[Intel-gfx] [PATCH 2/4] drm/i915: Unset reset pch handshake when PCH is not present in one place

[Souza, Jose]

On Thu, 2018-09-13 at 14:51 -0700, Rodrigo Vivi wrote:
> On Thu, Sep 13, 2018 at 02:22:49PM -0700, José Roberto de Souza
> wrote:
> > Right now RESET_PCH_HANDSHAKE_ENABLE is enabled all the times
> > inside
> > of intel_power_domains_init_hw() and if PCH is NOP it is unsed in
> > i915_gem_init_hw().
> > So making skl_pch_reset_handshake() handle both cases and calling
> > it for the missing gens in intel_power_domains_init_hw().
> > Ivybridge have a different register and bits but with the same
> > objective so moving it too.
> > 
> > Signed-off-by: José Roberto de Souza <jose.souza at intel.com>
> > ---
> >  drivers/gpu/drm/i915/i915_gem.c         | 12 ------------
> >  drivers/gpu/drm/i915/intel_runtime_pm.c | 12 +++++++++++-
> >  2 files changed, 11 insertions(+), 13 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/i915_gem.c
> > b/drivers/gpu/drm/i915/i915_gem.c
> > index 89834ce19acd..b389e084c8c6 100644
> > --- a/drivers/gpu/drm/i915/i915_gem.c
> > +++ b/drivers/gpu/drm/i915/i915_gem.c
> > @@ -5296,18 +5296,6 @@ int i915_gem_init_hw(struct drm_i915_private
> > *dev_priv)
> >  		I915_WRITE(MI_PREDICATE_RESULT_2, IS_HSW_GT3(dev_priv)
> > ?
> >  			   LOWER_SLICE_ENABLED : LOWER_SLICE_DISABLED);
> >  
> > -	if (HAS_PCH_NOP(dev_priv)) {
> > -		if (IS_IVYBRIDGE(dev_priv)) {
> > -			u32 temp = I915_READ(GEN7_MSG_CTL);
> > -			temp &= ~(WAIT_FOR_PCH_FLR_ACK |
> > WAIT_FOR_PCH_RESET_ACK);
> > -			I915_WRITE(GEN7_MSG_CTL, temp);
> > -		} else if (INTEL_GEN(dev_priv) >= 7) {
> > -			u32 temp = I915_READ(HSW_NDE_RSTWRN_OPT);
> > -			temp &= ~RESET_PCH_HANDSHAKE_ENABLE;
> > -			I915_WRITE(HSW_NDE_RSTWRN_OPT, temp);
> > -		}
> > -	}
> > -
> >  	intel_gt_workarounds_apply(dev_priv);
> >  
> >  	i915_gem_init_swizzling(dev_priv);
> > diff --git a/drivers/gpu/drm/i915/intel_runtime_pm.c
> > b/drivers/gpu/drm/i915/intel_runtime_pm.c
> > index 8bcb33367d0d..369a292cafac 100644
> > --- a/drivers/gpu/drm/i915/intel_runtime_pm.c
> > +++ b/drivers/gpu/drm/i915/intel_runtime_pm.c
> > @@ -3239,11 +3239,12 @@ static void icl_mbus_init(struct
> > drm_i915_private *dev_priv)
> >  	I915_WRITE(MBUS_ABOX_CTL, val);
> >  }
> >  
> > +/* Actually it is hsw+ but until skl it was not required to set it
> > */
> >  static void skl_pch_reset_handshake(struct drm_i915_private
> > *dev_priv)
> >  {
> >  	u32 val = I915_READ(HSW_NDE_RSTWRN_OPT);
> >  
> > -	if (HAS_PCH_SPLIT(dev_priv))
> > +	if (HAS_PCH_SPLIT(dev_priv) && !HAS_PCH_NOP(dev_priv))
> 
> now I saw why you wanted the unset bit here as well ;)
> 
> >  		val |= RESET_PCH_HANDSHAKE_ENABLE;
> >  	else
> >  		val &= ~RESET_PCH_HANDSHAKE_ENABLE;
> > @@ -3758,6 +3759,15 @@ void intel_power_domains_init_hw(struct
> > drm_i915_private *dev_priv, bool resume)
> >  		mutex_lock(&power_domains->lock);
> >  		vlv_cmnlane_wa(dev_priv);
> >  		mutex_unlock(&power_domains->lock);
> > +	} else if (IS_IVYBRIDGE(dev_priv)) {
> > +		if (HAS_PCH_NOP(dev_priv)) {
> > +			u32 val = I915_READ(GEN7_MSG_CTL);
> > +
> > +			val &= ~(WAIT_FOR_PCH_FLR_ACK |
> > WAIT_FOR_PCH_RESET_ACK);
> > +			I915_WRITE(GEN7_MSG_CTL, val);
> > +		}
> > +	} else if (INTEL_GEN(dev_priv) >= 7) {
> > +		skl_pch_reset_handshake(dev_priv);
> 
> well... skl_ doesn't make sense anymore...
> maybe s/skl/hsw ?
> 
> but what about the remaining block for Ivybridge?
> I feel it got little unbalanced here...
> 
> maybe we move everything inside to a
> intel_pch_reset_handshake(dev_priv, bool set) ?

I guess move everything to intel_pch_reset_handshake() sounds better
but without the bool set, as the HAS_PCH_NOP can be set to all those
platforms depending if display is on or off.



> 
> >  	}
> >  
> >  	/*
> > -- 
> > 2.19.0
> > 
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx at lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/intel-gfx