[PATCH v3 0/2] block: cope with WRITE ZEROES failing in blkdev_issue_zeroout()

Hi Christoph, Martin,

blkdev_issue_zeroout() now checks for any error.  This required a minor
refactor, so I dropped the stable tag, Jens can add it back if needed.

v2 -> v3:
- another code flow change in blkdev_issue_zeroout() suggested by
  Christoph -- no functional changes

v1 -> v2:
- changed code flow in blkdev_issue_zeroout() according to Christoph's
  suggestion
- this required adding additional checks to blkdev_issue_zeroout() and
  __blkdev_issue_zero_pages(), the latter is no longer void

Previous version at

  https://marc.info/?l=linux-block&m=150712940922089&w=2

Thanks,

                Ilya


Ilya Dryomov (2):
  block: factor out __blkdev_issue_zero_pages()
  block: cope with WRITE ZEROES failing in blkdev_issue_zeroout()

 block/blk-lib.c | 108 +++++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 72 insertions(+), 36 deletions(-)

-- 
2.4.3

blkdev_issue_zeroout() will use this in !BLKDEV_ZERO_NOFALLBACK case.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 block/blk-lib.c | 63 +++++++++++++++++++++++++++++++++------------------------
 1 file changed, 37 insertions(+), 26 deletions(-)

diff --git a/block/blk-lib.c b/block/blk-lib.c
index 62240f8832ca..9d2ab8bba52a 100644
--- a/block/blk-lib.c
+++ b/block/blk-lib.c

@@ -274,6 +274,40 @@ static unsigned int __blkdev_sectors_to_bio_pages(sector_t nr_sects)
 	return min(pages, (sector_t)BIO_MAX_PAGES);
 }
 
+static int __blkdev_issue_zero_pages(struct block_device *bdev,
+		sector_t sector, sector_t nr_sects, gfp_t gfp_mask,
+		struct bio **biop)
+{
+	struct request_queue *q = bdev_get_queue(bdev);
+	struct bio *bio = *biop;
+	int bi_size = 0;
+	unsigned int sz;
+
+	if (!q)
+		return -ENXIO;
+
+	while (nr_sects != 0) {
+		bio = next_bio(bio, __blkdev_sectors_to_bio_pages(nr_sects),
+			       gfp_mask);
+		bio->bi_iter.bi_sector = sector;
+		bio_set_dev(bio, bdev);
+		bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
+
+		while (nr_sects != 0) {
+			sz = min((sector_t) PAGE_SIZE, nr_sects << 9);
+			bi_size = bio_add_page(bio, ZERO_PAGE(0), sz, 0);
+			nr_sects -= bi_size >> 9;
+			sector += bi_size >> 9;
+			if (bi_size < sz)
+				break;
+		}
+		cond_resched();
+	}
+
+	*biop = bio;
+	return 0;
+}
+
 /**
  * __blkdev_issue_zeroout - generate number of zero filed write bios
  * @bdev:	blockdev to issue

@@ -304,9 +338,6 @@ int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
 		unsigned flags)
 {
 	int ret;
-	int bi_size = 0;
-	struct bio *bio = *biop;
-	unsigned int sz;
 	sector_t bs_mask;
 
 	bs_mask = (bdev_logical_block_size(bdev) >> 9) - 1;

@@ -316,30 +347,10 @@ int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
 	ret = __blkdev_issue_write_zeroes(bdev, sector, nr_sects, gfp_mask,
 			biop, flags);
 	if (ret != -EOPNOTSUPP || (flags & BLKDEV_ZERO_NOFALLBACK))
-		goto out;
-
-	ret = 0;
-	while (nr_sects != 0) {
-		bio = next_bio(bio, __blkdev_sectors_to_bio_pages(nr_sects),
-			       gfp_mask);
-		bio->bi_iter.bi_sector = sector;
-		bio_set_dev(bio, bdev);
-		bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
-
-		while (nr_sects != 0) {
-			sz = min((sector_t) PAGE_SIZE, nr_sects << 9);
-			bi_size = bio_add_page(bio, ZERO_PAGE(0), sz, 0);
-			nr_sects -= bi_size >> 9;
-			sector += bi_size >> 9;
-			if (bi_size < sz)
-				break;
-		}
-		cond_resched();
-	}
+		return ret;
 
-	*biop = bio;
-out:
-	return ret;
+	return __blkdev_issue_zero_pages(bdev, sector, nr_sects, gfp_mask,
+					 biop);
 }
 EXPORT_SYMBOL(__blkdev_issue_zeroout);
 

-- 
2.4.3

Looks fine,

Reviewed-by: Christoph Hellwig <hch@lst.de>

sd_config_write_same() ignores ->max_ws_blocks == 0 and resets it to
permit trying WRITE SAME on older SCSI devices, unless ->no_write_same
is set.  Because REQ_OP_WRITE_ZEROES is implemented in terms of WRITE
SAME, blkdev_issue_zeroout() may fail with -EREMOTEIO:

  $ fallocate -zn -l 1k /dev/sdg
  fallocate: fallocate failed: Remote I/O error
  $ fallocate -zn -l 1k /dev/sdg  # OK
  $ fallocate -zn -l 1k /dev/sdg  # OK

The following calls succeed because sd_done() sets ->no_write_same in
response to a sense that would become BLK_STS_TARGET/-EREMOTEIO, causing
__blkdev_issue_zeroout() to fall back to generating ZERO_PAGE bios.

This means blkdev_issue_zeroout() must cope with WRITE ZEROES failing
and fall back to manually zeroing, unless BLKDEV_ZERO_NOFALLBACK is
specified.  For BLKDEV_ZERO_NOFALLBACK case, return -EOPNOTSUPP if
sd_done() has just set ->no_write_same thus indicating lack of offload
support.

Fixes: c20cfc27a473 ("block: stop using blkdev_issue_write_same for zeroing")
Cc: Christoph Hellwig <hch@lst.de>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 block/blk-lib.c | 45 +++++++++++++++++++++++++++++++++++----------
 1 file changed, 35 insertions(+), 10 deletions(-)

diff --git a/block/blk-lib.c b/block/blk-lib.c
index 9d2ab8bba52a..23411ec7f3cc 100644
--- a/block/blk-lib.c
+++ b/block/blk-lib.c

@@ -321,12 +321,6 @@ static int __blkdev_issue_zero_pages(struct block_device *bdev,
  *  Zero-fill a block range, either using hardware offload or by explicitly
  *  writing zeroes to the device.
  *
- *  Note that this function may fail with -EOPNOTSUPP if the driver signals
- *  zeroing offload support, but the device fails to process the command (for
- *  some devices there is no non-destructive way to verify whether this
- *  operation is actually supported).  In this case the caller should call
- *  retry the call to blkdev_issue_zeroout() and the fallback path will be used.
- *
  *  If a device is using logical block provisioning, the underlying space will
  *  not be released if %flags contains BLKDEV_ZERO_NOUNMAP.
  *

@@ -370,18 +364,49 @@ EXPORT_SYMBOL(__blkdev_issue_zeroout);
 int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
 		sector_t nr_sects, gfp_t gfp_mask, unsigned flags)
 {
-	int ret;
-	struct bio *bio = NULL;
+	int ret = 0;
+	sector_t bs_mask;
+	struct bio *bio;
 	struct blk_plug plug;
+	bool try_write_zeroes = !!bdev_write_zeroes_sectors(bdev);
 
+	bs_mask = (bdev_logical_block_size(bdev) >> 9) - 1;
+	if ((sector | nr_sects) & bs_mask)
+		return -EINVAL;
+
+retry:
+	bio = NULL;
 	blk_start_plug(&plug);
-	ret = __blkdev_issue_zeroout(bdev, sector, nr_sects, gfp_mask,
-			&bio, flags);
+	if (try_write_zeroes) {
+		ret = __blkdev_issue_write_zeroes(bdev, sector, nr_sects,
+						  gfp_mask, &bio, flags);
+	} else if (!(flags & BLKDEV_ZERO_NOFALLBACK)) {
+		ret = __blkdev_issue_zero_pages(bdev, sector, nr_sects,
+						gfp_mask, &bio);
+	} else {
+		/* No zeroing offload support */
+		ret = -EOPNOTSUPP;
+	}
 	if (ret == 0 && bio) {
 		ret = submit_bio_wait(bio);
 		bio_put(bio);
 	}
 	blk_finish_plug(&plug);
+	if (ret && try_write_zeroes) {
+		if (!(flags & BLKDEV_ZERO_NOFALLBACK)) {
+			try_write_zeroes = false;
+			goto retry;
+		}
+		if (!bdev_write_zeroes_sectors(bdev)) {
+			/*
+			 * Zeroing offload support was indicated, but the
+			 * device reported ILLEGAL REQUEST (for some devices
+			 * there is no non-destructive way to verify whether
+			 * WRITE ZEROES is actually supported).
+			 */
+			ret = -EOPNOTSUPP;
+		}
+	}
 
 	return ret;
 }

-- 
2.4.3

Looks fine,

Reviewed-by: Christoph Hellwig <hch@lst.de>

On Mon, Oct 16, 2017 at 3:59 PM, Ilya Dryomov [off-list ref] wrote:

Hi Christoph, Martin,

blkdev_issue_zeroout() now checks for any error.  This required a minor
refactor, so I dropped the stable tag, Jens can add it back if needed.

v2 -> v3:
- another code flow change in blkdev_issue_zeroout() suggested by
  Christoph -- no functional changes

v1 -> v2:
- changed code flow in blkdev_issue_zeroout() according to Christoph's
  suggestion
- this required adding additional checks to blkdev_issue_zeroout() and
  __blkdev_issue_zero_pages(), the latter is no longer void

Previous version at

  https://marc.info/?l=linux-block&m=150712940922089&w=2

Thanks,

                Ilya


Ilya Dryomov (2):
  block: factor out __blkdev_issue_zero_pages()
  block: cope with WRITE ZEROES failing in blkdev_issue_zeroout()

 block/blk-lib.c | 108 +++++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 72 insertions(+), 36 deletions(-)

Jens, can you pick these up?  Technically this is a user-visible
regression (Fixes tag in 2/2), but we are late in the cycle so I'm not
sure what is the right thing to do here.

Thanks,

                Ilya

On 10/24/2017 11:27 PM, Ilya Dryomov wrote:

On Mon, Oct 16, 2017 at 3:59 PM, Ilya Dryomov [off-list ref] wrote:

quoted

Hi Christoph, Martin,

blkdev_issue_zeroout() now checks for any error.  This required a minor
refactor, so I dropped the stable tag, Jens can add it back if needed.

v2 -> v3:
- another code flow change in blkdev_issue_zeroout() suggested by
  Christoph -- no functional changes

v1 -> v2:
- changed code flow in blkdev_issue_zeroout() according to Christoph's
  suggestion
- this required adding additional checks to blkdev_issue_zeroout() and
  __blkdev_issue_zero_pages(), the latter is no longer void

Previous version at

  https://marc.info/?l=linux-block&m=150712940922089&w=2

Thanks,

                Ilya


Ilya Dryomov (2):
  block: factor out __blkdev_issue_zero_pages()
  block: cope with WRITE ZEROES failing in blkdev_issue_zeroout()

 block/blk-lib.c | 108 +++++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 72 insertions(+), 36 deletions(-)

Jens, can you pick these up?  Technically this is a user-visible
regression (Fixes tag in 2/2), but we are late in the cycle so I'm not
sure what is the right thing to do here.

It wasn't introduced in this cycle, so there's little point in trying
to rush it for 4.14. I have applied it for 4.15, thanks.

-- 
Jens Axboe