My quick answer is yes, this sounds cool.
From that time, that two attempts to network 40 class systems that I know about were:
- As I have mentioned in the past, in the late 1970s CMU had developed an 8085 later 8086 based Multibus board that would eventually begat the Cisco AGS, called the CMU distributed Front-End. It was an attempt to get IP connected to the 11/40 class systems, by splicing them into the 11's with an DR-11B or C (I've forgotten which). But the primary protocol processing was done on the FE and not in the UNIX kernel. Phil Karn might remember what happened to that stuff, I do not believe I still have it.
- Shortly thereafter @ Tektronix had 3Com's UNET. I was primarily running on the 11/70 and that was what we used to develop against the Cyber and VMS IP/TCP's we wrote. We also had it on an 11/44 but both of those are 45 class. We did try to run on an 11/60 - which is 40 class as you asked, and we did get it work as I recall after a fashion. But we did not push it very hard;as we had built a front end that in the same vein as the CMU device, called the NIBB - Network Interface Black Box, which was a Z80 that did the same sort of thing and the need was not there.
I was under the impression the 2.X folks may have tried to do it later with all the thunk code. But that was all after my work. Fred Canter at DEC might have tried something in later versions of Ultrix/PDP11, we should look at the sources in his stream had see what he did too.
Clem