Crashes in GIMP 2.6.6

Hello people.

I having frequent crashes in Gimp 2.6.6, the version distributed by the
Slackware 64 12.2 version.

It seems quite random. In most cases (all except one), the crash happened
while trying to save an image - just before the file dialog appeared. The
exception crashed when the dialog was already visible. First the text
entry box didn't accept text (cursor disappeared when moved over the box).
Then, after a couple of seconds, the crash happened.

I was running gimp in gdb, hoping to do a backtrace, but detected that
Gimp had been compiled without debug info. The text appearing is:


Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7ffcda5c66f0 (LWP 15429)]
0x00007ffcd8f02272 in _gtk_entry_completion_popup (completion=0x28270f0)
at gtkentrycompletion.c:1491
1491 gtkentrycompletion.c: No such file or directory.
in gtkentrycompletion.c
(gdb)
(script-fu:15434): LibGimpBase-WARNING **: script-fu: gimp_wire_read():
error

The backtrace only showed:

#0 0x00007ffcd8f02272 in _gtk_entry_completion_popup
#(completion=0x28270f0)
at gtkentrycompletion.c:1491
Cannot access memory at address 0x7fffe25f2288


I have used Gimp quite a bit the last days, and only had problems after
using a scaling script (in Guile). Could this be the problem?

John

On Fri, 3 Jul 2009 20:45:27 -0300
John Coppens <john@jcoppens.com> wrote:

> I was running gimp in gdb, hoping to do a backtrace, but detected that
> Gimp had been compiled without debug info.

I recompiled a new version of Gimp (still 2.6.6) and the error remains.
But now I can get a backtrace:

#0 0x00007f08f94df272 in _gtk_entry_completion_popup (
completion=0x7f08ede86800) at gtkentrycompletion.c:1491
#1 0x00007f08f94d3318 in gtk_entry_completion_timeout (
data=<value optimized out>) at gtkentry.c:9021
#2 0x00007f08f917b86b in gdk_threads_dispatch (data=0x7f08e8322300)
at gdk.c:498
#3 0x00007f08f64606cb in g_timeout_dispatch (source=0x29413b0,
callback=0xb64ed0, user_data=0xb64ed0) at gmain.c:3250
#4 0x00007f08f645ff5a in IA__g_main_context_dispatch (context=0xb0ca50)
at gmain.c:1814
#5 0x00007f08f64635d8 in g_main_context_iterate (context=0xb0ca50,
#block=1,
dispatch=1, self=<value optimized out>) at gmain.c:2445
#6 0x00007f08f6463acd in IA__g_main_loop_run (loop=0xf530a0) at
#gmain.c:2653 7 0x000000000046a2fd in app_run (full_prog_name=<value
#optimized out>,
filenames=0x0, alternate_system_gimprc=0x0, alternate_gimprc=0x0,
session_name=<value optimized out>, batch_interpreter=0x0,
batch_commands=0x0, as_new=0, no_interface=0, no_data=0, no_fonts=0,
no_splash=0, be_verbose=0, use_shm=1, use_cpu_accel=1,
console_messages=0, use_debug_handler=0,
stack_trace_mode=GIMP_STACK_TRACE_NEVER,
pdb_compat_mode=GIMP_PDB_COMPAT_ON) at app.c:246
#8 0x000000000046b28f in main (argc=1, argv=0x7fff02bcfc68) at main.c:406


It's not much, but it might help. Is this some error of mine, or should I
submit this as a bug report?

John

On 07/04/2009 06:33 AM, John Coppens wrote:
> I recompiled a new version of Gimp (still 2.6.6) and the error remains.
> But now I can get a backtrace:
>
> It's not much, but it might help. Is this some error of mine, or should I
> submit this as a bug report?
>

Yes please, submit a bug report:
bugs.gimp.org

Be sure to specify the GTK+ version you use, and also step-by-step
instructions on how to reproduce the problem

/ Martin

Hi,

On Fri, 2009-07-03 at 20:45 -0300, John Coppens wrote:

> I have used Gimp quite a bit the last days, and only had problems after
> using a scaling script (in Guile). Could this be the problem?

More likely the problem is that you are using unstable development
versions of GLib and GTK+ (as indicated in your bug report). Is there a
specific reason you are doing that? And why don't you first check if
using a stable version of the toolkit fixes your problem?


Sven

On Sun, 05 Jul 2009 23:23:53 +0200
Sven Neumann <sven@gimp.org> wrote:

> And why don't you first check if
> using a stable version of the toolkit fixes your problem?

I apologize for that, and for not seeing your answer immediately (you
did Cc: me, but that msg didn't arrive). In my defense, I did check the
README file of GTK +, and did not see any indication that 2.17 was an
unstable version:

---Readme:
This is GTK+ version 2.17.2. GTK+ is a multi-platform toolkit for
creating graphical user interfaces. Offering a complete set of widgets,
GTK+ is suitable for projects ranging from small one-off projects to
complete application suites.
---

I'd say the first line of the README would be a logical place to warn
people, wouldn't it? In hindsight, I _did_ know the odd versions were
development. I guess stress got me... Just switched everything to 64 bit
here.

Will test with 2.16.4 now. Again, sorry!

John