tests/amd_dispatch: add negative test

Started by Jesse Zhang on Oct 11, 2023 · 3 messages from 2 authors · Last reply Oct 11, 2023

Subsystems (2)

Maintainers (2)

STALE1122d

open the first message on its own page

Jesse Zhang

Message details and related content
Open message details and related content
Dispatch a bad program on gfx/compute ring.
Check wether they will hang.
Expose the additional parameter 'hung' to the upper layer.
Dispatch existent a binary shader that waits for the register to be changed,
but it does not occur and as a result, we have GPU reset and check this.

V3:
  -add detail description and fix code style(Kamil)
  -Improve test description(Vitaly)

Cc: Vitaly Prosyak <redacted>
Cc: Luben Tuikov <redacted>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Christian Koenig <christian.koenig@amd.com>
Cc: Kamil Konieczny <redacted>

Signed-off-by: Jesse zhang <redacted>
Reviewed-by : Vitaly Prosyak [off-list ref]
---
 lib/amdgpu/amd_dispatch.c   |  4 ++--
 lib/amdgpu/amd_dispatch.h   |  2 +-
 tests/amdgpu/amd_dispatch.c | 38 ++++++++++++++++++++++++++++++++-----
 3 files changed, 36 insertions(+), 8 deletions(-)

diff --git a/lib/amdgpu/amd_dispatch.c b/lib/amdgpu/amd_dispatch.c
index 9de3986ba..040381a11 100644
--- a/lib/amdgpu/amd_dispatch.c
+++ b/lib/amdgpu/amd_dispatch.c
@@ -524,7 +524,7 @@ amdgpu_dispatch_hang_slow_helper(amdgpu_device_handle device_handle,
 	}
 }
 
-void amdgpu_gfx_dispatch_test(amdgpu_device_handle device_handle, uint32_t ip_type)
+void amdgpu_gfx_dispatch_test(amdgpu_device_handle device_handle, uint32_t ip_type, int hang)
 {
 	int r;
 	struct drm_amdgpu_info_hw_ip info;
@@ -547,7 +547,7 @@ void amdgpu_gfx_dispatch_test(amdgpu_device_handle device_handle, uint32_t ip_ty
 		amdgpu_memset_dispatch_test(device_handle, ip_type, ring_id,
 					    version);
 		amdgpu_memcpy_dispatch_test(device_handle, ip_type, ring_id,
-					    version, 0);
+					    version, hang);
 	}
 }
 
diff --git a/lib/amdgpu/amd_dispatch.h b/lib/amdgpu/amd_dispatch.h
index c500b5b3f..4df8b1355 100644
--- a/lib/amdgpu/amd_dispatch.h
+++ b/lib/amdgpu/amd_dispatch.h
@@ -27,7 +27,7 @@
 #include <amdgpu.h>
 
 void amdgpu_gfx_dispatch_test(amdgpu_device_handle device_handle,
-			      uint32_t ip_type);
+			      uint32_t ip_type, int hang);
 
 void amdgpu_memcpy_dispatch_test(amdgpu_device_handle device_handle,
 					uint32_t ip_type,
diff --git a/tests/amdgpu/amd_dispatch.c b/tests/amdgpu/amd_dispatch.c
index 769f26cd0..323284306 100644
--- a/tests/amdgpu/amd_dispatch.c
+++ b/tests/amdgpu/amd_dispatch.c
@@ -24,6 +24,18 @@ amdgpu_dispatch_hang_slow_compute(amdgpu_device_handle device_handle)
 	amdgpu_dispatch_hang_slow_helper(device_handle, AMDGPU_HW_IP_COMPUTE);
 }
 
+static void
+amdgpu_dispatch_hang_gfx(amdgpu_device_handle device_handle)
+{
+	amdgpu_gfx_dispatch_test(device_handle, AMDGPU_HW_IP_GFX, 1);
+}
+
+static void
+amdgpu_dispatch_hang_compute(amdgpu_device_handle device_handle)
+{
+	amdgpu_gfx_dispatch_test(device_handle, AMDGPU_HW_IP_COMPUTE, 1);
+}
+
 static void
 amdgpu_gpu_reset_test(amdgpu_device_handle device_handle, int drm_amdgpu)
 {
@@ -54,8 +66,8 @@ amdgpu_gpu_reset_test(amdgpu_device_handle device_handle, int drm_amdgpu)
 	r = amdgpu_cs_ctx_free(context_handle);
 	igt_assert_eq(r, 0);
 
-	amdgpu_gfx_dispatch_test(device_handle, AMDGPU_HW_IP_GFX);
-	amdgpu_gfx_dispatch_test(device_handle, AMDGPU_HW_IP_COMPUTE);
+	amdgpu_gfx_dispatch_test(device_handle, AMDGPU_HW_IP_GFX, 0);
+	amdgpu_gfx_dispatch_test(device_handle, AMDGPU_HW_IP_COMPUTE, 0);
 }
 
 igt_main
@@ -85,7 +97,7 @@ igt_main
 		asic_rings_readness(device, 1, arr_cap);
 
