Linux RAID subsystem development
 help / color / mirror / Atom feed
* [PATCH] md/raid10: preserve recovery flags on reshape abort
       [not found] <6aec6738.059f5055.2712e4.0005.GAE@google.com>
@ 2026-11-10  5:44 ` Hristos Nenkov
  2026-11-20  7:05   ` yu kuai
  0 siblings, 1 reply; 6+ messages in thread
From: Hristos Nenkov @ 2026-11-10  5:44 UTC (permalink / raw)
  To: Song Liu, linux-raid
  Cc: Yu Kuai, Li Nan, Xiao Ni, NeilBrown, dm-devel, linux-kernel,
	Alasdair Kergon, Mike Snitzer, Mikulas Patocka,
	Benjamin Marzinski, syzbot+8b22a0459202f76b4096, Hristos Nenkov

raid10_start_reshape() can fail while resizing a bitmap before it
changes any recovery bits on its success path. Its abort path nevertheless
clears the entire mddev->recovery bitmap.

dm-raid keeps MD_RECOVERY_FROZEN set across suspend and resume. If RAID10
reshape setup fails, raid_preresume() logs the failure and continues with
resume. Clearing the bitmap removes MD_RECOVERY_FROZEN and makes
raid_resume() hit its invariant warning.

Do not clear mddev->recovery from the RAID10 abort path. The error paths
reach abort before raid10_start_reshape() modifies the recovery bits, so
there is no personality-owned recovery state to roll back there.

Fixes: bb63a7019df9 ("md/raid10: resize bitmap when required during reshape.")
Reported-by: syzbot+8b22a0459202f76b4096@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=8b22a0459202f76b4096
Signed-off-by: Hristos Nenkov <nenkov2004@gmail.com>
---
 drivers/md/raid10.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 066c43154..1fd3f5f4f 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -4490,7 +4490,6 @@ static int raid10_start_reshape(struct mddev *mddev)
 	return 0;
 
 abort:
-	mddev->recovery = 0;
 	spin_lock_irq(&conf->device_lock);
 	conf->geo = conf->prev;
 	mddev->raid_disks = conf->geo.raid_disks;

base-commit: b3dfbde62cf1981c4638939c159dfd087b7a72e9
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH] md/raid10: preserve recovery flags on reshape abort
  2026-11-10  5:44 ` [PATCH] md/raid10: preserve recovery flags on reshape abort Hristos Nenkov
@ 2026-11-20  7:05   ` yu kuai
  2026-11-20 11:17     ` [PATCH v2 0/2] Fix recovery flags on failed RAID10 reshape nenkov2004
  0 siblings, 1 reply; 6+ messages in thread
From: yu kuai @ 2026-11-20  7:05 UTC (permalink / raw)
  To: Hristos Nenkov, Song Liu, linux-raid, yu kuai
  Cc: Li Nan, Xiao Ni, NeilBrown, dm-devel, linux-kernel,
	Alasdair Kergon, Mike Snitzer, Mikulas Patocka,
	Benjamin Marzinski, syzbot+8b22a0459202f76b4096

Hi,

在 2026/11/10 13:44, Hristos Nenkov 写道:
> raid10_start_reshape() can fail while resizing a bitmap before it
> changes any recovery bits on its success path. Its abort path nevertheless
> clears the entire mddev->recovery bitmap.
>
> dm-raid keeps MD_RECOVERY_FROZEN set across suspend and resume. If RAID10
> reshape setup fails, raid_preresume() logs the failure and continues with
> resume. Clearing the bitmap removes MD_RECOVERY_FROZEN and makes
> raid_resume() hit its invariant warning.
>
> Do not clear mddev->recovery from the RAID10 abort path. The error paths
> reach abort before raid10_start_reshape() modifies the recovery bits, so
> there is no personality-owned recovery state to roll back there.
>
> Fixes: bb63a7019df9 ("md/raid10: resize bitmap when required during reshape.")
> Reported-by: syzbot+8b22a0459202f76b4096@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=8b22a0459202f76b4096
> Signed-off-by: Hristos Nenkov <nenkov2004@gmail.com>
> ---
>   drivers/md/raid10.c | 1 -
>   1 file changed, 1 deletion(-)
>
> diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
> index 066c43154..1fd3f5f4f 100644
> --- a/drivers/md/raid10.c
> +++ b/drivers/md/raid10.c
> @@ -4490,7 +4490,6 @@ static int raid10_start_reshape(struct mddev *mddev)
>   	return 0;
>   
>   abort:
> -	mddev->recovery = 0;

This is not quite related to this patch, but take a look at rs_start_reshape(),
MD_RECOVERY_WAIT is set before calling pers->start_reshape(), and it's never
cleared from error path.

This should probably be fixed first, and then it's safe to remove the above line.

>   	spin_lock_irq(&conf->device_lock);
>   	conf->geo = conf->prev;
>   	mddev->raid_disks = conf->geo.raid_disks;
>
> base-commit: b3dfbde62cf1981c4638939c159dfd087b7a72e9

-- 
Thanks,
Kuai

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH v2 0/2] Fix recovery flags on failed RAID10 reshape
  2026-11-20  7:05   ` yu kuai
@ 2026-11-20 11:17     ` nenkov2004
  2026-11-20 11:17       ` [PATCH v2 1/2] dm raid: Clear MD_RECOVERY_WAIT on reshape setup failure nenkov2004
  2026-11-20 11:17       ` [PATCH v2 2/2] md/raid10: Preserve recovery flags on reshape abort nenkov2004
  0 siblings, 2 replies; 6+ messages in thread
From: nenkov2004 @ 2026-11-20 11:17 UTC (permalink / raw)
  To: linux-raid, yukuai, song
  Cc: dm-devel, linux-kernel, magiclinan, xiao, neil, agk, snitzer,
	mpatocka, bmarzins, syzbot+8b22a0459202f76b4096, Hristos Nenkov

From: Hristos Nenkov <nenkov2004@gmail.com>

Hi Kuai,

Thanks for the review. This revision implements the ordering you
suggested. Patch 1 clears MD_RECOVERY_WAIT in dm-raid on all three
reshape setup errors. Patch 2 removes the recovery-word reset from
the RAID10 abort path.

Checks: checkpatch reported no errors or warnings for either diff.
Both affected kernel objects built in WSL. Five focused userspace
checks passed for the dm-raid flag handling. I did not run a
live-kernel reshape regression test.

I used OpenAI Codex to apply the review feedback, prepare the v2
changelog and run the checks above.

Changes in v2:
- Add the dm-raid error-path fix first, as suggested by Yu Kuai.
- Keep the original RAID10 change as the second patch.

Hristos Nenkov (2):
  dm raid: Clear MD_RECOVERY_WAIT on reshape setup failure
  md/raid10: Preserve recovery flags on reshape abort

 drivers/md/dm-raid.c | 10 +++++++---
 drivers/md/raid10.c  |  1 -
 2 files changed, 7 insertions(+), 4 deletions(-)


base-commit: b3dfbde62cf1981c4638939c159dfd087b7a72e9

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH v2 1/2] dm raid: Clear MD_RECOVERY_WAIT on reshape setup failure
  2026-11-20 11:17     ` [PATCH v2 0/2] Fix recovery flags on failed RAID10 reshape nenkov2004
@ 2026-11-20 11:17       ` nenkov2004
  2026-11-20 11:22         ` sashiko-bot
  2026-11-20 11:17       ` [PATCH v2 2/2] md/raid10: Preserve recovery flags on reshape abort nenkov2004
  1 sibling, 1 reply; 6+ messages in thread
From: nenkov2004 @ 2026-11-20 11:17 UTC (permalink / raw)
  To: linux-raid, yukuai, song
  Cc: dm-devel, linux-kernel, magiclinan, xiao, neil, agk, snitzer,
	mpatocka, bmarzins, syzbot+8b22a0459202f76b4096, Hristos Nenkov

From: Hristos Nenkov <nenkov2004@gmail.com>

rs_start_reshape() sets MD_RECOVERY_WAIT to keep the sync thread
from running before the table reload. If setup, check_reshape() or
start_reshape() fails, raid_preresume() logs the error and continues
without a reshape. There is no reshape-related table reload to wait for.

Clear MD_RECOVERY_WAIT on those error paths. Keep the bit set when
reshape setup succeeds.

Suggested-by: Yu Kuai <yukuai@fygo.io>
Link: https://lore.kernel.org/r/4d13edef-f454-437b-8910-3a905697303e@fygo.io
Signed-off-by: Hristos Nenkov <nenkov2004@gmail.com>
---
 drivers/md/dm-raid.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/drivers/md/dm-raid.c b/drivers/md/dm-raid.c
index ac019a0f5..e132951f3 100644
--- a/drivers/md/dm-raid.c
+++ b/drivers/md/dm-raid.c
@@ -4004,7 +4004,7 @@ static int rs_start_reshape(struct raid_set *rs)
 
 	r = rs_setup_reshape(rs);
 	if (r)
-		return r;
+		goto clear_recovery_wait;
 
 	/*
 	 * Check any reshape constraints enforced by the personalility
@@ -4014,7 +4014,7 @@ static int rs_start_reshape(struct raid_set *rs)
 	r = pers->check_reshape(mddev);
 	if (r) {
 		rs->ti->error = "pers->check_reshape() failed";
-		return r;
+		goto clear_recovery_wait;
 	}
 
 	/*
@@ -4025,7 +4025,7 @@ static int rs_start_reshape(struct raid_set *rs)
 		r = pers->start_reshape(mddev);
 		if (r) {
 			rs->ti->error = "pers->start_reshape() failed";
-			return r;
+			goto clear_recovery_wait;
 		}
 	}
 
@@ -4037,6 +4037,10 @@ static int rs_start_reshape(struct raid_set *rs)
 	rs_update_sbs(rs);
 
 	return 0;
+
+clear_recovery_wait:
+	clear_bit(MD_RECOVERY_WAIT, &mddev->recovery);
+	return r;
 }
 
 static int raid_preresume(struct dm_target *ti)

^ permalink raw reply related	[flat|nested] 6+ messages in thread

* [PATCH v2 2/2] md/raid10: Preserve recovery flags on reshape abort
  2026-11-20 11:17     ` [PATCH v2 0/2] Fix recovery flags on failed RAID10 reshape nenkov2004
  2026-11-20 11:17       ` [PATCH v2 1/2] dm raid: Clear MD_RECOVERY_WAIT on reshape setup failure nenkov2004
@ 2026-11-20 11:17       ` nenkov2004
  1 sibling, 0 replies; 6+ messages in thread
From: nenkov2004 @ 2026-11-20 11:17 UTC (permalink / raw)
  To: linux-raid, yukuai, song
  Cc: dm-devel, linux-kernel, magiclinan, xiao, neil, agk, snitzer,
	mpatocka, bmarzins, syzbot+8b22a0459202f76b4096, Hristos Nenkov

From: Hristos Nenkov <nenkov2004@gmail.com>

raid10_start_reshape() can fail before it updates the recovery flags.
Its abort path clears the entire recovery word, including the
MD_RECOVERY_FROZEN bit that dm-raid requires until raid_resume().

Keep the recovery word intact when rolling back reshape setup. The
preceding dm-raid change clears MD_RECOVERY_WAIT on failure in the
caller that set it.

Fixes: bb63a7019df9 ("md/raid10: resize bitmap when required during reshape.")
Reported-by: syzbot+8b22a0459202f76b4096@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=8b22a0459202f76b4096
Signed-off-by: Hristos Nenkov <nenkov2004@gmail.com>
---
 drivers/md/raid10.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 066c43154..1fd3f5f4f 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -4490,7 +4490,6 @@ static int raid10_start_reshape(struct mddev *mddev)
 	return 0;
 
 abort:
-	mddev->recovery = 0;
 	spin_lock_irq(&conf->device_lock);
 	conf->geo = conf->prev;
 	mddev->raid_disks = conf->geo.raid_disks;

^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH v2 1/2] dm raid: Clear MD_RECOVERY_WAIT on reshape setup failure
  2026-11-20 11:17       ` [PATCH v2 1/2] dm raid: Clear MD_RECOVERY_WAIT on reshape setup failure nenkov2004
@ 2026-11-20 11:22         ` sashiko-bot
  0 siblings, 0 replies; 6+ messages in thread
From: sashiko-bot @ 2026-11-20 11:22 UTC (permalink / raw)
  To: nenkov2004, linux-raid, yukuai, song
  Cc: dm-devel, magiclinan, xiao, neil, agk, snitzer, mpatocka,
	bmarzins, syzbot+8b22a0459202f76b4096

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Low] dm raid: missing Fixes trailer for rs_start_reshape() error
  cleanup
--

commit 6a83155083baad91c84f61814a141b78b22b8537
Author: Hristos Nenkov <nenkov2004@gmail.com>

> dm raid: Clear MD_RECOVERY_WAIT on reshape setup failure
>
> rs_start_reshape() sets MD_RECOVERY_WAIT to keep the sync thread
> from running before the table reload. If setup, check_reshape() or
> start_reshape() fails, raid_preresume() logs the error and continues
> without a reshape. There is no reshape-related table reload to wait for.
>
> Clear MD_RECOVERY_WAIT on those error paths. Keep the bit set when
> reshape setup succeeds.
>
> Suggested-by: Yu Kuai <yukuai@fygo.io>
> Link: https://lore.kernel.org/r/4d13edef-f454-437b-8910-3a905697303e@fygo.io
> Signed-off-by: Hristos Nenkov <nenkov2004@gmail.com>

[Severity: Low]
Commit 644e2537fdc7 ("dm raid: fix stripe adding reshape deadlock") added
set_bit(MD_RECOVERY_WAIT, &mddev->recovery) at the start of
rs_start_reshape() without clearing it on the error return paths.  Should
this include:

  Fixes: 644e2537fdc7 ("dm raid: fix stripe adding reshape deadlock")

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261120111757.4705-1-nenkov2004@gmail.com?part=1

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2026-11-20 11:22 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <6aec6738.059f5055.2712e4.0005.GAE@google.com>
2026-11-10  5:44 ` [PATCH] md/raid10: preserve recovery flags on reshape abort Hristos Nenkov
2026-11-20  7:05   ` yu kuai
2026-11-20 11:17     ` [PATCH v2 0/2] Fix recovery flags on failed RAID10 reshape nenkov2004
2026-11-20 11:17       ` [PATCH v2 1/2] dm raid: Clear MD_RECOVERY_WAIT on reshape setup failure nenkov2004
2026-11-20 11:22         ` sashiko-bot
2026-11-20 11:17       ` [PATCH v2 2/2] md/raid10: Preserve recovery flags on reshape abort nenkov2004

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox