[Intel-gfx] Pushing commits to drm-intel-next-queued and questions about CI

[Hans de Goede]

Hi,

On 03-11-17 18:40, Rodrigo Vivi wrote:
> Hi Hans,
> 
> On Fri, Nov 03, 2017 at 10:41:55AM +0000, Hans de Goede wrote:
>> Hi Daniel,
>>
>> A while ago I was granted commit rights to the drm-intel repo.
>> So far I've not used these, but since no-one seems to be
>> pushing these 2 sets, I guess now might be a good time to
>> learn how to push things myself.
>>
>> I'm talking about these 2 patch-sets:
>>
>> https://patchwork.freedesktop.org/series/32274/
> 
> For this one we got the ack from Ingo, but I missed the reviewed-by
> on the second one that is the i915 part of it.
> Do we need it or just that ack was enough for both patches?

With "this one" I assume you mean the one below:

>> https://patchwork.freedesktop.org/series/32288/

As that is the one which has an ack from Ingo, the second
one has had a reviewed-by from Imre Deak for a few
versions already, it is right there in patch work.

Thank you for rescheduling the tests, this one passes
Fi.CI.BAT now, but I did not see Fi.CI.IGD getting run?


As for the other series:

https://patchwork.freedesktop.org/series/32274/

That one passed Fi.CI.BAT before (and had warnings
in Fi.CI.IGD which seem to be normal as they happen more
often) but now failed in Fi.CI.BAT, with almost all tests
for fi-gdg-551 failing and everything else succeeeding.

> ... I just triggered both retests for you here...
>
>> This means the other tests never ran, so I guess
>> I should reschedule the tests for this set.
>>
>> Given that all the patches have been reviewed and acked can
>> I simply push these to drm-intel-next-queued once the
>> tests results are sorted out?
> 
> If you have permission and if you are follow the guidelines
> and docs, you know what you are doing ;)

I'm not entirely sure I know what I'm doing, hence this mail :)

My understanding is: that once all patches are reviewed and
a set passes Fi.CI.BAT and Fi.CI.IGD it can be pushed to
drm-intel-next-queued, is that correct?

And what about pesky CI failures ?  Note I do see the usefulness
of CI, it did catch a real bug in another patch-set of mine,
just wondering how to deal with CI given that its results
seem to be varying...

Regards,

Hans