Re: [Intel-gfx] [CI 15/19] drm/i915/bigjoiner: atomic commit changes for uncompressed joiner

[Jani Nikula <jani.nikula@linux.intel.com>]

On Thu, 03 Jun 2021, "Manna, Animesh" <animesh.manna@intel.com> wrote:
>> -----Original Message-----
>> From: Jani Nikula <jani.nikula@linux.intel.com>
>> Sent: Thursday, June 3, 2021 3:10 PM
>> To: Roper, Matthew D <matthew.d.roper@intel.com>; intel-
>> gfx@lists.freedesktop.org
>> Cc: Manna, Animesh <animesh.manna@intel.com>; Navare, Manasi D
>> <manasi.d.navare@intel.com>; Kulkarni, Vandita <vandita.kulkarni@intel.com>
>> Subject: Re: [Intel-gfx] [CI 15/19] drm/i915/bigjoiner: atomic commit changes
>> for uncompressed joiner
>> 
>> On Fri, 14 May 2021, Matt Roper <matthew.d.roper@intel.com> wrote:
>> > From: Animesh Manna <animesh.manna@intel.com>
>> >
>> > Respective bit for master or slave to be set for uncompressed
>> > bigjoiner in dss_ctl1 register.
>> 
>> I was looking at the changes here due to a static checker complaint, and I think
>> there are a number of issues here. Some more serious than others, and some
>> predate the patch.
>> 
>> Comments inline.
>> 
>> > Cc: Manasi Navare <manasi.d.navare@intel.com>
>> > Signed-off-by: Animesh Manna <animesh.manna@intel.com>
>> > Signed-off-by: Clinton Taylor <Clinton.A.Taylor@intel.com>
>> > Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
>> > Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>
>> > ---
>> >  drivers/gpu/drm/i915/display/intel_display.c |  6 +++
>> >  drivers/gpu/drm/i915/display/intel_vdsc.c    | 40 +++++++++++++++++++-
>> >  drivers/gpu/drm/i915/display/intel_vdsc.h    |  2 +
>> >  drivers/gpu/drm/i915/i915_reg.h              |  2 +
>> >  4 files changed, 49 insertions(+), 1 deletion(-)
>> >
>> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c
>> > b/drivers/gpu/drm/i915/display/intel_display.c
>> > index b5fd721137d3..422b59ebf6dc 100644
>> > --- a/drivers/gpu/drm/i915/display/intel_display.c
>> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
>> > @@ -3411,6 +3411,7 @@ static void icl_ddi_bigjoiner_pre_enable(struct
>> intel_atomic_state *state,
>> >  					 const struct intel_crtc_state
>> *crtc_state)  {
>> >  	struct intel_crtc *master = to_intel_crtc(crtc_state->uapi.crtc);
>> > +	struct drm_i915_private *dev_priv = to_i915(master->base.dev);
>> >  	struct intel_crtc_state *master_crtc_state;
>> >  	struct drm_connector_state *conn_state;
>> >  	struct drm_connector *conn;
>> > @@ -3444,6 +3445,9 @@ static void icl_ddi_bigjoiner_pre_enable(struct
>> intel_atomic_state *state,
>> >  		/* and DSC on slave */
>> >  		intel_dsc_enable(NULL, crtc_state);
>> >  	}
>> > +
>> > +	if (DISPLAY_VER(dev_priv) >= 13)
>> > +		intel_uncompressed_joiner_enable(crtc_state);
>> >  }
>> >
>> >  static void hsw_crtc_enable(struct intel_atomic_state *state, @@
>> > -6250,6 +6254,8 @@ static bool hsw_get_pipe_config(struct intel_crtc *crtc,
>> >  	}
>> >
>> >  	intel_dsc_get_config(pipe_config);
>> > +	if (DISPLAY_VER(dev_priv) >= 13 && !pipe_config-
>> >dsc.compression_enable)
>> > +		intel_uncompressed_joiner_get_config(pipe_config);
>> >
>> >  	if (!active) {
>> >  		/* bigjoiner slave doesn't enable transcoder */ diff --git
>> > a/drivers/gpu/drm/i915/display/intel_vdsc.c
>> > b/drivers/gpu/drm/i915/display/intel_vdsc.c
>> > index adcd6752f919..efc3184d8315 100644
>> > --- a/drivers/gpu/drm/i915/display/intel_vdsc.c
>> > +++ b/drivers/gpu/drm/i915/display/intel_vdsc.c
>> > @@ -1021,6 +1021,22 @@ static i915_reg_t dss_ctl2_reg(const struct
>> intel_crtc_state *crtc_state)
>> >  	return is_pipe_dsc(crtc_state) ? ICL_PIPE_DSS_CTL2(pipe) : DSS_CTL2;
>> > }
>> >
>> > +void intel_uncompressed_joiner_enable(const struct intel_crtc_state
>> > +*crtc_state)
>> 
>> Naming. Basically for any new function, the function name prefix should match
>> the file name. intel_vdsc.[ch] should have functions prefixed intel_vdsc_*(). This
>> is where we're headed to increase clarity.
>> 
>> intel_uncompressed_*() is something completely different.
>> 
>> Granted, here we already have intel_dsc_*() in intel_vdsc.c. We should probably
>> stick with intel_dsc_*(). A possible function or file rename is not out of the
>> question, but that's a separate matter.
>
> As there is not separate register for uncompressed joiner, using bitfield of dsc register only the function name can be changed to intel_dsc_uncompressed_joiner_enable().
>
>> 
>> > +{
>> > +	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
>> > +	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
>> > +	u32 dss_ctl1_val = 0;
>> > +
>> > +	if (crtc_state->bigjoiner && !crtc_state->dsc.compression_enable) {
>> > +		if (crtc_state->bigjoiner_slave)
>> > +			dss_ctl1_val |= UNCOMPRESSED_JOINER_SLAVE;
>> > +		else
>> > +			dss_ctl1_val |= UNCOMPRESSED_JOINER_MASTER;
>> > +
>> > +		intel_de_write(dev_priv, dss_ctl1_reg(crtc_state), dss_ctl1_val);
>> > +	}
>> > +}
>> > +
>> >  void intel_dsc_enable(struct intel_encoder *encoder,
>> >  		      const struct intel_crtc_state *crtc_state)  { @@ -1060,13
>> > +1076,35 @@ 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)
>> > +	if (!(old_crtc_state->dsc.compression_enable &&
>> > +	      old_crtc_state->bigjoiner))
>> 
>> This fails to disable compression if we only have compression but no bigjoiner,
>> which is the more common case!
>> 
>> See also:
>> 
>> https://gitlab.freedesktop.org/drm/intel/-/issues/3537
>> https://patchwork.freedesktop.org/patch/msgid/20210603065356.15435-1-
>> vandita.kulkarni@intel.com
>> 
>
> We may need to remove both the condition check.
> In uncompressed bigjoiner, compression_enable flag will be 0 and may not clear the bit of dss_ctl1_reg.
> So can we remove both the checks, hoping it will not harm reseting the register even if it is not set. 

Now it only disables if *both* were enabled. It needs to be disabled if
*either* was enabled. But if *neither* was enabled, we don't need to do
this.


>
>> >  		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);  }
>> >
>> > +void intel_uncompressed_joiner_get_config(struct intel_crtc_state
>> > +*crtc_state) {
>> > +	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
>> > +	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
>> > +	u32 dss_ctl1;
>> > +
>> > +	dss_ctl1 = intel_de_read(dev_priv, dss_ctl1_reg(crtc_state));
>> > +	if (dss_ctl1 & UNCOMPRESSED_JOINER_MASTER) {
>> > +		crtc_state->bigjoiner = true;
>> > +		if (!WARN_ON(INTEL_NUM_PIPES(dev_priv) == crtc->pipe + 1))
>> > +			crtc_state->bigjoiner_linked_crtc =
>> > +				intel_get_crtc_for_pipe(dev_priv, crtc->pipe +
>> 1);
>> > +	} else if (dss_ctl1 & UNCOMPRESSED_JOINER_SLAVE) {
>> > +		crtc_state->bigjoiner = true;
>> > +		crtc_state->bigjoiner_slave = true;
>> > +		if (!WARN_ON(crtc->pipe == PIPE_A))
>> > +			crtc_state->bigjoiner_linked_crtc =
>> > +				intel_get_crtc_for_pipe(dev_priv, crtc->pipe -
>> 1);
>> > +	}
>> 
>> Nitpick: This duplicates a bunch of logic for figuring out master/slave.
>> 
>> The static checker warning was about crtc->pipe + 1 usage. Since
>> INTEL_NUM_PIPES() looks at the hamming weight of i915->pipe_mask, the
>> checker has a hard time figuring out it does not overflow
>> i915->pipe_to_crtc_mapping[] in intel_get_crtc_for_pipe().
>> 
>> So here in intel_vdsc.c the checks are for overflowing/underflowing the pipe
>> range. In intel_get_crtc_for_pipe() there's a check for the pipe actually existing -
>> the pipe numbering might not be contiguous.
>> 
>> Superficially the static checker warning is bogus, as in we won't overflow
>> anything. However, deep down there are issues in the consistency of the checks
>> and how to handle non-contigouous pipe numbering.
>> 
>> Indeed, this does not *need* the number. We should figure out the next *crtc*,
>> not the next pipe *number*, which may or may not be pipe + 1.
>
> dss_ctl1 value is used to identify master or slave crtc. If needed WARN_ON check can be removed... but pipe enum values like PIPE_A/B/C/D is used to get the master/slave crtc and always bigjoiner is possible with two adjacent pipes.

I sent the patch for this and you reviewed it... but are adjacent pipes
a requirement for bigjoiner? Would e.g. A+C work if B was fused off? If
not, my patch 2/2 is incorrect.

BR,
Jani.


>
> Regards,
> Animesh
>
>> 
>> 
>> BR,
>> Jani.
>> 
>> > +}
>> > +
>> >  void intel_dsc_get_config(struct intel_crtc_state *crtc_state)  {
>> >  	struct drm_dsc_config *vdsc_cfg = &crtc_state->dsc.config; diff
>> > --git a/drivers/gpu/drm/i915/display/intel_vdsc.h
>> > b/drivers/gpu/drm/i915/display/intel_vdsc.h
>> > index 65d301c23580..fe4d45561253 100644
>> > --- a/drivers/gpu/drm/i915/display/intel_vdsc.h
>> > +++ b/drivers/gpu/drm/i915/display/intel_vdsc.h
>> > @@ -12,11 +12,13 @@ struct intel_encoder;  struct intel_crtc_state;
>> >
>> >  bool intel_dsc_source_support(const struct intel_crtc_state
>> > *crtc_state);
>> > +void intel_uncompressed_joiner_enable(const struct intel_crtc_state
>> > +*crtc_state);
>> >  void intel_dsc_enable(struct intel_encoder *encoder,
>> >  		      const struct intel_crtc_state *crtc_state);  void
>> > intel_dsc_disable(const struct intel_crtc_state *crtc_state);  int
>> > intel_dsc_compute_params(struct intel_encoder *encoder,
>> >  			     struct intel_crtc_state *pipe_config);
>> > +void intel_uncompressed_joiner_get_config(struct intel_crtc_state
>> > +*crtc_state);
>> >  void intel_dsc_get_config(struct intel_crtc_state *crtc_state);  enum
>> > intel_display_power_domain  intel_dsc_power_domain(const struct
>> > intel_crtc_state *crtc_state); diff --git
>> > a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
>> > index 31bc413dbba1..dd6e0bae9573 100644
>> > --- a/drivers/gpu/drm/i915/i915_reg.h
>> > +++ b/drivers/gpu/drm/i915/i915_reg.h
>> > @@ -11493,6 +11493,8 @@ enum skl_power_gate {
>> >  #define  SPLITTER_CONFIGURATION_MASK		REG_GENMASK(26, 25)
>> >  #define  SPLITTER_CONFIGURATION_2_SEGMENT
>> 	REG_FIELD_PREP(SPLITTER_CONFIGURATION_MASK, 0)
>> >  #define  SPLITTER_CONFIGURATION_4_SEGMENT
>> 	REG_FIELD_PREP(SPLITTER_CONFIGURATION_MASK, 1)
>> > +#define  UNCOMPRESSED_JOINER_MASTER		(1 << 21)
>> > +#define  UNCOMPRESSED_JOINER_SLAVE		(1 << 20)
>> >
>> >  #define _ICL_PIPE_DSS_CTL2_PB			0x78204
>> >  #define _ICL_PIPE_DSS_CTL2_PC			0x78404
>> 
>> --
>> Jani Nikula, Intel Open Source Graphics Center

-- 
Jani Nikula, Intel Open Source Graphics Center
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx