Re: m68k boot-floppies
> checking packages needed in the root disk
> Can't find package pmac-fdisk-cross
> checking packages needed in the base system
> Can't find package pmac-fdisk-cross
> E: ./check_mirror.sh abort
> make: *** [check_mirror] Error 255
> I removed pmac again from the list, Michaels upload did not get through to
> potato (I mirrored today noon). I should get his package and put it into
<brown paper bag mode on>
My fault - I simply hadn't re-uploaded the package yet. It's done now; I
hope the installer doesn't screw up because there's been the incomplete
0.1-3 upload already.
<brown paper bag mode off>
> /debian/local. I should probably also put the 2.2 kernel images (amiga,
> atari and mac) there. Michael, do you still have atari and mac images?
Did I have an Atari image? Seems I built one, and tested it myself only.
Hope it works for others, too.
I do have a Mac kernel image / modules / sysmap tarball which has been
tested by a few Mac users (got some feedback, built a second one and
that was OK apparently).
Both in ~schmitz/build on kullervo now.
> make release: (Kver set to 2.0.36, dont build atari files, I dont have the
> support tar.gz here?)
In /usr/local/build/schmitz on kullervo.
> I: copying required libraries to root filesystem
> tar: ./lib/libresolv-2.1.2.so: Cannot stat: No such file or directory
> -rw-r--r-- 1 root root 42216 Mar 11 06:32 /lib/libresolv-2.1.3.so
> tar: ./lib/libnss_dns-2.1.2.so: Cannot stat: No such file or directory
> -rw-r--r-- 1 root root 9920 Mar 11 06:32 /lib/libnss_dns-2.1.3.so
> tar: Error exit delayed from previous errors
> but the build continues?
Dunno, it should abort if these are important, seems like they aren't.
> I: adding busybox and links to it
> [...]
> I: making loop filesystem in /tmp/fimgQ3FDMY, size 1440k
> 1440+0 records in
> 1440+0 records out
> I: formatting and conditioning the new image
> mkdosfs 2.5 (07 Jan 2000)
> I: mounting new disk image at /var/tmp/boot-floppies/mnt.rescue
> I: installing kernel 'linuxamiga'
> I: rescue image too small to include root.bin
> I: copying help messages
> I: installing boot loader
> Not writing syslinux for amiga !
> I: unmounting, cleaning up
> I: rescue image complete: -rw-r--r-- 1 root root 1474560 Mar 22 20:45
> resc1440amiga.bin
> I: making detached root disk root1440amiga.bin from rootamiga.bin
> 1440+0 records in
> 1440+0 records out
> 974+1 records in
> 974+1 records out
> make[1]: *** No rule to make target `drv1440amiga.bin'. Stop.
> make[1]: Leaving directory /build/boot-floppies'
> make: *** [build] Error 2
>
> Que pasa?
Somebody screwed with the makefile in the meantime, maybe the
drv1440amiga.bin is called something else now but the m68k main target
still wants this file. Look for all the drv1440* targets, or for whatever
depends on the modules.tgz file (maybe that's hidden in a separate shell
script again?).
> I built a base2_2.tgz anyway, maybe somebody wants to try a test install,
> but I dont know how far you can get without a drv image.
You will fail the 'install kernels and modules' step. Who cares. You can
always fake a drv1440.bin by just dropping the new modules.tgz into the
old, loop mounted drv1440.bin. Not that I advocate this over fixing the
makefile but it provides a quick way to complete the set of files to get
them out for testing.
> As for the non-amiga stuff, if somebody integrates it in the Makefile, I can
> give it another try, but I have no time fix that myself now.
Just build the ramdisks (root*.bin) for the other subarchs, the rest is
pretty much irrelevant for testing (the resc1440/drv1440 files are utterly
pointless for a m68k install _except_ for the modules install which can be
done at a later stage unless you need ethercard modules to install ->
people please fix as needed). The root.bin stuff is the only thing crucial
for testing the installs.
We'd need to post that stuff on some FTP site strictly for people to test;
I'd not upload it to potato to avoid confusion ('see there's boot-floppies
for potato now, no need to worry anymore').
> Question:
> base-contents.txt lists this:
> ii xviddetect 0.2-10 XFree86 installation helper
> Description: XFree86 installation helper
> A simple script that will detect which X server to install, by
> running a non-invasive PCI scan of your system and matching against a
> database of known cards
ROTFLAK. I just hope this script reduces to $ echo "You need XF68_FBDev"
on m68k.
> Does this make any sense on m68k? We will allways install fbdev. Why is it
> in the base system at all?
For the graphical, X-based installer perhaps?
But seriously, running such a thing before starting to download X package
is a brilliant idea. Just Not For M68k (tm).
If you want to hear my opinion on XFree and PCI probing, ask me. That's a
long story.
Michael
Reply to: