Re: libpam segfault when passwd passes NULL pamh (was Re: gcc -O2 produces invalid object code (x86_64, netbsd-5 branch))

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Tue, Mar 09, 2010 at 01:56:33PM -0500, Richard Hansen wrote:
> Joerg Sonnenberger wrote:
> >On Tue, Mar 09, 2010 at 01:05:19AM -0500, Richard Hansen wrote:
> >>>Anything more than a plain number and fixed error message is
> >>>not possible with the constraints of the interface.
> >>
> >>I'm sorry, I'm still not following.  Are you arguing that
> >>pam_strerror() isn't needed for pam_start() failures because
> >>there's no useful information to convey to the user?
> >
> >No, just that the help it can give is very limited.
> 
> Ah, yes.  pam_strerror() could return a pointer to a static buffer
> that it fills with a rich error message, but that approach is
> problematic (concurrency, string lifetime, etc.).

Exactly.

> In the meantime, it'd be nice if passwd was fixed (see attached
> patch).  Ideally the fix would be cherry picked to the netbsd-5
> branch.  Should I submit a new problem report?  Or start a new
> thread on the appropriate mailing list?

I've committed a patch earlier and send a pullup request :)

Joerg