Bug 7141 - Samba server crash (unsure of reason, includes backtrace)
Summary: Samba server crash (unsure of reason, includes backtrace)
Status: RESOLVED FIXED
Alias: None
Product: Samba 4.0
Classification: Unclassified
Component: Other (show other bugs)
Version: unspecified
Hardware: x86 Linux
: P3 critical (vote)
Target Milestone: ---
Assignee: Andrew Bartlett
QA Contact: samba4-qa@samba.org
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-02-16 19:03 UTC by James Rhodes
Modified: 2010-02-26 10:33 UTC (History)
1 user (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description James Rhodes 2010-02-16 19:03:02 UTC
The Samba server crashed (unsure of the reason).  Here is the backtrace found in the logs:

[Tue Feb 16 22:46:36 2010 EST, 0 lib/cmdline/popt_common.c:58:popt_s4_talloc_log_fn()]
Bad talloc magic value - double free
[Tue Feb 16 22:46:36 2010 EST, 0 ../lib/util/fault.c:143:smb_panic()]
PANIC: Bad talloc magic value - double free
[Tue Feb 16 22:46:36 2010 EST, 0 ../lib/util/fault.c:62:call_backtrace()]
BACKTRACE: 47 stack frames:
 #0 /usr/local/samba/sbin/samba(call_backtrace+0x2b) [0x8a1e9ff]
 #1 /usr/local/samba/sbin/samba(smb_panic+0x226) [0x8a1ecd9]
 #2 /usr/local/samba/sbin/samba [0x8a37ad7]
 #3 /usr/local/samba/sbin/samba [0x8a37b80]
 #4 /usr/local/samba/sbin/samba [0x8a37c8e]
 #5 /usr/local/samba/sbin/samba(_talloc_reference_loc+0x37) [0x8a3812c]
 #6 /usr/local/samba/sbin/samba(ndr_pull_init_blob+0x7d) [0x8a04a5b]
 #7 /usr/local/samba/sbin/samba(ndr_pull_struct_blob+0x2b) [0x8a06755]
 #8 /usr/local/samba/sbin/samba [0x83ee649]
 #9 /usr/local/samba/sbin/samba(notify_trigger+0x2e) [0x83ef6a3]
 #10 /usr/local/samba/sbin/samba [0x83d279d]
 #11 /usr/local/samba/sbin/samba [0x8a38959]
 #12 /usr/local/samba/sbin/samba [0x8a38ae1]
 #13 /usr/local/samba/sbin/samba [0x8a38ae1]
 #14 /usr/local/samba/sbin/samba [0x8a38ae1]
 #15 /usr/local/samba/sbin/samba(_talloc_free+0xad) [0x8a392a7]
 #16 /usr/local/samba/sbin/samba [0x839a2a6]
 #17 /usr/local/samba/sbin/samba [0x8a38959]
 #18 /usr/local/samba/sbin/samba [0x8a38ae1]
 #19 /usr/local/samba/sbin/samba [0x8a38ae1]
 #20 /usr/local/samba/sbin/samba [0x8a38ae1]
 #21 /usr/local/samba/sbin/samba [0x8a38ae1]
 #22 /usr/local/samba/sbin/samba(_talloc_free+0xad) [0x8a392a7]
 #23 /usr/local/samba/sbin/samba [0x8a14f74]
 #24 /usr/local/samba/sbin/samba(stream_terminate_connection+0xe4) [0x80fd84c]
 #25 /usr/local/samba/sbin/samba [0x80fd900]
 #26 /usr/local/samba/sbin/samba [0x80fd94a]
 #27 /usr/local/samba/sbin/samba [0x8a3edd8]
 #28 /usr/local/samba/sbin/samba [0x8a3f445]
 #29 /usr/local/samba/sbin/samba(_tevent_loop_once+0xdf) [0x8a3b70a]
 #30 /usr/local/samba/sbin/samba(tevent_common_loop_wait+0x26) [0x8a3b92b]
 #31 /usr/local/samba/sbin/samba(_tevent_loop_wait+0x1d) [0x8a3b9e9]
 #32 /usr/local/samba/sbin/samba [0x8a14d47]
 #33 /usr/local/samba/sbin/samba [0x80fdfa7]
 #34 /usr/local/samba/sbin/samba [0x8a3edd8]
 #35 /usr/local/samba/sbin/samba [0x8a3f445]
 #36 /usr/local/samba/sbin/samba(_tevent_loop_once+0xdf) [0x8a3b70a]
 #37 /usr/local/samba/sbin/samba(tevent_common_loop_wait+0x26) [0x8a3b92b]
 #38 /usr/local/samba/sbin/samba(_tevent_loop_wait+0x1d) [0x8a3b9e9]
 #39 /usr/local/samba/sbin/samba [0x8a14ed8]
 #40 /usr/local/samba/sbin/samba(task_server_startup+0x89) [0x80ffc98]
 #41 /usr/local/samba/sbin/samba [0x80fd588]
 #42 /usr/local/samba/sbin/samba(server_service_startup+0x109) [0x80fd6b7]
 #43 /usr/local/samba/sbin/samba [0x80f711b]
 #44 /usr/local/samba/sbin/samba(main+0x38) [0x80f7207]
 #45 /lib/libc.so.6(__libc_start_main+0xe5) [0xb7a90705]
 #46 /usr/local/samba/sbin/samba [0x80f60a1]
Comment 1 Matthias Dieter Wallnöfer 2010-02-17 12:37:34 UTC
Is the crash deterministic reproducible? It's possible for you to either run the samba daemon under valgrind and provide a logfile? Or to recompile all with "./configure.developer" and run it under gdb (when the crash happens type "bt" or "bt full" (better) and provide us the output)?
Comment 2 Matthias Dieter Wallnöfer 2010-02-17 12:38:32 UTC
tridge, would be nice if you could help us with this.
Comment 3 James Rhodes 2010-02-17 18:37:58 UTC
No there was no reason for the crash, it occurred (I believe) while someone was attempting to use Offline Files Synchronization, but I can't be sure.  If it's any help, this particular Offline Files Synchronization contained warnings about certain file types not being able to be made offline.
Comment 4 Matthias Dieter Wallnöfer 2010-02-18 06:55:31 UTC
James, is your samba binary update? We fixed a similar crash not so long ago (see bug #7053). So if you're s4 deployment is a bit older please update first and retry that what could have caused the crash.
Comment 5 James Rhodes 2010-02-18 07:03:03 UTC
I installed it about 4-5 days ago, from the GIT repository (I presume installing from GIT gets the most up-to-date version).

