Re: Regression on Chromebook Pixel 2015 due to i915 fastboot always-on

[Jani Nikula <jani.nikula@linux.intel.com>]

On Tue, 17 Nov 2015, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> On Tue, Nov 17, 2015 at 7:18 PM, Linus Torvalds
> <torvalds@linux-foundation.org> wrote:
>> On Tue, Nov 17, 2015 at 9:53 AM, Olof Johansson <olof@lixom.net> wrote:
>>>
>>> The problem as I see it is that it's unknown how many machines depends
>>> on previous behavior. If it's only Pixel 2015 then I think a whitelist
>>> would be just fine.
>>
>> Considering how many problems we historically have had with backlight
>> handling, I would strongly urge people to *not* start going down the
>> whitelist approach.
>>
>> If the backlight doesn't get set up correctly, the machine might as
>> well be considered dead. Very few people are going to give good
>> reports of it. So the backlight code needs to bend oevr backwards in
>> being robust even more so than most other code, and "whitelist
>> known-working setups" is absolutely the reverse of robust. It's a
>> hack, and it's guaranteed to not be maintainable.
>>
>> Yes, yes, we have whitelists for other things. I hate them in other
>> places too. But things like "this device has very odd audio
>> configuration" is very different from "this machine appears dead on
>> boot", for example.
>>
>> So reverting quickly is definitely the right thing to do. Or applying
>> the patch that apparently fixes it for Olof, and hopefully fixes it in
>> general - without any kind of random "on _this_ machine we do _that_"
>> crap.
>>
>> If drm people don't want the revert, send me a pull request with the fix.
>
> Imo revert. With all the QA awol fail we've suffered the past few
> months we've become a bit too lax imo with reverting fast, and the
> point of the split-out commit was to allow exactly that.

Based on the logs from Olof, looks like a modeset would be required to
enable backlight, instead of just cranking up the brightness. So agreed
on the revert.

> On top I don't really like the casting Maarten's current hack does, we
> probably need a per-encoder ->sanitize hook for this stuff. Better to
> retry for 4.5. Can you pls push the revert?

Moreover, you can't just enable the backlight at will, it needs to
follow the panel power sequence. You have to enable the PWM first, and
toggle the power sequencer backlight bit after that. Encoder specific
hooks can handle that. Though might still be safest to just force a
modeset on machines in weird state at driver load.

BR,
Jani.


-- 
Jani Nikula, Intel Open Source Technology Center