i2400mu *i2400mu = container_of(i2400m, struct i2400mu, i2400m); + i2400mu_init(i2400mu); + i2400m_netdev_setup(net_dev); +} + + +/* + * Debug levels control; see debug.h + */ +struct d_level D_LEVEL[] = { + D_SUBMODULE_DEFINE(usb), + D_SUBMODULE_DEFINE(fw), + D_SUBMODULE_DEFINE(notif), + D_SUBMODULE_DEFINE(rx), + D_SUBMODULE_DEFINE(tx), +}; +size_t D_LEVEL_SIZE = ARRAY_SIZE(D_LEVEL); + + +#define __debugfs_register(prefix, name, parent) \ +do { \ + result = d_level_register_debugfs(prefix, name, parent); \ + if (result < 0) \ + goto error; \ +} while (0) + + +static +int i2400mu_debugfs_add(struct i2400mu *i2400mu) +{ + int result; + struct device *dev = &i2400mu->usb_iface->dev; + struct dentry *dentry = i2400mu->i2400m.wimax_dev.debugfs_dentry; + struct dentry *fd; + + dentry = debugfs_create_dir("i2400m-usb", dentry); + result = PTR_ERR(dentry); + if (IS_ERR(dentry)) { + if (result == -ENODEV) + result = 0; /* No debugfs support */ + goto error; + } + i2400mu->debugfs_dentry = dentry; + __debugfs_register("dl_", usb, dentry); + __debugfs_register("dl_", fw, dentry); + __debugfs_register("dl_", notif, dentry); + __debugfs_register("dl_", rx, dentry); + __debugfs_register("dl_", tx, dentry); + + /* Don't touch these if you don't know what you are doing */ + fd = debugfs_create_u8("rx_size_auto_shrink", 0600, dentry, + &i2400mu->rx_size_auto_shrink); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "rx_size_auto_shrink: %d\n", result); + goto error; + } + + fd = debugfs_create_size_t("rx_size", 0600, dentry, + &i2400mu->rx_size); + result = PTR_ERR(fd); + if (IS_ERR(fd) && result != -ENODEV) { + dev_err(dev, "Can't create debugfs entry " + "rx_size: %d\n", result); + goto error; + } + + return 0; + +error: + debugfs_remove_recursive(i2400mu->debugfs_dentry); + return result; +} + + +/* + * Probe a i2400m interface and register it + * + * @iface: USB interface to link to + * @id: USB class/subclass/protocol id + * @returns: 0 if ok, < 0 errno code on error. + * + * Alloc a net device, initialize the bus-specific details and then + * calls the bus-generic initialization routine. That will register + * the wimax and netdev devices, upload the firmware [using + * _bus_bm_*()], call _bus_dev_start() to finalize the setup of the + * communication with the device and then will start to talk to it to + * finnish setting it up. + */ +static +int i2400mu_probe(struct usb_interface *iface, + const struct usb_device_id *id) +{ + int result; + struct net_device *net_dev; + struct device *dev = &iface->dev; + struct i2400m *i2400m; + struct i2400mu *i2400mu; + struct usb_device *usb_dev = interface_to_usbdev(iface); + + if (usb_dev->speed != USB_SPEED_HIGH) + dev_err(dev, "device not connected as high speed\n"); + + /* Allocate instance [calls i2400m_netdev_setup() on it]. */ + result = -ENOMEM; + net_dev = alloc_netdev(sizeof(*i2400mu), "wmx%d", + i2400mu_netdev_setup); + if (net_dev == NULL) { + dev_err(dev, "no memory for network device instance\n"); + goto error_alloc_netdev; + } + SET_NETDEV_DEV(net_dev, dev); + i2400m = net_dev_to_i2400m(net_dev); + i2400mu = container_of(i2400m, struct i2400mu, i2400m); + i2400m->wimax_dev.net_dev = net_dev; + i2400mu->usb_dev = usb_get_dev(usb_dev); + i2400mu->usb_iface = iface; + usb_set_intfdata(iface, i2400mu); + + i2400m->bus_tx_block_size = I2400MU_BLK_SIZE; + i2400m->bus_pl_size_max = I2400MU_PL_SIZE_MAX; + i2400m->bus_dev_start = i2400mu_bus_dev_start; + i2400m->bus_dev_stop = i2400mu_bus_dev_stop; + i2400m->bus_tx_kick = i2400mu_bus_tx_kick; + i2400m->bus_reset = i2400mu_bus_reset; + i2400m->bus_bm_cmd_send = i2400mu_bus_bm_cmd_send; + i2400m->bus_bm_wait_for_ack = i2400mu_bus_bm_wait_for_ack; + i2400m->bus_fw_name = I2400MU_FW_FILE_NAME; + i2400m->bus_bm_mac_addr_impaired = 0; + + iface->needs_remote_wakeup = 1; /* autosuspend (15s delay) */ + device_init_wakeup(dev, 1); + usb_autopm_enable(i2400mu->usb_iface); + usb_dev->autosuspend_delay = 15 * HZ; + usb_dev->autosuspend_disabled = 0; + + result = i2400m_setup(i2400m, I2400M_BRI_MAC_REINIT); + if (result < 0) { + dev_err(dev, "cannot setup device: %d\n", result); + goto error_setup; + } + result = i2400mu_debugfs_add(i2400mu); + if (result < 0) { + dev_err(dev, "Can't register i2400mu's debugfs: %d\n", result); + goto error_debugfs_add; + } + return 0; + +error_debugfs_add: + i2400m_release(i2400m); +error_setup: + usb_set_intfdata(iface, NULL); + usb_put_dev(i2400mu->usb_dev); + free_netdev(net_dev); +error_alloc_netdev: + return result; +} + + +/* + * Disconect a i2400m from the system. + * + * i2400m_stop() has been called before, so al the rx and tx contexts + * have been taken down already. Make sure the queue is stopped, + * unregister netdev and i2400m, free and kill. + */ +static +void i2400mu_disconnect(struct usb_interface *iface) +{ + struct i2400mu *i2400mu = usb_get_intfdata(iface); + struct i2400m *i2400m = &i2400mu->i2400m; + struct net_device *net_dev = i2400m->wimax_dev.net_dev; + struct device *dev = &iface->dev; + + d_fnstart(3, dev, "(iface %p i2400m %p)\n", iface, i2400m); + + debugfs_remove_recursive(i2400mu->debugfs_dentry); + i2400m_release(i2400m); + usb_set_intfdata(iface, NULL); + usb_put_dev(i2400mu->usb_dev); + free_netdev(net_dev); + d_fnend(3, dev, "(iface %p i2400m %p) = void\n", iface, i2400m); +} + + +/* + * Get the device ready for USB port or system standby and hibernation + * + * USB port and system standby are handled the same. + * + * When the system hibernates, the USB device is powered down and then + * up, so we don't really have to do much here, as it will be seen as + * a reconnect. Still for simplicity we consider this case the same as + * suspend, so that the device has a chance to do notify the base + * station (if connected). + * + * So at the end, the three cases require common handling. + * + * If at the time of this call the device's firmware is not loaded, + * nothing has to be done. + * + * If the firmware is loaded, we need to: + * + * - tell the device to go into host interface power save mode, wait + * for it to ack + * + * This is quite more interesting than it is; we need to execute a + * command, but this time, we don't want the code in usb-{tx,rx}.c + * to call the usb_autopm_get/put_interface() barriers as it'd + * deadlock, so we need to decrement i2400mu->do_autopm, that acts + * as a poor man's semaphore. Ugly, but it works. + * + * As well, the device might refuse going to sleep for whichever + * reason. In this case we just fail. For system suspend/hibernate, + * we *can't* fail. We look at usb_dev->auto_pm to see if the + * suspend call comes from the USB stack or from the system and act + * in consequence. + * + * - stop the notification endpoint polling + */ +static +int i2400mu_suspend(struct usb_interface *iface, pm_message_t pm_msg) +{ + int result = 0; + struct device *dev = &iface->dev; + struct i2400mu *i2400mu = usb_get_intfdata(iface); + struct usb_device *usb_dev = i2400mu->usb_dev; + struct i2400m *i2400m = &i2400mu->i2400m; + + d_fnstart(3, dev, "(iface %p pm_msg %u)\n", iface, pm_msg.event); + if (i2400m->updown == 0) + goto no_firmware; + d_printf(1, dev, "fw up, requesting standby\n"); + atomic_dec(&i2400mu->do_autopm); + result = i2400m_cmd_enter_powersave(i2400m); + atomic_inc(&i2400mu->do_autopm); + if (result < 0 && usb_dev->auto_pm == 0) { + /* System suspend, can't fail */ + dev_err(dev, "failed to suspend, will reset on resume\n"); + result = 0; + } + if (result < 0) + goto error_enter_powersave; + i2400mu_notification_release(i2400mu); + d_printf(1, dev, "fw up, got standby\n"); +error_enter_powersave: +no_firmware: + d_fnend(3, dev, "(iface %p pm_msg %u) = %d\n", + iface, pm_msg.event, result); + return result; +} + + +static +int i2400mu_resume(struct usb_interface *iface) +{ + int ret = 0; + struct device *dev = &iface->dev; + struct i2400mu *i2400mu = usb_get_intfdata(iface); + struct i2400m *i2400m = &i2400mu->i2400m; + + d_fnstart(3, dev, "(iface %p)\n", iface); + if (i2400m->updown == 0) { + d_printf(1, dev, "fw was down, no resume neeed\n"); + goto out; + } + d_printf(1, dev, "fw was up, resuming\n"); + i2400mu_notification_setup(i2400mu); + /* USB has flow control, so we don't need to give it time to + * come back; otherwise, we'd use something like a get-state + * command... */ +out: + d_fnend(3, dev, "(iface %p) = %d\n", iface, ret); + return ret; +} + + +static +struct usb_device_id i2400mu_id_table[] = { + { USB_DEVICE(0x8086, 0x0181) }, + { USB_DEVICE(0x8086, 0x1403) }, + { USB_DEVICE(0x8086, 0x1405) }, + { USB_DEVICE(0x8086, 0x0180) }, + { USB_DEVICE(0x8086, 0x0182) }, + { USB_DEVICE(0x8086, 0x1406) }, + { USB_DEVICE(0x8086, 0x1403) }, + { }, +}; +MODULE_DEVICE_TABLE(usb, i2400mu_id_table); + + +static +struct usb_driver i2400mu_driver = { + .name = KBUILD_MODNAME, + .suspend = i2400mu_suspend, + .resume = i2400mu_resume, + .probe = i2400mu_probe, + .disconnect = i2400mu_disconnect, + .id_table = i2400mu_id_table, + .supports_autosuspend = 1, +}; + +static +int __init i2400mu_driver_init(void) +{ + return usb_register(&i2400mu_driver); +} +module_init(i2400mu_driver_init); + + +static +void __exit i2400mu_driver_exit(void) +{ + flush_scheduled_work(); /* for the stuff we schedule from sysfs.c */ + usb_deregister(&i2400mu_driver); +} +module_exit(i2400mu_driver_exit); + +MODULE_AUTHOR("Intel Corporation "); +MODULE_DESCRIPTION("Intel 2400M WiMAX networking for USB"); +MODULE_LICENSE("GPL"); +MODULE_FIRMWARE(I2400MU_FW_FILE_NAME); From 795038107b0078ee5ad3ad33327fe1c3520f6bf2 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:52 -0800 Subject: [PATCH 115/123] i2400m/USB: firmware upload backend This implements the backends for the generic driver (i2400m) to be able to load firmware to the USB device. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/usb-fw.c | 340 ++++++++++++++++++++++++++++++ 1 file changed, 340 insertions(+) create mode 100644 drivers/net/wimax/i2400m/usb-fw.c diff --git a/drivers/net/wimax/i2400m/usb-fw.c b/drivers/net/wimax/i2400m/usb-fw.c new file mode 100644 index 00000000000..5ad287c228b --- /dev/null +++ b/drivers/net/wimax/i2400m/usb-fw.c @@ -0,0 +1,340 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Firmware uploader's USB specifics + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Yanir Lubetkin + * Inaky Perez-Gonzalez + * - Initial implementation + * + * Inaky Perez-Gonzalez + * - bus generic/specific split + * + * THE PROCEDURE + * + * See fw.c for the generic description of this procedure. + * + * This file implements only the USB specifics. It boils down to how + * to send a command and waiting for an acknowledgement from the + * device. + * + * This code (and process) is single threaded. It assumes it is the + * only thread poking around (guaranteed by fw.c). + * + * COMMAND EXECUTION + * + * A write URB is posted with the buffer to the bulk output endpoint. + * + * ACK RECEPTION + * + * We just post a URB to the notification endpoint and wait for + * data. We repeat until we get all the data we expect (as indicated + * by the call from the bus generic code). + * + * The data is not read from the bulk in endpoint for boot mode. + * + * ROADMAP + * + * i2400mu_bus_bm_cmd_send + * i2400m_bm_cmd_prepare... + * i2400mu_tx_bulk_out + * + * i2400mu_bus_bm_wait_for_ack + * i2400m_notif_submit + */ +#include +#include "i2400m-usb.h" + + +#define D_SUBMODULE fw +#include "usb-debug-levels.h" + + +/* + * Synchronous write to the device + * + * Takes care of updating EDC counts and thus, handle device errors. + */ +static +ssize_t i2400mu_tx_bulk_out(struct i2400mu *i2400mu, void *buf, size_t buf_size) +{ + int result; + struct device *dev = &i2400mu->usb_iface->dev; + int len; + struct usb_endpoint_descriptor *epd; + int pipe, do_autopm = 1; + + result = usb_autopm_get_interface(i2400mu->usb_iface); + if (result < 0) { + dev_err(dev, "BM-CMD: can't get autopm: %d\n", result); + do_autopm = 0; + } + epd = usb_get_epd(i2400mu->usb_iface, I2400MU_EP_BULK_OUT); + pipe = usb_sndbulkpipe(i2400mu->usb_dev, epd->bEndpointAddress); +retry: + result = usb_bulk_msg(i2400mu->usb_dev, pipe, buf, buf_size, &len, HZ); + switch (result) { + case 0: + if (len != buf_size) { + dev_err(dev, "BM-CMD: short write (%u B vs %zu " + "expected)\n", len, buf_size); + result = -EIO; + break; + } + result = len; + break; + case -EINVAL: /* while removing driver */ + case -ENODEV: /* dev disconnect ... */ + case -ENOENT: /* just ignore it */ + case -ESHUTDOWN: /* and exit */ + case -ECONNRESET: + result = -ESHUTDOWN; + break; + case -ETIMEDOUT: /* bah... */ + break; + default: /* any other? */ + if (edc_inc(&i2400mu->urb_edc, + EDC_MAX_ERRORS, EDC_ERROR_TIMEFRAME)) { + dev_err(dev, "BM-CMD: maximum errors in " + "URB exceeded; resetting device\n"); + usb_queue_reset_device(i2400mu->usb_iface); + result = -ENODEV; + break; + } + dev_err(dev, "BM-CMD: URB error %d, retrying\n", + result); + goto retry; + } + result = len; + if (do_autopm) + usb_autopm_put_interface(i2400mu->usb_iface); + return result; +} + + +/* + * Send a boot-mode command over the bulk-out pipe + * + * Command can be a raw command, which requires no preparation (and + * which might not even be following the command format). Checks that + * the right amount of data was transfered. + * + * To satisfy USB requirements (no onstack, vmalloc or in data segment + * buffers), we copy the command to i2400m->bm_cmd_buf and send it from + * there. + * + * @flags: pass thru from i2400m_bm_cmd() + * @return: cmd_size if ok, < 0 errno code on error. + */ +ssize_t i2400mu_bus_bm_cmd_send(struct i2400m *i2400m, + const struct i2400m_bootrom_header *_cmd, + size_t cmd_size, int flags) +{ + ssize_t result; + struct device *dev = i2400m_dev(i2400m); + struct i2400mu *i2400mu = container_of(i2400m, struct i2400mu, i2400m); + int opcode = _cmd == NULL ? -1 : i2400m_brh_get_opcode(_cmd); + struct i2400m_bootrom_header *cmd; + size_t cmd_size_a = ALIGN(cmd_size, 16); /* USB restriction */ + + d_fnstart(8, dev, "(i2400m %p cmd %p size %zu)\n", + i2400m, _cmd, cmd_size); + result = -E2BIG; + if (cmd_size > I2400M_BM_CMD_BUF_SIZE) + goto error_too_big; + memcpy(i2400m->bm_cmd_buf, _cmd, cmd_size); + cmd = i2400m->bm_cmd_buf; + if (cmd_size_a > cmd_size) /* Zero pad space */ + memset(i2400m->bm_cmd_buf + cmd_size, 0, cmd_size_a - cmd_size); + if ((flags & I2400M_BM_CMD_RAW) == 0) { + if (WARN_ON(i2400m_brh_get_response_required(cmd) == 0)) + dev_warn(dev, "SW BUG: response_required == 0\n"); + i2400m_bm_cmd_prepare(cmd); + } + result = i2400mu_tx_bulk_out(i2400mu, i2400m->bm_cmd_buf, cmd_size); + if (result < 0) { + dev_err(dev, "boot-mode cmd %d: cannot send: %zd\n", + opcode, result); + goto error_cmd_send; + } + if (result != cmd_size) { /* all was transferred? */ + dev_err(dev, "boot-mode cmd %d: incomplete transfer " + "(%zu vs %zu submitted)\n", opcode, result, cmd_size); + result = -EIO; + goto error_cmd_size; + } +error_cmd_size: +error_cmd_send: +error_too_big: + d_fnend(8, dev, "(i2400m %p cmd %p size %zu) = %zd\n", + i2400m, _cmd, cmd_size, result); + return result; +} + + +static +void __i2400mu_bm_notif_cb(struct urb *urb) +{ + complete(urb->context); +} + + +/* + * submit a read to the notification endpoint + * + * @i2400m: device descriptor + * @urb: urb to use + * @completion: completion varible to complete when done + * + * Data is always read to i2400m->bm_ack_buf + */ +static +int i2400mu_notif_submit(struct i2400mu *i2400mu, struct urb *urb, + struct completion *completion) +{ + struct i2400m *i2400m = &i2400mu->i2400m; + struct usb_endpoint_descriptor *epd; + int pipe; + + epd = usb_get_epd(i2400mu->usb_iface, I2400MU_EP_NOTIFICATION); + pipe = usb_rcvintpipe(i2400mu->usb_dev, epd->bEndpointAddress); + usb_fill_int_urb(urb, i2400mu->usb_dev, pipe, + i2400m->bm_ack_buf, I2400M_BM_ACK_BUF_SIZE, + __i2400mu_bm_notif_cb, completion, + epd->bInterval); + return usb_submit_urb(urb, GFP_KERNEL); +} + + +/* + * Read an ack from the notification endpoint + * + * @i2400m: + * @_ack: pointer to where to store the read data + * @ack_size: how many bytes we should read + * + * Returns: < 0 errno code on error; otherwise, amount of received bytes. + * + * Submits a notification read, appends the read data to the given ack + * buffer and then repeats (until @ack_size bytes have been + * received). + */ +ssize_t i2400mu_bus_bm_wait_for_ack(struct i2400m *i2400m, + struct i2400m_bootrom_header *_ack, + size_t ack_size) +{ + ssize_t result = -ENOMEM; + struct device *dev = i2400m_dev(i2400m); + struct i2400mu *i2400mu = container_of(i2400m, struct i2400mu, i2400m); + struct urb notif_urb; + void *ack = _ack; + size_t offset, len; + long val; + int do_autopm = 1; + DECLARE_COMPLETION_ONSTACK(notif_completion); + + d_fnstart(8, dev, "(i2400m %p ack %p size %zu)\n", + i2400m, ack, ack_size); + BUG_ON(_ack == i2400m->bm_ack_buf); + result = usb_autopm_get_interface(i2400mu->usb_iface); + if (result < 0) { + dev_err(dev, "BM-ACK: can't get autopm: %d\n", (int) result); + do_autopm = 0; + } + usb_init_urb(¬if_urb); /* ready notifications */ + usb_get_urb(¬if_urb); + offset = 0; + while (offset < ack_size) { + init_completion(¬if_completion); + result = i2400mu_notif_submit(i2400mu, ¬if_urb, + ¬if_completion); + if (result < 0) + goto error_notif_urb_submit; + val = wait_for_completion_interruptible_timeout( + ¬if_completion, HZ); + if (val == 0) { + result = -ETIMEDOUT; + usb_kill_urb(¬if_urb); /* Timedout */ + goto error_notif_wait; + } + if (val == -ERESTARTSYS) { + result = -EINTR; /* Interrupted */ + usb_kill_urb(¬if_urb); + goto error_notif_wait; + } + result = notif_urb.status; /* How was the ack? */ + switch (result) { + case 0: + break; + case -EINVAL: /* while removing driver */ + case -ENODEV: /* dev disconnect ... */ + case -ENOENT: /* just ignore it */ + case -ESHUTDOWN: /* and exit */ + case -ECONNRESET: + result = -ESHUTDOWN; + goto error_dev_gone; + default: /* any other? */ + usb_kill_urb(¬if_urb); /* Timedout */ + if (edc_inc(&i2400mu->urb_edc, + EDC_MAX_ERRORS, EDC_ERROR_TIMEFRAME)) + goto error_exceeded; + dev_err(dev, "BM-ACK: URB error %d, " + "retrying\n", notif_urb.status); + continue; /* retry */ + } + if (notif_urb.actual_length == 0) { + d_printf(6, dev, "ZLP received, retrying\n"); + continue; + } + /* Got data, append it to the buffer */ + len = min(ack_size - offset, (size_t) notif_urb.actual_length); + memcpy(ack + offset, i2400m->bm_ack_buf, len); + offset += len; + } + result = offset; +error_notif_urb_submit: +error_notif_wait: +error_dev_gone: +out: + if (do_autopm) + usb_autopm_put_interface(i2400mu->usb_iface); + d_fnend(8, dev, "(i2400m %p ack %p size %zu) = %zd\n", + i2400m, ack, ack_size, result); + return result; + +error_exceeded: + dev_err(dev, "bm: maximum errors in notification URB exceeded; " + "resetting device\n"); + usb_queue_reset_device(i2400mu->usb_iface); + goto out; +} From a8ebf98f541463107bb9544a1b611981dc2477e7 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:53 -0800 Subject: [PATCH 116/123] i2400m/USB: TX and RX path backends Implements the backend so that the generic driver can TX/RX to/from the USB device. TX is implemented with a kthread sitting in a never-ending loop that when kicked by the generic driver's TX code will pull data from the TX FIFO and send it to the device until it drains it. Then it goes back sleep, waiting for another kick. RX is implemented in a similar fashion, but reads are kicked in by the device notifying in the interrupt endpoint that data is ready. Device reset notifications are also sent via the notification endpoint. We need a thread contexts to run USB autopm functions (blocking) and to process the received data (can get to be heavy in processing time). Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/usb-notif.c | 269 +++++++++++++++++ drivers/net/wimax/i2400m/usb-rx.c | 417 +++++++++++++++++++++++++++ drivers/net/wimax/i2400m/usb-tx.c | 229 +++++++++++++++ 3 files changed, 915 insertions(+) create mode 100644 drivers/net/wimax/i2400m/usb-notif.c create mode 100644 drivers/net/wimax/i2400m/usb-rx.c create mode 100644 drivers/net/wimax/i2400m/usb-tx.c diff --git a/drivers/net/wimax/i2400m/usb-notif.c b/drivers/net/wimax/i2400m/usb-notif.c new file mode 100644 index 00000000000..9702c22b249 --- /dev/null +++ b/drivers/net/wimax/i2400m/usb-notif.c @@ -0,0 +1,269 @@ +/* + * Intel Wireless WiMAX Connection 2400m over USB + * Notification handling + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Yanir Lubetkin + * Inaky Perez-Gonzalez + * - Initial implementation + * + * + * The notification endpoint is active when the device is not in boot + * mode; in here we just read and get notifications; based on those, + * we act to either reinitialize the device after a reboot or to + * submit a RX request. + * + * ROADMAP + * + * i2400mu_usb_notification_setup() + * + * i2400mu_usb_notification_release() + * + * i2400mu_usb_notification_cb() Called when a URB is ready + * i2400mu_notif_grok() + * i2400m_dev_reset_handle() + * i2400mu_rx_kick() + */ +#include +#include "i2400m-usb.h" + + +#define D_SUBMODULE notif +#include "usb-debug-levels.h" + + +static const +__le32 i2400m_ZERO_BARKER[4] = { 0, 0, 0, 0 }; + + +/* + * Process a received notification + * + * In normal operation mode, we can only receive two types of payloads + * on the notification endpoint: + * + * - a reboot barker, we do a bootstrap (the device has reseted). + * + * - a block of zeroes: there is pending data in the IN endpoint + */ +static +int i2400mu_notification_grok(struct i2400mu *i2400mu, const void *buf, + size_t buf_len) +{ + int ret; + struct device *dev = &i2400mu->usb_iface->dev; + struct i2400m *i2400m = &i2400mu->i2400m; + + d_fnstart(4, dev, "(i2400m %p buf %p buf_len %zu)\n", + i2400mu, buf, buf_len); + ret = -EIO; + if (buf_len < sizeof(i2400m_NBOOT_BARKER)) + /* Not a bug, just ignore */ + goto error_bad_size; + if (!memcmp(i2400m_NBOOT_BARKER, buf, sizeof(i2400m_NBOOT_BARKER)) + || !memcmp(i2400m_SBOOT_BARKER, buf, sizeof(i2400m_SBOOT_BARKER))) + ret = i2400m_dev_reset_handle(i2400m); + else if (!memcmp(i2400m_ZERO_BARKER, buf, sizeof(i2400m_ZERO_BARKER))) { + i2400mu_rx_kick(i2400mu); + ret = 0; + } else { /* Unknown or unexpected data in the notif message */ + char prefix[64]; + ret = -EIO; + dev_err(dev, "HW BUG? Unknown/unexpected data in notification " + "message (%zu bytes)\n", buf_len); + snprintf(prefix, sizeof(prefix), "%s %s: ", + dev_driver_string(dev) , dev->bus_id); + if (buf_len > 64) { + print_hex_dump(KERN_ERR, prefix, DUMP_PREFIX_OFFSET, + 8, 4, buf, 64, 0); + printk(KERN_ERR "%s... (only first 64 bytes " + "dumped)\n", prefix); + } else + print_hex_dump(KERN_ERR, prefix, DUMP_PREFIX_OFFSET, + 8, 4, buf, buf_len, 0); + } +error_bad_size: + d_fnend(4, dev, "(i2400m %p buf %p buf_len %zu) = %d\n", + i2400mu, buf, buf_len, ret); + return ret; +} + + +/* + * URB callback for the notification endpoint + * + * @urb: the urb received from the notification endpoint + * + * This function will just process the USB side of the transaction, + * checking everything is fine, pass the processing to + * i2400m_notification_grok() and resubmit the URB. + */ +static +void i2400mu_notification_cb(struct urb *urb) +{ + int ret; + struct i2400mu *i2400mu = urb->context; + struct device *dev = &i2400mu->usb_iface->dev; + + d_fnstart(4, dev, "(urb %p status %d actual_length %d)\n", + urb, urb->status, urb->actual_length); + ret = urb->status; + switch (ret) { + case 0: + ret = i2400mu_notification_grok(i2400mu, urb->transfer_buffer, + urb->actual_length); + if (ret == -EIO && edc_inc(&i2400mu->urb_edc, EDC_MAX_ERRORS, + EDC_ERROR_TIMEFRAME)) + goto error_exceeded; + if (ret == -ENOMEM) /* uff...power cycle? shutdown? */ + goto error_exceeded; + break; + case -EINVAL: /* while removing driver */ + case -ENODEV: /* dev disconnect ... */ + case -ENOENT: /* ditto */ + case -ESHUTDOWN: /* URB killed */ + case -ECONNRESET: /* disconnection */ + goto out; /* Notify around */ + default: /* Some error? */ + if (edc_inc(&i2400mu->urb_edc, + EDC_MAX_ERRORS, EDC_ERROR_TIMEFRAME)) + goto error_exceeded; + dev_err(dev, "notification: URB error %d, retrying\n", + urb->status); + } + usb_mark_last_busy(i2400mu->usb_dev); + ret = usb_submit_urb(i2400mu->notif_urb, GFP_ATOMIC); + switch (ret) { + case 0: + case -EINVAL: /* while removing driver */ + case -ENODEV: /* dev disconnect ... */ + case -ENOENT: /* ditto */ + case -ESHUTDOWN: /* URB killed */ + case -ECONNRESET: /* disconnection */ + break; /* just ignore */ + default: /* Some error? */ + dev_err(dev, "notification: cannot submit URB: %d\n", ret); + goto error_submit; + } + d_fnend(4, dev, "(urb %p status %d actual_length %d) = void\n", + urb, urb->status, urb->actual_length); + return; + +error_exceeded: + dev_err(dev, "maximum errors in notification URB exceeded; " + "resetting device\n"); +error_submit: + usb_queue_reset_device(i2400mu->usb_iface); +out: + d_fnend(4, dev, "(urb %p status %d actual_length %d) = void\n", + urb, urb->status, urb->actual_length); + return; +} + + +/* + * setup the notification endpoint + * + * @i2400m: device descriptor + * + * This procedure prepares the notification urb and handler for receiving + * unsolicited barkers from the device. + */ +int i2400mu_notification_setup(struct i2400mu *i2400mu) +{ + struct device *dev = &i2400mu->usb_iface->dev; + int usb_pipe, ret = 0; + struct usb_endpoint_descriptor *epd; + char *buf; + + d_fnstart(4, dev, "(i2400m %p)\n", i2400mu); + buf = kmalloc(I2400MU_MAX_NOTIFICATION_LEN, GFP_KERNEL | GFP_DMA); + if (buf == NULL) { + dev_err(dev, "notification: buffer allocation failed\n"); + ret = -ENOMEM; + goto error_buf_alloc; + } + + i2400mu->notif_urb = usb_alloc_urb(0, GFP_KERNEL); + if (!i2400mu->notif_urb) { + ret = -ENOMEM; + dev_err(dev, "notification: cannot allocate URB\n"); + goto error_alloc_urb; + } + epd = usb_get_epd(i2400mu->usb_iface, I2400MU_EP_NOTIFICATION); + usb_pipe = usb_rcvintpipe(i2400mu->usb_dev, epd->bEndpointAddress); + usb_fill_int_urb(i2400mu->notif_urb, i2400mu->usb_dev, usb_pipe, + buf, I2400MU_MAX_NOTIFICATION_LEN, + i2400mu_notification_cb, i2400mu, epd->bInterval); + ret = usb_submit_urb(i2400mu->notif_urb, GFP_KERNEL); + if (ret != 0) { + dev_err(dev, "notification: cannot submit URB: %d\n", ret); + goto error_submit; + } + d_fnend(4, dev, "(i2400m %p) = %d\n", i2400mu, ret); + return ret; + +error_submit: + usb_free_urb(i2400mu->notif_urb); +error_alloc_urb: + kfree(buf); +error_buf_alloc: + d_fnend(4, dev, "(i2400m %p) = %d\n", i2400mu, ret); + return ret; +} + + +/* + * Tear down of the notification mechanism + * + * @i2400m: device descriptor + * + * Kill the interrupt endpoint urb, free any allocated resources. + * + * We need to check if we have done it before as for example, + * _suspend() call this; if after a suspend() we get a _disconnect() + * (as the case is when hibernating), nothing bad happens. + */ +void i2400mu_notification_release(struct i2400mu *i2400mu) +{ + struct device *dev = &i2400mu->usb_iface->dev; + + d_fnstart(4, dev, "(i2400mu %p)\n", i2400mu); + if (i2400mu->notif_urb != NULL) { + usb_kill_urb(i2400mu->notif_urb); + kfree(i2400mu->notif_urb->transfer_buffer); + usb_free_urb(i2400mu->notif_urb); + i2400mu->notif_urb = NULL; + } + d_fnend(4, dev, "(i2400mu %p)\n", i2400mu); +} diff --git a/drivers/net/wimax/i2400m/usb-rx.c b/drivers/net/wimax/i2400m/usb-rx.c new file mode 100644 index 00000000000..074cc1f8985 --- /dev/null +++ b/drivers/net/wimax/i2400m/usb-rx.c @@ -0,0 +1,417 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * USB RX handling + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Yanir Lubetkin + * - Initial implementation + * Inaky Perez-Gonzalez + * - Use skb_clone(), break up processing in chunks + * - Split transport/device specific + * - Make buffer size dynamic to exert less memory pressure + * + * + * This handles the RX path on USB. + * + * When a notification is received that says 'there is RX data ready', + * we call i2400mu_rx_kick(); that wakes up the RX kthread, which + * reads a buffer from USB and passes it to i2400m_rx() in the generic + * handling code. The RX buffer has an specific format that is + * described in rx.c. + * + * We use a kernel thread in a loop because: + * + * - we want to be able to call the USB power management get/put + * functions (blocking) before each transaction. + * + * - We might get a lot of notifications and we don't want to submit + * a zillion reads; by serializing, we are throttling. + * + * - RX data processing can get heavy enough so that it is not + * appropiate for doing it in the USB callback; thus we run it in a + * process context. + * + * We provide a read buffer of an arbitrary size (short of a page); if + * the callback reports -EOVERFLOW, it means it was too small, so we + * just double the size and retry (being careful to append, as + * sometimes the device provided some data). Every now and then we + * check if the average packet size is smaller than the current packet + * size and if so, we halve it. At the end, the size of the + * preallocated buffer should be following the average received + * transaction size, adapting dynamically to it. + * + * ROADMAP + * + * i2400mu_rx_kick() Called from notif.c when we get a + * 'data ready' notification + * i2400mu_rxd() Kernel RX daemon + * i2400mu_rx() Receive USB data + * i2400m_rx() Send data to generic i2400m RX handling + * + * i2400mu_rx_setup() called from i2400mu_bus_dev_start() + * + * i2400mu_rx_release() called from i2400mu_bus_dev_stop() + */ +#include +#include +#include "i2400m-usb.h" + + +#define D_SUBMODULE rx +#include "usb-debug-levels.h" + +/* + * Dynamic RX size + * + * We can't let the rx_size be a multiple of 512 bytes (the RX + * endpoint's max packet size). On some USB host controllers (we + * haven't been able to fully characterize which), if the device is + * about to send (for example) X bytes and we only post a buffer to + * receive n*512, it will fail to mark that as babble (so that + * i2400mu_rx() [case -EOVERFLOW] can resize the buffer and get the + * rest). + * + * So on growing or shrinking, if it is a multiple of the + * maxpacketsize, we remove some (instead of incresing some, so in a + * buddy allocator we try to waste less space). + * + * Note we also need a hook for this on i2400mu_rx() -- when we do the + * first read, we are sure we won't hit this spot because + * i240mm->rx_size has been set properly. However, if we have to + * double because of -EOVERFLOW, when we launch the read to get the + * rest of the data, we *have* to make sure that also is not a + * multiple of the max_pkt_size. + */ + +static +size_t i2400mu_rx_size_grow(struct i2400mu *i2400mu) +{ + struct device *dev = &i2400mu->usb_iface->dev; + size_t rx_size; + const size_t max_pkt_size = 512; + + rx_size = 2 * i2400mu->rx_size; + if (rx_size % max_pkt_size == 0) { + rx_size -= 8; + d_printf(1, dev, + "RX: expected size grew to %zu [adjusted -8] " + "from %zu\n", + rx_size, i2400mu->rx_size); + } else + d_printf(1, dev, + "RX: expected size grew to %zu from %zu\n", + rx_size, i2400mu->rx_size); + return rx_size; +} + + +static +void i2400mu_rx_size_maybe_shrink(struct i2400mu *i2400mu) +{ + const size_t max_pkt_size = 512; + struct device *dev = &i2400mu->usb_iface->dev; + + if (unlikely(i2400mu->rx_size_cnt >= 100 + && i2400mu->rx_size_auto_shrink)) { + size_t avg_rx_size = + i2400mu->rx_size_acc / i2400mu->rx_size_cnt; + size_t new_rx_size = i2400mu->rx_size / 2; + if (avg_rx_size < new_rx_size) { + if (new_rx_size % max_pkt_size == 0) { + new_rx_size -= 8; + d_printf(1, dev, + "RX: expected size shrank to %zu " + "[adjusted -8] from %zu\n", + new_rx_size, i2400mu->rx_size); + } else + d_printf(1, dev, + "RX: expected size shrank to %zu " + "from %zu\n", + new_rx_size, i2400mu->rx_size); + i2400mu->rx_size = new_rx_size; + i2400mu->rx_size_cnt = 0; + i2400mu->rx_size_acc = i2400mu->rx_size; + } + } +} + +/* + * Receive a message with payloads from the USB bus into an skb + * + * @i2400mu: USB device descriptor + * @rx_skb: skb where to place the received message + * + * Deals with all the USB-specifics of receiving, dynamically + * increasing the buffer size if so needed. Returns the payload in the + * skb, ready to process. On a zero-length packet, we retry. + * + * On soft USB errors, we retry (until they become too frequent and + * then are promoted to hard); on hard USB errors, we reset the + * device. On other errors (skb realloacation, we just drop it and + * hope for the next invocation to solve it). + * + * Returns: pointer to the skb if ok, ERR_PTR on error. + * NOTE: this function might realloc the skb (if it is too small), + * so always update with the one returned. + * ERR_PTR() is < 0 on error. + */ +static +struct sk_buff *i2400mu_rx(struct i2400mu *i2400mu, struct sk_buff *rx_skb) +{ + int result = 0; + struct device *dev = &i2400mu->usb_iface->dev; + int usb_pipe, read_size, rx_size, do_autopm; + struct usb_endpoint_descriptor *epd; + const size_t max_pkt_size = 512; + + d_fnstart(4, dev, "(i2400mu %p)\n", i2400mu); + do_autopm = atomic_read(&i2400mu->do_autopm); + result = do_autopm ? + usb_autopm_get_interface(i2400mu->usb_iface) : 0; + if (result < 0) { + dev_err(dev, "RX: can't get autopm: %d\n", result); + do_autopm = 0; + } + epd = usb_get_epd(i2400mu->usb_iface, I2400MU_EP_BULK_IN); + usb_pipe = usb_rcvbulkpipe(i2400mu->usb_dev, epd->bEndpointAddress); +retry: + rx_size = skb_end_pointer(rx_skb) - rx_skb->data - rx_skb->len; + if (unlikely(rx_size % max_pkt_size == 0)) { + rx_size -= 8; + d_printf(1, dev, "RX: rx_size adapted to %d [-8]\n", rx_size); + } + result = usb_bulk_msg( + i2400mu->usb_dev, usb_pipe, rx_skb->data + rx_skb->len, + rx_size, &read_size, HZ); + usb_mark_last_busy(i2400mu->usb_dev); + switch (result) { + case 0: + if (read_size == 0) + goto retry; /* ZLP, just resubmit */ + skb_put(rx_skb, read_size); + break; + case -EINVAL: /* while removing driver */ + case -ENODEV: /* dev disconnect ... */ + case -ENOENT: /* just ignore it */ + case -ESHUTDOWN: + case -ECONNRESET: + break; + case -EOVERFLOW: { /* too small, reallocate */ + struct sk_buff *new_skb; + rx_size = i2400mu_rx_size_grow(i2400mu); + if (rx_size <= (1 << 16)) /* cap it */ + i2400mu->rx_size = rx_size; + else if (printk_ratelimit()) { + dev_err(dev, "BUG? rx_size up to %d\n", rx_size); + result = -EINVAL; + goto out; + } + skb_put(rx_skb, read_size); + new_skb = skb_copy_expand(rx_skb, 0, rx_size - rx_skb->len, + GFP_KERNEL); + if (new_skb == NULL) { + if (printk_ratelimit()) + dev_err(dev, "RX: Can't reallocate skb to %d; " + "RX dropped\n", rx_size); + kfree(rx_skb); + result = 0; + goto out; /* drop it...*/ + } + kfree_skb(rx_skb); + rx_skb = new_skb; + i2400mu->rx_size_cnt = 0; + i2400mu->rx_size_acc = i2400mu->rx_size; + d_printf(1, dev, "RX: size changed to %d, received %d, " + "copied %d, capacity %ld\n", + rx_size, read_size, rx_skb->len, + (long) (skb_end_pointer(new_skb) - new_skb->head)); + goto retry; + } + /* In most cases, it happens due to the hardware scheduling a + * read when there was no data - unfortunately, we have no way + * to tell this timeout from a USB timeout. So we just ignore + * it. */ + case -ETIMEDOUT: + dev_err(dev, "RX: timeout: %d\n", result); + result = 0; + break; + default: /* Any error */ + if (edc_inc(&i2400mu->urb_edc, + EDC_MAX_ERRORS, EDC_ERROR_TIMEFRAME)) + goto error_reset; + dev_err(dev, "RX: error receiving URB: %d, retrying\n", result); + goto retry; + } +out: + if (do_autopm) + usb_autopm_put_interface(i2400mu->usb_iface); + d_fnend(4, dev, "(i2400mu %p) = %p\n", i2400mu, rx_skb); + return rx_skb; + +error_reset: + dev_err(dev, "RX: maximum errors in URB exceeded; " + "resetting device\n"); + usb_queue_reset_device(i2400mu->usb_iface); + rx_skb = ERR_PTR(result); + goto out; +} + + +/* + * Kernel thread for USB reception of data + * + * This thread waits for a kick; once kicked, it will allocate an skb + * and receive a single message to it from USB (using + * i2400mu_rx()). Once received, it is passed to the generic i2400m RX + * code for processing. + * + * When done processing, it runs some dirty statistics to verify if + * the last 100 messages received were smaller than half of the + * current RX buffer size. In that case, the RX buffer size is + * halved. This will helps lowering the pressure on the memory + * allocator. + * + * Hard errors force the thread to exit. + */ +static +int i2400mu_rxd(void *_i2400mu) +{ + int result = 0; + struct i2400mu *i2400mu = _i2400mu; + struct i2400m *i2400m = &i2400mu->i2400m; + struct device *dev = &i2400mu->usb_iface->dev; + struct net_device *net_dev = i2400m->wimax_dev.net_dev; + size_t pending; + int rx_size; + struct sk_buff *rx_skb; + + d_fnstart(4, dev, "(i2400mu %p)\n", i2400mu); + while (1) { + d_printf(2, dev, "TX: waiting for messages\n"); + pending = 0; + wait_event_interruptible( + i2400mu->rx_wq, + (kthread_should_stop() /* check this first! */ + || (pending = atomic_read(&i2400mu->rx_pending_count))) + ); + if (kthread_should_stop()) + break; + if (pending == 0) + continue; + rx_size = i2400mu->rx_size; + d_printf(2, dev, "RX: reading up to %d bytes\n", rx_size); + rx_skb = __netdev_alloc_skb(net_dev, rx_size, GFP_KERNEL); + if (rx_skb == NULL) { + dev_err(dev, "RX: can't allocate skb [%d bytes]\n", + rx_size); + msleep(50); /* give it some time? */ + continue; + } + + /* Receive the message with the payloads */ + rx_skb = i2400mu_rx(i2400mu, rx_skb); + result = PTR_ERR(rx_skb); + if (IS_ERR(rx_skb)) + goto out; + atomic_dec(&i2400mu->rx_pending_count); + if (rx_skb->len == 0) { /* some ignorable condition */ + kfree_skb(rx_skb); + continue; + } + + /* Deliver the message to the generic i2400m code */ + i2400mu->rx_size_cnt++; + i2400mu->rx_size_acc += rx_skb->len; + result = i2400m_rx(i2400m, rx_skb); + if (result == -EIO + && edc_inc(&i2400mu->urb_edc, + EDC_MAX_ERRORS, EDC_ERROR_TIMEFRAME)) { + goto error_reset; + } + + /* Maybe adjust RX buffer size */ + i2400mu_rx_size_maybe_shrink(i2400mu); + } + result = 0; +out: + d_fnend(4, dev, "(i2400mu %p) = %d\n", i2400mu, result); + return result; + +error_reset: + dev_err(dev, "RX: maximum errors in received buffer exceeded; " + "resetting device\n"); + usb_queue_reset_device(i2400mu->usb_iface); + goto out; +} + + +/* + * Start reading from the device + * + * @i2400m: device instance + * + * Notify the RX thread that there is data pending. + */ +void i2400mu_rx_kick(struct i2400mu *i2400mu) +{ + struct i2400m *i2400m = &i2400mu->i2400m; + struct device *dev = &i2400mu->usb_iface->dev; + + d_fnstart(3, dev, "(i2400mu %p)\n", i2400m); + atomic_inc(&i2400mu->rx_pending_count); + wake_up_all(&i2400mu->rx_wq); + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); +} + + +int i2400mu_rx_setup(struct i2400mu *i2400mu) +{ + int result = 0; + struct i2400m *i2400m = &i2400mu->i2400m; + struct device *dev = &i2400mu->usb_iface->dev; + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + + i2400mu->rx_kthread = kthread_run(i2400mu_rxd, i2400mu, "%s-rx", + wimax_dev->name); + if (IS_ERR(i2400mu->rx_kthread)) { + result = PTR_ERR(i2400mu->rx_kthread); + dev_err(dev, "RX: cannot start thread: %d\n", result); + } + return result; +} + +void i2400mu_rx_release(struct i2400mu *i2400mu) +{ + kthread_stop(i2400mu->rx_kthread); +} + diff --git a/drivers/net/wimax/i2400m/usb-tx.c b/drivers/net/wimax/i2400m/usb-tx.c new file mode 100644 index 00000000000..dfd893356f4 --- /dev/null +++ b/drivers/net/wimax/i2400m/usb-tx.c @@ -0,0 +1,229 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * USB specific TX handling + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Yanir Lubetkin + * - Initial implementation + * Inaky Perez-Gonzalez + * - Split transport/device specific + * + * + * Takes the TX messages in the i2400m's driver TX FIFO and sends them + * to the device until there are no more. + * + * If we fail sending the message, we just drop it. There isn't much + * we can do at this point. We could also retry, but the USB stack has + * already retried and still failed, so there is not much of a + * point. As well, most of the traffic is network, which has recovery + * methods for dropped packets. + * + * For sending we just obtain a FIFO buffer to send, send it to the + * USB bulk out, tell the TX FIFO code we have sent it; query for + * another one, etc... until done. + * + * We use a thread so we can call usb_autopm_enable() and + * usb_autopm_disable() for each transaction; this way when the device + * goes idle, it will suspend. It also has less overhead than a + * dedicated workqueue, as it is being used for a single task. + * + * ROADMAP + * + * i2400mu_tx_setup() + * i2400mu_tx_release() + * + * i2400mu_bus_tx_kick() - Called by the tx.c code when there + * is new data in the FIFO. + * i2400mu_txd() + * i2400m_tx_msg_get() + * i2400m_tx_msg_sent() + */ +#include "i2400m-usb.h" + + +#define D_SUBMODULE tx +#include "usb-debug-levels.h" + + +/* + * Get the next TX message in the TX FIFO and send it to the device + * + * Note that any iteration consumes a message to be sent, no matter if + * it succeeds or fails (we have no real way to retry or complain). + * + * Return: 0 if ok, < 0 errno code on hard error. + */ +static +int i2400mu_tx(struct i2400mu *i2400mu, struct i2400m_msg_hdr *tx_msg, + size_t tx_msg_size) +{ + int result = 0; + struct i2400m *i2400m = &i2400mu->i2400m; + struct device *dev = &i2400mu->usb_iface->dev; + int usb_pipe, sent_size, do_autopm; + struct usb_endpoint_descriptor *epd; + + d_fnstart(4, dev, "(i2400mu %p)\n", i2400mu); + do_autopm = atomic_read(&i2400mu->do_autopm); + result = do_autopm ? + usb_autopm_get_interface(i2400mu->usb_iface) : 0; + if (result < 0) { + dev_err(dev, "TX: can't get autopm: %d\n", result); + do_autopm = 0; + } + epd = usb_get_epd(i2400mu->usb_iface, I2400MU_EP_BULK_OUT); + usb_pipe = usb_sndbulkpipe(i2400mu->usb_dev, epd->bEndpointAddress); +retry: + result = usb_bulk_msg(i2400mu->usb_dev, usb_pipe, + tx_msg, tx_msg_size, &sent_size, HZ); + usb_mark_last_busy(i2400mu->usb_dev); + switch (result) { + case 0: + if (sent_size != tx_msg_size) { /* Too short? drop it */ + dev_err(dev, "TX: short write (%d B vs %zu " + "expected)\n", sent_size, tx_msg_size); + result = -EIO; + } + break; + case -EINVAL: /* while removing driver */ + case -ENODEV: /* dev disconnect ... */ + case -ENOENT: /* just ignore it */ + case -ESHUTDOWN: /* and exit */ + case -ECONNRESET: + result = -ESHUTDOWN; + break; + default: /* Some error? */ + if (edc_inc(&i2400mu->urb_edc, + EDC_MAX_ERRORS, EDC_ERROR_TIMEFRAME)) { + dev_err(dev, "TX: maximum errors in URB " + "exceeded; resetting device\n"); + usb_queue_reset_device(i2400mu->usb_iface); + } else { + dev_err(dev, "TX: cannot send URB; retrying. " + "tx_msg @%zu %zu B [%d sent]: %d\n", + (void *) tx_msg - i2400m->tx_buf, + tx_msg_size, sent_size, result); + goto retry; + } + } + if (do_autopm) + usb_autopm_put_interface(i2400mu->usb_iface); + d_fnend(4, dev, "(i2400mu %p) = result\n", i2400mu); + return result; +} + + +/* + * Get the next TX message in the TX FIFO and send it to the device + * + * Note we exit the loop if i2400mu_tx() fails; that funtion only + * fails on hard error (failing to tx a buffer not being one of them, + * see its doc). + * + * Return: 0 + */ +static +int i2400mu_txd(void *_i2400mu) +{ + int result = 0; + struct i2400mu *i2400mu = _i2400mu; + struct i2400m *i2400m = &i2400mu->i2400m; + struct device *dev = &i2400mu->usb_iface->dev; + struct i2400m_msg_hdr *tx_msg; + size_t tx_msg_size; + + d_fnstart(4, dev, "(i2400mu %p)\n", i2400mu); + + while (1) { + d_printf(2, dev, "TX: waiting for messages\n"); + tx_msg = NULL; + wait_event_interruptible( + i2400mu->tx_wq, + (kthread_should_stop() /* check this first! */ + || (tx_msg = i2400m_tx_msg_get(i2400m, &tx_msg_size))) + ); + if (kthread_should_stop()) + break; + WARN_ON(tx_msg == NULL); /* should not happen...*/ + d_printf(2, dev, "TX: submitting %zu bytes\n", tx_msg_size); + d_dump(5, dev, tx_msg, tx_msg_size); + /* Yeah, we ignore errors ... not much we can do */ + i2400mu_tx(i2400mu, tx_msg, tx_msg_size); + i2400m_tx_msg_sent(i2400m); /* ack it, advance the FIFO */ + if (result < 0) + break; + } + d_fnend(4, dev, "(i2400mu %p) = %d\n", i2400mu, result); + return result; +} + + +/* + * i2400m TX engine notifies us that there is data in the FIFO ready + * for TX + * + * If there is a URB in flight, don't do anything; when it finishes, + * it will see there is data in the FIFO and send it. Else, just + * submit a write. + */ +void i2400mu_bus_tx_kick(struct i2400m *i2400m) +{ + struct i2400mu *i2400mu = container_of(i2400m, struct i2400mu, i2400m); + struct device *dev = &i2400mu->usb_iface->dev; + + d_fnstart(3, dev, "(i2400m %p) = void\n", i2400m); + wake_up_all(&i2400mu->tx_wq); + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); +} + + +int i2400mu_tx_setup(struct i2400mu *i2400mu) +{ + int result = 0; + struct i2400m *i2400m = &i2400mu->i2400m; + struct device *dev = &i2400mu->usb_iface->dev; + struct wimax_dev *wimax_dev = &i2400m->wimax_dev; + + i2400mu->tx_kthread = kthread_run(i2400mu_txd, i2400mu, "%s-tx", + wimax_dev->name); + if (IS_ERR(i2400mu->tx_kthread)) { + result = PTR_ERR(i2400mu->tx_kthread); + dev_err(dev, "TX: cannot start thread: %d\n", result); + } + return result; +} + +void i2400mu_tx_release(struct i2400mu *i2400mu) +{ + kthread_stop(i2400mu->tx_kthread); +} From 17d559af963995e483a51ec26697034431bcf2b9 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:54 -0800 Subject: [PATCH 117/123] i2400m/SDIO: header for the SDIO subdriver This contains the common function declaration and constants for the SDIO driver for the 2400m Wireless WiMAX Connection and it's debug level settings. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/i2400m-sdio.h | 132 +++++++++++++++++++ drivers/net/wimax/i2400m/sdio-debug-levels.h | 22 ++++ 2 files changed, 154 insertions(+) create mode 100644 drivers/net/wimax/i2400m/i2400m-sdio.h create mode 100644 drivers/net/wimax/i2400m/sdio-debug-levels.h diff --git a/drivers/net/wimax/i2400m/i2400m-sdio.h b/drivers/net/wimax/i2400m/i2400m-sdio.h new file mode 100644 index 00000000000..08c2fb73923 --- /dev/null +++ b/drivers/net/wimax/i2400m/i2400m-sdio.h @@ -0,0 +1,132 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * SDIO-specific i2400m driver definitions + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Brian Bian + * Dirk Brandewie + * Inaky Perez-Gonzalez + * Yanir Lubetkin + * - Initial implementation + * + * + * This driver implements the bus-specific part of the i2400m for + * SDIO. Check i2400m.h for a generic driver description. + * + * ARCHITECTURE + * + * This driver sits under the bus-generic i2400m driver, providing the + * connection to the device. + * + * When probed, all the function pointers are setup and then the + * bus-generic code called. The generic driver will then use the + * provided pointers for uploading firmware (i2400ms_bus_bm*() in + * sdio-fw.c) and then setting up the device (i2400ms_dev_*() in + * sdio.c). + * + * Once firmware is uploaded, TX functions (sdio-tx.c) are called when + * data is ready for transmission in the TX fifo; then the SDIO IRQ is + * fired and data is available (sdio-rx.c), it is sent to the generic + * driver for processing with i2400m_rx. + */ + +#ifndef __I2400M_SDIO_H__ +#define __I2400M_SDIO_H__ + +#include "i2400m.h" + +/* Host-Device interface for SDIO */ +enum { + I2400MS_BLK_SIZE = 256, + I2400MS_PL_SIZE_MAX = 0x3E00, + + I2400MS_DATA_ADDR = 0x0, + I2400MS_INTR_STATUS_ADDR = 0x13, + I2400MS_INTR_CLEAR_ADDR = 0x13, + I2400MS_INTR_ENABLE_ADDR = 0x14, + I2400MS_INTR_GET_SIZE_ADDR = 0x2C, + /* The number of ticks to wait for the device to signal that + * it is ready */ + I2400MS_INIT_SLEEP_INTERVAL = 10, +}; + + +/** + * struct i2400ms - descriptor for a SDIO connected i2400m + * + * @i2400m: bus-generic i2400m implementation; has to be first (see + * it's documentation in i2400m.h). + * + * @func: pointer to our SDIO function + * + * @tx_worker: workqueue struct used to TX data when the bus-generic + * code signals packets are pending for transmission to the device. + * + * @tx_workqueue: workqeueue used for data TX; we don't use the + * system's workqueue as that might cause deadlocks with code in + * the bus-generic driver. + */ +struct i2400ms { + struct i2400m i2400m; /* FIRST! See doc */ + struct sdio_func *func; + + struct work_struct tx_worker; + struct workqueue_struct *tx_workqueue; + char tx_wq_name[32]; + + struct dentry *debugfs_dentry; +}; + + +static inline +void i2400ms_init(struct i2400ms *i2400ms) +{ + i2400m_init(&i2400ms->i2400m); +} + + +extern int i2400ms_rx_setup(struct i2400ms *); +extern void i2400ms_rx_release(struct i2400ms *); +extern ssize_t __i2400ms_rx_get_size(struct i2400ms *); + +extern int i2400ms_tx_setup(struct i2400ms *); +extern void i2400ms_tx_release(struct i2400ms *); +extern void i2400ms_bus_tx_kick(struct i2400m *); + +extern ssize_t i2400ms_bus_bm_cmd_send(struct i2400m *, + const struct i2400m_bootrom_header *, + size_t, int); +extern ssize_t i2400ms_bus_bm_wait_for_ack(struct i2400m *, + struct i2400m_bootrom_header *, + size_t); +#endif /* #ifndef __I2400M_SDIO_H__ */ diff --git a/drivers/net/wimax/i2400m/sdio-debug-levels.h b/drivers/net/wimax/i2400m/sdio-debug-levels.h new file mode 100644 index 00000000000..c5199874130 --- /dev/null +++ b/drivers/net/wimax/i2400m/sdio-debug-levels.h @@ -0,0 +1,22 @@ +/* + * debug levels control file for the i2400m module's + */ +#ifndef __debug_levels__h__ +#define __debug_levels__h__ + +/* Maximum compile and run time debug level for all submodules */ +#define D_MODULENAME i2400m_sdio +#define D_MASTER CONFIG_WIMAX_I2400M_DEBUG_LEVEL + +#include + +/* List of all the enabled modules */ +enum d_module { + D_SUBMODULE_DECLARE(main), + D_SUBMODULE_DECLARE(tx), + D_SUBMODULE_DECLARE(rx), + D_SUBMODULE_DECLARE(fw) +}; + + +#endif /* #ifndef __debug_levels__h__ */ From a0848826bfaf0815ec9654d78c218a40f755ccd4 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:55 -0800 Subject: [PATCH 118/123] i2400m/SDIO: probe/disconnect, dev init/shutdown and reset backends Implements probe/disconnect for the SDIO device, as well as main backends for the generic driver to control the SDIO device (bus_dev_start(), bus_dev_stop() and bus_reset()). Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/sdio.c | 511 ++++++++++++++++++++++++++++++++ 1 file changed, 511 insertions(+) create mode 100644 drivers/net/wimax/i2400m/sdio.c diff --git a/drivers/net/wimax/i2400m/sdio.c b/drivers/net/wimax/i2400m/sdio.c new file mode 100644 index 00000000000..1bfa283bbd8 --- /dev/null +++ b/drivers/net/wimax/i2400m/sdio.c @@ -0,0 +1,511 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Linux driver model glue for the SDIO device, reset & fw upload + * + * + * Copyright (C) 2007-2008 Intel Corporation + * Dirk Brandewie + * Inaky Perez-Gonzalez + * Yanir Lubetkin + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA. + * + * + * See i2400m-sdio.h for a general description of this driver. + * + * This file implements driver model glue, and hook ups for the + * generic driver to implement the bus-specific functions (device + * communication setup/tear down, firmware upload and resetting). + * + * ROADMAP + * + * i2400m_probe() + * alloc_netdev() + * i2400ms_netdev_setup() + * i2400ms_init() + * i2400m_netdev_setup() + * i2400ms_enable_function() + * i2400m_setup() + * + * i2400m_remove() + * i2400m_release() + * free_netdev(net_dev) + * + * i2400ms_bus_reset() Called by i2400m->bus_reset + * __i2400ms_reset() + * __i2400ms_send_barker() + * + * i2400ms_bus_dev_start() Called by i2400m_dev_start() [who is + * i2400ms_tx_setup() called by i2400m_setup()] + * i2400ms_rx_setup() + * + * i2400ms_bus_dev_stop() Called by i2400m_dev_stop() [who is + * i2400ms_rx_release() is called by i2400m_release()] + * i2400ms_tx_release() + * + */ + +#include +#include +#include +#include "i2400m-sdio.h" +#include + +#define D_SUBMODULE main +#include "sdio-debug-levels.h" + +/* IOE WiMAX function timeout in seconds */ +static int ioe_timeout = 2; +module_param(ioe_timeout, int, 0); + +/* Our firmware file name */ +#define I2400MS_FW_FILE_NAME "i2400m-fw-sdio-" I2400M_FW_VERSION ".sbcf" + +/* + * Enable the SDIO function + * + * Tries to enable the SDIO function; might fail if it is still not + * ready (in some hardware, the SDIO WiMAX function is only enabled + * when we ask it to explicitly doing). Tries until a timeout is + * reached. + * + * The reverse of this is...sdio_disable_function() + * + * Returns: 0 if the SDIO function was enabled, < 0 errno code on + * error (-ENODEV when it was unable to enable the function). + */ +static +int i2400ms_enable_function(struct sdio_func *func) +{ + u64 timeout; + int err; + struct device *dev = &func->dev; + + d_fnstart(3, dev, "(func %p)\n", func); + /* Setup timeout (FIXME: This needs to read the CIS table to + * get a real timeout) and then wait for the device to signal + * it is ready */ + timeout = get_jiffies_64() + ioe_timeout * HZ; + err = -ENODEV; + while (err != 0 && time_before64(get_jiffies_64(), timeout)) { + sdio_claim_host(func); + err = sdio_enable_func(func); + if (0 == err) { + sdio_release_host(func); + d_printf(2, dev, "SDIO function enabled\n"); + goto function_enabled; + } + d_printf(2, dev, "SDIO function failed to enable: %d\n", err); + sdio_disable_func(func); + sdio_release_host(func); + msleep(I2400MS_INIT_SLEEP_INTERVAL); + } + /* If timed out, device is not there yet -- get -ENODEV so + * the device driver core will retry later on. */ + if (err == -ETIME) { + dev_err(dev, "Can't enable WiMAX function; " + " has the function been enabled?\n"); + err = -ENODEV; + } +function_enabled: + d_fnend(3, dev, "(func %p) = %d\n", func, err); + return err; +} + + +/* + * Setup driver resources needed to communicate with the device + * + * The fw needs some time to settle, and it was just uploaded, + * so give it a break first. I'd prefer to just wait for the device to + * send something, but seems the poking we do to enable SDIO stuff + * interferes with it, so just give it a break before starting... + */ +static +int i2400ms_bus_dev_start(struct i2400m *i2400m) +{ + int result; + struct i2400ms *i2400ms = container_of(i2400m, struct i2400ms, i2400m); + struct sdio_func *func = i2400ms->func; + struct device *dev = &func->dev; + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + msleep(200); + result = i2400ms_rx_setup(i2400ms); + if (result < 0) + goto error_rx_setup; + result = i2400ms_tx_setup(i2400ms); + if (result < 0) + goto error_tx_setup; + d_fnend(3, dev, "(i2400m %p) = %d\n", i2400m, result); + return result; + + i2400ms_tx_release(i2400ms); +error_tx_setup: + i2400ms_rx_release(i2400ms); +error_rx_setup: + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); + return result; +} + + +static +void i2400ms_bus_dev_stop(struct i2400m *i2400m) +{ + struct i2400ms *i2400ms = container_of(i2400m, struct i2400ms, i2400m); + struct sdio_func *func = i2400ms->func; + struct device *dev = &func->dev; + + d_fnstart(3, dev, "(i2400m %p)\n", i2400m); + i2400ms_rx_release(i2400ms); + i2400ms_tx_release(i2400ms); + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); +} + + +/* + * Sends a barker buffer to the device + * + * This helper will allocate a kmalloced buffer and use it to transmit + * (then free it). Reason for this is that the SDIO host controller + * expects alignment (unknown exactly which) which the stack won't + * really provide and certain arches/host-controller combinations + * cannot use stack/vmalloc/text areas for DMA transfers. + */ +static +int __i2400ms_send_barker(struct i2400ms *i2400ms, + const __le32 *barker, size_t barker_size) +{ + int ret; + struct sdio_func *func = i2400ms->func; + struct device *dev = &func->dev; + void *buffer; + + ret = -ENOMEM; + buffer = kmalloc(I2400MS_BLK_SIZE, GFP_KERNEL); + if (buffer == NULL) + goto error_kzalloc; + + memcpy(buffer, barker, barker_size); + sdio_claim_host(func); + ret = sdio_memcpy_toio(func, 0, buffer, I2400MS_BLK_SIZE); + sdio_release_host(func); + + if (ret < 0) + d_printf(0, dev, "E: barker error: %d\n", ret); + + kfree(buffer); +error_kzalloc: + return ret; +} + + +/* + * Reset a device at different levels (warm, cold or bus) + * + * @i2400ms: device descriptor + * @reset_type: soft, warm or bus reset (I2400M_RT_WARM/SOFT/BUS) + * + * FIXME: not tested -- need to confirm expected effects + * + * Warm and cold resets get an SDIO reset if they fail (unimplemented) + * + * Warm reset: + * + * The device will be fully reset internally, but won't be + * disconnected from the USB bus (so no reenumeration will + * happen). Firmware upload will be neccessary. + * + * The device will send a reboot barker in the notification endpoint + * that will trigger the driver to reinitialize the state + * automatically from notif.c:i2400m_notification_grok() into + * i2400m_dev_bootstrap_delayed(). + * + * Cold and bus (USB) reset: + * + * The device will be fully reset internally, disconnected from the + * USB bus an a reenumeration will happen. Firmware upload will be + * neccessary. Thus, we don't do any locking or struct + * reinitialization, as we are going to be fully disconnected and + * reenumerated. + * + * Note we need to return -ENODEV if a warm reset was requested and we + * had to resort to a bus reset. See i2400m_op_reset(), wimax_reset() + * and wimax_dev->op_reset. + * + * WARNING: no driver state saved/fixed + */ +static +int i2400ms_bus_reset(struct i2400m *i2400m, enum i2400m_reset_type rt) +{ + int result; + struct i2400ms *i2400ms = + container_of(i2400m, struct i2400ms, i2400m); + struct device *dev = i2400m_dev(i2400m); + static const __le32 i2400m_WARM_BOOT_BARKER[4] = { + __constant_cpu_to_le32(I2400M_WARM_RESET_BARKER), + __constant_cpu_to_le32(I2400M_WARM_RESET_BARKER), + __constant_cpu_to_le32(I2400M_WARM_RESET_BARKER), + __constant_cpu_to_le32(I2400M_WARM_RESET_BARKER), + }; + static const __le32 i2400m_COLD_BOOT_BARKER[4] = { + __constant_cpu_to_le32(I2400M_COLD_RESET_BARKER), + __constant_cpu_to_le32(I2400M_COLD_RESET_BARKER), + __constant_cpu_to_le32(I2400M_COLD_RESET_BARKER), + __constant_cpu_to_le32(I2400M_COLD_RESET_BARKER), + }; + + if (rt == I2400M_RT_WARM) + result = __i2400ms_send_barker(i2400ms, i2400m_WARM_BOOT_BARKER, + sizeof(i2400m_WARM_BOOT_BARKER)); + else if (rt == I2400M_RT_COLD) + result = __i2400ms_send_barker(i2400ms, i2400m_COLD_BOOT_BARKER, + sizeof(i2400m_COLD_BOOT_BARKER)); + else if (rt == I2400M_RT_BUS) { +do_bus_reset: + dev_err(dev, "FIXME: SDIO bus reset not implemented\n"); + result = rt == I2400M_RT_WARM ? -ENODEV : -ENOSYS; + } else + BUG(); + if (result < 0 && rt != I2400M_RT_BUS) { + dev_err(dev, "%s reset failed (%d); trying SDIO reset\n", + rt == I2400M_RT_WARM ? "warm" : "cold", result); + rt = I2400M_RT_BUS; + goto do_bus_reset; + } + return result; +} + + +static +void i2400ms_netdev_setup(struct net_device *net_dev) +{ + struct i2400m *i2400m = net_dev_to_i2400m(net_dev); + struct i2400ms *i2400ms = container_of(i2400m, struct i2400ms, i2400m); + i2400ms_init(i2400ms); + i2400m_netdev_setup(net_dev); +} + + +/* + * Debug levels control; see debug.h + */ +struct d_level D_LEVEL[] = { + D_SUBMODULE_DEFINE(main), + D_SUBMODULE_DEFINE(tx), + D_SUBMODULE_DEFINE(rx), + D_SUBMODULE_DEFINE(fw), +}; +size_t D_LEVEL_SIZE = ARRAY_SIZE(D_LEVEL); + + +#define __debugfs_register(prefix, name, parent) \ +do { \ + result = d_level_register_debugfs(prefix, name, parent); \ + if (result < 0) \ + goto error; \ +} while (0) + + +static +int i2400ms_debugfs_add(struct i2400ms *i2400ms) +{ + int result; + struct dentry *dentry = i2400ms->i2400m.wimax_dev.debugfs_dentry; + + dentry = debugfs_create_dir("i2400m-usb", dentry); + result = PTR_ERR(dentry); + if (IS_ERR(dentry)) { + if (result == -ENODEV) + result = 0; /* No debugfs support */ + goto error; + } + i2400ms->debugfs_dentry = dentry; + __debugfs_register("dl_", main, dentry); + __debugfs_register("dl_", tx, dentry); + __debugfs_register("dl_", rx, dentry); + __debugfs_register("dl_", fw, dentry); + + return 0; + +error: + debugfs_remove_recursive(i2400ms->debugfs_dentry); + return result; +} + + +/* + * Probe a i2400m interface and register it + * + * @func: SDIO function + * @id: SDIO device ID + * @returns: 0 if ok, < 0 errno code on error. + * + * Alloc a net device, initialize the bus-specific details and then + * calls the bus-generic initialization routine. That will register + * the wimax and netdev devices, upload the firmware [using + * _bus_bm_*()], call _bus_dev_start() to finalize the setup of the + * communication with the device and then will start to talk to it to + * finnish setting it up. + * + * Initialization is tricky; some instances of the hw are packed with + * others in a way that requires a third driver that enables the WiMAX + * function. In those cases, we can't enable the SDIO function and + * we'll return with -ENODEV. When the driver that enables the WiMAX + * function does its thing, it has to do a bus_rescan_devices() on the + * SDIO bus so this driver is called again to enumerate the WiMAX + * function. + */ +static +int i2400ms_probe(struct sdio_func *func, + const struct sdio_device_id *id) +{ + int result; + struct net_device *net_dev; + struct device *dev = &func->dev; + struct i2400m *i2400m; + struct i2400ms *i2400ms; + + /* Allocate instance [calls i2400m_netdev_setup() on it]. */ + result = -ENOMEM; + net_dev = alloc_netdev(sizeof(*i2400ms), "wmx%d", + i2400ms_netdev_setup); + if (net_dev == NULL) { + dev_err(dev, "no memory for network device instance\n"); + goto error_alloc_netdev; + } + SET_NETDEV_DEV(net_dev, dev); + i2400m = net_dev_to_i2400m(net_dev); + i2400ms = container_of(i2400m, struct i2400ms, i2400m); + i2400m->wimax_dev.net_dev = net_dev; + i2400ms->func = func; + sdio_set_drvdata(func, i2400ms); + + i2400m->bus_tx_block_size = I2400MS_BLK_SIZE; + i2400m->bus_pl_size_max = I2400MS_PL_SIZE_MAX; + i2400m->bus_dev_start = i2400ms_bus_dev_start; + i2400m->bus_dev_stop = i2400ms_bus_dev_stop; + i2400m->bus_tx_kick = i2400ms_bus_tx_kick; + i2400m->bus_reset = i2400ms_bus_reset; + i2400m->bus_bm_cmd_send = i2400ms_bus_bm_cmd_send; + i2400m->bus_bm_wait_for_ack = i2400ms_bus_bm_wait_for_ack; + i2400m->bus_fw_name = I2400MS_FW_FILE_NAME; + i2400m->bus_bm_mac_addr_impaired = 1; + + result = i2400ms_enable_function(i2400ms->func); + if (result < 0) { + dev_err(dev, "Cannot enable SDIO function: %d\n", result); + goto error_func_enable; + } + + sdio_claim_host(func); + result = sdio_set_block_size(func, I2400MS_BLK_SIZE); + if (result < 0) { + dev_err(dev, "Failed to set block size: %d\n", result); + goto error_set_blk_size; + } + sdio_release_host(func); + + result = i2400m_setup(i2400m, I2400M_BRI_NO_REBOOT); + if (result < 0) { + dev_err(dev, "cannot setup device: %d\n", result); + goto error_setup; + } + + result = i2400ms_debugfs_add(i2400ms); + if (result < 0) { + dev_err(dev, "cannot create SDIO debugfs: %d\n", + result); + goto error_debugfs_add; + } + return 0; + +error_debugfs_add: + i2400m_release(i2400m); +error_setup: + sdio_set_drvdata(func, NULL); + sdio_claim_host(func); +error_set_blk_size: + sdio_disable_func(func); + sdio_release_host(func); +error_func_enable: + free_netdev(net_dev); +error_alloc_netdev: + return result; +} + + +static +void i2400ms_remove(struct sdio_func *func) +{ + struct device *dev = &func->dev; + struct i2400ms *i2400ms = sdio_get_drvdata(func); + struct i2400m *i2400m = &i2400ms->i2400m; + struct net_device *net_dev = i2400m->wimax_dev.net_dev; + + d_fnstart(3, dev, "SDIO func %p\n", func); + debugfs_remove_recursive(i2400ms->debugfs_dentry); + i2400m_release(i2400m); + sdio_set_drvdata(func, NULL); + sdio_claim_host(func); + sdio_disable_func(func); + sdio_release_host(func); + free_netdev(net_dev); + d_fnend(3, dev, "SDIO func %p\n", func); +} + +enum { + I2400MS_INTEL_VID = 0x89, +}; + +static +const struct sdio_device_id i2400ms_sdio_ids[] = { + /* Intel: i2400m WiMAX over SDIO */ + { SDIO_DEVICE(I2400MS_INTEL_VID, 0x1402) }, + { }, /* end: all zeroes */ +}; +MODULE_DEVICE_TABLE(sdio, i2400ms_sdio_ids); + + +static +struct sdio_driver i2400m_sdio_driver = { + .name = KBUILD_MODNAME, + .probe = i2400ms_probe, + .remove = i2400ms_remove, + .id_table = i2400ms_sdio_ids, +}; + + +static +int __init i2400ms_driver_init(void) +{ + return sdio_register_driver(&i2400m_sdio_driver); +} +module_init(i2400ms_driver_init); + + +static +void __exit i2400ms_driver_exit(void) +{ + flush_scheduled_work(); /* for the stuff we schedule */ + sdio_unregister_driver(&i2400m_sdio_driver); +} +module_exit(i2400ms_driver_exit); + + +MODULE_AUTHOR("Intel Corporation "); +MODULE_DESCRIPTION("Intel 2400M WiMAX networking for SDIO"); +MODULE_LICENSE("GPL"); +MODULE_FIRMWARE(I2400MS_FW_FILE_NAME); From 020bb6f30b636d563d4268116107d592550ddd0c Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:56 -0800 Subject: [PATCH 119/123] i2400m/SDIO: firmware upload backend This implements the backends for the generic driver (i2400m) to be able to load firmware to the SDIO device. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/sdio-fw.c | 224 +++++++++++++++++++++++++++++ 1 file changed, 224 insertions(+) create mode 100644 drivers/net/wimax/i2400m/sdio-fw.c diff --git a/drivers/net/wimax/i2400m/sdio-fw.c b/drivers/net/wimax/i2400m/sdio-fw.c new file mode 100644 index 00000000000..3487205d8f5 --- /dev/null +++ b/drivers/net/wimax/i2400m/sdio-fw.c @@ -0,0 +1,224 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * Firmware uploader's SDIO specifics + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Yanir Lubetkin + * Inaky Perez-Gonzalez + * - Initial implementation + * + * Inaky Perez-Gonzalez + * - Bus generic/specific split for USB + * + * Dirk Brandewie + * - Initial implementation for SDIO + * + * Inaky Perez-Gonzalez + * - SDIO rehash for changes in the bus-driver model + * + * THE PROCEDURE + * + * See fw.c for the generic description of this procedure. + * + * This file implements only the SDIO specifics. It boils down to how + * to send a command and waiting for an acknowledgement from the + * device. We do polled reads. + * + * COMMAND EXECUTION + * + * THe generic firmware upload code will call i2400m_bus_bm_cmd_send() + * to send commands. + * + * The SDIO devices expects things in 256 byte blocks, so it will pad + * it, compute the checksum (if needed) and pass it to SDIO. + * + * ACK RECEPTION + * + * This works in polling mode -- the fw loader says when to wait for + * data and for that it calls i2400ms_bus_bm_wait_for_ack(). + * + * This will poll the device for data until it is received. We need to + * receive at least as much bytes as where asked for (although it'll + * always be a multiple of 256 bytes). + */ +#include +#include "i2400m-sdio.h" + + +#define D_SUBMODULE fw +#include "sdio-debug-levels.h" + +/* + * Send a boot-mode command to the SDIO function + * + * We use a bounce buffer (i2400m->bm_cmd_buf) because we need to + * touch the header if the RAW flag is not set. + * + * @flags: pass thru from i2400m_bm_cmd() + * @return: cmd_size if ok, < 0 errno code on error. + * + * Note the command is padded to the SDIO block size for the device. + */ +ssize_t i2400ms_bus_bm_cmd_send(struct i2400m *i2400m, + const struct i2400m_bootrom_header *_cmd, + size_t cmd_size, int flags) +{ + ssize_t result; + struct device *dev = i2400m_dev(i2400m); + struct i2400ms *i2400ms = container_of(i2400m, struct i2400ms, i2400m); + int opcode = _cmd == NULL ? -1 : i2400m_brh_get_opcode(_cmd); + struct i2400m_bootrom_header *cmd; + /* SDIO restriction */ + size_t cmd_size_a = ALIGN(cmd_size, I2400MS_BLK_SIZE); + + d_fnstart(5, dev, "(i2400m %p cmd %p size %zu)\n", + i2400m, _cmd, cmd_size); + result = -E2BIG; + if (cmd_size > I2400M_BM_CMD_BUF_SIZE) + goto error_too_big; + + memcpy(i2400m->bm_cmd_buf, _cmd, cmd_size); /* Prep command */ + cmd = i2400m->bm_cmd_buf; + if (cmd_size_a > cmd_size) /* Zero pad space */ + memset(i2400m->bm_cmd_buf + cmd_size, 0, cmd_size_a - cmd_size); + if ((flags & I2400M_BM_CMD_RAW) == 0) { + if (WARN_ON(i2400m_brh_get_response_required(cmd) == 0)) + dev_warn(dev, "SW BUG: response_required == 0\n"); + i2400m_bm_cmd_prepare(cmd); + } + d_printf(4, dev, "BM cmd %d: %zu bytes (%zu padded)\n", + opcode, cmd_size, cmd_size_a); + d_dump(5, dev, cmd, cmd_size); + + sdio_claim_host(i2400ms->func); /* Send & check */ + result = sdio_memcpy_toio(i2400ms->func, I2400MS_DATA_ADDR, + i2400m->bm_cmd_buf, cmd_size_a); + sdio_release_host(i2400ms->func); + if (result < 0) { + dev_err(dev, "BM cmd %d: cannot send: %ld\n", + opcode, (long) result); + goto error_cmd_send; + } + result = cmd_size; +error_cmd_send: +error_too_big: + d_fnend(5, dev, "(i2400m %p cmd %p size %zu) = %d\n", + i2400m, _cmd, cmd_size, (int) result); + return result; +} + + +/* + * Read an ack from the device's boot-mode (polling) + * + * @i2400m: + * @_ack: pointer to where to store the read data + * @ack_size: how many bytes we should read + * + * Returns: < 0 errno code on error; otherwise, amount of received bytes. + * + * The ACK for a BM command is always at least sizeof(*ack) bytes, so + * check for that. We don't need to check for device reboots + * + * NOTE: We do an artificial timeout of 1 sec over the SDIO timeout; + * this way we have control over it...there is no way that I know + * of setting an SDIO transaction timeout. + */ +ssize_t i2400ms_bus_bm_wait_for_ack(struct i2400m *i2400m, + struct i2400m_bootrom_header *ack, + size_t ack_size) +{ + int result; + ssize_t rx_size; + u64 timeout; + struct i2400ms *i2400ms = container_of(i2400m, struct i2400ms, i2400m); + struct sdio_func *func = i2400ms->func; + struct device *dev = &func->dev; + + BUG_ON(sizeof(*ack) > ack_size); + + d_fnstart(5, dev, "(i2400m %p ack %p size %zu)\n", + i2400m, ack, ack_size); + + timeout = get_jiffies_64() + 2 * HZ; + sdio_claim_host(func); + while (1) { + if (time_after64(get_jiffies_64(), timeout)) { + rx_size = -ETIMEDOUT; + dev_err(dev, "timeout waiting for ack data\n"); + goto error_timedout; + } + + /* Find the RX size, check if it fits or not -- it if + * doesn't fit, fail, as we have no way to dispose of + * the extra data. */ + rx_size = __i2400ms_rx_get_size(i2400ms); + if (rx_size < 0) + goto error_rx_get_size; + result = -ENOSPC; /* Check it fits */ + if (rx_size < sizeof(*ack)) { + rx_size = -EIO; + dev_err(dev, "HW BUG? received is too small (%zu vs " + "%zu needed)\n", sizeof(*ack), rx_size); + goto error_too_small; + } + if (rx_size > I2400M_BM_ACK_BUF_SIZE) { + dev_err(dev, "SW BUG? BM_ACK_BUF is too small (%u vs " + "%zu needed)\n", I2400M_BM_ACK_BUF_SIZE, + rx_size); + goto error_too_small; + } + + /* Read it */ + result = sdio_memcpy_fromio(func, i2400m->bm_ack_buf, + I2400MS_DATA_ADDR, rx_size); + if (result == -ETIMEDOUT || result == -ETIME) + continue; + if (result < 0) { + dev_err(dev, "BM SDIO receive (%zu B) failed: %d\n", + rx_size, result); + goto error_read; + } else + break; + } + rx_size = min((ssize_t)ack_size, rx_size); + memcpy(ack, i2400m->bm_ack_buf, rx_size); +error_read: +error_too_small: +error_rx_get_size: +error_timedout: + sdio_release_host(func); + d_fnend(5, dev, "(i2400m %p ack %p size %zu) = %ld\n", + i2400m, ack, ack_size, (long) rx_size); + return rx_size; +} From 514ec71f7289c942f801bdbd309428c470bfc071 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:57 -0800 Subject: [PATCH 120/123] i2400m/SDIO: TX and RX path backends Implements the backend so that the generic driver can TX/RX to/from the SDIO device. For RX, when data is ready the SDIO IRQ is fired and that will allocate an skb, put all the data there and then pass it to the generic driver RX code for processing and delivery. TX, when kicked by the generic driver, will schedule work on a driver-specific workqueue that pulls data from the TX FIFO and sends it to the device until it drains it. Thread contexts are needed as SDIO functions are blocking. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/wimax/i2400m/sdio-rx.c | 255 +++++++++++++++++++++++++++++ drivers/net/wimax/i2400m/sdio-tx.c | 153 +++++++++++++++++ 2 files changed, 408 insertions(+) create mode 100644 drivers/net/wimax/i2400m/sdio-rx.c create mode 100644 drivers/net/wimax/i2400m/sdio-tx.c diff --git a/drivers/net/wimax/i2400m/sdio-rx.c b/drivers/net/wimax/i2400m/sdio-rx.c new file mode 100644 index 00000000000..a3008b904f7 --- /dev/null +++ b/drivers/net/wimax/i2400m/sdio-rx.c @@ -0,0 +1,255 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * SDIO RX handling + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Dirk Brandewie + * - Initial implementation + * + * + * This handles the RX path on SDIO. + * + * The SDIO bus driver calls the "irq" routine when data is available. + * This is not a traditional interrupt routine since the SDIO bus + * driver calls us from its irq thread context. Because of this + * sleeping in the SDIO RX IRQ routine is okay. + * + * From there on, we obtain the size of the data that is available, + * allocate an skb, copy it and then pass it to the generic driver's + * RX routine [i2400m_rx()]. + * + * ROADMAP + * + * i2400ms_irq() + * i2400ms_rx() + * __i2400ms_rx_get_size() + * i2400m_rx() + * + * i2400ms_rx_setup() + * + * i2400ms_rx_release() + */ +#include +#include +#include +#include +#include +#include "i2400m-sdio.h" + +#define D_SUBMODULE rx +#include "sdio-debug-levels.h" + + +/* + * Read and return the amount of bytes available for RX + * + * The RX size has to be read like this: byte reads of three + * sequential locations; then glue'em together. + * + * sdio_readl() doesn't work. + */ +ssize_t __i2400ms_rx_get_size(struct i2400ms *i2400ms) +{ + int ret, cnt, val; + ssize_t rx_size; + unsigned xfer_size_addr; + struct sdio_func *func = i2400ms->func; + struct device *dev = &i2400ms->func->dev; + + d_fnstart(7, dev, "(i2400ms %p)\n", i2400ms); + xfer_size_addr = I2400MS_INTR_GET_SIZE_ADDR; + rx_size = 0; + for (cnt = 0; cnt < 3; cnt++) { + val = sdio_readb(func, xfer_size_addr + cnt, &ret); + if (ret < 0) { + dev_err(dev, "RX: Can't read byte %d of RX size from " + "0x%08x: %d\n", cnt, xfer_size_addr + cnt, ret); + rx_size = ret; + goto error_read; + } + rx_size = rx_size << 8 | (val & 0xff); + } + d_printf(6, dev, "RX: rx_size is %ld\n", (long) rx_size); +error_read: + d_fnend(7, dev, "(i2400ms %p) = %ld\n", i2400ms, (long) rx_size); + return rx_size; +} + + +/* + * Read data from the device (when in normal) + * + * Allocate an SKB of the right size, read the data in and then + * deliver it to the generic layer. + * + * We also check for a reboot barker. That means the device died and + * we have to reboot it. + */ +static +void i2400ms_rx(struct i2400ms *i2400ms) +{ + int ret; + struct sdio_func *func = i2400ms->func; + struct device *dev = &func->dev; + struct i2400m *i2400m = &i2400ms->i2400m; + struct sk_buff *skb; + ssize_t rx_size; + + d_fnstart(7, dev, "(i2400ms %p)\n", i2400ms); + rx_size = __i2400ms_rx_get_size(i2400ms); + if (rx_size < 0) { + ret = rx_size; + goto error_get_size; + } + ret = -ENOMEM; + skb = alloc_skb(rx_size, GFP_ATOMIC); + if (NULL == skb) { + dev_err(dev, "RX: unable to alloc skb\n"); + goto error_alloc_skb; + } + + ret = sdio_memcpy_fromio(func, skb->data, + I2400MS_DATA_ADDR, rx_size); + if (ret < 0) { + dev_err(dev, "RX: SDIO data read failed: %d\n", ret); + goto error_memcpy_fromio; + } + /* Check if device has reset */ + if (!memcmp(skb->data, i2400m_NBOOT_BARKER, + sizeof(i2400m_NBOOT_BARKER)) + || !memcmp(skb->data, i2400m_SBOOT_BARKER, + sizeof(i2400m_SBOOT_BARKER))) { + ret = i2400m_dev_reset_handle(i2400m); + kfree_skb(skb); + } else { + skb_put(skb, rx_size); + i2400m_rx(i2400m, skb); + } + d_fnend(7, dev, "(i2400ms %p) = void\n", i2400ms); + return; + +error_memcpy_fromio: + kfree_skb(skb); +error_alloc_skb: +error_get_size: + d_fnend(7, dev, "(i2400ms %p) = %d\n", i2400ms, ret); + return; +} + + +/* + * Process an interrupt from the SDIO card + * + * FIXME: need to process other events that are not just ready-to-read + * + * Checks there is data ready and then proceeds to read it. + */ +static +void i2400ms_irq(struct sdio_func *func) +{ + int ret; + struct i2400ms *i2400ms = sdio_get_drvdata(func); + struct i2400m *i2400m = &i2400ms->i2400m; + struct device *dev = &func->dev; + int val; + + d_fnstart(6, dev, "(i2400ms %p)\n", i2400ms); + val = sdio_readb(func, I2400MS_INTR_STATUS_ADDR, &ret); + if (ret < 0) { + dev_err(dev, "RX: Can't read interrupt status: %d\n", ret); + goto error_no_irq; + } + if (!val) { + dev_err(dev, "RX: BUG? got IRQ but no interrupt ready?\n"); + goto error_no_irq; + } + sdio_writeb(func, 1, I2400MS_INTR_CLEAR_ADDR, &ret); + if (WARN_ON(i2400m->boot_mode != 0)) + dev_err(dev, "RX: SW BUG? boot mode and IRQ is up?\n"); + else + i2400ms_rx(i2400ms); +error_no_irq: + d_fnend(6, dev, "(i2400ms %p) = void\n", i2400ms); + return; +} + + +/* + * Setup SDIO RX + * + * Hooks up the IRQ handler and then enables IRQs. + */ +int i2400ms_rx_setup(struct i2400ms *i2400ms) +{ + int result; + struct sdio_func *func = i2400ms->func; + struct device *dev = &func->dev; + + d_fnstart(5, dev, "(i2400ms %p)\n", i2400ms); + sdio_claim_host(func); + result = sdio_claim_irq(func, i2400ms_irq); + if (result < 0) { + dev_err(dev, "Cannot claim IRQ: %d\n", result); + goto error_irq_claim; + } + result = 0; + sdio_writeb(func, 1, I2400MS_INTR_ENABLE_ADDR, &result); + if (result < 0) { + sdio_release_irq(func); + dev_err(dev, "Failed to enable interrupts %d\n", result); + } +error_irq_claim: + sdio_release_host(func); + d_fnend(5, dev, "(i2400ms %p) = %d\n", i2400ms, result); + return result; +} + + +/* + * Tear down SDIO RX + * + * Disables IRQs in the device and removes the IRQ handler. + */ +void i2400ms_rx_release(struct i2400ms *i2400ms) +{ + int result; + struct sdio_func *func = i2400ms->func; + struct device *dev = &func->dev; + + d_fnstart(5, dev, "(i2400ms %p)\n", i2400ms); + sdio_claim_host(func); + sdio_writeb(func, 0, I2400MS_INTR_ENABLE_ADDR, &result); + sdio_release_irq(func); + sdio_release_host(func); + d_fnend(5, dev, "(i2400ms %p) = %d\n", i2400ms, result); +} diff --git a/drivers/net/wimax/i2400m/sdio-tx.c b/drivers/net/wimax/i2400m/sdio-tx.c new file mode 100644 index 00000000000..5105a5ebc44 --- /dev/null +++ b/drivers/net/wimax/i2400m/sdio-tx.c @@ -0,0 +1,153 @@ +/* + * Intel Wireless WiMAX Connection 2400m + * SDIO TX transaction backends + * + * + * Copyright (C) 2007-2008 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * + * Intel Corporation + * Dirk Brandewie + * - Initial implementation + * + * + * Takes the TX messages in the i2400m's driver TX FIFO and sends them + * to the device until there are no more. + * + * If we fail sending the message, we just drop it. There isn't much + * we can do at this point. Most of the traffic is network, which has + * recovery methods for dropped packets. + * + * The SDIO functions are not atomic, so we can't run from the context + * where i2400m->bus_tx_kick() [i2400ms_bus_tx_kick()] is being called + * (some times atomic). Thus, the actual TX work is deferred to a + * workqueue. + * + * ROADMAP + * + * i2400ms_bus_tx_kick() + * i2400ms_tx_submit() [through workqueue] + * + * i2400m_tx_setup() + * + * i2400m_tx_release() + */ +#include +#include "i2400m-sdio.h" + +#define D_SUBMODULE tx +#include "sdio-debug-levels.h" + + +/* + * Pull TX transations from the TX FIFO and send them to the device + * until there are no more. + */ +static +void i2400ms_tx_submit(struct work_struct *ws) +{ + int result; + struct i2400ms *i2400ms = container_of(ws, struct i2400ms, tx_worker); + struct i2400m *i2400m = &i2400ms->i2400m; + struct sdio_func *func = i2400ms->func; + struct device *dev = &func->dev; + struct i2400m_msg_hdr *tx_msg; + size_t tx_msg_size; + + d_fnstart(4, dev, "(i2400ms %p, i2400m %p)\n", i2400ms, i2400ms); + + while (NULL != (tx_msg = i2400m_tx_msg_get(i2400m, &tx_msg_size))) { + d_printf(2, dev, "TX: submitting %zu bytes\n", tx_msg_size); + d_dump(5, dev, tx_msg, tx_msg_size); + + sdio_claim_host(func); + result = sdio_memcpy_toio(func, 0, tx_msg, tx_msg_size); + sdio_release_host(func); + + i2400m_tx_msg_sent(i2400m); + + if (result < 0) { + dev_err(dev, "TX: cannot submit TX; tx_msg @%zu %zu B:" + " %d\n", (void *) tx_msg - i2400m->tx_buf, + tx_msg_size, result); + } + + d_printf(2, dev, "TX: %zub submitted\n", tx_msg_size); + } + + d_fnend(4, dev, "(i2400ms %p) = void\n", i2400ms); +} + + +/* + * The generic driver notifies us that there is data ready for TX + * + * Schedule a run of i2400ms_tx_submit() to handle it. + */ +void i2400ms_bus_tx_kick(struct i2400m *i2400m) +{ + struct i2400ms *i2400ms = container_of(i2400m, struct i2400ms, i2400m); + struct device *dev = &i2400ms->func->dev; + + d_fnstart(3, dev, "(i2400m %p) = void\n", i2400m); + + /* schedule tx work, this is because tx may block, therefore + * it has to run in a thread context. + */ + queue_work(i2400ms->tx_workqueue, &i2400ms->tx_worker); + + d_fnend(3, dev, "(i2400m %p) = void\n", i2400m); +} + +int i2400ms_tx_setup(struct i2400ms *i2400ms) +{ + int result; + struct device *dev = &i2400ms->func->dev; + struct i2400m *i2400m = &i2400ms->i2400m; + + d_fnstart(5, dev, "(i2400ms %p)\n", i2400ms); + + INIT_WORK(&i2400ms->tx_worker, i2400ms_tx_submit); + snprintf(i2400ms->tx_wq_name, sizeof(i2400ms->tx_wq_name), + "%s-tx", i2400m->wimax_dev.name); + i2400ms->tx_workqueue = + create_singlethread_workqueue(i2400ms->tx_wq_name); + if (NULL == i2400ms->tx_workqueue) { + dev_err(dev, "TX: failed to create workqueue\n"); + result = -ENOMEM; + } else + result = 0; + d_fnend(5, dev, "(i2400ms %p) = %d\n", i2400ms, result); + return result; +} + +void i2400ms_tx_release(struct i2400ms *i2400ms) +{ + destroy_workqueue(i2400ms->tx_workqueue); +} From 143ee2d5557c0598a97f3089eb29e8226e0e8cee Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Tue, 23 Dec 2008 16:18:48 -0800 Subject: [PATCH 121/123] i2400m: Makefile and Kconfig Integrate the i2400m driver into the kernel's build and Kconfig. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- drivers/net/Kconfig | 2 ++ drivers/net/Makefile | 1 + drivers/net/wimax/Kconfig | 17 +++++++++++ drivers/net/wimax/Makefile | 5 ++++ drivers/net/wimax/i2400m/Kconfig | 49 +++++++++++++++++++++++++++++++ drivers/net/wimax/i2400m/Makefile | 29 ++++++++++++++++++ 6 files changed, 103 insertions(+) create mode 100644 drivers/net/wimax/Kconfig create mode 100644 drivers/net/wimax/Makefile create mode 100644 drivers/net/wimax/i2400m/Kconfig create mode 100644 drivers/net/wimax/i2400m/Makefile diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig index 9a18270c108..97ea7c60e00 100644 --- a/drivers/net/Kconfig +++ b/drivers/net/Kconfig @@ -2614,6 +2614,8 @@ source "drivers/net/tokenring/Kconfig" source "drivers/net/wireless/Kconfig" +source "drivers/net/wimax/Kconfig" + source "drivers/net/usb/Kconfig" source "drivers/net/pcmcia/Kconfig" diff --git a/drivers/net/Makefile b/drivers/net/Makefile index e5c34b46421..a3c5c002f22 100644 --- a/drivers/net/Makefile +++ b/drivers/net/Makefile @@ -263,3 +263,4 @@ obj-$(CONFIG_NIU) += niu.o obj-$(CONFIG_VIRTIO_NET) += virtio_net.o obj-$(CONFIG_SFC) += sfc/ +obj-$(CONFIG_WIMAX) += wimax/ diff --git a/drivers/net/wimax/Kconfig b/drivers/net/wimax/Kconfig new file mode 100644 index 00000000000..565018ec1e3 --- /dev/null +++ b/drivers/net/wimax/Kconfig @@ -0,0 +1,17 @@ +# +# WiMAX LAN device drivers configuration +# + + +comment "Enable WiMAX (Networking options) to see the WiMAX drivers" + depends on WIMAX = n + +if WIMAX + +menu "WiMAX Wireless Broadband devices" + +source "drivers/net/wimax/i2400m/Kconfig" + +endmenu + +endif diff --git a/drivers/net/wimax/Makefile b/drivers/net/wimax/Makefile new file mode 100644 index 00000000000..992bc02bc01 --- /dev/null +++ b/drivers/net/wimax/Makefile @@ -0,0 +1,5 @@ + +obj-$(CONFIG_WIMAX_I2400M) += i2400m/ + +# (from Sam Ravnborg) force kbuild to create built-in.o +obj- := dummy.o diff --git a/drivers/net/wimax/i2400m/Kconfig b/drivers/net/wimax/i2400m/Kconfig new file mode 100644 index 00000000000..d623b3d99a4 --- /dev/null +++ b/drivers/net/wimax/i2400m/Kconfig @@ -0,0 +1,49 @@ + +config WIMAX_I2400M + tristate + depends on WIMAX + select FW_LOADER + +comment "Enable USB support to see WiMAX USB drivers" + depends on USB = n + +comment "Enable MMC support to see WiMAX SDIO drivers" + depends on MMC = n + +config WIMAX_I2400M_USB + tristate "Intel Wireless WiMAX Connection 2400 over USB (including 5x50)" + depends on WIMAX && USB + select WIMAX_I2400M + help + Select if you have a device based on the Intel WiMAX + Connection 2400 over USB (like any of the Intel Wireless + WiMAX/WiFi Link 5x50 series). + + If unsure, it is safe to select M (module). + +config WIMAX_I2400M_SDIO + tristate "Intel Wireless WiMAX Connection 2400 over SDIO" + depends on WIMAX && MMC + select WIMAX_I2400M + help + Select if you have a device based on the Intel WiMAX + Connection 2400 over SDIO. + + If unsure, it is safe to select M (module). + +config WIMAX_I2400M_DEBUG_LEVEL + int "WiMAX i2400m debug level" + depends on WIMAX_I2400M + default 8 + help + + Select the maximum debug verbosity level to be compiled into + the WiMAX i2400m driver code. + + By default, this is disabled at runtime and can be + selectively enabled at runtime for different parts of the + code using the sysfs debug-levels file. + + If set at zero, this will compile out all the debug code. + + It is recommended that it is left at 8. diff --git a/drivers/net/wimax/i2400m/Makefile b/drivers/net/wimax/i2400m/Makefile new file mode 100644 index 00000000000..1696e936cf5 --- /dev/null +++ b/drivers/net/wimax/i2400m/Makefile @@ -0,0 +1,29 @@ + +obj-$(CONFIG_WIMAX_I2400M) += i2400m.o +obj-$(CONFIG_WIMAX_I2400M_USB) += i2400m-usb.o +obj-$(CONFIG_WIMAX_I2400M_SDIO) += i2400m-sdio.o + +i2400m-y := \ + control.o \ + driver.o \ + fw.o \ + op-rfkill.o \ + netdev.o \ + tx.o \ + rx.o + +i2400m-$(CONFIG_DEBUG_FS) += debugfs.o + +i2400m-usb-y := \ + usb-fw.o \ + usb-notif.o \ + usb-tx.o \ + usb-rx.o \ + usb.o + + +i2400m-sdio-y := \ + sdio.o \ + sdio-tx.o \ + sdio-fw.o \ + sdio-rx.o From e30698743419d20dce03d033761f203b4d847ab0 Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:57:59 -0800 Subject: [PATCH 122/123] wimax: export linux/wimax.h and linux/wimax/i2400m.h with headers_install These two files are what user space can use to establish communication with the WiMAX kernel API and to speak the Intel 2400m Wireless WiMAX connection's control protocol. Signed-off-by: Inaky Perez-Gonzalez Cc: David Woodhouse Signed-off-by: Greg Kroah-Hartman --- include/linux/Kbuild | 2 ++ include/linux/wimax/Kbuild | 1 + 2 files changed, 3 insertions(+) create mode 100644 include/linux/wimax/Kbuild diff --git a/include/linux/Kbuild b/include/linux/Kbuild index a3323f337e4..12e9a2957ca 100644 --- a/include/linux/Kbuild +++ b/include/linux/Kbuild @@ -371,3 +371,5 @@ unifdef-y += xattr.h unifdef-y += xfrm.h objhdr-y += version.h +header-y += wimax.h +header-y += wimax/ diff --git a/include/linux/wimax/Kbuild b/include/linux/wimax/Kbuild new file mode 100644 index 00000000000..3cb4f269bb0 --- /dev/null +++ b/include/linux/wimax/Kbuild @@ -0,0 +1 @@ +header-y += i2400m.h From 8a70da82edc50aa7a4b54864babf2d72538ba1bb Mon Sep 17 00:00:00 2001 From: Inaky Perez-Gonzalez Date: Sat, 20 Dec 2008 16:58:00 -0800 Subject: [PATCH 123/123] wimax/i2400m: add CREDITS and MAINTAINERS entries This patch adds entries for the original developers of the i2400m drivers and up-to-date maintainer entries. Signed-off-by: Inaky Perez-Gonzalez Signed-off-by: Greg Kroah-Hartman --- CREDITS | 17 +++++++++++++++++ MAINTAINERS | 16 ++++++++++++++++ 2 files changed, 33 insertions(+) diff --git a/CREDITS b/CREDITS index abe05a0be4e..9f1736a8d5a 100644 --- a/CREDITS +++ b/CREDITS @@ -464,6 +464,11 @@ S: 1200 Goldenrod Dr. S: Nampa, Idaho 83686 S: USA +N: Dirk J. Brandewie +E: dirk.j.brandewie@intel.com +E: linux-wimax@intel.com +D: Intel Wireless WiMAX Connection 2400 SDIO driver + N: Derrick J. Brashear E: shadow@dementia.org W: http://www.dementia.org/~shadow @@ -2119,6 +2124,11 @@ N: H.J. Lu E: hjl@gnu.ai.mit.edu D: GCC + libraries hacker +N: Yanir Lubetkin +E: yanirx.lubatkin@intel.com +E: linux-wimax@intel.com +D: Intel Wireless WiMAX Connection 2400 driver + N: Michal Ludvig E: michal@logix.cz E: michal.ludvig@asterisk.co.nz @@ -2693,6 +2703,13 @@ S: RR #5, 497 Pole Line Road S: Thunder Bay, Ontario S: CANADA P7C 5M9 +N: Inaky Perez-Gonzalez +E: inaky.perez-gonzalez@intel.com +E: linux-wimax@intel.com +E: inakypg@yahoo.com +D: WiMAX stack +D: Intel Wireless WiMAX Connection 2400 driver + N: Yuri Per E: yuri@pts.mipt.ru D: Some smbfs fixes diff --git a/MAINTAINERS b/MAINTAINERS index f62fd6ac45a..2b03fb5f06a 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -2305,6 +2305,14 @@ W: http://lists.sourceforge.net/mailman/listinfo/ipw2100-devel W: http://ipw2200.sourceforge.net S: Supported +INTEL WIRELESS WIMAX CONNECTION 2400 +P: Inaky Perez-Gonzalez +M: inaky.perez-gonzalez@intel.com +M: linux-wimax@intel.com +L: wimax@linuxwimax.org +S: Supported +W: http://linuxwimax.org + INTEL WIRELESS WIFI LINK (iwlwifi) P: Zhu Yi M: yi.zhu@intel.com @@ -4733,6 +4741,14 @@ M: zaga@fly.cc.fer.hr L: linux-scsi@vger.kernel.org S: Maintained +WIMAX STACK +P: Inaky Perez-Gonzalez +M: inaky.perez-gonzalez@intel.com +M: linux-wimax@intel.com +L: wimax@linuxwimax.org +S: Supported +W: http://linuxwimax.org + WIMEDIA LLC PROTOCOL (WLP) SUBSYSTEM P: David Vrabel M: david.vrabel@csr.com