Re: [PATCH v2] drm/i915: State readout and cross-checking for dp_m2_n2

[Vandana Kannan <vandana.kannan@intel.com>]

On May-15-2014 2:48 PM, Vandana Kannan wrote:
> On May-13-2014 3:10 PM, Vandana Kannan wrote:
>> On May-13-2014 2:28 PM, Daniel Vetter wrote:
>>> On Tue, May 13, 2014 at 01:56:04PM +0530, Vandana Kannan wrote:
>>>> On May-12-2014 3:57 PM, Ville Syrjälä wrote:
>>>>> On Mon, May 05, 2014 at 01:49:31PM +0530, Vandana Kannan wrote:
>>>>>> Adding relevant read out comparison code, in check_crtc_state, for the new
>>>>>> member of crtc_config, dp_m2_n2, which was introduced to store link_m_n
>>>>>> values for a DP downclock mode (if available). Suggested by Daniel.
>>>>>>
>>>>>> v2: Changed patch title.
>>>>>> Daniel's review comments incorporated.
>>>>>> Added relevant state readout code for M2_N2. dp_m2_n2 comparison to be done
>>>>>> only when high RR is not in use (This is because alternate m_n register
>>>>>> programming will be done only when low RR is being used).
>>>>>>
>>>>>> Signed-off-by: Vandana Kannan <vandana.kannan@intel.com>
>>>>>> Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
>>>>>> ---
>>>>>>  drivers/gpu/drm/i915/intel_ddi.c     |  1 +
>>>>>>  drivers/gpu/drm/i915/intel_display.c | 72 +++++++++++++++++++++++++++++++++---
>>>>>>  drivers/gpu/drm/i915/intel_dp.c      |  2 +
>>>>>>  drivers/gpu/drm/i915/intel_drv.h     |  2 +
>>>>>>  4 files changed, 72 insertions(+), 5 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/gpu/drm/i915/intel_ddi.c b/drivers/gpu/drm/i915/intel_ddi.c
>>>>>> index 0ad4e96..6784f0b 100644
>>>>>> --- a/drivers/gpu/drm/i915/intel_ddi.c
>>>>>> +++ b/drivers/gpu/drm/i915/intel_ddi.c
>>>>>> @@ -1587,6 +1587,7 @@ void intel_ddi_get_config(struct intel_encoder *encoder,
>>>>>>  	case TRANS_DDI_MODE_SELECT_DP_MST:
>>>>>>  		pipe_config->has_dp_encoder = true;
>>>>>>  		intel_dp_get_m_n(intel_crtc, pipe_config);
>>>>>> +		intel_dp_get_m2_n2(intel_crtc, pipe_config);
>>>>>>  		break;
>>>>>>  	default:
>>>>>>  		break;
>>>>>> diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
>>>>>> index 797f01c..2e625eb 100644
>>>>>> --- a/drivers/gpu/drm/i915/intel_display.c
>>>>>> +++ b/drivers/gpu/drm/i915/intel_display.c
>>>>>> @@ -6670,6 +6670,34 @@ void intel_dp_get_m_n(struct intel_crtc *crtc,
>>>>>>  					     &pipe_config->dp_m_n);
>>>>>>  }
>>>>>>  
>>>>>> +void intel_dp_get_m2_n2(struct intel_crtc *crtc,
>>>>>> +		      struct intel_crtc_config *pipe_config)
>>>>>> +{
>>>>>> +	struct drm_device *dev = crtc->base.dev;
>>>>>> +	struct drm_i915_private *dev_priv = dev->dev_private;
>>>>>> +	enum transcoder transcoder = pipe_config->cpu_transcoder;
>>>>>> +
>>>>>> +	if (INTEL_INFO(dev)->gen >= 8) {
>>>>>> +		intel_cpu_transcoder_get_m_n(crtc, transcoder,
>>>>>> +						&pipe_config->dp_m_n);
>>>>>
>>>>> dm_m2_n2 surely? And why do we even want to do this?
>>>>>
>>>> My miss, will change this.
>>>> For Gen8, there is only one set of MN registers to be programmed for
>>>> high and low RR. Hence this check.