Also, is there a way to maintain all of the settings during an upgrade?  If I have to upgrade, I don't want to lose all of the settings/users/file permissions.
Comment 6 Matthias Dieter Wallnöfer 2010-02-18 07:55:41 UTC
Okay - that release seems to be recent enough.

First run a "make uninstall install" after the build. If the directory (AD) structure changed (new schema, new system objects...) or new configuration files were added you don't need to run a full provision.
Do simply use the the "upgradeprovision" script instead. It is located under "scripting/bin" in the s4 source tree. It's suggested to use the "--full" option if possible (this does more checks and updates - without it only the most essential changes are applied).
Comment 7 Andrew Bartlett 2010-02-22 05:00:27 UTC
Just to note that upgradeprovision isn't normally required, but if it won't work after the upgrade, that's where to start looking. 

Grabbing the backtrace from a developer build would be the most helpful thing at this point.  I need to know the calling functions in more detail.
Comment 8 James Rhodes 2010-02-22 05:35:14 UTC
Problem is, even if I did rebuild it as a developer build, I wouldn't even know how to start reproducing the bug - I wasn't there when it occurred and I didn't find out until the next day.
Comment 9 Matthias Dieter Wallnöfer 2010-02-25 05:20:46 UTC
Additional question: when exactly did you check out this failing samba release? Do you have the GIT revision number somewhere? Was it before Feb 13th?
Comment 10 Matthias Dieter Wallnöfer 2010-02-26 10:33:42 UTC
Let us make it like this: since it's very likely that this problem has been fixed already (I discussed this with my team colleaugue tridge) I mark this as "FIXED". If you face this problem again, please REOPEN this bug report.