home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!corton!enst!ulysse!philipp
- From: philipp@ulysse.enst.fr (Philippe-Andre Prindeville)
- Newsgroups: comp.sys.hp
- Subject: Bug in LLA/SNAP and porting Appletalk to HP-UX 8.05/7
- Keywords: appletalk, LLA, SNAP
- Message-ID: <2392@ulysse.enst.fr>
- Date: 26 Jul 92 21:42:02 GMT
- References: <92206.15:50:38.186934.B03@vm.urz.uni-heidelberg.de>
- Reply-To: philipp@res.enst.fr (Philippe-Andre Prindeville)
- Organization: Telecom Paris, France
- Lines: 51
-
- I'm trying to port CAP 6.0 (Columbia University's Appletalk
- package) to HP-UX 8.0. There is an option to use either IP
- encapsulated Appletalk (called IPTalk) or to "go native" and
- use Appletalk directly over Ethernet (called EtherTalk).
-
- Being a brave network hacker (cough), I rolled up my sleeves
- and said, "Well, let's get this done before dinner."
-
- Not quite. First off, I found out that Ethertalk uses SNAP
- encoding (SSAP&DSAP == 0xaa). Good. 8 bytes wasted for
- nothing. Oh well. So then I changed my code to use SAP
- encoding. The doc says that codes 6, 248, and 252 are reserved.
- Fine. I want 170 (0xaa). I run my code and... Device busy.
-
- Hmmm. Probably that brain-dead habit of using SNAP encapsulation
- of IP, I say to myself. Quickly remedied:
-
- % lanconfig lan0 ether -ieee
-
- and I try it again: Device busy.
-
- I looked for background daemons that might be using this SSAP,
- and none are present. Or none that are obvious. I went into a
- rage and killed everything that might be tying up the device.
- No such luck.
-
- Other SSAPs work fine. Strange. So, my conclusion is that
- something in the kernel (probably IP) gets its hooks into the
- device driver and doesn't free it.
-
- Since SNAP encapsulation is used often enough, you would think that
- HP could add one more IOCTL/NETCTRL to their list that allows a
- user to specify the TYPE field of SNAP (ie. provide a client hook
- to SNAP like they did for raw SSAP/DSAPs). IP doesn't use *all*
- of them, only 0x0800 and the HPprobe-thingy.
-
- You would also think that writev()s work with LLA sockets. No
- where in the doc does it say they do, but thankfully, they do.
-
- I think SNAP encapsulation is brain-dead, especially if you don't
- have FDDI. But anyway... Does anyone have a work around to this
- SNAP conflict? It would be nice if multiple clients could use SNAP.
-
- I've noticed that there is a libatalk.a file in /etc/conf. If you
- add "atalk" to your dfile, it gets compiled in. But no where is
- there an API on programming this code. The code seems to have been
- lifted from Kinetics or Stanford. Does anyone have a handle on
- docs? If I can't get LLA/SNAP working, I might have to use the (no
- doubt ancient) version of Appletalk distributed by HP.
-
- -Philip
-