Thread (11 messages) 11 messages, 2 authors, 2023-08-22

Re: [PATCH -next 3/7] md: don't rely on 'mddev->pers' to be set in mddev_suspend()

From: Yu Kuai <hidden>
Date: 2023-08-22 02:08:58
Also in: lkml

Hi,

在 2023/08/22 8:13, Song Liu 写道:
On Thu, Aug 3, 2023 at 6:30 AM Yu Kuai [off-list ref] wrote:
quoted
From: Yu Kuai <redacted>

Now that active io is initialized when mddev is allocated, it's safe to
call mddev_suspend() before 'mddev->pers' is set.

This also prevent null-ptr-def in some cases that caller doesn't
guarantee 'mddev->pers' to be set.
This description is a little confusing (to me at least). Please revise it.
Sorry about this, how about this:

'active_io' used to be initialized while the array is running, and
'mddev->pers' is set while the array is running as well. Hence caller
must hold 'reconfig_mutex' and guarantee 'mddev->pers' is set before
calling mddev_suspend().

Now that 'active_io' is initialized when mddev is allocated, such
restriction doesn't exist anymore. In the meantime, follow up patches
will refactor mddev_suspend(), hence add checking for 'mddev->pers' to
prevent null-ptr-deref.

Thanks,
Kuai
Other than that, the set looks good to me.

Thanks,
Song
quoted
Signed-off-by: Yu Kuai <redacted>
---
  drivers/md/md.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 897e94a9e47d..f14f2f0a9484 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -448,7 +448,7 @@ void mddev_suspend(struct mddev *mddev)
         set_bit(MD_ALLOW_SB_UPDATE, &mddev->flags);
         percpu_ref_kill(&mddev->active_io);

-       if (mddev->pers->prepare_suspend)
+       if (mddev->pers && mddev->pers->prepare_suspend)
                 mddev->pers->prepare_suspend(mddev);

         wait_event(mddev->sb_wait, percpu_ref_is_zero(&mddev->active_io));
--
2.39.2
.
  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help