On Wed, 2018-07-04 at 10:39 -0400, Mikulas Patocka wrote:
This is backport of the upstream patch 41c73a49df31151f4ff868f28fe4f129f113fa2c. It should be backported to stable kernels because this performance problem was seen on the android 4.4 kernel.
Also queued this up for 3.16, thanks.
Ben.
commit 41c73a49df31151f4ff868f28fe4f129f113fa2c Author: Mikulas Patocka mpatocka@redhat.com Date: Wed Nov 23 17:04:00 2016 -0500
dm bufio: drop the lock when doing GFP_NOIO allocationIf the first allocation attempt using GFP_NOWAIT fails, drop the lock and retry using GFP_NOIO allocation (lock is dropped because the allocation can take some time). Note that we won't do GFP_NOIO allocation when we loop for the second time, because the lock shouldn't be dropped between __wait_for_free_buffer and __get_unclaimed_buffer. Signed-off-by: Mikulas Patocka mpatocka@redhat.com Signed-off-by: Mike Snitzer snitzer@redhat.com
diff --git a/drivers/md/dm-bufio.c b/drivers/md/dm-bufio.c index 9ef88f0e2382..1c2e1dd7ca16 100644 --- a/drivers/md/dm-bufio.c +++ b/drivers/md/dm-bufio.c @@ -822,6 +822,7 @@ enum new_flag { static struct dm_buffer *__alloc_buffer_wait_no_callback(struct dm_bufio_client *c, enum new_flag nf) { struct dm_buffer *b;
bool tried_noio_alloc = false;/* * dm-bufio is resistant to allocation failures (it just keeps @@ -846,6 +847,15 @@ static struct dm_buffer *__alloc_buffer_wait_no_callback(struct dm_bufio_client if (nf == NF_PREFETCH) return NULL;
if (dm_bufio_cache_size_latch != 1 && !tried_noio_alloc) {dm_bufio_unlock(c);b = alloc_buffer(c, GFP_NOIO | __GFP_NORETRY | __GFP_NOMEMALLOC | __GFP_NOWARN);dm_bufio_lock(c);if (b)return b;tried_noio_alloc = true;}if (!list_empty(&c->reserved_buffers)) { b = list_entry(c->reserved_buffers.next, struct dm_buffer, lru_list);