[Intel-gfx] [PATCH 2/3] drm/i915: Stop sleeping from inside gen6_bsd_submit_request()

[Chris Wilson]

On Mon, Apr 10, 2017 at 05:23:35PM +0200, Michal Wajdeczko wrote:
> On Mon, Apr 10, 2017 at 04:02:06PM +0100, Chris Wilson wrote:
> > submit_request() is called from an atomic context, it's not allowed to
> > sleep. We have to be careful in our parameters to
> > intel_uncore_wait_for_register() to limit ourselves to the atomic wait
> > loop and not incur the wrath of our warnings.
> > 
> > Fixes: 6976e74b5fa1 ("drm/i915: Don't allow overuse of __intel_wait_for_register_fw()")
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > Cc: Michal Wajdeczko <michal.wajdeczko at intel.com>
> > Cc: Joonas Lahtinen <joonas.lahtinen at linux.intel.com>
> > Link: http://patchwork.freedesktop.org/patch/msgid/20170410143807.22725-1-chris@chris-wilson.co.uk
> > ---
> >  drivers/gpu/drm/i915/intel_ringbuffer.c | 10 +++++-----
> >  1 file changed, 5 insertions(+), 5 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
> > index c98acc27279a..331da59a1eb5 100644
> > --- a/drivers/gpu/drm/i915/intel_ringbuffer.c
> > +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
> > @@ -1729,11 +1729,11 @@ static void gen6_bsd_submit_request(struct drm_i915_gem_request *request)
> >  	I915_WRITE64_FW(GEN6_BSD_RNCID, 0x0);
> >  
> >  	/* Wait for the ring not to be idle, i.e. for it to wake up. */
> > -	if (intel_wait_for_register_fw(dev_priv,
> > -				       GEN6_BSD_SLEEP_PSMI_CONTROL,
> > -				       GEN6_BSD_SLEEP_INDICATOR,
> > -				       0,
> > -				       50))
> > +	if (__intel_wait_for_register_fw(dev_priv,
> > +					 GEN6_BSD_SLEEP_PSMI_CONTROL,
> > +					 GEN6_BSD_SLEEP_INDICATOR,
> > +					 0,
> > +					 1000, 0, NULL))
> 
> Hmm, it's now 50x smaller timeout, but let's hope it's still enough, hence

We have already told the device to leave rc6 (we hold forcewake) so this
should be a formality. For the last few months we have been running with
a sleep after 2us, and have never seen the warning about sleeping from
here in CI -- so I think a 1ms (i.e. 500x longer) timeout is safe.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre