[Intel-gfx] [PATCH] drm/i915: apply the PCI_D0/D3 hibernation workaround everywhere on pre GEN6

[Pavel Machek]

On Wed 2015-07-01 11:35:48, Jani Nikula wrote:
> On Tue, 30 Jun 2015, Pavel Machek <pavel at ucw.cz> wrote:
> > Hi!
> >
> >> commit da2bc1b9db3351addd293e5b82757efe1f77ed1d
> >> Author: Imre Deak <imre.deak at intel.com>
> >> Date:   Thu Oct 23 19:23:26 2014 +0300
> >> 
> >>     drm/i915: add poweroff_late handler
> >> 
> >> introduced a regression on old platforms during hibernation. A workaround was
> >> added in
> >> 
> >> commit ab3be73fa7b43f4c3648ce29b5fd649ea54d3adb
> >> Author: Imre Deak <imre.deak at intel.com>
> >> Date:   Mon Mar 2 13:04:41 2015 +0200
> >> 
> >>     drm/i915: gen4: work around hang during hibernation
> >> 
> >> using an explicit blacklist for the GENs/BIOS vendors where the issue was
> >> reported. Later there we had reports of the same failure on platforms not on
> >> this list.
> >> 
> >> To my best knowledge the correct thing to do is still to put the device to PCI
> >> D3 state during hibernation, see [1] and [2] for the reasons. This
> >> also aligns
> >
> > Hmm, so the reasons according to you are:
> >
> >> - ACPI mandates that the OSPM (the kernel in our case) puts all
> >>   devices
> >>   into D3 that are not wake-up sources (i915 is not) (Kudos to Ville
> >>   for
> >>   pointing this out)
> >
> > Clearly, BIOS vendors did not read this, and pretty clearly Windows
> > do not follow the specs, either. That means that it is bad idea for us
> > to follow the specs, and trigger BIOS bugs. 
> >
> >> - Embedded panels have a well defined shutdown sequence. We don't
> >>   have
> >>   any good reason to not follow this, in fact for some panels the
> >>   subsequent reinitialization could be problematic in case of a hard
> >>   power-off. (Thanks to Jani for this info)
> >
> > Please cite concrete example. I have yet to see machine that would not
> > power up on forced power down. In fact, I argue that such machine
> > would be very broken, and that such machine does not exist. While we
> > have these real machines broken:
> 
> I was originally referring to reboots, which might not be applicable
> here. Anyway, we have to go out of our way to handle that properly in
> some cases:
> 
> commit 01527b3127997ef6370d5ad4fa25d96847fbf12a
> Author: Clint Taylor <clinton.a.taylor at intel.com>
> Date:   Mon Jul 7 13:01:46 2014 -0700
> 
>     drm/i915/vlv: T12 eDP panel timing enforcement during reboot

So we now have single argument "ACPI specs says so". Unfortunately,
BIOSes can not read, and Windows does it the other way around...

What is special about gen6?
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html