From c7b92516a9c68fa5403879225a5a19974a801ef6 Mon Sep 17 00:00:00 2001 From: Christoph Lameter Date: Mon, 6 Dec 2010 11:16:28 -0600 Subject: fs: Use this_cpu_xx operations in buffer.c Optimize various per cpu area operations through these new percpu operations. These operations avoid address calculations through the use of segment prefixes and multiple memory references through RMW instructions etc. Reduces code size: Before: christoph@linux-2.6$ size fs/buffer.o text data bss dec hex filename 19169 80 28 19277 4b4d fs/buffer.o After: christoph@linux-2.6$ size fs/buffer.o text data bss dec hex filename 19138 80 28 19246 4b2e fs/buffer.o V3->V4: - Move the use of this_cpu_inc_return into a later patch so that this one can go in without percpu infrastructure changes. Cc: Wu Fengguang Cc: Christoph Hellwig Acked-by: H. Peter Anvin Signed-off-by: Christoph Lameter Signed-off-by: Tejun Heo --- fs/buffer.c | 35 +++++++++++++++++------------------ 1 file changed, 17 insertions(+), 18 deletions(-) (limited to 'fs') diff --git a/fs/buffer.c b/fs/buffer.c index 5930e382959b..137d9de00e24 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -1270,12 +1270,10 @@ static inline void check_irqs_on(void) static void bh_lru_install(struct buffer_head *bh) { struct buffer_head *evictee = NULL; - struct bh_lru *lru; check_irqs_on(); bh_lru_lock(); - lru = &__get_cpu_var(bh_lrus); - if (lru->bhs[0] != bh) { + if (__this_cpu_read(bh_lrus.bhs[0]) != bh) { struct buffer_head *bhs[BH_LRU_SIZE]; int in; int out = 0; @@ -1283,7 +1281,8 @@ static void bh_lru_install(struct buffer_head *bh) get_bh(bh); bhs[out++] = bh; for (in = 0; in < BH_LRU_SIZE; in++) { - struct buffer_head *bh2 = lru->bhs[in]; + struct buffer_head *bh2 = + __this_cpu_read(bh_lrus.bhs[in]); if (bh2 == bh) { __brelse(bh2); @@ -1298,7 +1297,7 @@ static void bh_lru_install(struct buffer_head *bh) } while (out < BH_LRU_SIZE) bhs[out++] = NULL; - memcpy(lru->bhs, bhs, sizeof(bhs)); + memcpy(__this_cpu_ptr(&bh_lrus.bhs), bhs, sizeof(bhs)); } bh_lru_unlock(); @@ -1313,23 +1312,22 @@ static struct buffer_head * lookup_bh_lru(struct block_device *bdev, sector_t block, unsigned size) { struct buffer_head *ret = NULL; - struct bh_lru *lru; unsigned int i; check_irqs_on(); bh_lru_lock(); - lru = &__get_cpu_var(bh_lrus); for (i = 0; i < BH_LRU_SIZE; i++) { - struct buffer_head *bh = lru->bhs[i]; + struct buffer_head *bh = __this_cpu_read(bh_lrus.bhs[i]); if (bh && bh->b_bdev == bdev && bh->b_blocknr == block && bh->b_size == size) { if (i) { while (i) { - lru->bhs[i] = lru->bhs[i - 1]; + __this_cpu_write(bh_lrus.bhs[i], + __this_cpu_read(bh_lrus.bhs[i - 1])); i--; } - lru->bhs[0] = bh; + __this_cpu_write(bh_lrus.bhs[0], bh); } get_bh(bh); ret = bh; @@ -3205,20 +3203,21 @@ static void recalc_bh_state(void) if (__get_cpu_var(bh_accounting).ratelimit++ < 4096) return; - __get_cpu_var(bh_accounting).ratelimit = 0; + __this_cpu_write(bh_accounting.ratelimit, 0); for_each_online_cpu(i) tot += per_cpu(bh_accounting, i).nr; buffer_heads_over_limit = (tot > max_buffer_heads); } - + struct buffer_head *alloc_buffer_head(gfp_t gfp_flags) { struct buffer_head *ret = kmem_cache_zalloc(bh_cachep, gfp_flags); if (ret) { INIT_LIST_HEAD(&ret->b_assoc_buffers); - get_cpu_var(bh_accounting).nr++; + preempt_disable(); + __this_cpu_inc(bh_accounting.nr); recalc_bh_state(); - put_cpu_var(bh_accounting); + preempt_enable(); } return ret; } @@ -3228,9 +3227,10 @@ void free_buffer_head(struct buffer_head *bh) { BUG_ON(!list_empty(&bh->b_assoc_buffers)); kmem_cache_free(bh_cachep, bh); - get_cpu_var(bh_accounting).nr--; + preempt_disable(); + __this_cpu_dec(bh_accounting.nr); recalc_bh_state(); - put_cpu_var(bh_accounting); + preempt_enable(); } EXPORT_SYMBOL(free_buffer_head); @@ -3243,9 +3243,8 @@ static void buffer_exit_cpu(int cpu) brelse(b->bhs[i]); b->bhs[i] = NULL; } - get_cpu_var(bh_accounting).nr += per_cpu(bh_accounting, cpu).nr; + this_cpu_add(bh_accounting.nr, per_cpu(bh_accounting, cpu).nr); per_cpu(bh_accounting, cpu).nr = 0; - put_cpu_var(bh_accounting); } static int buffer_cpu_notify(struct notifier_block *self, -- cgit v1.2.3-59-g8ed1b From ee1be8626355e6a1f3f8c44e2351ff2661c5998d Mon Sep 17 00:00:00 2001 From: Christoph Lameter Date: Mon, 6 Dec 2010 11:40:05 -0600 Subject: fs: Use this_cpu_inc_return in buffer.c __this_cpu_inc can create a single instruction with the same effect as the _get_cpu_var(..)++ construct in buffer.c. Cc: Wu Fengguang Cc: Christoph Hellwig Acked-by: H. Peter Anvin Signed-off-by: Christoph Lameter Signed-off-by: Tejun Heo --- fs/buffer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/buffer.c b/fs/buffer.c index 137d9de00e24..2219a76e2caf 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -3201,7 +3201,7 @@ static void recalc_bh_state(void) int i; int tot = 0; - if (__get_cpu_var(bh_accounting).ratelimit++ < 4096) + if (__this_cpu_inc_return(bh_accounting.ratelimit) - 1 < 4096) return; __this_cpu_write(bh_accounting.ratelimit, 0); for_each_online_cpu(i) -- cgit v1.2.3-59-g8ed1b