Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 24 May 2012 21:39:09 +0200
From: "Indan Zupancic" <indan@....nu>
To: "H. Peter Anvin" <hpa@...or.com>
Cc: "Roland McGrath" <mcgrathr@...gle.com>,
 "Will Drewry" <wad@...omium.org>,
 linux-kernel@...r.kernel.org,
 netdev@...isplace.org,
 linux-security-module@...r.kernel.org,
 kernel-hardening@...ts.openwall.com,
 mingo@...hat.com,
 oleg@...hat.com,
 peterz@...radead.org,
 rdunlap@...otime.net,
 tglx@...utronix.de,
 luto@....edu,
 serge.hallyn@...onical.com,
 pmoore@...hat.com,
 akpm@...ux-foundation.org,
 corbet@....net,
 markus@...omium.org,
 coreyb@...ux.vnet.ibm.com,
 keescook@...omium.org,
 viro@...iv.linux.org.uk,
 jmorris@...ei.org
Subject: Re: [RFC PATCH 0/3] move the secure_computing call

On Thu, May 24, 2012 20:45, H. Peter Anvin wrote:
> On 05/24/2012 11:27 AM, Indan Zupancic wrote:
>>
>> If so, then the seccomp check needs to be redone after any ptrace
>> changes, or we should give up and just do the seccomp check first,
>> instead of possibly looping forever. PTRACE_EVENT_SECCOMP has the
>> same problem.
>>
>> If a seccomp filtered task can do ptrace(), it can easily bypass
>> the seccomp filter by ptracing any task not under the same filter
>> but from the same user. And then it can puppeteer the victim into
>> doing anything it wishes. So pretending seccomp can make a ptracer
>> secure is futile, I think. Perhaps it's better to keep it simple and
>> always do the seccomp test first and ignore ptrace changes, however
>> sad that may seem. Seccomp had the power to stop ptrace(). It didn't,
>> so it shouldn't try to do it afterwards either.
>>
>> It's a bit fuzzy though, only reason why doing seccomp first is more
>> convenient is because seccomp can generate ptrace events. I don't
>> think it will make a difference in practice because ptrace(2) won't
>> be allowed by seccomp filters anyway, so it's a bit of a theoretical
>> problem.
>>
>
> No, that's not the reason to do seccomp first.  The reason to do seccomp
> first is that a seccomp filter can be part of the process execution and
> can completely transform the system call picture.

How? All that seccomp can do is deny system calls and kill the task.
What you describe sounds more like ptrace.

>
> Consider UML, for example.  It uses ptrace to capture system calls and
> execute them on the behalf of the process.  It needs to know what system
> calls *actually* are done by the virtual process.

Seccomp can't change system calls, only prevent them, so I miss how it
can change anything for UML in that way.

>
> (Note: that being said, UML might very well be better done using seccomp
> filters *instead* of ptrace, but that's another matter.)

Well, obviously it will use seccomp filters instead of ptrace when possible
and do the more complicated stuff via PTRACE_SECCOMP_EVENT when seccomp isn't
sufficient. But mainly for performance reasons.

>
> I agree with you, if the process is traceable it is rather questionable
> to claim any kind of security; more likely consider that a debugging
> mode and tell people to lock out ptrace for real sandboxing.

"process is traceable" should be "process is able to trace".

Greetings,

Indan


Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.