From: "Ville Syrjälä" <ville.syrjala@linux.intel.com>
To: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 08/14] drm/i915: Skip useless watermark/FIFO related work on VLV/CHV when not needed
Date: Thu, 16 Feb 2017 19:54:24 +0200 [thread overview]
Message-ID: <20170216175424.GL31595@intel.com> (raw)
In-Reply-To: <a24a02b6-da42-e7b6-5cf9-c7f2865bdabc@linux.intel.com>
On Thu, Dec 15, 2016 at 04:37:53PM +0100, Maarten Lankhorst wrote:
> Op 12-12-16 om 21:35 schreef ville.syrjala@linux.intel.com:
> > From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> >
> > Check whether anything relevant has actually change when we compute new
> > watermarks for each plane in the state. If the watermarks for no
> > primary/sprite planes changed we don't have to recompute the FIFO split
> > or reprogram the DSBARB registers. And even the cursor watermarks didn't
> > change we can skip the merge+invert step between all the planes on
> > the pipe as well.
> >
> > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > ---
> > drivers/gpu/drm/i915/intel_atomic.c | 1 +
> > drivers/gpu/drm/i915/intel_drv.h | 1 +
> > drivers/gpu/drm/i915/intel_pm.c | 73 +++++++++++++++++++++++++++++--------
> > 3 files changed, 60 insertions(+), 15 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/intel_atomic.c b/drivers/gpu/drm/i915/intel_atomic.c
> > index c5a166752eda..df33f270b459 100644
> > --- a/drivers/gpu/drm/i915/intel_atomic.c
> > +++ b/drivers/gpu/drm/i915/intel_atomic.c
> > @@ -99,6 +99,7 @@ intel_crtc_duplicate_state(struct drm_crtc *crtc)
> > crtc_state->update_wm_pre = false;
> > crtc_state->update_wm_post = false;
> > crtc_state->fb_changed = false;
> > + crtc_state->fifo_changed = false;
> > crtc_state->wm.need_postvbl_update = false;
> > crtc_state->fb_bits = 0;
> This flag is only used in intel_pm.c, maybe put it in crtc_state->wm.vlv.fifo_changed, and clear it in the beginning of the wm recalculation?
I don't really like that. The state we get from duplicate_state() should
be directly useable IMO. It was already checked after all. So all these
kinds of flags should be reset immediately I think.
I guess modeset is a bit of an exception to my logic there. That is I'm
forcing the flag back to true wheever there's a modeset. Alternatively
I could restore the registers whenever the power well gets turned back
on, but this way is a little simpler I think.
> > diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
> > index 66668c18a47a..a92857864ee8 100644
> > --- a/drivers/gpu/drm/i915/intel_drv.h
> > +++ b/drivers/gpu/drm/i915/intel_drv.h
> > @@ -560,6 +560,7 @@ struct intel_crtc_state {
> > bool disable_cxsr;
> > bool update_wm_pre, update_wm_post; /* watermarks are updated */
> > bool fb_changed; /* fb on any of the planes is changed */
> > + bool fifo_changed; /* FIFO split is changed */
> >
> > /* Pipe source size (ie. panel fitter input size)
> > * All planes will be positioned inside this space,
> > diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
> > index f68b46eed224..c7cc62cf51f6 100644
> > --- a/drivers/gpu/drm/i915/intel_pm.c
> > +++ b/drivers/gpu/drm/i915/intel_pm.c
> > @@ -1103,31 +1103,36 @@ static u16 vlv_invert_wm_value(u16 wm, u16 fifo_size)
> > }
> >
> > /* starting from 'level' set all higher levels to 'value' */
> > -static void vlv_plane_wm_set(struct intel_crtc_state *crtc_state,
> > +static bool vlv_plane_wm_set(struct intel_crtc_state *crtc_state,
> > int level, enum plane_id plane_id, u16 value)
> > {
> > struct drm_i915_private *dev_priv = to_i915(crtc_state->base.crtc->dev);
> > int num_levels = vlv_num_wm_levels(dev_priv);
> > + bool dirty = false;
> >
> > for (; level < num_levels; level++) {
> > struct vlv_pipe_wm *noninverted =
> > &crtc_state->wm.vlv.noninverted[level];
> >
> > + dirty |= noninverted->plane[plane_id] != value;
> > noninverted->plane[plane_id] = value;
> > }
> > +
> > + return dirty;
> > }
> >
> > -static void vlv_plane_wm_compute(struct intel_crtc_state *crtc_state,
> > +static bool vlv_plane_wm_compute(struct intel_crtc_state *crtc_state,
> > const struct intel_plane_state *plane_state)
> > {
> > struct intel_plane *plane = to_intel_plane(plane_state->base.plane);
> > enum plane_id plane_id = plane->id;
> > int num_levels = vlv_num_wm_levels(to_i915(plane->base.dev));
> > int level;
> > + bool dirty = false;
> >
> > if (!plane_state->base.visible) {
> > - vlv_plane_wm_set(crtc_state, 0, plane_id, 0);
> > - return;
> > + dirty |= vlv_plane_wm_set(crtc_state, 0, plane_id, 0);
> > + goto out;
> > }
> >
> > for (level = 0; level < num_levels; level++) {
> > @@ -1143,17 +1148,22 @@ static void vlv_plane_wm_compute(struct intel_crtc_state *crtc_state,
> > if (wm > max_wm)
> > break;
> >
> > + dirty |= noninverted->plane[plane_id] != wm;
> > noninverted->plane[plane_id] = wm;
> > }
> >
> > /* mark all higher levels as invalid */
> > - vlv_plane_wm_set(crtc_state, level, plane_id, USHRT_MAX);
> > + dirty |= vlv_plane_wm_set(crtc_state, level, plane_id, USHRT_MAX);
> >
> > - DRM_DEBUG_KMS("%s wms: [0]=%d,[1]=%d,[2]=%d\n",
> > - plane->base.name,
> > - crtc_state->wm.vlv.noninverted[VLV_WM_LEVEL_PM2].plane[plane_id],
> > - crtc_state->wm.vlv.noninverted[VLV_WM_LEVEL_PM5].plane[plane_id],
> > - crtc_state->wm.vlv.noninverted[VLV_WM_LEVEL_DDR_DVFS].plane[plane_id]);
> > +out:
> > + if (dirty)
> > + DRM_DEBUG_KMS("%s wms: [0]=%d,[1]=%d,[2]=%d\n",
> > + plane->base.name,
> > + crtc_state->wm.vlv.noninverted[VLV_WM_LEVEL_PM2].plane[plane_id],
> > + crtc_state->wm.vlv.noninverted[VLV_WM_LEVEL_PM5].plane[plane_id],
> > + crtc_state->wm.vlv.noninverted[VLV_WM_LEVEL_DDR_DVFS].plane[plane_id]);
> > +
> > + return dirty;
> > }
> >
> > static bool vlv_plane_wm_is_valid(const struct intel_crtc_state *crtc_state,
> > @@ -1186,10 +1196,12 @@ static int vlv_compute_pipe_wm(struct intel_crtc_state *crtc_state)
> > &crtc_state->wm.vlv.fifo_state;
> > int num_active_planes = hweight32(crtc_state->active_planes &
> > ~BIT(PLANE_CURSOR));
> > + bool needs_modeset = drm_atomic_crtc_needs_modeset(&crtc_state->base);
> > struct intel_plane_state *plane_state;
> > struct intel_plane *plane;
> > enum plane_id plane_id;
> > int level, ret, i;
> > + unsigned int dirty = 0;
> >
> > for_each_intel_plane_in_state(state, plane, plane_state, i) {
> > const struct intel_plane_state *old_plane_state =
> > @@ -1199,7 +1211,37 @@ static int vlv_compute_pipe_wm(struct intel_crtc_state *crtc_state)
> > old_plane_state->base.crtc != &crtc->base)
> > continue;
> >
> > - vlv_plane_wm_compute(crtc_state, plane_state);
> > + if (vlv_plane_wm_compute(crtc_state, plane_state))
> > + dirty |= BIT(plane->id);
> > + }
> > +
> > + /*
> > + * DSPARB registers may have been reset due to the
> > + * power well being turned off. Make sure we restore
> > + * them to a consistent state even if no primary/sprite
> > + * planes are initially active.
> > + */
> > + if (needs_modeset)
> > + crtc_state->fifo_changed = true;
> > +
> > + if (!dirty)
> > + return 0;
> > +
> > + /* cursor changes don't warrant a FIFO recompute */
> > + if (dirty & ~BIT(PLANE_CURSOR)) {
> > + const struct intel_crtc_state *old_crtc_state =
> > + to_intel_crtc_state(crtc->base.state);
> > + const struct vlv_fifo_state *old_fifo_state =
> > + &old_crtc_state->wm.vlv.fifo_state;
> > +
> > + ret = vlv_compute_fifo(crtc_state);
> > + if (ret)
> > + return ret;
> > +
> > + if (needs_modeset ||
> > + memcmp(old_fifo_state, fifo_state,
> > + sizeof(*fifo_state)) != 0)
> > + crtc_state->fifo_changed = true;
> > }
> >
> > /* initially allow all levels */
> > @@ -1212,10 +1254,6 @@ static int vlv_compute_pipe_wm(struct intel_crtc_state *crtc_state)
> > wm_state->cxsr = crtc->pipe != PIPE_C &&
> > crtc->wm.cxsr_allowed && num_active_planes == 1;
> >
> > - ret = vlv_compute_fifo(crtc_state);
> > - if (ret)
> > - return ret;
> > -
> > for (level = 0; level < wm_state->num_levels; level++) {
