linux-kernel - Re: [PATCH v3 1/3] fs: Support setting a minimum fd for "lowest available fd" allocation

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:   Wed, 8 Apr 2020 20:17:08 -0700
From:   Josh Triplett <josh@...htriplett.org>
To:     Aleksa Sarai <cyphar@...har.com>
Cc:     linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
        io-uring@...r.kernel.org, linux-arch@...r.kernel.org,
        Alexander Viro <viro@...iv.linux.org.uk>,
        Arnd Bergmann <arnd@...db.de>, Jens Axboe <axboe@...nel.dk>
Subject: Re: [PATCH v3 1/3] fs: Support setting a minimum fd for "lowest
 available fd" allocation

On Wed, Apr 08, 2020 at 10:00:40PM +1000, Aleksa Sarai wrote:
> On 2020-04-07, Josh Triplett <josh@...htriplett.org> wrote:
> > Some applications want to prevent the usual "lowest available fd"
> > allocation from allocating certain file descriptors. For instance, they
> > may want to prevent allocation of a closed fd 0, 1, or 2 other than via
> > dup2/dup3, or reserve some low file descriptors for other purposes.
> > 
> > Add a prctl to increase the minimum fd and return the previous minimum.
> > 
> > System calls that allocate a specific file descriptor, such as
> > dup2/dup3, ignore this minimum.
> > 
> > exec resets the minimum fd, to prevent one program from interfering with
> > another program's expectations about fd allocation.
> 
> Why is it implemented as an "increase the value" interface? It feels
> like this is meant to avoid some kind of security trap (with a library
> reducing the value) but it means that if you want to temporarily raise
> the minimum fd number it's not possible (without re-exec()ing yourself,
> which is hardly a fun thing to do).
> 
> Then again, this might've been discussed before and I missed it...

It was: the previous version was a "get" and "set" interface. That
interface didn't allow for the possibility that something else in the
process had already set a minimum. This new atomic increase interface
(which also serves as a "get" interface if you pass 0) makes it possible
for a userspace library to reserve a range. (You have no guarantee about
previously allocated descriptors in that range, but you know that no
*new* automatically allocated descriptors will appear in that range,
which suffices; userspace can do the rest.)

- Josh Triplett