From 15d0d3b3371227f846b9f644547fde081c7e1c0c Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Wed, 25 Feb 2009 06:22:45 +0100 Subject: [PATCH 1/9] generic IPI: simplify barriers and locking Simplify the barriers in generic remote function call interrupt code. Firstly, just unconditionally take the lock and check the list in the generic_call_function_single_interrupt IPI handler. As we've just taken an IPI here, the chances are fairly high that there will be work on the list for us, so do the locking unconditionally. This removes the tricky lockless list_empty check and dubious barriers. The change looks bigger than it is because it is just removing an outer loop. Secondly, clarify architecture specific IPI locking rules. Generic code has no tools to impose any sane ordering on IPIs if they go outside normal cache coherency, ergo the arch code must make them appear to obey cache coherency as a "memory operation" to initiate an IPI, and a "memory operation" to receive one. This way at least they can be reasoned about in generic code, and smp_mb used to provide ordering. The combination of these two changes means that explict barriers can be taken out of queue handling for the single case -- shared data is explicitly locked, and ipi ordering must conform to that, so no barriers needed. An extra barrier is needed in the many handler, so as to ensure we load the list element after the IPI is received. Does any architecture actually *need* these barriers? For the initiator I could see it, but for the handler I would be surprised. So the other thing we could do for simplicity is just to require that, rather than just matching with cache coherency, we just require a full barrier before generating an IPI, and after receiving an IPI. In which case, the smp_mb()s can go away. But just for now, we'll be on the safe side and use the barriers (they're in the slow case anyway). Signed-off-by: Nick Piggin Acked-by: Peter Zijlstra Cc: linux-arch@vger.kernel.org Cc: Andrew Morton Cc: Linus Torvalds Cc: Jens Axboe Cc: Oleg Nesterov Cc: Suresh Siddha Signed-off-by: Ingo Molnar --- kernel/smp.c | 83 ++++++++++++++++++++++++++++------------------------ 1 file changed, 44 insertions(+), 39 deletions(-) diff --git a/kernel/smp.c b/kernel/smp.c index bbedbb7efe3..6ecf4b9895d 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -74,9 +74,16 @@ static void generic_exec_single(int cpu, struct call_single_data *data) spin_unlock_irqrestore(&dst->lock, flags); /* - * Make the list addition visible before sending the ipi. + * The list addition should be visible before sending the IPI + * handler locks the list to pull the entry off it because of + * normal cache coherency rules implied by spinlocks. + * + * If IPIs can go out of order to the cache coherency protocol + * in an architecture, sufficient synchronisation should be added + * to arch code to make it appear to obey cache coherency WRT + * locking and barrier primitives. Generic code isn't really equipped + * to do the right thing... */ - smp_mb(); if (ipi) arch_send_call_function_single_ipi(cpu); @@ -103,6 +110,14 @@ void generic_smp_call_function_interrupt(void) struct call_function_data *data; int cpu = get_cpu(); + /* + * Ensure entry is visible on call_function_queue after we have + * entered the IPI. See comment in smp_call_function_many. + * If we don't have this, then we may miss an entry on the list + * and never get another IPI to process it. + */ + smp_mb(); + /* * It's ok to use list_for_each_rcu() here even though we may delete * 'pos', since list_del_rcu() doesn't clear ->next @@ -154,49 +169,37 @@ void generic_smp_call_function_single_interrupt(void) { struct call_single_queue *q = &__get_cpu_var(call_single_queue); LIST_HEAD(list); + unsigned int data_flags; - /* - * Need to see other stores to list head for checking whether - * list is empty without holding q->lock - */ - smp_read_barrier_depends(); - while (!list_empty(&q->list)) { - unsigned int data_flags; + spin_lock(&q->lock); + list_replace_init(&q->list, &list); + spin_unlock(&q->lock); - spin_lock(&q->lock); - list_replace_init(&q->list, &list); - spin_unlock(&q->lock); + while (!list_empty(&list)) { + struct call_single_data *data; - while (!list_empty(&list)) { - struct call_single_data *data; + data = list_entry(list.next, struct call_single_data, + list); + list_del(&data->list); - data = list_entry(list.next, struct call_single_data, - list); - list_del(&data->list); - - /* - * 'data' can be invalid after this call if - * flags == 0 (when called through - * generic_exec_single(), so save them away before - * making the call. - */ - data_flags = data->flags; - - data->func(data->info); - - if (data_flags & CSD_FLAG_WAIT) { - smp_wmb(); - data->flags &= ~CSD_FLAG_WAIT; - } else if (data_flags & CSD_FLAG_LOCK) { - smp_wmb(); - data->flags &= ~CSD_FLAG_LOCK; - } else if (data_flags & CSD_FLAG_ALLOC) - kfree(data); - } /* - * See comment on outer loop + * 'data' can be invalid after this call if + * flags == 0 (when called through + * generic_exec_single(), so save them away before + * making the call. */ - smp_read_barrier_depends(); + data_flags = data->flags; + + data->func(data->info); + + if (data_flags & CSD_FLAG_WAIT) { + smp_wmb(); + data->flags &= ~CSD_FLAG_WAIT; + } else if (data_flags & CSD_FLAG_LOCK) { + smp_wmb(); + data->flags &= ~CSD_FLAG_LOCK; + } else if (data_flags & CSD_FLAG_ALLOC) + kfree(data); } } @@ -375,6 +378,8 @@ void smp_call_function_many(const struct cpumask *mask, /* * Make the list addition visible before sending the ipi. + * (IPIs must obey or appear to obey normal Linux cache coherency + * rules -- see comment in generic_exec_single). */ smp_mb(); From 8969a5ede0f9e17da4b943712429aef2c9bcd82b Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 25 Feb 2009 13:59:47 +0100 Subject: [PATCH 2/9] generic-ipi: remove kmalloc() Remove the use of kmalloc() from the smp_call_function_*() calls. Steven's generic-ipi patch (d7240b98: generic-ipi: use per cpu data for single cpu ipi calls) started the discussion on the use of kmalloc() in this code and fixed the smp_call_function_single(.wait=0) fallback case. In this patch we complete this by also providing means for the _many() call, which fully removes the need for kmalloc() in this code. The problem with the _many() call is that other cpus might still be observing our entry when we're done with it. It solved this by dynamically allocating data elements and RCU-freeing it. We solve it by using a single per-cpu entry which provides static storage and solves one half of the problem (avoiding referencing freed data). The other half, ensuring the queue iteration it still possible, is done by placing re-used entries at the head of the list. This means that if someone was still iterating that entry when it got moved, he will now re-visit the entries on the list he had already seen, but avoids skipping over entries like would have happened had we placed the new entry at the end. Furthermore, visiting entries twice is not a problem, since we remove our cpu from the entry's cpumask once its called. Many thanks to Oleg for his suggestions and him poking holes in my earlier attempts. Signed-off-by: Peter Zijlstra Cc: Oleg Nesterov Cc: Linus Torvalds Cc: Nick Piggin Cc: Jens Axboe Cc: "Paul E. McKenney" Cc: Rusty Russell Signed-off-by: Ingo Molnar --- kernel/smp.c | 264 ++++++++++++++++++++++++++++++++------------------- 1 file changed, 166 insertions(+), 98 deletions(-) diff --git a/kernel/smp.c b/kernel/smp.c index 6ecf4b9895d..7a0ce25829d 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -10,23 +10,28 @@ #include #include #include +#include static DEFINE_PER_CPU(struct call_single_queue, call_single_queue); -static LIST_HEAD(call_function_queue); -__cacheline_aligned_in_smp DEFINE_SPINLOCK(call_function_lock); + +static struct { + struct list_head queue; + spinlock_t lock; +} call_function __cacheline_aligned_in_smp = { + .queue = LIST_HEAD_INIT(call_function.queue), + .lock = __SPIN_LOCK_UNLOCKED(call_function.lock), +}; enum { CSD_FLAG_WAIT = 0x01, - CSD_FLAG_ALLOC = 0x02, - CSD_FLAG_LOCK = 0x04, + CSD_FLAG_LOCK = 0x02, }; struct call_function_data { struct call_single_data csd; spinlock_t lock; unsigned int refs; - struct rcu_head rcu_head; - unsigned long cpumask_bits[]; + cpumask_var_t cpumask; }; struct call_single_queue { @@ -34,8 +39,45 @@ struct call_single_queue { spinlock_t lock; }; +static DEFINE_PER_CPU(struct call_function_data, cfd_data) = { + .lock = __SPIN_LOCK_UNLOCKED(cfd_data.lock), +}; + +static int +hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu) +{ + long cpu = (long)hcpu; + struct call_function_data *cfd = &per_cpu(cfd_data, cpu); + + switch (action) { + case CPU_UP_PREPARE: + case CPU_UP_PREPARE_FROZEN: + if (!alloc_cpumask_var_node(&cfd->cpumask, GFP_KERNEL, + cpu_to_node(cpu))) + return NOTIFY_BAD; + break; + +#ifdef CONFIG_CPU_HOTPLUG + case CPU_UP_CANCELED: + case CPU_UP_CANCELED_FROZEN: + + case CPU_DEAD: + case CPU_DEAD_FROZEN: + free_cpumask_var(cfd->cpumask); + break; +#endif + }; + + return NOTIFY_OK; +} + +static struct notifier_block __cpuinitdata hotplug_cfd_notifier = { + .notifier_call = hotplug_cfd, +}; + static int __cpuinit init_call_single_data(void) { + void *cpu = (void *)(long)smp_processor_id(); int i; for_each_possible_cpu(i) { @@ -44,18 +86,69 @@ static int __cpuinit init_call_single_data(void) spin_lock_init(&q->lock); INIT_LIST_HEAD(&q->list); } + + hotplug_cfd(&hotplug_cfd_notifier, CPU_UP_PREPARE, cpu); + register_cpu_notifier(&hotplug_cfd_notifier); + return 0; } early_initcall(init_call_single_data); -static void csd_flag_wait(struct call_single_data *data) +/* + * csd_wait/csd_complete are used for synchronous ipi calls + */ +static void csd_wait_prepare(struct call_single_data *data) { - /* Wait for response */ - do { - if (!(data->flags & CSD_FLAG_WAIT)) - break; + data->flags |= CSD_FLAG_WAIT; +} + +static void csd_complete(struct call_single_data *data) +{ + if (data->flags & CSD_FLAG_WAIT) { + /* + * ensure we're all done before saying we are + */ + smp_mb(); + data->flags &= ~CSD_FLAG_WAIT; + } +} + +static void csd_wait(struct call_single_data *data) +{ + while (data->flags & CSD_FLAG_WAIT) cpu_relax(); - } while (1); +} + +/* + * csd_lock/csd_unlock used to serialize access to per-cpu csd resources + * + * For non-synchronous ipi calls the csd can still be in use by the previous + * function call. For multi-cpu calls its even more interesting as we'll have + * to ensure no other cpu is observing our csd. + */ +static void csd_lock(struct call_single_data *data) +{ + while (data->flags & CSD_FLAG_LOCK) + cpu_relax(); + data->flags = CSD_FLAG_LOCK; + + /* + * prevent CPU from reordering the above assignment to ->flags + * with any subsequent assignments to other fields of the + * specified call_single_data structure. + */ + + smp_mb(); +} + +static void csd_unlock(struct call_single_data *data) +{ + WARN_ON(!(data->flags & CSD_FLAG_LOCK)); + /* + * ensure we're all done before releasing data + */ + smp_mb(); + data->flags &= ~CSD_FLAG_LOCK; } /* @@ -89,16 +182,7 @@ static void generic_exec_single(int cpu, struct call_single_data *data) arch_send_call_function_single_ipi(cpu); if (wait) - csd_flag_wait(data); -} - -static void rcu_free_call_data(struct rcu_head *head) -{ - struct call_function_data *data; - - data = container_of(head, struct call_function_data, rcu_head); - - kfree(data); + csd_wait(data); } /* @@ -122,41 +206,35 @@ void generic_smp_call_function_interrupt(void) * It's ok to use list_for_each_rcu() here even though we may delete * 'pos', since list_del_rcu() doesn't clear ->next */ - rcu_read_lock(); - list_for_each_entry_rcu(data, &call_function_queue, csd.list) { + list_for_each_entry_rcu(data, &call_function.queue, csd.list) { int refs; - if (!cpumask_test_cpu(cpu, to_cpumask(data->cpumask_bits))) + spin_lock(&data->lock); + if (!cpumask_test_cpu(cpu, data->cpumask)) { + spin_unlock(&data->lock); continue; + } + cpumask_clear_cpu(cpu, data->cpumask); + spin_unlock(&data->lock); data->csd.func(data->csd.info); spin_lock(&data->lock); - cpumask_clear_cpu(cpu, to_cpumask(data->cpumask_bits)); WARN_ON(data->refs == 0); - data->refs--; - refs = data->refs; + refs = --data->refs; + if (!refs) { + spin_lock(&call_function.lock); + list_del_rcu(&data->csd.list); + spin_unlock(&call_function.lock); + } spin_unlock(&data->lock); if (refs) continue; - spin_lock(&call_function_lock); - list_del_rcu(&data->csd.list); - spin_unlock(&call_function_lock); - - if (data->csd.flags & CSD_FLAG_WAIT) { - /* - * serialize stores to data with the flag clear - * and wakeup - */ - smp_wmb(); - data->csd.flags &= ~CSD_FLAG_WAIT; - } - if (data->csd.flags & CSD_FLAG_ALLOC) - call_rcu(&data->rcu_head, rcu_free_call_data); + csd_complete(&data->csd); + csd_unlock(&data->csd); } - rcu_read_unlock(); put_cpu(); } @@ -192,14 +270,14 @@ void generic_smp_call_function_single_interrupt(void) data->func(data->info); - if (data_flags & CSD_FLAG_WAIT) { - smp_wmb(); - data->flags &= ~CSD_FLAG_WAIT; - } else if (data_flags & CSD_FLAG_LOCK) { - smp_wmb(); - data->flags &= ~CSD_FLAG_LOCK; - } else if (data_flags & CSD_FLAG_ALLOC) - kfree(data); + if (data_flags & CSD_FLAG_WAIT) + csd_complete(data); + + /* + * Unlocked CSDs are valid through generic_exec_single() + */ + if (data_flags & CSD_FLAG_LOCK) + csd_unlock(data); } } @@ -218,7 +296,9 @@ static DEFINE_PER_CPU(struct call_single_data, csd_data); int smp_call_function_single(int cpu, void (*func) (void *info), void *info, int wait) { - struct call_single_data d; + struct call_single_data d = { + .flags = 0, + }; unsigned long flags; /* prevent preemption and reschedule on another processor, as well as CPU removal */ @@ -239,13 +319,11 @@ int smp_call_function_single(int cpu, void (*func) (void *info), void *info, /* * We are calling a function on a single CPU * and we are not going to wait for it to finish. - * We first try to allocate the data, but if we - * fail, we fall back to use a per cpu data to pass - * the information to that CPU. Since all callers - * of this code will use the same data, we must - * synchronize the callers to prevent a new caller - * from corrupting the data before the callee - * can access it. + * We use a per cpu data to pass the information to + * that CPU. Since all callers of this code will + * use the same data, we must synchronize the + * callers to prevent a new caller from corrupting + * the data before the callee can access it. * * The CSD_FLAG_LOCK is used to let us know when * the IPI handler is done with the data. @@ -255,18 +333,11 @@ int smp_call_function_single(int cpu, void (*func) (void *info), void *info, * will make sure the callee is done with the * data before a new caller will use it. */ - data = kmalloc(sizeof(*data), GFP_ATOMIC); - if (data) - data->flags = CSD_FLAG_ALLOC; - else { - data = &per_cpu(csd_data, me); - while (data->flags & CSD_FLAG_LOCK) - cpu_relax(); - data->flags = CSD_FLAG_LOCK; - } + data = &__get_cpu_var(csd_data); + csd_lock(data); } else { data = &d; - data->flags = CSD_FLAG_WAIT; + csd_wait_prepare(data); } data->func = func; @@ -326,14 +397,14 @@ void smp_call_function_many(const struct cpumask *mask, { struct call_function_data *data; unsigned long flags; - int cpu, next_cpu; + int cpu, next_cpu, me = smp_processor_id(); /* Can deadlock when called with interrupts disabled */ WARN_ON(irqs_disabled()); /* So, what's a CPU they want? Ignoring this one. */ cpu = cpumask_first_and(mask, cpu_online_mask); - if (cpu == smp_processor_id()) + if (cpu == me) cpu = cpumask_next_and(cpu, mask, cpu_online_mask); /* No online cpus? We're done. */ if (cpu >= nr_cpu_ids) @@ -341,7 +412,7 @@ void smp_call_function_many(const struct cpumask *mask, /* Do we have another CPU which isn't us? */ next_cpu = cpumask_next_and(cpu, mask, cpu_online_mask); - if (next_cpu == smp_processor_id()) + if (next_cpu == me) next_cpu = cpumask_next_and(next_cpu, mask, cpu_online_mask); /* Fastpath: do that cpu by itself. */ @@ -350,31 +421,28 @@ void smp_call_function_many(const struct cpumask *mask, return; } - data = kmalloc(sizeof(*data) + cpumask_size(), GFP_ATOMIC); - if (unlikely(!data)) { - /* Slow path. */ - for_each_online_cpu(cpu) { - if (cpu == smp_processor_id()) - continue; - if (cpumask_test_cpu(cpu, mask)) - smp_call_function_single(cpu, func, info, wait); - } - return; - } + data = &__get_cpu_var(cfd_data); + csd_lock(&data->csd); - spin_lock_init(&data->lock); - data->csd.flags = CSD_FLAG_ALLOC; + spin_lock_irqsave(&data->lock, flags); if (wait) - data->csd.flags |= CSD_FLAG_WAIT; + csd_wait_prepare(&data->csd); + data->csd.func = func; data->csd.info = info; - cpumask_and(to_cpumask(data->cpumask_bits), mask, cpu_online_mask); - cpumask_clear_cpu(smp_processor_id(), to_cpumask(data->cpumask_bits)); - data->refs = cpumask_weight(to_cpumask(data->cpumask_bits)); + cpumask_and(data->cpumask, mask, cpu_online_mask); + cpumask_clear_cpu(me, data->cpumask); + data->refs = cpumask_weight(data->cpumask); - spin_lock_irqsave(&call_function_lock, flags); - list_add_tail_rcu(&data->csd.list, &call_function_queue); - spin_unlock_irqrestore(&call_function_lock, flags); + spin_lock(&call_function.lock); + /* + * Place entry at the _HEAD_ of the list, so that any cpu still + * observing the entry in generic_smp_call_function_interrupt() will + * not miss any other list entries. + */ + list_add_rcu(&data->csd.list, &call_function.queue); + spin_unlock(&call_function.lock); + spin_unlock_irqrestore(&data->lock, flags); /* * Make the list addition visible before sending the ipi. @@ -384,11 +452,11 @@ void smp_call_function_many(const struct cpumask *mask, smp_mb(); /* Send a message to all CPUs in the map */ - arch_send_call_function_ipi_mask(to_cpumask(data->cpumask_bits)); + arch_send_call_function_ipi_mask(data->cpumask); /* optionally wait for the CPUs to complete */ if (wait) - csd_flag_wait(&data->csd); + csd_wait(&data->csd); } EXPORT_SYMBOL(smp_call_function_many); @@ -418,20 +486,20 @@ EXPORT_SYMBOL(smp_call_function); void ipi_call_lock(void) { - spin_lock(&call_function_lock); + spin_lock(&call_function.lock); } void ipi_call_unlock(void) { - spin_unlock(&call_function_lock); + spin_unlock(&call_function.lock); } void ipi_call_lock_irq(void) { - spin_lock_irq(&call_function_lock); + spin_lock_irq(&call_function.lock); } void ipi_call_unlock_irq(void) { - spin_unlock_irq(&call_function_lock); + spin_unlock_irq(&call_function.lock); } From 6e2756376c706e4da3454a272947983f92e80a7e Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 25 Feb 2009 13:59:48 +0100 Subject: [PATCH 3/9] generic-ipi: remove CSD_FLAG_WAIT Oleg noticed that we don't strictly need CSD_FLAG_WAIT, rework the code so that we can use CSD_FLAG_LOCK for both purposes. Signed-off-by: Peter Zijlstra Cc: Oleg Nesterov Cc: Linus Torvalds Cc: Nick Piggin Cc: Jens Axboe Cc: "Paul E. McKenney" Cc: Rusty Russell Signed-off-by: Ingo Molnar --- block/blk-softirq.c | 2 +- include/linux/smp.h | 3 +- kernel/sched.c | 2 +- kernel/smp.c | 92 ++++++++++++--------------------------------- kernel/softirq.c | 2 +- 5 files changed, 29 insertions(+), 72 deletions(-) diff --git a/block/blk-softirq.c b/block/blk-softirq.c index ce0efc6b26d..ee9c2160222 100644 --- a/block/blk-softirq.c +++ b/block/blk-softirq.c @@ -64,7 +64,7 @@ static int raise_blk_irq(int cpu, struct request *rq) data->info = rq; data->flags = 0; - __smp_call_function_single(cpu, data); + __smp_call_function_single(cpu, data, 0); return 0; } diff --git a/include/linux/smp.h b/include/linux/smp.h index 715196b09d6..00866d7fdf3 100644 --- a/include/linux/smp.h +++ b/include/linux/smp.h @@ -82,7 +82,8 @@ smp_call_function_mask(cpumask_t mask, void(*func)(void *info), void *info, return 0; } -void __smp_call_function_single(int cpuid, struct call_single_data *data); +void __smp_call_function_single(int cpuid, struct call_single_data *data, + int wait); /* * Generic and arch helpers diff --git a/kernel/sched.c b/kernel/sched.c index 410eec40413..d4c2749a299 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1093,7 +1093,7 @@ static void hrtick_start(struct rq *rq, u64 delay) if (rq == this_rq()) { hrtimer_restart(timer); } else if (!rq->hrtick_csd_pending) { - __smp_call_function_single(cpu_of(rq), &rq->hrtick_csd); + __smp_call_function_single(cpu_of(rq), &rq->hrtick_csd, 0); rq->hrtick_csd_pending = 1; } } diff --git a/kernel/smp.c b/kernel/smp.c index 7a0ce25829d..f5308258891 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -23,8 +23,7 @@ static struct { }; enum { - CSD_FLAG_WAIT = 0x01, - CSD_FLAG_LOCK = 0x02, + CSD_FLAG_LOCK = 0x01, }; struct call_function_data { @@ -94,31 +93,6 @@ static int __cpuinit init_call_single_data(void) } early_initcall(init_call_single_data); -/* - * csd_wait/csd_complete are used for synchronous ipi calls - */ -static void csd_wait_prepare(struct call_single_data *data) -{ - data->flags |= CSD_FLAG_WAIT; -} - -static void csd_complete(struct call_single_data *data) -{ - if (data->flags & CSD_FLAG_WAIT) { - /* - * ensure we're all done before saying we are - */ - smp_mb(); - data->flags &= ~CSD_FLAG_WAIT; - } -} - -static void csd_wait(struct call_single_data *data) -{ - while (data->flags & CSD_FLAG_WAIT) - cpu_relax(); -} - /* * csd_lock/csd_unlock used to serialize access to per-cpu csd resources * @@ -126,10 +100,15 @@ static void csd_wait(struct call_single_data *data) * function call. For multi-cpu calls its even more interesting as we'll have * to ensure no other cpu is observing our csd. */ -static void csd_lock(struct call_single_data *data) +static void csd_lock_wait(struct call_single_data *data) { while (data->flags & CSD_FLAG_LOCK) cpu_relax(); +} + +static void csd_lock(struct call_single_data *data) +{ + csd_lock_wait(data); data->flags = CSD_FLAG_LOCK; /* @@ -155,11 +134,12 @@ static void csd_unlock(struct call_single_data *data) * Insert a previously allocated call_single_data element for execution * on the given CPU. data must already have ->func, ->info, and ->flags set. */ -static void generic_exec_single(int cpu, struct call_single_data *data) +static +void generic_exec_single(int cpu, struct call_single_data *data, int wait) { struct call_single_queue *dst = &per_cpu(call_single_queue, cpu); - int wait = data->flags & CSD_FLAG_WAIT, ipi; unsigned long flags; + int ipi; spin_lock_irqsave(&dst->lock, flags); ipi = list_empty(&dst->list); @@ -182,7 +162,7 @@ static void generic_exec_single(int cpu, struct call_single_data *data) arch_send_call_function_single_ipi(cpu); if (wait) - csd_wait(data); + csd_lock_wait(data); } /* @@ -232,7 +212,6 @@ void generic_smp_call_function_interrupt(void) if (refs) continue; - csd_complete(&data->csd); csd_unlock(&data->csd); } @@ -270,9 +249,6 @@ void generic_smp_call_function_single_interrupt(void) data->func(data->info); - if (data_flags & CSD_FLAG_WAIT) - csd_complete(data); - /* * Unlocked CSDs are valid through generic_exec_single() */ @@ -313,36 +289,16 @@ int smp_call_function_single(int cpu, void (*func) (void *info), void *info, func(info); local_irq_restore(flags); } else if ((unsigned)cpu < nr_cpu_ids && cpu_online(cpu)) { - struct call_single_data *data; + struct call_single_data *data = &d; - if (!wait) { - /* - * We are calling a function on a single CPU - * and we are not going to wait for it to finish. - * We use a per cpu data to pass the information to - * that CPU. Since all callers of this code will - * use the same data, we must synchronize the - * callers to prevent a new caller from corrupting - * the data before the callee can access it. - * - * The CSD_FLAG_LOCK is used to let us know when - * the IPI handler is done with the data. - * The first caller will set it, and the callee - * will clear it. The next caller must wait for - * it to clear before we set it again. This - * will make sure the callee is done with the - * data before a new caller will use it. - */ + if (!wait) data = &__get_cpu_var(csd_data); - csd_lock(data); - } else { - data = &d; - csd_wait_prepare(data); - } + + csd_lock(data); data->func = func; data->info = info; - generic_exec_single(cpu, data); + generic_exec_single(cpu, data, wait); } else { err = -ENXIO; /* CPU not online */ } @@ -362,12 +318,15 @@ EXPORT_SYMBOL(smp_call_function_single); * instance. * */ -void __smp_call_function_single(int cpu, struct call_single_data *data) +void __smp_call_function_single(int cpu, struct call_single_data *data, + int wait) { - /* Can deadlock when called with interrupts disabled */ - WARN_ON((data->flags & CSD_FLAG_WAIT) && irqs_disabled()); + csd_lock(data); - generic_exec_single(cpu, data); + /* Can deadlock when called with interrupts disabled */ + WARN_ON(wait && irqs_disabled()); + + generic_exec_single(cpu, data, wait); } /* FIXME: Shim for archs using old arch_send_call_function_ipi API. */ @@ -425,9 +384,6 @@ void smp_call_function_many(const struct cpumask *mask, csd_lock(&data->csd); spin_lock_irqsave(&data->lock, flags); - if (wait) - csd_wait_prepare(&data->csd); - data->csd.func = func; data->csd.info = info; cpumask_and(data->cpumask, mask, cpu_online_mask); @@ -456,7 +412,7 @@ void smp_call_function_many(const struct cpumask *mask, /* optionally wait for the CPUs to complete */ if (wait) - csd_wait(&data->csd); + csd_lock_wait(&data->csd); } EXPORT_SYMBOL(smp_call_function_many); diff --git a/kernel/softirq.c b/kernel/softirq.c index bdbe9de9cd8..48c3d5d627a 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -496,7 +496,7 @@ static int __try_remote_softirq(struct call_single_data *cp, int cpu, int softir cp->flags = 0; cp->priv = softirq; - __smp_call_function_single(cpu, cp); + __smp_call_function_single(cpu, cp, 0); return 0; } return 1; From 0b13fda1e0936b3d64c4c407f183d33fa6bd2ad4 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 25 Feb 2009 16:52:11 +0100 Subject: [PATCH 4/9] generic-ipi: cleanups Andrew pointed out that there's some small amount of style rot in kernel/smp.c. Clean it up. Reported-by: Andrew Morton Cc: Nick Piggin Cc: Jens Axboe Cc: Peter Zijlstra Signed-off-by: Ingo Molnar --- kernel/smp.c | 166 +++++++++++++++++++++++++++------------------------ 1 file changed, 88 insertions(+), 78 deletions(-) diff --git a/kernel/smp.c b/kernel/smp.c index f5308258891..7ad2262d2ec 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -2,13 +2,12 @@ * Generic helpers for smp ipi calls * * (C) Jens Axboe 2008 - * */ -#include -#include -#include #include #include +#include +#include +#include #include #include @@ -17,29 +16,30 @@ static DEFINE_PER_CPU(struct call_single_queue, call_single_queue); static struct { struct list_head queue; spinlock_t lock; -} call_function __cacheline_aligned_in_smp = { - .queue = LIST_HEAD_INIT(call_function.queue), - .lock = __SPIN_LOCK_UNLOCKED(call_function.lock), -}; +} call_function __cacheline_aligned_in_smp = + { + .queue = LIST_HEAD_INIT(call_function.queue), + .lock = __SPIN_LOCK_UNLOCKED(call_function.lock), + }; enum { CSD_FLAG_LOCK = 0x01, }; struct call_function_data { - struct call_single_data csd; - spinlock_t lock; - unsigned int refs; - cpumask_var_t cpumask; + struct call_single_data csd; + spinlock_t lock; + unsigned int refs; + cpumask_var_t cpumask; }; struct call_single_queue { - struct list_head list; - spinlock_t lock; + struct list_head list; + spinlock_t lock; }; static DEFINE_PER_CPU(struct call_function_data, cfd_data) = { - .lock = __SPIN_LOCK_UNLOCKED(cfd_data.lock), + .lock = __SPIN_LOCK_UNLOCKED(cfd_data.lock), }; static int @@ -71,7 +71,7 @@ hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu) } static struct notifier_block __cpuinitdata hotplug_cfd_notifier = { - .notifier_call = hotplug_cfd, + .notifier_call = hotplug_cfd, }; static int __cpuinit init_call_single_data(void) @@ -96,9 +96,9 @@ early_initcall(init_call_single_data); /* * csd_lock/csd_unlock used to serialize access to per-cpu csd resources * - * For non-synchronous ipi calls the csd can still be in use by the previous - * function call. For multi-cpu calls its even more interesting as we'll have - * to ensure no other cpu is observing our csd. + * For non-synchronous ipi calls the csd can still be in use by the + * previous function call. For multi-cpu calls its even more interesting + * as we'll have to ensure no other cpu is observing our csd. */ static void csd_lock_wait(struct call_single_data *data) { @@ -112,27 +112,29 @@ static void csd_lock(struct call_single_data *data) data->flags = CSD_FLAG_LOCK; /* - * prevent CPU from reordering the above assignment to ->flags - * with any subsequent assignments to other fields of the - * specified call_single_data structure. + * prevent CPU from reordering the above assignment + * to ->flags with any subsequent assignments to other + * fields of the specified call_single_data structure: */ - smp_mb(); } static void csd_unlock(struct call_single_data *data) { WARN_ON(!(data->flags & CSD_FLAG_LOCK)); + /* - * ensure we're all done before releasing data + * ensure we're all done before releasing data: */ smp_mb(); + data->flags &= ~CSD_FLAG_LOCK; } /* - * Insert a previously allocated call_single_data element for execution - * on the given CPU. data must already have ->func, ->info, and ->flags set. + * Insert a previously allocated call_single_data element + * for execution on the given CPU. data must already have + * ->func, ->info, and ->flags set. */ static void generic_exec_single(int cpu, struct call_single_data *data, int wait) @@ -154,10 +156,9 @@ void generic_exec_single(int cpu, struct call_single_data *data, int wait) * If IPIs can go out of order to the cache coherency protocol * in an architecture, sufficient synchronisation should be added * to arch code to make it appear to obey cache coherency WRT - * locking and barrier primitives. Generic code isn't really equipped - * to do the right thing... + * locking and barrier primitives. Generic code isn't really + * equipped to do the right thing... */ - if (ipi) arch_send_call_function_single_ipi(cpu); @@ -183,8 +184,8 @@ void generic_smp_call_function_interrupt(void) smp_mb(); /* - * It's ok to use list_for_each_rcu() here even though we may delete - * 'pos', since list_del_rcu() doesn't clear ->next + * It's ok to use list_for_each_rcu() here even though we may + * delete 'pos', since list_del_rcu() doesn't clear ->next */ list_for_each_entry_rcu(data, &call_function.queue, csd.list) { int refs; @@ -219,14 +220,14 @@ void generic_smp_call_function_interrupt(void) } /* - * Invoked by arch to handle an IPI for call function single. Must be called - * from the arch with interrupts disabled. + * Invoked by arch to handle an IPI for call function single. Must be + * called from the arch with interrupts disabled. */ void generic_smp_call_function_single_interrupt(void) { struct call_single_queue *q = &__get_cpu_var(call_single_queue); - LIST_HEAD(list); unsigned int data_flags; + LIST_HEAD(list); spin_lock(&q->lock); list_replace_init(&q->list, &list); @@ -235,22 +236,20 @@ void generic_smp_call_function_single_interrupt(void) while (!list_empty(&list)) { struct call_single_data *data; - data = list_entry(list.next, struct call_single_data, - list); + data = list_entry(list.next, struct call_single_data, list); list_del(&data->list); /* - * 'data' can be invalid after this call if - * flags == 0 (when called through - * generic_exec_single(), so save them away before - * making the call. + * 'data' can be invalid after this call if flags == 0 + * (when called through generic_exec_single()), + * so save them away before making the call: */ data_flags = data->flags; data->func(data->info); /* - * Unlocked CSDs are valid through generic_exec_single() + * Unlocked CSDs are valid through generic_exec_single(): */ if (data_flags & CSD_FLAG_LOCK) csd_unlock(data); @@ -276,34 +275,41 @@ int smp_call_function_single(int cpu, void (*func) (void *info), void *info, .flags = 0, }; unsigned long flags; - /* prevent preemption and reschedule on another processor, - as well as CPU removal */ - int me = get_cpu(); + int this_cpu; int err = 0; + /* + * prevent preemption and reschedule on another processor, + * as well as CPU removal + */ + this_cpu = get_cpu(); + /* Can deadlock when called with interrupts disabled */ WARN_ON(irqs_disabled()); - if (cpu == me) { + if (cpu == this_cpu) { local_irq_save(flags); func(info); local_irq_restore(flags); - } else if ((unsigned)cpu < nr_cpu_ids && cpu_online(cpu)) { - struct call_single_data *data = &d; - - if (!wait) - data = &__get_cpu_var(csd_data); - - csd_lock(data); - - data->func = func; - data->info = info; - generic_exec_single(cpu, data, wait); } else { - err = -ENXIO; /* CPU not online */ + if ((unsigned)cpu < nr_cpu_ids && cpu_online(cpu)) { + struct call_single_data *data = &d; + + if (!wait) + data = &__get_cpu_var(csd_data); + + csd_lock(data); + + data->func = func; + data->info = info; + generic_exec_single(cpu, data, wait); + } else { + err = -ENXIO; /* CPU not online */ + } } put_cpu(); + return err; } EXPORT_SYMBOL(smp_call_function_single); @@ -313,10 +319,9 @@ EXPORT_SYMBOL(smp_call_function_single); * @cpu: The CPU to run on. * @data: Pre-allocated and setup data structure * - * Like smp_call_function_single(), but allow caller to pass in a pre-allocated - * data structure. Useful for embedding @data inside other structures, for - * instance. - * + * Like smp_call_function_single(), but allow caller to pass in a + * pre-allocated data structure. Useful for embedding @data inside + * other structures, for instance. */ void __smp_call_function_single(int cpu, struct call_single_data *data, int wait) @@ -329,10 +334,11 @@ void __smp_call_function_single(int cpu, struct call_single_data *data, generic_exec_single(cpu, data, wait); } -/* FIXME: Shim for archs using old arch_send_call_function_ipi API. */ +/* Deprecated: shim for archs using old arch_send_call_function_ipi API. */ + #ifndef arch_send_call_function_ipi_mask -#define arch_send_call_function_ipi_mask(maskp) \ - arch_send_call_function_ipi(*(maskp)) +# define arch_send_call_function_ipi_mask(maskp) \ + arch_send_call_function_ipi(*(maskp)) #endif /** @@ -340,7 +346,8 @@ void __smp_call_function_single(int cpu, struct call_single_data *data, * @mask: The set of cpus to run on (only runs on online subset). * @func: The function to run. This must be fast and non-blocking. * @info: An arbitrary pointer to pass to the function. - * @wait: If true, wait (atomically) until function has completed on other CPUs. + * @wait: If true, wait (atomically) until function has completed + * on other CPUs. * * If @wait is true, then returns once @func has returned. Note that @wait * will be implicitly turned on in case of allocation failures, since @@ -351,27 +358,27 @@ void __smp_call_function_single(int cpu, struct call_single_data *data, * must be disabled when calling this function. */ void smp_call_function_many(const struct cpumask *mask, - void (*func)(void *), void *info, - bool wait) + void (*func)(void *), void *info, bool wait) { struct call_function_data *data; unsigned long flags; - int cpu, next_cpu, me = smp_processor_id(); + int cpu, next_cpu, this_cpu = smp_processor_id(); /* Can deadlock when called with interrupts disabled */ WARN_ON(irqs_disabled()); - /* So, what's a CPU they want? Ignoring this one. */ + /* So, what's a CPU they want? Ignoring this one. */ cpu = cpumask_first_and(mask, cpu_online_mask); - if (cpu == me) + if (cpu == this_cpu) cpu = cpumask_next_and(cpu, mask, cpu_online_mask); + /* No online cpus? We're done. */ if (cpu >= nr_cpu_ids) return; /* Do we have another CPU which isn't us? */ next_cpu = cpumask_next_and(cpu, mask, cpu_online_mask); - if (next_cpu == me) + if (next_cpu == this_cpu) next_cpu = cpumask_next_and(next_cpu, mask, cpu_online_mask); /* Fastpath: do that cpu by itself. */ @@ -387,30 +394,31 @@ void smp_call_function_many(const struct cpumask *mask, data->csd.func = func; data->csd.info = info; cpumask_and(data->cpumask, mask, cpu_online_mask); - cpumask_clear_cpu(me, data->cpumask); + cpumask_clear_cpu(this_cpu, data->cpumask); data->refs = cpumask_weight(data->cpumask); spin_lock(&call_function.lock); /* * Place entry at the _HEAD_ of the list, so that any cpu still - * observing the entry in generic_smp_call_function_interrupt() will - * not miss any other list entries. + * observing the entry in generic_smp_call_function_interrupt() + * will not miss any other list entries: */ list_add_rcu(&data->csd.list, &call_function.queue); spin_unlock(&call_function.lock); + spin_unlock_irqrestore(&data->lock, flags); /* * Make the list addition visible before sending the ipi. - * (IPIs must obey or appear to obey normal Linux cache coherency - * rules -- see comment in generic_exec_single). + * (IPIs must obey or appear to obey normal Linux cache + * coherency rules -- see comment in generic_exec_single). */ smp_mb(); /* Send a message to all CPUs in the map */ arch_send_call_function_ipi_mask(data->cpumask); - /* optionally wait for the CPUs to complete */ + /* Optionally wait for the CPUs to complete */ if (wait) csd_lock_wait(&data->csd); } @@ -420,7 +428,8 @@ EXPORT_SYMBOL(smp_call_function_many); * smp_call_function(): Run a function on all other CPUs. * @func: The function to run. This must be fast and non-blocking. * @info: An arbitrary pointer to pass to the function. - * @wait: If true, wait (atomically) until function has completed on other CPUs. + * @wait: If true, wait (atomically) until function has completed + * on other CPUs. * * Returns 0. * @@ -436,6 +445,7 @@ int smp_call_function(void (*func)(void *), void *info, int wait) preempt_disable(); smp_call_function_many(cpu_online_mask, func, info, wait); preempt_enable(); + return 0; } EXPORT_SYMBOL(smp_call_function); From 641cd4cfcdc71ce01535b31cc4d57d59a1fae1fc Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 13 Mar 2009 10:47:34 +0100 Subject: [PATCH 5/9] generic-ipi: eliminate WARN_ON()s during oops/panic Do not output smp-call related warnings in the oops/panic codepath. Reported-by: Jan Beulich Acked-by: Peter Zijlstra LKML-Reference: <49B91A7E.76E4.0078.0@novell.com> Signed-off-by: Ingo Molnar --- kernel/smp.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/kernel/smp.c b/kernel/smp.c index 7ad2262d2ec..858baac568e 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -5,6 +5,7 @@ */ #include #include +#include #include #include #include @@ -285,7 +286,7 @@ int smp_call_function_single(int cpu, void (*func) (void *info), void *info, this_cpu = get_cpu(); /* Can deadlock when called with interrupts disabled */ - WARN_ON(irqs_disabled()); + WARN_ON_ONCE(irqs_disabled() && !oops_in_progress); if (cpu == this_cpu) { local_irq_save(flags); @@ -329,7 +330,7 @@ void __smp_call_function_single(int cpu, struct call_single_data *data, csd_lock(data); /* Can deadlock when called with interrupts disabled */ - WARN_ON(wait && irqs_disabled()); + WARN_ON_ONCE(wait && irqs_disabled() && !oops_in_progress); generic_exec_single(cpu, data, wait); } @@ -365,7 +366,7 @@ void smp_call_function_many(const struct cpumask *mask, int cpu, next_cpu, this_cpu = smp_processor_id(); /* Can deadlock when called with interrupts disabled */ - WARN_ON(irqs_disabled()); + WARN_ON_ONCE(irqs_disabled() && !oops_in_progress); /* So, what's a CPU they want? Ignoring this one. */ cpu = cpumask_first_and(mask, cpu_online_mask); From ffd71da4e3f323b7673b061e6f7e0d0c12dc2b49 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 13 Mar 2009 10:54:24 +0100 Subject: [PATCH 6/9] panic: decrease oops_in_progress only after having done the panic Impact: eliminate secondary warnings during panic() We can panic() in a number of difficult, atomic contexts, hence we use bust_spinlocks(1) in panic() to increase oops_in_progress, which prevents various debug checks we have in place. But in practice this protection only covers the first few printk's done by panic() - it does not cover the later attempt to stop all other CPUs and kexec(). If a secondary warning triggers in one of those facilities that can make the panic message scroll off. So do bust_spinlocks(0) only much later in panic(). (which code is only reached if panic policy is relaxed that it can return after a warning message) Reported-by: Jan Beulich LKML-Reference: <49B91A7E.76E4.0078.0@novell.com> Signed-off-by: Ingo Molnar --- kernel/panic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/panic.c b/kernel/panic.c index 32fe4eff1b8..57fb005de54 100644 --- a/kernel/panic.c +++ b/kernel/panic.c @@ -77,7 +77,6 @@ NORET_TYPE void panic(const char * fmt, ...) #ifdef CONFIG_DEBUG_BUGVERBOSE dump_stack(); #endif - bust_spinlocks(0); /* * If we have crashed and we have a crash kernel loaded let it handle @@ -136,6 +135,7 @@ NORET_TYPE void panic(const char * fmt, ...) mdelay(1); i++; } + bust_spinlocks(0); } EXPORT_SYMBOL(panic); From d1dedb52acd98bd5e13e1ff4c4d045d58bbd16fe Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 13 Mar 2009 11:14:06 +0100 Subject: [PATCH 7/9] panic, smp: provide smp_send_stop() wrapper on UP too Impact: cleanup, no code changed Remove an ugly #ifdef CONFIG_SMP from panic(), by providing an smp_send_stop() wrapper on UP too. LKML-Reference: <49B91A7E.76E4.0078.0@novell.com> Signed-off-by: Ingo Molnar --- include/linux/smp.h | 4 +++- kernel/panic.c | 2 -- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/include/linux/smp.h b/include/linux/smp.h index 2d3bcb6b37f..a69db820eed 100644 --- a/include/linux/smp.h +++ b/include/linux/smp.h @@ -38,7 +38,7 @@ int smp_call_function_single(int cpuid, void (*func) (void *info), void *info, /* * main cross-CPU interfaces, handles INIT, TLB flush, STOP, etc. * (defined in asm header): - */ + */ /* * stops all CPUs but the current one: @@ -122,6 +122,8 @@ extern unsigned int setup_max_cpus; #else /* !SMP */ +static inline void smp_send_stop(void) { } + /* * These macros fold the SMP functionality into a single CPU system */ diff --git a/kernel/panic.c b/kernel/panic.c index 57fb005de54..ca75e819d0e 100644 --- a/kernel/panic.c +++ b/kernel/panic.c @@ -85,14 +85,12 @@ NORET_TYPE void panic(const char * fmt, ...) */ crash_kexec(NULL); -#ifdef CONFIG_SMP /* * Note smp_send_stop is the usual smp shutdown function, which * unfortunately means it may not be hardened to work in a panic * situation. */ smp_send_stop(); -#endif atomic_notifier_call_chain(&panic_notifier_list, 0, buf); From c95dbf27e201587b2a6cf63f8501a0313d9b4801 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 13 Mar 2009 11:14:06 +0100 Subject: [PATCH 8/9] panic: clean up kernel/panic.c Impact: cleanup, no code changed Clean up kernel/panic.c some more and make it more consistent. LKML-Reference: <49B91A7E.76E4.0078.0@novell.com> Signed-off-by: Ingo Molnar --- kernel/panic.c | 115 ++++++++++++++++++++++++++----------------------- 1 file changed, 61 insertions(+), 54 deletions(-) diff --git a/kernel/panic.c b/kernel/panic.c index ca75e819d0e..3fd8c5bf8b3 100644 --- a/kernel/panic.c +++ b/kernel/panic.c @@ -8,19 +8,19 @@ * This function is used through-out the kernel (including mm and fs) * to indicate a major problem. */ -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include #include -#include +#include #include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include #include int panic_on_oops; @@ -52,19 +52,15 @@ EXPORT_SYMBOL(panic_blink); * * This function never returns. */ - NORET_TYPE void panic(const char * fmt, ...) { - long i; static char buf[1024]; va_list args; -#if defined(CONFIG_S390) - unsigned long caller = (unsigned long) __builtin_return_address(0); -#endif + long i; /* - * It's possible to come here directly from a panic-assertion and not - * have preempt disabled. Some functions called from here want + * It's possible to come here directly from a panic-assertion and + * not have preempt disabled. Some functions called from here want * preempt to be disabled. No point enabling it later though... */ preempt_disable(); @@ -99,19 +95,21 @@ NORET_TYPE void panic(const char * fmt, ...) if (panic_timeout > 0) { /* - * Delay timeout seconds before rebooting the machine. - * We can't use the "normal" timers since we just panicked.. - */ - printk(KERN_EMERG "Rebooting in %d seconds..",panic_timeout); + * Delay timeout seconds before rebooting the machine. + * We can't use the "normal" timers since we just panicked. + */ + printk(KERN_EMERG "Rebooting in %d seconds..", panic_timeout); + for (i = 0; i < panic_timeout*1000; ) { touch_nmi_watchdog(); i += panic_blink(i); mdelay(1); i++; } - /* This will not be a clean reboot, with everything - * shutting down. But if there is a chance of - * rebooting the system it will be rebooted. + /* + * This will not be a clean reboot, with everything + * shutting down. But if there is a chance of + * rebooting the system it will be rebooted. */ emergency_restart(); } @@ -124,10 +122,15 @@ NORET_TYPE void panic(const char * fmt, ...) } #endif #if defined(CONFIG_S390) - disabled_wait(caller); + { + unsigned long caller; + + caller = (unsigned long)__builtin_return_address(0); + disabled_wait(caller); + } #endif local_irq_enable(); - for (i = 0;;) { + for (i = 0; ; ) { touch_softlockup_watchdog(); i += panic_blink(i); mdelay(1); @@ -140,23 +143,23 @@ EXPORT_SYMBOL(panic); struct tnt { - u8 bit; - char true; - char false; + u8 bit; + char true; + char false; }; static const struct tnt tnts[] = { - { TAINT_PROPRIETARY_MODULE, 'P', 'G' }, - { TAINT_FORCED_MODULE, 'F', ' ' }, - { TAINT_UNSAFE_SMP, 'S', ' ' }, - { TAINT_FORCED_RMMOD, 'R', ' ' }, - { TAINT_MACHINE_CHECK, 'M', ' ' }, - { TAINT_BAD_PAGE, 'B', ' ' }, - { TAINT_USER, 'U', ' ' }, - { TAINT_DIE, 'D', ' ' }, - { TAINT_OVERRIDDEN_ACPI_TABLE, 'A', ' ' }, - { TAINT_WARN, 'W', ' ' }, - { TAINT_CRAP, 'C', ' ' }, + { TAINT_PROPRIETARY_MODULE, 'P', 'G' }, + { TAINT_FORCED_MODULE, 'F', ' ' }, + { TAINT_UNSAFE_SMP, 'S', ' ' }, + { TAINT_FORCED_RMMOD, 'R', ' ' }, + { TAINT_MACHINE_CHECK, 'M', ' ' }, + { TAINT_BAD_PAGE, 'B', ' ' }, + { TAINT_USER, 'U', ' ' }, + { TAINT_DIE, 'D', ' ' }, + { TAINT_OVERRIDDEN_ACPI_TABLE, 'A', ' ' }, + { TAINT_WARN, 'W', ' ' }, + { TAINT_CRAP, 'C', ' ' }, }; /** @@ -193,7 +196,8 @@ const char *print_tainted(void) *s = 0; } else snprintf(buf, sizeof(buf), "Not tainted"); - return(buf); + + return buf; } int test_taint(unsigned flag) @@ -209,7 +213,8 @@ unsigned long get_taint(void) void add_taint(unsigned flag) { - debug_locks = 0; /* can't trust the integrity of the kernel anymore */ + /* can't trust the integrity of the kernel anymore: */ + debug_locks = 0; set_bit(flag, &tainted_mask); } EXPORT_SYMBOL(add_taint); @@ -264,8 +269,8 @@ static void do_oops_enter_exit(void) } /* - * Return true if the calling CPU is allowed to print oops-related info. This - * is a bit racy.. + * Return true if the calling CPU is allowed to print oops-related info. + * This is a bit racy.. */ int oops_may_print(void) { @@ -274,20 +279,22 @@ int oops_may_print(void) /* * Called when the architecture enters its oops handler, before it prints - * anything. If this is the first CPU to oops, and it's oopsing the first time - * then let it proceed. + * anything. If this is the first CPU to oops, and it's oopsing the first + * time then let it proceed. * - * This is all enabled by the pause_on_oops kernel boot option. We do all this - * to ensure that oopses don't scroll off the screen. It has the side-effect - * of preventing later-oopsing CPUs from mucking up the display, too. + * This is all enabled by the pause_on_oops kernel boot option. We do all + * this to ensure that oopses don't scroll off the screen. It has the + * side-effect of preventing later-oopsing CPUs from mucking up the display, + * too. * - * It turns out that the CPU which is allowed to print ends up pausing for the - * right duration, whereas all the other CPUs pause for twice as long: once in - * oops_enter(), once in oops_exit(). + * It turns out that the CPU which is allowed to print ends up pausing for + * the right duration, whereas all the other CPUs pause for twice as long: + * once in oops_enter(), once in oops_exit(). */ void oops_enter(void) { - debug_locks_off(); /* can't trust the integrity of the kernel anymore */ + /* can't trust the integrity of the kernel anymore: */ + debug_locks_off(); do_oops_enter_exit(); } From 70f454408e68fdba2c2529ab7d6ec3c3525e59f2 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Fri, 20 Mar 2009 09:24:10 +0100 Subject: [PATCH 9/9] s390: remove arch specific smp_send_stop() Impact: build fix on s390 !CONFIG_SMP Remove arch specific smp_send_stop for !CONFIG_SMP since it conflicts with a new generic version. Reported-by: Stephen Rothwell Signed-off-by: Heiko Carstens Cc: Martin Schwidefsky LKML-Reference: <20090320092410.30d2bac3@osiris.boeblingen.de.ibm.com> Signed-off-by: Ingo Molnar --- arch/s390/include/asm/smp.h | 6 ------ 1 file changed, 6 deletions(-) diff --git a/arch/s390/include/asm/smp.h b/arch/s390/include/asm/smp.h index 024b91e0623..f89e2d5c060 100644 --- a/arch/s390/include/asm/smp.h +++ b/arch/s390/include/asm/smp.h @@ -97,12 +97,6 @@ extern void arch_send_call_function_ipi(cpumask_t mask); #endif #ifndef CONFIG_SMP -static inline void smp_send_stop(void) -{ - /* Disable all interrupts/machine checks */ - __load_psw_mask(psw_kernel_bits & ~PSW_MASK_MCHECK); -} - #define hard_smp_processor_id() 0 #define smp_cpu_not_running(cpu) 1 #endif