[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#31450: makedev patch for m68k boot floppies



Enrique Zanardi wrote:
> > > /dev/tty* /dev/sda*, /dev/sdb* /dev/loop*
> >
> > > with obvious results when booting the ramdisk or installed base
> > > system.  Patch attached.
> >
> > Yikes, thanks.  I'll let Enrique handle this; it's over my level of
> > competency.  I can't tell if it's a makedev bug or a boot-floppies
> > bug; sure looks like boot-floppies is the culprit.
> 
> I guess "generic-m68k" should build those devices, like it did in
> previous versions. Don't you think so? Oh, wait, there's no generic-m68k
> target in /dev/MAKEDEV , just generic-{i386,powerpc} ...

There's no simple 'generic' catchall either it seems. 

> OTOH, I prefer using specific device names on rootdisks.sh and basedisk.sh.
> That's more robust than relying on makedev aliases.

Agreed. I just have to figure out what other devices are missing now. And 
we should have a makedev alias in case someone messes up /dev/ ...
> > > - mac-us-ext keymap isn't a valid binary map (we had trouble with
> > > _incomplete_ binary maps that wiped the kernel keytable on Atari
> > > before; this is new).
> 
> I guess it's a bug in writemaps. It's strange that it have gone unnoticed
> all this time. Perhaps something has changed in the mac keymaps recently?

Nope, the Atari map isn't accepted either. Last change there was what made
it
work for the 2.0 boot floppies (Roman created a full map, the Atari maps had 
been partial maps before). No change since then AFAIK.

> > > - the 'configure modules' step complains about 'sed: not found',
> > > configuring modules is impossible, dinstall keeps prompting that
> > > step again and again.
> >
> > I thought this was fixed in modconf 0.2.21?
> 
> Indeed. Which version are you using for m68k?

Beats me. I'll check that as soon as I've rebooted that Mac in Linux. Might 
not happen until real late today, it's in use :-(

I'll send a few other diffs (update for the basedisks makedev arguments, 
and a slight enhancement to check_mirror.sh to report _all_ missing packages
in one single run instead of only one) later anyway. And I'm puzzled over
the 
lowmem business. Care to explain the magic there? I've built lowmemrd.bin, 
booted it and nothing happens. 

	Michael


Reply to: