> Charles moved the boot loader to 7 MB because the INSTALL kernel is
> apparently now so large it was overwriting the boot loader as the
> kernel was loaded.

As I mentioned, this means trouble on some machines because they may
have as little as 4M contiguous RAM at the beginning of physical space.

> However, 7 MB is obviously bad from the point of view of SS1/SS2/IPX
> machines (and probably IPCs, too); it would be better to load just
> above 16 MB, so it works on machines with 4 MB/bank.

...and breaks on machines with 16M in the form of one 16M SIMM in the
first socket.

And none of this helps a machine with only 4M of RAM total.  (I could
live with the INSTALL kernel not working on such a machine;
desupporting it entirely would rather bother me, as I have a real use
for such a thing.)

> One option might be to provide multiple boot loaders in /usr/mdec,
> for varying configurations.

Ooo, I really hate the notion that frobbing the memory setup means
reinstalling the bootblocks.

> Another would be to have the boot loader load below 4 MB but try to
> find a place to relocate itself to.

This sounds good to me: load the bootloader at (say) 300000 and have it
ask the PROM for memory configuration and then relocate itself as high
as possible - or not, if the kernel is <3M.

But what about kernels >4M on machines with >4M total but a hole
starting at 400000?  Like an ELC with a 4M SIMM in the first slot?  4M
in each slot is 16M, and it's not an unreasonable configuration to
expect to be able to boot the INSTALL kernel on.

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B