Date: Tue, 15 Jul 1997 23:22:22 +0600 (ESD) From: "Serge A. Babkin" <babkin@hq.icb.chel.su> To: terry@lambert.org (Terry Lambert) Cc: msmith@atrad.adelaide.edu.au, jhay@mikom.csir.co.za, hackers@FreeBSD.ORG Subject: Re: IPX routing? Message-ID: <199707151722.XAA19286@hq.icb.chel.su> In-Reply-To: <199707151637.JAA03601@phaeton.artisoft.com> from "Terry Lambert" at Jul 15, 97 09:37:13 am
next in thread | previous in thread | raw e-mail | index | archive | help
> > > I've looked ayt the Linux code and it seems to me that they > > just allow the applied software to decide which encapsulation > > type to use: they just fill the full IPX header (that includes > > Ethernet header) in user-level software rather than in driver. > > It would be not bad to make the IPX implementation more > > Linux-like so it would be easy to port Netware emulators. > > If this is true, then they are not doing full 802.3 encapsulation > for 802.3. If this code is not special cased to be IPX specific > (it would seem to me that it could not be, if their IPX is in > user space), then their 802.3 implementation is not useful for > 802.3 packets other than IPX. This is because the IPX packets > are only partially encapsulated (Novell misunderstood the 802.3 > encapsulation requirements, and never corrected their implementation). I'm not quite sure, I looked that in the NWE sources. May be the kernel then repackages the packets, I don't know. But I can say for sure that the applied program can write the port numbers and addresses right into the packet (not into the address structure or by binding) and then transfer it to kernel. -SB
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199707151722.XAA19286>