Bug 164914 – [GTK] Web/database process crash in gdk_event_source_check

• Status: RESOLVED
• Resolution: WONTFIX
• Component: WebKitGTK

[Michael Catanzaro]

Web process crash in gdk_event_source_check:

Truncated backtrace:
Thread no. 1 (9 frames)
 #0 _g_log_abort at gmessages.c:487
 #1 g_log_writer_default at gmessages.c:2382
 #2 g_log_structured_array at gmessages.c:1770
 #3 g_log_structured at gmessages.c:1597
 #4 gdk_event_source_check at gdkeventsource.c:92
 #5 g_main_context_check at gmain.c:3762
 #8 WTF::RunLoop::run at /usr/src/debug/webkitgtk-2.14.1/Source/WTF/wtf/glib/RunLoopGLib.cpp:94
 #9 WebKit::ChildProcessMain<WebKit::WebProcess, WebKit::WebProcessMain> at /usr/src/debug/webkitgtk-2.14.1/Source/WebKit2/Shared/unix/ChildProcessMain.h:61
 #11 _start

It's crashing on this error condition:

      if (source->pfd.revents & G_IO_IN)
        {
          if (wl_display_read_events (display_wayland->wl_display) < 0)
            g_error ("Error reading events from display: %s", g_strerror (errno));
        }

[Michael Catanzaro]

(Full backtrace downstream.)

[Michael Catanzaro]

And the error condition is "broken pipe"

[Michael Catanzaro]

OK, same crash and same error condition in WebKitDatabaseProcess as well. Why are we running GTK+ in the database process???

[Michael Catanzaro]

(Note: 46 reports of this.)

[Michael Catanzaro]

Looks very similar to this gnome-terminal crash when the user logs out of GNOME (i.e. when the Wayland display goes away): https://bugzilla.redhat.com/show_bug.cgi?id=1395389

[Michael Catanzaro]

Looks like it affects a bunch of other GTK+ applications:

https://retrace.fedoraproject.org/faf/problems/2506226/

I doubt it's our bug.

[Michael Catanzaro]

Ahhh, turns out this is just the infamous https://bugzilla.redhat.com/show_bug.cgi?id=1366897

[Michael Catanzaro]

(In reply to comment #3)
> OK, same crash and same error condition in WebKitDatabaseProcess as well.
> Why are we running GTK+ in the database process???

Fixed in bug #165138.