From: "Juha-Pekka Heikkilä" <juhapekka.heikkila@gmail.com>
To: Sowmiya S <sowmiya.s@intel.com>, igt-dev@lists.freedesktop.org
Cc: swati2.sharma@intel.com, chaitanya.kumar.borah@intel.com
Subject: Re: [PATCH i-g-t v2 2/3] tests/intel/kms_pipe_stress: fix plane source crop and separate cursor loop
Date: Mon, 29 Jun 2026 22:28:51 +0300 [thread overview]
Message-ID: <ae1eb7d5-2563-42fd-8026-6856395ba364@gmail.com> (raw)
In-Reply-To: <20260602162422.3360253-3-sowmiya.s@intel.com>
This original test look bit messy to my eye but I think this change does
what it says and is all ok.
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
On 02/06/2026 19.24, Sowmiya S wrote:
> Pass 3/4 destination size as initial source crop to
> universal_plane_set_fb() to avoid an immediate
> downscale. Keep source in sync with igt_fb_set_size()
> in the reduction loop so the downscale ratio never
> exceeds the hardware limit (~3x).
>
> Move cursor handling out of the overlay loop into a
> dedicated second for_each_plane_on_crtc loop with a
> DRM_MODE_ATOMIC_TEST_ONLY commit. If the test commit
> rejects the cursor (e.g. unsupported format or modifier),
> disable it gracefully so the final commit with only overlay
> planes can succeed.
>
> Signed-off-by: Sowmiya S <sowmiya.s@intel.com>
> ---
> tests/intel/kms_pipe_stress.c | 110 +++++++++++++++++++++-------------
> 1 file changed, 68 insertions(+), 42 deletions(-)
>
> diff --git a/tests/intel/kms_pipe_stress.c b/tests/intel/kms_pipe_stress.c
> index 8724a674c..a84433b9d 100644
> --- a/tests/intel/kms_pipe_stress.c
> +++ b/tests/intel/kms_pipe_stress.c
> @@ -425,57 +425,83 @@ static int crtc_stress(struct data *data, igt_output_t *output,
> if (!data->num_planes[crtc->pipe] || !new_mode)
> return 0;
>
> - for_each_plane_on_crtc(crtc,
> - plane) {
> + /* Set up overlay/primary planes up to the configured limit. */
> + for_each_plane_on_crtc(crtc, plane) {
> int plane_width, plane_height;
> - if (plane->type == DRM_PLANE_TYPE_CURSOR) {
> - cursor_plane_set_fb(plane,
> - &data->cursor_fb[crtc->pipe],
> - cursor_width, cursor_height);
> - plane_width = cursor_width;
> - plane_height = cursor_height;
> - } else {
> - universal_plane_set_fb(plane,
> - &data->fb[crtc->pipe * MAX_PLANES + i],
> - mode->hdisplay, mode->vdisplay);
>
> - plane_width = (mode->hdisplay * 3) / 4;
> - plane_height = (mode->vdisplay * 3) / 4;
> + if (plane->type == DRM_PLANE_TYPE_CURSOR)
> + continue;
>
> - ret = try_plane_scaling(data, plane, plane_width, plane_height);
> + plane_width = (mode->hdisplay * 3) / 4;
> + plane_height = (mode->vdisplay * 3) / 4;
>
> - while (ret) {
> - if (plane_width <= cursor_width || plane_height <= cursor_height)
> - break;
> + /*
> + * Set the source crop to match the initial destination
> + * size (no upscaling). Keeping source == destination
> + * prevents the downscale ratio from exceeding the
> + * hardware limit (~3×) when the destination is halved
> + * in the reduction loop below.
> + */
> + universal_plane_set_fb(plane,
> + &data->fb[crtc->pipe * MAX_PLANES + i],
> + plane_width, plane_height);
>
> - plane_width /= 2;
> - plane_height /= 2;
> + ret = try_plane_scaling(data, plane, plane_width, plane_height);
>
> - ret = try_plane_scaling(data, plane, plane_width, plane_height);
> + while (ret) {
> + if (plane_width <= cursor_width || plane_height <= cursor_height)
> + break;
>
> - igt_info("Reduced plane %d size to %dx%d\n",
> - plane->index, plane_width, plane_height);
> - }
> - if (ret) {
> - igt_info("Plane %d pipe %d try commit failed, exiting\n", i,
> - crtc->pipe);
> - data->num_planes[crtc->pipe] = i;
> - igt_info("Max num planes for pipe %d set to %d\n",
> - crtc->pipe, i);
> - /*
> - * We have now determined max amount of full sized planes, we will just
> - * keep it in mind and be smarter next time. Also lets remove unneeded fbs.
> - * Don't destroy cursor_fb as will take care about it at the end.
> - */
> + plane_width /= 2;
> + plane_height /= 2;
> +
> + /* Keep source crop in sync with the reduced destination. */
> + igt_fb_set_size(&data->fb[crtc->pipe * MAX_PLANES + i],
> + plane, plane_width, plane_height);
> +
> + ret = try_plane_scaling(data, plane, plane_width, plane_height);
> +
> + igt_info("Reduced plane %d size to %dx%d\n",
> + plane->index, plane_width, plane_height);
> + }
> + if (ret) {
> + igt_info("Plane %d pipe %d try commit failed, exiting\n", i,
> + crtc->pipe);
> + data->num_planes[crtc->pipe] = i;
> + igt_info("Max num planes for pipe %d set to %d\n",
> + crtc->pipe, i);
> + /*
> + * We have now determined max amount of full sized planes, we will just
> + * keep it in mind and be smarter next time. Also lets remove unneeded fbs.
> + * Don't destroy cursor_fb as will take care about it at the end.
> + */
> + igt_plane_set_fb(plane, NULL);
> + cleanup_plane_fbs(data, crtc, i, MAX_PLANES);
> + }
> +
> + if (++i >= data->num_planes[crtc->pipe])
> + break;
> + }
> +
> + /*
> + * Always attempt to enable the cursor plane. First validate it with
> + * a TEST_ONLY commit: some format/modifier combinations (e.g.
> + * XRGB8888, 4-tiled) are not accepted by the cursor hardware.
> + * If the test commit rejects the cursor, gracefully disable it so
> + * the final actual commit with only overlay planes can succeed.
> + */
> + for_each_plane_on_crtc(crtc, plane) {
> + if (plane->type == DRM_PLANE_TYPE_CURSOR) {
> + cursor_plane_set_fb(plane,
> + &data->cursor_fb[crtc->pipe],
> + cursor_width, cursor_height);
> + if (igt_display_try_commit_atomic(&data->display,
> + DRM_MODE_ATOMIC_TEST_ONLY |
> + DRM_MODE_ATOMIC_ALLOW_MODESET, NULL)) {
> igt_plane_set_fb(plane, NULL);
> - cleanup_plane_fbs(data,
> - crtc,
> - i,
> - MAX_PLANES);
> + igt_info("Cursor fb not supported by hw (format/modifier), disabling\n");
> }
> -
> - if (++i >= data->num_planes[crtc->pipe])
> - break;
> + break;
> }
> }
>
next prev parent reply other threads:[~2026-06-29 19:29 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-02 16:24 [PATCH i-g-t v2 0/3] tests/intel/kms_pipe_stress: fix plane setup and cursor handling Sowmiya S
2026-06-02 16:24 ` [PATCH i-g-t v2 1/3] tests/intel/kms_pipe_stress: fix inverted condition skipping plane setup Sowmiya S
2026-06-02 16:24 ` [PATCH i-g-t v2 2/3] tests/intel/kms_pipe_stress: fix plane source crop and separate cursor loop Sowmiya S
2026-06-29 19:28 ` Juha-Pekka Heikkilä [this message]
2026-06-02 16:24 ` [PATCH i-g-t v2 3/3] tests/intel/kms_pipe_stress: use ARGB8888/LINEAR for cursor framebuffer Sowmiya S
2026-06-04 7:22 ` Borah, Chaitanya Kumar
2026-06-03 6:10 ` ✓ Xe.CI.BAT: success for tests/intel/kms_pipe_stress: fix plane setup and cursor handling (rev2) Patchwork
2026-06-03 6:23 ` ✓ i915.CI.BAT: " Patchwork
2026-06-03 17:46 ` ✗ Xe.CI.FULL: failure " Patchwork
2026-06-04 2:29 ` ✗ i915.CI.Full: " Patchwork
2026-06-09 15:51 ` ✓ i915.CI.Full: success " Patchwork
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=ae1eb7d5-2563-42fd-8026-6856395ba364@gmail.com \
--to=juhapekka.heikkila@gmail.com \
--cc=chaitanya.kumar.borah@intel.com \
--cc=igt-dev@lists.freedesktop.org \
--cc=sowmiya.s@intel.com \
--cc=swati2.sharma@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