Subject: Re: AppleTalk
To: Donald Lee <MacPPC1@caution.icompute.com>
From: Timo Schoeler <timo.schoeler@riscworks.net>
List: port-macppc
Date: 09/29/2005 17:55:58
>>>>>I am including the previous post because I want to make sure people know
>>>>>where I am coming from. It was my understanding that Appletalk was not
>>>>>supported in NetBSD. Am I wrong? Specifically I am interested in
>>>>>appletalk over the serial lines.
>>>>
>>>>AppleTalk as in the network protocol is supported (include "options
>>>>NETATALK" in your kernel config file). However, that's the software
>>>>side of things... On the hardware side, LocalTalk (AppleTalk running
>>>>over a RS-422 serial port at 230Kbps) is not supported. So in
>>>>practice, NetBSD only supports AppleTalk over Ethernet. If you want to
>>>>hook up a LocalTalk device, you'd have to use a Localtalk to Ethernet
>>>>bridge of some sort (e.g., Cayman GatorBox, Shiva FastPath, a Mac
>>>>running classic MacOS and Apple's router software that I forget the
>>>>name of, etc...)
>>>>-- Name: Dave Huang         |  Mammal, mammal / their names are called /
>>>
>>>Caveat: Last time I tried to install netatalk, the TCP/IP side of AFP
>>>worked, but the appletalk part did not.
>>>Snippet from my logs:
>>>Aug 27 03:04:50 mercy ntpd[180]: frequency initialized -25.967 from /var/db/ntp.drift
>>>Aug 27 03:04:50 mercy ntpd[180]: using kernel phase-lock loop 0041  Aug 27 03:04:52 mercy atalkd[174]: zip_getnetinfo for tlp0
>>>Aug 27 03:05:09 mercy last message repeated 2 times
>>>Aug 27 03:05:19 mercy atalkd[174]: config for no router
>>>Aug 27 03:05:21 mercy atalkd[174]: as_timer: can't route 200.168 to loopback: Invalid argument
>>>Aug 27 03:05:22 mercy atalkd: difaddr(0.0): Can't assign requested address
>>>Aug 27 03:05:22 mercy atalkd: difaddr(0.0): Can't assign requested address
>>>Aug 27 03:05:22 mercy afpd[236]: main: atp_open: Can't assign requested address
>>>Aug 27 03:05:22 mercy afpd[236]: ASIP started on 209.46.8.67:548(1) (1.4b2+asun2.1.3)
>>>Aug 27 03:09:16 mercy ntpd[180]: time reset -11.325214 s
>>>Aug 27 03:09:16 mercy ntpd[180]: kernel pll status change 41
>>>Aug 27 03:57:43 mercy named[107]: ns_resp: sendto([2001:dc0:1:0:4777::140].53):No route to host
>>>Aug 27 04:00:03 mercy afpd[1707]: ASIP session:548(1) from 209.46.8.68:49195(0)
>>>Aug 27 04:00:03 mercy afpd[236]: server_child[1] 1707 done
>>>Aug 27 04:00:03 mercy afpd[1708]: ASIP session:548(1) from 209.46.8.68:49196(0)
>>>Aug 27 04:00:03 mercy afpd[1708]: randnum/rand2num login: backup
>>>Aug 27 04:00:03 mercy afpd[1708]: login backup (uid 1006, gid 0)
>>>This behavior seems to have started in 1.6.x.  It worked fine in 1.5.x.
>>
>>i see this behaviour too on sparc64 and macppc machines; AFAICS it's due
>>to db4 failing (see [1] e.g.).
>>
>>timo
>>
>>[1] -- http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=29746
> 
> 
> The one I just re-built (with gcc 3 and the "builtin" compiler
> on 1.6.2 (2.95.3)) is netatalk+asun-2.1.3.
> 
> Same problem.
> 
> Can I fix this by re-building db4?  AFAIK, 1.6.2 is all built with 2.95.3.
> 
> ?
> 
> -dgl-

from hacks.mk (see db4's directory in pkgsrc):

### [Sun Mar 14 19:32:40 UTC 2004 : jlam]
### GCC 3.3.x and older versions have an optimization bug on powerpc that's
### tickled by the db4 source code, so remove optimization flags in that
### case. This fixes PR pkg/30647 by Ian Spray.

so, the newer the compiler, the better it should work. in theory.