smp: Avoid using two cache lines for struct call_single_data

struct call_single_data is used in IPIs to transfer information between
CPUs.  Its size is bigger than sizeof(unsigned long) and less than
cache line size.  Currently it is not allocated with any explicit alignment
requirements.  This makes it possible for allocated call_single_data to
cross two cache lines, which results in double the number of the cache lines
that need to be transferred among CPUs.

This can be fixed by requiring call_single_data to be aligned with the
size of call_single_data. Currently the size of call_single_data is the
power of 2.  If we add new fields to call_single_data, we may need to
add padding to make sure the size of new definition is the power of 2
as well.

Fortunately, this is enforced by GCC, which will report bad sizes.

To set alignment requirements of call_single_data to the size of
call_single_data, a struct definition and a typedef is used.

To test the effect of the patch, I used the vm-scalability multiple
thread swap test case (swap-w-seq-mt).  The test will create multiple
threads and each thread will eat memory until all RAM and part of swap
is used, so that huge number of IPIs are triggered when unmapping
memory.  In the test, the throughput of memory writing improves ~5%
compared with misaligned call_single_data, because of faster IPIs.

Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Huang, Ying <ying.huang@intel.com>
[ Add call_single_data_t and align with size of call_single_data. ]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/87bmnqd6lz.fsf@yhuang-mobile.sh.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

3 files changed, 19 insertions, 17 deletions

diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index eeef1a3086d1..f29a7d2b57e1 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -769,7 +769,7 @@ struct rq {
 #ifdef CONFIG_SCHED_HRTICK
 #ifdef CONFIG_SMP
 	int hrtick_csd_pending;
-	struct call_single_data hrtick_csd;
+	call_single_data_t hrtick_csd;
 #endif
 	struct hrtimer hrtick_timer;
 #endif
diff --git a/kernel/smp.c b/kernel/smp.c
index 3061483cb3ad..81cfca9b4cc3 100644
--- a/kernel/smp.c
+++ b/kernel/smp.c
@@ -28,7 +28,7 @@ enum {
 };
 
 struct call_function_data {
-	struct call_single_data	__percpu *csd;
+	call_single_data_t	__percpu *csd;
 	cpumask_var_t		cpumask;
 	cpumask_var_t		cpumask_ipi;
 };
@@ -51,7 +51,7 @@ int smpcfd_prepare_cpu(unsigned int cpu)
 		free_cpumask_var(cfd->cpumask);
 		return -ENOMEM;
 	}
-	cfd->csd = alloc_percpu(struct call_single_data);
+	cfd->csd = alloc_percpu(call_single_data_t);
 	if (!cfd->csd) {
 		free_cpumask_var(cfd->cpumask);
 		free_cpumask_var(cfd->cpumask_ipi);
@@ -103,12 +103,12 @@ void __init call_function_init(void)
  * 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 __always_inline void csd_lock_wait(struct call_single_data *csd)
+static __always_inline void csd_lock_wait(call_single_data_t *csd)
 {
 	smp_cond_load_acquire(&csd->flags, !(VAL & CSD_FLAG_LOCK));
 }
 
-static __always_inline void csd_lock(struct call_single_data *csd)
+static __always_inline void csd_lock(call_single_data_t *csd)
 {
 	csd_lock_wait(csd);
 	csd->flags |= CSD_FLAG_LOCK;
@@ -116,12 +116,12 @@ static __always_inline void csd_lock(struct call_single_data *csd)
 	/*
 	 * prevent CPU from reordering the above assignment
 	 * to ->flags with any subsequent assignments to other
-	 * fields of the specified call_single_data structure:
+	 * fields of the specified call_single_data_t structure:
 	 */
 	smp_wmb();
 }
 
-static __always_inline void csd_unlock(struct call_single_data *csd)
+static __always_inline void csd_unlock(call_single_data_t *csd)
 {
 	WARN_ON(!(csd->flags & CSD_FLAG_LOCK));
 
@@ -131,14 +131,14 @@ static __always_inline void csd_unlock(struct call_single_data *csd)
 	smp_store_release(&csd->flags, 0);
 }
 
-static DEFINE_PER_CPU_SHARED_ALIGNED(struct call_single_data, csd_data);
+static DEFINE_PER_CPU_SHARED_ALIGNED(call_single_data_t, csd_data);
 
 /*
- * Insert a previously allocated call_single_data element
+ * Insert a previously allocated call_single_data_t element
  * for execution on the given CPU. data must already have
  * ->func, ->info, and ->flags set.
  */
-static int generic_exec_single(int cpu, struct call_single_data *csd,
+static int generic_exec_single(int cpu, call_single_data_t *csd,
 			       smp_call_func_t func, void *info)
 {
 	if (cpu == smp_processor_id()) {
@@ -210,7 +210,7 @@ static void flush_smp_call_function_queue(bool warn_cpu_offline)
 {
 	struct llist_head *head;
 	struct llist_node *entry;
-	struct call_single_data *csd, *csd_next;
+	call_single_data_t *csd, *csd_next;
 	static bool warned;
 
 	WARN_ON(!irqs_disabled());
@@ -268,8 +268,10 @@ static void flush_smp_call_function_queue(bool warn_cpu_offline)
 int smp_call_function_single(int cpu, smp_call_func_t func, void *info,
 			     int wait)
 {
-	struct call_single_data *csd;
-	struct call_single_data csd_stack = { .flags = CSD_FLAG_LOCK | CSD_FLAG_SYNCHRONOUS };
+	call_single_data_t *csd;
+	call_single_data_t csd_stack = {
+		.flags = CSD_FLAG_LOCK | CSD_FLAG_SYNCHRONOUS,
+	};
 	int this_cpu;
 	int err;
 
@@ -321,7 +323,7 @@ EXPORT_SYMBOL(smp_call_function_single);
  * NOTE: Be careful, there is unfortunately no current debugging facility to
  * validate the correctness of this serialization.
  */
-int smp_call_function_single_async(int cpu, struct call_single_data *csd)
+int smp_call_function_single_async(int cpu, call_single_data_t *csd)
 {
 	int err = 0;
 
@@ -444,7 +446,7 @@ void smp_call_function_many(const struct cpumask *mask,
 
 	cpumask_clear(cfd->cpumask_ipi);
 	for_each_cpu(cpu, cfd->cpumask) {
-		struct call_single_data *csd = per_cpu_ptr(cfd->csd, cpu);
+		call_single_data_t *csd = per_cpu_ptr(cfd->csd, cpu);
 
 		csd_lock(csd);
 		if (wait)
@@ -460,7 +462,7 @@ void smp_call_function_many(const struct cpumask *mask,
 
 	if (wait) {
 		for_each_cpu(cpu, cfd->cpumask) {
-			struct call_single_data *csd;
+			call_single_data_t *csd;
 
 			csd = per_cpu_ptr(cfd->csd, cpu);
 			csd_lock_wait(csd);
diff --git a/kernel/up.c b/kernel/up.c
index ee81ac9af4ca..42c46bf3e0a5 100644
--- a/kernel/up.c
+++ b/kernel/up.c
@@ -23,7 +23,7 @@ int smp_call_function_single(int cpu, void (*func) (void *info), void *info,
 }
 EXPORT_SYMBOL(smp_call_function_single);
 
-int smp_call_function_single_async(int cpu, struct call_single_data *csd)
+int smp_call_function_single_async(int cpu, call_single_data_t *csd)
 {
 	unsigned long flags;