All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] btrfs: don't leak extent_map in btrfs_get_io_geometry()
@ 2019-07-15 13:16 Johannes Thumshirn
  2019-07-15 13:22 ` Nikolay Borisov
  2019-07-17 14:58 ` David Sterba
  0 siblings, 2 replies; 3+ messages in thread
From: Johannes Thumshirn @ 2019-07-15 13:16 UTC (permalink / raw)
  To: David Sterba; +Cc: Nikolay Borisov, Linux BTRFS Mailinglist, Johannes Thumshirn

btrfs_get_io_geometry() calls btrfs_get_chunk_map() to acquire a reference
on a extent_map, but on normal operation it does not drop this reference
anymore.

This leads to excessive kmemleak reports.

Always call free_extent_map(), not just in the error case.

Fixes: 5f1411265e16 ("btrfs: Introduce btrfs_io_geometry infrastructure")
Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>
---
 fs/btrfs/volumes.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index a13ddba1ebc3..d74b74ca07af 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -5941,6 +5941,7 @@ int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 	u64 stripe_len;
 	u64 raid56_full_stripe_start = (u64)-1;
 	int data_stripes;
+	int ret = 0;
 
 	ASSERT(op != BTRFS_MAP_DISCARD);
 
@@ -5961,8 +5962,8 @@ int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 		btrfs_crit(fs_info,
 "stripe math has gone wrong, stripe_offset=%llu offset=%llu start=%llu logical=%llu stripe_len=%llu",
 			stripe_offset, offset, em->start, logical, stripe_len);
-		free_extent_map(em);
-		return -EINVAL;
+		ret = -EINVAL;
+		goto out;
 	}
 
 	/* stripe_offset is the offset of this block in its stripe */
@@ -6009,7 +6010,10 @@ int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 	io_geom->stripe_offset = stripe_offset;
 	io_geom->raid56_stripe_offset = raid56_full_stripe_start;
 
-	return 0;
+out:
+	/* once for us */
+	free_extent_map(em);
+	return ret;
 }
 
 static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
-- 
2.16.4


^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH] btrfs: don't leak extent_map in btrfs_get_io_geometry()
  2019-07-15 13:16 [PATCH] btrfs: don't leak extent_map in btrfs_get_io_geometry() Johannes Thumshirn
@ 2019-07-15 13:22 ` Nikolay Borisov
  2019-07-17 14:58 ` David Sterba
  1 sibling, 0 replies; 3+ messages in thread
From: Nikolay Borisov @ 2019-07-15 13:22 UTC (permalink / raw)
  To: Johannes Thumshirn, David Sterba; +Cc: Linux BTRFS Mailinglist



On 15.07.19 г. 16:16 ч., Johannes Thumshirn wrote:
> btrfs_get_io_geometry() calls btrfs_get_chunk_map() to acquire a reference
> on a extent_map, but on normal operation it does not drop this reference
> anymore.
> 
> This leads to excessive kmemleak reports.
> 
> Always call free_extent_map(), not just in the error case.
> 
> Fixes: 5f1411265e16 ("btrfs: Introduce btrfs_io_geometry infrastructure")
> Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>

My bad:

Reviewed-by: Nikolay Borisov <nborisov@suse.com>


> ---
>  fs/btrfs/volumes.c | 10 +++++++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
> 
> diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
> index a13ddba1ebc3..d74b74ca07af 100644
> --- a/fs/btrfs/volumes.c
> +++ b/fs/btrfs/volumes.c
> @@ -5941,6 +5941,7 @@ int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
>  	u64 stripe_len;
>  	u64 raid56_full_stripe_start = (u64)-1;
>  	int data_stripes;
> +	int ret = 0;
>  
>  	ASSERT(op != BTRFS_MAP_DISCARD);
>  
> @@ -5961,8 +5962,8 @@ int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
>  		btrfs_crit(fs_info,
>  "stripe math has gone wrong, stripe_offset=%llu offset=%llu start=%llu logical=%llu stripe_len=%llu",
>  			stripe_offset, offset, em->start, logical, stripe_len);
> -		free_extent_map(em);
> -		return -EINVAL;
> +		ret = -EINVAL;
> +		goto out;
>  	}
>  
>  	/* stripe_offset is the offset of this block in its stripe */
> @@ -6009,7 +6010,10 @@ int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
>  	io_geom->stripe_offset = stripe_offset;
>  	io_geom->raid56_stripe_offset = raid56_full_stripe_start;
>  
> -	return 0;
> +out:
> +	/* once for us */
> +	free_extent_map(em);
> +	return ret;
>  }
>  
>  static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
> 

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] btrfs: don't leak extent_map in btrfs_get_io_geometry()
  2019-07-15 13:16 [PATCH] btrfs: don't leak extent_map in btrfs_get_io_geometry() Johannes Thumshirn
  2019-07-15 13:22 ` Nikolay Borisov
@ 2019-07-17 14:58 ` David Sterba
  1 sibling, 0 replies; 3+ messages in thread
From: David Sterba @ 2019-07-17 14:58 UTC (permalink / raw)
  To: Johannes Thumshirn; +Cc: David Sterba, Nikolay Borisov, Linux BTRFS Mailinglist

On Mon, Jul 15, 2019 at 03:16:12PM +0200, Johannes Thumshirn wrote:
> btrfs_get_io_geometry() calls btrfs_get_chunk_map() to acquire a reference
> on a extent_map, but on normal operation it does not drop this reference
> anymore.
> 
> This leads to excessive kmemleak reports.
> 
> Always call free_extent_map(), not just in the error case.
> 
> Fixes: 5f1411265e16 ("btrfs: Introduce btrfs_io_geometry infrastructure")
> Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>

Added to 5.3 queue, thanks.

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2019-07-17 14:58 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-07-15 13:16 [PATCH] btrfs: don't leak extent_map in btrfs_get_io_geometry() Johannes Thumshirn
2019-07-15 13:22 ` Nikolay Borisov
2019-07-17 14:58 ` David Sterba

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.