[Intel-gfx] [RFC] Runtime display PM for VLV/BYT

[Imre Deak]

On Wed, 2013-10-16 at 00:09 +0200, Daniel Vetter wrote:
> On Tue, Oct 15, 2013 at 8:15 PM, Imre Deak <imre.deak at intel.com> wrote:
> > Related to this: I made intel_encoder_get_hw_state() only check if the
> > power well is on and return false if it's not to indicate that the
> > encoder is off. I also thought of doing the same as you and take a ref
> > instead, not sure what's the right way. Maybe doing the readout only if
> > the power is on, but also making sure we have a reference in this case?
> > So with a new helper we'd have in intel_encoder_get_hw_state():
> 
> I think the approach we've quickly discussed in today's call is
> probably simplest: We grab a temporary reference to all the display
> power wells around all the dpms/modeset functions and ignore any power
> well checks on top of that. The hw will (well, should) be in the power
> on default state, so nothing should magically turn on if we don't want
> that.

Ok, I'm fine with this too for now, later it can be improved if needed. 
I added a new POWER_DOMAIN_INIT for keeping all power wells on through
driver init->first modeset and suspend->first modeset, that could be
used for this purpose too.

--Imre
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: This is a digitally signed message part
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20131016/27841385/attachment.sig>