Make sure to correctly notify of end of a ServiceWorkerJob when the context is stopped

• Component: Service Workers
• Status: RESOLVED FIXED

[youenn fablet]

Make sure to correctly notify of end of job when the context is stopped

[youenn fablet]

Created attachment 363279 [details]
Patch

[Alex Christensen]

Comment on attachment 363279 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=363279&action=review

> Source/WebCore/ChangeLog:3
> +        Make sure to correctly notify of end of job when the context is stopped

This title should mention ServiceWorkerJob.  Otherwise it's way too vague.

> Source/WebCore/ChangeLog:14
> +        Covered by existing tests not crashing anymore, in particular imported/w3c/web-platform-tests/service-workers/service-worker/skip-waiting.https.html.

Should we unskip this test?

> Source/WebCore/workers/service/ServiceWorkerJob.cpp:-172
> -    if (!m_scriptLoader)

Why can this no longer be null?

[youenn fablet]

(In reply to Alex Christensen from comment #2)
> Comment on attachment 363279 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=363279&action=review
> 
> > Source/WebCore/ChangeLog:3
> > +        Make sure to correctly notify of end of job when the context is stopped
> 
> This title should mention ServiceWorkerJob.  Otherwise it's way too vague.
> 
> > Source/WebCore/ChangeLog:14
> > +        Covered by existing tests not crashing anymore, in particular imported/w3c/web-platform-tests/service-workers/service-worker/skip-waiting.https.html.
> 
> Should we unskip this test?
> 
> > Source/WebCore/workers/service/ServiceWorkerJob.cpp:-172
> > -    if (!m_scriptLoader)
> 
> Why can this no longer be null?

[youenn fablet]

> > > Source/WebCore/ChangeLog:3
> > > +        Make sure to correctly notify of end of job when the context is stopped
> > 
> > This title should mention ServiceWorkerJob.  Otherwise it's way too vague.

OK

> > > Source/WebCore/ChangeLog:14
> > > +        Covered by existing tests not crashing anymore, in particular imported/w3c/web-platform-tests/service-workers/service-worker/skip-waiting.https.html.
> > 
> > Should we unskip this test?

It is unskilled but crashes in debug sometimes.

> > > Source/WebCore/workers/service/ServiceWorkerJob.cpp:-172
> > > -    if (!m_scriptLoader)
> > 
> > Why can this no longer be null?

It is called in one place and we check for this before calling this method.

[Chris Dumez]

Comment on attachment 363279 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=363279&action=review

> Source/WebCore/workers/service/ServiceWorkerContainer.h:151
> +    uint64_t m_ongoingSettledRegistrationCount { 0 };

Bad name, when I saw it I was wondering why you needed this member and did not rely on m_ongoingSettledRegistrations.size() instead.
We probably want to call it m_lastOngoingSettledRegistrationIdentifier, like the data member above.

[youenn fablet]

Created attachment 363352 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 363352 [details]
Patch

Clearing flags on attachment: 363352

Committed r242372: <https://trac.webkit.org/changeset/242372>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/48572995>