[Intel-gfx] [PATCH] drm/i915/dsc: Remove redundant checks in DSC disable

Thu Jun 3 18:45:42 UTC 2021

On Thu, Jun 03, 2021 at 08:37:22AM -0700, Kulkarni, Vandita wrote:
> > -----Original Message-----
> > From: Manna, Animesh <animesh.manna at intel.com>
> > Sent: Thursday, June 3, 2021 7:24 PM
> > To: Kulkarni, Vandita <vandita.kulkarni at intel.com>; Nikula, Jani
> > <jani.nikula at intel.com>; Saarinen, Jani <jani.saarinen at intel.com>; intel-
> > gfx at lists.freedesktop.org
> > Cc: Navare, Manasi D <manasi.d.navare at intel.com>
> > Subject: RE: [Intel-gfx] [PATCH] drm/i915/dsc: Remove redundant checks in
> > DSC disable
> > 
> > 
> > 
> > > -----Original Message-----
> > > From: Kulkarni, Vandita <vandita.kulkarni at intel.com>
> > > Sent: Thursday, June 3, 2021 4:55 PM
> > > To: Nikula, Jani <jani.nikula at intel.com>; Saarinen, Jani
> > > <jani.saarinen at intel.com>; intel-gfx at lists.freedesktop.org
> > > Cc: Manna, Animesh <animesh.manna at intel.com>; Navare, Manasi D
> > > <manasi.d.navare at intel.com>
> > > Subject: RE: [Intel-gfx] [PATCH] drm/i915/dsc: Remove redundant checks
> > > in DSC disable
> > >
> > > > -----Original Message-----
> > > > From: Nikula, Jani <jani.nikula at intel.com>
> > > > Sent: Thursday, June 3, 2021 3:11 PM
> > > > To: Kulkarni, Vandita <vandita.kulkarni at intel.com>; Saarinen, Jani
> > > > <jani.saarinen at intel.com>; intel-gfx at lists.freedesktop.org
> > > > Cc: Manna, Animesh <animesh.manna at intel.com>; Navare, Manasi D
> > > > <manasi.d.navare at intel.com>
> > > > Subject: RE: [Intel-gfx] [PATCH] drm/i915/dsc: Remove redundant
> > > > checks in DSC disable
> > > >
> > > > On Thu, 03 Jun 2021, "Kulkarni, Vandita"
> > > > <vandita.kulkarni at intel.com>
> > > > wrote:
> > > > >> -----Original Message-----
> > > > >> From: Saarinen, Jani <jani.saarinen at intel.com>
> > > > >> Sent: Thursday, June 3, 2021 1:07 PM
> > > > >> To: Kulkarni, Vandita <vandita.kulkarni at intel.com>; intel-
> > > > >> gfx at lists.freedesktop.org
> > > > >> Cc: Nikula, Jani <jani.nikula at intel.com>
> > > > >> Subject: RE: [Intel-gfx] [PATCH] drm/i915/dsc: Remove redundant
> > > > >> checks in DSC disable
> > > > >>
> > > > >> Hi,
> > > > >> > -----Original Message-----
> > > > >> > From: Intel-gfx <intel-gfx-bounces at lists.freedesktop.org> On
> > > > >> > Behalf Of Vandita Kulkarni
> > > > >> > Sent: torstai 3. kesäkuuta 2021 9.54
> > > > >> > To: intel-gfx at lists.freedesktop.org
> > > > >> > Cc: Nikula, Jani <jani.nikula at intel.com>
> > > > >> > Subject: [Intel-gfx] [PATCH] drm/i915/dsc: Remove redundant
> > > > >> > checks in DSC disable
> > > > >> >
> > > > >> > There can be a chance that pre os has enabled DSC and driver's
> > > > >> > compute config would not need dsc to be enabled, in such case
> > > > >> > if we check on compute config's compression state to disable,
> > > > >> > we might end up in state
> > > > >> mismatch.
> > > > >>
> > > > >> I assume this fixes real gitlab issue too?
> > > > > Okay, will add the tag
> > > > > Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/3537
> > > >
> > > > See https://lore.kernel.org/r/87fsxzp9qx.fsf@intel.com
> > > >
> > > > The problem is with ->bigjoiner, not the entire statement.
> > > Thanks for pointing this out, true that bigjoiner not being enabled
> > > will stop dsc disabling.
> > > The bigjoiner check was making the entire condition check unnecessary.
> > >
> > > Will update and refloat.
> > 
> > Hi Jani/Vandita,
> > 
> > For uncompressed bigjoiner case if we want to use the same function to
> > clear the dsc_ctrl1 register we may need to remove both the condition
> > check.
> > As for uncompressed bigjoiner case, compression_enable Will be 0 and will
> > block in clearing the dss_ctl1_reg.
> 
> Yes, I was going through and found that bit 20 and 21 of dss_ctl1 are being used
> for uncompressed joiner.
> So when dsc is not enabled to avoid writing the register we could add
> below code .
> 
> if (dsc)
> 	clear dss_ctl2
> if ( bigjoiner | dsc)
> 	clear dss_ctl1;
> return;
> 
> bigjoiner = 1 and dsc = 0  - uncompressed , I think there is no harm in clearing dsc bits again
> bigjoiner = 1 and dsc = 1 - compressed - uncompressed bits are already 0
> bigjoiner = 0 and dsc= 1 - just dsc  - clear dsc rest are 0s already
> bigjoiner = 0 and dsc = 0  do nothing, return
> 
> If I have missed any corner case, please let me know.
> 
> Thanks,
> Vandita

