February 16, 2018 UPDATE

Today, after some heavy searching, I've found sources to a number of
utilities customized by CMU. In particular, I've found the CMU-modified
sources for the version of csh(1) in MACH 2.5. This is important, because
there is at least one bug ("cd ../.." at root will crash csh). I've found
the problem, fixed it, and rebuilt csh. The problem is resolved now.
Additionally, the binary sizes are so close that it gives me confidence that
the found csh source code really is that which was used in Mach 2.5:

root@mod43:/ 5 # ls -l /usr/build/obj/usr/cs/bin/csh/csh /bin/csh
-rwxr-xr-x 1 mtxinu 122904 Jan 27 1991 /bin/csh
-rwxr-xr-x 1 root 122912 Feb 16 16:56 /usr/build/obj/usr/cs/bin/csh/csh
root@mod43:/ 6 # size /usr/build/obj/usr/cs/bin/csh/csh /bin/csh
text data bss dec hex
118752 4128 15692 138572 21d4c /usr/build/obj/usr/cs/bin/csh/csh
118752 4120 15752 138624 21d80 /bin/csh

Work continues with the distribution-generation utility for sysload(8).

LATE update:

Suddenly, after many weeks of working on Mach386 using my lone ISA-based
computer, the system suddenly became unstable and generally wouldn't boot
or even begin the BIOS POST test.

Examination of the motherboard revealed why: Almost every one of the electrolytic
capacitors on the motherboard had suddenly swelled, with many having almost popped
the can off. There was no physical leakage onto the motherboard, however. All of
this literally occurred within a 12-hour period; I specifically examine them periodically.

While I have a spare motherboard and CPU, it only has one ISA slot. I need
at least two ISA slots to accommodate the Adaptec SCSI and the SMC 80x3 ether
boards being used for Mach386.

I've ordered new capacitors from Amazon at a very good price, and they're
expected by tomorrow evening--free next day shipping with my Amazon Prime.