btrfs: defer I/O completion based on the btrfs_raid_bio

Instead of attaching an extra allocation an indirect call to each
low-level bio issued by the RAID code, add a work_struct to struct
btrfs_raid_bio and only defer the per-rbio completion action.  The
per-bio action for all the I/Os are trivial and can be safely done
from interrupt context.

As a nice side effect this also allows sharing the boilerplate code
for the per-bio completions

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>

1 files changed, 40 insertions, 62 deletions

diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c
index baba435692d2..00cd9e8db7ae 100644
--- a/fs/btrfs/raid56.c
+++ b/fs/btrfs/raid56.c
@@ -1488,15 +1488,7 @@ static void set_bio_pages_uptodate(struct btrfs_raid_bio *rbio, struct bio *bio)
 	}
 }
 
-/*
- * end io for the read phase of the rmw cycle.  All the bios here are physical
- * stripe bios we've read from the disk so we can recalculate the parity of the
- * stripe.
- *
- * This will usually kick off finish_rmw once all the bios are read in, but it
- * may trigger parity reconstruction if we had any errors along the way
- */
-static void raid_rmw_end_io(struct bio *bio)
+static void raid56_bio_end_io(struct bio *bio)
 {
 	struct btrfs_raid_bio *rbio = bio->bi_private;
 
@@ -1507,23 +1499,34 @@ static void raid_rmw_end_io(struct bio *bio)
 
 	bio_put(bio);
 
-	if (!atomic_dec_and_test(&rbio->stripes_pending))
-		return;
+	if (atomic_dec_and_test(&rbio->stripes_pending))
+		queue_work(rbio->bioc->fs_info->endio_raid56_workers,
+			   &rbio->end_io_work);
+}
 
-	if (atomic_read(&rbio->error) > rbio->bioc->max_errors)
-		goto cleanup;
+/*
+ * End io handler for the read phase of the RMW cycle.  All the bios here are
+ * physical stripe bios we've read from the disk so we can recalculate the
+ * parity of the stripe.
+ *
+ * This will usually kick off finish_rmw once all the bios are read in, but it
+ * may trigger parity reconstruction if we had any errors along the way
+ */
+static void raid56_rmw_end_io_work(struct work_struct *work)
+{
+	struct btrfs_raid_bio *rbio =
+		container_of(work, struct btrfs_raid_bio, end_io_work);
+
+	if (atomic_read(&rbio->error) > rbio->bioc->max_errors) {
+		rbio_orig_end_io(rbio, BLK_STS_IOERR);
+		return;
+	}
 
 	/*
-	 * this will normally call finish_rmw to start our write
-	 * but if there are any failed stripes we'll reconstruct
-	 * from parity first
+	 * This will normally call finish_rmw to start our write but if there
+	 * are any failed stripes we'll reconstruct from parity first.
 	 */
 	validate_rbio_for_rmw(rbio);
-	return;
-
-cleanup:
-
-	rbio_orig_end_io(rbio, BLK_STS_IOERR);
 }
 
 /*
@@ -1598,10 +1601,9 @@ static int raid56_rmw_stripe(struct btrfs_raid_bio *rbio)
 	 * touch it after that.
 	 */
 	atomic_set(&rbio->stripes_pending, bios_to_read);
+	INIT_WORK(&rbio->end_io_work, raid56_rmw_end_io_work);
 	while ((bio = bio_list_pop(&bio_list))) {
-		bio->bi_end_io = raid_rmw_end_io;
-
-		btrfs_bio_wq_end_io(rbio->bioc->fs_info, bio, BTRFS_WQ_ENDIO_RAID56);
+		bio->bi_end_io = raid56_bio_end_io;
 
 		if (trace_raid56_read_partial_enabled()) {
 			struct raid56_bio_trace_info trace_info = { 0 };
@@ -2076,25 +2078,13 @@ cleanup_io:
 }
 
 /*
- * This is called only for stripes we've read from disk to
- * reconstruct the parity.
+ * This is called only for stripes we've read from disk to reconstruct the
+ * parity.
  */
-static void raid_recover_end_io(struct bio *bio)
+static void raid_recover_end_io_work(struct work_struct *work)
 {
-	struct btrfs_raid_bio *rbio = bio->bi_private;
-
-	/*
-	 * we only read stripe pages off the disk, set them
-	 * up to date if there were no errors
-	 */
-	if (bio->bi_status)
-		fail_bio_stripe(rbio, bio);
-	else
-		set_bio_pages_uptodate(rbio, bio);
-	bio_put(bio);
-
-	if (!atomic_dec_and_test(&rbio->stripes_pending))
-		return;
+	struct btrfs_raid_bio *rbio =
+		container_of(work, struct btrfs_raid_bio, end_io_work);
 
 	if (atomic_read(&rbio->error) > rbio->bioc->max_errors)
 		rbio_orig_end_io(rbio, BLK_STS_IOERR);
@@ -2177,10 +2167,9 @@ static int __raid56_parity_recover(struct btrfs_raid_bio *rbio)
 	 * touch it after that.
 	 */
 	atomic_set(&rbio->stripes_pending, bios_to_read);
+	INIT_WORK(&rbio->end_io_work, raid_recover_end_io_work);
 	while ((bio = bio_list_pop(&bio_list))) {
-		bio->bi_end_io = raid_recover_end_io;
-
-		btrfs_bio_wq_end_io(rbio->bioc->fs_info, bio, BTRFS_WQ_ENDIO_RAID56);
+		bio->bi_end_io = raid56_bio_end_io;
 
 		if (trace_raid56_scrub_read_recover_enabled()) {
 			struct raid56_bio_trace_info trace_info = { 0 };
@@ -2650,24 +2639,14 @@ cleanup:
  * This will usually kick off finish_rmw once all the bios are read in, but it
  * may trigger parity reconstruction if we had any errors along the way
  */
-static void raid56_parity_scrub_end_io(struct bio *bio)
+static void raid56_parity_scrub_end_io_work(struct work_struct *work)
 {
-	struct btrfs_raid_bio *rbio = bio->bi_private;
-
-	if (bio->bi_status)
-		fail_bio_stripe(rbio, bio);
-	else
-		set_bio_pages_uptodate(rbio, bio);
-
-	bio_put(bio);
-
-	if (!atomic_dec_and_test(&rbio->stripes_pending))
-		return;
+	struct btrfs_raid_bio *rbio =
+		container_of(work, struct btrfs_raid_bio, end_io_work);
 
 	/*
-	 * this will normally call finish_rmw to start our write
-	 * but if there are any failed stripes we'll reconstruct
-	 * from parity first
+	 * This will normally call finish_rmw to start our write, but if there
+	 * are any failed stripes we'll reconstruct from parity first
 	 */
 	validate_rbio_for_parity_scrub(rbio);
 }
@@ -2737,10 +2716,9 @@ static void raid56_parity_scrub_stripe(struct btrfs_raid_bio *rbio)
 	 * touch it after that.
 	 */
 	atomic_set(&rbio->stripes_pending, bios_to_read);
+	INIT_WORK(&rbio->end_io_work, raid56_parity_scrub_end_io_work);
 	while ((bio = bio_list_pop(&bio_list))) {
-		bio->bi_end_io = raid56_parity_scrub_end_io;
-
-		btrfs_bio_wq_end_io(rbio->bioc->fs_info, bio, BTRFS_WQ_ENDIO_RAID56);
+		bio->bi_end_io = raid56_bio_end_io;
 
 		if (trace_raid56_scrub_read_enabled()) {
 			struct raid56_bio_trace_info trace_info = { 0 };