Re: [PATCH] jfs: fix uaf in jfs_evict_inode

[Dave Kleikamp]

On 10/31/23 12:39AM, Edward Adam Davis wrote:
> When the execution of diMount(ipimap) fails, the object ipimap that has been
> released may be accessed in diFreeSpecial(). Asynchronous ipimap release occurs
> when rcu_core() calls jfs_free_node().
>
> Therefore, when diMount(ipimap) fails, sbi->ipimap should not be initialized as
> ipimap.
>
> Reported-and-tested-by: syzbot+01cf2dbcbe2022454388@xxxxxxxxxxxxxxxxxxxxxxxxx
> Signed-off-by: Edward Adam Davis <eadavis@xxxxxx>

Looks good.

Thanks,
Shaggy

> ---
>   fs/jfs/jfs_mount.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/fs/jfs/jfs_mount.c b/fs/jfs/jfs_mount.c
> index 415eb65a36ff..9b5c6a20b30c 100644
> --- a/fs/jfs/jfs_mount.c
> +++ b/fs/jfs/jfs_mount.c
> @@ -172,15 +172,15 @@ int jfs_mount(struct super_block *sb)
>   	}
>   	jfs_info("jfs_mount: ipimap:0x%p", ipimap);
>   
> -	/* map further access of per fileset inodes by the fileset inode */
> -	sbi->ipimap = ipimap;
> -
>   	/* initialize fileset inode allocation map */
>   	if ((rc = diMount(ipimap))) {
>   		jfs_err("jfs_mount: diMount failed w/rc = %d", rc);
>   		goto err_ipimap;
>   	}
>   
> +	/* map further access of per fileset inodes by the fileset inode */
> +	sbi->ipimap = ipimap;
> +
>   	return rc;
>   
>   	/*