The Samba-Bugzilla – Bug 220
smbd reloads smb.conf changes -- then segfaults
Last modified: 2005-08-24 10:26:14 UTC
Whenever I make a change to the smb.conf on my 3.0.0beta1 PDC/fileserver, the
processes corresponding to open client connections begin to segfault. This
tends to be a problem when a user has a locking-sensitive application in use at
Example gdb backtrace (no debugging symbols, unfortunately):
#0 0x401e6b89 in wait4 () from /lib/libc.so.6
#1 0x4025a000 in sys_sigabbrev () from /lib/libc.so.6
#2 0x40189712 in strtold_l () from /lib/libc.so.6
#3 0x40189815 in system () from /lib/libc.so.6
#4 0x0817c5cd in smb_panic ()
#5 0x0816d6d2 in dbgtext ()
#6 0x0816d725 in dbgtext ()
#7 0x401739d8 in sigaction () from /lib/libc.so.6
#8 0x08181b7b in alloc_sub_basic ()
#9 0x0807686b in lp_talloc_free ()
#10 0x0807779f in lp_servicename ()
#11 0x080b87ed in make_connection ()
#12 0x08093df3 in reply_tcon_and_X ()
#13 0x080b5c2e in respond_to_all_remaining_local_messages ()
#14 0x080b5cce in respond_to_all_remaining_local_messages ()
#15 0x080b5f73 in process_smb ()
#16 0x080b68e2 in smbd_process ()
#17 0x081d21bf in main ()
#18 0x40162a51 in __libc_start_main () from /lib/libc.so.6
I could not reproduce it here, I think we need more info. For example your
smb.conf and the exact changes you have made.
Feel free to reopen this bug with more info, please don't feel offended by the
Any change at all causes the problem -- even a change to the debug level. I'll
have to check over the smb.conf on this machine and sanitize it before
forwarding it on.
Some factors which might help explain why it's not easily reproducible -- so
far, I've only had the problem with 3.0.0beta1, and haven't been able to upgrade
yet to beta2, so it may be fixed; the segfaults are due to clients connected at
the time the smb.conf is reloaded; and the number of segfaults does *not*
correspond to the number of clients actually connected at the time, it's always
All of which contributes to my not having set up a reproducible test case yet.
I noticed after filing the bug that one of the reports on the Debian packages,
http://bugs.debian.org/180711, seems to correspond to this bug as well -- so
it's not just me... :)
This has either ben fixed or is _really_ hard to reproduce. In either
case, it doesn't look like a P3. Steve said it would be a couple
of weeks before he could try out a new snapshot on the problematic server.
Sorry Steve. Beta1 is really old and we've seen no signs
of reproducing this on newer releases.
originally reported against 3.0.0beta1. CLeaning out
non-production release versions.
sorry for the same, cleaning up the database to prevent unecessary reopens of bugs.