[i-g-t] tests/intel-ci/xe: Block kms_rmfb at close-fd test

Lucas De Marchi lucas.demarchi at intel.com
Thu Sep 19 19:30:29 UTC 2024


On Wed, Jul 31, 2024 at 11:45:12AM GMT, Bhanuprakash Modem wrote:
>Upon reopening the driver, the 'kms_rmfb at close-fd' test expects
>that 'getcrtc' should return an FB ID of 0. However, due to the
>default modeset, 'getcrtc' always returns a non-zero FB ID, which
>is expected behavior.

I'm trying to understand the issue here. Do you know when this default
modeset is triggered?

>
>In the case of the i915 driver, the 'i915_gem_drop_caches' clears
>the cache and sets the FB ID to 0. Currently, there is no equivalent

not sure if this is really related to gem_drop_caches. To me it's more
related to something preventing the modeset

>mechanism in the XE driver.

doing some tests, it seems we have:

commit
close-fd
   drop caches
check-fb-id
modeset kicks in

The check for fb_id then works because it happens before the modeset.
In fact, I can reproduce the same issue with xe and the other variant
of the test (igt at kms_rmfb@rmfb-ioctl) by simply adding a sleep(1)
before the last drmModeFreeCrtc(). I think we should either check that
fb_id != prev_fb_id or provide the same guarantee in xe that it will
remain 0 until the process exit.

AFAICS the latter can be guaranteed by doing this in the fixture:

drm_reopen_driver(data.drm_fd);

Let me send a patch to double check in CI...

Lucas De Marchi

>
>Therefore, we should block this test from XE execution until an
>alternative solution is implemented.
>
>Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/294
>Signed-off-by: Bhanuprakash Modem <bhanuprakash.modem at intel.com>
>---
> tests/intel-ci/xe.blocklist.txt | 4 ++++
> 1 file changed, 4 insertions(+)
>
>diff --git a/tests/intel-ci/xe.blocklist.txt b/tests/intel-ci/xe.blocklist.txt
>index f9ef612c4..4ee8fb0d4 100644
>--- a/tests/intel-ci/xe.blocklist.txt
>+++ b/tests/intel-ci/xe.blocklist.txt
>@@ -78,6 +78,10 @@ igt at kms_frontbuffer_tracking@(fbcpsrdrrs|psrdrrs).*
> # KMS: Yet to add/missing XE support
> ##################################################################
> igt at kms_prime@.*
>+
>+# Expected failure on XE:
>+# https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/294
>+igt at kms_rmfb@close-fd
> ##################################################################
> # KMS: Non-Intel tests
> ##################################################################
>-- 
>2.43.2
>


More information about the igt-dev mailing list