[389-users] segfault while moving entry to non-existent LDAP container

[Vladimir Elisseev]

We've done some more tests, and it appears that we can reproduce this
segfault only while using a particular LDAP account, but if we use
directory manager to perform the same change using ldapmodrdn,
everything works as expected (no segfault, but error). I think this is a
result of bad ACI's. I'm going to check what is wrong there. However, I
think that too restrictive ACI shouldn't lead to segfault anyway.

Vlad.


On Wed, 2012-11-14 at 07:09 -0700, Rich Megginson wrote:
> On 11/14/2012 02:05 AM, Vladimir Elisseev wrote:
> > Obviously, I've tried ldapmodify and, as expected, it ends with an
> > error, no segfaults. However, I just tried
> > ldapmodrdn -r -h localhost -p 389 -D "cn=xxx" -W "cn=00000000000002000007,ou=1,ou=Users,ou=132252,ou=Tenants,dc=CIDS" "cn=00000000000002000007" -s "ou=DeletedUsers,ou=132245,ou=Tenants,DC=CIDS"
> > where the target superior entry doesn't exist, and, to my surprise, this
> > leads to the same segfault... I don't think it's normal, isn't it?
> > BTW, we've opened a case at Red Hat (we have RHDS subscription)
> > regarding this issue, so I suppose we have top stop discussing this
> > problem here, right?
> 
> No, we do not have to stop discussing this problem here, but the Red Hat 
> support team should be aware of this email thread so that they can 
> follow it.
> 
> 
> 
> >
> > Regards,
> > Vladimir.
> >
> > On Tue, 2012-11-13 at 09:58 -0800, Noriko Hosoi wrote:
> >> (2012/11/13 05:22), Rich Megginson wrote:
> >>> On 11/13/2012 03:30 AM, Vladimir Elisseev wrote:
> >>>> Hello,
> >>>>
> >>>> First of all I'd say that most likely this segfault is a result of
> >>>> badly designed application and/or bad coding. The segfault occurs while
> >>>> this application tries to move an entry to non-existing LDAP container.
> >>>> Unfortunately I don't have access to the source code of this app. The
> >>>> segfault is below with backtrace from dgb:
> >>>>
> >>>> ns-slapd[4983]: segfault at 18 ip 00007f2ed4a60759 sp
> >>>> 00007f2e955e13e0 error 4 in libback-ldbm.so[7f2ed4a34000+8f000]
> >>>>
> >>>>
> >>>> #0  0x00007f2ed4a60759 in id2entry_add_ext () from
> >>>> /usr/lib64/dirsrv/plugins/libback-ldbm.so
> >>>> #1  0x00007f2ed4a8a34c in modify_update_all () from
> >>>> /usr/lib64/dirsrv/plugins/libback-ldbm.so
> >>>> #2  0x00007f2ed4a8eb4f in ldbm_back_modrdn () from
> >>>> /usr/lib64/dirsrv/plugins/libback-ldbm.so
> >>>> #3  0x00007f2eddbecdaa in ?? () from /usr/lib64/dirsrv/libslapd.so.0
> >>>> #4  0x00007f2eddbed66c in do_modrdn () from
> >>>> /usr/lib64/dirsrv/libslapd.so.0
> >>>> #5  0x0000000000413904 in ?? ()
> >>>> #6  0x00007f2edc0369e3 in ?? () from /lib64/libnspr4.so
> >>>> #7  0x00007f2edb9d9851 in start_thread () from /lib64/libpthread.so.0
> >>>> #8  0x00007f2edb72711d in clone () from /lib64/libc.so.6
> >>>>
> >>>> I'd appreciate any thoughts regarding what kind of (bad) things this
> >>>> application is doing. Is it possible to have a kind of protection in
> >>>> this case on directory server?
> >>> rpm -q 389-ds-base
> >>> Can you provide a full stack trace based on the instructions at
> >>> http://port389.org/wiki/FAQ#Debugging_Crashes ?
> >> Also, can we have the modrdn operation you executed?  Command line
> >> history and/or the snippet of the access log would be helpful.
> >>
> >> I tried these modrdns, but it failed with the expected errors... And the
> >> server is up and running after that.
> >> $ ldapmodify ...
> >> dn: cn=HR,ou=Groups,dc=example,dc=com
> >> changetype: modrdn
> >> newrdn: cn=HR
> >> deleteoldrdn: 1
> >> newsuperior: ou=bogus,dc=example,dc=com
> >>
> >> modifying rdn of entry "cn=HR,ou=Groups,dc=example,dc=com"
> >> ldap_rename: No such object (32)
> >>       matched DN: dc=example,dc=com
> >>
> >> $ ldapmodify ...
> >> dn: cn=HR,ou=Groups,dc=example,dc=com
> >> changetype: modrdn
> >> newrdn: cn=HR
> >> deleteoldrdn: 1
> >> newsuperior: o=bogus.com
> >>
> >> modifying rdn of entry "cn=HR,ou=Groups,dc=example,dc=com"
> >> ldap_rename: Operation affects multiple DSAs (71)
> >>       additional info: Cannot move entries across backends
> >>
> >> --
> >> 389 users mailing list
> >> 389-users at lists.fedoraproject.org
> >> https://admin.fedoraproject.org/mailman/listinfo/389-users
> >
> > --
> > 389 users mailing list
> > 389-users at lists.fedoraproject.org
> > https://admin.fedoraproject.org/mailman/listinfo/389-users
>