From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id E4D5928ED for ; Mon, 12 Dec 2022 13:35:19 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 5AA15C433D2; Mon, 12 Dec 2022 13:35:19 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1670852119; bh=OvIH+/rzbE4iIJMXArCK3MshVpQL1tOZGfQMHYX+yj0=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=yVo7LKQWIVGP5+wU2pRr2ksEdUk7Ls3h8SmZEQdxlGlUbqTH3pKW1YEi92Atwe7gd yT0wfjPHBndNGB5/eswsTg08r018RH3a1Cc32iyZ9QbgGnEjrBFKcfURaHy5spq//m mMTuNj3p58wVFinG0ui1/klytVxoWW14UBSTJRzw= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, Jann Horn , Al Viro , Sasha Levin Subject: [PATCH 6.0 015/157] fs: use acquire ordering in __fget_light() Date: Mon, 12 Dec 2022 14:16:03 +0100 Message-Id: <20221212130935.047230694@linuxfoundation.org> X-Mailer: git-send-email 2.38.1 In-Reply-To: <20221212130934.337225088@linuxfoundation.org> References: <20221212130934.337225088@linuxfoundation.org> User-Agent: quilt/0.67 Precedence: bulk X-Mailing-List: patches@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit From: Jann Horn [ Upstream commit 7ee47dcfff1835ff75a794d1075b6b5f5462cfed ] We must prevent the CPU from reordering the files->count read with the FD table access like this, on architectures where read-read reordering is possible: files_lookup_fd_raw() close_fd() put_files_struct() atomic_read(&files->count) I would like to mark this for stable, but the stable rules explicitly say "no theoretical races", and given that the FD table pointer and files->count are explicitly stored in the same cacheline, this sort of reordering seems quite unlikely in practice... Signed-off-by: Jann Horn Signed-off-by: Al Viro Signed-off-by: Sasha Levin --- fs/file.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) diff --git a/fs/file.c b/fs/file.c index 3bcc1ecc314a..57af5f8375fd 100644 --- a/fs/file.c +++ b/fs/file.c @@ -1002,7 +1002,16 @@ static unsigned long __fget_light(unsigned int fd, fmode_t mask) struct files_struct *files = current->files; struct file *file; - if (atomic_read(&files->count) == 1) { + /* + * If another thread is concurrently calling close_fd() followed + * by put_files_struct(), we must not observe the old table + * entry combined with the new refcount - otherwise we could + * return a file that is concurrently being freed. + * + * atomic_read_acquire() pairs with atomic_dec_and_test() in + * put_files_struct(). + */ + if (atomic_read_acquire(&files->count) == 1) { file = files_lookup_fd_raw(files, fd); if (!file || unlikely(file->f_mode & mask)) return 0; -- 2.35.1