[PATCH v3] btrfs: not a disk error if the bio_add_page fails

[PATCH v3] btrfs: not a disk error if the bio_add_page fails

[Anand Jain]

bio_add_page() can fail for logical reasons as from the bio_add_page()
comments:- 'This will only fail if either
bio->bi_vcnt == bio->bi_max_vecs or it's a cloned bio.'
Here as we have just allocated the bio, so both of those failures can't
occur. So drop the check for fail. Which would also drop write error
statistics.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
v2->v3: Drop check for bio_add_page return value. Thanks Liubo.
v1->v2: Add missing bio_put(). Thanks Filipe.

 fs/btrfs/scrub.c | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index ecfe3118d9dd..5030898247db 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -4600,7 +4600,6 @@ static int write_page_nocow(struct scrub_ctx *sctx,
 {
 	struct bio *bio;
 	struct btrfs_device *dev;
-	int ret;
 
 	dev = sctx->wr_tgtdev;
 	if (!dev)
@@ -4615,17 +4614,15 @@ static int write_page_nocow(struct scrub_ctx *sctx,
 	bio->bi_iter.bi_sector = physical_for_dev_replace >> 9;
 	bio_set_dev(bio, dev->bdev);
 	bio->bi_opf = REQ_OP_WRITE | REQ_SYNC;
-	ret = bio_add_page(bio, page, PAGE_SIZE, 0);
-	if (ret != PAGE_SIZE) {
-leave_with_eio:
+	/* bio_add_page won't fail here */
+	bio_add_page(bio, page, PAGE_SIZE, 0);
+
+	if (btrfsic_submit_bio_wait(bio)) {
 		bio_put(bio);
 		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_WRITE_ERRS);
 		return -EIO;
 	}
 
-	if (btrfsic_submit_bio_wait(bio))
-		goto leave_with_eio;
-
 	bio_put(bio);
 	return 0;
 }
-- 
2.15.0

Re: [PATCH v3] btrfs: not a disk error if the bio_add_page fails

[David Sterba]

On Fri, Jan 05, 2018 at 10:47:07AM +0800, Anand Jain wrote:
> bio_add_page() can fail for logical reasons as from the bio_add_page()
> comments:- 'This will only fail if either
> bio->bi_vcnt == bio->bi_max_vecs or it's a cloned bio.'
> Here as we have just allocated the bio, so both of those failures can't
> occur. So drop the check for fail. Which would also drop write error
> statistics.
> 
> Signed-off-by: Anand Jain <anand.jain@oracle.com>

Reviewed-by: David Sterba <dsterba@suse.com>