[Freedreno] [PATCH] drm/msm: dpu: Allow planes to extend past active display

[Sean Paul]

On Wed, Aug 22, 2018 at 02:56:22PM -0700, Jeykumar Sankaran wrote:
> On 2018-08-22 13:41, Sean Paul wrote:
> > From: Sean Paul <seanpaul at chromium.org>
> > 
> > The atomic_check is a bit too aggressive with respect to planes which
> > leave the active area. This caused a bunch of log spew when the cursor
> > got to the edge of the screen and stopped it from going all the way.
> > 
> > This patch removes the conservative bounds checks from atomic and clips
> > the dst rect such that we properly display planes which go off the
> > screen.
> > 
> > Cc: Sravanthi Kollukuduru <skolluku at codeaurora.org>
> > Signed-off-by: Sean Paul <seanpaul at chromium.org>
> > ---
> >  drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c  |  3 +--
> >  drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c | 10 +++++++---
> >  2 files changed, 8 insertions(+), 5 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
> > b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
> > index 07c2d15b45f2..f0a5e776ba32 100644
> > --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
> > +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
> > @@ -1551,8 +1551,7 @@ static int dpu_crtc_atomic_check(struct drm_crtc
> > *crtc,
> >  		cnt++;
> > 
> >  		dst = drm_plane_state_dest(pstate);
> > -		if (!drm_rect_intersect(&clip, &dst) ||
> > -		    !drm_rect_equals(&clip, &dst)) {
> > +		if (!drm_rect_intersect(&clip, &dst)) {
> >  			DPU_ERROR("invalid vertical/horizontal
> > destination\n");
> >  			DPU_ERROR("display: " DRM_RECT_FMT " plane: "
> >  				  DRM_RECT_FMT "\n",
> > DRM_RECT_ARG(&crtc_rect),
> > diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
> > b/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
> > index efdf9b200dd9..720005a8ea25 100644
> > --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
> > +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
> > @@ -1254,7 +1254,7 @@ static int dpu_plane_sspp_atomic_update(struct
> > drm_plane *plane,
> >  	const struct dpu_format *fmt;
> >  	struct drm_crtc *crtc;
> >  	struct drm_framebuffer *fb;
> > -	struct drm_rect src, dst;
> > +	struct drm_rect clip = { 0 }, src, dst;
> > 
> >  	if (!plane) {
> >  		DPU_ERROR("invalid plane\n");
> > @@ -1300,14 +1300,18 @@ static int dpu_plane_sspp_atomic_update(struct
> > drm_plane *plane,
> > 
> >  	dst = drm_plane_state_dest(state);
> > 
> > +	clip.x2 = crtc->state->adjusted_mode.hdisplay;
> > +	clip.y2 = crtc->state->adjusted_mode.vdisplay;
> > +	drm_rect_intersect(&clip, &dst);
> > +
> >  	DPU_DEBUG_PLANE(pdpu, "FB[%u] " DRM_RECT_FMT "->crtc%u "
> > DRM_RECT_FMT
> >  			", %4.4s ubwc %d\n", fb->base.id,
> > DRM_RECT_ARG(&src),
> > -			crtc->base.id, DRM_RECT_ARG(&dst),
> > +			crtc->base.id, DRM_RECT_ARG(&clip),
> >  			(char *)&fmt->base.pixel_format,
> >  			DPU_FORMAT_IS_UBWC(fmt));
> > 
> >  	pdpu->pipe_cfg.src_rect = src;
> > -	pdpu->pipe_cfg.dst_rect = dst;
> > +	pdpu->pipe_cfg.dst_rect = clip;
> 
> If clip width/height is not equal to dst(which will be same as src for
> cursor)
> width/height it will require scaling which is not supported by DMA pipe
> (used for cursor planes).

It doesn't require scaling, it just requires that the scanout stops short.

Perhaps some ascii-art will explain better. This is what u/s is trying to do:

+----------------------+
|     active display   |
|                      |
|                  +---------+
|                  |  plane  |
|                  |         |
|                  +---------+
|                      |
+----------------------+

This is what the patch does:

+----------------------+
|     active display   |
|                      |
|                  +---+
|                  |  p|
|                  |   |
|                  +---+
|                      |
+----------------------+

So we're not scaling anything, just stopping short. On my device, the first
scenario results in immediate blue screen when the cursor exits the active area
(underrun?). With the patch, it works just fine since everything is confined to
the screen.

Sean


> Ideally, we should suppress the error log and reset the cursor x/y to w/h
> offset from the boundaries.
> 
> Jeykumar S.
> 
> > 
> >  	_dpu_plane_setup_scaler(pdpu, pstate, fmt, false);
> 
> -- 
> Jeykumar S

-- 
Sean Paul, Software Engineer, Google / Chromium OS