[Intel-gfx] [PATCH 0/4] drm/i915/dp: native aux read return value fixes and cleanup

[Daniel Vetter]

On Thu, Jan 30, 2014 at 4:05 PM, Jani Nikula <jani.nikula at intel.com> wrote:
> On Thu, 30 Jan 2014, Daniel Vetter <daniel at ffwll.ch> wrote:
>> On Thu, Jan 30, 2014 at 11:37:00AM +0200, Jani Nikula wrote:
>>> Fix and cleanup the rather confused native aux read return values and
>>> return value checks. Maybe we'll get them right from now on. Just maybe.
>>>
>>> Did not have a chance to test this, I'm afraid, but it's pretty
>>> straightforward.
>>
>> How does this fit in with Thierry's dp aux helper improvements? I prefer
>> if we start to converge on that sooner than later ...
>
> For the i2c-over-aux conversion
>
> commit 8a5e6aeb30ecaf8f11a99c0d008c8935cd6fba9f
> Author: Paulo Zanoni <paulo.r.zanoni at intel.com>
> Date:   Wed Oct 30 19:50:26 2013 -0200
>
>     drm/i915: turn the eDP VDD on for any i2c transactions
>
> will be a PITA, as we lose control of that level in the stack.
>
> So we'd need to move the vdd enable/disable higher or lower. Per that
> commit, higher was problematic. If we move it lower, we may need to add
> refcounting to the vdd, but then we'll lose the warn for "eDP VDD
> already requested on\n".
>
> Ideas welcome.

Nasty. Although this was just to make the /dev i2c access work from
userspace, so I don't expect more places to pop up. Hence I think a
panel_vdd_force_on which returns the current state + panel_vdd_restore
to undo that in the low-level dp aux to be sufficient. But not sure.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch