On Tue, 2014-01-07 at 16:58 -0800, Steve Langasek wrote: > On Wed, Jan 08, 2014 at 01:54:32PM +1300, Andrew Bartlett wrote: > > > > > Reading symbols from /usr/lib/x86_64-linux-gnu/libhdb.so.9...Reading > > > > symbols from > > > > /usr/lib/debug/usr/lib/x86_64-linux-gnu/libhdb.so.9.2.0...done. > > > > done. > > > > Loaded symbols for /usr/lib/x86_64-linux-gnu/libhdb.so.9 > > > > > That happened with the samba server in PDC mode, running "passwd" > > > > via sssd or changing the password via Windows 8. > > > > I'm sorry to say that this stack trace shows signs of stack corruption. > > > "0x0000000000000000" and "0x0000000000000008" are not valid addresses to > > > appear in a backtrace. > > > > You may have better luck debugging by running under valgrind, which can > > > often detect stack corruption errors at runtime /before/ they clobber the > > > stack. > > > I strongly suspect in this and in 734355 that Heimdal and Samba have got > > out of sync. We use libhdb (and other parts) from Heimdal in a quite > > intimate manner. > > By "intimate", do you mean "using private interfaces"?
Samba is developed and tested with the in-tree version of Heimdal. The way it's packaged in debian isn't how we develop or test it, but allows us to comply with policies regarding bundled libraries. > What are the relevant values of "in-sync-ness" for heimdal? I would prefer it was exactly the version we have in-tree, but that's now quite old. Andrew Bartlett -- Andrew Bartlett http://samba.org/~abartlet/ Authentication Developer, Samba Team http://samba.org Samba Developer, Catalyst IT http://catalyst.net.nz/services/samba
signature.asc
Description: This is a digitally signed message part