> > const struct vlv_pipe_wm *noninverted =
> > &crtc_state->wm.vlv.noninverted[level];
> > @@ -1265,6 +1303,9 @@ static void vlv_atomic_update_fifo(struct intel_atomic_state *state,
> > &crtc_state->wm.vlv.fifo_state;
> > int sprite0_start, sprite1_start, fifo_size;
> >
> > + if (!crtc_state->fifo_changed)
> > + return;
> > +
> > sprite0_start = fifo_state->plane[PLANE_PRIMARY];
> > sprite1_start = fifo_state->plane[PLANE_SPRITE0] + sprite0_start;
> > fifo_size = fifo_state->plane[PLANE_SPRITE1] + sprite1_start;
> > @@ -4631,6 +4672,8 @@ void vlv_wm_get_hw_state(struct drm_device *dev)
> > active->num_levels = wm->level + 1;
> > active->cxsr = wm->cxsr;
> >
> > + vlv_get_fifo_size(crtc_state);
> > +
> > /* FIXME sanitize things more */
> > for (level = 0; level < active->num_levels; level++) {
> > struct vlv_pipe_wm *noninverted =
>
--
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2017-02-16 17:54 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-12-12 20:35 [PATCH 00/14] drm/i915: VLV/CHV two-stage watermarks ville.syrjala
2016-12-12 20:35 ` [PATCH 01/14] drm/i915: Track visible planes in a bitmask ville.syrjala
2016-12-15 14:56 ` Maarten Lankhorst
2016-12-15 15:02 ` Ville Syrjälä
2016-12-15 17:11 ` Maarten Lankhorst
2016-12-15 17:20 ` Ville Syrjälä
2016-12-12 20:35 ` [PATCH 02/14] drm/i915: Track plane fifo sizes under intel_crtc ville.syrjala
2016-12-15 14:58 ` Maarten Lankhorst
2017-02-16 17:48 ` Ville Syrjälä
2016-12-12 20:35 ` [PATCH 03/14] drm/i915: Move vlv wms from crtc->wm_state to crtc->wm.active.vlv ville.syrjala
2016-12-12 20:35 ` [PATCH 04/14] drm/i915: Plop vlv wm state into crtc_state ville.syrjala
2016-12-12 20:35 ` [PATCH 05/14] drm/i915: Plop vlv/chv fifo sizes into crtc state ville.syrjala
2016-12-12 20:35 ` [PATCH 06/14] drm/i915: Compute VLV/CHV FIFO sizes based on the PM2 watermarks ville.syrjala
2016-12-12 20:35 ` [PATCH 07/14] drm/i915: Compute vlv/chv wms the atomic way ville.syrjala
2016-12-15 15:30 ` Maarten Lankhorst
2016-12-15 15:38 ` Ville Syrjälä
2016-12-15 15:45 ` Maarten Lankhorst
2016-12-15 16:09 ` Ville Syrjälä
2016-12-15 17:12 ` Maarten Lankhorst
2016-12-15 17:17 ` Ville Syrjälä
2016-12-29 15:42 ` Maarten Lankhorst
2016-12-12 20:35 ` [PATCH 08/14] drm/i915: Skip useless watermark/FIFO related work on VLV/CHV when not needed ville.syrjala
2016-12-15 15:37 ` Maarten Lankhorst
2017-02-16 17:54 ` Ville Syrjälä [this message]
2016-12-12 20:35 ` [PATCH 09/14] drm/i915: Compute proper intermediate wms for vlv/cvh ville.syrjala
2016-12-12 20:35 ` [PATCH 10/14] drm/i915: Nuke crtc->wm.cxsr_allowed ville.syrjala
2016-12-12 20:35 ` [PATCH 11/14] drm/i915: Only use update_wm_{pre, post} for pre-ilk platforms ville.syrjala
2016-12-12 20:35 ` [PATCH 12/14] drm/i915: Sanitize VLV/CHV watermarks properly ville.syrjala
2016-12-12 20:35 ` [PATCH 13/14] drm/i915: Workaround VLV/CHV sprite1->sprite0 enable underrun ville.syrjala
2016-12-15 15:34 ` Maarten Lankhorst
2016-12-15 15:47 ` Ville Syrjälä
2016-12-15 16:13 ` Maarten Lankhorst
2016-12-15 16:18 ` Ville Syrjälä
2016-12-12 20:35 ` [PATCH 14/14] drm/i915: Kill level 0 wm hack for VLV/CHV ville.syrjala
2016-12-15 17:10 ` Maarten Lankhorst
2016-12-12 20:35 ` [PATCH i-g-t] tests: Add kms_plane_blinker ville.syrjala
2016-12-15 15:17 ` Maarten Lankhorst
2016-12-15 15:23 ` Ville Syrjälä
2016-12-15 15:28 ` Maarten Lankhorst
2016-12-15 15:36 ` Ville Syrjälä
2016-12-15 15:41 ` Maarten Lankhorst
2016-12-15 16:42 ` Ville Syrjälä
2016-12-15 17:26 ` Ville Syrjälä
2016-12-19 7:07 ` Maarten Lankhorst
2016-12-12 21:15 ` ✗ Fi.CI.BAT: warning for drm/i915: VLV/CHV two-stage watermarks Patchwork
2016-12-13 7:42 ` Saarinen, Jani
2016-12-13 8:40 ` Chris Wilson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170216175424.GL31595@intel.com \
--to=ville.syrjala@linux.intel.com \
--cc=intel-gfx@lists.freedesktop.org \
--cc=maarten.lankhorst@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox