> The RT 4.3 port was called AOS (for the, "Academic Operating System"). It
> was mostly Tahoe with NFS and came with most of the sources, but some bits
> were distributed only as object code: I believe some of the MM bits?
> Perhaps the MMU code? I vaguely recall this being one of the things people
> had a hard time with when trying to port Reno and 4.4 to the RT.
What was delivered as binary was the Advanced Floating-Point Accelerator
microcode.
Thanks. My memory of all of this is decaying over time. I'd forgotten about the AFPA; I believe our RTs either had a 68881 or 68882, but it's been so long the details are fuzzy: I definitely remember a Motorola FPU, but no longer remember the model.
At the end of the AOS work circa 1996, most of the kernel was 4.4,
except for the network stack which was 4.3-Reno, and the VM system which
was still 4.3 (hence no mmap).
This happened outside of IBM, didn't it? What prevented the rest of the VM code being ported?
> The port was fairly faithful; the C compiler was a bit strange "High C" or
> "Hi C", bit GCC was available after a while, but had some bug and could not
> compile the kernel.
The compiler was Metaware High C. GCC could not be used to compile the
kernel sources unchanged, because one of the locore->trap.c paths was
relying upon the stack layout used by the compiler. With that fixed, gcc
could be used to build a working kernel.
I vaguely remember that happening, but by then we had retired the RTs.
I vaguely remember Metaware being somewhat religiously extreme, but again the details are fuzzy now. Was there some kind of ecclesiastical reference in the man page?
- Dan C.