[Date Prev][Date Next]
Re: (ITS#3796) slapd core on slapcat -b cn=conf
This is most likely the same problem as ITS#3793. It is purely a slapd
code issue, and not related to BerkeleyDB in any way.
The slap tools (slapcat et al) are all single-threaded programs, so none
of the thread-related functions should ever get used. It seems we need
some additional init/de-init checks to make sure of that.
Instead of "bt full" please do a "bt 20" on your core file and reply to
this ITS with that info, thanks.
> Full_Name: Dusty Doris
> Version: 2.3
> OS: Freebsd 5.4
> URL: ftp://ftp.openldap.org/incoming/
> Submission from: (NULL) (22.214.171.124)
> Running slapcat on cn=config causes a slapd core.
> I understand that doing this isn't needed, as you can just read the ldif files
> in slapd.d or do an ldapsearch. If slapcat was never intended to run on
> cn=config, then perhaps the answer is simply to adjust slapcat so it does not
> run on that base.
> OS: FreeBSD 5.4
> BDB: 4.3.28 or 4.2.52_4 (from ports)
> OpenLDAP: 2.3.4 (from ports)
> Command run to cause core:
> $ slapcat -b cn=config or slapcat -n 0
> End of output from slapcat:
> slapcat shutdown: initiated
> slapcat destroy: freeing system resources.
> Assertion failed: (rw != NULL), function ldap_pvt_thread_rdwr_destroy,
> file rdwr.c, line 99.
> Abort trap (core dumped)
> $ gdb -c slapd.core
> Core was generated by `slapd'.
> Program terminated with signal 6, Aborted.
> #0 0x282e831b in ?? ()
> End of bt full:
> #1513 0x00000011 in ?? ()
> No symbol table info available.
> Cannot access memory at address 0xbfc00000
> Can reproduce anytime slapcat -b cn=config or slapcat -n 0 is run.
-- Howard Chu
Chief Architect, Symas Corp. Director, Highland Sun
Symas: Premier OpenSource Development and Support