Win10 FYI Release (NVIDIA Vulkan) flakes due to device lost (multiple tests) |
||||||
Issue descriptionSample failure: https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Win10%20FYI%20Release%20%28NVIDIA%29/3137 gpu_tests.webgl_conformance_integration_test.WebGLConformanceIntegrationTest.WebglConformance_conformance_uniforms_uniform_default_values failed unexpectedly 1.1750s: Traceback (most recent call last): _RunGpuTest at content\test\gpu\gpu_tests\gpu_integration_test.py:155 self.RunActualGpuTest(url, *args) RunActualGpuTest at content\test\gpu\gpu_tests\webgl_conformance_integration_test.py:190 getattr(self, test_name)(test_path, *args[1:]) _RunConformanceTest at content\test\gpu\gpu_tests\webgl_conformance_integration_test.py:210 self._CheckTestCompletion() _CheckTestCompletion at content\test\gpu\gpu_tests\webgl_conformance_integration_test.py:206 self.fail(self._WebGLTestMessages(self.tab)) fail at .swarming_module\bin\Lib\unittest\case.py:410 raise self.failureException(msg) AssertionError: Unable to fetch WebGL rendering context for Canvas FAIL Unable to fetch WebGL rendering context for Canvas Uncaught TypeError: Cannot read property 'createBuffer' of null Locals: msg : u"Unable to fetch WebGL rendering context for Canvas\nFAIL Unable to fetch WebGL rendering context for Canvas\nUncaught TypeError: Cannot read property 'createBuffer' of null\n"
,
Nov 21
Looks like the context is being lost: [418/467] gpu_tests.webgl_conformance_integration_test.WebGLConformanceIntegrationTest.WebglConformance_conformance_uniforms_no_over_optimization_on_uniform_array_17 passed 0.9410s [2696:6364:1121/065819.200:INFO:CONSOLE(1905)] "Synchronous XMLHttpRequest on the main thread is deprecated because of its detrimental effects to the end user's experience. For more help, check https://xhr.spec.whatwg.org/.", source: http://127.0.0.1:63969/third_party/webgl/src/sdk/tests/js/webgl-test-utils.js (1905) [419/467] gpu_tests.webgl_conformance_integration_test.WebGLConformanceIntegrationTest.WebglConformance_conformance_uniforms_null_uniform_location passed 0.0760s [7440:5616:1121/065820.857:WARNING:angle_platform_impl.cc(52)] handleError(1238): Internal Vulkan error: The logical or physical device has been lost., in ../../third_party/angle/src/libANGLE/renderer/vulkan/RendererVk.cpp, line 930. [7440:5616:1121/065820.857:ERROR:angle_platform_impl.cc(47)] DebugReportCallback(125): [ VUID-vkDestroyFence-fence-01120 ] Object: 0x8601 (Type = 7) | Fence 0x8601 is in use. The Vulkan spec states: All queue submission commands that refer to fence must have completed execution (https://www.khronos.org/registry/vulkan/specs/1.1-extensions/html/vkspec.html#VUID-vkDestroyFence-fence-01120) [7440:5616:1121/065820.857:ERROR:angle_platform_impl.cc(47)] DebugReportCallback(125): [ VUID-vkDestroyCommandPool-commandPool-00041 ] Object: 0xa70c280 (Type = 6) | Attempt to destroy command pool with command buffer (0xa70c280) which is in use. The Vulkan spec states: All VkCommandBuffer objects allocated from commandPool must not be in the pending state. (https://www.khronos.org/registry/vulkan/specs/1.1-extensions/html/vkspec.html#VUID-vkDestroyCommandPool-commandPool-00041) The real error is somewhat farther up in the stack. samans, when you report the bugs it's also helpful to report what graphics driver these were running on. In this case, Vulkan. Though it's unclear in this case what the root cause of the device lost was.
,
Nov 21
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/53faf8eea7fea887bb9a5496c973a7545935e82a commit 53faf8eea7fea887bb9a5496c973a7545935e82a Author: Saman Sami <samans@chromium.org> Date: Wed Nov 21 19:09:55 2018 Suppress flaky WebGL test WebglConformance_conformance_uniforms_uniform_default_values is flaking on Win10 FYI Release (NVIDIA). TBR=kbr@chromium.org Bug: 907544 Change-Id: Ib41ff367449b0fd3c424d95fdf39f4fdcf7e4633 Reviewed-on: https://chromium-review.googlesource.com/c/1346673 Reviewed-by: Robert Kroeger <rjkroege@chromium.org> Commit-Queue: Saman Sami <samans@chromium.org> Cr-Commit-Position: refs/heads/master@{#610119} [modify] https://crrev.com/53faf8eea7fea887bb9a5496c973a7545935e82a/content/test/gpu/gpu_tests/webgl2_conformance_expectations.py
,
Nov 22
This seems to only affect the passthrough tests so cc'ing geofflang@.
,
Nov 28
Might warrant filing an ANGLE bug for this one so it's tracked under the Vulkan work. Jamie, can you take this for tracking purposes?
,
Nov 28
https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Win10%20FYI%20Release%20%28NVIDIA%29/3192 Another device lost flake. I don't think suppressing the test is necessary. Seems like a low-probability flake that can occur during some (or all) tests. If we're lucky it could be something that's fixed with the driver upgrade.
,
Nov 28
Actually, we could mark the test as skip. And for what it's worth, I see at least 3 flakes on https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Win10%20FYI%20Release%20%28NVIDIA%29?limit=200 but zero flakes on https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Win10%20FYI%20Exp%20Release%20%28NVIDIA%29?limit=200 So that's promising. Tentatively blocking on driver upgrade. |
||||||
►
Sign in to add a comment |
||||||
Comment 1 by samans@chromium.org
, Nov 21