Re: [PATCH 06/25] drm/tests: hdmi: Convert to drmm_connector_hdmi_init()

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found no new issues
in this patch, but noticed 1 related pre-existing issue it does not
introduce:
- [Medium] drm/tests: ww_mutex leak in
  drm_test_check_reject_audio_infoframe() [1]
--

commit d078fa85b984429f42b17f7efe1e93d8055ccfb2
Author: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>

drm/tests: hdmi: Convert to drmm_connector_hdmi_init()

This commit converts the HDMI state helper KUnit tests to the new
drmm_connector_hdmi_init() signature and stores each connector's
drm_connector_hdmi_funcs copy in drm_atomic_helper_connector_hdmi_priv so
it outlives the connector.

> diff --git a/drivers/gpu/drm/tests/drm_hdmi_state_helper_test.c b/drivers/gpu/drm/tests/drm_hdmi_state_helper_test.c
> index 336bb644cb7c9..f1bdd197429c6 100644
> --- a/drivers/gpu/drm/tests/drm_hdmi_state_helper_test.c
> +++ b/drivers/gpu/drm/tests/drm_hdmi_state_helper_test.c

[ ... ]

> @@ -3314,8 +3310,6 @@ static void drm_test_check_reject_audio_infoframe(struct kunit *test)
>  	int ret;
>  
>  	priv = drm_kunit_helper_connector_hdmi_init_with_edid_funcs(test,
> -				BIT(DRM_OUTPUT_COLOR_FORMAT_RGB444),
> -				8,
>  				&reject_audio_infoframe_hdmi_funcs,
>  				test_edid_hdmi_1080p_rgb_max_200mhz);
>  	KUNIT_ASSERT_NOT_NULL(test, priv);

[Severity: Medium]
This problem wasn't introduced by this patch, but can a failed assertion in
drm_test_check_reject_audio_infoframe() leak the modeset ww_mutex locks in
ctx [1]?

drivers/gpu/drm/tests/drm_hdmi_state_helper_test.c:drm_test_check_reject_audio_infoframe() {
    ...
	drm_modeset_acquire_init(&ctx, 0);

retry_conn_enable:
	ret = drm_kunit_helper_enable_crtc_connector(test, drm,
						     crtc, conn,
						     preferred,
						     &ctx);
    ...
	KUNIT_ASSERT_EQ(test, ret, 0);
    ...
	drm_modeset_drop_locks(&ctx);
	drm_modeset_acquire_fini(&ctx);
}

KUNIT_ASSERT_* returns immediately on failure, bypassing
drm_modeset_drop_locks() and drm_modeset_acquire_fini() at the end of the
function.  Because ctx lives on the stack, returning with locks held leaves
dangling lock list nodes pointing to out-of-scope stack memory, triggering
lockdep warnings and potential crashes during subsequent tests or DRM
device cleanup.

[1] https://sashiko.dev/#/bug/linux-1258d4b2-95ea-43e2-a892-e2011a232ee8

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261114-drm-conn-hdmi-cleanup-v1-0-20e8a33a2286@collabora.com?part=6