Bug 223624
| Summary: | REGRESSION(r r274711) [GLIB] webgl/conformance/extensions/khr-parallel-shader-compile.html is timing out | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Lauro Moura <lmoura> |
| Component: | WebGL | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED DUPLICATE | ||
| Severity: | Normal | CC: | bugs-noreply, dino, jdarpinian, kbr, kkinnunen, kpiddington, webkit-bug-importer |
| Priority: | P2 | Keywords: | InRadar |
| Version: | WebKit Nightly Build | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
Lauro Moura
webgl/conformance/extensions/khr-parallel-shader-compile.html
Updated in r274711, timing out in GLIB since, even with [ Slow ].
Checking the first loop of the test (~120), it seems the compile operations are not being offloaded to another thread, as there is a big wait in the compileShader calls. In the 7th iteration of the loop:
extraCode size 2949120
shaderSource(vertexSource) took 18ms
shaderSource(fragmentSource) took 17ms
compileShader(vs) took 1460ms
compileShader(fs) took 1392ms
linkProgram(program) took 1ms
This is causing the measuredCompileDuration to always be smaller than the threshold (500ms), as it's actually just a get operation, and the code never triggers the loop exit condition.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Kenneth Russell
+jdarpinian who recently strengthened this test upstream in https://github.com/KhronosGroup/WebGL .
Should probably skip this test in TestExpectations on GLIB while the implementation's being fixed. Note it will likely take some time to roll an updated ANGLE back downstream.
Radar WebKit Bug Importer
<rdar://problem/75989686>
Kimmo Kinnunen
*** This bug has been marked as a duplicate of bug 251514 ***