I'm so close to being able to build a kernel that I can almost taste it! My
system is a uVax II with 13MB or RAM. I've got 16MB for it, but when I try
to run with 16MB it won't come up to multiuser. :-( I grabbed the tar files
out of current (4/23) and unpacked them, then did a make build. It got
through the includes target before it hung so that was good (complete set
of includes in DESTDIR)

Then I went to /usr/src/arch/vax/conf and did a config GENERIC and that
worked (first time that's happened) and went over to ../compile/GENERIC and
started a make (remembering what the faq said about not running make
depend). The kernel has been building for most of the day. The system
however hangs randomly, usually after compiling a file that used some swap
space (like vfs_syscalls) for example. Big switch statements are especially
hard on the compiler. 

Anyway, when it hangs, if I break using the console I get:

?02 EXT HLT 
PC = 800006D7

The interesting thing is that the PC is always sitting in the 80000000
page. The last three digits vary but always the first page in 2GB. My guess
is that somewhere the VM system is handing back a virtual address on this
page and something about the MMU on the uVax II (KA630) can't handle it and
it goes into la-la land. Would one of you folks with a 630 CPU manual take
a look and tell me what is supposed to happen? (I'd expect a page fault but
ya never know.)

If I can get through a successful kernel compile then the next step will be
to add the 'break into ddb' feature so that I can go poking around to see
where the heck the system thinks it should be when this happens.

--Chuck