Gstreamer & GLib memory management/profiling

[Olivier Crête]

Hi,

I should also mention that this is a known upstream bug:

https://bugzilla.gnome.org/show_bug.cgi?id=701694

Olivier

On 31/10/14 11:55 AM, Olivier Crête wrote:
> Hi,
> 
> If you attach GDB, you'll see that this is actually a GLib bug. The
> g_signal_nodes table is allocated from the ctor, so before your main is
> called. Then when it tries to re-alloc that with the profiler, it fail.
> The g_mem_set_vtable() is actually incompatible with GObject itself...
> 
> Olivier
> 
> On 31/10/14 11:42 AM, Sergei Vorobyov wrote:
>> I "forgot" the line
>>
>> extern GMemVTable *glib_mem_profiler_table;
>>
>> before int main ....
>>
>> (by a nasty habit of C-x C-s, which saves in emacs but kills the line
>> and saves in Netbeans)
>>
>> But it *does not change anything*, since it is already defined somewhere
>> in GLib headers.
>>
>>
>>
>> On Fri, Oct 31, 2014 at 4:34 PM, Sergei Vorobyov
>> <sergei.vorobyov at facilitylabs.com
>> <mailto:sergei.vorobyov at facilitylabs.com>> wrote:
>>
>>     > I don't see why anything in GStreamer should be "incompatible" with that
>>     > GLib API.
>>
>>     Neither do I. However, try a *FULL* program below:
>>
>>     // Compile with:
>>     // gcc -std=c11 -Wall -pthread test.c -o test `pkg-config --cflags
>>     --libs glib-2.0 gstreamer-1.0 gstreamer-pbutils-1.0 gstreamer-video-1.0`
>>     // Run and enjoy
>>
>>     #include <stdio.h>
>>     #include <stdlib.h>
>>
>>     #include <string.h>
>>     #include <unistd.h>
>>     #include <error.h>
>>     #include <errno.h>
>>     #include <math.h>
>>
>>     #include <glib.h>
>>     #include <glib/gstdio.h>
>>     #include <glib/gprintf.h>
>>
>>     #include <gst/gst.h>
>>     #include <gst/pbutils/pbutils.h>
>>     #include <gst/video/videooverlay.h>
>>
>>     int main(int argc, char** argv) {
>>         GError *e;
>>         g_mem_set_vtable (glib_mem_profiler_table);
>>         if (g_mem_is_system_malloc ())
>>             g_print ("system malloc is used\n");
>>         else
>>             g_print ("NOT system malloc is used\n");
>>         g_print ("********************* point 1\n");
>>         if (!gst_init_check (&argc, &argv, &e))
>>             g_print ("failed gst_init_check: %s\n", e->message);
>>         g_print ("********************* point 2\n");
>>         g_mem_profile ();
>>         return (EXIT_SUCCESS);
>>     }
>>
>>     It produces, as expected:
>>
>>     fl at nuc5:~/NetBeansProjects/test$ ./test 
>>     NOT system malloc is used
>>     ********************* point 1
>>
>>     (test:3987): GLib-WARNING **: free(0x1a84fd0): memory has been freed
>>     12297829382473034410 times already
>>     *** Error in `./test': realloc(): invalid pointer:
>>     0x0000000001a87e80 ***
>>     Aborted (core dumped)
>>
>>
>>     On Fri, Oct 31, 2014 at 2:20 PM, Tim Müller <tim at centricular.com
>>     <mailto:tim at centricular.com>> wrote:
>>
>>         On Fri, 2014-10-31 at 13:46 +0100, Sergei Vorobyov wrote:
>>
>>         Hi,
>>
>>         > GLib's memory profiling seems incompatible with GStreamer. The
>>         > snippet:
>>
>>         I don't see why anything in GStreamer should be "incompatible"
>>         with that
>>         GLib API.
>>
>>         > extern GMemVTable *glib_mem_profiler_table;
>>         >
>>         >
>>         > gint main (gint argn, gchar** args) {
>>         >     GError *e;
>>         >     g_mem_set_vtable (glib_mem_profiler_table);
>>         >     if (g_mem_is_system_malloc ())
>>         >         g_print ("system malloc is used\n");
>>         >     else
>>         >         g_print ("NOT system malloc is used\n");
>>         >     g_print ("********************* point 1\n");
>>         >     if (!gst_init_check (&argn, &args, &e)) {
>>         >         g_print ("failed gst_init_check: %s\n", e->message);
>>         >     }
>>         >     g_print ("********************* point 2\n");
>>         >
>>         >     g_mem_profile ();
>>         >
>>         >
>>         >     ...
>>         >
>>         >
>>         > crashes at gst_init producing:
>>         >
>>         >
>>         > fl at nuc5:~/bin$ ./mp-v01.020
>>         > NOT system malloc is used
>>         > ********************* point 1
>>         >
>>         >
>>         > (mp-v01.020:15561): GLib-WARNING **: free(0x1b32fd0): memory has been
>>         > freed 12297829382473034410 times already
>>         > *** Error in `./mp-v01.020': realloc(): invalid pointer:
>>         > 0x0000000001b35e80 ***
>>         > Aborted (core dumped)
>>
>>         I would guess this is just a bug somewhere (mixing glib and non-glib
>>         alloc/realloc/free API somewhere). Do you have a full stack
>>         trace from
>>         gdb?
>>
>>         Cheers
>>          -Tim
>>
>>
>>         _______________________________________________
>>         gstreamer-devel mailing list
>>         gstreamer-devel at lists.freedesktop.org
>>         <mailto:gstreamer-devel at lists.freedesktop.org>
>>         http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
>>
>>
>>
>>
>>
>> _______________________________________________
>> gstreamer-devel mailing list
>> gstreamer-devel at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
>>
> 

-- 
Olivier Crête
olivier.crete at collabora.com