From: Coly Li Date: Tue, 24 May 2022 10:23:34 +0000 (+0800) Subject: bcache: improve multithreaded bch_sectors_dirty_init() X-Git-Tag: microblaze-v5.20~52^2~14 X-Git-Url: http://git.monstr.eu/?a=commitdiff_plain;h=4dc34ae1b45fe26e772a44379f936c72623dd407;p=linux-2.6-microblaze.git bcache: improve multithreaded bch_sectors_dirty_init() Commit b144e45fc576 ("bcache: make bch_sectors_dirty_init() to be multithreaded") makes bch_sectors_dirty_init() to be much faster when counting dirty sectors by iterating all dirty keys in the btree. But it isn't in ideal shape yet, still can be improved. This patch does the following changes to improve current parallel dirty keys iteration on the btree, - Add read lock to root node when multiple threads iterating the btree, to prevent the root node gets split by I/Os from other registered bcache devices. - Remove local variable "char name[32]" and generate kernel thread name string directly when calling kthread_run(). - Allocate "struct bch_dirty_init_state state" directly on stack and avoid the unnecessary dynamic memory allocation for it. - Decrease BCH_DIRTY_INIT_THRD_MAX from 64 to 12 which is enough indeed. - Increase &state->started to count created kernel thread after it succeeds to create. - When wait for all dirty key counting threads to finish, use wait_event() to replace wait_event_interruptible(). With the above changes, the code is more clear, and some potential error conditions are avoided. Fixes: b144e45fc576 ("bcache: make bch_sectors_dirty_init() to be multithreaded") Signed-off-by: Coly Li Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20220524102336.10684-3-colyli@suse.de Signed-off-by: Jens Axboe --- diff --git a/drivers/md/bcache/writeback.c b/drivers/md/bcache/writeback.c index 9ee0005874cd..d24c09490f8e 100644 --- a/drivers/md/bcache/writeback.c +++ b/drivers/md/bcache/writeback.c @@ -948,10 +948,10 @@ void bch_sectors_dirty_init(struct bcache_device *d) struct btree_iter iter; struct sectors_dirty_init op; struct cache_set *c = d->c; - struct bch_dirty_init_state *state; - char name[32]; + struct bch_dirty_init_state state; /* Just count root keys if no leaf node */ + rw_lock(0, c->root, c->root->level); if (c->root->level == 0) { bch_btree_op_init(&op.op, -1); op.inode = d->id; @@ -961,54 +961,42 @@ void bch_sectors_dirty_init(struct bcache_device *d) for_each_key_filter(&c->root->keys, k, &iter, bch_ptr_invalid) sectors_dirty_init_fn(&op.op, c->root, k); + rw_unlock(0, c->root); return; } - state = kzalloc(sizeof(struct bch_dirty_init_state), GFP_KERNEL); - if (!state) { - pr_warn("sectors dirty init failed: cannot allocate memory\n"); - return; - } - - state->c = c; - state->d = d; - state->total_threads = bch_btre_dirty_init_thread_nr(); - state->key_idx = 0; - spin_lock_init(&state->idx_lock); - atomic_set(&state->started, 0); - atomic_set(&state->enough, 0); - init_waitqueue_head(&state->wait); - - for (i = 0; i < state->total_threads; i++) { - /* Fetch latest state->enough earlier */ + state.c = c; + state.d = d; + state.total_threads = bch_btre_dirty_init_thread_nr(); + state.key_idx = 0; + spin_lock_init(&state.idx_lock); + atomic_set(&state.started, 0); + atomic_set(&state.enough, 0); + init_waitqueue_head(&state.wait); + + for (i = 0; i < state.total_threads; i++) { + /* Fetch latest state.enough earlier */ smp_mb__before_atomic(); - if (atomic_read(&state->enough)) + if (atomic_read(&state.enough)) break; - state->infos[i].state = state; - atomic_inc(&state->started); - snprintf(name, sizeof(name), "bch_dirty_init[%d]", i); - - state->infos[i].thread = - kthread_run(bch_dirty_init_thread, - &state->infos[i], - name); - if (IS_ERR(state->infos[i].thread)) { + state.infos[i].state = &state; + state.infos[i].thread = + kthread_run(bch_dirty_init_thread, &state.infos[i], + "bch_dirtcnt[%d]", i); + if (IS_ERR(state.infos[i].thread)) { pr_err("fails to run thread bch_dirty_init[%d]\n", i); for (--i; i >= 0; i--) - kthread_stop(state->infos[i].thread); + kthread_stop(state.infos[i].thread); goto out; } + atomic_inc(&state.started); } - /* - * Must wait for all threads to stop. - */ - wait_event_interruptible(state->wait, - atomic_read(&state->started) == 0); - out: - kfree(state); + /* Must wait for all threads to stop. */ + wait_event(state.wait, atomic_read(&state.started) == 0); + rw_unlock(0, c->root); } void bch_cached_dev_writeback_init(struct cached_dev *dc) diff --git a/drivers/md/bcache/writeback.h b/drivers/md/bcache/writeback.h index 02b2f9df73f6..31df716951f6 100644 --- a/drivers/md/bcache/writeback.h +++ b/drivers/md/bcache/writeback.h @@ -20,7 +20,7 @@ #define BCH_WRITEBACK_FRAGMENT_THRESHOLD_MID 57 #define BCH_WRITEBACK_FRAGMENT_THRESHOLD_HIGH 64 -#define BCH_DIRTY_INIT_THRD_MAX 64 +#define BCH_DIRTY_INIT_THRD_MAX 12 /* * 14 (16384ths) is chosen here as something that each backing device * should be a reasonable fraction of the share, and not to blow up