1411528 - wpt redirected-response.https.htm service-worker test crashes in debug

Status: RESOLVED FIXED

(Core :: DOM: Service Workers, defect)

Description

[James Graham [:jgraham]]

On Windows it also causes OOM trying to call mozcrash, so disabling the test test for now.

Example log:

https://treeherder.mozilla.org/logviewer.html#?job_id=139311778&repo=try&lineNumber=39769

Comment 1

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/c80fbf59afde
Disable service worker test for OOM during crash on Windows, a=testonly

Comment 2

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

https://hg.mozilla.org/mozilla-central/rev/c80fbf59afde

Comment 3

[Ben Kelly [:bkelly, not reviewing]]

Investigating.

Comment 4

[Ben Kelly [:bkelly, not reviewing]]

I believe this is happening due to changes I made in bug 1390638.  We didn't have this test coverage at the time, so I didn't catch it.

Comment 5

[Ben Kelly [:bkelly, not reviewing]]

Attachment: Clear the intercept listening when synthesizing a redirect in HttpChannelChild. r=asuth

https://treeherder.mozilla.org/#/jobs?repo=try&revision=43d3866ab92e8cd59637a76048af50a24504400b

Comment 6

[Ben Kelly [:bkelly, not reviewing]]

Comment on attachment 8922489 [details] [diff] [review]
Clear the intercept listening when synthesizing a redirect in HttpChannelChild. r=asuth

Andrew, the WPT test case that is crashing is doing :

  `respondWith(Response.redirect(url2))`

For an outer `fetch(url1, { redirect: 'error' })`.

This means we are trying to synthesize a redirect, but that the channel will end up being canceled because of the RequestRedirect being set to "error".

When Cancel() is called during the "maybe synthesizing a redirect" step we end up calling AsyncAbort() because mInterceptListener is set.  However, we also have a remote actor here that also is canceling.  So we end up getting two sets of OnStartRequest/OnStopRequest.

The fix in this patch is to simply clear the mInterceptListener when we synthesized the redirect.  We know we're not actually synthesizing a body since we are following a redirect instead.

And of course, it will be nice when we can get rid of the child side intercept code because this will get easier to understand.

Comment 7

[Andrew Sutherland [:asuth] (he/him)]

Comment on attachment 8922489 [details] [diff] [review]
Clear the intercept listening when synthesizing a redirect in HttpChannelChild. r=asuth

Review of attachment 8922489 [details] [diff] [review]:
-----------------------------------------------------------------

Thank you for the very detailed explanation.

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/6f9a97d89dfd
Clear the intercept listening when synthesizing a redirect in HttpChannelChild. r=asuth

Comment 9

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

https://hg.mozilla.org/mozilla-central/rev/6f9a97d89dfd