On Wed, 12 Dec 2012, Martin K. Petersen wrote: > >>>>> "Joe" == Joe Lawrence <Joe.Lawrence@xxxxxxxxxxx> writes: > > Joe> I can confirm the same issue with 3.7.0. If anyone else is running > Joe> with raid1 and disks that support write same, can you give this a > Joe> try? > > Your patch looks good to me (the do_same one). We'll need raid10.c and > raid5.c to be fixed up in a similar fashion. > > -- > Martin K. Petersen Oracle Linux Engineering Hi Martin, I took a look at raid5 and I don't think it suffers from the same problem (ie, cloned write bios missing the flag). A quick mkfs/mount test showed that the blkdev_issue_write_same() calls all succeeded anyway. So I added the same logic to raid10 and it similarly passes my quick tests. I don't know what else might create WRITE SAME cmds at the moment (I tried dd'ing a bunch of zeros and that didn't seem to spawn any), so all I did was to mkfs/mount/fio/umount/fsck. MD recovery seemed happy if did this with a degraded array and brought the partner in later. One question I do have though, I'm not sure about any write bitmap implications of this. I noticed in raid0_run you call: blk_queue_max_write_same_sectors(mddev->queue, mddev->chunk_sectors); which should keep the acceptable LBA range inside a bitmap 'chunk'? Am I right in understanding that this would keep any write same from ranging across bitmap bits? In my testing, my MD chunksize was 512K but my SAS disks write_same_sectors was only 64K... so I think I inadvertently missed this necessary step. Thanks, -- Joe >From c3ebb7a21850f1ff83c5498655e4f5a18aa883fd Mon Sep 17 00:00:00 2001 From: Joe Lawrence <joe.lawrence@xxxxxxxxxxx> Date: Wed, 12 Dec 2012 17:03:40 -0500 Subject: [PATCH] md: raid1,10: Copy REQ_WRITE_SAME flag in cloned write bios If the mddev's max_write_same_sectors are non-zero, the block layer may send WRITE_SAME requests. When cloning these bios in raid1,10 write cases, make sure we add this flag to the new bios. Signed-off-by: Joe Lawrence <joe.lawrence@xxxxxxxxxxx> --- drivers/md/raid1.c | 4 +++- drivers/md/raid10.c | 7 +++++-- 2 files changed, 8 insertions(+), 3 deletions(-) diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index a0f7309..85aba6a 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -1001,6 +1001,7 @@ static void make_request(struct mddev *mddev, struct bio * bio) const unsigned long do_flush_fua = (bio->bi_rw & (REQ_FLUSH | REQ_FUA)); const unsigned long do_discard = (bio->bi_rw & (REQ_DISCARD | REQ_SECURE)); + const unsigned long do_same = (bio->bi_rw & REQ_WRITE_SAME); struct md_rdev *blocked_rdev; struct blk_plug_cb *cb; struct raid1_plug_cb *plug = NULL; @@ -1302,7 +1303,8 @@ read_again: conf->mirrors[i].rdev->data_offset); mbio->bi_bdev = conf->mirrors[i].rdev->bdev; mbio->bi_end_io = raid1_end_write_request; - mbio->bi_rw = WRITE | do_flush_fua | do_sync | do_discard; + mbio->bi_rw = + WRITE | do_flush_fua | do_sync | do_discard | do_same; mbio->bi_private = r1_bio; atomic_inc(&r1_bio->remaining); diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index c9acbd7..fdb4a6e 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -1106,6 +1106,7 @@ static void make_request(struct mddev *mddev, struct bio * bio) const unsigned long do_fua = (bio->bi_rw & REQ_FUA); const unsigned long do_discard = (bio->bi_rw & (REQ_DISCARD | REQ_SECURE)); + const unsigned long do_same = (bio->bi_rw & REQ_WRITE_SAME); unsigned long flags; struct md_rdev *blocked_rdev; struct blk_plug_cb *cb; @@ -1461,7 +1462,8 @@ retry_write: rdev)); mbio->bi_bdev = rdev->bdev; mbio->bi_end_io = raid10_end_write_request; - mbio->bi_rw = WRITE | do_sync | do_fua | do_discard; + mbio->bi_rw = + WRITE | do_sync | do_fua | do_discard | do_same; mbio->bi_private = r10_bio; atomic_inc(&r10_bio->remaining); @@ -1503,7 +1505,8 @@ retry_write: r10_bio, rdev)); mbio->bi_bdev = rdev->bdev; mbio->bi_end_io = raid10_end_write_request; - mbio->bi_rw = WRITE | do_sync | do_fua | do_discard; + mbio->bi_rw = + WRITE | do_sync | do_fua | do_discard | do_same; mbio->bi_private = r10_bio; atomic_inc(&r10_bio->remaining); -- 1.7.11.7 -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
