[PATCH 2/2] drm/i915/psr: Implment WA to help reach PC10
Kandpal, Suraj
suraj.kandpal at intel.com
Fri Aug 23 06:18:09 UTC 2024
> -----Original Message-----
> From: Hogander, Jouni <jouni.hogander at intel.com>
> Sent: Friday, August 23, 2024 10:54 AM
> To: Kandpal, Suraj <suraj.kandpal at intel.com>; intel-gfx at lists.freedesktop.org
> Cc: Murthy, Arun R <arun.r.murthy at intel.com>; Manna, Animesh
> <animesh.manna at intel.com>; jani.nikula at linux.intel.com
> Subject: Re: [PATCH 2/2] drm/i915/psr: Implment WA to help reach PC10
>
> On Fri, 2024-08-23 at 04:54 +0000, Kandpal, Suraj wrote:
> >
> >
> > > -----Original Message-----
> > > From: Hogander, Jouni <jouni.hogander at intel.com>
> > > Sent: Thursday, August 22, 2024 2:16 PM
> > > To: Kandpal, Suraj <suraj.kandpal at intel.com>; intel-
> > > gfx at lists.freedesktop.org
> > > Cc: Murthy, Arun R <arun.r.murthy at intel.com>; Manna, Animesh
> > > <animesh.manna at intel.com>; jani.nikula at linux.intel.com
> > > Subject: Re: [PATCH 2/2] drm/i915/psr: Implment WA to help reach
> > > PC10
> > >
> > > On Wed, 2024-06-19 at 10:07 +0530, Suraj Kandpal wrote:
> > > > To reach PC10 when PKG_C_LATENCY is configure we must do the
> > > following
> > > > things
> > > > 1) Enter PSR1 only when delayed_vblank < 6 lines and DC5 can be
> > > > entered
> > > > 2) Allow PSR2 deep sleep when DC5 can be entered
> > > > 3) DC5 can be entered when all transocoder have either PSR1, PSR2
> > > > or eDP 1.5 PR ALPM enabled and VBI is disabled and flips and
> > > > pushes are not happening.
> > > >
> > > > --v2
> > > > -Switch condition and do an early return [Jani] -Do some checks in
> > > > compute_config [Jani] -Do not use register reads as a method of
> > > > checking states for DPKGC or delayed vblank [Jani] -Use another
> > > > way to see is vblank interrupts are disabled or not [Jani]
> > > >
> > > > WA: 16023497226
> > > > Signed-off-by: Suraj Kandpal <suraj.kandpal at intel.com>
> > > > ---
> > > > .../drm/i915/display/intel_display_types.h | 2 +
> > > > drivers/gpu/drm/i915/display/intel_psr.c | 82
> > > > ++++++++++++++++++-
> > > > 2 files changed, 83 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h
> > > > b/drivers/gpu/drm/i915/display/intel_display_types.h
> > > > index 46b3cbeb4a82..031f8e889b65 100644
> > > > --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> > > > +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> > > > @@ -1708,6 +1708,8 @@ struct intel_psr {
> > > > bool sink_support;
> > > > bool source_support;
> > > > bool enabled;
> > > > + bool delayed_vblank;
> > > > + bool is_dpkgc_configured;
> > > > bool paused;
> > > > enum pipe pipe;
> > > > enum transcoder transcoder; diff --git
> > > > a/drivers/gpu/drm/i915/display/intel_psr.c
> > > > b/drivers/gpu/drm/i915/display/intel_psr.c
> > > > index 080bf5e51148..4ddea6737386 100644
> > > > --- a/drivers/gpu/drm/i915/display/intel_psr.c
> > > > +++ b/drivers/gpu/drm/i915/display/intel_psr.c
> > > > @@ -808,6 +808,73 @@ static u8 psr_compute_idle_frames(struct
> > > intel_dp
> > > > *intel_dp)
> > > > return idle_frames;
> > > > }
> > > >
> > > > +static bool intel_psr_check_delayed_vblank_limit(struct
> > > > intel_crtc_state *crtc_state)
> > > > +{
> > > > + struct drm_display_mode *adjusted_mode = &crtc_state-
> > > > > hw.adjusted_mode;
> > > > +
> > > > + return (adjusted_mode->crtc_vblank_start - adjusted_mode-
> > > > > crtc_vdisplay) >= 6;
> > > > +}
> > > > +
> > > > +/*
> > > > + * PKG_C_LATENCY is configured only when DISPLAY_VER >= 20 and
> > > > + * VRR is not enabled
> > > > + */
> > > > +static bool intel_psr_is_dpkgc_configured(struct
> > > > drm_i915_private
> > > > *i915)
> > > > +{
> > > > + struct intel_crtc *intel_crtc;
> > > > +
> > > > + if (DISPLAY_VER(i915) < 20)
> > > > + return false;
> > > > +
> > > > + for_each_intel_crtc(&i915->drm, intel_crtc) {
> > > > + struct intel_crtc_state *crtc_state;
> > > > +
> > > > + if (!intel_crtc->active)
> > > > + continue;
> > > > +
> > > > + crtc_state = intel_crtc->config;
> > > > +
> > > > + if (crtc_state->vrr.enable)
> > > > + return false;
> > > > + }
> > > > +
> > > > + return true;
> > > > +}
> > > > +
> > > > +/*
> > > > + * DC5 entry is only possible if vblank interrupt is disabled
> > > > + * and either psr1, psr2, edp 1.5 pr alpm is enabled on all
> > > > + * enabled encoders.
> > > > + */
> > > > +static bool intel_psr_is_dc5_entry_possible(struct
> > > > drm_i915_private
> > > > *i915)
> > > > +{
> > > > + struct intel_crtc *intel_crtc;
> > > > +
> > > > + for_each_intel_crtc(&i915->drm, intel_crtc) {
> > > > + struct intel_encoder *encoder;
> > > > + struct drm_crtc *crtc = &intel_crtc->base;
> > > > + struct drm_vblank_crtc *vblank;
> > > > +
> > > > + if (!intel_crtc->active)
> > > > + continue;
> > > > +
> > > > + vblank = drm_crtc_vblank_crtc(crtc);
> > > > +
> > > > + if (vblank->enabled)
> > > > + return false;
> > > > +
> > > > + for_each_encoder_on_crtc(&i915->drm, crtc,
> > > > encoder) {
> > > > + struct intel_dp *intel_dp =
> > > > enc_to_intel_dp(encoder);
> > > > + struct intel_psr *psr = &intel_dp->psr;
> > > > +
> > > > + if (!psr->enabled)
> > > > + return false;
> > > > + }
> > > > + }
> > > > +
> > > > + return true;
> > > > +}
> > > > +
> > > > static bool hsw_activate_psr1(struct intel_dp *intel_dp)
> > > > {
> > > > struct drm_i915_private *dev_priv = dp_to_i915(intel_dp);
> > > > @@
> > > > -815,6 +882,14 @@ static bool hsw_activate_psr1(struct intel_dp
> > > > *intel_dp)
> > > > u32 max_sleep_time = 0x1f;
> > > > u32 val = EDP_PSR_ENABLE;
> > > >
> > > > + /* WA: 16023497226*/
> > > > + if (intel_dp->psr.is_dpkgc_configured &&
> > > > + (intel_dp->psr.delayed_vblank ||
> > > > intel_psr_is_dc5_entry_possible(dev_priv))) {
> > > > + drm_dbg_kms(&dev_priv->drm,
> > > > + "PSR1 not activated as it doesn't
> > > > meet
> > > > requirements of WA:16023497226\n");
> > > > + return false;
> > > > + }
> > > > +
> > >
> > > I would recommend doing this in intel_psr_compute_config as a last
> > > step and drop patch 1. Doing it this way would be safer as it's not
> > > opening new sequence/state where psr.enabled = true and psr.active =
> > > false after intel_psr_enable_locked.
> >
> > The reason for this was I wanted to disable only psr1 based on if dc5
> > entry is possible or not.
> > Even if I call the dc5_entry_is_possible function from compute_config
> > and save it in the intel_psr state we would still end up with the seq
> > psr.enabled = true and psr.active = false unless you see a param which
> > will only activate psr2 and not psr1 in such scenario ?
> >
>
> I was thinking doing it like this:
>
> +static void wa_16023497226(struct intel_crtc_state * crtc_state) {
> + /* PSR2 not handled here. Wa not needed for Panel Replay */
> + if (crtc_state->has_sel_update || crtc_state-
> >has_panel_replay)
> + return;
> +
> + if (intel_dp->psr.is_dpkgc_configured &&
> + (intel_dp->psr.delayed_vblank ||
> + intel_psr_is_dc5_entry_possible(dev_priv))) {
> + drm_dbg_kms(&dev_priv->drm,
> + "PSR1 not enabled as it doesn't meet
> + requirements of WA:16023497226\n");
> + crtc_state->has_psr = false;
> + }
> +}
> +
> void intel_psr_compute_config(struct intel_dp *intel_dp,
> struct intel_crtc_state *crtc_state,
> struct drm_connector_state *conn_state) @@ -
> 1635,6 +1651,8 @@ void intel_psr_compute_config(struct intel_dp *intel_dp,
> return;
>
> crtc_state->has_sel_update =
> intel_sel_update_config_valid(intel_dp, crtc_state);
> +
> + wa_16023497226(crtc_state);
> }
>
> void intel_psr_get_config(struct intel_encoder *encoder,
>
> Do you see this would be possible? Current PSR2 handling in your patches is ok
> to me.
Even if has_psr as false I see that hsw_psr1_activate can be invoked since there is
No real check stopping it from getting activated
/* psr1, psr2 and panel-replay are mutually exclusive.*/
if (intel_dp->psr.panel_replay_enabled)
dg2_activate_panel_replay(intel_dp);
else if (intel_dp->psr.sel_update_enabled)
hsw_activate_psr2(intel_dp);
else
ret = hsw_activate_psr1(intel_dp);
so if it isn't psr2 or panel replay we will activate psr1 do we need to add an else if statement here in that case.
Regards,
Suraj Kandpal
>
> BR,
>
> Jouni Högander
>
> > Regards,
> > Suraj Kandpal
> >
> > >
> > > BR,
> > >
> > > Jouni Högander
> > >
> > > > val |=
> > > > EDP_PSR_IDLE_FRAMES(psr_compute_idle_frames(intel_dp));
> > > >
> > > > if (DISPLAY_VER(dev_priv) < 20) @@ -907,7 +982,10 @@
> > > > static void hsw_activate_psr2(struct intel_dp
> > > > *intel_dp)
> > > > u32 val = EDP_PSR2_ENABLE;
> > > > u32 psr_val = 0;
> > > >
> > > > - val |=
> > > > EDP_PSR2_IDLE_FRAMES(psr_compute_idle_frames(intel_dp));
> > > > + /* WA: 16023497226*/
> > > > + if (intel_dp->psr.is_dpkgc_configured &&
> > > > + intel_psr_is_dc5_entry_possible(dev_priv))
> > > > + val |=
> > > > EDP_PSR2_IDLE_FRAMES(psr_compute_idle_frames(intel_dp));
> > > >
> > > > if (DISPLAY_VER(dev_priv) < 14 &&
> > > > !IS_ALDERLAKE_P(dev_priv))
> > > > val |= EDP_SU_TRACK_ENABLE; @@ -1502,6 +1580,8 @@
> > > > void intel_psr_compute_config(struct intel_dp *intel_dp,
> > > > return;
> > > >
> > > > crtc_state->has_sel_update =
> > > > intel_sel_update_config_valid(intel_dp, crtc_state);
> > > > + intel_dp->psr.delayed_vblank =
> > > > intel_psr_check_delayed_vblank_limit(crtc_state);
> > > > + intel_dp->psr.is_dpkgc_configured =
> > > > intel_psr_is_dpkgc_configured(dev_priv);
> > > > }
> > > >
> > > > void intel_psr_get_config(struct intel_encoder *encoder,
> >
More information about the Intel-gfx
mailing list