From: NeilBrown <neilb@suse.com>
To: Milan Broz <gmazyland@gmail.com>,
Mike Snitzer <snitzer@redhat.com>,
device-mapper development <dm-devel@redhat.com>
Cc: Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: [RFC PATCH] dm: don't assign zero to ->bi_status of an active bio.
Date: Thu, 15 Feb 2018 20:01:31 +1100 [thread overview]
Message-ID: <87h8qipqxg.fsf@notabene.neil.brown.name> (raw)
In-Reply-To: <70cda2a3-f246-d45b-f600-1f9d15ba22ff@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 4354 bytes --]
Between the moment when generic_make_request() is first
called on a bio, and when bio_endio() finally gets past
bio_remaining_done(), a bio might have chained children, and might
get ->bi_status set asynchronously.
So during this time it is not safe to set it to zero.
It *is* safe to set it to any other error status as for most
purposes, one error is as good as another.
This patch is untested and RFC only. It identifies a number of
possible problem areas in dm and often suggests fixes.
Please don't apply without careful review.
Reviewed-by: Nobody yet.
Signed-off-by: NeilBrown <neilb@suse.com>
---
drivers/md/dm-bio-prison-v1.c | 3 ++-
drivers/md/dm-bufio.c | 6 ++++--
drivers/md/dm-crypt.c | 3 ++-
drivers/md/dm-mpath.c | 3 +++
drivers/md/dm-thin.c | 3 ++-
drivers/md/dm-verity-target.c | 3 ++-
drivers/md/dm-zoned-target.c | 3 ++-
7 files changed, 17 insertions(+), 7 deletions(-)
diff --git a/drivers/md/dm-bio-prison-v1.c b/drivers/md/dm-bio-prison-v1.c
index 874841f0fc83..68cb4fe98e9f 100644
--- a/drivers/md/dm-bio-prison-v1.c
+++ b/drivers/md/dm-bio-prison-v1.c
@@ -238,7 +238,8 @@ void dm_cell_error(struct dm_bio_prison *prison,
dm_cell_release(prison, cell, &bios);
while ((bio = bio_list_pop(&bios))) {
- bio->bi_status = error;
+ if (error)
+ bio->bi_status = error;
bio_endio(bio);
}
}
diff --git a/drivers/md/dm-bufio.c b/drivers/md/dm-bufio.c
index 414c9af54ded..80131e098650 100644
--- a/drivers/md/dm-bufio.c
+++ b/drivers/md/dm-bufio.c
@@ -565,7 +565,8 @@ static void dmio_complete(unsigned long error, void *context)
{
struct dm_buffer *b = context;
- b->bio.bi_status = error ? BLK_STS_IOERR : 0;
+ if (error)
+ b->bio.bi_status = BLK_STS_IOERR;
b->bio.bi_end_io(&b->bio);
}
@@ -614,7 +615,8 @@ static void inline_endio(struct bio *bio)
*/
bio_reset(bio);
- bio->bi_status = status;
+ if (status)
+ bio->bi_status = status;
end_fn(bio);
}
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index 8168f737590e..3e7db4a1093e 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -1488,7 +1488,8 @@ static void crypt_dec_pending(struct dm_crypt_io *io)
else
kfree(io->integrity_metadata);
- base_bio->bi_status = error;
+ if (error)
+ base_bio->bi_status = error;
bio_endio(base_bio);
}
diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index 7d3e572072f5..6f793d7d29f0 100644
--- a/drivers/md/dm-mpath.c
+++ b/drivers/md/dm-mpath.c
@@ -651,6 +651,9 @@ static int __multipath_map_bio(struct multipath *m, struct bio *bio,
mpio->pgpath = pgpath;
+ /* FIXME If the bio was chained, this could
+ * over-write an error... is that possible?
+ */
bio->bi_status = 0;
bio_set_dev(bio, pgpath->path.dev->bdev);
bio->bi_opf |= REQ_FAILFAST_TRANSPORT;
diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 629c555890c1..16d4c386cc56 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -563,7 +563,8 @@ static void error_bio_list(struct bio_list *bios, blk_status_t error)
struct bio *bio;
while ((bio = bio_list_pop(bios))) {
- bio->bi_status = error;
+ if (error)
+ bio->bi_status = error;
bio_endio(bio);
}
}
diff --git a/drivers/md/dm-verity-target.c b/drivers/md/dm-verity-target.c
index aedb8222836b..86a1d501d915 100644
--- a/drivers/md/dm-verity-target.c
+++ b/drivers/md/dm-verity-target.c
@@ -503,7 +503,8 @@ static void verity_finish_io(struct dm_verity_io *io, blk_status_t status)
struct bio *bio = dm_bio_from_per_bio_data(io, v->ti->per_io_data_size);
bio->bi_end_io = io->orig_bi_end_io;
- bio->bi_status = status;
+ if (status)
+ bio->bi_status = status;
verity_fec_finish_io(io);
diff --git a/drivers/md/dm-zoned-target.c b/drivers/md/dm-zoned-target.c
index caff02caf083..cfdacaf79492 100644
--- a/drivers/md/dm-zoned-target.c
+++ b/drivers/md/dm-zoned-target.c
@@ -637,7 +637,8 @@ static int dmz_end_io(struct dm_target *ti, struct bio *bio, blk_status_t *error
return DM_ENDIO_INCOMPLETE;
/* Done */
- bio->bi_status = bioctx->status;
+ if (bioctx->status)
+ bio->bi_status = bioctx->status;
if (bioctx->zone) {
struct dm_zone *zone = bioctx->zone;
--
2.14.0.rc0.dirty
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]
next prev parent reply other threads:[~2018-02-15 9:01 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-02-14 13:02 DM Regression in 4.16-rc1 - read() returns data when it shouldn't Milan Broz
2018-02-14 20:39 ` NeilBrown
2018-02-14 23:05 ` Mike Snitzer
2018-02-15 0:07 ` [dm-devel] " NeilBrown
2018-02-15 7:37 ` Milan Broz
2018-02-15 8:52 ` NeilBrown
2018-02-15 9:00 ` [PATCH] dm: correctly handle chained bios in dec_pending() NeilBrown
2018-02-15 9:01 ` NeilBrown [this message]
2018-02-15 9:09 ` [PATCH] block: be more careful about status in __bio_chain_endio NeilBrown
2019-02-22 21:10 ` Mike Snitzer
2019-02-22 22:46 ` Jens Axboe
2019-02-22 23:55 ` Mike Snitzer
2019-02-23 2:02 ` John Dorminy
2019-02-23 2:44 ` Mike Snitzer
2019-02-23 3:10 ` John Dorminy
2019-06-12 2:56 ` John Dorminy
2019-06-12 7:01 ` Christoph Hellwig
2019-06-17 7:32 ` Hannes Reinecke
2018-02-19 13:44 ` DM Regression in 4.16-rc1 - read() returns data when it shouldn't Thorsten Leemhuis
2018-02-19 17:15 ` Mike Snitzer
2018-02-26 10:14 ` Thorsten Leemhuis
2018-02-26 11:01 ` NeilBrown
2018-02-26 17:31 ` Thorsten Leemhuis
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87h8qipqxg.fsf@notabene.neil.brown.name \
--to=neilb@suse.com \
--cc=dm-devel@redhat.com \
--cc=gmazyland@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=snitzer@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®