I'm away from my test net until tomorrow or Monday, but the pieces I can
answer for now...

On Sat, Feb 27, 2010 at 06:04:54PM -0800, ottomeister wrote:
> On Fri, Feb 26, 2010 at 10:04 PM, Michael Jinks <[email protected]> wrote:
> > Sun Ray info:
> > VERSION: ?4.2_17,REV=2009.06.24.11.48
> 
> That's not the 4.2 product release.  It's probably the first Early
> Access build,
> which is now obsolete.  Sun Support won't take a call on an EA version, and
> there were more than a few fixes that went into 4.2 between EA and the final
> release. Offhand I don't recall any fixes that would have a bearing on this
> PAM issue but you should plan to move to the final product release anyway.

Good point.  Trying to hunt up the download now.  If nothing else, it's
something I can work on remotely...

> As far as diagnosing the problem goes, I'd start by looking for clues in
> /var/adm/messages and in /var/dt/Xerrors.

In the messages file, the last reference I find to anything pam-like is
from a time when I had our module in the config file but not copied into
place yet.  Once it was in the path where the config file looks for it,
no further mention.

Nothing in the Xerrors file at all, as far as I can tell, just routine
stuff.

> If that doesn't help then you might
> try turning on PAM debug reporting; see
> http://blogs.sun.com/wittyman/entry/debugging_pam_in_solaris for
> instructions.

Sounds promising.  Will do that when I get back to the office in a day
or two.

> If that doesn't identify the problem then you could resort to attaching a
> 'truss -f -upam' to the 'dtlogin' master process (the one whose parent PID is
> 1) before you attempt to start a new Sun Ray session and log in. (If you
> already have a session, choose 'Reset' from the greeter after launching the
> 'truss'. I think that'll be enough to get you a fresh greeter that
> will be tracked
> by the 'truss'.) This will generate tons of output, so be sure to send
> it to a file
> (by using the -o option) for later analysis.

Yeah, I was afraid it might come to this... but I didn't know exactly
which process to truss, so thanks for that.

> Just for fun, you might compare the output of 'nm' on your module against the
> output from the Sun Ray modules and see whether there's a symbol name
> clash between them.

At that point I'm in over my head.

By a symbol clash, do you mean symbol names which show up in both our
module and the Sun Ray ones?  Clearly that's not what you mean because
if so, there are "clashes" between the Sun Ray modules themselves.  Can
you clue me up a little, or tell me where I should RTFM to find out what
a symbol clash is for a shared library?  So far Google and Wikipedia
haven't helped.

> Your module is built as a 32-bit binary, right?

When I read this I thought, AHA!  I hadn't taken any steps to prevent
building 64-bit.  But no, it's 32-bit ELF.
_______________________________________________
SunRay-Users mailing list
[email protected]
http://www.filibeto.org/mailman/listinfo/sunray-users

Reply via email to