Bug 104779 – REGRESSION (r137286-r137296): unit test failure in WebKit2APITests/TestWebKitWebContext

• Component: WebKitGTK
• Status: RESOLVED FIXED

[Zan Dobersek]

A regression was introduced between r137286 and r137296, with the following unit test failing:

TEST: ./Tools/gtk/../Scripts/../../WebKitBuild/Release/Programs/WebKit2APITests/TestWebKitWebContext... (pid=31822)

  /webkit2/WebKitWebContext/default-context:                           OK

  /webkit2/WebKitWebContext/get-plugins:                               OK

  /webkit2/WebKitWebContext/uri-scheme:                                **

ERROR:../../Source/WebKit2/UIProcess/API/gtk/tests/WebViewTest.cpp:206:void resourceGetDataCallback(GObject*, GAsyncResult*, gpointer): assertion failed: (data)

FAIL

The two related builds:
http://build.webkit.org/builders/GTK%20Linux%2064-bit%20Release/builds/32032
http://build.webkit.org/builders/GTK%20Linux%2064-bit%20Release/builds/32033

The commit range:
http://trac.webkit.org/log/?verbose=on&rev=137296&stop_rev=137286

I'll just do the common thing and for starters blame the libsoup bump in r137286.
http://trac.webkit.org/changeset/137286

[Zan Dobersek]

Skipped in r137434.
http://trac.webkit.org/changeset/137434

[Zan Dobersek]

Was treating this as a regression after r137198, but that patch has since been rolled out and the test is still failing.
http://trac.webkit.org/changeset/137198

Perhaps this is related to the recent MainResourceLoader modifications, for instance r137333 and similar patches.
http://trac.webkit.org/changeset/137333

Good thing I have time for a bisection!

[Zan Dobersek]

Tested this with r144170 and the failure still occurs.

[Manuel Rego Casasnovas]

*** Bug 112421 has been marked as a duplicate of this bug. ***

[Manuel Rego Casasnovas]

I've been taking a look to this issue, I don't have too much information but I'd like to share it here.

On one hand, the test works for me in the stable branch, however it's failing in trunk (and it was already failing when I reported the duplicated bug #112421).

On the other hand, it seems that the problem is that DocumentLoader::mainResourceData() is returning NULL. This means that dataReference in WebPage::getMainResourceDataOfFrame() is not properly initialized.

[Carlos Garcia Campos]

It doesn't fail for me with a recent trunk (r146868)

[Manuel Rego Casasnovas]

(In reply to comment #6)
> It doesn't fail for me with a recent trunk (r146868)

It's working for me too in r147306.

[Carlos Garcia Campos]

Could someone unskip it and close this, please?

[Manuel Rego Casasnovas]

Created attachment 195947 [details]
Patch

[WebKit Review Bot]

Comment on attachment 195947 [details]
Patch

Clearing flags on attachment: 195947

Committed r147319: <http://trac.webkit.org/changeset/147319>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[WebKit Review Bot]

Re-opened since this is blocked by bug 113707

[Zan Dobersek]

(In reply to comment #12)
> Re-opened since this is blocked by bug 113707

The test is still failing on the 64-bit release builder, so rolling the unskip patch out. The failure is still the same.

[Carlos Garcia Campos]

(In reply to comment #13)
> (In reply to comment #12)
> > Re-opened since this is blocked by bug 113707
> 
> The test is still failing on the 64-bit release builder, so rolling the unskip patch out. The failure is still the same.

hmm, I managed to reproduce the issue in the 64 bit release bot. It seems the scheme handlers are registered correctly, but libsoup is not handling them. I've just built a new libsoup in the bot and it works, so this will be fixed when bug #113282 lands.

[Zan Dobersek]

Comment on attachment 195947 [details]
Patch

Bug #113282 landed, perhaps just flipping the cq flag on this patch will do all the magic of landing, removing the need to reupload the same patch.

[Carlos Garcia Campos]

Comment on attachment 195947 [details]
Patch

Wait a bit bots still don't build, and the patch has Reviewed by NOBODY

[Zan Dobersek]

Ugh, right. Sorry for jumping forward with it.

[Carlos Garcia Campos]

Comment on attachment 195947 [details]
Patch

Sorry, it seems the 64 bit release bot is building fine now.

[WebKit Review Bot]

Comment on attachment 195947 [details]
Patch

Clearing flags on attachment: 195947

Committed r147559: <http://trac.webkit.org/changeset/147559>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.