Re: docs on compiling kernels for miboot?
Ethan Benson wrote:
> On Thu, Jun 28, 2001 at 08:22:40AM -0700, Andrew Sharp wrote:
> > Are you sure it's an elf? Now you've done it. I'm going to have to
> > fire up the mac to see where I went wrong on this. Oh, what a
> > bother. I coulda sworn ....
> eb@socrates boot-floppies$ ls -l linux*
> -rw-r--r-- 1 root root 867199 Jun 28 06:25 linuxapus
> -rw-r--r-- 1 root root 888549 Jun 28 06:34 linuxchrp
> -rw-r--r-- 1 root root 1028605 Jun 28 06:43 linuxpmac.coff
> -rw-r--r-- 1 root root 1141274 Jun 28 06:43 linuxpmac.gz
> -rw-r--r-- 1 root root 731418 Jun 28 06:51 linuxprep
> eb@socrates boot-floppies$ file linuxpmac.coff
> linuxpmac.coff: executable (RISC System/6000 V3.1) or obj module not stripped
> eb@socrates boot-floppies$ zcat linuxpmac.gz > linuxpmac
> eb@socrates boot-floppies$ file linuxpmac
> linuxpmac: ELF 32-bit MSB executable, PowerPC or cisco 4500, version 1, statically linked, not stripped
> eb@socrates boot-floppies$
> now lets go look at what the miboot script uses again....
> eb@socrates boot-floppies$ grep linux powerpc-specials/miBoot/mkboot.sh
> hcopy -r linuxpmac.gz :zImage
> yup im sure.
No really, are you sure? I think someone's just rubbing it in....
> > One thing about the OF is that it isn't readily available on some
> > machines, like my first pmac, the 7200. I tried very hard, and
> > failed, to get to the OF on that machine, although I have read many
> > successes on this list. I guess I never was able to get the cable
> > right. Anyway, it's a pain, so there is always going to be that one
> > person emailing in: "...but I'm on a 7200, and I don't see an OF
> > prompt...." Perhaps we should just say that such machines aren't
> > supported anymore, but if you want to take a crack at it, there's
> > this old miboot thang, and quik, go knock yourself out. For a
> > couple of months.
> woody boot-floppies should make 7200's bootable.
The current crop of potato boot floppies work on 7200s, but the OF
isn't available without the right cable and a terminal or machine
running a terminal emulator. A real pain for me.
> > One thing that makes SILO so sweet is the relatively competent OF
> > job that Sun, and it's consultants, did. I was under the vague
> relatively competent? Sun's OpenBoot/OpenFirmware implementations are GREAT.
> > impression that yaboot was the closest to SILO that we could get
> > dealing with apple's OF implementation. I worked next to the same
> no its not. Apple's NewWorld OF is complete enough to have the full
> set of silo's features, and there is little reason for a code fork.
> silo now has much cleaner filesystem support making it much easier to
> drop new filesystems in (without breaking ext2 ahem). as well as nice
> things like transparent gzip support, a working message= variable,
> built in ls, and cat commands, tab key completion, etc.
> when looking to port these to yaboot i have come to the conclusion
> that it would be far simpler to add powerpc support to silo proper
> then to spend ages decrufting (read rewriting) yaboot. basing yaboot
> on quik was a good way to get us a quick and dirty working bootloader,
> but its not a good thing long term.
OK you have me drooling now, even though I don't have newworld.
Yet. Is anybody working on this??
> > guys that did Sun's OF; they were hired to do
> > PowerHouse/FirePower/StarMax's OF implementation, so I suppose that
> > some ppc machines have better boot roms than others.
> i thought the clones just got thier rom from apple?
Apple cloners did, but Powerhouse was an NT clone. That's right,
prep or chrp (I can never keep those two separate in my head). We
had reference motherboards from IBM called sandalwood that were
never released to the public and came with a port of NT on them.
Apple's rom was not a consideration for an NT machine that had to
pass a very specific data structure to the HAL. OK, OK, I did the
HAL and the scsi and ethernet drivers. I took my lashes for that
transgression years ago. The guys that did the OF rom programming
were pretty good, but had a lot of nasty grumbling going on when
asked about "...so how do you like this NT thing?" They were part
time because they were simultaneously doing a number of
implementations for Sun. In those days, apple, ibm, motorola were
all going to switch over to prep/chrp, whichever, and the powerpc
world was gonna be great! At least, apple and motorola were. IBM
was gonna add a prep line. Then IBM politics took over and that was
dead. The A/IX people thought of themselves as pretty exclusive
group, and didn't want an IBM/PPC product that ran NT and was cheap,
very cheap, by comparison to an RS/6000 product. It was thought
people would stop paying exorbitant prices for RS/6000 workstations
if they had this choice. I don't know what made them think that.
Motorola dropped out, and we all know what a bunch of boneheads
Apple is, I don't think they even started to work on it. For five
minutes an arch was dreamed of that might stand a chance of at least
standing up to the hegemony of the PC/x86 arch and at least be an
alternative. You would be able to buy a box and run NT or mockos on
it, or both, you're choice! With an open spec! At the time, there
were a lot of rumours that Powerhouse's boxes would run NeXTStep,
but that was not the main intention. Jobs' attitude was, yeah,
sure, if you pay us a gazillion dollars to port it. At that point
Rothstein, now a top VP at apple and the main founder of Powerhouse
really hated Jobs for cancelling the hardware at NeXT and several
other things. No one even thought about Linux. It was still little
more than a twinkle in Linus' eyes in those days.
The funniest thing is that the hardware and software guys that they
hired from Apple were the worst bunch of clowns I've ever seen. The
hardware team which they took almost intact from NeXT were seriously
good. They threw together a working dual 604 motherboard before
even Motorola had their reference board done. Motorola kept calling
them with questions!