Re: [PATCH v3 5/5] drm/i915/dp: Disable the AUX DPCD probe quirk if it's not required

[Imre Deak <imre.deak@intel.com>]

On Fri, Jun 06, 2025 at 04:44:37PM +0300, Jani Nikula wrote:
> On Thu, 05 Jun 2025, Imre Deak <imre.deak@intel.com> wrote:
> > Reading DPCD registers has side-effects and some of these can cause a
> > problem for instance during link training. Based on this it's better to
> > avoid the probing quirk done before each DPCD register read, limiting
> > this to the monitor which requires it. The only known problematic
> > monitor is an external SST sink, so keep the quirk disabled always for
> > eDP and MST sinks. Reenable the quirk after a hotplug event and after
> > resuming from a power state without hotplug support, until the
> > subsequent EDID based detection.
> >
> > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > Cc: Jani Nikula <jani.nikula@linux.intel.com>
> > Signed-off-by: Imre Deak <imre.deak@intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_dp.c      | 11 +++++++++--
> >  drivers/gpu/drm/i915/display/intel_dp_aux.c  |  2 ++
> >  drivers/gpu/drm/i915/display/intel_hotplug.c | 10 ++++++++++
> >  3 files changed, 21 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c
> > index 208a953b04a2f..d65a18fc1aeb9 100644
> > --- a/drivers/gpu/drm/i915/display/intel_dp.c
> > +++ b/drivers/gpu/drm/i915/display/intel_dp.c
> > @@ -5747,6 +5747,11 @@ intel_dp_set_edid(struct intel_dp *intel_dp)
> >  	/* Below we depend on display info having been updated */
> >  	drm_edid_connector_update(&connector->base, drm_edid);
> >  
> > +	if (!intel_dp_is_edp(intel_dp))
> 
> Why does this depend on !edp?
> 
> Feels like unnecessary optimization based on your knowledge of that one
> specific display.

The detection itself requires probing before each DPCD access. I want to
avoid it whenever possible and since the quirk is relevant only the
particular HP external display, we can avoid the probing on eDP
completely.

> > +		drm_dp_dpcd_set_probe_quirk(&intel_dp->aux,
> > +					    drm_edid_has_quirk(&connector->base,
> > +							       DRM_EDID_QUIRK_DP_DPCD_PROBE));
> > +
> >  	vrr_capable = intel_vrr_is_capable(connector);
> >  	drm_dbg_kms(display->drm, "[CONNECTOR:%d:%s] VRR capable: %s\n",
> >  		    connector->base.base.id, connector->base.name, str_yes_no(vrr_capable));
> > @@ -5881,6 +5886,7 @@ intel_dp_detect(struct drm_connector *_connector,
> >  	intel_dp_print_rates(intel_dp);
> >  
> >  	if (intel_dp->is_mst) {
> > +		drm_dp_dpcd_set_probe_quirk(&intel_dp->aux, false);
> 
> Isn't this excessive? Haven't we already set the quirk state?

No, this is the MST root connector's detection and we don't read the EDID
for it (we read it for MST non-root connectors, but those are not
relavant in any case). So this should be set here explicitly, with the
same justification as above for eDP (on MST the probing is never needed,
so we can avoid it on such links completely).

> 
> >  		/*
> >  		 * If we are in MST mode then this connector
> >  		 * won't appear connected or have anything
> > @@ -6321,10 +6327,11 @@ intel_dp_hpd_pulse(struct intel_digital_port *dig_port, bool long_hpd)
> >  	 * complete the DP tunnel BW request for the latter connector/encoder
> >  	 * waiting for this encoder's DPRX read, perform a dummy read here.
> >  	 */
> > -	if (long_hpd)
> > +	if (long_hpd) {
> > +		drm_dp_dpcd_set_probe_quirk(&intel_dp->aux, true);
> > +
> >  		intel_dp_read_dprx_caps(intel_dp, dpcd);
> >  
> > -	if (long_hpd) {
> >  		intel_dp->reset_link_params = true;
> >  		intel_dp_invalidate_source_oui(intel_dp);
> >  
> > diff --git a/drivers/gpu/drm/i915/display/intel_dp_aux.c b/drivers/gpu/drm/i915/display/intel_dp_aux.c
> > index bf8e8e0cc19c9..410252ba6fd16 100644
> > --- a/drivers/gpu/drm/i915/display/intel_dp_aux.c
> > +++ b/drivers/gpu/drm/i915/display/intel_dp_aux.c
> > @@ -835,6 +835,8 @@ void intel_dp_aux_init(struct intel_dp *intel_dp)
> >  
> >  	intel_dp->aux.transfer = intel_dp_aux_transfer;
> >  	cpu_latency_qos_add_request(&intel_dp->pm_qos, PM_QOS_DEFAULT_VALUE);
> > +
> > +	drm_dp_dpcd_set_probe_quirk(&intel_dp->aux, !intel_dp_is_edp(intel_dp));
> >  }
> >  
> >  static enum aux_ch default_aux_ch(struct intel_encoder *encoder)
> > diff --git a/drivers/gpu/drm/i915/display/intel_hotplug.c b/drivers/gpu/drm/i915/display/intel_hotplug.c
> > index 74fe398663d63..1093c6c3714c0 100644
> > --- a/drivers/gpu/drm/i915/display/intel_hotplug.c
> > +++ b/drivers/gpu/drm/i915/display/intel_hotplug.c
> > @@ -33,6 +33,7 @@
> >  #include "intel_display_core.h"
> >  #include "intel_display_rpm.h"
> >  #include "intel_display_types.h"
> > +#include "intel_dp.h"
> >  #include "intel_hdcp.h"
> >  #include "intel_hotplug.h"
> >  #include "intel_hotplug_irq.h"
> > @@ -906,9 +907,18 @@ void intel_hpd_poll_enable(struct intel_display *display)
> >   */
> >  void intel_hpd_poll_disable(struct intel_display *display)
> >  {
> > +	struct intel_encoder *encoder;
> > +
> >  	if (!HAS_DISPLAY(display))
> >  		return;
> >  
> > +	for_each_intel_dp(display->drm, encoder) {
> > +		struct intel_dp *intel_dp = enc_to_intel_dp(encoder);
> > +
> > +		if (!intel_dp_is_edp(intel_dp))
> > +			drm_dp_dpcd_set_probe_quirk(&intel_dp->aux, true);
> > +	}
> > +
> >  	WRITE_ONCE(display->hotplug.poll_enabled, false);
> >  
> >  	spin_lock_irq(&display->irq.lock);
> 
> -- 
> Jani Nikula, Intel