Bug 71859 – MHTML documents containing references to resources they don't include never finish loading

• Status: REOPENED
• Component: WebCore Misc.

[Jay Civelli]

Save a page such as cnn.com to MHTML.
Load the generated MHTML.
The page gets loaded, but we never get a notification that the load stopped (and the progress never gets to 100%).

[Jay Civelli]

Created attachment 114159 [details]
Patch

[Jay Civelli]

Created attachment 114163 [details]
Patch

[Adam Barth]

Comment on attachment 114163 [details]
Patch

Test?

[Jay Civelli]

Created attachment 114442 [details]
Patch

[Jay Civelli]

Added a Chromium test for this (as this cannot be tested with a LayoutTest).

[Adam Barth]

Comment on attachment 114442 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=114442&action=review

> Source/WebCore/loader/DocumentLoader.cpp:660
> +        return true;
> +    }
> +    return false;

I would have combined these to "return failLoad", but that's just a nit.

[WebKit Review Bot]

Comment on attachment 114442 [details]
Patch

Clearing flags on attachment: 114442

Committed r99826: <http://trac.webkit.org/changeset/99826>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Tony Gentilcore]

Reopening as I'm rolling this out in bug 72109. Explanation in bug 72039.

[Jay Civelli]

Created attachment 146168 [details]
Patch

[Jay Civelli]

Reviving this bug.
Uploaded a newer patch.

[Adam Barth]

@japhet: Any interest in reviewing this patch?

[Nate Chapin]

Comment on attachment 146168 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=146168&action=review

> Source/WebCore/loader/DocumentLoader.cpp:694
> +        // Schedule this loader without a resource so that resource completion/failure notifications are still fired.
> +        m_pendingSubstituteResources.set(loader, 0);
> +        deliverSubstituteResourcesAfterDelay();

It might make sense to be a little more explicit here that didFail() will get called via substituteResourceDeliveryTimerFired() for a null SubstituteResource. It wasn't immediately obvious to me that this was safe.

[Jay Civelli]

Created attachment 149348 [details]
Patch for landing

[WebKit Review Bot]

Comment on attachment 149348 [details]
Patch for landing

Clearing flags on attachment: 149348

Committed r121206: <http://trac.webkit.org/changeset/121206>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[WebKit Review Bot]

Re-opened since this is blocked by 89935