On Fri, Dec 06, 2019 at 05:18:30PM -0800, José Roberto de Souza wrote:
> Due to DDB overlaps the pipe enabling sequence is not always crescent.
> As the previous patch selects the first pipe/transcoder in the MST
> stream to the MST master and it needs to be enabled first this
> changes were needed to guarantee that.
>
> So here it will first loop through all the MST masters and other
> pipes that do not have pipe dependencies and enabling then, as when
> the master is being enabled all the slaves are also going to a full
> modeset they will not overlap with each other.
> Then on the second loop it will enable all the MST slaves.
>
> I have tried to put port sync pipes into those two loops but
> intel_update_trans_port_sync_crtcs() is doing way more than just
> enable pipes, reading spec I guess it could be accomplish but I will
> leave it to people working on port sync.
> At least now the port sync pipes are enabled by last so the slave DDB
> allocation will not overlap with other pipes.
>
> Cc: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> Cc: Maarten Lankhorst <maarten.lankhorst@xxxxxxxxxxxxxxx>
> Cc: Matt Roper <matthew.d.roper@xxxxxxxxx>
> Cc: Manasi Navare <manasi.d.navare@xxxxxxxxx>
> Signed-off-by: José Roberto de Souza <jose.souza@xxxxxxxxx>
> ---
> drivers/gpu/drm/i915/display/intel_display.c | 120 ++++++++++++++++---
> 1 file changed, 105 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> index f89494c849ce..2f74c0bfb2a8 100644
> --- a/drivers/gpu/drm/i915/display/intel_display.c
> +++ b/drivers/gpu/drm/i915/display/intel_display.c
> @@ -14576,6 +14576,39 @@ static void intel_update_trans_port_sync_crtcs(struct intel_crtc *crtc,
> state);
> }
>
> +static void
> +skl_commit_modeset_enable_pipe(struct intel_crtc *crtc,
> + struct intel_crtc_state *old_crtc_state,
> + struct intel_crtc_state *new_crtc_state,
> + unsigned int *updated, bool *progress,
> + struct skl_ddb_entry *entry)
> +{
> + struct intel_atomic_state *state = to_intel_atomic_state(old_crtc_state->uapi.state);
> + struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
> + bool vbl_wait = false;
> +
> + *updated = *updated | BIT(crtc->pipe);
> + *progress = true;
> + *entry = new_crtc_state->wm.skl.ddb;
> +
> + /*
> + * If this is an already active pipe, it's DDB changed,
> + * and this isn't the last pipe that needs updating
> + * then we need to wait for a vblank to pass for the
> + * new ddb allocation to take effect.
> + */
> + if (!needs_modeset(new_crtc_state) &&
> + state->wm_results.dirty_pipes != *updated &&
> + !skl_ddb_entry_equal(&new_crtc_state->wm.skl.ddb,
> + &old_crtc_state->wm.skl.ddb))
> + vbl_wait = true;
> +
> + intel_update_crtc(crtc, state, old_crtc_state, new_crtc_state);
> +
> + if (vbl_wait)
> + intel_wait_for_vblank(dev_priv, crtc->pipe);
> +}
> +
> static void skl_commit_modeset_enables(struct intel_atomic_state *state)
> {
> struct drm_i915_private *dev_priv = to_i915(state->base.dev);
> @@ -14600,18 +14633,84 @@ static void skl_commit_modeset_enables(struct intel_atomic_state *state)
> /*
> * Whenever the number of active pipes changes, we need to make sure we
> * update the pipes in the right order so that their ddb allocations
> - * never overlap with eachother inbetween CRTC updates. Otherwise we'll
> + * never overlap with each other between CRTC updates. Otherwise we'll
> * cause pipe underruns and other bad stuff.
> + *
> + * First enable all the pipes that do not depends on other pipes while
> + * respecting the DDB allocation overlaps.
> + *
> + * TODO: integrate port sync to the loops bellow.
> + * Port sync is not respecting the DDB allocation overlaps as it
> + * was not checking for the slave port overlaps and there is more than
> + * just a pipe enable in intel_update_trans_port_sync_crtcs()
> */
> do {
> progress = false;
>
> - for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
> + for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
> + new_crtc_state, i) {
> + if (updated & BIT(crtc->pipe) ||
> + !new_crtc_state->hw.active)
> + continue;
> +
> + if (intel_dp_mst_is_slave_trans(new_crtc_state) ||
> + is_trans_port_sync_mode(new_crtc_state))
> + continue;
> +
> + if (skl_ddb_allocation_overlaps(&new_crtc_state->wm.skl.ddb,
> + entries,
> + INTEL_NUM_PIPES(dev_priv), i))
> + continue;
> +
> + skl_commit_modeset_enable_pipe(crtc, old_crtc_state,
> + new_crtc_state, &updated,
> + &progress, &entries[i]);
> + }
> + } while (progress);
> +
> + /*
> + * Now enable all the pipes that depends on other pipe aka MST slaves
> + */
> + do {
> + progress = false;
I think we probably want to split the modeset from the other updates
now so that we can avoid repeating all this stuff for modesets.
Something like:
do {
for_each_crtc() {
if (needs_modeset()
continue;
// current thing
}
} while (progress);
for_each_crtc() {
if (!needs_modeset())
continue;
if (port_sync || mst_slave)
continue;
WARN_ON(allocaiton_overlaps());
enable_crtc();
}
for_each_crtc() {
if (!needs_modeset())
continue;
if (port_sync)
continue;
WARN_ON(allocaiton_overlaps());
enable_crtc();
}
for_each_crtc() {
if (!needs_modeset())
continue;
if (!port_sync_master)
continue;
WARN_ON(allocaiton_overlaps());
enable_port_sync();
}
Still repetitive, but at leeast we don't have to repeat all the ddb
overlap avoidance stuff.
> +
> + for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
> + new_crtc_state, i) {
> + if (updated & BIT(crtc->pipe) ||
> + !new_crtc_state->hw.active)
> + continue;
> +
> + if (is_trans_port_sync_mode(new_crtc_state))
> + continue;
> +
> + if (skl_ddb_allocation_overlaps(&new_crtc_state->wm.skl.ddb,
> + entries,
> + INTEL_NUM_PIPES(dev_priv), i))
> + continue;
> +
> + skl_commit_modeset_enable_pipe(crtc, old_crtc_state,
> + new_crtc_state, &updated,
> + &progress, &entries[i]);
> + }
> + } while (progress);
> +
> + /* Port sync loop */
> + do {
> + progress = false;
> +
> + for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
> + new_crtc_state, i) {
> enum pipe pipe = crtc->pipe;
> bool vbl_wait = false;
> bool modeset = needs_modeset(new_crtc_state);
>
> - if (updated & BIT(crtc->pipe) || !new_crtc_state->hw.active)
> + if (updated & BIT(pipe) || !new_crtc_state->hw.active)
> + continue;
> +
> + if (!is_trans_port_sync_master(new_crtc_state))
> + continue;
> +
> + if (!modeset)
> continue;
>
> if (skl_ddb_allocation_overlaps(&new_crtc_state->wm.skl.ddb,
> @@ -14634,18 +14733,9 @@ static void skl_commit_modeset_enables(struct intel_atomic_state *state)
> state->wm_results.dirty_pipes != updated)
> vbl_wait = true;
>
> - if (modeset && is_trans_port_sync_mode(new_crtc_state)) {
> - if (is_trans_port_sync_master(new_crtc_state))
> - intel_update_trans_port_sync_crtcs(crtc,
> - state,
> - old_crtc_state,
> - new_crtc_state);
> - else
> - continue;
> - } else {
> - intel_update_crtc(crtc, state, old_crtc_state,
> - new_crtc_state);
> - }
> + intel_update_trans_port_sync_crtcs(crtc, state,
> + old_crtc_state,
> + new_crtc_state);
>
> if (vbl_wait)
> intel_wait_for_vblank(dev_priv, pipe);
> --
> 2.24.0
--
Ville Syrjälä
Intel
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
