[Intel-gfx] [PATCH i-g-t 3/3] perf_pmu: Refine requirement testing for engine-busy-stats

Chris Wilson chris at chris-wilson.co.uk
Thu Jul 4 16:15:50 UTC 2019


Now that we report whether the accurate per-engine utilisation
statistics are available (albeit via the scheduler caps) put it to to
use to selectively enable the high accuracy tests where we expect it to
work.

Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
---
 tests/perf_pmu.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tests/perf_pmu.c b/tests/perf_pmu.c
index 72b9166af..c3573b7a1 100644
--- a/tests/perf_pmu.c
+++ b/tests/perf_pmu.c
@@ -1468,7 +1468,7 @@ test_enable_race(int gem_fd, const struct intel_execution_engine2 *e)
 	struct drm_i915_gem_execbuffer2 eb = { };
 	int fd;
 
-	igt_require(gem_has_execlists(gem_fd));
+	igt_require(gem_scheduler_has_engine_busy_stats(gem_fd));
 	igt_require(gem_context_has_engine(gem_fd, 0, e->flags));
 
 	obj.handle = gem_create(gem_fd, 4096);
@@ -1538,7 +1538,7 @@ accuracy(int gem_fd, const struct intel_execution_engine2 *e,
 	int fd;
 
 	/* Sampling platforms cannot reach the high accuracy criteria. */
-	igt_require(gem_has_execlists(gem_fd));
+	igt_require(gem_scheduler_has_engine_busy_stats(gem_fd));
 
 	/* Aim for approximately 100 iterations for calibration */
 	cycle_us = min_test_us / target_iters;
-- 
2.20.1



More information about the Intel-gfx mailing list