Interesting review. Have to point out the first, published in 1961, was Burroughs B5000:
Anyone on a security forum should be impressed to death by that design appearing in the Fortran era where machines were giant calculators. It had a high-level language that aimed for safety/maintenance, a CPU built for one, bounds-checks, some pointer protection, CPU-checked function calls (bit too heavy there), code vs data tagging to only allow code execution, OS/apps came in source w/ interface checks during compile, virtual memory, multiple CPU’s, simple IDS, and could freeze/fix/thaw faulty programs. Holy. Shit.
Just imagine if market demand was for something other than price or performance… reasonable for the mini/microcomputers I’ll give them… just imagine though had that been the starting point for INFOSEC. I mean, it was given Anderson from Burroughs invented INFOSEC & another put the isolation primitives into Intel CPU’s. Still, just imagine if we at least kept all the checks in the CPU’s, code/data separation, and type-safe language for OS’s. Imagine how few stories about botnets we’d have seen until ROP was invented. Alright, now that alternate history was fun, wasn’t it? :)
Top project going the Burroughs route on CPU side:
Top project trying to design a safer PDP-11 to protect UNIX legacy: