Bug 158474 – [GTK] Layout Test editing/execCommand/print.html fails as timeout

• Status: NEW
• Component: WebKitGTK

[Fujii Hironori]

Following test cases are constantly failing on my Debian 8 box, while they succeeds on BuildBot.

>  editing/execCommand/print.html [ Timeout ]
>  fast/media/print-restores-previous-mediatype.html [ Timeout ]

I set up Debian 8 on VirtualBox and executed following commands:

> $ ./Tools/gtk/install-dependencies
> $ ./Tools/Scripts/update-webkitgtk-libs
> $ ./Tools/Scripts/build-webkit --gtk --release --64-bit
> $ ./Tools/Scripts/run-webkit-tests --gtk --release --no-new-test-results

[Fujii Hironori]

Swiching to Fedora 24, but still this problem happens.

This is a callstack:

> #0  0x00007f3cad37832d in poll () from /lib64/libc.so.6
> #1  0x00007f3cb22b0b4d in g_poll () from /home/fujii/work/webkit/w1/WebKitBuild/DependenciesGTK/Root/lib/libglib-2.0.so.0
> #2  0x00007f3cb229fc54 in g_main_context_poll () from /home/fujii/work/webkit/w1/WebKitBuild/DependenciesGTK/Root/lib/libglib-2.0.so.0
> #3  0x00007f3cb229f5e7 in g_main_context_iterate () from /home/fujii/work/webkit/w1/WebKitBuild/DependenciesGTK/Root/lib/libglib-2.0.so.0
> #4  0x00007f3cb229fa38 in g_main_loop_run () from /home/fujii/work/webkit/w1/WebKitBuild/DependenciesGTK/Root/lib/libglib-2.0.so.0
> #5  0x00007f3caf0ee68e in gtk_enumerate_printers () from /home/fujii/work/webkit/w1/WebKitBuild/DependenciesGTK/Root/lib/libgtk-3.so.0
> #6  0x00007f3cbf928b5b in (anonymous namespace)::PrinterListGtk::PrinterListGtk (this=0x7f3c9ece68b8)
>     at ../../Source/WebKit2/WebProcess/WebPage/gtk/PrinterListGtk.cpp:55
> #7  0x00007f3cbf9289e5 in (anonymous namespace)::PrinterListGtk::getOrCreate ()
>     at ../../Source/WebKit2/WebProcess/WebPage/gtk/PrinterListGtk.cpp:42
> #8  0x00007f3cbf71ecf4 in (anonymous namespace)::WebChromeClient::print (this=0x2430aa0, frame=0x7f3c9eda0000)
>     at ../../Source/WebKit2/WebProcess/WebCoreSupport/WebChromeClient.cpp:616
> #9  0x00007f3cc06dd4d6 in (anonymous namespace)::Chrome::print (this=0x7f3c9edea330, frame=0x7f3c9eda0000)
>     at ../../Source/WebCore/page/Chrome.cpp:427
> #10 0x00007f3cc025075b in (anonymous namespace)::executePrint (frame=...) at ../../Source/WebCore/editing/EditorCommand.cpp:937
> #11 0x00007f3cc0252ae8 in (anonymous namespace)::Editor::Command::execute (this=0x7ffd6f8d61b0, parameter=..., triggeringEvent=0x0)
>     at ../../Source/WebCore/editing/EditorCommand.cpp:1779
> #12 0x00007f3cc00b9c82 in (anonymous namespace)::Document::execCommand (this=0x7f3c9ed43000, commandName=..., userInterface=true, 
>     value=...) at ../../Source/WebCore/dom/Document.cpp:4922
> #13 0x00007f3cc136c4ef in (anonymous namespace)::jsDocumentPrototypeFunctionExecCommand (state=0x7ffd6f8d6290)
>     at DerivedSources/WebCore/JSDocument.cpp:4552
> #14 0x00007f3c5e288028 in ?? ()
> #15 0x00007ffd6f8d6320 in ?? ()
> #16 0x00007f3cb927f48b in llint_entry () from /home/fujii/work/webkit/w1/WebKitBuild/Debug/lib/libjavascriptcoregtk-4.0.so.18
> Backtrace stopped: frame did not save the PC

[Fujii Hironori]

Bug 686838 – gtk_enumerate_printers() hangs
https://bugzilla.gnome.org/show_bug.cgi?id=686838

There is a workaround to enable cloudprint.

$ sudo dnf install rest-devel
$ ./Tools/jhbuild/jhbuild-wrapper --gtk buildone -f -a -c gtk+

Solved.

[Carlos Garcia Campos]

Oh, so it's the gtk_enumerate_printers hang, I had forgotten that GTK+ bug. We should fix that in GTK+ then

[Carlos Garcia Campos]

Marek's patch in https://bugzilla.gnome.org/show_bug.cgi?id=686838 is good, we can probably add it to our jhbuild to make the test pass.