Re: [PATCH 1/2] md/raid0: Free r0conf memory when register integrity failed

On Fri, Dec 10, 2021 at 9:22 AM Guoqing Jiang [off-list ref] wrote:

On 12/10/21 2:02 AM, Song Liu wrote:

quoted

On Wed, Dec 8, 2021 at 9:55 PM Xiao Ni[off-list ref]  wrote:

quoted

It doesn't free memory when register integrity failed. And move
free conf codes into a single function.

Signed-off-by: Xiao Ni<redacted>
---
  drivers/md/raid0.c | 18 +++++++++++++++---
  1 file changed, 15 insertions(+), 3 deletions(-)

diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index 62c8b6adac70..3fa47df1c60e 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c

@@ -356,6 +356,7 @@ static sector_t raid0_size(struct mddev *mddev, sector_t sectors, int raid_disks
         return array_sectors;
  }

+static void free_conf(struct r0conf *conf);
  static void raid0_free(struct mddev *mddev, void *priv);

  static int raid0_run(struct mddev *mddev)

@@ -413,19 +414,30 @@ static int raid0_run(struct mddev *mddev)
         dump_zones(mddev);

         ret = md_integrity_register(mddev);
+       if (ret)
+               goto free;

         return ret;
+
+free:
+       free_conf(conf);

Can we just use raid0_free() here? Also, after freeing conf,
we should set mddev->private to NULL.

Agree, like what raid1_run did. And we might need to check the
return value of pers->run in level_store as well.

Yes. It needs to check the return value and try to reback to the original
state. I planed to fix this in the following days not this patch. This patch
only fix the NULL reference problem after reshape.

In fact, no only we need to check pers->run in level_store, we also need
to check integrity during reshape. Now integrity only supports
raid0/raid1/raid10,
so it needs to do some jobs related with integrity (unregister/register)

I plan to fix these two problems in the next patches. Is it OK?

Regards
Xiao