Hello,
I was wondering if anybody is having the same problem. On a workstation install mc keeps segfaulting on any command that I issue. This is in a tty, not in an Xterm or konsole.
Funny thing is this only happens for root and leonard(uid 500), but not for users test(501) and test1(502).
Any suggestions on how to debug this issue? Maybe it is just a corrupt file somewhere but where should I look?
Bye, Leonard.
-- How clean is a war when you shoot around nukelar waste? Stop the use of depleted uranium ammo! End all weapons of mass destruction.
On Wed, Jul 30, 2003 at 08:22:26PM +0200, Leonard den Ottolander wrote:
Hello,
I was wondering if anybody is having the same problem. On a workstation install mc keeps segfaulting on any command that I issue. This is in a tty, not in an Xterm or konsole.
Funny thing is this only happens for root and leonard(uid 500), but not for users test(501) and test1(502).
Any suggestions on how to debug this issue? Maybe it is just a corrupt file somewhere but where should I look?
rm -rf ~/.mc
??? Try to just rename to some other name.
Hello Milan,
rm -rf ~/.mc
??? Try to just rename to some other name.
No, that doesn't solve the issue. After mv ~/.mc ~/.mc.000 mc in a tty still segfaults when running any command as root or leonard.
Bye, Leonard.
-- How clean is a war when you shoot around nukelar waste? Stop the use of depleted uranium ammo! End all weapons of mass destruction.
I would say that you should try the following:
1) Add a new user so it has default settings in all . directories 2) run mc and see if it segfaults. 3) run strace on mc with the output to a file and see if it still segfaults and where. 4) Open a bugzilla and send that info in.
On Wed, 2003-07-30 at 14:52, Leonard den Ottolander wrote:
Hello Milan,
rm -rf ~/.mc
??? Try to just rename to some other name.
No, that doesn't solve the issue. After mv ~/.mc ~/.mc.000 mc in a tty still segfaults when running any command as root or leonard.
Bye, Leonard.
-- How clean is a war when you shoot around nukelar waste? Stop the use of depleted uranium ammo! End all weapons of mass destruction.
-- Rhl-beta-list mailing list Rhl-beta-list@redhat.com http://www.redhat.com/mailman/listinfo/rhl-beta-list
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 30 Jul 2003 15:04:46 -0600, Stephen Smoogen wrote:
I would say that you should try the following:
- Add a new user so it has default settings in all . directories
- run mc and see if it segfaults.
- run strace on mc with the output to a file and see if it still
segfaults and where. 4) Open a bugzilla and send that info in.
2.1) Disable service gpm and try again. 2.2) Verify the mc package or re-install it. 2.3) Verify some of the libraries used by mc.
- --
Stephen Smoogen wrote:
I would say that you should try the following:
- Add a new user so it has default settings in all . directories
- run mc and see if it segfaults.
- run strace on mc with the output to a file and see if it still segfaults and where.
- Open a bugzilla and send that info in.
rawhide exibit the same behavior.
The behavior describe seen the same. In a xterm window, mc does not segfault.
At a console, for root and for user mc behave correctly until ctrl-o, - the first ctrl-o give the console and behave - the second ctrl-o to switch back to mc panel produce the segfault (gdb) bt #0 0x40287673 in strlen () from /lib/libc.so.6 #1 0x4019335e in Gpm_Open () from /usr/lib/libgpm.so.1
$ rpm -q gpm gpm-1.20.1-35
On Wed, Jul 30, 2003 at 06:30:42PM -0400, Gilles J. Seguin wrote:
(gdb) bt #0 0x40287673 in strlen () from /lib/libc.so.6 #1 0x4019335e in Gpm_Open () from /usr/lib/libgpm.so.1
See https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=101104 Mirek
Hi Mirek,
#1 0x4019335e in Gpm_Open () from /usr/lib/libgpm.so.1
See https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=101104
Although I could not reproduce the crash in vi reported here upgrading gpm to 1.20.1-36 seems to have solved the problem I had with mc. Thanks.
Bye, Leonard.
-- How clean is a war when you shoot around nukelar waste? Stop the use of depleted uranium ammo! End all weapons of mass destruction.
Leonard den Ottolander wrote:
Hi Mirek,
#1 0x4019335e in Gpm_Open () from /usr/lib/libgpm.so.1
See https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=101104
Although I could not reproduce the crash in vi reported here upgrading gpm to 1.20.1-36 seems to have solved the problem I had with mc. Thanks.
Yes, segfault is resolve with this upgrade. mc is displaying error messages at the console, before writing the panel over it.
Do we have a simple way to send the result somewhere so I can read what it is .
Hi,
On first sight upgrading gpm to 1.20.1-36 seemed to solve my mc crashes, but as it turns out it didn't. Today again I experienced crashes in mc as user leonard (currently not reproducible) and as root (still reproducable) that dissappeared when stopping gpm. Obviously the patch didn't solve the entire problem.
I do have core files, but no experience with analyzing these. Should I submit them? Or can you give me some clues on where and how to look?
Bye, Leonard.
-- How clean is a war when you shoot around nukelar waste? Stop the use of depleted uranium ammo! End all weapons of mass destruction.
Leonard den Ottolander wrote:
Hi,
On first sight upgrading gpm to 1.20.1-36 seemed to solve my mc crashes, but as it turns out it didn't.
Idem
Today again I experienced crashes in mc as user leonard (currently not reproducible) and as root (still reproducable) that dissappeared when stopping gpm.
Idem
Obviously the patch didn't solve the entire problem.
I do have core files, but no experience with analyzing these. Should I submit them? Or can you give me some clues on where and how to look?
Allow core file to be saved # ulimit -c 99999; # unit are in KB ... # gdb mc core.1234 (gdb) bt #0 0x40287673 in strlen () from /lib/libc.so.6 #1 0x401933e2 in Gpm_Open () from /usr/lib/libgpm.so.1 #2 0x4032f91d in __libc_ptyname2 () from /lib/libc.so.6
The back trace is different, the behavior from working again to segfault is puzzling, at best.
The backtrace being different let me beleive that it is trigger from a different case.
I filed the bug for the crash in vim. I haven't had a chance to test gpm-1.20.1-36 yet, but looking at the code it seems the patch was not enough. A few lines after where it crashed before there is another strlen(option.consolename) (option.consolename is the null pointer that was dereferenced for the previous crash too). Based on my reading of the gpm documentation as far as how the Gpm_Open function is supposed to operate I have a proper solution in mind. I'll reopen the bug and see if my idea works and attach the patch to the bug. If you add yourself to the CC list on the bug (https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=101104) it'll mail you when I post the patch.
- Geoff Reedy
Hello Gilles,
# gdb mc core.1234
Aha. That's the way to load the symbols. Should have probably looked that up in the man page, but thanks.
(gdb) bt #0 0x40287673 in strlen () from /lib/libc.so.6 #1 0x401933e2 in Gpm_Open () from /usr/lib/libgpm.so.1 #2 0x4032f91d in __libc_ptyname2 () from /lib/libc.so.6
The back trace is different, the behavior from working again to segfault is puzzling, at best.
Yes. I didn't do a lot of testing, but I was quite surprised to see the segfaults back. By the way, they are now reproducible for user leonard again as well.
The backtrace being different let me beleive that it is trigger from a different case.
Mine seem to be similar to the original posting (some slightly different, shorter): #0 0x00ef7173 in strlen () from /lib/tls/libc.so.6 #1 0x002c73e2 in Gpm_Open () from /usr/lib/libgpm.so.1 #2 0x08072dde in strcpy () #3 0x0807eefb in getch () #4 0x0806e959 in strcpy () ... #15 0x08071595 in strcpy () #16 0x00e95678 in __libc_start_main () from /lib/tls/libc.so.6
I requested bug 100836 to be reopened, and just received this:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=100836
------- Additional Comments From aoliva@redhat.com 2003-08-01 17:48 --- ---- It sure solves the problem that I reported. If you still have problems, please file a new bug report describing how to duplicate it.
Answered that I am still seeing the exact problem that he was(?) having.
Bye, Leonard.
-- How clean is a war when you shoot around nukelar waste? Stop the use of depleted uranium ammo! End all weapons of mass destruction.
-
Geoff Reedy -
Gilles J. Seguin -
Leonard den Ottolander -
Michael Schwendt -
Milan Kerslager -
Miloslav Trmač -
Stephen Smoogen