From patchwork Mon Oct 20 17:03:54 2008 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mike Travis X-Patchwork-Id: 5104 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from ozlabs.org (localhost [127.0.0.1]) by ozlabs.org (Postfix) with ESMTP id 28C7648E49 for ; Tue, 21 Oct 2008 04:12:32 +1100 (EST) X-Original-To: linuxppc-dev@ozlabs.org Delivered-To: linuxppc-dev@ozlabs.org Received: from relay.sgi.com (relay1.sgi.com [192.48.171.29]) by ozlabs.org (Postfix) with ESMTP id 9B101DDF10 for ; Tue, 21 Oct 2008 04:03:26 +1100 (EST) Received: from polaris-admin.engr.sgi.com (polaris-admin.engr.sgi.com [192.102.97.111]) by relay1.corp.sgi.com (Postfix) with ESMTP id 890488F8129; Mon, 20 Oct 2008 10:03:24 -0700 (PDT) Received: by polaris-admin.engr.sgi.com (Postfix, from userid 5508) id 831CD517A548; Mon, 20 Oct 2008 10:03:24 -0700 (PDT) Message-Id: <20081020170324.412681000@polaris-admin.engr.sgi.com> References: <20081020170319.539427000@polaris-admin.engr.sgi.com> User-Agent: quilt/0.46-1 Date: Mon, 20 Oct 2008 10:03:54 -0700 From: Mike Travis To: Ingo Molnar , Rusty Russell Subject: [PATCH 35/35] cpumask: smp_call_function_many() Content-Disposition: inline; filename=cpumask:smp_call_function_many.patch Cc: davej@codemonkey.org.uk, Jeremy Fitzhardinge , Jes Sorensen , IA64 , S390 , peterz@infradead.org, Jack Steiner , linux-kernel@vger.kernel.org, Eric Dumazet , PowerPC , Andi Kleen , Thomas Gleixner , Yinghai Lu , "H. Peter Anvin" , SPARC , Andrew Morton , David Miller X-BeenThere: linuxppc-dev@ozlabs.org X-Mailman-Version: 2.1.11 Precedence: list List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: linuxppc-dev-bounces+patchwork-incoming=ozlabs.org@ozlabs.org Errors-To: linuxppc-dev-bounces+patchwork-incoming=ozlabs.org@ozlabs.org Transition from cpumask_t-taking smp_call_function_mask() to a new smp_call_function_many() which takes a struct cpumask *. (Naming is inspired by smp_call_function_single). Note that the new one returns void: the old one couldn't fail either unless there was a logic bug. From: Rusty Russell Signed-off-by: Rusty Russell Signed-off-by: Mike Travis --- arch/s390/include/asm/smp.h | 3 - arch/s390/kernel/smp.c | 30 +++++++++++-------- include/linux/smp.h | 13 +++++++- kernel/smp.c | 68 +++++++++++++++++++++++--------------------- 4 files changed, 64 insertions(+), 50 deletions(-) --- test-compile.orig/arch/s390/include/asm/smp.h +++ test-compile/arch/s390/include/asm/smp.h @@ -90,9 +90,6 @@ extern int __cpu_up (unsigned int cpu); extern struct mutex smp_cpu_state_mutex; extern int smp_cpu_polarization[]; - -extern int smp_call_function_mask(cpumask_t mask, void (*func)(void *), - void *info, int wait); #endif #ifndef CONFIG_SMP --- test-compile.orig/arch/s390/kernel/smp.c +++ test-compile/arch/s390/kernel/smp.c @@ -103,7 +103,7 @@ static void do_call_function(void) } static void __smp_call_function_map(void (*func) (void *info), void *info, - int wait, cpumask_t map) + int wait, struct cpumask *map) { struct call_data_struct data; int cpu, local = 0; @@ -163,14 +163,16 @@ out: * You must not call this function with disabled interrupts, from a * hardware interrupt handler or from a bottom half. */ + +/* protected by call_lock */ +static DEFINE_BITMAP(smp_call_map, CONFIG_NR_CPUS); + int smp_call_function(void (*func) (void *info), void *info, int wait) { - cpumask_t map; - spin_lock(&call_lock); - map = cpu_online_map; - cpu_clear(smp_processor_id(), map); - __smp_call_function_map(func, info, wait, map); + cpumask_copy(to_cpumask(smp_call_map), cpu_online_mask); + cpumask_clear_cpu(smp_processor_id(), to_cpumask(smp_call_map)); + __smp_call_function_map(func, info, wait, to_cpumask(smp_call_map)); spin_unlock(&call_lock); return 0; } @@ -192,14 +194,15 @@ int smp_call_function_single(int cpu, vo int wait) { spin_lock(&call_lock); - __smp_call_function_map(func, info, wait, cpumask_of_cpu(cpu)); + cpumask_copy(to_cpumask(smp_call_map), cpumask_of(cpu)); + __smp_call_function_map(func, info, wait, cpumask_of(cpu)); spin_unlock(&call_lock); return 0; } EXPORT_SYMBOL(smp_call_function_single); /** - * smp_call_function_mask(): Run a function on a set of other CPUs. + * smp_call_function_many(): Run a function on a set of other CPUs. * @mask: The set of cpus to run on. Must not include the current cpu. * @func: The function to run. This must be fast and non-blocking. * @info: An arbitrary pointer to pass to the function. @@ -213,16 +216,17 @@ EXPORT_SYMBOL(smp_call_function_single); * You must not call this function with disabled interrupts or from a * hardware interrupt handler or from a bottom half handler. */ -int smp_call_function_mask(cpumask_t mask, void (*func)(void *), void *info, - int wait) +int smp_call_function_many(const struct cpumask *mask, + void (*func)(void *), void *info, bool wait) { spin_lock(&call_lock); - cpu_clear(smp_processor_id(), mask); - __smp_call_function_map(func, info, wait, mask); + cpumask_copy(to_cpumask(smp_call_map), cpu_online_mask); + cpumask_clear_cpu(smp_processor_id(), to_cpumask(smp_call_map)); + __smp_call_function_map(func, info, wait, to_cpumask(smp_call_map)); spin_unlock(&call_lock); return 0; } -EXPORT_SYMBOL(smp_call_function_mask); +EXPORT_SYMBOL(smp_call_function_many); void smp_send_stop(void) { --- test-compile.orig/include/linux/smp.h +++ test-compile/include/linux/smp.h @@ -62,12 +62,21 @@ extern void smp_cpus_done(unsigned int m * Call a function on all other processors */ int smp_call_function(void(*func)(void *info), void *info, int wait); -int smp_call_function_mask(cpumask_t mask, void(*func)(void *info), void *info, - int wait); +void smp_call_function_many(const struct cpumask *mask, + void(*func)(void *info), void *info, bool wait); int smp_call_function_single(int cpuid, void (*func) (void *info), void *info, int wait); void __smp_call_function_single(int cpuid, struct call_single_data *data); +/* Use smp_call_function_many, which takes a pointer to the mask. */ +static inline int __deprecated +smp_call_function_mask(cpumask_t mask, void(*func)(void *info), void *info, + int wait) +{ + smp_call_function_many(&mask, func, info, wait); + return 0; +} + /* * Generic and arch helpers */ --- test-compile.orig/kernel/smp.c +++ test-compile/kernel/smp.c @@ -24,7 +24,7 @@ struct call_function_data { struct call_single_data csd; spinlock_t lock; unsigned int refs; - cpumask_t cpumask; + struct cpumask *cpumask; struct rcu_head rcu_head; }; @@ -109,13 +109,13 @@ void generic_smp_call_function_interrupt list_for_each_entry_rcu(data, &call_function_queue, csd.list) { int refs; - if (!cpu_isset(cpu, data->cpumask)) + if (!cpumask_test_cpu(cpu, data->cpumask)) continue; data->csd.func(data->csd.info); spin_lock(&data->lock); - cpu_clear(cpu, data->cpumask); + cpumask_clear_cpu(cpu, data->cpumask); WARN_ON(data->refs == 0); data->refs--; refs = data->refs; @@ -273,7 +273,7 @@ static void quiesce_dummy(void *unused) /* * Ensure stack based data used in call function mask is safe to free. * - * This is needed by smp_call_function_mask when using on-stack data, because + * This is needed by smp_call_function_many when using on-stack data, because * a single call function queue is shared by all CPUs, and any CPU may pick up * the data item on the queue at any time before it is deleted. So we need to * ensure that all CPUs have transitioned through a quiescent state after @@ -287,7 +287,7 @@ static void quiesce_dummy(void *unused) * If a faster scheme can be made, we could go back to preferring stack based * data -- the data allocation/free is non-zero cost. */ -static void smp_call_function_mask_quiesce_stack(cpumask_t mask) +static void smp_call_function_mask_quiesce_stack(const struct cpumask *mask) { struct call_single_data data; int cpu; @@ -295,21 +295,19 @@ static void smp_call_function_mask_quies data.func = quiesce_dummy; data.info = NULL; - for_each_cpu_mask(cpu, mask) { + for_each_cpu(cpu, mask) { data.flags = CSD_FLAG_WAIT; generic_exec_single(cpu, &data); } } /** - * smp_call_function_mask(): Run a function on a set of other CPUs. - * @mask: The set of cpus to run on. + * smp_call_function_many(): Run a function on a set of other CPUs. + * @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. * - * Returns 0 on success, else a negative status code. - * * If @wait is true, then returns once @func has returned. Note that @wait * will be implicitly turned on in case of allocation failures, since * we fall back to on-stack allocation. @@ -318,12 +316,13 @@ static void smp_call_function_mask_quies * hardware interrupt handler or from a bottom half handler. Preemption * must be disabled when calling this function. */ -int smp_call_function_mask(cpumask_t mask, void (*func)(void *), void *info, - int wait) +void smp_call_function_many(const struct cpumask *mask, + void (*func)(void *), void *info, + bool wait) { struct call_function_data d; struct call_function_data *data = NULL; - cpumask_t allbutself; + cpumask_var_t allbutself; unsigned long flags; int cpu, num_cpus; int slowpath = 0; @@ -331,21 +330,28 @@ int smp_call_function_mask(cpumask_t mas /* Can deadlock when called with interrupts disabled */ WARN_ON(irqs_disabled()); - cpu = smp_processor_id(); - allbutself = cpu_online_map; - cpu_clear(cpu, allbutself); - cpus_and(mask, mask, allbutself); - num_cpus = cpus_weight(mask); + if (!alloc_cpumask_var(&allbutself, GFP_ATOMIC)) { + /* Slow path. */ + for_each_online_cpu(cpu) { + if (cpumask_test_cpu(cpu, mask)) + smp_call_function_single(cpu, func, info, wait); + } + return; + } + cpumask_and(allbutself, cpu_online_mask, mask); + cpumask_clear_cpu(smp_processor_id(), allbutself); + num_cpus = cpumask_weight(allbutself); /* * If zero CPUs, return. If just a single CPU, turn this request * into a targetted single call instead since it's faster. */ if (!num_cpus) - return 0; + return; else if (num_cpus == 1) { - cpu = first_cpu(mask); - return smp_call_function_single(cpu, func, info, wait); + cpu = cpumask_first(allbutself); + smp_call_function_single(cpu, func, info, wait); + goto out; } data = kmalloc(sizeof(*data), GFP_ATOMIC); @@ -364,25 +370,25 @@ int smp_call_function_mask(cpumask_t mas data->csd.func = func; data->csd.info = info; data->refs = num_cpus; - data->cpumask = mask; + data->cpumask = allbutself; spin_lock_irqsave(&call_function_lock, flags); list_add_tail_rcu(&data->csd.list, &call_function_queue); spin_unlock_irqrestore(&call_function_lock, flags); /* Send a message to all CPUs in the map */ - arch_send_call_function_ipi(mask); + arch_send_call_function_ipi((cpumask_t)*allbutself); /* optionally wait for the CPUs to complete */ if (wait) { csd_flag_wait(&data->csd); if (unlikely(slowpath)) - smp_call_function_mask_quiesce_stack(mask); + smp_call_function_mask_quiesce_stack(allbutself); } - - return 0; +out: + free_cpumask_var(allbutself); } -EXPORT_SYMBOL(smp_call_function_mask); +EXPORT_SYMBOL(smp_call_function_many); /** * smp_call_function(): Run a function on all other CPUs. @@ -390,7 +396,7 @@ EXPORT_SYMBOL(smp_call_function_mask); * @info: An arbitrary pointer to pass to the function. * @wait: If true, wait (atomically) until function has completed on other CPUs. * - * Returns 0 on success, else a negative status code. + * Returns 0. * * If @wait is true, then returns once @func has returned; otherwise * it returns just before the target cpu calls @func. In case of allocation @@ -401,12 +407,10 @@ EXPORT_SYMBOL(smp_call_function_mask); */ int smp_call_function(void (*func)(void *), void *info, int wait) { - int ret; - preempt_disable(); - ret = smp_call_function_mask(cpu_online_map, func, info, wait); + smp_call_function_many(cpu_online_mask, func, info, wait); preempt_enable(); - return ret; + return 0; } EXPORT_SYMBOL(smp_call_function);