[git pull] drm for v4.17-rc1

[Daniel Vetter]

On Tue, Apr 3, 2018 at 3:23 PM, Jani Nikula <jani.nikula at linux.intel.com> wrote:
> On Tue, 03 Apr 2018, Lucas Stach <dev at lynxeye.de> wrote:
>> My _feeling_ is that the review economy in drm-misc, which gets DRM the
>> bragging rights of 80% reviewed patches, has already lowered the weight
>> associated with those reviews, as most of them are really shallow. This
>> might be okay with you and I'm certainly not trying to change the way
>> drm-misc is handled, but I doubt that this is the universal gold
>> standard which should be applied to everything.
>
> I think you need to substantiate your claims about rubber stamping
> reviews. I'm not seeing that. And I do pay attention to the reviews that
> happen on i915 and drm display parts, kind of review-of-review. I'm
> personally pretty diligent about review, and I'm honestly *more* ashamed
> of patches I reviewed regressing than patches I wrote. Looking around, I
> don't think I'm alone.

Yeah, a thing to keep in mind is that we've had this "forced review"
in drm-intel since well over 5 years (so much longer than commit
rights). And the same rules apply to any core drm patches that get
merged into drm-misc. Similar for amd drivers, and Dave Airlie as
subsystem maintainer. Small drivers are explictly excempt from strict
review requirements in drm-misc, there we just want an Acked-by: to
signal a 2nd person at least looked at the patch. But even there a lot
of the patches got through full review scrutinies, with a bunch of
revisions until the patch is right. And all this has defacto run on a
"you review mine, I review yours" review economy all this time.

That amounts to 50+ people, some who've done this for 5+ years, you
accused of doing rubber stamp reviews. I agree with Jani that this
deserves some more concrete data than your personal feelings.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch