Re: [PATCH] staging: tidspbridge: protect dmm_map properly

From: Felipe Contreras
Date: Mon Mar 07 2011 - 16:48:41 EST


On Mon, Mar 7, 2011 at 9:29 PM, Ramirez Luna, Omar <omar.ramirez@xxxxxx> wrote:
> On Mon, Mar 7, 2011 at 12:02 PM, Felipe Contreras
> <felipe.contreras@xxxxxxxxx> wrote:
>> On Mon, Dec 20, 2010 at 7:12 PM, Felipe Contreras
>> <felipe.contreras@xxxxxxxxx> wrote:
>>> We need to protect not only the dmm_map list, but the individual
>>> map_obj's, otherwise, we might be building the scatter-gather list with
>>> garbage. So, use the existing proc_lock for that.
>>>
>>> I observed race conditions which caused kernel panics while running
>>> stress tests. This patch fixes those.
>>
>> I just heard that Tuomas Kulve is getting a lot of panics on Gumstix
>> Overo. I propose we apply this patch on the stable tree ASAP, and if
>> there's no better proposals, also on .38.
>
> Can you or Tuomas share the bug report data (panic log, test case
> maybe)? I would like to discard issues affected by timing that could
> be hidden with this patch.

I got this from Tuomas:
http://pastie.org/1643677

It seems it's very easy to reproduce on Gumstix Overo with a
gst-launch command. The way I reproduced it was very tedious; running
a full blown Maemo test suite with GBs of clips.

I think the issue is very clear; if you build a sg list with garbage
memory, problems are expected =/

> I agree that for the time being this needs to be sent upstream, even
> if in paper Ohad's patch solves the issue without side effects of
> locking.

Perhaps. I don't remember if I ack'ed Ohad's patch, but even if it's
ok, I think it can be applied on top of my patch.

Cheers.

--
Felipe Contreras
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/