[Intel-gfx] [PATCH 1/3] drm/i915: implement new pm ops for i915

[Jesse Barnes]

On Wed, 16 Dec 2009 13:36:10 +0800
Zhenyu Wang <zhenyuw at linux.intel.com> wrote:

> On 2009.12.02 09:36:48 -0800, Jesse Barnes wrote:
> > 
> > > On 2009.11.25 13:09:37 +0800, Zhenyu Wang wrote:
> > > > One problem in i915 hibernate with current legacy pci pm ops is
> > > > that after we do freeze, we'll be forced to do resume once
> > > > again, which re-init some resources and do modesetting again,
> > > > that is unnecessary for hibernate. This patch trys to bypass
> > > > that.
> > > > 
> > > > We can't resolve this within legacy pm framework, but can do it
> > > > easily with new pm ops. Suspend (S3) process has also been kept
> > > > without change.
> > > 
> > > Any comment or concern about this one?
> > > 
> > > This patch only changes for hibernate (S4), and removes screen
> > > flicking (extra modesetting) during hibernate before power off
> > > machine. If any other device failed in the hibernate process, the
> > > driver for it should be fixed and worst case is a VT switch to
> > > bring back X.
> > 
> > Yeah sorry meant to look at this but it got lost.  It looks ok, the
> > core DRM should probably be converted too.  I had some patches for
> > that awhile back but I think the interfaces have changed since
> > then...
> > 
> 
> Looks it should be specified by driver in KMS case as always true now.
> Updated one below.
> 
> From 2caaf2f129ff1176c2aef9c7e8ad3ed04d529329 Mon Sep 17 00:00:00 2001
> From: Zhenyu Wang <zhenyuw at linux.intel.com>
> Date: Wed, 16 Dec 2009 13:30:34 +0800
> Subject: [PATCH] drm/i915: implement new pm ops for i915
> 
> One problem in i915 hibernate with current legacy pci pm ops is
> that after we do freeze, we'll be forced to do resume once again,
> which re-init some resources and do modesetting again, that is
> unnecessary for hibernate. This patch trys to bypass that.
> 
> We can't resolve this within legacy pm framework, but can do it
> easily with new pm ops. Suspend (S3) process has also been kept
> without change.
> 
> Signed-off-by: Zhenyu Wang <zhenyuw at linux.intel.com>

Looks good to me.

Reviewed-by: Jesse Barnes <jbarnes at virtuousgeek.org>

-- 
Jesse Barnes, Intel Open Source Technology Center