Bug 147106

Summary: [GTK] SIGABRT in WebKitWebProcess WebKitGtk+ 2.8.4 and 2.10.x when visiting some pages on dispatchDecidePolicyForNavigationAction
Product: WebKit Reporter: Andres Gomez Garcia <agomez>
Component: WebKitGTKAssignee: Nobody <webkit-unassigned>
Status: NEW ---    
Severity: Normal CC: agomez, bugs-noreply, cgarcia, mcatanzaro
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
See Also: https://bugs.webkit.org/show_bug.cgi?id=147111
Attachments:
Description Flags
BT from gdb
none
Another similar BT from gdb
none
Yet another similar BT from gdb
none
And yet another similar BT from gdb none

Description Andres Gomez Garcia 2015-07-20 05:03:00 PDT 
I'm using WebKitGtk+ with my own JHBuild setting:
https://github.com/tanty/jhbuild-epiphany/tree/wkgtk-devel

Epiphany 3.17.1 and WebKit 2.8.4

I'm running Epiphany with the dconf key:

"process-model" = "shared-secondary-process"

When visiting several of the entries at http://www.eljueves.es/feeds/rss.html in different tabs, WebKitGtk+ receives a SIGABRT.

This bug is not reproducible in a predictable way.
Comment 1 Andres Gomez Garcia 2015-07-20 05:04:23 PDT 
Created attachment 257088 [details]
BT from gdb

I also have a 1.2G core, if needed.
Comment 2 Andres Gomez Garcia 2015-07-20 05:05:39 PDT 
I also have a BT from epiphany and a ~600M SIGSEV core from epiphany but I suppose that is caused by the SIGABRT in the WebKitGtk+ SIGBART in the UI process.
Comment 3 Michael Catanzaro 2015-07-20 07:04:56 PDT 
#6  0x00007fc69eef04e2 in std::__throw_bad_function_call() () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
#7  0x00007fc6a7e46b53 in std::function<void (WebCore::PolicyAction)>::operator()(WebCore::PolicyAction) const () from /opt/gnome-os/tanty/epiphany/install/lib/libwebkit2gtk-4.0.so.37

Looks like the std::function was null.

Also, "Thread 637" looks really bad. We shouldn't make more than 20-25 threads.

(In reply to comment #2)
> I also have a BT from epiphany and a ~600M SIGSEV core from epiphany but I
> suppose that is caused by the SIGABRT in the WebKitGtk+ SIGBART in the UI
> process.

That is probably a separate bug worth reporting: the UI process should not crash during a web process crash (what you reported here is a web process crash).
Comment 4 Andres Gomez Garcia 2015-07-20 08:52:52 PDT 
(In reply to comment #3)
...
> That is probably a separate bug worth reporting: the UI process should not
> crash during a web process crash (what you reported here is a web process
> crash).

Oh, right. Didn't realize the crashing process was <path_to>/libexec/webkit2gtk-4.0/WebKitWebProcess .

Check bug 147111.
Comment 5 Andres Gomez Garcia 2015-10-30 02:16:38 PDT 
Created attachment 264385 [details]
Another similar BT from gdb

This happen when using WebKitGtk+ with my own JHBuild setting:
https://github.com/tanty/jhbuild-epiphany/tree/wkgtk-devel

Epiphany 3.18.0 and WebKit 2.10.0

I'm running Epiphany with the dconf key:

"process-model" = "shared-secondary-process"

The compilation was done with CMake args:

"-DPORT=GTK -DCMAKE_BUILD_TYPE=Debug -DDEVELOPER_MODE=ON -DCMAKE_C_FLAGS_DEBUG=-g1 -DCMAKE_CXX_FLAGS_DEBUG=-g1"

WebKitWebProcess crashed as described on bug 150670. Epiphany didn't crash and re-spawned another WebProcess. This second WebProcess crashed again in this way.

This bug is not reproducible in a predictable way.
Comment 6 Andres Gomez Garcia 2015-10-30 04:57:40 PDT 
Created attachment 264392 [details]
Yet another similar BT from gdb

This one was obtained just running MiniBrowser.
Comment 7 Andres Gomez Garcia 2015-11-04 02:24:30 PST 
Happens also in 2.10.3
Comment 8 Andres Gomez Garcia 2015-12-19 08:37:37 PST 
Created attachment 267688 [details]
And yet another similar BT from gdb

BT obtained from 2.10.4 with CMake args:

'-DPORT=GTK -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_FLAGS_RELEASE="-O0 -g1 -DNDEBUG -DG_DISABLE_CAST_CHECKS" -DCMAKE_CXX_FLAGS_RELEASE="-O0 -g1 -DNDEBUG -DG_DISABLE_CAST_CHECKS"'