1673378 - Crash running a non trivial mochitest involving service workers on Linux debug builds

Status: RESOLVED WORKSFORME

(Core :: DOM: Content Processes, defect, P1)

Description

[Nicolas Chevobbe [:nchevobbe]]

While working on some worker debugging support for DevTools, I modified a test so it was a bit more involved (creating multiple documents, with iframes on different origins, removing those, terminating workers, reloading ...).
As a result, some TRY jobs were crashing and I couldn't see how DevTools was causing this.

As a last attempt, I tried to copy the test and removed all the DevTools instructions from it, so it now only performs "regular" operations (waiting for some time between each interaction).

And I still get a crash:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=52f0e3319c174bdc47d0d0ee5d454546bcde6a7d&selectedTaskRun=IMQkJlXwQemkRqsm4fupgg.0

You can see the test here https://hg.mozilla.org/try/rev/2d918c56e85914b3c257b03b6476ac817c501885 (it's only within devtools because the original test was in this folder).

Comment 1

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

The severity field is not set for this bug.
:jstutte, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 2

[Jens Stutte [:jstutte]]

Hi Nicolas, is this still happening?

Comment 3

[Nicolas Chevobbe [:nchevobbe]]

(In reply to Jens Stutte [:jstutte] from comment #2)

Hi Nicolas, is this still happening?

yes, it still does. I rebased my patch and pushed to TRY again this morning: https://treeherder.mozilla.org/jobs?repo=try&revision=422ce65aa4df6913b9aeb26462f885b317f60b00&selectedTaskRun=FDopYC3nQtGfnwSTLx1MYA.0

Comment 4

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

The severity field is not set for this bug.
:ytausky, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 5

[Yaron Tausky [:ytausky]]

The crashing jobs show two assertions being triggered in the parent process; the first one is related to (what I think is) interned strings and the second one is about the list of ContentParents. Both are very general and not related to service workers. I'm not sure what the right component for those would be, I have to check.

Comment 6

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

The severity field is not set for this bug.
:ytausky, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 7

[Jens Stutte [:jstutte]]

Tentatively moving to DOM: Content Processes as this is where UniquePtr<LinkedList<ContentParent>> ContentParent::sContentParents; lives. It would be interesting to see if this still happens now with fission enabled.

Olli, can you please suggest a severity here? Thanks!

Comment 8

[Olli Pettay [:smaug][[e-mail]]]

This smells bad. Possibly even S1 bad. I'm trying to get the testcase running again.

Comment 9

[Olli Pettay [:smaug][[e-mail]]]

Attachment: testcase tweaked

Apparently some directories have moved in devtools

Comment 10

[Olli Pettay [:smaug][[e-mail]]]

But locally this didn't assert anything.

Please reopen if you can still reproduce this.