I think in the original code the condition was just reversed, instead it should be  :

if !(dsc_en || bigjoiner_en) {
	write 0 to dss ctl 1
	write 0 to dss ctl 2
}

So here basically it meets all the conditions you mentioned Vandita:

- only when both dsc and bigjoiner are 0, it will do nothing
- In all other cases DSC + Bigjoiner : Clear all bits including uncompressed bits which shd be 0 already
- In dsc = 0, bigjoiner = 1 (uncompressed), it will clear both again which is okay since dsc bits are already 0

Does this make sense?

Regards
Manasi

> > 
> > Regards,
> > Animesh
> > >
> > > Thanks,
> > > Vandita
> > > >
> > > >
> > > > BR,
> > > > Jani.
> > > >
> > > > >
> > > > > Thanks,
> > > > > Vandita
> > > > >>
> > > > >> >
> > > > >> > Signed-off-by: Vandita Kulkarni <vandita.kulkarni at intel.com>
> > > > >> > ---
> > > > >> >  drivers/gpu/drm/i915/display/intel_vdsc.c | 4 ----
> > > > >> >  1 file changed, 4 deletions(-)
> > > > >> >
> > > > >> > diff --git a/drivers/gpu/drm/i915/display/intel_vdsc.c
> > > > >> > b/drivers/gpu/drm/i915/display/intel_vdsc.c
> > > > >> > index 19cd9531c115..b05a96011d93 100644
> > > > >> > --- a/drivers/gpu/drm/i915/display/intel_vdsc.c
> > > > >> > +++ b/drivers/gpu/drm/i915/display/intel_vdsc.c
> > > > >> > @@ -1161,10 +1161,6 @@ void intel_dsc_disable(const struct
> > > > >> > intel_crtc_state
> > > > >> > *old_crtc_state)
> > > > >> >  	struct intel_crtc *crtc = to_intel_crtc(old_crtc_state-
> > >uapi.crtc);
> > > > >> >  	struct drm_i915_private *dev_priv = to_i915(crtc-
> > >base.dev);
> > > > >> >
> > > > >> > -	if (!(old_crtc_state->dsc.compression_enable &&
> > > > >> > -	      old_crtc_state->bigjoiner))
> > > > >> > -		return;
> > > > >> > -
> > > > >> >  	intel_de_write(dev_priv, dss_ctl1_reg(old_crtc_state), 0);
> > > > >> >  	intel_de_write(dev_priv, dss_ctl2_reg(old_crtc_state), 0);  }
> > > > >> > --
> > > > >> > 2.21.0.5.gaeb582a
> > > > >> >
> > > > >> > _______________________________________________
> > > > >> > Intel-gfx mailing list
> > > > >> > Intel-gfx at lists.freedesktop.org
> > > > >> > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
> > > >
> > > > --
> > > > Jani Nikula, Intel Open Source Graphics Center