[OpenAFS-devel] pr_Initialize falis when called

[Jeffrey Hutzelman]

On   wrote:

>
> <Pine.GSO.4.61-042.0408061202140.26136@johnstown.andrew.cmu.edu><35461.35
> .10.2.90.1091810490.squirrel@vampyrus.envirobat.org>
>

Gee, it would sure help if whoeever is using a mailer that inserts random 
blank lines in the middle of headers would fix it...

> The debugger messages being displayed around the vicinity of the error
> take place within the module "rx/rx.c" on line number 758 (that is with
> the inclusion of some debug print statements) right here:
> printf ( "in rx2\n" );
>     RXS_NewConnection(securityObject, conn);    <------
> printf ( "in rx3\n" );

Sounds like someone is calling rx_NewConnection with a bad security object.
RXS_NewConnection is a method-invocation wrapper.  It looks inside the 
security object to find a pointer to the "new connection" method for that 
object, and calls it.

If your program is actually crashing on this line, and not inside 
rxkad_NewConnection, then either the secobj contains a bad method pointer, 
or more likely, the secobj pointer itself is bad.

Unfortunately, the backtrace you sent was somewhat mangled by cut-and-paste 
and by being quoted-printable encoded in a message that doesn't include an 
appropriate content-transfer-encoding header (again, due to the extraneous 
blank lines).  Worse, it appears that some of the data in the backtrace is 
bogus.

I would suggest you do the following:

- rebuild the relevant binaries (including the AFS libraries) with
  debugging support (-g) and without any optimization, if you have not
  already done so.
- Put the core file and the executable that generated it somewhere
  where we can see them (maybe somewhere in AFS?)
- Change your password


-- Jeffrey T. Hutzelman (N3NHS) <jhutz+@cmu.edu>
   Sr. Research Systems Programmer
   School of Computer Science - Research Computing Facility
   Carnegie Mellon University - Pittsburgh, PA