From 12938a9220a38d555e38dc9b40021e664b99a1f1 Mon Sep 17 00:00:00 2001 From: Christoph Lameter Date: Mon, 6 Dec 2010 11:16:20 -0600 Subject: vmstat: Optimize zone counter modifications through the use of this cpu operations this cpu operations can be used to slightly optimize the function. The changes will avoid some address calculations and replace them with the use of the percpu segment register. If one would have this_cpu_inc_return and this_cpu_dec_return then it would be possible to optimize inc_zone_page_state and dec_zone_page_state even more. V1->V2: - Fix __dec_zone_state overflow handling - Use s8 variables for temporary storage. V2->V3: - Put __percpu annotations in correct places. Reviewed-by: Pekka Enberg Acked-by: H. Peter Anvin Signed-off-by: Christoph Lameter Signed-off-by: Tejun Heo --- mm/vmstat.c | 48 ++++++++++++++++++++++++++++-------------------- 1 file changed, 28 insertions(+), 20 deletions(-) (limited to 'mm/vmstat.c') diff --git a/mm/vmstat.c b/mm/vmstat.c index 8f62f17ee1c..3ad909d9600 100644 --- a/mm/vmstat.c +++ b/mm/vmstat.c @@ -167,18 +167,20 @@ static void refresh_zone_stat_thresholds(void) void __mod_zone_page_state(struct zone *zone, enum zone_stat_item item, int delta) { - struct per_cpu_pageset *pcp = this_cpu_ptr(zone->pageset); - - s8 *p = pcp->vm_stat_diff + item; + struct per_cpu_pageset __percpu *pcp = zone->pageset; + s8 __percpu *p = pcp->vm_stat_diff + item; long x; + long t; + + x = delta + __this_cpu_read(*p); - x = delta + *p; + t = __this_cpu_read(pcp->stat_threshold); - if (unlikely(x > pcp->stat_threshold || x < -pcp->stat_threshold)) { + if (unlikely(x > t || x < -t)) { zone_page_state_add(x, zone, item); x = 0; } - *p = x; + __this_cpu_write(*p, x); } EXPORT_SYMBOL(__mod_zone_page_state); @@ -221,16 +223,19 @@ EXPORT_SYMBOL(mod_zone_page_state); */ void __inc_zone_state(struct zone *zone, enum zone_stat_item item) { - struct per_cpu_pageset *pcp = this_cpu_ptr(zone->pageset); - s8 *p = pcp->vm_stat_diff + item; + struct per_cpu_pageset __percpu *pcp = zone->pageset; + s8 __percpu *p = pcp->vm_stat_diff + item; + s8 v, t; - (*p)++; + __this_cpu_inc(*p); - if (unlikely(*p > pcp->stat_threshold)) { - int overstep = pcp->stat_threshold / 2; + v = __this_cpu_read(*p); + t = __this_cpu_read(pcp->stat_threshold); + if (unlikely(v > t)) { + s8 overstep = t >> 1; - zone_page_state_add(*p + overstep, zone, item); - *p = -overstep; + zone_page_state_add(v + overstep, zone, item); + __this_cpu_write(*p, -overstep); } } @@ -242,16 +247,19 @@ EXPORT_SYMBOL(__inc_zone_page_state); void __dec_zone_state(struct zone *zone, enum zone_stat_item item) { - struct per_cpu_pageset *pcp = this_cpu_ptr(zone->pageset); - s8 *p = pcp->vm_stat_diff + item; + struct per_cpu_pageset __percpu *pcp = zone->pageset; + s8 __percpu *p = pcp->vm_stat_diff + item; + s8 v, t; - (*p)--; + __this_cpu_dec(*p); - if (unlikely(*p < - pcp->stat_threshold)) { - int overstep = pcp->stat_threshold / 2; + v = __this_cpu_read(*p); + t = __this_cpu_read(pcp->stat_threshold); + if (unlikely(v < - t)) { + s8 overstep = t >> 1; - zone_page_state_add(*p - overstep, zone, item); - *p = overstep; + zone_page_state_add(v - overstep, zone, item); + __this_cpu_write(*p, overstep); } } -- cgit v1.2.3 From 908ee0f122bf2a67414854af5b90c6621d186a71 Mon Sep 17 00:00:00 2001 From: Christoph Lameter Date: Mon, 6 Dec 2010 11:40:02 -0600 Subject: vmstat: Use this_cpu_inc_return for vm statistics this_cpu_inc_return() saves us a memory access there. Code size does not change. V1->V2: - Fixed the location of the __per_cpu pointer attributes - Sparse checked V2->V3: - Move fixes to __percpu attribute usage to earlier patch Reviewed-by: Pekka Enberg Acked-by: H. Peter Anvin Signed-off-by: Christoph Lameter Signed-off-by: Tejun Heo --- mm/vmstat.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) (limited to 'mm/vmstat.c') diff --git a/mm/vmstat.c b/mm/vmstat.c index 3ad909d9600..f9a7bc89fd1 100644 --- a/mm/vmstat.c +++ b/mm/vmstat.c @@ -227,9 +227,7 @@ void __inc_zone_state(struct zone *zone, enum zone_stat_item item) s8 __percpu *p = pcp->vm_stat_diff + item; s8 v, t; - __this_cpu_inc(*p); - - v = __this_cpu_read(*p); + v = __this_cpu_inc_return(*p); t = __this_cpu_read(pcp->stat_threshold); if (unlikely(v > t)) { s8 overstep = t >> 1; @@ -251,9 +249,7 @@ void __dec_zone_state(struct zone *zone, enum zone_stat_item item) s8 __percpu *p = pcp->vm_stat_diff + item; s8 v, t; - __this_cpu_dec(*p); - - v = __this_cpu_read(*p); + v = __this_cpu_dec_return(*p); t = __this_cpu_read(pcp->stat_threshold); if (unlikely(v < - t)) { s8 overstep = t >> 1; -- cgit v1.2.3 From 7c83912062c801738d7d19acaf8f7fec25ea663c Mon Sep 17 00:00:00 2001 From: Christoph Lameter Date: Tue, 14 Dec 2010 10:28:46 -0600 Subject: vmstat: User per cpu atomics to avoid interrupt disable / enable Currently the operations to increment vm counters must disable interrupts in order to not mess up their housekeeping of counters. So use this_cpu_cmpxchg() to avoid the overhead. Since we can no longer count on preremption being disabled we still have some minor issues. The fetching of the counter thresholds is racy. A threshold from another cpu may be applied if we happen to be rescheduled on another cpu. However, the following vmstat operation will then bring the counter again under the threshold limit. The operations for __xxx_zone_state are not changed since the caller has taken care of the synchronization needs (and therefore the cycle count is even less than the optimized version for the irq disable case provided here). The optimization using this_cpu_cmpxchg will only be used if the arch supports efficient this_cpu_ops (must have CONFIG_CMPXCHG_LOCAL set!) The use of this_cpu_cmpxchg reduces the cycle count for the counter operations by %80 (inc_zone_page_state goes from 170 cycles to 32). Signed-off-by: Christoph Lameter --- mm/vmstat.c | 101 +++++++++++++++++++++++++++++++++++++++++++++++++++--------- 1 file changed, 87 insertions(+), 14 deletions(-) (limited to 'mm/vmstat.c') diff --git a/mm/vmstat.c b/mm/vmstat.c index f9a7bc89fd1..7329eb8a08a 100644 --- a/mm/vmstat.c +++ b/mm/vmstat.c @@ -184,20 +184,6 @@ void __mod_zone_page_state(struct zone *zone, enum zone_stat_item item, } EXPORT_SYMBOL(__mod_zone_page_state); -/* - * For an unknown interrupt state - */ -void mod_zone_page_state(struct zone *zone, enum zone_stat_item item, - int delta) -{ - unsigned long flags; - - local_irq_save(flags); - __mod_zone_page_state(zone, item, delta); - local_irq_restore(flags); -} -EXPORT_SYMBOL(mod_zone_page_state); - /* * Optimized increment and decrement functions. * @@ -265,6 +251,92 @@ void __dec_zone_page_state(struct page *page, enum zone_stat_item item) } EXPORT_SYMBOL(__dec_zone_page_state); +#ifdef CONFIG_CMPXCHG_LOCAL +/* + * If we have cmpxchg_local support then we do not need to incur the overhead + * that comes with local_irq_save/restore if we use this_cpu_cmpxchg. + * + * mod_state() modifies the zone counter state through atomic per cpu + * operations. + * + * Overstep mode specifies how overstep should handled: + * 0 No overstepping + * 1 Overstepping half of threshold + * -1 Overstepping minus half of threshold +*/ +static inline void mod_state(struct zone *zone, + enum zone_stat_item item, int delta, int overstep_mode) +{ + struct per_cpu_pageset __percpu *pcp = zone->pageset; + s8 __percpu *p = pcp->vm_stat_diff + item; + long o, n, t, z; + + do { + z = 0; /* overflow to zone counters */ + + /* + * The fetching of the stat_threshold is racy. We may apply + * a counter threshold to the wrong the cpu if we get + * rescheduled while executing here. However, the following + * will apply the threshold again and therefore bring the + * counter under the threshold. + */ + t = this_cpu_read(pcp->stat_threshold); + + o = this_cpu_read(*p); + n = delta + o; + + if (n > t || n < -t) { + int os = overstep_mode * (t >> 1) ; + + /* Overflow must be added to zone counters */ + z = n + os; + n = -os; + } + } while (this_cpu_cmpxchg(*p, o, n) != o); + + if (z) + zone_page_state_add(z, zone, item); +} + +void mod_zone_page_state(struct zone *zone, enum zone_stat_item item, + int delta) +{ + mod_state(zone, item, delta, 0); +} +EXPORT_SYMBOL(mod_zone_page_state); + +void inc_zone_state(struct zone *zone, enum zone_stat_item item) +{ + mod_state(zone, item, 1, 1); +} + +void inc_zone_page_state(struct page *page, enum zone_stat_item item) +{ + mod_state(page_zone(page), item, 1, 1); +} +EXPORT_SYMBOL(inc_zone_page_state); + +void dec_zone_page_state(struct page *page, enum zone_stat_item item) +{ + mod_state(page_zone(page), item, -1, -1); +} +EXPORT_SYMBOL(dec_zone_page_state); +#else +/* + * Use interrupt disable to serialize counter updates + */ +void mod_zone_page_state(struct zone *zone, enum zone_stat_item item, + int delta) +{ + unsigned long flags; + + local_irq_save(flags); + __mod_zone_page_state(zone, item, delta); + local_irq_restore(flags); +} +EXPORT_SYMBOL(mod_zone_page_state); + void inc_zone_state(struct zone *zone, enum zone_stat_item item) { unsigned long flags; @@ -295,6 +367,7 @@ void dec_zone_page_state(struct page *page, enum zone_stat_item item) local_irq_restore(flags); } EXPORT_SYMBOL(dec_zone_page_state); +#endif /* * Update the zone counters for one cpu. -- cgit v1.2.3