Bug 205739 – webgl/1.0.3/conformance/glsl/misc/shader-with-reserved-words.html is timing out on some hardware configurations

• Component: WebGL
• Status: RESOLVED FIXED

[Truitt Savell]

webgl/1.0.3/conformance/glsl/misc/shader-with-reserved-words.html

Description:
This test began failing after r253926. this may just be a rebase

History:
https://results.webkit.org/?suite=layout-tests&test=webgl%2F1.0.3%2Fconformance%2Fglsl%2Fmisc%2Fshader-with-reserved-words.html

Diff:
--- /Volumes/Data/slave/catalina-debug-tests-wk2-webgl/build/layout-test-results/webgl/1.0.3/conformance/glsl/misc/shader-with-reserved-words-expected.txt
+++ /Volumes/Data/slave/catalina-debug-tests-wk2-webgl/build/layout-test-results/webgl/1.0.3/conformance/glsl/misc/shader-with-reserved-words-actual.txt
@@ -1,5 +1,5 @@
-#PID UNRESPONSIVE - WebKitTestRunner (pid 96218)
-FAIL: Timed out waiting for notifyDone to be called
+This test runs the WebGL Test listed below in an iframe and reports PASS or FAIL.
 
-#EOF
-#EOF
+Test: ../../../resources/webgl_test_files/conformance/glsl/misc/shader-with-reserved-words.html
+[ PASS ] All tests passed
+

[Radar WebKit Bug Importer]

<rdar://problem/58302584>

[Alexey Proskuryakov]

This is a tricky one, but there is definitely something wrong going on here, both with and without ANGLE.

This test is skipped in most configurations, and marked as a flaky timeout in the few configurations where it runs. But it also appears to be skipped on internal bots (although I don't understand how), so we don't have results from different hardware configurations.

LayoutTests/TestExpectations:3823:webgl/1.0.3/conformance/glsl/misc/shader-with-reserved-words.html [ Skip ]
LayoutTests/platform/mac-wk1/TestExpectations:583:webkit.org/b/170877 [ Debug ] webgl/1.0.3/conformance/glsl/misc/shader-with-reserved-words.html [ Pass Timeout ]
LayoutTests/platform/mac/TestExpectations:1728:webkit.org/b/170877 [ Debug ] webgl/1.0.3/conformance/glsl/misc/shader-with-reserved-words.html [ Pass Timeout ]

Anyway, this test times out on Mac minis with or without ANGLE. It passes on iMac16,2, also with or without. It presumably timed out in a different way on Dean's iMac Pro (?), making him land this "#PID UNRESPONSIVE" expectation.

[Alexey Proskuryakov]

Reusing this but to track all issues with this test.

[James Darpinian]

I've identified a regression in shader compilation performance with ANGLE enabled, so I will fix that. After fixing the regression the performance should be the same with and without ANGLE, so if it was timing out before then it will probably still timeout.

[James Darpinian]

Created attachment 389921 [details]
Optimize disabling ANGLE_texture_rectangle

[WebKit Commit Bot]

Comment on attachment 389921 [details]
Optimize disabling ANGLE_texture_rectangle

Clearing flags on attachment: 389921

Committed r255990: <https://trac.webkit.org/changeset/255990>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.