Assertion failure (halloc.c:75) "value" in GTKG-0.96.3 with GTK2

[uberpinguin]

Hello.  Trying to figure out the cause of this - it's apparently an
uncommon problem (though not completely unique:
http://www.nabble.com/I've-been-making-a-hash-of-halloc...-t2576447.html).
When starting a GTKG built with GTK2 support, it crashes with an
assertion failure in halloc.c.  I typically use '-fsigned-char
-march=pentium4 -O2 -pipe' for my CFLAGS, but get the same error with
just '-O2 -g' and '-O0'.
GTKG built with GTK1 does not exhibit this behavior.

I am using Gentoo Linux, but I get the same error when I compile
gtk-gnutella directly from the source available on sourceforge.

Here is my 'emerge --info' (system profile information):
        Portage 2.1.2-r9 (default-linux/x86/2006.1, gcc-4.1.2,
        glibc-2.5-r0, 2.6.19-beyond2 i686)
        =================================================================
        System uname: 2.6.19-beyond2 i686 Mobile Intel(R) Pentium(R) 4 -
        M CPU 2.20GHz
        Gentoo Base System release 1.12.9
        Timestamp of tree: Fri, 16 Feb 2007 00:59:01 +0000
        distcc 2.18.3 i686-pc-linux-gnu (protocols 1 and 2) (default
        port 3632) [disabled]
        dev-java/java-config: 1.3.7, 2.0.31-r3
        dev-lang/python:     2.4.4
        dev-python/pycrypto: 2.0.1-r5
        sys-apps/sandbox:    1.2.18.1
        sys-devel/autoconf:  2.13, 2.61
        sys-devel/automake:  1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3,
        1.9.6-r2, 1.10
        sys-devel/binutils:  2.17
        sys-devel/gcc-config: 1.3.14
        sys-devel/libtool:   1.5.22
        virtual/os-headers:  2.6.20
        ACCEPT_KEYWORDS="x86 ~x86"
        AUTOCLEAN="yes"
        CBUILD="i686-pc-linux-gnu"
        CFLAGS="-fsigned-char -march=pentium4 -O2 -pipe"
        CHOST="i686-pc-linux-gnu"
        CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config"
        CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/java-config/vms/ /etc/revdep-rebuild /etc/terminfo /etc/texmf/web2c"
        CXXFLAGS="-fsigned-char -march=pentium4 -O2 -pipe"
        DISTDIR="/usr/portage/distfiles"
        FEATURES="autoconfig ccache distlocks metadata-transfer sandbox
        sfperms strict"
        GENTOO_MIRRORS="http://distfiles.gentoo.org
        http://distro.ibiblio.org/pub/linux/distributions/gentoo"
        LANG="en_US.utf8"
        LC_ALL="en_US.utf8"
        MAKEOPTS="-j3"
       

gtk-gnutella is configured with dbus, gnutls and nls support.  I am also
including the output from running GTKG configured with -Oders
-Dgtkversion=2 -Doptimize='-O0 -g3' under gdb:

        GNU gdb 6.6
        Copyright (C) 2006 Free Software Foundation, Inc.
        GDB is free software, covered by the GNU General Public License,
        and you are
        welcome to change it and/or distribute copies of it under
        certain conditions.
        Type "show copying" to see the conditions.
        There is absolutely no warranty for GDB.  Type "show warranty"
        for details.
        This GDB was configured as "i686-pc-linux-gnu"...
        Using host libthread_db library "/lib/libthread_db.so.1".
        (gdb) run
        Starting
        program: /home/ubrpngn/build/gtk-gnutella-0.96.3/src/gtk-gnutella
        [Thread debugging using libthread_db enabled]
        [New Thread -1208277328 (LWP 22639)]
        07-02-18 21:01:16 (WARNING): This is an unofficial build which
        accesses files in this directory:
        "/home/ubrpngn/build/gtk-gnutella-0.96.3"
        07-02-18 21:01:16 (MESSAGE): language code: "en"
        07-02-18 21:01:16 (MESSAGE): using locale character set "UTF-8"
        07-02-18 21:01:16 (MESSAGE): primary filename character set
        "UTF-8"
        07-02-18 21:01:16 (MESSAGE): gtk-gnutella/0.96.3 (2006-11-09;
        r12293; GTK2; Linux i686)
        07-02-18 21:01:16 (MESSAGE): D-BUS set up and ready for use.
        07-02-18 21:01:16 (MESSAGE): Sent D-BUS signal 'Events': started
        07-02-18 21:01:16 (WARNING): can't read file
        "/home/ubrpngn/.gtk-gnutella/config_gnet": No such file or
        directory
        [New Thread -1213371504 (LWP 22644)]
        [New Thread -1221764208 (LWP 22645)]
        [Thread -1213371504 (LWP 22644) exited]
        [Thread -1221764208 (LWP 22645) exited]
        07-02-18 21:01:16 (WARNING):
        gnutls_certificate_set_x509_key_file() failed: Error while
        reading file.
        07-02-18 21:01:16 (WARNING): [hosts] unable to retrieve: no
        alternate locations known
        07-02-18 21:01:16 (WARNING): [hosts] unable to retrieve: no
        alternate locations known
        07-02-18 21:01:16 (WARNING): [hostile IP addresses (private)]
        unable to retrieve: no alternate locations known
        07-02-18 21:01:16 (WARNING): [Spam database] unable to retrieve:
        no alternate locations known
        07-02-18 21:01:16 (WARNING): can't read file
        "/home/ubrpngn/.gtk-gnutella/sha1_cache": No such file or
        directory
        07-02-18 21:01:16 (WARNING): [download mesh] unable to retrieve:
        no alternate locations known
        07-02-18 21:01:16 (WARNING): [banned mesh] unable to retrieve:
        no alternate locations known
        07-02-18 21:01:16 (WARNING): [Host Whitelist] unable to
        retrieve: no alternate locations known
        07-02-18 21:01:16 (WARNING): [PARQ upload queue data] unable to
        retrieve: no alternate locations known
        07-02-18 21:01:16 (WARNING): can't read file
        "/home/ubrpngn/.gtk-gnutella/config_gui": No such file or
        directory
       
        Assertion failure (halloc.c:75) "value"
       
        Program received signal SIGABRT, Aborted.
        [Switching to Thread -1208277328 (LWP 22639)]
        0xb7fe0410 in __kernel_vsyscall ()
        (gdb) bt full
        #0  0xb7fe0410 in __kernel_vsyscall ()
        No symbol table info available.
        #1  0x4120fb97 in raise () from /lib/libc.so.6
        No symbol table info available.
        #2  0x4121114e in abort () from /lib/libc.so.6
        No symbol table info available.
        #3  0x00000002 in ?? ()
        No symbol table info available.
        #4  0x00000000 in ?? ()
        No symbol table info available.
        (gdb) cont
        Continuing.
       
        Program terminated with signal SIGABRT, Aborted.
        The program no longer exists.
       
Please let me know if there is any more information I can provide.
Thank you.

Will Keaney
--
Build a person a fire, and they may be warm for a day.
Set that person on fire, and they will be warm for the rest of their
life.
GnuPG Public Key ID 1024D/203C7431.  
Public Key Available From http://keyserv.nic-se.se:11371/#extract


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Bill Pringlemeir]

On 18 Feb 2007, keaneyw@... wrote:

> I am using Gentoo Linux, but I get the same error when I compile
> gtk-gnutella directly from the source available on sourceforge.

> Here is my 'emerge --info' (system profile information): Portage
> 2.1.2-r9 (default-linux/x86/2006.1, gcc-4.1.2, glibc-2.5-r0,
> 2.6.19-beyond2 i686)

gcc-4.1.2?  That is five days old.

> gtk-gnutella is configured with dbus, gnutls and nls support.  I am
> also including the output from running GTKG configured with -Oders
> -Dgtkversion=2 -Doptimize='-O0 -g3' under gdb:

> Assertion failure (halloc.c:75) "value"

I have similar systems with GTKG and it is fine.  The stack trace
doesn't look good.  Have you compiled anything else with this GCC?

You can set this before running,

export G_SLICE=always-malloc

That will make Glib use malloc.

Also, you can compile with "-DUSE_MALLOC" to avoid the custom GTKG
allocators.  However, it won't be as fast.

Are you certain the core file you were using is matching the binary
for your core trace?  There were problems in SVN, but them seemed to
have resolved for me (and others) about a month ago.

Perhaps you will get better advice when Europe wakes up ;-).

hth,
Bill Pringlemeir.



-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

Bill Pringlemeir wrote:

> On 18 Feb 2007, keaneyw@... wrote:
>
> > I am using Gentoo Linux, but I get the same error when I compile
> > gtk-gnutella directly from the source available on sourceforge.
>
> > Here is my 'emerge --info' (system profile information): Portage
> > 2.1.2-r9 (default-linux/x86/2006.1, gcc-4.1.2, glibc-2.5-r0,
> > 2.6.19-beyond2 i686)
>
> gcc-4.1.2?  That is five days old.

I've been using that for quite some time.

> > gtk-gnutella is configured with dbus, gnutls and nls support.  I am
> > also including the output from running GTKG configured with -Oders
> > -Dgtkversion=2 -Doptimize='-O0 -g3' under gdb:
 
> > Assertion failure (halloc.c:75) "value"

Typically that would mean something tried to free something that was not
allocated. However, without a valid stacktrace it's pretty difficult to
figure out where this happened.

Modify the build script to use this instead: -Dccflags='-O0 -g'
Then you should be able to get a proper stacktrace with gdb.

--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[uberpinguin]

Some parts of this message have been removed. Learn more about Nabble's security policy.

Oh, and gmail's automagic reply feature fails to catch the fact that I'm trying to reply to a mailing list.  So I'm forwarding this reply back to the list.

---------- Forwarded message ----------
From: William Keaney <keaneyw@...>
Date: Feb 19, 2007 9:19 AM
Subject: Re: [Gtk-gnutella-users] Assertion failure (halloc.c:75) "value" in GTKG-0.96.3 with GTK2
To: Christian Biere <christianbiere@...>

Thanks, Christian and Bill.
Bill: 
I wasn't tracing a core, I was actually running GTKG within GDB.  I did, however, generate a core and examine that as well.  Same results.
Christian:
I rebuilt with the CFLAGS you suggested; the resulting BT is just as verbose as the last one :(
I also read http://www.gentoo.org/proj/en/qa/backtraces.xml and tried building GTKG with -Dccflags='-O1 -ggdb'.  Still no useful BT.

Based on Bill's comment that it works in at least one other system, I am going to go over my own and see if there is anything wrong that might influence this.  Perhaps my GTK2 is broken in some way that has yet to surface, though other GTK2 apps work fine (such as Evolution, XChat, Seamonkey, grdesktop...).

Will
(GPG sig missing 'cause gmail sucks that way, and I can't use Evolution behind my employer's HTTP proxy)

On 2/19/07, Christian Biere <christianbiere@...> wrote:

Bill Pringlemeir wrote:

> On 18 Feb 2007, keaneyw@... wrote:
>
> > I am using Gentoo Linux, but I get the same error when I compile
> > gtk-gnutella directly from the source available on sourceforge.
>
> > Here is my 'emerge --info' (system profile information): Portage
> > 2.1.2-r9 (default-linux/x86/2006.1, gcc-4.1.2, glibc-2.5-r0,
> > 2.6.19-beyond2 i686)
>
> gcc-4.1.2 ?  That is five days old.

I've been using that for quite some time.

> > gtk-gnutella is configured with dbus, gnutls and nls support.  I am
> > also including the output from running GTKG configured with -Oders
> > -Dgtkversion=2 -Doptimize='-O0 -g3' under gdb:

> > Assertion failure (halloc.c:75) "value"

Typically that would mean something tried to free something that was not
allocated. However, without a valid stacktrace it's pretty difficult to
figure out where this happened.

Modify the build script to use this instead: -Dccflags='-O0 -g'
Then you should be able to get a proper stacktrace with gdb.

--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Bill Pringlemeir]

On 19 Feb 2007, keaneyw@... wrote:

> Based on Bill's comment that it works in at least one other system,
> I am going to go over my own and see if there is anything wrong that
> might influence this.  Perhaps my GTK2 is broken in some way that
> has yet to surface, though other GTK2 apps work fine (such as
> Evolution, XChat, Seamonkey, grdesktop...).

Sorry, that is not true.  I just saw x86, and linux 2.6.19-xx.  I have
a similar system, but not Gentoo (and I am rather clueless, I think
you also have a Mac?  I have a no-name PC).

If the GTK2 is suspect, I would try running with,

      export G_SLICE=always-malloc;  gtk-gnutella <whatever...>

There may be some conflict between the GTKG allocator and the GTK2
allocator.  However, I have a newer version of GTK2 and it doesn't seem
to have a problem.

hth,
Bill Pringlemeir.


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

Bill Pringlemeir wrote:
> If the GTK2 is suspect, I would try running with,
>
>       export G_SLICE=always-malloc;  gtk-gnutella <whatever...>
>
> There may be some conflict between the GTKG allocator and the GTK2
> allocator.  However, I have a newer version of GTK2 and it doesn't seem
> to have a problem.

Gtk+ 2.x has a clear interface for overriding the memory allocators. So there
cannot be any voodoo bugs. If there are "conflicts" then it's clearly because
someone accidently uses for example malloc() instead of g_malloc(), a
double-free bug or similar. Without a stacktrace it's hardly possible to
find the culprit though.

--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

William Keaney wrote:
> I wasn't tracing a core, I was actually running GTKG within GDB.  I did,
> however, generate a core and examine that as well.  Same results.  Christian:
> I rebuilt with the CFLAGS you suggested; the resulting BT is just as verbose
> as the last one :( I also read
> http://www.gentoo.org/proj/en/qa/backtraces.xml and tried building GTKG with
> -Dccflags='-O1 -ggdb'.  Still no useful BT.

The problem is probably that the executable is stripped.

 $ file /path/to/gtk-gnutella

can usually tell whether it's stripped or not. Instead of installing it,
just use the executable where it has been compiled. When you invoke gdb
make sure you pass the full path, not just "gtk-gnutella". Otherwise,
it might pick an unstripped gtk-gnutella found in your PATH.


--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[uberpinguin]

On Mon, 2007-02-19 at 19:07 +0100, Christian Biere wrote:
> The problem is probably that the executable is stripped.
>
>  $ file /path/to/gtk-gnutella
>
> can usually tell whether it's stripped or not. Instead of installing it,
> just use the executable where it has been compiled. When you invoke gdb
> make sure you pass the full path, not just "gtk-gnutella". Otherwise,
> it might pick an unstripped gtk-gnutella found in your PATH.
I am running it directly from the build directory, './src/gtk-gnutella'.
        $ file src/gtk-gnutella
        src/gtk-gnutella: ELF 32-bit LSB executable, Intel 80386,
        version 1 (SYSV), for GNU/Linux 2.6.9, dynamically linked (uses
        shared libs), not stripped

On Mon, 2007-02-19 at 12:46 -0500, Bill Pringlemeir wrote:
> Sorry, that is not true.  I just saw x86, and linux 2.6.19-xx.  I have
> a similar system, but not Gentoo (and I am rather clueless, I think
> you also have a Mac?  I have a no-name PC).
No, I am using a p4-based PC (Dell Latitude C840 to be precise).  I
suppose it's hair-splitting and off-topic to note that Macs are PCs
with a different archictecture... ;-)
>
> If the GTK2 is suspect, I would try running with,
>
>       export G_SLICE=always-malloc;  gtk-gnutella <whatever...>
Thanks for the suggestion; it produces the same result :/
> There may be some conflict between the GTKG allocator and the GTK2
> allocator.  However, I have a newer version of GTK2 and it doesn't seem
> to have a problem.
What version of GTK2 do you have?  I am using 2.10.9.

Thanks for your time, both of you.
Will
--
Build a person a fire, and they may be warm for a day.
Set that person on fire, and they will be warm for the rest of their
life.
GnuPG Public Key ID 1024D/203C7431.  
Public Key Available From http://keyserv.nic-se.se:11371/#extract


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

Will Keaney wrote:
> I am running it directly from the build directory, './src/gtk-gnutella'.
>         $ file src/gtk-gnutella
>         src/gtk-gnutella: ELF 32-bit LSB executable, Intel 80386,
>         version 1 (SYSV), for GNU/Linux 2.6.9, dynamically linked (uses
>         shared libs), not stripped

Could you try passing -Ud_fast_assert to Configure and see whether you
get a useful stacktrace then?

--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[uberpinguin]

On Mon, 2007-02-19 at 20:17 +0100, Christian Biere wrote:
> Will Keaney wrote:
> > I am running it directly from the build directory, './src/gtk-gnutella'.
> >         $ file src/gtk-gnutella
> >         src/gtk-gnutella: ELF 32-bit LSB executable, Intel 80386,
> >         version 1 (SYSV), for GNU/Linux 2.6.9, dynamically linked (uses
> >         shared libs), not stripped
>
> Could you try passing -Ud_fast_assert to Configure and see whether you
> get a useful stacktrace then?
Still nothing useful.  I also tried passing -Dldflags="-nopie", but
 that didn't help either.
--
Build a person a fire, and they may be warm for a day.
Set that person on fire, and they will be warm for the rest of their
life.
GnuPG Public Key ID 1024D/203C7431.  
Public Key Available From http://keyserv.nic-se.se:11371/#extract


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

Will Keaney wrote:
> On Mon, 2007-02-19 at 20:17 +0100, Christian Biere wrote:
> > Could you try passing -Ud_fast_assert to Configure and see whether you
> > get a useful stacktrace then?
> Still nothing useful.  I also tried passing -Dldflags="-nopie", but
>  that didn't help either.

If you add -DUSE_MALLOC to your CFLAGS, this code should be disabled and
the crash shouldn't occur.

--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[uberpinguin]

On Mon, 2007-02-19 at 21:39 +0100, Christian Biere wrote:
> Will Keaney wrote:
> > On Mon, 2007-02-19 at 20:17 +0100, Christian Biere wrote:
> > > Could you try passing -Ud_fast_assert to Configure and see whether you
> > > get a useful stacktrace then?
> > Still nothing useful.  I also tried passing -Dldflags="-nopie", but
> >  that didn't help either.
>
> If you add -DUSE_MALLOC to your CFLAGS, this code should be disabled and
> the crash shouldn't occur.
Unfortunately, GTKG segfaults with that flag.  Still no debugging
symbols, either.  I am going to do a system-wide recompile this week
anyway, so I'll just make do with the GTK1 interface until that is
done; at that point I will resume this line of investigation.
Thanks for all your help tracking this down.
--
Build a person a fire, and they may be warm for a day.
Set that person on fire, and they will be warm for the rest of their
life.
GnuPG Public Key ID 1024D/203C7431.  
Public Key Available From http://keyserv.nic-se.se:11371/#extract


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

Will Keaney wrote:
> On Mon, 2007-02-19 at 21:39 +0100, Christian Biere wrote:
> > If you add -DUSE_MALLOC to your CFLAGS, this code should be disabled and
> > the crash shouldn't occur.
> Unfortunately, GTKG segfaults with that flag.

Ok, this means it has nothing to do with our custom allocator. You remove the
-DUSE_MALLOC again and change line 75 of halloc.c to this:

        g_return_if_fail(value);

Then it will emit a warning but not crash. If that works it would be interesting
to see which messages are on the console (or the log) before and after this
warning. That might give a clue which part of the code this issue derives from.

--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[uberpinguin]

On Mon, 2007-02-19 at 22:23 +0100, Christian Biere wrote:

> Will Keaney wrote:
> > On Mon, 2007-02-19 at 21:39 +0100, Christian Biere wrote:
> > > If you add -DUSE_MALLOC to your CFLAGS, this code should be disabled and
> > > the crash shouldn't occur.
> > Unfortunately, GTKG segfaults with that flag.
>
> Ok, this means it has nothing to do with our custom allocator. You remove the
> -DUSE_MALLOC again and change line 75 of halloc.c to this:
>
> g_return_if_fail(value);
>
> Then it will emit a warning but not crash. If that works it would be interesting
> to see which messages are on the console (or the log) before and after this
> warning. That might give a clue which part of the code this issue derives from.

I think we're getting somewhere...

        (gdb) run
        Starting
        program: /home/ubrpngn/build/gtk-gnutella-0.96.3/src/gtk-gnutella
        [Thread debugging using libthread_db enabled]
        [New Thread -1208416592 (LWP 6901)]
        07-02-19 16:33:09 (WARNING): This is an unofficial build which
        accesses files in this directory:
        "/home/ubrpngn/build/gtk-gnutella-0.96.3"
        07-02-19 16:33:09 (MESSAGE): language code: "en"
        07-02-19 16:33:09 (MESSAGE): using locale character set "UTF-8"
        07-02-19 16:33:09 (MESSAGE): primary filename character set
        "UTF-8"
        07-02-19 16:33:09 (MESSAGE): gtk-gnutella/0.96.3 (2006-11-09;
        r12293; GTK2; Linux i686)
        07-02-19 16:33:09 (MESSAGE): D-BUS set up and ready for use.
        07-02-19 16:33:09 (MESSAGE): Sent D-BUS signal 'Events': started
        [New Thread -1213609072 (LWP 6904)]
        [Thread -1213609072 (LWP 6904) exited]
        [New Thread -1222001776 (LWP 6905)]
        [Thread -1222001776 (LWP 6905) exited]
        07-02-19 16:33:09 (WARNING):
        gnutls_certificate_set_x509_key_file() failed: Error while
        reading file.
        07-02-19 16:33:09 (WARNING): [hostile IP addresses (private)]
        unable to retrieve: no alternate locations known
        07-02-19 16:33:09 (WARNING): [Spam database] unable to retrieve:
        no alternate locations known
        07-02-19 16:33:09 (WARNING): [Host Whitelist] unable to
        retrieve: no alternate locations known
       
        Assertion failure (halloc.c:75) "value"
       
        ** ERROR **: file vmm.c: line 262 (free_pages): assertion
        failed: (round_pagesize_fast((size_t) p) == (size_t) p)
        aborting...
       
        Program received signal SIGTRAP, Trace/breakpoint trap.
        [Switching to Thread -1208416592 (LWP 6901)]
        0x47e5a088 in g_logv () from /usr/lib/libglib-2.0.so.0
        (gdb) thread apply all bt full
       
        Thread 1 (Thread -1208416592 (LWP 6901)):
        #0  0x47e5a088 in g_logv () from /usr/lib/libglib-2.0.so.0
        No symbol table info available.
        #1  0x47e5a2f8 in g_log () from /usr/lib/libglib-2.0.so.0
        No symbol table info available.
        #2  0x47e5a36b in g_assert_warning ()
        from /usr/lib/libglib-2.0.so.0
        No symbol table info available.
        #3  0x081e8611 in free_pages (p=0xb7b07690, size=1206609152) at
        vmm.c:262
        n = <value optimized out>
        __PRETTY_FUNCTION__ = "free_pages"
        #4  0x081c1052 in hfree (p=0xb7b07690) at halloc.c:129
        No locals.
        #5  0x47e58ee7 in g_free () from /usr/lib/libglib-2.0.so.0
        No symbol table info available.
        #6  0xb7c6f90a in ?? ()
        from /usr/lib/gtk-2.0/2.10.0/engines/libsmooth.so
        No symbol table info available.
        #7  0xb7b07690 in ?? ()
        No symbol table info available.
        #8  0xb7c5527c in ?? ()
        No symbol table info available.
        #9  0x00000604 in ?? ()
        No symbol table info available.
        ---Type <return> to continue, or q <return> to quit---
        #10 0xbfa91db4 in ?? ()
        No symbol table info available.
        #11 0xb7f99700 in ?? ()
        No symbol table info available.
        #12 0x00000000 in ?? ()
        No symbol table info available.

Will
--
Build a person a fire, and they may be warm for a day.
Set that person on fire, and they will be warm for the rest of their
life.
GnuPG Public Key ID 1024D/203C7431.  
Public Key Available From http://keyserv.nic-se.se:11371/#extract


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

Will Keaney wrote:

> On Mon, 2007-02-19 at 22:23 +0100, Christian Biere wrote:
> > Will Keaney wrote:
> > > On Mon, 2007-02-19 at 21:39 +0100, Christian Biere wrote:
> > > > If you add -DUSE_MALLOC to your CFLAGS, this code should be disabled and
> > > > the crash shouldn't occur.
> > > Unfortunately, GTKG segfaults with that flag.
> >
> > Ok, this means it has nothing to do with our custom allocator. You remove the
> > -DUSE_MALLOC again and change line 75 of halloc.c to this:
> >
> > g_return_if_fail(value);

Ok, I just looked at the code of 0.96.3. Just remove line 75 and instead put similar
into hfree() after line 124:

After this line:
        size = halloc_get_size(p);

Add this:
        g_return_if_fail(size);


--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[uberpinguin]

On Mon, 2007-02-19 at 22:48 +0100, Christian Biere wrote:

> Will Keaney wrote:
> > On Mon, 2007-02-19 at 22:23 +0100, Christian Biere wrote:
> > > Will Keaney wrote:
> > > > On Mon, 2007-02-19 at 21:39 +0100, Christian Biere wrote:
> > > > > If you add -DUSE_MALLOC to your CFLAGS, this code should be disabled and
> > > > > the crash shouldn't occur.
> > > > Unfortunately, GTKG segfaults with that flag.
> > >
> > > Ok, this means it has nothing to do with our custom allocator. You remove the
> > > -DUSE_MALLOC again and change line 75 of halloc.c to this:
> > >
> > > g_return_if_fail(value);
>
> Ok, I just looked at the code of 0.96.3. Just remove line 75 and instead put similar
> into hfree() after line 124:
>
> After this line:
> size = halloc_get_size(p);
>
> Add this:
> g_return_if_fail(size);
>

This gets as far as drawing most of the UI, then segfaults again

(screenshot: http://tinyurl.com/yrbrjj).  The debug output is less
informative this time, though:
        (gdb) run
        Starting
        program: /home/ubrpngn/build/gtk-gnutella-0.96.3/src/gtk-gnutella
        [Thread debugging using libthread_db enabled]
        [New Thread -1208490320 (LWP 7922)]
        07-02-19 18:18:37 (WARNING): This is an unofficial build which
        accesses files in this directory:
        "/home/ubrpngn/build/gtk-gnutella-0.96.3"
        07-02-19 18:18:37 (MESSAGE): language code: "en"
        07-02-19 18:18:37 (MESSAGE): using locale character set "UTF-8"
        07-02-19 18:18:37 (MESSAGE): primary filename character set
        "UTF-8"
        07-02-19 18:18:37 (MESSAGE): gtk-gnutella/0.96.3 (2006-11-09;
        r12293; GTK2; Linux i686)
        07-02-19 18:18:37 (MESSAGE): D-BUS set up and ready for use.
        07-02-19 18:18:37 (MESSAGE): Sent D-BUS signal 'Events': started
        [New Thread -1213584496 (LWP 7925)]
        [Thread -1213584496 (LWP 7925) exited]
        [New Thread -1221977200 (LWP 7926)]
        [Thread -1221977200 (LWP 7926) exited]
        07-02-19 18:18:37 (WARNING):
        gnutls_certificate_set_x509_key_file() failed: Error while
        reading file.
        07-02-19 18:18:37 (WARNING): [hostile IP addresses (private)]
        unable to retrieve: no alternate locations known
        07-02-19 18:18:37 (WARNING): [Spam database] unable to retrieve:
        no alternate locations known
        07-02-19 18:18:37 (WARNING): [Host Whitelist] unable to
        retrieve: no alternate locations known
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
        07-02-19 18:18:37 (WARNING): searches file does not
        exist: /home/ubrpngn/.gtk-gnutella/searches.xml
        07-02-19 18:18:37 (WARNING): retrieving searches
        from /home/ubrpngn/.gtk-gnutella/searches.xml.orig instead
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Assertion failure (halloc.c:124) "size"
       
        Program received signal SIGSEGV, Segmentation fault.
        [Switching to Thread -1208490320 (LWP 7922)]
        0x47ed37b9 in ?? () from /usr/lib/libgobject-2.0.so.0
        (gdb) thread apply all bt full
       
        Thread 1 (Thread -1208490320 (LWP 7922)):
        #0  0x47ed37b9 in ?? () from /usr/lib/libgobject-2.0.so.0
        No symbol table info available.
        #1  0xbf8dff28 in ?? ()
        No symbol table info available.
        #2  0x47ed76bf in ?? () from /usr/lib/libgobject-2.0.so.0
        No symbol table info available.
        #3  0xbf8dff04 in ?? ()
        No symbol table info available.
        #4  0x65da39e0 in ?? ()
        No symbol table info available.
        #5  0xb66491e0 in ?? ()
        No symbol table info available.
        #6  0x47ef56e8 in ?? () from /usr/lib/libgobject-2.0.so.0
        No symbol table info available.
        #7  0x00000000 in ?? ()
        No symbol table info available.

Will
--
Build a person a fire, and they may be warm for a day.
Set that person on fire, and they will be warm for the rest of their
life.
GnuPG Public Key ID 1024D/203C7431.  
Public Key Available From http://keyserv.nic-se.se:11371/#extract


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

Will Keaney wrote:
> This gets as far as drawing most of the UI, then segfaults again
> (screenshot: http://tinyurl.com/yrbrjj).  The debug output is less
> informative this time, though:

Do you use some gtk-engine or gtk-theme-engine? I suspect the culprit is some
kind of extension to Gtk+ (i.e., such an engine or similar) which does not
consistenly use g_malloc() and g_free(). If you use such an engine - your
screenshot looks like it - try to disable it.

Or just to be sure it's not D-BUS, maybe disable it by passing -Ddbus=false to
Configure.

--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[uberpinguin]

On Tue, 2007-02-20 at 00:41 +0100, Christian Biere wrote:
> Will Keaney wrote:
> > This gets as far as drawing most of the UI, then segfaults again
> > (screenshot: http://tinyurl.com/yrbrjj).  The debug output is less
> > informative this time, though:
>
> Do you use some gtk-engine or gtk-theme-engine? I suspect the culprit is some
> kind of extension to Gtk+ (i.e., such an engine or similar) which does not
> consistenly use g_malloc() and g_free(). If you use such an engine - your
> screenshot looks like it - try to disable it.
Well... sort of.  I'm running Enlightenment DR17, but using a
~/.gtkrc-2.0 file to load themes for GTK+ apps.  You are on the right
track, though - I was loading the Nuvola theme; GTKG appears to work
fine with that line commented out of ~/.gtkrc-2.0.  For the sake of
completeness, here's the file in question:
        # -- THEME AUTO-WRITTEN DO NOT EDIT
        include "/usr/share/themes/Nuvola/gtk-2.0/gtkrc"
       
        style "user-font" {
                font_name = "DejaVu Sans 9"
        }
       
        widget_class "*" style "user-font"
       
        gtk-font-name="DejaVu Sans 9"
       
        include "/home/ubrpngn/.gtkrc.mine"
       
        # -- THEME AUTO-WRITTEN DO NOT EDIT
Thanks very much for all your time on this.
--
Build a person a fire, and they may be warm for a day.
Set that person on fire, and they will be warm for the rest of their
life.
GnuPG Public Key ID 1024D/203C7431.  
Public Key Available From http://keyserv.nic-se.se:11371/#extract


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[uberpinguin]

On Tue, 2007-02-20 at 00:41 +0100, Christian Biere wrote:

> Will Keaney wrote:
> > This gets as far as drawing most of the UI, then segfaults again
> > (screenshot: http://tinyurl.com/yrbrjj).  The debug output is less
> > informative this time, though:
>
> Do you use some gtk-engine or gtk-theme-engine? I suspect the culprit is some
> kind of extension to Gtk+ (i.e., such an engine or similar) which does not
> consistenly use g_malloc() and g_free(). If you use such an engine - your
> screenshot looks like it - try to disable it.
>
> Or just to be sure it's not D-BUS, maybe disable it by passing -Ddbus=false to
> Configure.

Another follow-up: I did some additional testing, and it appears that

GTK2 themes based on SmoothGNOME all cause GTKG to fail with this
halloc error.  This includes SmoothGNOME itself.  GTKG works fine with
other themes, such as ClearLooks or ThinIce.
--
Build a person a fire, and they may be warm for a day.
Set that person on fire, and they will be warm for the rest of their
life.
GnuPG Public Key ID 1024D/203C7431.  
Public Key Available From http://keyserv.nic-se.se:11371/#extract


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users

[Christian Biere]

Will Keaney wrote:
> Another follow-up: I did some additional testing, and it appears that
> GTK2 themes based on SmoothGNOME all cause GTKG to fail with this
> halloc error.  This includes SmoothGNOME itself.  GTKG works fine with
> other themes, such as ClearLooks or ThinIce.

It's best if you reported this at http://bugzilla.gnome.org as bug in
gtk-engines because it's not a bug in gtk-gnutella, so there's little we can do
except documenting it.

--
Christian

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-users mailing list
Gtk-gnutella-users@...
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-users