linux-kernel - Re: NULL ptr deref in perf/filter

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: <CAOMGZ=FivHoLx_PnKg7rSNFq_fBTDsUZTsXhQAjx=NV5VfOGjA@mail.gmail.com>
Date:	Sat, 30 Jul 2016 00:36:45 +0200
From:	Vegard Nossum <vegard.nossum@...il.com>
To:	Peter Zijlstra <peterz@...radead.org>
Cc:	David Carrillo-Cisneros <davidcc@...gle.com>,
	Vineet Gupta <Vineet.Gupta1@...opsys.com>,
	Kan Liang <kan.liang@...el.com>,
	Arnaldo Carvalho de Melo <acme@...hat.com>,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: NULL ptr deref in perf/filter_match

On 29 July 2016 at 23:41, Vegard Nossum <vegard.nossum@...il.com> wrote:
> On 27 July 2016 at 16:15, Vegard Nossum <vegard.nossum@...il.com> wrote:
>> Hi,
>>
>> I'm seeing this on latest linus/master:
[...]
>> RIP: 0010:[<ffffffff81327820>]  [<ffffffff81327820>] perf_iterate_sb+0x1b0/0x6a0
[...]
>>
>> In particular, it looks to me like event->ctx is NULL.
>
> Digging a bit deeper into this, it seems the event itself is getting
> created by perf_event_open() and it gets added to the pmu_event_list
> through:
>
> perf_event_open()
>  - perf_event_alloc()
>     - account_event()
>        - account_pmu_sb_event()
>           - attach_sb_event()
>
> so at this point the event is being attached but its ->ctx is still
> NULL. It seems like ->ctx is set just a bit later in
> perf_event_open(), though.
>
> But before that, __schedule() comes along and creates a stack trace
> similar to the one above:
>
> __schedule()
>  - __perf_event_task_sched_out()
>    - perf_iterate_sb()
>      - perf_iterate_sb_cpu()
>         - event_filter_match()
>           - perf_cgroup_match()
>             - __get_cpu_context()
>               - (dereference ctx which is NULL)
>
> So I guess the question is... should the event be attached (= put on
> the list) before ->ctx gets set? Or should the cgroup code check for a
> NULL ->ctx?
>
> I'm seeing the NUL ptr deref in __perf_event_task_sched_in() as well, btw.
>
> I'm thinking this is probably where the bug was introduced:
>
> commit f2fb6bef92514432398a653df1c2f1041d79ac46
> Author: Kan Liang <kan.liang@...el.com>
> Date:   Wed Mar 23 11:24:37 2016 -0700
>
>    perf/core: Optimize side-band event delivery

Reverting aab5b71ef2b5c62323b9abe397e2db57b18e1f78 and
f2fb6bef92514432398a653df1c2f1041d79ac46 does indeed fix the issue for
me.

(Just to be clear, I'm not suggesting a revert as the final fix to
this issue, but it shows quite clearly where the problem is.)


Vegard

View attachment "0001-Revert-perf-core-Rename-the-perf_event_aux-APIs-to-p.patch" of type "text/x-patch" (8846 bytes)