[Intel-gfx] [PATCH 0/8] rps locking fixes v2

[Daniel Vetter]

On Thu, Aug 09, 2012 at 10:43:53AM +0100, Chris Wilson wrote:
> On Wed,  8 Aug 2012 23:35:32 +0200, Daniel Vetter <daniel.vetter at ffwll.ch> wrote:
> > Hi all,
> > 
> > Essentially just rebase, with Ben's review comments taking into account and one
> > WARN_ON(mutex_is_locked) moved around a bit.
> > 
> > Review&testing highly welcome.
> > 
> > Cheers, Daniel
> > 
> > Daniel Vetter (8):
> >   drm/i915: properly guard ilk ips state
> >   drm/i915: fixup up debugfs rps state handling
> >   drm/i915: move all rps state into dev_priv->rps
> >   drm/i915: kill dev_priv->mchdev_lock
> >   drm/i915: DE_PCU_EVENT irq is ilk-only
> >   drm/i915: fix up ilk drps/ips locking
> >   drm/ips: move drps/ips/ilk related variables into dev_priv->ips
> 1-7 look like good mechanical changes to improve readability and remove
> some superstition (multiple igfx perchance). Only the single comment,
> Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>
> 
> >   drm/i915: enable rc6 on ilk again
> 
> Optimist.
> 
> I think you need an ack from the guilty parties if they believe that
> the code is in good order first.

Well, I've dug around in the git history, but could only find "we have
still reports", no link nor mail address :( And my google-fu failed me,
too. But the locking fixes clearly close a hole - the pcu mbox needs
multiple writes to send out a message, and I can easily believe that we
blow up the hw if we race with different messages. So I guess it can't
hurt to retest, but since I couldn't get hold of any of the reporters I've
figured to be the optimist and just try ;-)

I'll poke Ben to at least ack it, since he last changed the ilk rps code,
he owns it ...
-Daniel
-- 
Daniel Vetter
Mail: daniel at ffwll.ch
Mobile: +41 (0)79 365 57 48