[dm-devel] [PATCH] dm-crypt: fix lost ioprio when queuing crypto bios from task with ioprio

[Eric Wheeler]

Since dm-crypt queues writes (and sometimes reads) to a different kernel
thread (workqueue), the bios will dispatch from tasks with different
io_context->ioprio settings than the submitting task, thus giving
incorrect ioprio hints to the io scheduler.

By assigning the ioprio to the bio before queuing to a workqueue, the
original submitting task's io_context->ioprio setting can be retained
through the life of the bio.  We only set the bio's ioprio in dm-crypt
if not already set (by somewhere else, higher in the stack).

Signed-off-by: Eric Wheeler <dm-devel at linux.ewheeler.net>
Cc: Mikulas Patocka <mpatocka at redhat.com>
Cc: Alasdair Kergon <agk at redhat.com>
Cc: Mike Snitzer <snitzer at redhat.com>
Cc: Shaohua Li <shli at kernel.org>
Cc: Jens Axboe <axboe at kernel.dk>
Cc: stable at vger.kernel.org
---
 block/bio.c           |  1 +
 drivers/md/dm-crypt.c | 11 +++++++++--
 include/linux/bio.h   | 23 +++++++++++++++++++++++
 3 files changed, 33 insertions(+), 2 deletions(-)

diff --git a/block/bio.c b/block/bio.c
index db85c57..1a529ff 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -584,6 +584,7 @@ void __bio_clone_fast(struct bio *bio, struct bio *bio_src)
 	bio->bi_iter = bio_src->bi_iter;
 	bio->bi_io_vec = bio_src->bi_io_vec;
 
+	bio_set_prio(bio, bio_prio(bio_src));
 	bio_clone_blkcg_association(bio, bio_src);
 }
 EXPORT_SYMBOL(__bio_clone_fast);
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index 9b99ee9..ea7e102 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -1133,6 +1133,7 @@ static void clone_init(struct dm_crypt_io *io, struct bio *clone)
 	clone->bi_private = io;
 	clone->bi_end_io  = crypt_endio;
 	clone->bi_bdev    = cc->dev->bdev;
+	bio_set_prio(clone, bio_prio(io->base_bio));
 	bio_set_op_attrs(clone, bio_op(io->base_bio), bio_flags(io->base_bio));
 }
 
@@ -2070,10 +2071,16 @@ static int crypt_map(struct dm_target *ti, struct bio *bio)
 	io->ctx.req = (struct skcipher_request *)(io + 1);
 
 	if (bio_data_dir(io->base_bio) == READ) {
-		if (kcryptd_io_read(io, GFP_NOWAIT))
+		if (kcryptd_io_read(io, GFP_NOWAIT)) {
+			if (!ioprio_valid(bio_prio(io->base_bio)))
+				bio_set_task_prio(io->base_bio, current);
 			kcryptd_queue_read(io);
-	} else
+		}
+	} else {
+		if (!ioprio_valid(bio_prio(io->base_bio)))
+			bio_set_task_prio(io->base_bio, current);
 		kcryptd_queue_crypt(io);
+	}
 
 	return DM_MAPIO_SUBMITTED;
 }
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 97cb48f..0b4f8bb 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -62,6 +62,29 @@
 #define bio_sectors(bio)	((bio)->bi_iter.bi_size >> 9)
 #define bio_end_sector(bio)	((bio)->bi_iter.bi_sector + bio_sectors((bio)))
 
+/* Set the bio's ioprio to that of the task's io_context->ioprio, if any.
+ * Always returns io_context->ioprio (or 0 if none); bio can be NULL.
+ */
+static inline unsigned short bio_set_task_prio(
+	struct bio *bio, struct task_struct *task)
+{
+	struct io_context *ioc;
+	unsigned short ioprio = 0;
+
+	ioc = get_task_io_context(current, GFP_NOIO, NUMA_NO_NODE);
+	if (ioc) {
+		if (ioprio_valid(ioc->ioprio)) {
+			ioprio = ioc->ioprio;
+			put_io_context(ioc);
+		}
+
+		if (bio != NULL && ioprio_valid(ioprio))
+			bio_set_prio(bio, ioc->ioprio);
+	}
+
+	return ioprio;
+}
+
 /*
  * Check whether this bio carries any data or not. A NULL bio is allowed.
  */
-- 
1.8.3.1