WebglConformance_conformance_glsl_misc_shader_with_non_reserved_words times out on Win7 Debug (New Intel) GPU.FYI bot |
||
Issue descriptionStarted failing here https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1742 The only suspicious change I could find in the blamelist is removing an optimization in v8, which I'm not even sure is related to this test. https://codereview.chromium.org/2544503002 There were also some WebKit changes, maybe they have slowed thing down somehow?
,
Dec 1 2016
Filed issue 670440 for the buildbot issue. Looking into this.
,
Dec 1 2016
I can't come up with anything substantial that might be causing these timeouts. Am trying a speculative revert in https://codereview.chromium.org/2548583003/
,
Dec 2 2016
I see a time increase from 42m17s to 44m10s that's a 5% regression on https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1740/steps/webgl_conformance_tests vs https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1742/steps/webgl_conformance_tests . Do we have any webgl benchmarks regressing in the same time?
,
Dec 5 2016
This seems to have gone green in https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1765. I don't see anything obvious in the build where it went green. There's a skia roll (nothing looks related), plus a test suppression for a different test by kbr@ in the range (seems unrelated, but maybe this one was collateral damage?) Feel free to reopen if it reoccurs.
,
Dec 5 2016
Looks like I've been wrong about https://codereview.chromium.org/2544503002 causing problems. https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1758 is the first one to have it reverted, but it's broken because of issue 670984 . But, https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1763 and https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1764 (which also have the revert) fail in the same way as https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1742 did. Comparing 1742 blamelist with 1765, I see there were some skia changes touching the same area: https://skia-review.googlesource.com/c/5353/, https://skia-review.googlesource.com/c/5544/. I don't know skia, so I can't say if this code runs during this test. I will take care of reverting the v8 revert.
,
Dec 6 2016
To double confirm that v8 CL was unrelated: https://build.chromium.org/p/chromium.gpu.fyi/builders/Win7%20Debug%20%28New%20Intel%29/builds/1779 has v8 CL relanded, and doesn't fail WebglConformance_conformance_glsl_misc_shader_with_non_reserved_words. |
||
►
Sign in to add a comment |
||
Comment 1 by jmad...@chromium.org
, Dec 1 2016Labels: GPU-Intel