[Intel-gfx] [PATCH 5/5] drm/i915: Add SPLL sharing support for DDI

[Daniel Vetter]

On Mon, Mar 03, 2014 at 11:56:08AM +0200, Ville Syrjälä wrote:
> On Fri, Feb 28, 2014 at 06:40:22PM -0300, Paulo Zanoni wrote:
> > Hi
> > 
> > 2014-02-27 9:23 GMT-03:00  <ville.syrjala at linux.intel.com>:
> > > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > >
> > > Write some theoretical SPLL sharing support for DDI. Currently that will
> > > never happens since we never use SPLL for anything but FDI. But having
> > > the code there makes it easier if we ever want to do it, and it might
> > > excercise the PLL sharing code a bit more.
> > 
> > One thing is that if we ever want to use SSC on eDP/DP/HDMI, we will
> > want to use either SPLL or WRPLL instead of LCPLL. I always wanted to
> > implement this, and recently I saw a bug that made me think about it
> > again.
> > 
> > So instead of this patch, why don't you write support to use SSC when
> > requested? We could do this through some i915.use_ssc command line
> > option that would just try to force SSC on everything, or maybe a
> > "debug connector property", or debugfs, or anything you might want.
> > With this, you would be able to actually test your patches, and we'd
> > have a useful feature.
> > 
> > If you disagree, please just say "no" and then I'll go back to review
> > this patch :)
> 
> I think improving the PLL selection code should be a separate patch.
> IIRC I even said as much in the cover letter.

wrt improved pll selection logic I'd welcome if we could port the ddi code
and refcounting to use the somewhat generic stuff I've created for pch
plls. There's not much logic in there really, but the one thing which is
abstracted is the refcount reconstruction for the hw state readout and
cross-checking.

That part is devilishly tricky to get right apparently (it took us a
pile of iterations to get there for both the pch and ddi pll code), so imo
the right thing to do before making this more complicated.

I've merged the two fixes from this series and punted on the others for
now.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch