From: "Manszewski, Christoph" <christoph.manszewski@intel.com>
To: nishit.sharma@intel.com, igt-dev@lists.freedesktop.org,
stuart.summers@intel.com
Subject: Re: [PATCH i-g-t] DONT_MERGE:tests/intel/xe_compute: Robust SR-IOV/VF/PF and per-GT ccs_mode attribute handling
Date: Fri, 13 Feb 2026 18:45:34 +0100 [thread overview]
Message-ID: <7a733c76-76c1-4bc8-8f22-cfba6173e332@intel.com> (raw)
In-Reply-To: <20260205041640.1357478-1-nishit.sharma@intel.com>
Hi Nishit,
On 2/5/26 05:16, nishit.sharma@intel.com wrote:
> From: Nishit Sharma <nishit.sharma@intel.com>
>
> Add explicit checks for VF devices using intel_is_vf_device().
> Changed test skip logic based on PF with enabled VFs
> Remove global sriov_enabled flag and always check device state at runtime.
> For multi-GT systems, only run ccs_mode tests on GTs where the ccs_mode
> sysfs attribute is present. This ensures tests are robust on all platforms
> and configurations.
> Once the KMD changes are merged this patch will be aligned and merged
>
> Signed-off-by: Nishit Sharma <nishit.sharma@intel.com>
> ---
> tests/intel/xe_compute.c | 23 ++++++++++++++++-------
> 1 file changed, 16 insertions(+), 7 deletions(-)
>
> diff --git a/tests/intel/xe_compute.c b/tests/intel/xe_compute.c
> index 310093fc5..47d4b49dd 100644
> --- a/tests/intel/xe_compute.c
> +++ b/tests/intel/xe_compute.c
> @@ -26,8 +26,6 @@
> #define DURATION_MARGIN 0.2
> #define MIN_BUSYNESS 95.0
>
> -bool sriov_enabled;
> -
> struct thread_data {
> pthread_t thread;
> pthread_mutex_t *mutex;
> @@ -101,6 +99,10 @@ test_ccs_mode(void)
>
> num_gt_with_ccs_mode++;
We count the gt as having ccs_mode...
> gt_fd = gt_sysfs_open(gt);
> + if (!igt_sysfs_has_attr(gt_fd, "ccs_mode")) {
> + close(gt_fd);
> + continue;
> + }
and then we check if it has ccs mode? Even when we move this check, is
it even needed here? Can there be no ccs_mode entry if the gt supports
ccs mode (just asking, I don't know much about this feature).
> igt_assert(igt_sysfs_printf(gt_fd, "ccs_mode", "%u", 0) < 0);
> for (m = 1; m <= num_slices; m++) {
> /* compute slices are to be equally distributed among enabled engines */
> @@ -177,6 +179,11 @@ test_compute_kernel_with_ccs_mode(void)
>
> num_gt_with_ccs_mode++;
> gt_fd = gt_sysfs_open(gt);
> + if (!igt_sysfs_has_attr(gt_fd, "ccs_mode")) {
> + close(gt_fd);
> + continue;
> + }
> +
Same as above.
> for (m = 1; m <= num_slices; m++) {
> if (num_slices % m)
> continue;
> @@ -354,7 +361,7 @@ static bool is_sriov_mode(int fd)
> {
> bool is_sriov = false;
>
> - if (igt_sriov_is_pf(fd) && igt_sriov_vfs_supported(fd))
> + if (intel_is_vf_device(fd) || (igt_sriov_is_pf(fd) && igt_sriov_get_enabled_vfs(fd) > 0))
this function could just return the if condition.
> is_sriov = true;
>
> return is_sriov;
> @@ -422,6 +429,10 @@ test_eu_busy(uint64_t duration_sec)
> continue;
This continue
>
> gt_fd = gt_sysfs_open(gt);
> + if (!igt_sysfs_has_attr(gt_fd, "ccs_mode")) {
> + close(gt_fd);
> + continue;
and this one, can lead to the skip after the loop checking unitialized
ccs_mode variable.
> + }
> igt_assert(igt_sysfs_printf(gt_fd, "ccs_mode", "%u", num_slices) > 0);
> igt_assert(igt_sysfs_scanf(gt_fd, "ccs_mode", "%u", &ccs_mode) > 0);
> close(gt_fd);
> @@ -514,6 +525,7 @@ int igt_main()
> {
> int xe, ccs_mode[4];
> unsigned int ip_ver;
> + bool sriov_enabled;
>
> igt_fixture() {
> xe = drm_open_driver(DRIVER_XE);
> @@ -530,14 +542,12 @@ int igt_main()
>
> /* ccs mode tests should be run without open gpu file handles */
> igt_subtest("ccs-mode-basic") {
> - /* skip if sriov enabled */
> if (sriov_enabled)
> igt_skip("Skipping test when SRIOV is enabled\n");
> test_ccs_mode();
> }
>
> igt_subtest("ccs-mode-compute-kernel") {
> - /* skip if sriov enabled */
> if (sriov_enabled)
> igt_skip("Skipping test when SRIOV is enabled\n");
> test_compute_kernel_with_ccs_mode();
> @@ -554,7 +564,6 @@ int igt_main()
>
> /* test to check available EU utilisation in multi-ccs case */
> igt_subtest("eu-busy-10s") {
> - /* skip if sriov enabled */
> if (sriov_enabled)
> igt_skip("Skipping test when SRIOV is enabled\n");
>
> @@ -566,7 +575,7 @@ int igt_main()
> }
>
> igt_fixture() {
> - if (!sriov_enabled)
> + if (sriov_enabled)
Why?
> igt_restore_ccs_mode(ccs_mode, ARRAY_SIZE(ccs_mode));
> }
> }
next prev parent reply other threads:[~2026-02-13 17:45 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-02-05 4:16 [PATCH i-g-t] DONT_MERGE:tests/intel/xe_compute: Robust SR-IOV/VF/PF and per-GT ccs_mode attribute handling nishit.sharma
2026-02-05 7:30 ` ✓ i915.CI.BAT: success for DONT_MERGE:tests/intel/xe_compute: Robust SR-IOV/VF/PF and per-GT ccs_mode attribute handling (rev2) Patchwork
2026-02-05 7:42 ` ✓ Xe.CI.BAT: " Patchwork
2026-02-05 21:41 ` ✗ i915.CI.Full: failure " Patchwork
2026-02-06 1:34 ` ✗ Xe.CI.FULL: " Patchwork
2026-02-13 17:45 ` Manszewski, Christoph [this message]
-- strict thread matches above, loose matches on Subject: below --
2026-02-03 3:57 [PATCH i-g-t] DONT_MERGE:tests/intel/xe_compute: Robust SR-IOV/VF/PF and per-GT ccs_mode attribute handling nishit.sharma
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=7a733c76-76c1-4bc8-8f22-cfba6173e332@intel.com \
--to=christoph.manszewski@intel.com \
--cc=igt-dev@lists.freedesktop.org \
--cc=nishit.sharma@intel.com \
--cc=stuart.summers@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