[PATCH i-g-t] tests/intel/kms_pm_dc: Add a new test to validate the deep sleep state during extended vblank
Kandpal, Suraj
suraj.kandpal at intel.com
Mon May 27 05:03:25 UTC 2024
> Subject: [PATCH i-g-t] tests/intel/kms_pm_dc: Add a new test to validate the
> deep sleep state during extended vblank
>
> Add a new test to validate deep sleep states during extended vblank scenarios,
> where two frames are committed simultaneously for a give time with reduced
Typo here *given
> refresh rate.
> v2: dealy of one frame added to simulate extended vblank.
> remove vrr related debug checks.
>
> Signed-off-by: Jeevan B <jeevan.b at intel.com>
> ---
> tests/intel/kms_pm_dc.c | 49
> +++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 49 insertions(+)
>
> diff --git a/tests/intel/kms_pm_dc.c b/tests/intel/kms_pm_dc.c index
> 7766d34d7..9f89e537d 100644
> --- a/tests/intel/kms_pm_dc.c
> +++ b/tests/intel/kms_pm_dc.c
> @@ -584,6 +584,46 @@ static unsigned int read_pkgc_counter(int
> debugfs_root_fd)
> return get_dc_counter(str);
> }
>
> +static void test_deep_pkgc_state(data_t *data) {
> + unsigned int pre_val = 0, cur_val = 0;
> + time_t start = time(NULL), duration = 2, delay;
> + enum pipe pipe;
> + bool pkgc_flag;
> + igt_display_t *display = &data->display;
> + igt_plane_t *primary;
> + igt_output_t *output;
> +
> + for_each_pipe_with_valid_output(display, pipe, output) {
> + igt_output_set_pipe(output, pipe);
> + data->output = output;
> + data->mode = igt_output_get_mode(output);
> + setup_videoplayback(data);
> + igt_require(igt_output_has_prop(output,
> IGT_CONNECTOR_VRR_CAPABLE));
> + igt_require(igt_output_get_prop(output,
> IGT_CONNECTOR_VRR_CAPABLE));
Why are there two igt_require checking the same thing here?
> +
> + primary = igt_output_get_plane_type(data->output,
> +
> DRM_PLANE_TYPE_PRIMARY);
> + pre_val = read_pkgc_counter(data->debugfs_root_fd);
> + delay = 1 * ((1000 * 1000) / data->mode->vrefresh);
If 1000*1000 is constant why not use a #define would look cleaner
> + while (time(NULL) - start < duration) {
> + igt_plane_set_fb(primary, &data->fb_rgb);
> + igt_display_commit(&data->display);
> + usleep(delay);
> +
> + igt_plane_set_fb(primary, &data->fb_rgr);
> + igt_display_commit(&data->display);
> + cur_val = read_pkgc_counter(data->debugfs_root_fd);
> + if (cur_val > pre_val) {
> + pkgc_flag = true;
> + continue;
> + }
> + }
> + }
> + cleanup_dc3co_fbs(data);
> + igt_assert_f(pkgc_flag, "PKGC10 is not achieved.\n"); }
> +
> static void test_pkgc_state_dpms(data_t *data) {
> unsigned int timeout_sec = 6;
> @@ -687,6 +727,15 @@ igt_main
> test_dc_state_psr(&data, CHECK_DC6);
> }
>
> + igt_describe("This test validates display engine entry to DC8 state "
> + "while extended vblank");
*during extended vblank
Regards,
Suraj Kandpal
> + igt_subtest("deep-pkgc") {
> +
> igt_require_f(igt_pm_pc8_plus_residencies_enabled(data.msr_fd),
> + "PC8+ residencies not supported\n");
> + igt_require(intel_display_ver(data.devid) >= 20);
> + test_deep_pkgc_state(&data);
> + }
> +
> igt_describe("This test validates display engine entry to DC5 state "
> "while all connectors's DPMS property set to OFF");
> igt_subtest("dc5-dpms") {
> --
> 2.25.1
More information about the igt-dev
mailing list