| Message ID | 20230215083832.287519-3-vinod.govindapillai@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Provision to ignore long HPDs in CI systems | expand |
On Wed, Feb 15, 2023 at 10:38:32AM +0200, Vinod Govindapillai wrote: > If the ignore long HPD flag is set, ignore the link training > failures as well. Because of spurious HPDs, some unexpected link > training failures are happening while executing IGT test cases. > Ignore the link training failures for the time being if the long > HPDs are also ignored in the environments like CI. > > Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com> Looks ok to me, on the patchset: Reviewed-by: Imre Deak <imre.deak@intel.com> > --- > .../drm/i915/display/intel_dp_link_training.c | 24 +++++++++++++++++++ > 1 file changed, 24 insertions(+) > > diff --git a/drivers/gpu/drm/i915/display/intel_dp_link_training.c b/drivers/gpu/drm/i915/display/intel_dp_link_training.c > index 3d3efcf02011..f90c627ab553 100644 > --- a/drivers/gpu/drm/i915/display/intel_dp_link_training.c > +++ b/drivers/gpu/drm/i915/display/intel_dp_link_training.c > @@ -1433,7 +1433,11 @@ intel_dp_128b132b_link_train(struct intel_dp *intel_dp, > void intel_dp_start_link_train(struct intel_dp *intel_dp, > const struct intel_crtc_state *crtc_state) > { > + struct drm_i915_private *i915 = dp_to_i915(intel_dp); > + struct intel_connector *connector = intel_dp->attached_connector; > + struct intel_encoder *encoder = &dp_to_dig_port(intel_dp)->base; > bool passed; > + > /* > * TODO: Reiniting LTTPRs here won't be needed once proper connector > * HW state readout is added. > @@ -1451,6 +1455,26 @@ void intel_dp_start_link_train(struct intel_dp *intel_dp, > else > passed = intel_dp_link_train_all_phys(intel_dp, crtc_state, lttpr_count); > > + /* > + * Ignore the link failure in CI > + * > + * In fixed enviroments like CI, sometimes unexpected long HPDs are > + * generated by the displays. If ignore_long_hpd flag is set, such long > + * HPDs are ignored. And probably as a consequence of these ignored > + * long HPDs, subsequent link trainings are failed resulting into CI > + * execution failures. > + * > + * For test cases which rely on the link training or processing of HPDs > + * ignore_long_hpd flag can unset from the testcase. > + */ > + if (!passed && i915->display.hotplug.ignore_long_hpd) { > + drm_dbg_kms(&i915->drm, > + "[CONNECTOR:%d:%s][ENCODER:%d:%s] Ignore the link failure\n", > + connector->base.base.id, connector->base.name, > + encoder->base.base.id, encoder->base.name); > + return; > + } > + > if (!passed) > intel_dp_schedule_fallback_link_training(intel_dp, crtc_state); > } > -- > 2.34.1 >
diff --git a/drivers/gpu/drm/i915/display/intel_dp_link_training.c b/drivers/gpu/drm/i915/display/intel_dp_link_training.c index 3d3efcf02011..f90c627ab553 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_link_training.c +++ b/drivers/gpu/drm/i915/display/intel_dp_link_training.c @@ -1433,7 +1433,11 @@ intel_dp_128b132b_link_train(struct intel_dp *intel_dp, void intel_dp_start_link_train(struct intel_dp *intel_dp, const struct intel_crtc_state *crtc_state) { + struct drm_i915_private *i915 = dp_to_i915(intel_dp); + struct intel_connector *connector = intel_dp->attached_connector; + struct intel_encoder *encoder = &dp_to_dig_port(intel_dp)->base; bool passed; + /* * TODO: Reiniting LTTPRs here won't be needed once proper connector * HW state readout is added. @@ -1451,6 +1455,26 @@ void intel_dp_start_link_train(struct intel_dp *intel_dp, else passed = intel_dp_link_train_all_phys(intel_dp, crtc_state, lttpr_count); + /* + * Ignore the link failure in CI + * + * In fixed enviroments like CI, sometimes unexpected long HPDs are + * generated by the displays. If ignore_long_hpd flag is set, such long + * HPDs are ignored. And probably as a consequence of these ignored + * long HPDs, subsequent link trainings are failed resulting into CI + * execution failures. + * + * For test cases which rely on the link training or processing of HPDs + * ignore_long_hpd flag can unset from the testcase. + */ + if (!passed && i915->display.hotplug.ignore_long_hpd) { + drm_dbg_kms(&i915->drm, + "[CONNECTOR:%d:%s][ENCODER:%d:%s] Ignore the link failure\n", + connector->base.base.id, connector->base.name, + encoder->base.base.id, encoder->base.name); + return; + } + if (!passed) intel_dp_schedule_fallback_link_training(intel_dp, crtc_state); }
If the ignore long HPD flag is set, ignore the link training failures as well. Because of spurious HPDs, some unexpected link training failures are happening while executing IGT test cases. Ignore the link training failures for the time being if the long HPDs are also ignored in the environments like CI. Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com> --- .../drm/i915/display/intel_dp_link_training.c | 24 +++++++++++++++++++ 1 file changed, 24 insertions(+)