Commit dfcc34c9 authored by Nate Dailey's avatar Nate Dailey Committed by Song Liu
Browse files

md/raid1: don't clear bitmap bits on interrupted recovery. 



sync_request_write no longer submits writes to a Faulty device. This has
the unfortunate side effect that bitmap bits can be incorrectly cleared
if a recovery is interrupted (previously, end_sync_write would have
prevented this). This means the next recovery may not copy everything
it should, potentially corrupting data.

Add a function for doing the proper md_bitmap_end_sync, called from
end_sync_write and the Faulty case in sync_request_write.

backport note to 4.14: s/md_bitmap_end_sync/bitmap_end_sync
Cc: stable@vger.kernel.org 4.14+
Fixes: 0c9d5b12 ("md/raid1: avoid reusing a resync bio after error handling.")
Reviewed-by: default avatarJack Wang <jinpu.wang@cloud.ionos.com>
Tested-by: default avatarJack Wang <jinpu.wang@cloud.ionos.com>
Signed-off-by: default avatarNate Dailey <nate.dailey@stratus.com>
Signed-off-by: default avatarSong Liu <songliubraving@fb.com>
parent 96d7cb93

drivers/md/raid1.c

+18 −10
Original line number Diff line number Diff line
@@ -1863,6 +1863,20 @@ static void end_sync_read(struct bio *bio) 
		reschedule_retry(r1_bio);

}



static void abort_sync_write(struct mddev *mddev, struct r1bio *r1_bio)

{

	sector_t sync_blocks = 0;

	sector_t s = r1_bio->sector;

	long sectors_to_go = r1_bio->sectors;



	/* make sure these bits don't get cleared. */

	do {

		md_bitmap_end_sync(mddev->bitmap, s, &sync_blocks, 1);

		s += sync_blocks;

		sectors_to_go -= sync_blocks;

	} while (sectors_to_go > 0);

}



static void end_sync_write(struct bio *bio)

{

	int uptodate = !bio->bi_status;
@@ -1874,15 +1888,7 @@ static void end_sync_write(struct bio *bio) 
	struct md_rdev *rdev = conf->mirrors[find_bio_disk(r1_bio, bio)].rdev;



	if (!uptodate) {

		sector_t sync_blocks = 0;

		sector_t s = r1_bio->sector;

		long sectors_to_go = r1_bio->sectors;

		/* make sure these bits doesn't get cleared. */

		do {

			md_bitmap_end_sync(mddev->bitmap, s, &sync_blocks, 1);

			s += sync_blocks;

			sectors_to_go -= sync_blocks;

		} while (sectors_to_go > 0);

		abort_sync_write(mddev, r1_bio);

		set_bit(WriteErrorSeen, &rdev->flags);

		if (!test_and_set_bit(WantReplacement, &rdev->flags))

			set_bit(MD_RECOVERY_NEEDED, &
@@ -2172,8 +2178,10 @@ static void sync_request_write(struct mddev *mddev, struct r1bio *r1_bio) 
		     (i == r1_bio->read_disk ||

		      !test_bit(MD_RECOVERY_SYNC, &mddev->recovery))))

			continue;

		if (test_bit(Faulty, &conf->mirrors[i].rdev->flags))

		if (test_bit(Faulty, &conf->mirrors[i].rdev->flags)) {

			abort_sync_write(mddev, r1_bio);

			continue;

		}



		bio_set_op_attrs(wbio, REQ_OP_WRITE, 0);

		if (test_bit(FailFast, &conf->mirrors[i].rdev->flags))

CRA Git | Maintained and supported by SUSTech CRA and CCSE