Re: Bug#23143: make-kpkg fails on Alpha
On Tue, Jun 09, 1998 at 02:18:15PM +0100, Nikita Schmidt wrote:
> Absolutely. Not even mentioning that on systems where MILO can't keep
> settings in the NVRAM it is plain impossible to boot up Linux without
> having to type a long `boot ...' line (thus, no automatic reboot).
I didn't realize that that situation even existed, but that would also be a
place where this configuration file would be helpful.
> However, a question: which disk and partition the config file should be
> read from? Shall we employ RTC memory or reserve a place inside MILO's
> executable image? Or can parameters be passed to MILO from SRM or ARC? (I
> believe both SRM and ARC can store their settings in some kind of
> non-volatile memory on all types of systems.) If they can, do we want it
> done that way? (It will prevent people on whose systems MILO works OK even
> when run straight from the NVRAM from installing it there.)
I think the best solution would be d) all of the above.
You can pass parameters to MILO from ARC (and I assume SRM), so it should be
settable that way. You should also be able to compile in a default, I
think, to accomodate those who have it flashed in.
I think the project will come down to a parser for the config file (steal
from LILO?), a way to take input (also from LILO), and then some connecting
glue between the new stuff and existing MILO routines.
> Next idea in this direction is that it would be nice to have MILO
> load the kernel automatically even on the first run, from the boot
> floppy or CD or whatever. Or at least read a config file with the
> choices. Not a problem with ARC (when MILO is on a FAT floppy, isn't
> it? - so the config file can be there as well), but an SRM bootable
> floppy can't have a reasonable filesystem unless partitioned, and
> (correct me if I'm wrong) Linux doesn't support partitioned floppies.
Your supposition about MILO being run from ARC is correct. I don't know
about the rest.
> So we probably need to be able to supply a raw block number instead of
> partition number along with the disk specification. Any ideas?
Oh, this is getting beyond my rather picayune intentions, I'm willing to
consider anything.
> If nobody else wants or has started already, I can try to hack MILO
> accordingly (I've just finished the project that was eating up 110% of
> my time). However, as Mike has already shown his potential interest in
> doing this, I leave it up to him for the moment.
If you have time, work on it. I had been thinking about this for the last
few weeks, but that represented a commitment of daydreaming, rather than
working.
Wasn't Chris talking about getting it to compile? I've heard that's a
bitch, so you might want to talk with him about it.
I may at least get the source and look it over, though.
Mike.
--
To UNSUBSCRIBE, email to debian-alpha-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: