diff mbox

btrfs: add missing initialization in btrfs_check_shared

Message ID 20180314150311.20152-2-enadolski@suse.com (mailing list archive)
State New, archived
Headers show

Commit Message

Edmund Nadolski March 14, 2018, 3:03 p.m. UTC
btrfs_check_shared calls find_parent_nodes in a loop. In each
iteration it passes in a share_check struct to request a check for
a shared extent.  The share_check::share_count must be re-initialized
to zero for each iteration in order to avoid using a stale count
value from the previous iteration, thus causing a false
shared extent indication.

Signed-off-by: Edmund Nadolski <enadolski@suse.com>
---
 fs/btrfs/backref.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Nikolay Borisov March 14, 2018, 3:09 p.m. UTC | #1
On 14.03.2018 17:03, Edmund Nadolski wrote:
> btrfs_check_shared calls find_parent_nodes in a loop. In each
> iteration it passes in a share_check struct to request a check for
> a shared extent.  The share_check::share_count must be re-initialized
> to zero for each iteration in order to avoid using a stale count
> value from the previous iteration, thus causing a false
> shared extent indication.
> 
> Signed-off-by: Edmund Nadolski <enadolski@suse.com>

Is there a particular commit which introduced this regression (i.e. one
of the backref walking series)? It would be useful to know so that this
can be tagged with a Fixes: tag and perhaps stable.

Also your cover letter should go as a changelog in the patch so that we
have a proper history of how this fix came to be :)

> ---
>  fs/btrfs/backref.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
> index 4e89598..4a33448 100644
> --- a/fs/btrfs/backref.c
> +++ b/fs/btrfs/backref.c
> @@ -1519,6 +1519,7 @@ int btrfs_check_shared(struct btrfs_root *root, u64 inum, u64 bytenr)
>  		if (!node)
>  			break;
>  		bytenr = node->val;
> +		shared.share_count = 0;
>  		cond_resched();
>  	}
>  
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Sterba March 14, 2018, 3:24 p.m. UTC | #2
On Wed, Mar 14, 2018 at 09:03:11AM -0600, Edmund Nadolski wrote:
> btrfs_check_shared calls find_parent_nodes in a loop. In each
> iteration it passes in a share_check struct to request a check for
> a shared extent.  The share_check::share_count must be re-initialized
> to zero for each iteration in order to avoid using a stale count
> value from the previous iteration, thus causing a false
> shared extent indication.
> 
> Signed-off-by: Edmund Nadolski <enadolski@suse.com>

I've replaced the changelog with text from cover letter and added the
Fixes: reference to 3ec4d3238ab that introduces 'shared'.
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index 4e89598..4a33448 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -1519,6 +1519,7 @@  int btrfs_check_shared(struct btrfs_root *root, u64 inum, u64 bytenr)
 		if (!node)
 			break;
 		bytenr = node->val;
+		shared.share_count = 0;
 		cond_resched();
 	}