>>>>>> +	} else if (INTEL_INFO(dev)->gen > 6) {
>>>>>> +		pipe_config->dp_m2_n2.link_m =
>>>>>> +					I915_READ(PIPE_LINK_M2(transcoder));
>>>>>> +		pipe_config->dp_m2_n2.link_n =
>>>>>> +					I915_READ(PIPE_LINK_N2(transcoder));
>>>>>> +		pipe_config->dp_m2_n2.gmch_m =
>>>>>> +					I915_READ(PIPE_DATA_M2(transcoder))
>>>>>> +					& ~TU_SIZE_MASK;
>>>>>> +		pipe_config->dp_m2_n2.gmch_n =
>>>>>> +					I915_READ(PIPE_DATA_N2(transcoder));
>>>>>> +		pipe_config->dp_m2_n2.tu =
>>>>>> +					((I915_READ(PIPE_DATA_M2(transcoder))
>>>>>> +					& TU_SIZE_MASK) >> TU_SIZE_SHIFT) + 1;
>>>>>> +	}
>>>>>> +
>>>>>> +}
>>>>>>
>>>>>> +
>>>>>>  static void ironlake_get_fdi_m_n_config(struct intel_crtc *crtc,
>>>>>>  					struct intel_crtc_config *pipe_config)
>>>>>>  {
>>>>>> @@ -9169,6 +9197,15 @@ static void intel_dump_pipe_config(struct intel_crtc *crtc,
>>>>>>  		      pipe_config->dp_m_n.gmch_m, pipe_config->dp_m_n.gmch_n,
>>>>>>  		      pipe_config->dp_m_n.link_m, pipe_config->dp_m_n.link_n,
>>>>>>  		      pipe_config->dp_m_n.tu);
>>>>>> +
>>>>>> +	DRM_DEBUG_KMS("dp: %i, gmch_m2: %u, gmch_n2: %u, link_m2: %u, link_n2: %u, tu2: %u\n",
>>>>>> +		      pipe_config->has_dp_encoder,
>>>>>> +		      pipe_config->dp_m2_n2.gmch_m,
>>>>>> +		      pipe_config->dp_m2_n2.gmch_n,
>>>>>> +		      pipe_config->dp_m2_n2.link_m,
>>>>>> +		      pipe_config->dp_m2_n2.link_n,
>>>>>> +		      pipe_config->dp_m2_n2.tu);
>>>>>> +
>>>>>>  	DRM_DEBUG_KMS("requested mode:\n");
>>>>>>  	drm_mode_debug_printmodeline(&pipe_config->requested_mode);
>>>>>>  	DRM_DEBUG_KMS("adjusted mode:\n");
>>>>>> @@ -9533,6 +9570,14 @@ intel_pipe_config_compare(struct drm_device *dev,
>>>>>>  			  struct intel_crtc_config *current_config,
>>>>>>  			  struct intel_crtc_config *pipe_config)
>>>>>>  {
>>>>>> +	struct drm_i915_private *dev_priv = dev->dev_private;
>>>>>> +	struct intel_connector *intel_connector = dev_priv->drrs.connector;
>>>>>> +	struct intel_encoder *encoder = (intel_connector != NULL) ?
>>>>>> +			intel_attached_encoder(&intel_connector->base) :
>>>>>> +			NULL;
>>>>>> +	struct intel_dp *intel_dp = (encoder != NULL) ?
>>>>>> +			enc_to_intel_dp(&encoder->base) : NULL;
>>>>>> +
>>>>>>  #define PIPE_CONF_CHECK_X(name)	\
>>>>>>  	if (current_config->name != pipe_config->name) { \
>>>>>>  		DRM_ERROR("mismatch in " #name " " \
>>>>>> @@ -9583,11 +9628,28 @@ intel_pipe_config_compare(struct drm_device *dev,
>>>>>>  	PIPE_CONF_CHECK_I(fdi_m_n.tu);
>>>>>>  
>>>>>>  	PIPE_CONF_CHECK_I(has_dp_encoder);
>>>>>> -	PIPE_CONF_CHECK_I(dp_m_n.gmch_m);
>>>>>> -	PIPE_CONF_CHECK_I(dp_m_n.gmch_n);
>>>>>> -	PIPE_CONF_CHECK_I(dp_m_n.link_m);
>>>>>> -	PIPE_CONF_CHECK_I(dp_m_n.link_n);
>>>>>> -	PIPE_CONF_CHECK_I(dp_m_n.tu);
>>>>>> +
>>>>>> +
>>>>>> +	/* DP M1_N1 registers are used when DRRS is disabled or when high RR
>>>>>> +	 * is used. Else, DP M2_N2 registers are used. The following check
>>>>>> +	 * has been added to make sure a mismatch (if any) is displayed only
>>>>>> +	 * for a real difference and not because of DRRS state.
>>>>>> +	 */
>>>>>> +	if ((dev_priv->vbt.drrs_type == DRRS_NOT_SUPPORTED) ||
>>>>>> +		(dev_priv->vbt.drrs_type != DRRS_NOT_SUPPORTED && intel_dp &&
>>>>>> +		intel_dp->drrs_state.refresh_rate_type == DRRS_HIGH_RR)) {
>>>>>> +			PIPE_CONF_CHECK_I(dp_m_n.gmch_m);
>>>>>> +			PIPE_CONF_CHECK_I(dp_m_n.gmch_n);
>>>>>> +			PIPE_CONF_CHECK_I(dp_m_n.link_m);
>>>>>> +			PIPE_CONF_CHECK_I(dp_m_n.link_n);
>>>>>> +			PIPE_CONF_CHECK_I(dp_m_n.tu);
>>>>>> +	} else {
>>>>>> +		PIPE_CONF_CHECK_I(dp_m2_n2.gmch_m);
>>>>>> +		PIPE_CONF_CHECK_I(dp_m2_n2.gmch_n);
>>>>>> +		PIPE_CONF_CHECK_I(dp_m2_n2.link_m);
>>>>>> +		PIPE_CONF_CHECK_I(dp_m2_n2.link_n);
>>>>>> +		PIPE_CONF_CHECK_I(dp_m2_n2.tu);
>>>>>> +	}
>>>>>
>>>>> Can't we just check both dp_m_n and dp_m2_n2 always?
>>>>>
>>>> Daniel/Ville,
>>>>
>>>> pipe_config->dp_m2_n2 is populated along with pipe_config->dp_m_n in
>>>> compute_config(). while pipe_config dp_m_n values are programmed into MN
>>>> registers during the mode set, dp_m2_n2 values are programmed only when
>>>> low RR is used.
>>>> Given this, when PIPE_CONF_CHECK is done for dp_m2_n2, it would show a
>>>> mismatch (struct contains values to be programmed but registers still
>>>> have 0) and return.
>>>> The reason these checks were added was to avoid the scenario in which a
>>>> mismatch is highlighted when there is no real failure at all. For
>>>> example, DRRS is enabled on boot but the scenario to trigger low RR has
>>>> not yet occurred, dp_m2_n2 has been computed but M2_N2 registers are not
>>>> programmed in this case. There will be a mismatch in the pipe_config check.
>>>>
>>>> Changes can be made to set M2_N2 registers when dp_m2_n2 is populated
>>>> and toggle the bit (PIPECONF bit) to switch between high and low RR, but
>>>> this becomes invalid for gen8 and above where only 1 set of MN registers
>>>> are available.
>>>>
>>>> Please suggest some other way to handle this.
>>>> May be a quirk like the following?
>>>>
>>>>         /*
>>>>          * FIXME: BIOS likes to set up a cloned config with lvds+external
>>>>          * screen. Since we don't yet re-compute the pipe config when moving
>>>>          * just the lvds port away to another pipe the sw tracking won't
>>>> match.
>>>>          *
>>>>          * Proper atomic modesets with recomputed global state will fix
>>>> this.
>>>>          * Until then just don't check gmch state for inherited modes.
>>>>          */
>>>>         if (!PIPE_CONF_QUIRK(PIPE_CONFIG_QUIRK_INHERITED_MODE)) {
>>>>
>>>> Or go ahead with checking dp_m_n and dp_m2_n2 always ?
>>>> Please let me know..
>>>
>>> I think for BDW we should check whether the hw values match _either_
>>> dp_m_n or dp_m2_n2. But for hsw and earlier we should check both.
>>>
>>> That leaves us with the slightly awkward issue of recovering DRRS state on
>>> bdw. But I think since we sprinkle idle/busy calls all over the place this
>>> is a problem we can just ignore ;-)
>>> -Daniel
>>>
>>
>> Ok, So I will make the following changes and resend..
>> 1. in compute_config(), for gen < 8, set M2_N2 registers (instead of
>> doing it in intel_dp_set_drrs_state()).
>> 2. in intel_pipe_config_compare(), include your inputs given above..
>> -Vandana
> 
> hi Daniel,
> I see a few issues with this approach of setting M2 N2 registers ahead
> and just toggling the bit to switch between refresh rates (for gen 7 and
> below)..
> 
> writing into M2_N2 registers in compute_config() would fail when system
> resumes from sleep. to overcome this, setting M2 N2 can be done in
> crtc_mode_set functions (similar to set_m_n).
> 
> given this setting, suppose user space sets a refresh rate based on
> usage scenario (future implementation maybe), which would come through
> mode_set calls, M2 N2 registers can be appropriately programmed with
> user space requested RR. now, (if kernel idleness detection for DRRS is
> enabled) if the screen image is idle, then additional changes would be
> required to program M2 N2 registers with the correct low RR values. For
> example, set_m2_n2 would have to be called just before set_drrs for
> lowest RR is called.
> 
> in summary, the alternative approach to avoid DRRS related checks in
> pipe_config_compare() is doable but creates other inconveniences w.r.t
> the overall design and interaction with user space.
> 
> please let me know your opinion on this..
> 
> -Vandana
> 
Hi Daniel,
Please let me know your inputs on this..
Given that making changes to avoid DRRS related checks in
pipe_config_compare affects the overall design and future
implementations related to RR switching, I think that using a quirk for
downclock_mode to compare dp_m_n and dp_m2_n2 based on the RR in use,

if (!low RR)
	compare dp_m_n
else
	compare dp_m2_n2

would avoid unrelated DRRS checks and make sure extension of DRRS
implementation would be possible.
Only thing would be that both dp_m_n and dp_m2_n2 cant be compared always..
-Vandana
>>>>
>>>> -Vandana
>>>>
>>>>>>  
>>>>>>  	PIPE_CONF_CHECK_I(adjusted_mode.crtc_hdisplay);
>>>>>>  	PIPE_CONF_CHECK_I(adjusted_mode.crtc_htotal);
>>>>>> diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
>>>>>> index 34ed143..9aa4dcd 100644
>>>>>> --- a/drivers/gpu/drm/i915/intel_dp.c
>>>>>> +++ b/drivers/gpu/drm/i915/intel_dp.c
>>>>>> @@ -1511,6 +1511,8 @@ static void intel_dp_get_config(struct intel_encoder *encoder,
>>>>>>  
>>>>>>  	intel_dp_get_m_n(crtc, pipe_config);
>>>>>>  
>>>>>> +	intel_dp_get_m2_n2(crtc, pipe_config);
>>>>>> +
>>>>>>  	if (port == PORT_A) {
>>>>>>  		if ((I915_READ(DP_A) & DP_PLL_FREQ_MASK) == DP_PLL_FREQ_160MHZ)
>>>>>>  			pipe_config->port_clock = 162000;
>>>>>> diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
>>>>>> index d8b540b..1013f70 100644
>>>>>> --- a/drivers/gpu/drm/i915/intel_drv.h
>>>>>> +++ b/drivers/gpu/drm/i915/intel_drv.h
>>>>>> @@ -763,6 +763,8 @@ void hsw_enable_pc8(struct drm_i915_private *dev_priv);
>>>>>>  void hsw_disable_pc8(struct drm_i915_private *dev_priv);
>>>>>>  void intel_dp_get_m_n(struct intel_crtc *crtc,
>>>>>>  		      struct intel_crtc_config *pipe_config);
>>>>>> +void intel_dp_get_m2_n2(struct intel_crtc *crtc,
>>>>>> +		      struct intel_crtc_config *pipe_config);
>>>>>>  int intel_dotclock_calculate(int link_freq, const struct intel_link_m_n *m_n);
>>>>>>  void
>>>>>>  ironlake_check_encoder_dotclock(const struct intel_crtc_config *pipe_config,
>>>>>> -- 
>>>>>> 1.9.2
>>>>>>
>>>>>> _______________________________________________
>>>>>> Intel-gfx mailing list
>>>>>> Intel-gfx@lists.freedesktop.org
>>>>>> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>>>>>
>>>>
>>>
>>
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>