 	}
-	igt_describe("Test-GPU-reset-using-a-binary-shader-to-hang-the-job-on-compute-ring");
+	igt_describe("Test GPU reset using a binary shader to slow hang the job on compute ring");
 	igt_subtest_with_dynamic("amdgpu-dispatch-test-compute-with-IP-COMPUTE") {
 		if (arr_cap[AMD_IP_COMPUTE]) {
 			igt_dynamic_f("amdgpu-dispatch-test-compute")
@@ -93,7 +105,7 @@ igt_main
 		}
 	}
 
-	igt_describe("Test-GPU-reset-using-a-binary-shader-to-hang-the-job-on-gfx-ring");
+	igt_describe("Test GPU reset using a binary shader to slow hang the job on gfx ring");
 	igt_subtest_with_dynamic("amdgpu-dispatch-test-gfx-with-IP-GFX") {
 		if (arr_cap[AMD_IP_GFX]) {
 			igt_dynamic_f("amdgpu-dispatch-test-gfx")
@@ -101,7 +113,23 @@ igt_main
 		}
 	}
 
-	igt_describe("Test-GPU-reset-using-amdgpu-debugfs-to-hang-the-job-on-gfx-ring");
+	igt_describe("Test GPU reset using a binary shader to hang the job on gfx ring");
+	igt_subtest_with_dynamic("amdgpu-dispatch-hang-test-gfx-with-IP-GFX") {
+		if (arr_cap[AMD_IP_GFX]) {
+			igt_dynamic_f("amdgpu-dispatch-hang-test-gfx")
+			amdgpu_dispatch_hang_gfx(device);
+		}
+	}
+
+	igt_describe("Test GPU reset using a binary shader to hang the job on compute ring");
+	igt_subtest_with_dynamic("amdgpu-dispatch-hang-test-compute-with-IP-COMPUTE") {
+		if (arr_cap[AMD_IP_COMPUTE]) {
+			igt_dynamic_f("amdgpu-dispatch-hang-test-compute")
+			amdgpu_dispatch_hang_compute(device);
+		}
+	}
+
+	igt_describe("Test GPU reset using amdgpu debugfs to hang the job on gfx ring");
 	igt_subtest_with_dynamic("amdgpu-reset-test-gfx-with-IP-GFX-and-COMPUTE") {
 		if (arr_cap[AMD_IP_GFX] && arr_cap[AMD_IP_COMPUTE]) {
 			igt_dynamic_f("amdgpu-reset-gfx-compute")
-- 
2.25.1

[igt-dev] ✓ CI.xeBAT: success for tests/amd_dispatch: add negative test (rev3)

Patchwork replied to Jesse Zhang

Message details and related content
Open message details and related content
== Series Details ==

Series: tests/amd_dispatch: add negative test (rev3)
URL   : https://patchwork.freedesktop.org/series/124802/
State : success

== Summary ==

CI Bug Log - changes from XEIGT_7527_BAT -> XEIGTPW_9967_BAT
====================================================

Summary
-------

  **SUCCESS**

  No regressions found.

  

Participating hosts (4 -> 4)
------------------------------

  No changes in participating hosts

Known issues
------------

  Here are the changes found in XEIGTPW_9967_BAT that come from known issues:

### IGT changes ###

#### Possible fixes ####

  * igt@kms_flip@basic-flip-vs-wf_vblank@c-edp1:
    - bat-adlp-7:         [FAIL][1] ([Intel XE#480]) -> [PASS][2] +1 other test pass
   [1]: https://intel-gfx-ci.01.org/tree/intel-xe/IGT_7527/bat-adlp-7/igt@kms_flip@basic-flip-vs-wf_vblank@c-edp1.html
   [2]: https://intel-gfx-ci.01.org/tree/intel-xe/IGTPW_9967/bat-adlp-7/igt@kms_flip@basic-flip-vs-wf_vblank@c-edp1.html

  * {igt@xe_create@create-execqueues-noleak}:

27 more lines

[igt-dev] ✗ Fi.CI.BAT: failure for tests/amd_dispatch: add negative test (rev3)

Patchwork replied to Jesse Zhang

Message details and related content
Open message details and related content
== Series Details ==

Series: tests/amd_dispatch: add negative test (rev3)
URL   : https://patchwork.freedesktop.org/series/124802/
State : failure

== Summary ==

CI Bug Log - changes from CI_DRM_13737 -> IGTPW_9967
====================================================

Summary
-------

  **FAILURE**

  Serious unknown changes coming with IGTPW_9967 absolutely need to be
  verified manually.
  
  If you think the reported changes have nothing to do with the changes
  introduced in IGTPW_9967, please notify your bug team (lgci.bug.filing@intel.com) to allow them
  to document this new failure mode, which will reduce false positives in CI.

  External URL: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_9967/index.html

Participating hosts (37 -> 38)
------------------------------

  Additional (1): bat-dg2-8 

Possible new issues
-------------------

  Here are the unknown changes that may have been introduced in IGTPW_9967:

### IGT changes ###

#### Possible regressions ####

  * igt@kms_flip@basic-flip-vs-dpms@a-dp1:

150 more lines

Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help