After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 579029 - soup crashes at home page load
soup crashes at home page load
Status: RESOLVED DUPLICATE of bug 580193
Product: libsoup
Classification: Core
Component: HTTP Transport
2.26.x
Other Linux
: Normal major
: ---
Assigned To: libsoup-maint@gnome.bugs
libsoup-maint@gnome.bugs
Depends on:
Blocks:
 
 
Reported: 2009-04-15 09:48 UTC by Josh Saddler
Modified: 2009-05-05 22:28 UTC
See Also:
GNOME target: ---
GNOME version: 2.23/2.24



Description Josh Saddler 2009-04-15 09:48:31 UTC
Hey. I jumped on to #midori to investigate a crash whenever I loaded up my preferred homepage, http://www.signonsandiego.com. Seemed to not be a Midori problem; the folks on that channel indicated it was a libsoup bug. They were able to consistently reproduce the crash. The page can't finish loading; it gets most of the way done, then crashes the browser. No such issues in Firefox, but Firefox ain't using libsoup.

I can't find any special HTML or scripts on that page, except perhaps an embedded flash video (may be realmedia). There is an HTTP meta-refresh on a timed 10-minute interval buried somewhere in the code. Dunno if they're related to soup's failure. I built libsoup with support for SSL, but no other special --enable-foo configure options.

Anyway, here's the output:



New Thread 0xb40ebb90 (LWP 17391)]
[New Thread 0xb3febb90 (LWP 17392)]
[New Thread 0xb43ffb90 (LWP 17393)]

(midori:17375): GLib-GObject-CRITICAL **: g_object_unref: assertion `G_IS_OBJECT (object)' failed

(midori:17375): GLib-GObject-WARNING **: invalid uninstantiatable type `(null)' in cast to `SoupSession'

(midori:17375): GLib-GObject-WARNING **: instance of invalid non-instantiatable type `(null)'

Program received signal SIGSEGV, Segmentation fault.

Thread 3066398544 (LWP 17375)

  • #0 soup_session_get_queue
    at soup-session.c line 1198
  • #1 run_queue
    at soup-session-async.c line 240
  • #2 g_cclosure_marshal_VOID__VOID
    from /usr/lib/libgobject-2.0.so.0
  • #3 g_closure_invoke
    from /usr/lib/libgobject-2.0.so.0
  • #4 ??
    from /usr/lib/libgobject-2.0.so.0
  • #5 g_signal_emit_valist
    from /usr/lib/libgobject-2.0.so.0
  • #6 g_signal_emit
    from /usr/lib/libgobject-2.0.so.0
  • #7 soup_message_restarted
    at soup-message.c line 877
  • #8 soup_message_io_finished
    at soup-message-io.c line 168
  • #9 g_cclosure_marshal_VOID__VOID
    from /usr/lib/libgobject-2.0.so.0
  • #10 g_closure_invoke
    from /usr/lib/libgobject-2.0.so.0
  • #11 ??
    from /usr/lib/libgobject-2.0.so.0
  • #12 g_signal_emit_valist
    from /usr/lib/libgobject-2.0.so.0
  • #13 g_signal_emit
    from /usr/lib/libgobject-2.0.so.0
  • #14 socket_read_watch
    at soup-socket.c line 1116
  • #15 ??
    from /usr/lib/libglib-2.0.so.0
  • #16 g_main_context_dispatch
    from /usr/lib/libglib-2.0.so.0
  • #17 ??
    from /usr/lib/libglib-2.0.so.0
  • #18 g_main_loop_run
    from /usr/lib/libglib-2.0.so.0
  • #19 gtk_main
    from /usr/lib/libgtk-x11-2.0.so.0
  • #20 main
    at ../midori/main.c line 1753

Comment 1 Josh Saddler 2009-04-15 09:54:13 UTC
Just discovered something interesting: disabling all scripts within Midori, for a completely javascript/whatever-free experience, lets the page load just fine. All other pages work, too. The embedded videos are still perfectly viewable, too.

I guess . . . it's a bug in libsoup's script handling?
Comment 2 Dan Winship 2009-04-15 13:17:37 UTC
(In reply to comment #1)
> I guess . . . it's a bug in libsoup's script handling?

Nope, libsoup doesn't know anything about scripts. It's all just bytes.

> #7  0xb7a2897f in soup_message_restarted (msg=0x9e9ad38) at soup-message.c:877

this shows that the crash happens when a message is getting sent a second time. So presumably, when you enable javascript, one of the requests (either a js file itself or else something requested by a script) is getting a redirect response, and for some reason there's a refcounting bug or something that's making libsoup crash when it tries to load the redirected-to resource.
Comment 3 Alexander V. Butenko 2009-05-05 05:17:24 UTC
i was able to reproduse this bug with webkit 1.1.5 and it seems fixed with 1.1.6. 
Seems it was a webkit backend issue. 
Ticket can be closed.
Comment 4 Josh Saddler 2009-05-05 07:08:17 UTC
Yup, webkit-1.1.6 seems to have fixed the error. I dunno where libsoup's interaction fit in, but it all seems okay now. Can load the site with scripts working just fine.

My guess is something in webkit was broken that libsoup didn't like, but now webkit's fixed, so libsoup is a happy camper?

My take is that this bug can be closed, unless the libsoup folks have other plans.
Comment 5 Dan Winship 2009-05-05 14:07:22 UTC
oops, no, this is a dup of 580193, which got fixed in libsoup. (If you don't have the latest libsoup, then maybe kov put a workaround in WebKit too?)


*** This bug has been marked as a duplicate of 580193 ***
Comment 6 Alexander V. Butenko 2009-05-05 16:40:27 UTC
hehe. probably thats mine case. I have a -git version of libsoup.
Comment 7 Gustavo Noronha (kov) 2009-05-05 22:28:25 UTC
Not really. It may be a side-effect of the new error reporting code, that Jan wrote, though.