[arch-multilib] Memory access error with gtk
Hello all, A recent update brought a weird error with the standard gtk filerequester. When I try to use the scrollbars or scroll down in the requester with the mouse wheel the complete app (Real Studio 2011 R4.1) simply dies with a memory access error. No further error messages, the app just vanishes from the sreen. All other functionality of the requester is o.k. This has been introduced about a month ago. Downgrading lib32-gtk2 to an earlier version does not work as there are dependencies for libcups that need lib32-gtk2 in version 2.24. Any clues? Or at least some tips how I can debug this further? FWIW: In an archlinux 32 bit installation the app is just working... Harvey -- I am root. If you see me laughing, you'd better have a backup!
On 06.02.2012 17:12, Harvey wrote:
Hello all,
A recent update brought a weird error with the standard gtk filerequester. When I try to use the scrollbars or scroll down in the requester with the mouse wheel the complete app (Real Studio 2011 R4.1) simply dies with a memory access error. No further error messages, the app just vanishes from the sreen. All other functionality of the requester is o.k. This has been introduced about a month ago. Downgrading lib32-gtk2 to an earlier version does not work as there are dependencies for libcups that need lib32-gtk2 in version 2.24.
Any clues? Or at least some tips how I can debug this further?
You could try to get a backtrace: https://wiki.archlinux.org/index.php/Debug_-_Getting_Traces -- Florian Pritz
Florian,
You could try to get a backtrace: https://wiki.archlinux.org/index.php/Debug_-_Getting_Traces
Well, I did, but this only shows that the segfault is coming from gtk_dialog_run () from /usr/lib32/libgtk-x11-2.0.so.0 which belongs to lib32-gtk2. Full trace is here: http://pastebin.com/ank7PXPi This has been done with gdb. Do I get more useful information when recompiling lib32-gtk2 with -g, -O0 and -fbuiltin flags enabled or is this just a waste of time? Harvey -- I am root. If you see me laughing, you'd better have a backup!
On 07.02.2012 08:54, Harvey wrote:
Florian,
You could try to get a backtrace: https://wiki.archlinux.org/index.php/Debug_-_Getting_Traces
Well, I did, but this only shows that the segfault is coming from gtk_dialog_run () from /usr/lib32/libgtk-x11-2.0.so.0 which belongs to lib32-gtk2. Full trace is here: http://pastebin.com/ank7PXPi
This has been done with gdb. Do I get more useful information when recompiling lib32-gtk2 with -g, -O0 and -fbuiltin flags enabled or is this just a waste of time?
Do that and you will also see variable values and source code files/lines. -- Florian Pritz
Florian,
You could try to get a backtrace: https://wiki.archlinux.org/index.php/Debug_-_Getting_Traces
Well, I did, but this only shows that the segfault is coming from gtk_dialog_run () from /usr/lib32/libgtk-x11-2.0.so.0 which belongs to lib32-gtk2. Full trace is here: http://pastebin.com/ank7PXPi
This has been done with gdb. Do I get more useful information when recompiling lib32-gtk2 with -g, -O0 and -fbuiltin flags enabled or is this just a waste of time?
Do that and you will also see variable values and source code files/lines.
The package lib32-gtk2 doesn't compile. I only added '!strip' in the options and export CFLAGS="$CFLAGS -O0 -fbuiltin -g" export CXXFLAGS="$CXXFLAGS -O0 -fbuiltin -g" as first entries in the build() function of the PKGBUILD and get this error: t.c x11/xsettings-common.c libgdk-x11-2.0.la Makefile --output Gdk-2.0.gir Traceback (most recent call last): File "/usr/bin/g-ir-scanner", line 46, in <module> sys.exit(scanner_main(sys.argv)) File "/usr/lib/gobject-introspection/giscanner/scannermain.py", line 407, in scanner_main ss = create_source_scanner(options, args) File "/usr/lib/gobject-introspection/giscanner/scannermain.py", line 332, in create_source_scanner ss.parse_files(filenames) File "/usr/lib/gobject-introspection/giscanner/sourcescanner.py", line 249, in parse_files self._parse(headers) File "/usr/lib/gobject-introspection/giscanner/sourcescanner.py", line 282, in _parse stdout=subprocess.PIPE) File "/usr/lib/python2.7/subprocess.py", line 679, in __init__ errread, errwrite) File "/usr/lib/python2.7/subprocess.py", line 1228, in _execute_child raise child_exception OSError: [Errno 2] No such file or directory make[4]: *** [Gdk-2.0.gir] Fehler 1 make[4]: Leaving directory `/home/harvey/abs/local/lib32-gtk2/src/gtk+-2.24.9/gdk' make[3]: *** [all-recursive] Fehler 1 make[3]: Leaving directory `/home/harvey/abs/local/lib32-gtk2/src/gtk+-2.24.9/gdk' make[2]: *** [all] Fehler 2 make[2]: Leaving directory `/home/harvey/abs/local/lib32-gtk2/src/gtk+-2.24.9/gdk' make[1]: *** [all-recursive] Fehler 1 make[1]: Leaving directory `/home/harvey/abs/local/lib32-gtk2/src/gtk+-2.24.9' make: *** [all] Fehler 2 Edit: Even the 'normal' compile without changes to the PKGBUILD doesn't work :/ Harvey -- I am root. If you see me laughing, you'd better have a backup!
On 07.02.2012 11:54, Harvey wrote:
Florian,
You could try to get a backtrace: https://wiki.archlinux.org/index.php/Debug_-_Getting_Traces
Well, I did, but this only shows that the segfault is coming from gtk_dialog_run () from /usr/lib32/libgtk-x11-2.0.so.0 which belongs to lib32-gtk2. Full trace is here: http://pastebin.com/ank7PXPi
This has been done with gdb. Do I get more useful information when recompiling lib32-gtk2 with -g, -O0 and -fbuiltin flags enabled or is this just a waste of time?
Do that and you will also see variable values and source code files/lines.
The package lib32-gtk2 doesn't compile. I only added '!strip' in the options and
export CFLAGS="$CFLAGS -O0 -fbuiltin -g" export CXXFLAGS="$CXXFLAGS -O0 -fbuiltin -g"
as first entries in the build() function of the PKGBUILD and get this error: [...] Edit: Even the 'normal' compile without changes to the PKGBUILD doesn't work :/
Works fine in my chroot, maybe your system/chroot is broken. Here's the package I built (hopefully with debug symbols): http://paste.xinu.at/WrRiW/ -- Florian Pritz
Florian,
Works fine in my chroot, maybe your system/chroot is broken.
Here's the package I built (hopefully with debug symbols): http://paste.xinu.at/WrRiW/
I installed your package and guess what: this one does not segfault, the file requester works as it should. Weird. Too much stripped in the official package? Harvey -- I am root. If you see me laughing, you'd better have a backup!
On 07.02.2012 13:25, Harvey wrote:
Florian,
Works fine in my chroot, maybe your system/chroot is broken.
Here's the package I built (hopefully with debug symbols): http://paste.xinu.at/WrRiW/
I installed your package and guess what: this one does not segfault, the file requester works as it should. Weird.
Too much stripped in the official package?
Can you try if it is fixed in lib32-gtk2 2.24.10-1? (should on the mirrors soon) -- Florian Pritz
Florian,
Can you try if it is fixed in lib32-gtk2 2.24.10-1? (should on the mirrors soon)
No, it isn't. Still segfaults with lib32-gtk2 2.24.10-1. There must be something wrong with the official package. Harvey -- I am root. If you see me laughing, you'd better have a backup!
On 08.02.2012 08:46, Harvey wrote:
Florian,
Can you try if it is fixed in lib32-gtk2 2.24.10-1? (should on the mirrors soon)
No, it isn't. Still segfaults with lib32-gtk2 2.24.10-1. There must be something wrong with the official package.
I've rebuilt the current pkgbuild again without changing anything. Please test: http://paste.xinu.at/HSF/ -- Florian Pritz
Am 10.02.2012 19:56, schrieb Florian Pritz:
Can you try if it is fixed in lib32-gtk2 2.24.10-1? (should on the mirrors soon)
No, it isn't. Still segfaults with lib32-gtk2 2.24.10-1. There must be something wrong with the official package.
I've rebuilt the current pkgbuild again without changing anything. Please test: http://paste.xinu.at/HSF/
This one segfaults too :( So it has to be in the compiler flags, I guess. Stripping debug symbols seems to be unsuspicious, right? Maybe you built a test version with CXXFLAGS and CFLAGS -O0 -fbuiltin -g for me? Harvey -- I am root. If you see me laughing, you'd better have a backup!
On 10.02.2012 21:45, Harvey wrote:
Am 10.02.2012 19:56, schrieb Florian Pritz:
Can you try if it is fixed in lib32-gtk2 2.24.10-1? (should on the mirrors soon)
No, it isn't. Still segfaults with lib32-gtk2 2.24.10-1. There must be something wrong with the official package.
I've rebuilt the current pkgbuild again without changing anything. Please test: http://paste.xinu.at/HSF/
This one segfaults too :(
So it has to be in the compiler flags, I guess. Stripping debug symbols seems to be unsuspicious, right? Maybe you built a test version with CXXFLAGS and CFLAGS -O0 -fbuiltin -g for me?
http://paste.xinu.at/DHAZ/ -- Florian Pritz
Florian,
Can you try if it is fixed in lib32-gtk2 2.24.10-1? (should on the mirrors soon)
No, it isn't. Still segfaults with lib32-gtk2 2.24.10-1. There must be something wrong with the official package.
I've rebuilt the current pkgbuild again without changing anything. Please test: http://paste.xinu.at/HSF/
This one segfaults too :(
So it has to be in the compiler flags, I guess. Stripping debug symbols seems to be unsuspicious, right? Maybe you built a test version with CXXFLAGS and CFLAGS -O0 -fbuiltin -g for me?
This one is good. No segfault. So the compiler flags are to blame. Too much optimization? Harvey -- I am root. If you see me laughing, you'd better have a backup!
On 11.02.2012 13:55, Harvey wrote:
Florian,
Can you try if it is fixed in lib32-gtk2 2.24.10-1? (should on the mirrors soon)
No, it isn't. Still segfaults with lib32-gtk2 2.24.10-1. There must be something wrong with the official package.
I've rebuilt the current pkgbuild again without changing anything. Please test: http://paste.xinu.at/HSF/
This one segfaults too :(
So it has to be in the compiler flags, I guess. Stripping debug symbols seems to be unsuspicious, right? Maybe you built a test version with CXXFLAGS and CFLAGS -O0 -fbuiltin -g for me?
This one is good. No segfault. So the compiler flags are to blame. Too much optimization?
This one is stripped and built with -O0, but without -fbuilting -g. Whatever the outcome, please open a bug report about this. http://paste.xinu.at/6bn/ -- Florian Pritz
Florian,
This one is stripped and built with -O0, but without -fbuilting -g. Whatever the outcome, please open a bug report about this.
Still no segfaults. I opened a bug report: https://bugs.archlinux.org/task/28429 Harvey -- I am root. If you see me laughing, you'd better have a backup!
participants (2)
-
Florian Pritz
-
Harvey