Subject: Re: [netatalk-admins] Re: running on Solaris 2.6 x86
From: Thomas Tornblom (Thomas.Tornblom@Hax.SE)
Date: Wed Oct 07 1998 - 17:10:09 EDT
The "ifdef(sun)&&ifdef(i386)" are probably leftovers since I ported netatalk to
the Sun386i, aka RoadRunner.
This machine runs SunOS 4.0.2 and has absolutely nothing to do with Solaris 2.x
x86. The ifdefs were really issues with SunOS 4.0.2, not with i386. It seems Sun
changed some kernel interfaces between 4.0 and 4.1 which made some htons/ntohs
obsolete.
Thomas
> >Received: from terminator.rs.itd.umich.edu (terminator.rs.itd.umich.edu
[141.211.164.2]) by hellfire.englund.nu (8.8.8+Sun/8.8.8) with ESMTP id WAA17811
for <Thomas.Tornblom@Hax.SE>; Wed, 7 Oct 1998 22:02:26 +0200 (MET DST)
> X-Sender: john@donkey.oaty.com
> Mime-Version: 1.0
> To: netatalk-admins@umich.edu
> From: John Grant <jgrant@concordant.net>
> Subject: [netatalk-admins] Re: running on Solaris 2.6 x86
>
> Well after some snooping, it is obvious that (at the least) there is a
> byte-swap problem. Watching the chooser hit the server I can see packets
> received of type 9B80 (s.b. 809B).
>
> Looking at the souce there are lots of lines ifdef(sun)&&ifdef(i386)
> which (I don't think) can ever work, so I ripped out the ifs leaving all
> the calls to hton intact. (hton does byte-swap, I checked). Still no joy,
> the machine still sees the swap. I changed at.h to be 9B80 instead, but
> that made
> no difference. So how does 809b get itself registered to be APPLETALK ?
>
> Anyway, help would be appreciated - the source is all kind of shredded now...
>
> ---
>
> John
>
>
>
Real life: Thomas Törnblom Email: Thomas.Tornblom@Hax.SE
Snail mail: HB Hax Phone: +46 18 290 290
Banvallsvägen 14 Fax: +46 18 290 291
S - 754 40 Uppsala, Sweden Cellular: +46 73 986 6318
This archive was generated by hypermail 2b28 : Sat Dec 18 1999 - 16:33:25 EST