[PATCH i-g-t 0/2] Improve error report when there are less overlays than expected

[PATCH i-g-t 0/2] Improve error report when there are less overlays than expected

[Tomeu Vizoso]

Hi,

kms_plane_scaling would happily try to use the cursor plane as if it was an
overlay, and the first signal of it would be the kernel refusing the SetPlane
call because of the pixel format not matching.

To improve this situation, we check the format of the framebuffer before
calling SetPlane (printing the supported formats), and skip the test if there
aren't enough overlay planes.

Thanks,

Tomeu

Tomeu Vizoso (2):
  lib: Check format of framebuffer before SetPlane
  kms_plane_scaling: Skip if we don't have enough overlays

 lib/igt_kms.c             | 35 +++++++++++++++++++++++++++++++++++
 tests/kms_plane_scaling.c |  4 ++++
 2 files changed, 39 insertions(+)

-- 
2.7.4

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 1/2] lib: Check format of framebuffer before SetPlane

[Tomeu Vizoso]

Avoid setting a framebuffer in a format that the plane doesn't support,
so we have better debug output in IGT and we don't have to dig into
dmesg files to find out what's going on.

I found this issue when kms_plane_scaling tried to use a cursor plane as
a regular one in a Skylake machine with just one overlay.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
---
 lib/igt_kms.c | 35 +++++++++++++++++++++++++++++++++++
 1 file changed, 35 insertions(+)

diff --git a/lib/igt_kms.c b/lib/igt_kms.c
index 989704e14803..3234fa1bbdc4 100644
--- a/lib/igt_kms.c
+++ b/lib/igt_kms.c
@@ -1670,6 +1670,14 @@ static uint32_t igt_plane_get_fb_id(igt_plane_t *plane)
 		return 0;
 }
 
+static uint32_t igt_plane_get_fb_format(igt_plane_t *plane)
+{
+	if (plane->fb)
+		return plane->fb->drm_format;
+	else
+		return 0;
+}
+
 static uint32_t igt_plane_get_fb_gem_handle(igt_plane_t *plane)
 {
 	if (plane->fb)
@@ -1678,6 +1686,28 @@ static uint32_t igt_plane_get_fb_gem_handle(igt_plane_t *plane)
 		return 0;
 }
 
+static bool igt_plane_supports_format(igt_plane_t *plane, uint32_t format)
+{
+	int i;
+
+	for (i = 0; i < plane->drm_plane->count_formats; i++) {
+		if (format == plane->drm_plane->formats[i])
+			return true;
+	}
+
+	igt_debug("Plane %d in pipe %s doesn't support format %s.\n",
+		  plane->index,
+	          kmstest_pipe_name(plane->pipe->pipe),
+	          igt_format_str(format));
+
+	igt_debug("Formats supported are:");
+	for (i = 0; i < plane->drm_plane->count_formats; i++)
+		igt_debug(" %s", igt_format_str(plane->drm_plane->formats[i]));
+	igt_debug(".\n");
+
+	return false;
+}
+
 #define CHECK_RETURN(r, fail) {	\
 	if (r && !fail)		\
 		return r;	\
@@ -1772,6 +1802,7 @@ static int igt_drm_plane_commit(igt_plane_t *plane,
 	int32_t crtc_y;
 	uint32_t crtc_w;
 	uint32_t crtc_h;
+	uint32_t format;
 
 	igt_assert(plane->drm_plane);
 
@@ -1821,6 +1852,10 @@ static int igt_drm_plane_commit(igt_plane_t *plane,
 		    src_x >> 16, src_y >> 16, src_w >> 16, src_h >> 16,
 		    crtc_x, crtc_y, crtc_w, crtc_h);
 
+		/* it's an error to try an unsupported format */
+		format = igt_plane_get_fb_format(plane);
+		igt_assert(igt_plane_supports_format(plane, format));
+
 		ret = drmModeSetPlane(display->drm_fd,
 				      plane->drm_plane->plane_id,
 				      crtc_id,
-- 
2.7.4

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 2/2] kms_plane_scaling: Skip if we don't have enough overlays

[Tomeu Vizoso]

The test currently assumes that there are two overlay planes available,
but that's not generally true and from the error that returns the
kernel, it isn't obvious what's going on.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
---
 tests/kms_plane_scaling.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/tests/kms_plane_scaling.c b/tests/kms_plane_scaling.c
index 4546ce548515..0e2a875c6542 100644
--- a/tests/kms_plane_scaling.c
+++ b/tests/kms_plane_scaling.c
@@ -229,6 +229,8 @@ static void test_plane_scaling(data_t *d)
 
 		/* Set up fb2->plane2 mapping. */
 		d->plane2 = igt_output_get_plane(output, IGT_PLANE_2);
+		if(d->plane2->is_cursor)
+			continue;
 		igt_plane_set_fb(d->plane2, &d->fb2);
 
 		/* 2nd plane windowed */
@@ -265,6 +267,8 @@ static void test_plane_scaling(data_t *d)
 
 		/* Set up fb3->plane3 mapping. */
 		d->plane3 = igt_output_get_plane(output, IGT_PLANE_3);
+		if(d->plane3->is_cursor)
+			continue;
 		igt_plane_set_fb(d->plane3, &d->fb3);
 
 		/* 3rd plane windowed - no scaling */
-- 
2.7.4

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [PATCH i-g-t 0/2] Improve error report when there are less overlays than expected

[Robert Foss]

Hi tomeu,

This series looks good to me, feel free to add my r-b.


Rob.

> Hi,
> 
> kms_plane_scaling would happily try to use the cursor plane as if it
> was an
> overlay, and the first signal of it would be the kernel refusing the
> SetPlane
> call because of the pixel format not matching.
> 
> To improve this situation, we check the format of the framebuffer
> before
> calling SetPlane (printing the supported formats), and skip the test
> if there
> aren't enough overlay planes.
> 
> Thanks,
> 
> Tomeu
> 
> Tomeu Vizoso (2):
>   lib: Check format of framebuffer before SetPlane
>   kms_plane_scaling: Skip if we don't have enough overlays
> 
>  lib/igt_kms.c             | 35 +++++++++++++++++++++++++++++++++++
>  tests/kms_plane_scaling.c |  4 ++++
>  2 files changed, 39 insertions(+)
> 
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx