DM-Devel Archive on lore.kernel.org
 help / color / Atom feed
* [RFC] dm: fix imposition of queue_limits in dm_wq_work() thread
@ 2020-09-27 12:04 Jeffle Xu
  2020-09-28 16:03 ` Mike Snitzer
  0 siblings, 1 reply; 4+ messages in thread
From: Jeffle Xu @ 2020-09-27 12:04 UTC (permalink / raw)
  To: snitzer; +Cc: joseph.qi, dm-devel

Hi Mike, would you mind further expalin why bio processed by dm_wq_work()
always gets a previous ->submit_bio. Considering the following call graph:

->submit_bio, that is, dm_submit_bio
  DMF_BLOCK_IO_FOR_SUSPEND set, thus queue_io()

then worker thread dm_wq_work()
  dm_process_bio  // at this point. the input bio is the original bio
                     submitted to dm device

Please let me know if I missed something.

Thanks.
Jeffle


Original commit log:
dm_process_bio() can be called by dm_wq_work(), and if the currently
processing bio is the very beginning bio submitted to the dm device,
that is it has not been handled by previous ->submit_bio, then we also
need to impose the queue_limits when it's in thread (dm_wq_work()).

Fixes: cf9c37865557 ("dm: fix comment in dm_process_bio()")
Fixes: 568c73a355e0 ("dm: update dm_process_bio() to split bio if in ->make_request_fn()")
Signed-off-by: Jeffle Xu <jefflexu@linux.alibaba.com>
---
 drivers/md/dm.c | 16 ++++++----------
 1 file changed, 6 insertions(+), 10 deletions(-)

diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 6ed05ca65a0f..54471c75ddef 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1744,17 +1744,13 @@ static blk_qc_t dm_process_bio(struct mapped_device *md,
 	}
 
 	/*
-	 * If in ->submit_bio we need to use blk_queue_split(), otherwise
-	 * queue_limits for abnormal requests (e.g. discard, writesame, etc)
-	 * won't be imposed.
-	 * If called from dm_wq_work() for deferred bio processing, bio
-	 * was already handled by following code with previous ->submit_bio.
+	 * Call blk_queue_split() so that queue_limits for abnormal requests
+	 * (e.g. discard, writesame, etc) are ensured to be imposed, while
+	 * it's not needed for regular IO, since regular IO will be split by
+	 * following __split_and_process_bio.
 	 */
-	if (current->bio_list) {
-		if (is_abnormal_io(bio))
-			blk_queue_split(&bio);
-		/* regular IO is split by __split_and_process_bio */
-	}
+	if (is_abnormal_io(bio))
+		blk_queue_split(&bio);
 
 	if (dm_get_md_type(md) == DM_TYPE_NVME_BIO_BASED)
 		return __process_bio(md, map, bio, ti);
-- 
2.27.0

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, back to index

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-09-27 12:04 [RFC] dm: fix imposition of queue_limits in dm_wq_work() thread Jeffle Xu
2020-09-28 16:03 ` Mike Snitzer
2020-09-29  4:08   ` JeffleXu
2020-09-29 20:39   ` [PATCH] dm: fix missing imposition of queue_limits from " Mike Snitzer

DM-Devel Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/dm-devel/0 dm-devel/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dm-devel dm-devel/ https://lore.kernel.org/dm-devel \
		dm-devel@redhat.com
	public-inbox-index dm-devel

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/com.redhat.dm-devel


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git