Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <1542942953-93562-1-git-send-email-bianpan2016@163.com>
In-Reply-To: <1542942953-93562-1-git-send-email-bianpan2016@163.com>
From:   Amir Goldstein <amir73il@gmail.com>
Date:   Fri, 23 Nov 2018 07:58:15 +0200
Message-ID: <CAOQ4uxj74Xh6ZbJp7zPZOxOY-gkrm5K1yU31+DW1YMYzOWtKgA@mail.gmail.com>
Subject: Re: [PATCH] exportfs: do not read dentry after free
To:     bianpan2016@163.com
Cc:     Miklos Szeredi <mszeredi@redhat.com>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        Al Viro <viro@zeniv.linux.org.uk>,
        "J. Bruce Fields" <bfields@fieldses.org>,
        Christoph Hellwig <hch@lst.de>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Nov 23, 2018 at 5:16 AM Pan Bian <bianpan2016@163.com> wrote:
>
> The function dentry_connected calls dput(dentry) to drop the previously
> acquired reference to dentry. In this case, dentry can be released.
> After that, IS_ROOT(dentry) checks the condition
> (dentry == dentry->d_parent), which may result in a use-after-free bug.
> This patch directly compares dentry with its parent obtained before
> dropping the reference.
>
> Fixes: a056cc8934c("exportfs: stop retrying once we race with
> rename/remove")
>

CC Fixes patch author/reviewers

How did you find this? by code review or did this actually happen?

Normally a IS_ROOT dentry would be either DCACHE_DISCONNECTED or
pinned to some super block, but I guess there may be corner cases?

> Signed-off-by: Pan Bian <bianpan2016@163.com>
> ---
>  fs/exportfs/expfs.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/exportfs/expfs.c b/fs/exportfs/expfs.c
> index 645158d..a69aaf5 100644
> --- a/fs/exportfs/expfs.c
> +++ b/fs/exportfs/expfs.c
> @@ -77,7 +77,7 @@ static bool dentry_connected(struct dentry *dentry)
>                 struct dentry *parent = dget_parent(dentry);
>
>                 dput(dentry);
> -               if (IS_ROOT(dentry)) {
> +               if (dentry == parent) { /* is root entry */
>                         dput(parent);
>                         return false;
>                 }

The change itself looks right, but the name IS_ROOT is confusing
enough as it is. The explicit comment is just plain wrong.
If it was really a root dentry, it wouldn't have been DCACHE_DISCONNECTED
(unless it is a filesystem bug).

Thanks,
Amir.