Please keep [EMAIL PROTECTED] in the cc; this information should be recorded in the bug report.
Brent S. Elmer wrote: > I downloaded qt-x11-free-dbg and ran your debug version of > fireglcontrol. I didn't download the source to debug but here is the > gdb info anyway: As far as I can tell, it's Qt that's causing the exception, but to be on the safe side it's probably a good idea to do some more tests before we reassign the bug. First of all, the last file accessed in the strace was /home/brente/.qt/qtrc: try moving it out of the way and see it things get better. > [EMAIL PROTECTED]:~/firegl$ gdb fireglcontrol > GNU gdb 6.4.90-debian > 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 "i486-linux-gnu"...Using host libthread_db > library "/ lib/tls/i686/cmov/libthread_db.so.1". > > (gdb) r > Starting program: /home/brente/firegl/fireglcontrol > [Thread debugging using libthread_db enabled] > [New Thread -1222313760 (LWP 19558)] > Qt: gdb: -nograb added to command-line options. > Use the -dograb option to enforce grabbing. > > Program received signal SIGFPE, Arithmetic exception. > [Switching to Thread -1222313760 (LWP 19558)] > 0xb79a8bed in create_dpis () at kernel/qpaintdevice_x11.cpp:531 > 531 kernel/qpaintdevice_x11.cpp: No such file or directory. > in kernel/qpaintdevice_x11.cpp > (gdb) bt > #0 0xb79a8bed in create_dpis () at kernel/qpaintdevice_x11.cpp:531 > #1 0xb79a8cbb in QPaintDevice::x11AppDpiY (screen=-1) > at kernel/qpaintdevice_x11.cpp:653 > #2 0xb79a8d38 in QPaintDevice::x11AppDpiY () > at kernel/qpaintdevice_x11.cpp:675 > #3 0xb797f7ef in qt_init_internal (argcptr=0xbfaa7ac0, argv=0xbfaa7b34, > display=0x0, visual=0, colormap=0) at kernel/qapplication_x11.cpp:2162 > #4 0xb79802ce in qt_init (argcptr=0xbfaa7ac0, argv=0xbfaa7b34) > at kernel/qapplication_x11.cpp:2385 > #5 0xb79f822c in QApplication::construct (this=0xbfaa79f0, > [EMAIL PROTECTED], > argv=0xbfaa7b34, type=QApplication::GuiClient) > at kernel/qapplication.cpp:813 > #6 0xb79f85dc in QApplication (this=0xbfaa79f0, [EMAIL PROTECTED], > argv=0xbfaa7b34) at kernel/qapplication.cpp:728 > #7 0x08055561 in main (argc=1, argv=0x8087480) at main.cpp:32 > > Brent > > On Tue, 2006-09-26 at 00:09 +0200, Flavio Stanchina wrote: >> Brent S. Elmer wrote: >> > The kernel seems to be fine. Here is an strace -f for fireglcontrol. >> >> It looks like fireglcontrol breaks after reading qtrc, so I'd suspect a >> problem there. However, it's definitely not conclusive. >> >> I'm attaching a fireglcontrol binary compiled with debug info: please run >> it from gdb and tell me how it goes. You might also want to install >> qt-x11-free-dbg (careful, it's a 32 MB download) to get useful stack traces >> in the Qt code, if the problem turns out to be in there. >> Brent S. Elmer wrote: > In the strace I see a failure soon after fstat64 which is for large file > systems. I had CONFIG_LSF turned on in my kernel. I rebuilt a kernel > with it turned off to see if it helped. I still get the floating point > exception and I still see fstat64 in the strace. Is there some other > kernel config option for large file systems that I can turn off? I have > another qt application that is giving a floating point exception during > startup. I see fstat64 and stat64 in the strace of it also. Open > office .org also gives a floating point exception with stat64 and > fstat64 references in the strace. I am running an up to date Debian > etch system with a 2.6.17 kernel. I believe that fstat64 just happens to be the last system call before those programs cause the exception, and anyway, CONFIG_LSF has nothing to do with fstat64 and friends. The kernel supports files larger than 4GB (the 32-bit limit) even without CONFIG_LSF; that's only for files even larger, more than 2TB (i.e. 512 times more). That's even more evidence that the problem is not fireglconfig per se, however. -- Ciao, Flavio -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
