3.2.94-rc1 review patch. If anyone has any objections, please let me know. ------------------ From: Ernesto A. Fernández <ernesto.mnd.fernandez@xxxxxxxxx> commit d7d824966530acfe32b94d1ed672e6fe1638cd68 upstream. When changing a file's acl mask, btrfs_set_acl() will first set the group bits of i_mode to the value of the mask, and only then set the actual extended attribute representing the new acl. If the second part fails (due to lack of space, for example) and the file had no acl attribute to begin with, the system will from now on assume that the mask permission bits are actual group permission bits, potentially granting access to the wrong users. Prevent this by restoring the original mode bits if __btrfs_set_acl fails. Signed-off-by: Ernesto A. Fernández <ernesto.mnd.fernandez@xxxxxxxxx> Reviewed-by: David Sterba <dsterba@xxxxxxxx> Signed-off-by: David Sterba <dsterba@xxxxxxxx> [bwh: Backported to 3.2: adjust context] Signed-off-by: Ben Hutchings <ben@xxxxxxxxxxxxxxx> --- --- a/fs/btrfs/acl.c +++ b/fs/btrfs/acl.c @@ -156,6 +156,7 @@ static int btrfs_xattr_acl_set(struct de int ret; struct posix_acl *acl = NULL; struct inode *inode = dentry->d_inode; + umode_t old_mode = inode->i_mode; if (!inode_owner_or_capable(inode)) return -EPERM; @@ -181,6 +182,8 @@ static int btrfs_xattr_acl_set(struct de goto out; } ret = btrfs_set_acl(NULL, inode, acl, type); + if (ret) + inode->i_mode = old_mode; out: posix_acl_release(acl);