[PATCH i-g-t v2] tests/amdgpu: add timeout for queue reset

Jesse.zhang@amd.com jesse.zhang at amd.com
Fri Aug 16 03:23:00 UTC 2024


1. If the test case cannot trigger any reset on some ASIC,
It should be considered a failure.

2. Fix code style

V2: Fix some comments (Vitaly)
    Replace timeout failures with warnings.

Cc: Kamil Konieczny <kamil.konieczny at linux.intel.com>
Cc: Alex Deucher <alexander.deucher at amd.com>
Cc: Christian Koenig <christian.koenig at amd.com>
Cc: Vitaly Prosyak <vitaly.prosyak at amd.com>

Signed-off-by: Jesse Zhang <jesse.zhang at amd.com>
Reviewed-by: Vitaly Prosyak vitaly.prosyak at amd.com
---
 tests/amdgpu/amd_queue_reset.c | 22 ++++++++++++++++++++--
 1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/tests/amdgpu/amd_queue_reset.c b/tests/amdgpu/amd_queue_reset.c
index 6819892e0..918f90d6b 100644
--- a/tests/amdgpu/amd_queue_reset.c
+++ b/tests/amdgpu/amd_queue_reset.c
@@ -30,6 +30,7 @@
 #define SHARED_CHILD_DESCRIPTOR 3
 
 #define SHARED_MEM_NAME  "/queue_reset_shm"
+#define TEST_TIMEOUT 100 //100 seconds
 
 enum  process_type {
 	PROCESS_UNKNOWN,
@@ -49,12 +50,13 @@ enum error_code_bits {
 };
 
 enum reset_code_bits {
+	NO_RESET_SET_BIT,
 	QUEUE_RESET_SET_BIT,
 	GPU_RESET_BEGIN_SET_BIT,
 	GPU_RESET_END_SUCCESS_SET_BIT,
 	GPU_RESET_END_FAILURE_SET_BIT,
 
-	ALL_RESET_BITS = 0xf,
+	ALL_RESET_BITS = 0x1f,
 };
 
 struct shmbuf {
@@ -307,6 +309,8 @@ static void set_next_test_to_run(struct shmbuf *sh_mem, unsigned int error,
 	sync_point_enter(sh_mem);
 	wait_for_complete_iteration(sh_mem);
 	sync_point_exit(sh_mem);
+	igt_warn_on_f(sh_mem->reset_flags == 1U << NO_RESET_SET_BIT,
+		"Testing does not trigger reset \n");
 }
 
 static int
@@ -473,6 +477,9 @@ run_monitor_child(amdgpu_device_handle device, amdgpu_context_handle *arr_contex
 	int state_machine = 0;
 	int error_code;
 	unsigned int flags;
+	int64_t cnt = 0;
+	time_t start, end;
+	double elapsed = 0;
 
 	after_reset_state = after_reset_hangs = 0;
 	init_flags = in_process_flags = 0;
@@ -487,6 +494,7 @@ run_monitor_child(amdgpu_device_handle device, amdgpu_context_handle *arr_contex
 		error_code = 0;
 		flags = 0;
 		set_reset_state(sh_mem, false, ALL_RESET_BITS);
+		time(&start);
 		while (1) {
 			if (state_machine == 0) {
 				amdgpu_cs_query_reset_state2(arr_context[test_counter], &init_flags);
@@ -533,7 +541,17 @@ run_monitor_child(amdgpu_device_handle device, amdgpu_context_handle *arr_contex
 					break;
 				}
 			}
+			cnt++;
+			if (cnt % 1000000 == 0) {
+				time(&end);
+				elapsed = difftime(end, start);
+				if ( elapsed >= TEST_TIMEOUT) {
+					set_reset_state(sh_mem, true, NO_RESET_SET_BIT);
+					break;
+				}
+			}
 		}
+		elapsed = 0;
 		sync_point_exit(sh_mem);
 		num_of_tests--;
 		test_counter++;
@@ -1000,7 +1018,7 @@ igt_main
 			igt_describe("Stressful-and-multiple-cs-of-bad and good length-operations-using-multiple-processes");
 			igt_subtest_with_dynamic_f("amdgpu-%s-%s", ip_tests[i] == AMD_IP_COMPUTE ? "COMPUTE":"GRAFIX", it->name) {
 				if (arr_cap[ip_tests[i]] && get_next_rings(ring_id_good, info, &ring_id_good, &ring_id_bad, i)) {
-					igt_dynamic_f("amdgpu-%s-ring-good-%d-bad-%d-%s", it->name,ring_id_good, ring_id_bad, ip_tests[i] == AMD_IP_COMPUTE ? "COMPUTE":"GRAFIX")
+					igt_dynamic_f("amdgpu-%s-ring-good-%d-bad-%d-%s", it->name, ring_id_good, ring_id_bad, ip_tests[i] == AMD_IP_COMPUTE ? "COMPUTE":"GRAFIX")
 					set_next_test_to_run(sh_mem, it->test, ip_background, ip_tests[i], ring_id_good, ring_id_bad);
 				}
 			}
-- 
2.25.1



More information about the igt-dev mailing list