[Intel-gfx] [PATCH] drm/i915/adl_p: Increase CDCLK by 15% if PSR2 is used

Lisovskiy, Stanislav stanislav.lisovskiy at intel.com
Fri Mar 18 14:22:52 UTC 2022


On Fri, Mar 18, 2022 at 02:27:53PM +0200, Souza, Jose wrote:
> On Fri, 2022-03-18 at 05:22 -0700, José Roberto de Souza wrote:
> > On Fri, 2022-03-18 at 10:52 +0200, Stanislav Lisovskiy wrote:
> > > We are currently getting FIFO underruns, in particular
> > > when PSR2 is enabled. There seem to be no existing workaround
> > > or patches, which can fix that issue(were expecting some recent
> > > selective fetch update and DBuf bw/SAGV fixes to help,
> > > which unfortunately didn't).
> > > Current idea is that it looks like for some reason the
> > > DBuf prefill time isn't enough once we exit PSR2, despite its
> > > theoretically correct.
> > > So bump it up a bit by 15%(minimum experimental amount required
> > > to get it working), if PSR2 is enabled.
> > > For PSR1 there is no need in this hack, so we limit it only
> > > to PSR2 and Alderlake.
> > 
> > It this workaround meant to be permanent? If yes we should file a HSD and get hardware folks feedback.
> > 
> > > 
> > > Signed-off-by: Stanislav Lisovskiy <stanislav.lisovskiy at intel.com>
> > > ---
> > >  drivers/gpu/drm/i915/display/intel_cdclk.c | 13 +++++++++++++
> > >  1 file changed, 13 insertions(+)
> > > 
> > > diff --git a/drivers/gpu/drm/i915/display/intel_cdclk.c b/drivers/gpu/drm/i915/display/intel_cdclk.c
> > > index 8888fda8b701..095b79950788 100644
> > > --- a/drivers/gpu/drm/i915/display/intel_cdclk.c
> > > +++ b/drivers/gpu/drm/i915/display/intel_cdclk.c
> > > @@ -2325,6 +2325,19 @@ int intel_crtc_compute_min_cdclk(const struct intel_crtc_state *crtc_state)
> > >  					dev_priv->max_cdclk_freq));
> > >  	}
> > >  
> > 
> > Please add some comment in the code about this workaround.
> > 
> > 
> > > +	if (IS_ALDERLAKE_P(dev_priv)) {
> > > +		struct intel_encoder *encoder;
> > > +
> > > +		for_each_intel_encoder_with_psr(&dev_priv->drm, encoder) {
> > > +			struct intel_dp *intel_dp = enc_to_intel_dp(encoder);
> > > +
> > > +			if (intel_dp->psr.psr2_enabled) {
> > 
> > You should check the has_psr2 in the crtc_state, PSR2 could be disabled when this state is committed.
> 
> Ah and if a remember correctly those underruns only happens in a scenario with 4 pipes enabled? or it also happens with 2 and 3 pipes?
> Anyways would be better to narrow down the cases where the workaround is applied.
> So for at least in a case with a single pipe enabled we can have the lowest cdclk as possible. 

I was thinking the same initially, but this underrun is observed in lesser pipe cases, when PSR2 
is enabled.

Stan

> 
> > 
> > > +				min_cdclk = DIV_ROUND_UP(min_cdclk * 100, 85);
> > 
> > This is not increasing by 15%.
> > 
> > min_cdclk = 500
> > 500 * 100 = 50000
> > 50000 / 85 = 588.235294118
> > 
> > While 15% of 500 is 75.
> > 
> > Also if there is two CRTCs with PSR2 enabled you will bump min_cdclk twice.
> > 
> > > +				break;
> > > +			}
> > > +		}
> > > +	}
> > > +
> > >  	if (min_cdclk > dev_priv->max_cdclk_freq) {
> > >  		drm_dbg_kms(&dev_priv->drm,
> > >  			    "required cdclk (%d kHz) exceeds max (%d kHz)\n",
> > 
> 


More information about the Intel-gfx mailing list