[Piglit] add kernel message logging?

[Michal Suchanek]

On 1 September 2010 09:49, Michal Suchanek <hramrach at centrum.cz> wrote:
> On 1 September 2010 03:07, Dave Airlie <airlied at gmail.com> wrote:
>> On Wed, Sep 1, 2010 at 10:55 AM, Eric Anholt <eric at anholt.net> wrote:
>>> On Tue, 31 Aug 2010 16:10:09 +0200, Michal Suchanek <hramrach at centrum.cz> wrote:
>>>> Hello
>>>>
>>>> I was thinking that it would be nice if the results also logged kernel
>>>> messages genereated by the tests.
>>>>
>>>> Since the piglit tests often crash or lock up the GPU and this is
>>>> reported by the kernel it would add more useful information into the
>>>> logs.
>>>>
>>>> It should be quite easy to compare the dmesg output before and after the test.
>>>>
>>>> I don't know python so it would be quite difficult for me to add to
>>>> the test runner, unfortunately.
>>>
>>> Yuck, then when I roam wireless networks I'll get the wireless driver's
>>> spam logged in my test run?
>
> Yes, you probably will. Or you can always not log the messages.
>
>>
>> Even then, unless the kernel crash happens inline with the test its
>> hard to know if its the test that caused it.
>>
>> It could be something else on the desktop etc.
>>
>> At least for r600 we get a reset if we do something bad, but generally
>> the test also hangs pretty badly.
>>
>
> The logging is not done by the test and the test run does not stop for
> me when the GPU is reset but I have no idea what test caused it
> because I only see the screen flash and tens of tests logged.
>
> Thanks
>
> Michal
>

FWIW a patch which adds dmesg logging is at
http://repo.or.cz/w/piglit/hramrach.git/patch/0e9f0961abdd3f5df494c59cec290159a6f95c33

It logs the relevant dmesg output just after the test output for me so
it allows identifying tests which cause GPU lockups.

Thanks

Michal