On Fri, Oct 06, 2023 at 02:42:10PM +0300, Mika Kahola wrote: > Currently we are not using watchdog timers for PSR/PSR2. > The patch disables these timers so they are not in use.
I can't figure out what you're saying here. What bspec seems to be saying is that the max_sleep thing got nuked from the hw so we no longer need to mask it. > > BSpec: 69895 > > Signed-off-by: Mika Kahola <[email protected]> > --- > drivers/gpu/drm/i915/display/intel_psr.c | 24 +++++++++++++++++------- > 1 file changed, 17 insertions(+), 7 deletions(-) > > diff --git a/drivers/gpu/drm/i915/display/intel_psr.c > b/drivers/gpu/drm/i915/display/intel_psr.c > index 850b11f20285..13b58dceb2bf 100644 > --- a/drivers/gpu/drm/i915/display/intel_psr.c > +++ b/drivers/gpu/drm/i915/display/intel_psr.c > @@ -672,11 +672,15 @@ static void hsw_activate_psr1(struct intel_dp *intel_dp) > struct drm_i915_private *dev_priv = dp_to_i915(intel_dp); > enum transcoder cpu_transcoder = intel_dp->psr.transcoder; > u32 max_sleep_time = 0x1f; > - u32 val = EDP_PSR_ENABLE; > + u32 val = 0; > > - val |= EDP_PSR_IDLE_FRAMES(psr_compute_idle_frames(intel_dp)); > > - val |= EDP_PSR_MAX_SLEEP_TIME(max_sleep_time); > + if (DISPLAY_VER(dev_priv) < 20) { > + val = EDP_PSR_ENABLE; That would mean you never enable PSR1 on lnl+ > + val |= EDP_PSR_MAX_SLEEP_TIME(max_sleep_time); > + } > + > + val |= EDP_PSR_IDLE_FRAMES(psr_compute_idle_frames(intel_dp)); > if (IS_HASWELL(dev_priv)) > val |= EDP_PSR_MIN_LINK_ENTRY_TIME_8_LINES; > > @@ -1398,10 +1402,16 @@ static void intel_psr_enable_source(struct intel_dp > *intel_dp, > * runtime_pm besides preventing other hw tracking issues now we > * can rely on frontbuffer tracking. > */ > - mask = EDP_PSR_DEBUG_MASK_MEMUP | > - EDP_PSR_DEBUG_MASK_HPD | > - EDP_PSR_DEBUG_MASK_LPSP | > - EDP_PSR_DEBUG_MASK_MAX_SLEEP; > + if (DISPLAY_VER(dev_priv) >= 20) > + mask = EDP_PSR_DEBUG_MASK_MEMUP | > + EDP_PSR_DEBUG_MASK_HPD | > + EDP_PSR_DEBUG_MASK_LPSP; > + else > + mask = EDP_PSR_DEBUG_MASK_MEMUP | > + EDP_PSR_DEBUG_MASK_HPD | > + EDP_PSR_DEBUG_MASK_LPSP | > + EDP_PSR_DEBUG_MASK_MAX_SLEEP; The hpd mask bit also seems gone now, though there is no explanation why it disappeared. > + > > /* > * No separate pipe reg write mask on hsw/bdw, so have to unmask all > -- > 2.34.1 -- Ville Syrjälä Intel
