| Message ID | 20200322060305.70637-8-colyli@suse.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | bcache patches for Linux v5.7-rc1 | expand |
On 3/22/20 7:03 AM, Coly Li wrote: > The idea of this patch is from Davidlohr Bueso, he posts a patch > for bcache to optimize barrier usage for read-modify-write atomic > bitops. Indeed such optimization can also apply on other locations > where smp_mb() is used before or after an atomic operation. > > This patch replaces smp_mb() with smp_mb__before_atomic() or > smp_mb__after_atomic() in btree.c and writeback.c, where it is used > to synchronize memory cache just earlier on other cores. Although > the locations are not on hot code path, it is always not bad to mkae > things a little better. > > Signed-off-by: Coly Li <colyli@suse.de> > Cc: Davidlohr Bueso <dave@stgolabs.net> > --- > drivers/md/bcache/btree.c | 6 +++--- > drivers/md/bcache/writeback.c | 6 +++--- > 2 files changed, 6 insertions(+), 6 deletions(-) > Reviewed-by: Hannes Reinecke <hare@suse.de> Cheers, Hannes
diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c index 74d66b641169..72856e5f23a3 100644 --- a/drivers/md/bcache/btree.c +++ b/drivers/md/bcache/btree.c @@ -1947,7 +1947,7 @@ static int bch_btree_check_thread(void *arg) */ atomic_set(&check_state->enough, 1); /* Update check_state->enough earlier */ - smp_mb(); + smp_mb__after_atomic(); goto out; } skip_nr--; @@ -1972,7 +1972,7 @@ static int bch_btree_check_thread(void *arg) out: info->result = ret; /* update check_state->started among all CPUs */ - smp_mb(); + smp_mb__before_atomic(); if (atomic_dec_and_test(&check_state->started)) wake_up(&check_state->wait); @@ -2031,7 +2031,7 @@ int bch_btree_check(struct cache_set *c) */ for (i = 0; i < check_state->total_threads; i++) { /* fetch latest check_state->enough earlier */ - smp_mb(); + smp_mb__before_atomic(); if (atomic_read(&check_state->enough)) break; diff --git a/drivers/md/bcache/writeback.c b/drivers/md/bcache/writeback.c index 72ba6d015786..3f7641fb28d5 100644 --- a/drivers/md/bcache/writeback.c +++ b/drivers/md/bcache/writeback.c @@ -854,7 +854,7 @@ static int bch_dirty_init_thread(void *arg) else { atomic_set(&state->enough, 1); /* Update state->enough earlier */ - smp_mb(); + smp_mb__after_atomic(); goto out; } skip_nr--; @@ -873,7 +873,7 @@ static int bch_dirty_init_thread(void *arg) out: /* In order to wake up state->wait in time */ - smp_mb(); + smp_mb__before_atomic(); if (atomic_dec_and_test(&state->started)) wake_up(&state->wait); @@ -932,7 +932,7 @@ void bch_sectors_dirty_init(struct bcache_device *d) for (i = 0; i < state->total_threads; i++) { /* Fetch latest state->enough earlier */ - smp_mb(); + smp_mb__before_atomic(); if (atomic_read(&state->enough)) break;
The idea of this patch is from Davidlohr Bueso, he posts a patch for bcache to optimize barrier usage for read-modify-write atomic bitops. Indeed such optimization can also apply on other locations where smp_mb() is used before or after an atomic operation. This patch replaces smp_mb() with smp_mb__before_atomic() or smp_mb__after_atomic() in btree.c and writeback.c, where it is used to synchronize memory cache just earlier on other cores. Although the locations are not on hot code path, it is always not bad to mkae things a little better. Signed-off-by: Coly Li <colyli@suse.de> Cc: Davidlohr Bueso <dave@stgolabs.net> --- drivers/md/bcache/btree.c | 6 +++--- drivers/md/bcache/writeback.c | 6 +++--- 2 files changed, 6 insertions(+), 6 deletions(-)