diff mbox series

[v4,11/14] drm/atomic-helper: Set fence deadline for vblank

Message ID 20230218211608.1630586-12-robdclark@gmail.com
State New
Headers
Series dma-fence: Deadline awareness |

Commit Message

Rob Clark Feb. 18, 2023, 9:15 p.m. UTC 
  From: Rob Clark <robdclark@chromium.org>

For an atomic commit updating a single CRTC (ie. a pageflip) calculate
the next vblank time, and inform the fence(s) of that deadline.

v2: Comment typo fix (danvet)

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Rob Clark <robdclark@chromium.org>
---
 drivers/gpu/drm/drm_atomic_helper.c | 36 +++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

Comments

Luben Tuikov Feb. 22, 2023, 10:46 a.m. UTC | #1 
On 2023-02-18 16:15, Rob Clark wrote:
> From: Rob Clark <robdclark@chromium.org>
> 
> For an atomic commit updating a single CRTC (ie. a pageflip) calculate
> the next vblank time, and inform the fence(s) of that deadline.
> 
> v2: Comment typo fix (danvet)
> 
> Signed-off-by: Rob Clark <robdclark@chromium.org>
> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> Signed-off-by: Rob Clark <robdclark@chromium.org>
> ---
>  drivers/gpu/drm/drm_atomic_helper.c | 36 +++++++++++++++++++++++++++++
>  1 file changed, 36 insertions(+)
> 
> diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c
> index d579fd8f7cb8..35a4dc714920 100644
> --- a/drivers/gpu/drm/drm_atomic_helper.c
> +++ b/drivers/gpu/drm/drm_atomic_helper.c
> @@ -1511,6 +1511,40 @@ void drm_atomic_helper_commit_modeset_enables(struct drm_device *dev,
>  }
>  EXPORT_SYMBOL(drm_atomic_helper_commit_modeset_enables);
>  
> +/*
> + * For atomic updates which touch just a single CRTC, calculate the time of the
> + * next vblank, and inform all the fences of the deadline.
> + */
> +static void set_fence_deadline(struct drm_device *dev,
> +			       struct drm_atomic_state *state)
> +{
> +	struct drm_crtc *crtc, *wait_crtc = NULL;
> +	struct drm_crtc_state *new_crtc_state;
> +	struct drm_plane *plane;
> +	struct drm_plane_state *new_plane_state;
> +	ktime_t vbltime;

I've not looked at the latest language spec, but is AFAIR "vbltime"
would be uninitialized here. Has this changed?

> +	int i;
> +
> +	for_each_new_crtc_in_state (state, crtc, new_crtc_state, i) {
> +		if (wait_crtc)
> +			return;
> +		wait_crtc = crtc;
> +	}
> +
> +	/* If no CRTCs updated, then nothing to do: */
> +	if (!wait_crtc)
> +		return;
> +
> +	if (drm_crtc_next_vblank_time(wait_crtc, &vbltime))
> +		return;

We have a problem here in that we're adding the time remaining to the next
vblank event to an uninitialized local variable. As per my comment on patch 10,
I'd rather drm_crtc_next_vblank_time() yield the time remaining to the vblank event,
and we can do the arithmetic locally here in this function.
Rob Clark Feb. 22, 2023, 3:50 p.m. UTC | #2 
On Wed, Feb 22, 2023 at 2:46 AM Luben Tuikov <luben.tuikov@amd.com> wrote:
>
> On 2023-02-18 16:15, Rob Clark wrote:
> > From: Rob Clark <robdclark@chromium.org>
> >
> > For an atomic commit updating a single CRTC (ie. a pageflip) calculate
> > the next vblank time, and inform the fence(s) of that deadline.
> >
> > v2: Comment typo fix (danvet)
> >
> > Signed-off-by: Rob Clark <robdclark@chromium.org>
> > Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> > Signed-off-by: Rob Clark <robdclark@chromium.org>
> > ---
> >  drivers/gpu/drm/drm_atomic_helper.c | 36 +++++++++++++++++++++++++++++
> >  1 file changed, 36 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c
> > index d579fd8f7cb8..35a4dc714920 100644
> > --- a/drivers/gpu/drm/drm_atomic_helper.c
> > +++ b/drivers/gpu/drm/drm_atomic_helper.c
> > @@ -1511,6 +1511,40 @@ void drm_atomic_helper_commit_modeset_enables(struct drm_device *dev,
> >  }
> >  EXPORT_SYMBOL(drm_atomic_helper_commit_modeset_enables);
> >
> > +/*
> > + * For atomic updates which touch just a single CRTC, calculate the time of the
> > + * next vblank, and inform all the fences of the deadline.
> > + */
> > +static void set_fence_deadline(struct drm_device *dev,
> > +                            struct drm_atomic_state *state)
> > +{
> > +     struct drm_crtc *crtc, *wait_crtc = NULL;
> > +     struct drm_crtc_state *new_crtc_state;
> > +     struct drm_plane *plane;
> > +     struct drm_plane_state *new_plane_state;
> > +     ktime_t vbltime;
>
> I've not looked at the latest language spec, but is AFAIR "vbltime"
> would be uninitialized here. Has this changed?
>
> > +     int i;
> > +
> > +     for_each_new_crtc_in_state (state, crtc, new_crtc_state, i) {
> > +             if (wait_crtc)
> > +                     return;
> > +             wait_crtc = crtc;
> > +     }
> > +
> > +     /* If no CRTCs updated, then nothing to do: */
> > +     if (!wait_crtc)
> > +             return;
> > +
> > +     if (drm_crtc_next_vblank_time(wait_crtc, &vbltime))
> > +             return;
>
> We have a problem here in that we're adding the time remaining to the next
> vblank event to an uninitialized local variable. As per my comment on patch 10,
> I'd rather drm_crtc_next_vblank_time() yield the time remaining to the vblank event,
> and we can do the arithmetic locally here in this function.

if drm_crtc_next_vblank_time() returns 0 then it has initialized
vbltime, so no problem here

BR,
-R

> --
> Regards,
> Luben
>
> > +
> > +     for_each_new_plane_in_state (state, plane, new_plane_state, i) {
> > +             if (!new_plane_state->fence)
> > +                     continue;
> > +             dma_fence_set_deadline(new_plane_state->fence, vbltime);
> > +     }
> > +}
> > +
> >  /**
> >   * drm_atomic_helper_wait_for_fences - wait for fences stashed in plane state
> >   * @dev: DRM device
> > @@ -1540,6 +1574,8 @@ int drm_atomic_helper_wait_for_fences(struct drm_device *dev,
> >       struct drm_plane_state *new_plane_state;
> >       int i, ret;
> >
> > +     set_fence_deadline(dev, state);
> > +
> >       for_each_new_plane_in_state(state, plane, new_plane_state, i) {
> >               if (!new_plane_state->fence)
> >                       continue;
>
diff mbox series

Patch

diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c
index d579fd8f7cb8..35a4dc714920 100644
--- a/drivers/gpu/drm/drm_atomic_helper.c
+++ b/drivers/gpu/drm/drm_atomic_helper.c
@@ -1511,6 +1511,40 @@  void drm_atomic_helper_commit_modeset_enables(struct drm_device *dev,
 }
 EXPORT_SYMBOL(drm_atomic_helper_commit_modeset_enables);
 
+/*
+ * For atomic updates which touch just a single CRTC, calculate the time of the
+ * next vblank, and inform all the fences of the deadline.
+ */
+static void set_fence_deadline(struct drm_device *dev,
+			       struct drm_atomic_state *state)
+{
+	struct drm_crtc *crtc, *wait_crtc = NULL;
+	struct drm_crtc_state *new_crtc_state;
+	struct drm_plane *plane;
+	struct drm_plane_state *new_plane_state;
+	ktime_t vbltime;
+	int i;
+
+	for_each_new_crtc_in_state (state, crtc, new_crtc_state, i) {
+		if (wait_crtc)
+			return;
+		wait_crtc = crtc;
+	}
+
+	/* If no CRTCs updated, then nothing to do: */
+	if (!wait_crtc)
+		return;
+
+	if (drm_crtc_next_vblank_time(wait_crtc, &vbltime))
+		return;
+
+	for_each_new_plane_in_state (state, plane, new_plane_state, i) {
+		if (!new_plane_state->fence)
+			continue;
+		dma_fence_set_deadline(new_plane_state->fence, vbltime);
+	}
+}
+
 /**
  * drm_atomic_helper_wait_for_fences - wait for fences stashed in plane state
  * @dev: DRM device
@@ -1540,6 +1574,8 @@  int drm_atomic_helper_wait_for_fences(struct drm_device *dev,
 	struct drm_plane_state *new_plane_state;
 	int i, ret;
 
+	set_fence_deadline(dev, state);
+
 	for_each_new_plane_in_state(state, plane, new_plane_state, i) {
 		if (!new_plane_state->fence)
 			continue;