On Mon, Jan 20, 2025 at 06:53:24PM +0200, Jani Nikula wrote:
> On Thu, 16 Jan 2025, Ville Syrjala <ville.syrjala@xxxxxxxxxxxxxxx> wrote:
> > From: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> >
> > GOP might not agree with our idea of what the vblank delay should be.
> > Reuse the LRR codepaths to fix that up via a fastset.
> >
> > The relevant registers aren't actually double buffered so this is a
> > little bit dodgy. While I've not seen any real issues from frobbing
> > these live, let's limit this to just the fastboot case (by only
> > allowing it when old_crtc_state->inherited==true).
> >
> > Cc: Paz Zcharya <pazz@xxxxxxxxxxxx>
> > Signed-off-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> > ---
> > drivers/gpu/drm/i915/display/intel_display.c | 26 +++++++++++++++++---
> > 1 file changed, 22 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> > index 9b42d980ed7e..471fe7d80f30 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -5358,6 +5358,19 @@ pipe_config_cx0pll_mismatch(struct drm_printer *p, bool fastset,
> > intel_cx0pll_dump_hw_state(display, b);
> > }
> >
> > +static bool allow_vblank_delay_fastset(const struct intel_crtc_state *old_crtc_state)
> > +{
> > + struct intel_display *display = to_intel_display(old_crtc_state);
> > +
> > + /*
> > + * Allow fastboot to fix up vblank delay (handled via LRR
> > + * codepaths), a bit dodgy as the registers aren't
> > + * double buffered but seems to be working more or less...
> > + */
> > + return HAS_LRR(display) && old_crtc_state->inherited &&
> > + !intel_crtc_has_type(old_crtc_state, INTEL_OUTPUT_DSI);
> > +}
> > +
> > bool
> > intel_pipe_config_compare(const struct intel_crtc_state *current_config,
> > const struct intel_crtc_state *pipe_config,
> > @@ -5490,7 +5503,8 @@ intel_pipe_config_compare(const struct intel_crtc_state *current_config,
> > PIPE_CONF_CHECK_I(name.crtc_hsync_start); \
> > PIPE_CONF_CHECK_I(name.crtc_hsync_end); \
> > PIPE_CONF_CHECK_I(name.crtc_vdisplay); \
> > - PIPE_CONF_CHECK_I(name.crtc_vblank_start); \
> > + if (!fastset || !allow_vblank_delay_fastset(current_config)) \
> > + PIPE_CONF_CHECK_I(name.crtc_vblank_start); \
>
> Side note, part of me wants to change fastset to some variation of "full
> modeset" because *all* the compare checks are for !fastset.
fastset==true here indicates that we're checking whether fastset
is possible, fastset==false indicates we're running the actual
state checker. Not quite sure what the best naming for the
opposite situation would be.
>
> Reviewed-by: Jani Nikula <jani.nikula@xxxxxxxxx>
Ta.
>
> > PIPE_CONF_CHECK_I(name.crtc_vsync_start); \
> > PIPE_CONF_CHECK_I(name.crtc_vsync_end); \
> > if (!fastset || !pipe_config->update_lrr) { \
> > @@ -6084,7 +6098,8 @@ static int intel_modeset_checks(struct intel_atomic_state *state)
> > static bool lrr_params_changed(const struct drm_display_mode *old_adjusted_mode,
> > const struct drm_display_mode *new_adjusted_mode)
> > {
> > - return old_adjusted_mode->crtc_vblank_end != new_adjusted_mode->crtc_vblank_end ||
> > + return old_adjusted_mode->crtc_vblank_start != new_adjusted_mode->crtc_vblank_start ||
> > + old_adjusted_mode->crtc_vblank_end != new_adjusted_mode->crtc_vblank_end ||
> > old_adjusted_mode->crtc_vtotal != new_adjusted_mode->crtc_vtotal;
> > }
> >
> > @@ -6098,11 +6113,14 @@ static void intel_crtc_check_fastset(const struct intel_crtc_state *old_crtc_sta
> > if (old_crtc_state->vrr.in_range != new_crtc_state->vrr.in_range)
> > new_crtc_state->update_lrr = false;
> >
> > - if (!intel_pipe_config_compare(old_crtc_state, new_crtc_state, true))
> > + if (!intel_pipe_config_compare(old_crtc_state, new_crtc_state, true)) {
> > drm_dbg_kms(&i915->drm, "[CRTC:%d:%s] fastset requirement not met, forcing full modeset\n",
> > crtc->base.base.id, crtc->base.name);
> > - else
> > + } else {
> > + if (allow_vblank_delay_fastset(old_crtc_state))
> > + new_crtc_state->update_lrr = true;
> > new_crtc_state->uapi.mode_changed = false;
> > + }
> >
> > if (intel_compare_link_m_n(&old_crtc_state->dp_m_n,
> > &new_crtc_state->dp_m_n))
>
> --
> Jani Nikula, Intel
--
Ville Syrjälä
Intel
