: I must admit that I'm still running 1.2.11 on my Intel boxes
: and never tried 1.3.x.
Local Linux experts suggest that 1.3 is much faster than 1.2, at least
that's what I recall from the rapid fire discussion that took place at
pub-night a while ago. Not a big concern just yet.
: > This is a valid concern. I'm not sure why I can't get past the kernel
: > messages phase of the boot, but I'm not overly worried about it.
: Where is it stopping ? Perhaps you have a problem with page_color() ?
: This macro needs to be changed depending on your Icache linesize,
: when I remember right.
I don't know. Haven't taken the time to track down this problem. I'm
having difficulties generating a 188.8.131.52 tree to generate a patch
from, mostly disk space problems. I do know that it gets to my last
printk in floppy_init() and seems to return from it. I've not
quantified it further at this time.
: No, it should be on a ramdisk image, which itself must be on the
: boot floppy as long you don't have the floppy driver working. I
: believe there's such an image on ftp.waldorf-gmbh.de. Shall I
: check it for you ?
I do have a copy of that disk image, so you needn't do anything. I
think I need to do the hole in memory and the load at a high address
stuff first, since I'm right on the edge as it is with memory, and
something like a ramdisk image will push me over the edge into video
rom/ram :-(. Also, I'm starting to batch up patches, which I'm not
sure that I like doing, so I'd like to clear my queue, lest something
gets lost in the shuffle.
My game plan:
1) Iterate on patches to MILO and send them off to Andy as
part of 0.26.
2) Get the proper mappings working so I don't have to compile
with as many non-standard hacks (this is TLB stuff)
3) Generate Deskstation patches to date for the Linux kernel
4) Make Linux and MILO grok loading LINUX at 0x80010000 and
have Linux copy down the interrupt handlers, etc.
5) Make MILO grok compressed kernels
6) Start worrying about RAMDISKS and/or SCSI disks, etc.