[Intel-gfx] [PATCH i-g-t v2 1/5] lib/igt_kms: Fix drm_plane leak

[Brian Starkey]

Hi,

tsa at freenode was kind enough to run some piglit testing on this.

I gave the following test list:

igt at kms_cursor_legacy@basic-busy-flip-before-cursor-atomic
igt at kms_cursor_legacy@basic-busy-flip-before-cursor-legacy
igt at kms_plane_multiple@legacy-pipe-A-tiling-none
igt at kms_plane_multiple@atomic-pipe-A-tiling-none
igt at kms_plane_multiple@legacy-pipe-A-tiling-x
igt at kms_plane_multiple@atomic-pipe-A-tiling-x
igt at kms_chv_cursor_fail@pipe-A-128x128-bottom-edge
igt at kms_chv_cursor_fail@pipe-A-128x128-left-edge
igt at kms_chv_cursor_fail@pipe-A-128x128-right-edge
igt at kms_chv_cursor_fail@pipe-A-128x128-top-edge
igt at kms_ccs@bad-rotation-90
igt at kms_busy@flip-blt-A
igt at kms_mmap_write_crc

There were no changes to any test on 11 hosts (ILK to KBL).
(3 fails on bad-rotation-90 stays just as they were)

Thanks,
Brian

On Mon, Feb 20, 2017 at 11:02:44AM +0000, Brian Starkey wrote:
>In the loop looking for planes on a pipe, we always want to free up
>the drm_plane afterwards.
>
>Fixes: 36656239ef96 lib/igt_kms: Implement dynamic plane count support
>Signed-off-by: Brian Starkey <brian.starkey at arm.com>
>Reviewed-by: Robert Foss <robert.foss at collabora.com>
>---
> lib/igt_kms.c |    8 +++-----
> 1 file changed, 3 insertions(+), 5 deletions(-)
>
>diff --git a/lib/igt_kms.c b/lib/igt_kms.c
>index d6b7d677ed8d..9e59e35f2e2f 100644
>--- a/lib/igt_kms.c
>+++ b/lib/igt_kms.c
>@@ -1613,12 +1613,10 @@ void igt_display_init(igt_display_t *display, int drm_fd)
> 						    plane_resources->planes[j]);
> 			igt_assert(drm_plane);
>
>-			if (!(drm_plane->possible_crtcs & (1 << i))) {
>-				drmModeFreePlane(drm_plane);
>-				continue;
>-			}
>+			if (drm_plane->possible_crtcs & (1 << i))
>+				n_planes++;
>
>-			n_planes++;
>+			drmModeFreePlane(drm_plane);
> 		}
>
> 		igt_assert_lte(0, n_planes);
>-- 
>1.7.9.5
>