On Tue, 10 Mar 1998, Henry B. Hotz wrote:

> >Do you have access to a copy of EtherPeek? Probably not if you're snubbing
> >the Net.gods. :-) It'd be a good thing to see what's up. Also, try running
> >tcpdump. I don't know how to configure it, but it supposedly can do
> >appletalk.
> 
> Well it's listed in JPL's software catalog at $630.01.  At that price I
> think I'll give tcpdump a try first.  No, I won't because the man page says
> it only supports kip-encapsulated ddp, not native ethertalk packets.

Oh well. I guess I was told wrong. None of the Networking folks would loan
you a copy?

> Backup:  there is an Appletalk-aware version of Solaris snoop on
> playground.sun.com, but I need to move my Solaris machine to the same
> subnet first.  Since this subnet uses switched repeaters I may need to
> temporarily install my own passive repeater as well (and hope the Net.gods
> don't notice and charge me $1500 for it).  *sigh*

At least around here, the Networking folks are happy w/ being support
staff. Sounds like these folks are more interested in their own little
empires. Or their boss is.

> Are they the same as the ones used by LaserWriter 8?  Probably I should
> just  disable papd.  I'd rather run a router than a gateway.  Except for
> the performance issue, of course.

AFAIK, yes they are.

> Repeating my comment in private email:  Making atalkd.conf read-only makes
> some sense since the machine is on all the time, but it may be required to
> change the node assignment on the primary interface and a read-only config
> file would probably prevent that.

I don't think that'd be a problem. As I understand it, atalkd will try the
given address, and pick a different one if it's in use. W/ making the file
r/o, next boot it will try the address which didn't work (and thuss
possably change to another one again) rather than trying the one which
did work.

> >I can send you patches to try in private EMail. It's not fully fixed yet.
> >Basically we end up enabling interrupts which shouldn't be enabled, and
> >the open inputs end up ringing.
> 
> That fits.  I was hoping there was a workaround like disabling something or
> setting the baud rate way low.  Would it help to monkey with the booter
> settings, or is the problem too late in the process for that to help?

Nope. But I think it's a simple fix to z8530tty.c

> Better yet can you suggest a loopback wiring for the serial plug that will
> hold all the relevant signals in place?  Ready signals may strobe as the
> interface is configured.  Perhaps wiring Tx data + or - to all the input
> lines?

I think the problem's ringing inputs, so wiring HSKi and GPi to ANYTHING
would/should fix the problem. If so, maybe set the line to "raw" mode so
transitions get ignored.

> I can help you test patches as long as 1-week turnaround is OK.  Otherwise
> I'd rather get a hardware solution that works.

Sure.

Take care,

Bill