[PATCH i-g-t 5/6] tests/intel/xe_oa: Reduce test_non_zero_reason execution time

[Umesh Nerlige Ramappa]

On Tue, Apr 08, 2025 at 11:12:09AM -0700, Ashutosh Dixit wrote:
>Only check every 128th report to reduce test execution time. Without this

Since you have already read the entire buffer into memory, you could 
just check 2 reports at random location. I am thinking random would be 
better than fixed intervals to catch strange issues.

Regards,
Umesh

>patch execution times for buffer-size subtest for OA buffer sizes > 16
>M (with PEC report sanity checks) are excessive. E.g. for 128 MB size:
>
>	Starting subtest: buffer-size
>	Starting dynamic subtest: ccs-0-128M
>	Dynamic subtest ccs-0-128M: SUCCESS (133.654s)
>	Subtest buffer-size: SUCCESS (133.654s)
>
>With this patch:
>
>	Starting subtest: buffer-size
>	Starting dynamic subtest: ccs-0-128M
>	Dynamic subtest ccs-0-128M: SUCCESS (11.209s)
>	Subtest buffer-size: SUCCESS (11.209s)
>
>Signed-off-by: Ashutosh Dixit <ashutosh.dixit at intel.com>
>---
> tests/intel/xe_oa.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
>diff --git a/tests/intel/xe_oa.c b/tests/intel/xe_oa.c
>index cc9702322d..f3ce37ff5a 100644
>--- a/tests/intel/xe_oa.c
>+++ b/tests/intel/xe_oa.c
>@@ -2646,13 +2646,14 @@ test_non_zero_reason(const struct drm_xe_engine_class_instance *hwe, size_t oa_b
> 	igt_debug("Got %u bytes\n", total_len);
>
> 	last_report = NULL;
>-	for (uint32_t offset = 0; offset < total_len; offset += report_size) {
>+	for (uint32_t offset = 0, n = 0; offset < total_len; offset += report_size) {
> 		const uint32_t *report = (void *) (buf + offset);
> 		uint32_t reason = (report[0] >> OAREPORT_REASON_SHIFT) & OAREPORT_REASON_MASK;
>
> 		igt_assert_neq(reason, 0);
>
>-		if (last_report) {
>+		/* Only check every 128th report to reduce test execution time */
>+		if (last_report && !(n++ % 128)) {
> 			sanity_check_reports(last_report, report, fmt);
> 			pec_sanity_check_reports(last_report, report, metric_set(hwe));
> 		}
>-- 
>2.48.1
>