Re: (alpha) YES, we have working boot-floppies

[Falk Hueffner <falk.hueffner@student.uni-tuebingen.de>]

Stefan Schroepfer <whoknows@onlinehome.de> writes:

> Falk Hueffner wrote:
> > Hm, I'm open for suggestions about which MILOs to take.  Should I just
> > use those from potato?  I'm not sure they understand sparse
> > superblocks...
> 
> Ummm, no they don't. More precise: I played around a while with
> potato Milo and a PC164, creating new file systems on spare space
> on one of my hard disks, and got the following results:
> 
> PC164 Milo from potato
> - is not able to read from a file system with sparse superblocks
> - is not able to read *any* filesystem created using potato (I
>   tried "mkfs -O none <device>")

Well, at least this is explainable: due to bug #108165, mke2fs ignores
-O none and made a sparse superblocks file system anyway.

> - is able to read from a ext2 filesystem created from within my
>   "somehow" updated RedHat 5.1(!) installation (sorry, I have
>   nothing "in between")
> 
> That leaves us ARC users (if it wasn't a pure coincidence that
> Milo-2.2.17 didn't work for Don Spoon on his XLT) in a rather
> weird situation:
> - All "Reinauer" Milos do not work.
> - Milo from potato does work, but can't read "up-to-date" ext2
>   filesystems (maybe some other change before the introduction
>   of sparse superblocks is to be blamed or maybe current impls
>   simply are not able to create backwards compatible filesystems).

Hmm, this sucks.  I think I'll then use the potato milo for pc164 and
xlt, and document it's inability to read newer ext2 systems.  As an
(IMHO) acceptable workaround, you can always put your kernel on the
FAT partition which contains the MILO (if you made it large enough).

        Falk