inspector/cpu-profiler/threads.html has become a flaky failure on macOS WK1 bots --- /Volumes/Data/slave/mojave-release-tests-wk1/build/layout-test-results/inspector/cpu-profiler/threads-expected.txt +++ /Volumes/Data/slave/mojave-release-tests-wk1/build/layout-test-results/inspector/cpu-profiler/threads-actual.txt @@ -13,10 +13,16 @@ PASS: Event should have threads. PASS: Every thread has between 0 and 100 usage. PASS: Event should have 1 main thread. -PASS: Event should have 2 worker threads. +FAIL: Event should have 2 worker threads. + Expected: 2 + Actual: 7 PASS: Event should have other worker threads. -PASS: Worker 1 usage should be greater than zero. -PASS: Worker 2 usage should be greater than zero. +FAIL: Worker 1 usage should be greater than zero. + Expected: greater than 0 + Actual: 0 +FAIL: Worker 2 usage should be greater than zero. + Expected: greater than 0 + Actual: 0 PASS: Total usage should be greater than or equal to the sum of both worker threads. CPUProfiler.trackingComplete https://results.webkit.org/?suite=layout-tests&test=inspector%2Fcpu-profiler%2Fthreads.html
It looks like this started after https://trac.webkit.org/changeset/268897/webkit landed
<rdar://problem/70732755>
I have not yet been able to reproduce this locally by running this test alone in iterations.
Created attachment 412936 [details] Patch
Comment on attachment 412936 [details] Patch r=me I kinda wish there were a thread.type for worker (and perhaps a distinct thread.type for ServiceWorker).
Committed r269257: <https://trac.webkit.org/changeset/269257> All reviewed patches have been landed. Closing bug and clearing flags on attachment 412936 [details].
Comment on attachment 412936 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=412936&action=review > LayoutTests/ChangeLog:10 > + Check the thread name to identify the 2 worker threads. This is needed so that worklet thread are > + not identified as worker threads after r268897. Should we be identifying Worklet threads and have inspector tests for that?
(In reply to Joseph Pecoraro from comment #7) > Comment on attachment 412936 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=412936&action=review > > > LayoutTests/ChangeLog:10 > > + Check the thread name to identify the 2 worker threads. This is needed so that worklet thread are > > + not identified as worker threads after r268897. > > Should we be identifying Worklet threads and have inspector tests for that? Probably? Web Inspector support for AudioWorklet is still mostly missing: https://bugs.webkit.org/show_bug.cgi?id=217724