Bug#323182: debian-installer: boot.img for sarge doesn't, on PPC
On Tue, Aug 16, 2005 at 04:24:53PM -0400, SR, ESC wrote:
> Le lun 2005-08-15 a 17:44:46 -0400, Sven Luther <email@example.com> a dit:
> > On Mon, Aug 15, 2005 at 06:13:12PM +0200, Geert Stappers wrote:
> > > severity 323182 normal
> > > thanks mate
> > >
> > >
> > Ok, please ignore my previous mail, it is stupid.
> then disregard my reply to it :).
> > The boot.img are the miboot floppies for oldworld pmacs. These are saddly not
> > useable in sarge, because miboot is non-free (due to the one boot block which
> > is coming directly from apple and has a couple tens of m68k assembly
> > instructions nobody could be bothered to reverse-enginneer).
> um, i would, if i had the skills. and yes they still would be useful.
> i also have other ideas, such as pulling the rom off oldworlds, and
> sticking an eeprom chip on them, and turning them into frankenworlds.
> "oldworld" mac ppcs with as-new-or-newer-firmware-than-newworlds.
> would take some work to figure out what to keep on the old MacROM, to
> have them still function, but we could get rid of some issues with
> these things - like no cd linux booting, buggy firmware on the beige
> g3s such that they don't quik boot properly, etc. ...
Yep. If we go that way you could as well put one of the free OF
implementqations on them and transform them in CHRP boxes.
> as to miboot + the bootable codemaybe it's about time someone did
> something about it, instead of saying "it's not worth it, these are
> too old anyway". maybe they are, but why judge hardware based on what
> a corporation thinks?
It needs two people for a meaningfull reverse engineering, and i have
abstained from doing the first step to be free to hack on the second step.
The real problem is nobody has come up and did it.
There is some effort in alternate code though, maybe quick supporting floppies
or something such, we will see.
Anyway, for etch we are dropping 2.4, and 2.6 miboot was broken. (worked twice
and never after though, so no idea what the probmlem is.
> > As a result, the floppies present in sarge are entirely broken and will never
> > work. Also the 2.6 floppies, even with miboot present never worked fiably,
> > annd since we are doing away with the 2.4 kernels for etch, this means that
> > the miboot floppy target is going away then.
> um, maybe them not working should have been fixed instead of releasing
> them with sarge... isn't that irresponsible to do so?
Yep, but nobody came forward and did so. This should h&ave been mentioned in
the errata though at least.
> > Anyway, please try some of the older daily builds found on my site :
> > http://people.debian.org/~luther/d-i/sarge/images/
> > Chose a date, and go to powerpc/floppy-2.4 to find something working.
> i *know* those work, at least a lot of them do. what's so different
> with these, vs. the ones released?
Cool. The only difference is that the released ones don't contain non-free
miboot, and thus don't work. Well, if you consider the same date that is.
The current only way to boot an oldworld in a free manner is to drop to the
firmware and boot the vmlinuz-coff.initrd from it.
> > Again, sorry for my other stupid mail, and test these ones.
> > Maybe i should build a couple of miboot floppies accompanying the official
> > release ?
> that would be nice, and appreciated. quite often people installing
> linux want nothing to do with mock os, or apple. the oldworlds often
> give this opportunity, since you can do what you want to them, and not
> care about some stupid "warranty". and they work very wel, *despite*
> the bugs in OF, since a) there's ways around the orked bits of the
> firmware; b) a lot of them are very upgradable hardware-wise; c) they
> run linux better than they do mock os.
(Notice i did build some nubus kernel even, but i guess nobody even bothered
to test those.)