Re: UDB potato installation
Ulrich Teichert <ut@netsurf.de> writes:
> Afterwards, all went well until the mke2fs run (I decided to start
> from scratch), I choosed to use the non-compilant filesystem (*)
> and was unable to boot from that filesystem, because MILO (the old
> one and the one from 2.2) couldn't read the new filesystem.
> This weren't that bad, if I had known a way back! I was siting there,
> with a perfect running system which couldn't boot up from it's own
> root filesystem, which was mounting well.
>
> I failed miserably to get mke2fs to write a 2.0.x compilant filesystem
> manually, because I didn't found the right option - I really don't
> know what I should have done at this point.
mke2fs -O none -b 1024
The '-b 1024' is essential if you have a fairly large disk.
Somebody should really fix the ext2 code in MILO...
> I ended up writing the filesytem with an old debian installation
> disk, used that filesystem as it was, and all was well. I know
> that an installation on Alpha is harder than on i386, and that
> the UDB is obsolete as hell, but at (*) there should be warning
> for MILO users, IMHO.
Well, the intent was that dbootstrap would just not give you the
option to make a non-compliant filesystem, and I kind of thought that
I had tested this code and made sure it did that. But the code that
is supposed to do this looks fairly obviously wrong to me. I'm not
sure if this is my fault or a case of boot-floppies developers who
don't know C very well messing it up where it worked before.
I'll try to get around to fixing it this week and I will put up a test
version for MILO users when I do so.
--
David Huggins-Daines - dhd@debian.org
Reply to: