gtkpod crashes with SIGSEGV in gdk_window_set_geometry_hints

[Chaitanya Gupta]

Binary package hint: gtkpod

I was copying tracks from my iPod (2nd gen Nano) onto the filesystem (NTFS partition), when gtkpod crashed.

ProblemType: Bug
Architecture: amd64
Date: Wed Jan 27 02:25:38 2010
DistroRelease: Ubuntu 9.10
ExecutablePath: /usr/bin/gtkpod
InstallationMedia: Ubuntu 9.10 "Karmic Koala" - Release amd64 (20091027)
Package: gtkpod 0.99.14-2ubuntu3
ProcEnviron:
 PATH=(custom, user)
 LANG=en_IN
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.31-17.54-generic
SourcePackage: gtkpod
Uname: Linux 2.6.31-17-generic x86_64

[Chaitanya Gupta (mail-chaitanyagupta)]

Here's the backtrace from gdb for this process:

Program received signal SIGSEGV, Segmentation fault.
0x00007f0ec42cbe64 in gdk_window_set_geometry_hints () from /usr/lib/libgdk-x11-2.0.so.0
(gdb) bt
#0 0x00007f0ec42cbe64 in gdk_window_set_geometry_hints () from /usr/lib/libgdk-x11-2.0.so.0
#1 0x00007f0ec476b812 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#2 0x00007f0ec29035ae in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#3 0x00007f0ec291864d in ?? () from /usr/lib/libgobject-2.0.so.0
#4 0x00007f0ec2919d39 in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#5 0x00007f0ec291a283 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#6 0x00007f0ec45c4bd0 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#7 0x00007f0ec427d8c6 in ?? () from /usr/lib/libgdk-x11-2.0.so.0
#8 0x00007f0ec2466bce in g_main_context_dispatch () from /lib/libglib-2.0.so.0
#9 0x00007f0ec246a598 in ?? () from /lib/libglib-2.0.so.0
#10 0x00007f0ec246a6c0 in g_main_context_iteration () from /lib/libglib-2.0.so.0
#11 0x00007f0ec4646021 in gtk_main_iteration () from /usr/lib/libgtk-x11-2.0.so.0
#12 0x000000000044e825 in export_files_init ()
#13 0x00007f0ec29035ae in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#14 0x00007f0ec2918983 in ?? () from /usr/lib/libgobject-2.0.so.0
#15 0x00007f0ec2919d39 in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#16 0x00007f0ec291a283 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#17 0x00007f0ec475980e in gtk_widget_activate () from /usr/lib/libgtk-x11-2.0.so.0
#18 0x00007f0ec465a74d in gtk_menu_shell_activate_item () from /usr/lib/libgtk-x11-2.0.so.0
#19 0x00007f0ec465bf49 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#20 0x00007f0ec464c728 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#21 0x00007f0ec29035ae in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#22 0x00007f0ec291864d in ?? () from /usr/lib/libgobject-2.0.so.0
#23 0x00007f0ec2919bcc in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#24 0x00007f0ec291a283 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#25 0x00007f0ec475371f in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#26 0x00007f0ec4644cd3 in gtk_propagate_event () from /usr/lib/libgtk-x11-2.0.so.0
#27 0x00007f0ec4645ca3 in gtk_main_do_event () from /usr/lib/libgtk-x11-2.0.so.0
#28 0x00007f0ec42bacdc in ?? () from /usr/lib/libgdk-x11-2.0.so.0
#29 0x00007f0ec2466bce in g_main_context_dispatch () from /lib/libglib-2.0.so.0
#30 0x00007f0ec246a598 in ?? () from /lib/libglib-2.0.so.0
#31 0x00007f0ec246a9f5 in g_main_loop_run () from /lib/libglib-2.0.so.0
#32 0x00007f0ec4646177 in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0
#33 0x000000000045b352 in main ()

[Elladan (elladan)]

This happens 100% of the time when using the copy to filesystem feature for me. It makes the feature completely unusable.

[Matthew Horr (matthorr)]

Also affects me copying to filesystem with Nana 3G

[puzzler995 (emarsee1)]

Also affects me copying to filesystem with iPod Touch 2G with iPhone OS 3.1.2 installed

[Motin (motin)]

Same here. Anyone knows another way to copy files from iPod to filesystem with the track/video/filenames intact?

[Duane Waddle (duane-waddle)]

This problem appears to be related to a section of code where each file to be exported is handled in a separate worker thread. From what I can tell, when the worker thread does something that causes the warning dialog box to pop up (overwriting an existing file for example), this causes the crash.

I don't know enough about the app (or about GTK+ apps in general) to fix the underlying issue right now, but there was an #ifdef around the threaded version of the code and a #else to fallback to its single-thread equivalent.

The patch attached is an ugly, horrible workaround that rigs the #ifdef to where the single-thread version of the code is always used. It doesn't fix the underlying issue at all - but at least should allow folks who are being impacted to work around it.

[Drew Scott Daniels (drewdaniels)]

I backported version 1.0 from mavric (just a simple source download and build), and I didn't get this error, I got a different error. Unfortunately ubuntu-bug said the package wasn't from ubuntu so it wasn't easy for me to add data to Launchpad about the new failure.

First I downloaded the three source files into a directory by themselves, then I ran:
$ dpkg-source -x gtkpod_1.0.0-0ubuntu1.dsc
$ cd gtkpod-1.0.0/
$ dpkg-buildpackage -b
$ dpkg -i ../*.deb

After running gtkpod and extracting the files, it seemed
gdb showed:
gtkpod: ../../src/xcb_io.c:249: process_responses: Assertion `(((long) (dpy->last_request_read) - (long) (dpy->request)) <= 0)' failed.

Program received signal SIGABRT, Aborted.
[Switching to Thread 0x7fffe27fc700 (LWP 9684)]
0x00007ffff4794a75 in raise () from /lib/libc.so.6

I still want to check the source code to see if the relevant part was made thread safe, or I just didn't hit the race condition before hitting this other bug.

[Drew Scott Daniels (drewdaniels)]

file_export.c sometimes calls gtkpod_warning
misc_confirm.c's gtkpod_warning calls gtkpod_confirmation
confirmation.c's GtkResponseType gtkpod_confirmation is fairly big and finding potential race conditions there would be very time consuming for me. Since this code hasn't changed in the latest upstream release I think an upstream bug needs to be filed for this issue.
The upstream bug-tracker is at http://gtkpod.org/bugs/index.php?do=toplevel&project=0 and I haven't seen any bug filed that's similar to this one except one that claims there's no error messages for an export files crash:
http://gtkpod.org/bugs/index.php?do=details&task_id=24&project=1&status[0]=

I'm going to try to open an link in a new upstream bug.

[Drew Scott Daniels (drewdaniels)]

I've filed this as a new upstream bug at: http://gtkpod.org/bugs/index.php?do=details&task_id=37&project=1

[Duane Waddle (duane-waddle)]

Upstream bug was closed as 'fixed' on 2011-02-15. I've not been able to test it yet, but a quick glance at the source repository makes me think it was fixed in commit 40d3357adca13c532a706e86496f8d3410722484 .

http://gtkpod.git.sourceforge.net/git/gitweb.cgi?p=gtkpod/gtkpod;a=commit;h=40d3357adca13c532a706e86496f8d3410722484

[Gianfranco Costamagna (costamagnagianfranco)]

Ubuntu 9.10 Has reached his EOL, and this bug seems to be fixed upstream, so I'm closing.