Test boot floppies (from CVS)
I've gotten the full set of boot-floppies to build to my satisfaction and
sort of work - they boot on my PC164SX with SRM and dbootstrap runs (library
reduction was broken but I fixed it).
The set is at http://eradicator.org/debian/alpha-potato-boot/
1) I'm using MILO 2.2-14, so there are separate disks (bootable from
ARC/AlphaBIOS and SRM, or at least, I hope so :) with MILO and linload.exe
2) There is a single rescue disk, built from the generic 2.2.13 image (there
isn't a 2.2.14 package yet, I think I might build it). Two issues to keep
* IP autoconfiguration is turned on, and in 2.2 it takes MUCH longer for
BOOTP and RARP requests to time out :( so be patient ... next version
won't have this "feature".
* The kernel is compiled for SRM/aboot, so I don't know if MILO will boot
it. We may have to have two separate rescue disks in that case.
3) For SRM you can still boot directly from the rescue disk. I put an
aboot.conf on the disk but I'm not sure it will work right, as I haven't
tested it. The full boot command should be:
>>> boot dva0 -file linux -flags "root=/dev/fd0 load_ramdisk=1"
4) Yes, the base system is morbidly obese. This is partly because libc is
not stripped, but actually the main space waster is locales, it seems.
And of course, the usual caveats about this being CVS boot-floppies apply
... some of the installation sequence might be broken. Comments about
dbootstrap and such should probably go to debian-boot.
I'm going to do a test install on a spare disk here soon. Things I'd like
other people to test are:
- ARC and AlphaBIOS installs
- Installs on machines with defective SRM (Jensen, anyone?)
One big gotcha for SRM is that you'll be dropped into cfdisk for
partitioning. I'm thinking that there should be some way for dbootstrap to
notice that you're booting from SRM, and run fdisk instead.