Re: [PATCH 07/12] drm/i915: Move DP link retraining into intel_dp_detect()

[Manasi Navare <manasi.d.navare@intel.com>]

On Thu, Jul 28, 2016 at 11:15:22PM +0300, Ville Syrjälä wrote:
> On Thu, Jul 28, 2016 at 12:48:53PM -0700, Manasi Navare wrote:
> > On Thu, Jul 28, 2016 at 05:50:43PM +0300, ville.syrjala@linux.intel.com wrote:
> > > From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > 
> > > DP link retraining needs to grab some modeset locks to not race with
> > > modesets, so we can't really do it safely from the hpd_pulse, lest we
> > > risk deadlocking due to MST sideband stuff.
> > > 
> > > Move the link retraining to happen from the hotplug work instead.
> > > Doing at the end of intel_dp_detect() seems like a good place in case
> > > the sink already got disconnected, in which case retraining is
> > > pointless.
> > > 
> > > To determine if we need to schedule the hotplug work, we'll just check
> > > the sink lane status without locks from hpd_pulse. A little racy
> > > still eg. due to useing intel_dp->lane_count, but no less racy than
> > > what we already had. We'll repeat the check in from intel_dp_detect()
> > > with proper locking, where we'll also check if the link as actually
> > > active or not.
> > > 
> > > Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
> > > Cc: Jim Bride <jim.bride@linux.intel.com>
> > > Cc: Manasi D Navare <manasi.d.navare@intel.com>
> > > Cc: Durgadoss R <durgadoss.r@intel.com>
> > > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > ---
> > >  drivers/gpu/drm/i915/intel_dp.c | 154 +++++++++++++++++-----------------------
> > >  1 file changed, 66 insertions(+), 88 deletions(-)
> > > 
> > > diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
> > > index 4a4184c21989..675b83f57a07 100644
> > > --- a/drivers/gpu/drm/i915/intel_dp.c
> > > +++ b/drivers/gpu/drm/i915/intel_dp.c
> > > @@ -3842,15 +3842,6 @@ intel_dp_check_mst_status(struct intel_dp *intel_dp)
> > >  		bret = intel_dp_get_sink_irq_esi(intel_dp, esi);
> > >  go_again:
> > >  		if (bret == true) {
> > > -
> > > -			/* check link status - esi[10] = 0x200c */
> > > -			if (intel_dp->active_streams &&
> > > -			    !drm_dp_channel_eq_ok(&esi[10], intel_dp->lane_count)) {
> > > -				DRM_DEBUG_KMS("channel EQ not ok, retraining\n");
> > > -				intel_dp_start_link_train(intel_dp);
> > > -				intel_dp_stop_link_train(intel_dp);
> > > -			}
> > > -
> > >  			DRM_DEBUG_KMS("got esi %3ph\n", esi);
> > >  			ret = drm_dp_mst_hpd_irq(&intel_dp->mst_mgr, esi, &handled);
> > >  
> > > @@ -3886,34 +3877,42 @@ go_again:
> > >  	return -EINVAL;
> > >  }
> > >  
> > > -static void
> > > -intel_dp_check_link_status(struct intel_dp *intel_dp)
> > > +static bool
> > > +intel_dp_link_needs_retrain(struct intel_dp *intel_dp)
> > >  {
> > > -	struct intel_encoder *intel_encoder = &dp_to_dig_port(intel_dp)->base;
> > > -	struct drm_device *dev = intel_dp_to_dev(intel_dp);
> > >  	u8 link_status[DP_LINK_STATUS_SIZE];
> > >  
> > > -	WARN_ON(!drm_modeset_is_locked(&dev->mode_config.connection_mutex));
> > > +	if (intel_dp->active_streams == 0)
> > > +		return false;
> > > +
> > > +	if (intel_dp->compliance_test_type == DP_TEST_LINK_TRAINING)
> > > +		return true;
> > >  
> > >  	if (!intel_dp_get_link_status(intel_dp, link_status)) {
> > >  		DRM_ERROR("Failed to get link status\n");
> > > -		return;
> > > +		return false;
> > >  	}
> > >  
> > > -	if (!intel_encoder->base.crtc)
> > > -		return;
> > > +	return !drm_dp_channel_eq_ok(link_status, intel_dp->lane_count);
> > 
> > According to the DP spec, we should also check for the clock recovery bit in DPCD
> > We should also add a check drm_dp_clock_recovery_ok(link_status, intel_dp->lane_count)
> 
> I think drm_dp_channel_eq_ok() checks all the bits: INTERLANE_ALIGN_DONE,
> CR_DONE, EQ_DONE, and SYMBOL_LOCKED.

It only checks for EQ_DONE and INTERLANE_ALIGN_DONE.
drm_dp_clock_recovery_ok() checks for DP_LANE_CR_DONE bit.
> 
> Anyways, while I was looking at the short pulse handling a bit more, I came
> to the conclusion that we should rewrite it a bit.
> 
> The CTS docs seem to say that we're expected to read DPCD 0x200-0x205 on 
> short pulse, and use that information to determine what we should do.
> Currently we don't do that.
> 
> So I was thunking that we'd adjust the link_status stuff to read 0x200-0x207
> instead of 0x202-0x207, like so:
> -#define DP_LINK_STATUS_SIZE       6
> +#define DP_LINK_STATUS_SIZE       8
> 
>  static u8 dp_link_status(const u8 link_status[DP_LINK_STATUS_SIZE], int r)
>  {
> -       return link_status[r - DP_LANE0_1_STATUS];
> +       return link_status[r - DP_SINK_COUNT];
>  }
> 
> And then we'd use something like this as our short pulse handler:
> 
> static bool
> intel_dp_short_pulse(struct intel_dp *intel_dp)
> {
>         u8 link_status[DP_LINK_STATUS_SIZE], tmp;
> 
>         /*                                                                                                             
>          * Clearing compliance test variables to allow capturing                                                       
>          * of values for next automated test request.                                                                  
>          */
>         intel_dp->compliance_test_active = 0;
>         intel_dp->compliance_test_type = 0;
>         intel_dp->compliance_test_data = 0;
> 
>         if (!intel_dp_get_link_status(intel_dp, link_status))
>                 return false;
> 
>         tmp = drm_dp_link_status(link_status, DP_DEVICE_SERVICE_IRQ_VECTOR);
>         if (tmp)
>                 drm_dp_dpcd_writeb(&intel_dp->aux,
>                                    DP_DEVICE_SERVICE_IRQ_VECTOR,
>                                    tmp);
> 
>         if (tmp & DP_AUTOMATED_TEST_REQUEST)
>                 DRM_DEBUG_DRIVER("Test request in short pulse not handled\n");
>         if (tmp & (DP_CP_IRQ | DP_SINK_SPECIFIC_IRQ))
>                 DRM_DEBUG_DRIVER("CP or sink specific irq unhandled\n");
> 
>         tmp = drm_dp_link_status(link_status, DP_LANE_ALIGN_STATUS_UPDATED);
>         if (tmp & DP_DOWNSTREAM_PORT_STATUS_CHANGED)
>                 return false;
> 
>         tmp = drm_dp_link_status(link_status, DP_SINK_COUNT);
>         if (DP_GET_SINK_COUNT(tmp) != intel_dp->sink_count)
>                 return false;
> 
>         if (intel_dp->active_streams != 0 &&
>             !drm_dp_channel_eq_ok(link_status, intel_dp->lane_count))
>                 return false;
> 
>         return true;
> }
> 
> and obviously fill in whatever else sink irq handling we need.
> 
> Thoughts? Does that look sane?

Yes all the Link maintenance tests in CTS spec expect DPCD registers
0200h-0205h to be read on loss of symbol lock or loss of clock recovery.
So I think it makes sense to read 0200-0205h.

> 
> <snip>
> > >  
> > >  static enum drm_connector_status
> > >  intel_dp_detect(struct drm_connector *connector, bool force)
> > >  {
> > >  	struct intel_dp *intel_dp = intel_attached_dp(connector);
> > > -	struct intel_digital_port *intel_dig_port = dp_to_dig_port(intel_dp);
> > > -	struct intel_encoder *intel_encoder = &intel_dig_port->base;
> > >  	struct intel_connector *intel_connector = to_intel_connector(connector);
> > > +	enum drm_connector_status status;
> > >  
> > >  	DRM_DEBUG_KMS("[CONNECTOR:%d:%s]\n",
> > >  		      connector->base.id, connector->name);
> > >  
> > > -	if (intel_dp->is_mst) {
> > > -		/* MST devices are disconnected from a monitor POV */
> > > -		intel_dp_unset_edid(intel_dp);
> > > -		if (intel_encoder->type != INTEL_OUTPUT_EDP)
> > > -			intel_encoder->type = INTEL_OUTPUT_DP;
> > > -		return connector_status_disconnected;
> > > -	}
> > > -
> > > -	/* If full detect is not performed yet, do a full detect */
> > > -	if (!intel_dp->detect_done)
> > > -		intel_dp_long_pulse(intel_dp->attached_connector);
> > > +	status = intel_dp_long_pulse(intel_connector);
> > >  
> > > -	intel_dp->detect_done = false;
> > > +	if (status == connector_status_connected || intel_dp->is_mst)
> > 
> > Whya re we not checking for (!intel_encoder->crtc.base) before retraining?
> > I had seen issues with connected boot case when connector status is connected and if it
> > tries to retrain it fails with retries because there is no active pipe. In that
> > case it should not attempt to retrain.
> 
> I changed it to use intel_dp->active_streams instead. That'll work for MST
> as well as SST. Checking for the crtc would only cover SST.
> 
> > 
> > 
> > > +		intel_dp_link_retrain(intel_dp);
> > >  
> > > -	if (is_edp(intel_dp) || intel_connector->detect_edid)
> > > -		return connector_status_connected;
> > > -	else
> > > -		return connector_status_disconnected;
> > > +	return status;
> > >  }
> > >  
> > >  static void
> > > @@ -4706,39 +4682,41 @@ intel_dp_hpd_pulse(struct intel_digital_port *intel_dig_port, bool long_hpd)
> > >  		      port_name(intel_dig_port->port),
> > >  		      long_hpd ? "long" : "short");
> > >  
> > > +	if (long_hpd)
> > > +		return IRQ_NONE;
> > > +
> > >  	power_domain = intel_display_port_aux_power_domain(intel_encoder);
> > >  	intel_display_power_get(dev_priv, power_domain);
> > >  
> > > -	if (long_hpd) {
> > > -		intel_dp_long_pulse(intel_dp->attached_connector);
> > > -		if (intel_dp->is_mst)
> > > -			ret = IRQ_HANDLED;
> > > -		goto put_power;
> > > -
> > > -	} else {
> > > -		if (intel_dp->is_mst) {
> > > -			if (intel_dp_check_mst_status(intel_dp) == -EINVAL) {
> > > -				/*
> > > -				 * If we were in MST mode, and device is not
> > > -				 * there, get out of MST mode
> > > -				 */
> > > -				DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",
> > > -					      intel_dp->is_mst, intel_dp->mst_mgr.mst_state);
> > > -				intel_dp->is_mst = false;
> > > -				drm_dp_mst_topology_mgr_set_mst(&intel_dp->mst_mgr,
> > > -								intel_dp->is_mst);
> > > -				goto put_power;
> > > -			}
> > > -		}
> > > -
> > > -		if (!intel_dp->is_mst) {
> > > -			if (!intel_dp_short_pulse(intel_dp)) {
> > > -				intel_dp_long_pulse(intel_dp->attached_connector);
> > > -				goto put_power;
> > > -			}
> > > +	if (intel_dp->is_mst) {
> > > +		if (intel_dp_check_mst_status(intel_dp) == -EINVAL) {
> > > +			/*
> > > +			 * If we were in MST mode, and device is not
> > > +			 * there, get out of MST mode
> > > +			 */
> > > +			DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",
> > > +				      intel_dp->is_mst, intel_dp->mst_mgr.mst_state);
> > > +			intel_dp->is_mst = false;
> > > +			drm_dp_mst_topology_mgr_set_mst(&intel_dp->mst_mgr,
> > > +							intel_dp->is_mst);
> > > +			goto put_power;
> > >  		}
> > > +	} else {
> > > +		if (!intel_dp_short_pulse(intel_dp))
> > > +			goto put_power;
> > >  	}
> > >  
> > > +	/*
> > > +	 * Link retraining happens from the hotplug work,
> > > +	 * check if we might need to schdule it.
> > > +	 *
> > > +	 * There has been known issues of link loss
> > > +	 * triggerring long pulse, so let's check both
> > > +	 * for short and long pulse.
> > > +	 */
> > > +	if (intel_dp_link_needs_retrain(intel_dp))
> > > +		goto put_power;
> > > +
> > >  	ret = IRQ_HANDLED;
> > >  
> > >  put_power:
> > > -- 
> > > 2.7.4
> > > 
> 
> -- 
> Ville Syrjälä
> Intel OTC
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx