dm: avoid _hash_lock deadlock

Fix a reported deadlock if there are still unprocessed multipath events
on a device that is being removed.

_hash_lock is held during dev_remove while trying to send the
outstanding events.  Sending the events requests the _hash_lock
again in dm_copy_name_and_uuid.

This patch introduces a separate lock around regions that modify the
link to the hash table (dm_set_mdptr) or the name or uuid so that
dm_copy_name_and_uuid no longer needs _hash_lock.

Additionally, dm_copy_name_and_uuid can only be called if md exists
so we can drop the dm_get() and dm_put() which can lead to a BUG()
while md is being freed.

The deadlock:
 #0 [ffff8106298dfb48] schedule at ffffffff80063035
 #1 [ffff8106298dfc20] __down_read at ffffffff8006475d
 #2 [ffff8106298dfc60] dm_copy_name_and_uuid at ffffffff8824f740
 #3 [ffff8106298dfc90] dm_send_uevents at ffffffff88252685
 #4 [ffff8106298dfcd0] event_callback at ffffffff8824c678
 #5 [ffff8106298dfd00] dm_table_event at ffffffff8824dd01
 #6 [ffff8106298dfd10] __hash_remove at ffffffff882507ad
 #7 [ffff8106298dfd30] dev_remove at ffffffff88250865
 #8 [ffff8106298dfd60] ctl_ioctl at ffffffff88250d80
 #9 [ffff8106298dfee0] do_ioctl at ffffffff800418c4
#10 [ffff8106298dff00] vfs_ioctl at ffffffff8002fab9
#11 [ffff8106298dff40] sys_ioctl at ffffffff8004bdaf
#12 [ffff8106298dff80] tracesys at ffffffff8005d28d (via system_call)

Cc: stable@kernel.org
Reported-by: guy keren <choo@actcom.co.il>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
This commit is contained in:
Mikulas Patocka 2009-12-10 23:51:52 +00:00 committed by Alasdair G Kergon
parent 3067e02f8f
commit 6076905b5e
2 changed files with 17 additions and 9 deletions

View file

@ -56,6 +56,11 @@ static void dm_hash_remove_all(int keep_open_devices);
*/
static DECLARE_RWSEM(_hash_lock);
/*
* Protects use of mdptr to obtain hash cell name and uuid from mapped device.
*/
static DEFINE_MUTEX(dm_hash_cells_mutex);
static void init_buckets(struct list_head *buckets)
{
unsigned int i;
@ -206,7 +211,9 @@ static int dm_hash_insert(const char *name, const char *uuid, struct mapped_devi
list_add(&cell->uuid_list, _uuid_buckets + hash_str(uuid));
}
dm_get(md);
mutex_lock(&dm_hash_cells_mutex);
dm_set_mdptr(md, cell);
mutex_unlock(&dm_hash_cells_mutex);
up_write(&_hash_lock);
return 0;
@ -224,7 +231,9 @@ static void __hash_remove(struct hash_cell *hc)
/* remove from the dev hash */
list_del(&hc->uuid_list);
list_del(&hc->name_list);
mutex_lock(&dm_hash_cells_mutex);
dm_set_mdptr(hc->md, NULL);
mutex_unlock(&dm_hash_cells_mutex);
table = dm_get_table(hc->md);
if (table) {
@ -321,7 +330,9 @@ static int dm_hash_rename(uint32_t cookie, const char *old, const char *new)
*/
list_del(&hc->name_list);
old_name = hc->name;
mutex_lock(&dm_hash_cells_mutex);
hc->name = new_name;
mutex_unlock(&dm_hash_cells_mutex);
list_add(&hc->name_list, _name_buckets + hash_str(new_name));
/*
@ -1582,8 +1593,7 @@ int dm_copy_name_and_uuid(struct mapped_device *md, char *name, char *uuid)
if (!md)
return -ENXIO;
dm_get(md);
down_read(&_hash_lock);
mutex_lock(&dm_hash_cells_mutex);
hc = dm_get_mdptr(md);
if (!hc || hc->md != md) {
r = -ENXIO;
@ -1596,8 +1606,7 @@ int dm_copy_name_and_uuid(struct mapped_device *md, char *name, char *uuid)
strcpy(uuid, hc->uuid ? : "");
out:
up_read(&_hash_lock);
dm_put(md);
mutex_unlock(&dm_hash_cells_mutex);
return r;
}

View file

@ -139,13 +139,12 @@ void dm_send_uevents(struct list_head *events, struct kobject *kobj)
list_del_init(&event->elist);
/*
* Need to call dm_copy_name_and_uuid from here for now.
* Context of previous var adds and locking used for
* hash_cell not compatable.
* When a device is being removed this copy fails and we
* discard these unsent events.
*/
if (dm_copy_name_and_uuid(event->md, event->name,
event->uuid)) {
DMERR("%s: dm_copy_name_and_uuid() failed",
DMINFO("%s: skipping sending uevent for lost device",
__func__);
goto uevent_free;
}