Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230525100654.15069-1-jack@suse.cz> <20230525101624.15814-6-jack@suse.cz>
In-Reply-To: <20230525101624.15814-6-jack@suse.cz>
From:   Amir Goldstein <amir73il@gmail.com>
Date:   Fri, 26 May 2023 15:13:06 +0300
Message-ID: <CAOQ4uxhL0w+yqg2u_xW6r4J_gJX=_zoZjo3vh0ONqAbgxm2VTA@mail.gmail.com>
Subject: Re: [PATCH 6/6] fs: Restrict lock_two_nondirectories() to
 non-directory inodes
To:     Jan Kara <jack@suse.cz>
Cc:     Al Viro <viro@zeniv.linux.org.uk>, linux-fsdevel@vger.kernel.org,
        Christian Brauner <brauner@kernel.org>,
        Miklos Szeredi <miklos@szeredi.hu>,
        "Darrick J. Wong" <djwong@kernel.org>, Ted Tso <tytso@mit.edu>,
        Jaegeuk Kim <jaegeuk@kernel.org>, linux-ext4@vger.kernel.org,
        linux-f2fs-devel@lists.sourceforge.net, linux-xfs@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Thu, May 25, 2023 at 1:17=E2=80=AFPM Jan Kara <jack@suse.cz> wrote:
>
> Currently lock_two_nondirectories() is skipping any passed directories.
> After vfs_rename() uses lock_two_inodes(), all the remaining four users
> of this function pass only regular files to it. So drop the somewhat
> unusual "skip directory" logic and instead warn if anybody passes
> directory to it. This also allows us to use lock_two_inodes() in
> lock_two_nondirectories() to concentrate the lock ordering logic in less
> places.
>
> Signed-off-by: Jan Kara <jack@suse.cz>
> ---
>  fs/inode.c | 12 ++++--------
>  1 file changed, 4 insertions(+), 8 deletions(-)
>
> diff --git a/fs/inode.c b/fs/inode.c
> index 2015fa50d34a..accf5125a049 100644
> --- a/fs/inode.c
> +++ b/fs/inode.c
> @@ -1140,7 +1140,7 @@ void lock_two_inodes(struct inode *inode1, struct i=
node *inode2,
>  /**
>   * lock_two_nondirectories - take two i_mutexes on non-directory objects
>   *
> - * Lock any non-NULL argument that is not a directory.
> + * Lock any non-NULL argument. Passed objects must not be directories.
>   * Zero, one or two objects may be locked by this function.
>   *
>   * @inode1: first inode to lock
> @@ -1148,13 +1148,9 @@ void lock_two_inodes(struct inode *inode1, struct =
inode *inode2,
>   */
>  void lock_two_nondirectories(struct inode *inode1, struct inode *inode2)
>  {
> -       if (inode1 > inode2)
> -               swap(inode1, inode2);
> -
> -       if (inode1 && !S_ISDIR(inode1->i_mode))
> -               inode_lock(inode1);
> -       if (inode2 && !S_ISDIR(inode2->i_mode) && inode2 !=3D inode1)
> -               inode_lock_nested(inode2, I_MUTEX_NONDIR2);
> +       WARN_ON_ONCE(S_ISDIR(inode1->i_mode));
> +       WARN_ON_ONCE(S_ISDIR(inode2->i_mode));
> +       lock_two_inodes(inode1, inode2, I_MUTEX_NORMAL, I_MUTEX_NONDIR2);
>  }
>  EXPORT_SYMBOL(lock_two_nondirectories);
>

Need the same treatment to unlock_two_nondirectories() because now if
someone does pass a directory they will get a warning but also a leaked loc=
k.

Thanks,
Amir.