> -----Original Message-----
> From: Hogander, Jouni <jouni.hogander@xxxxxxxxx>
> Sent: Friday, September 20, 2024 11:06 AM
> To: Shankar, Uma <uma.shankar@xxxxxxxxx>; Kandpal, Suraj
> <suraj.kandpal@xxxxxxxxx>; intel-gfx@xxxxxxxxxxxxxxxxxxxxx
> Cc: Deak, Imre <imre.deak@xxxxxxxxx>
> Subject: Re: [PATCH] drm/i915/psr: Implment WA to help reach PC10
>
> On Thu, 2024-09-19 at 12:14 +0000, Shankar, Uma wrote:
> >
> >
> > > -----Original Message-----
> > > From: Kandpal, Suraj <suraj.kandpal@xxxxxxxxx>
> > > Sent: Monday, September 9, 2024 12:02 PM
> > > To: intel-gfx@xxxxxxxxxxxxxxxxxxxxx
> > > Cc: Shankar, Uma <uma.shankar@xxxxxxxxx>; Hogander, Jouni
> > > <jouni.hogander@xxxxxxxxx>; Deak, Imre <imre.deak@xxxxxxxxx>;
> > > Kandpal, Suraj <suraj.kandpal@xxxxxxxxx>
> > > Subject: [PATCH] drm/i915/psr: Implment WA to help reach PC10
> >
> > Not: Typo in implement
> >
> > > 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]
> > >
> > > --v3
> > > -Use has_psr to check if psr can be enabled or not for dc5_entry
> > > cond [Uma] - Move the dc5 entry computation to psr_compute_config
> > > [Jouni] -No need to change sequence of enabled and activate, so dont
> > > make hsw_psr1_activate return anything [Jouni] -Use has_psr to stop
> > > psr1 activation [Jouni] -Use lineage no. in WA -Add the display ver
> > > restrictions for WA
> > >
> > > --v4
> > > -use more appropriate name for check_vblank_limit() [Jouni] -Cover
> > > the case for idle frames when dpkgc is not configured [Jouni] -Check
> > > psr only for edp [Jouni]
> > >
> > > --v5
> > > -move psr1 handling to plane update [Jouni] -add todo for cases when
> > > vblank is enabled when psr enabled [Jouni] -use intel_display
> > > instead of drm_i915_private
> > >
> > > --v6
> > > -check target_dc_state [Jouni]
> > > -fix condition in pre/post plane update [Jouni]
> > >
> > > WA: 22019444797
> > > Signed-off-by: Suraj Kandpal <suraj.kandpal@xxxxxxxxx>
> > > ---
> > > .../drm/i915/display/intel_display_types.h | 3 +
> > > drivers/gpu/drm/i915/display/intel_psr.c | 112
> > > +++++++++++++++++-
> > > 2 files changed, 114 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 733de5edcfdb..59c81f0a3abd 100644
> > > --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> > > +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> > > @@ -1577,6 +1577,9 @@ struct intel_psr {
> > > #define I915_PSR_DEBUG_PANEL_REPLAY_DISABLE 0x40
> > >
> > > u32 debug;
> > > + bool is_dpkgc_configured;
> > > + bool is_dc5_entry_possible;
> > > + bool is_wa_delayed_vblank_limit;
> > > bool sink_support;
> > > bool source_support;
> > > bool enabled;
> > > diff --git a/drivers/gpu/drm/i915/display/intel_psr.c
> > > b/drivers/gpu/drm/i915/display/intel_psr.c
> > > index b30fa067ce6e..e50b476494a0 100644
> > > --- a/drivers/gpu/drm/i915/display/intel_psr.c
> > > +++ b/drivers/gpu/drm/i915/display/intel_psr.c
> > > @@ -26,6 +26,7 @@
> > > #include <drm/drm_atomic_helper.h>
> > > #include <drm/drm_damage_helper.h>
> > > #include <drm/drm_debugfs.h>
> > > +#include <drm/drm_vblank.h>
> > >
> > > #include "i915_drv.h"
> > > #include "i915_reg.h"
> > > @@ -874,6 +875,78 @@ static u8 psr_compute_idle_frames(struct
> > > intel_dp
> > > *intel_dp)
> > > return idle_frames;
> > > }
> > >
> > > +static bool
> > > +intel_psr_check_wa_delayed_vblank(const struct drm_display_mode
> > > +*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 intel_display
> > > *display,
> > > + struct intel_atomic_state
> > > *state) {
> > > + struct intel_crtc *intel_crtc;
> > > + struct intel_crtc_state *crtc_state;
> > > + int i;
> > > +
> > > + if (DISPLAY_VER(display) < 20)
> > > + return false;
> > > +
> > > + for_each_new_intel_crtc_in_state(state, intel_crtc,
> > > crtc_state, i) {
> > > + if (!intel_crtc->active)
> > > + continue;
> > > +
> > > + 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 intel_display *display,
> > > + struct intel_atomic_state *state) {
> > > + struct intel_crtc *intel_crtc;
> > > + struct intel_crtc_state *crtc_state;
> > > + int i;
> > > +
> > > + if ((display->power.domains.target_dc_state &
> > > + DC_STATE_EN_UPTO_DC5_DC6_MASK) == 0)
> > > + return false;
> > > +
> > > + for_each_new_intel_crtc_in_state(state, intel_crtc,
> > > crtc_state, i) {
> > > + struct drm_crtc *crtc = &intel_crtc->base;
> > > + struct drm_vblank_crtc *vblank;
> > > + struct intel_encoder *encoder;
> > > +
> > > + if (!intel_crtc->active)
> > > + continue;
> > > +
> > > + vblank = drm_crtc_vblank_crtc(crtc);
> > > +
> > > + if (vblank->enabled)
> > > + return false;
> > > +
> > > + if (crtc_state->has_psr)
> > > + return false;
> >
> > It should be !has_psr
> > > +
> > > + for_each_encoder_on_crtc(display->drm, crtc,
> > > encoder)
> > > + if (encoder->type != INTEL_OUTPUT_EDP)
> > > + return false;
>
> I'm not sure if we need to care about dual eDP case. One PSR and another non-
> PSR. That will return true from this function even thought it's not possible. That
> can be solved by checking CAN_PSR(intel_dp) here.
So ,
if (encoder->type != INTEL_OUTPUT_EDP && CAN_PSR(intel_dp))
return false;
>
> > > + }
> > > +
> > > + return true;
> > > +}
> > > +
> > > static void hsw_activate_psr1(struct intel_dp *intel_dp) {
> > > struct intel_display *display = to_intel_display(intel_dp);
> > > @@ -986,7
> > > +1059,15 @@ 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_22019444797
> > > + * TODO: Disable idle frames when vblank gets enabled while
> > > + * PSR2 is enabled
> > > + */
> > > + if (DISPLAY_VER(dev_priv) != 20 ||
> > > + !intel_dp->psr.is_dpkgc_configured ||
> >
> > Why ! for dpkgc, Here this can be enabled if dpkgc_configured right ?
> >
> > > + intel_dp->psr.is_dc5_entry_possible)
> > > + val |=
> > > EDP_PSR2_IDLE_FRAMES(psr_compute_idle_frames(intel_dp));
> > >
> > > if (DISPLAY_VER(display) < 14 && !IS_ALDERLAKE_P(dev_priv))
> > > val |= EDP_SU_TRACK_ENABLE; @@ -2667,10 +2748,20 @@
> > > void intel_psr_pre_plane_update(struct intel_atomic_state *state,
> > > const struct intel_crtc_state *new_crtc_state =
> > > intel_atomic_get_new_crtc_state(state, crtc);
> > > struct intel_encoder *encoder;
> > > + bool dpkgc_configured = false, dc5_entry_possible = false;
> > > + bool wa_delayed_vblank_limit = false;
> > >
> > > if (!HAS_PSR(display))
> > > return;
> > >
> > > + if (DISPLAY_VER(display) == 20) {
> > > + dpkgc_configured =
> > > intel_psr_is_dpkgc_configured(display,
> > > state);
> > > + dc5_entry_possible =
> > > + intel_psr_is_dc5_entry_possible(display,
> > > state);
> > > + wa_delayed_vblank_limit =
> > > + intel_psr_check_wa_delayed_vblank(&new_crtc
> > > _state-
> > > > hw.adjusted_mode);
> > > + }
> > > +
> > > for_each_intel_encoder_mask_with_psr(state->base.dev,
> > > encoder,
> > > old_crtc_state-
> > > >uapi.encoder_mask)
> > > {
> > > struct intel_dp *intel_dp =
> > > enc_to_intel_dp(encoder); @@ -
> > > 2679,6 +2770,12 @@ void intel_psr_pre_plane_update(struct
> > > intel_atomic_state *state,
> > >
> > > mutex_lock(&psr->lock);
> > >
> > > + if (DISPLAY_VER(i915) == 20) {
> > > + psr->is_dpkgc_configured =
> > > dpkgc_configured;
> > > + psr->is_dc5_entry_possible =
> > > dc5_entry_possible;
> > > + psr->is_wa_delayed_vblank_limit =
> > > wa_delayed_vblank_limit;
> >
> > We can drop the variables and directly assign this to psr->... and use
> > it subsequently.
> > Also it would be good to have this done in compute and than just use
> > it here.
> >
> > > + }
> > > +
> > > /*
> > > * Reasons to disable:
> > > * - PSR disabled in new state @@ -2686,6 +2783,7 @@
> > > void intel_psr_pre_plane_update(struct intel_atomic_state *state,
> > > * - Changing between PSR versions
> > > * - Region Early Transport changing
> > > * - Display WA #1136: skl, bxt
> > > + * - Display WA_22019444797
> > > */
> > > needs_to_disable |=
> > > intel_crtc_needs_modeset(new_crtc_state);
> > > needs_to_disable |= !new_crtc_state->has_psr; @@ -
> > > 2695,6
> > > +2793,10 @@ void intel_psr_pre_plane_update(struct
> > > intel_atomic_state *state,
> > > psr->su_region_et_enabled;
> > > needs_to_disable |= DISPLAY_VER(i915) < 11 &&
> > > new_crtc_state->wm_level_disabled;
> > > + /* TODO: Disable PSR1 when vblank gets enabled
> > > while PSR1 is
> > > enabled */
> > > + needs_to_disable |= DISPLAY_VER(display) == 20 &&
> > > dpkgc_configured &&
> > > + (wa_delayed_vblank_limit ||
> > > dc5_entry_possible) &&
> > > + !new_crtc_state->has_sel_update &&
> > > +!new_crtc_state->has_panel_replay;
> >
> > Good to move this to a small helper function which can be extended if
> > required as well.
> > Also seems used in post_plane as well.
> >
> > @Hogander, Jouni Can you also ack if this handling for PSR is ok.
>
> This need_to_disable/keep_disabled is ok. I think there is a bug in check itself:
>
> dc5_entry_possible should be !dc5_entry_possible
>
> What do you think?
>
Dc5_entry_possible returns true when we can enter dc5.
And the condition to disable ps1 is to have a delayed_vblank > 6 or
When dc5 can be entered so that check would be correct
Regards,
Suraj Kandpal
> BR,
>
> Jouni Högander
>
> >
> > > if (psr->enabled && needs_to_disable)
> > > intel_psr_disable_locked(intel_dp);
> > > @@ -2735,6 +2837,14 @@ void intel_psr_post_plane_update(struct
> > > intel_atomic_state *state,
> > > keep_disabled |= DISPLAY_VER(display) < 11 &&
> > > crtc_state->wm_level_disabled;
> > >
> > > + /*
> > > + * Wa_22019444797
> > > + * TODO: Disable PSR1 when vblank gets enabled
> > > while PSR1 is
> > > enabled
> > > + */
> > > + keep_disabled |= DISPLAY_VER(display) == 20 && psr-
> > > > is_dpkgc_configured &&
> > > + (psr->is_wa_delayed_vblank_limit || psr-
> > > > is_dc5_entry_possible) &&
> > > + !crtc_state->has_sel_update && !crtc_state-
> > > > has_panel_replay;
> > > +
> > > if (!psr->enabled && !keep_disabled)
> > > intel_psr_enable_locked(intel_dp,
> > > crtc_state);
> > > else if (psr->enabled && !crtc_state-
> > > >wm_level_disabled)
> > > --
> > > 2.43.2
> >
