linux-kernel - Re: [PATCH 0/3 v2] devpts: handle /dev/ptmx bind-mounts

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CA+55aFxH0O4tXtOux0kajrhCHGo0NcUT1jTvT-O0prfsegyp_A@mail.gmail.com>
Date:   Sun, 11 Mar 2018 14:46:26 -0700
From:   Linus Torvalds <torvalds@...ux-foundation.org>
To:     Christian Brauner <christian.brauner@...ntu.com>,
        Al Viro <viro@...iv.linux.org.uk>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        "Eric W. Biederman" <ebiederm@...ssion.com>
Subject: Re: [PATCH 0/3 v2] devpts: handle /dev/ptmx bind-mounts

On Sun, Mar 11, 2018 at 2:05 PM, Christian Brauner
<christian.brauner@...ntu.com> wrote:
>
> This is the second iteration of this patch.

This looks good to me. Just wondering how this should be merged, and
whether we should have a Cc: stable for it?

.. and, just in case, maybe Al can verify that there's nothing subtle
about follow_up() that we need to worry about. That said, NFS already
has that exact same loop for follow_to_parent(), just syntactically
slightly different version.

In fact, I wonder if we even need to do that

        if ((DEVPTS_SB(path.mnt->mnt_sb) == fsi) &&
            (path.mnt->mnt_root == fsi->ptmx_dentry)) {

and maybe we could do the follow_up() loop unconditionally?

Because if the ptmx dentry is *not* a bind mount, then the loop will
be a no-op, and if it *is* a bind-mount, then I'm not convinced we
should even try to just limit it to the devpts case - maybe somebody
did a bind-mount on just a legacy ptmx device node?

So that "if()" actually seems to be to be superfluous, and only limit
the "follow bind mounts' case unnecessarily. Hmm?

           Linus