1906816 - devtools test might run into race on process kill with forkserver

Status: RESOLVED FIXED

(DevTools :: Framework, defect)

Description

[:gerard-majax]

From backout https://bugzilla.mozilla.org/show_bug.cgi?id=1896054#c11

TEST-UNEXPECTED-FAIL | devtools/client/framework/test/browser_toolbox_many_toggles.js | uncaught exception - NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIProcessToolsService.kill] at manyTogglesWithContentProcessDestructionsInBackground/interval<@chrome://mochitests/content/browser/devtools/client/framework/test/browser_toolbox_many_toggles.js:33:20

it looks like we have the following cinematic of execution where when we enable forkserver, kill(PID) is done twice (as without forkserver) but the second one returns ESRCH.

Upon all investigation, I have not been able to prove those process would be left hanging around and it rather suggests they are disappearing a little bit faster than they used to, enough that we still see them in https://searchfox.org/mozilla-central/rev/ab488b411ff15c2e11dbd4d6be0455bb64bc8fc1/devtools/client/framework/test/browser_toolbox_many_toggles.js#27-29 but that by the time we kill https://searchfox.org/mozilla-central/rev/ab488b411ff15c2e11dbd4d6be0455bb64bc8fc1/devtools/client/framework/test/browser_toolbox_many_toggles.js#33 they are gone.

Comment 1

[:gerard-majax]

Attachment: Bug 1906816 - Do not fail devtools many toggles test when process died r?ochameau!

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

The Bugbug bot thinks this bug should belong to the 'DevTools::Framework' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 3

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/c8352c374507
Do not fail devtools many toggles test when process died r=ochameau,devtools-reviewers

Comment 4

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/c8352c374507