linux-kernel - Re: [PATCH v2 3/3] Make core

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]
Message-ID: <87lh5am8wu.fsf@x220.int.ebiederm.org>
Date:	Tue, 22 Mar 2016 17:42:41 -0500
From:	ebiederm@...ssion.com (Eric W. Biederman)
To:	Zhao Lei <zhaolei@...fujitsu.com>
Cc:	<linux-kernel@...r.kernel.org>,
	<containers@...ts.linux-foundation.org>,
	'Mateusz Guzik' <mguzik@...hat.com>,
	'Kamezawa Hiroyuki' <kamezawa.hiroyu@...fujitsu.com>
Subject: Re: [PATCH v2 3/3] Make core_pattern support namespace

Zhao Lei <zhaolei@...fujitsu.com> writes:

> Hi, Eric
>
>> -----Original Message-----
>> From: Eric W. Biederman [mailto:ebiederm@...ssion.com]
>> Sent: Tuesday, March 22, 2016 5:25 AM
>> To: Zhao Lei <zhaolei@...fujitsu.com>
>> Cc: linux-kernel@...r.kernel.org; containers@...ts.linux-foundation.org;
>> 'Mateusz Guzik' <mguzik@...hat.com>; 'Kamezawa Hiroyuki'
>> <kamezawa.hiroyu@...fujitsu.com>
>> Subject: Re: [PATCH v2 3/3] Make core_pattern support namespace
>> 
>> Zhao Lei <zhaolei@...fujitsu.com> writes:
>> 
>> > Hi, Eric
>> >
>> >> -----Original Message-----
>> >> From: Eric W. Biederman [mailto:ebiederm@...ssion.com]
>> 
>> > Let me make a summarize:
>> > You think this way is not acceptable, because the pipe program is running
>> > in the panic-process's namespace context.
>> 
>> Actually my view is that your patchset is not acceptable because it
>> is implemented in a way that is not backwards compatible (AKA it can
>> break existing configurations that remain unchanged) and your
>> implementation does not appear in the least safe from malicious users.
>> 
>> There is also a problem that your patchset is simply buggy for what it
>> tries to implement, as using pid_ns_for_children and the multiple kbuild
>> robot emails testifies.
>> 
>> > And in my view, a pipe program in the host's top level namespace is also
>> > a problem.
>> >
>> > Let us think a container, to make it act as a real machine, when a program
>> > panic, linux kernel should dump it into the container's filesystem.
>> >
>> > For the kernel, to keep the current way of forking pipe program by kthread,
>> > just let the pipe thread running in the container's namespace, instead the
>> host,
>> > may solve the problem in current kernel.
>> >
>> > What is your opinion?
>> >
>> > Btw, this patch is trying to solve the problem descripted in thread named:
>> > "piping core dump to a program escapes container" in
>> >
>> http://lists.linuxfoundation.org/pipermail/containers/2015-December/036476.
>> html
>> > Maybe using a userspace tool can make container dump to anywhere,
>> > but for kernel ifself, it is better to solve above problem if we can.
>> 
>> I think it would be great to find a way to run a core dump helper and
>> otherwise allow setting the core dump pattern in a container in a way
>> that is safe from malicious users and does not break existing setups.
>> 
> So, there is following problem:
> 1: safe from malicious users
>   We can try to find a way to fork process which have no relationship
>   with the panic process.
> 2: Bug in patch
>   It can be fixed, but I'd rather get a conclusion of this discussion
>   before fix.
> 3: Backwards compatible
>   It maybe the biggest problem in discussion, this patch is used to let
>   container dump files into container, it is different with current action.
>   Before patch:
>     File type dump_pattern: dump to container
>     Pipe type dump_pattern: dump to host
>   After patch:
>     File type dump_pattern: dump to container
>     Pipe type dump_pattern: dump to container
>   The second design seems better but not compatible with current kernel,
>   but this patch can not fix to keep compatible because it is the patch's
>   function.
>   Maybe we can make some workagound, as:
>   a. Add a kernel config to let the old style as default.
>   b. keep old style, and add "||" for core_pattern, as
>     echo "|| /root/container_dumper" >/proc/sys/kernel/core_pattern
>     to dump to container.
>
>   What is your opinion about it?

There are two parts to backwards compatibility.

1) How should core patterns that are set outside of any container be
   treated?

   The patchset under discussion imported core patterns set outside of
   containers into containers completely trivially changing their
   behavior and breaking backwards compatibility.  That is just not
   acceptable.

2) How should core patterns inside of containers be treated?

   There are corner cases that I am not thinking of that will cause
   regressions but I think we can reasonably assume that core patterns
   are not set inside of containers today.  Assuming that is true,
   then setting a core pattern inside of a container is the only thing
   that the kernel needs to detect to handle working inside of a
   container.

   The practical question I see for this case is which parent process
   needs to be used for your core dump helper, and which set of
   namespaces that parent helper has.

   I will also remind people thinking about these issues that containers
   can be run recursisvely.

Eric