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

Re: Stuck on install of 7600/120



On Wed, May 30, 2001 at 11:27:44AM -0500, Michael Shuey wrote:
> > > I picked "make bootable directly from hard drive" in the Debian installer
> > 
> > read the installation docs, its clearly stated that this step is
> > broken and cannot work on its own. =20
> 
> Okay, I'm kinda dense.  There is a paragraph in the install docs about this.
> It's not complete, but it did give me some pointers.

i know, there is also a brownbag, it says:

nvsetenv `ofpath /dev/sda3`

that should be 

nvsetenv boot-device `ofpath /dev/sda3`  

but there is another issue, some macs can't boot specific partitions,
only the pseudo partition 0.  and the ofpath in potato is broken
outputing some garbage in its names. so

ofpath /dev/sda

take its output and stick a 0 at the end of it, then use that for:

nvsetenv boot-device <path> (no <>)

> Why can't some of the necessary OF majik be run from the bootdisks?

the woody boot floppies will at the very least set the boot-device
variable, unfortunatly i think the 7200 is the only machine where that
is enough, most of the others require a ritual goat sacrifice which is
somewhat tricky to implement in C code.  

> > there is no vmlinuz on powermac, no powermac bootloader supports
> > compressed images, we just use the plain uncompressed ELF for now. =20
> 
> I noticed.  You should probably fix the bootdisks for the next potato
> re-release; they make quik.conf load /vmlinuz, which is wrong.

er, no i don't think so.. /vmlinux yes, which is still wrong since
thats a symlink and quik is broken about symlinks.  woody will fix
that though.  

> > then take that string and give it to nvsetenv:
> > 
> > nvsetenv boot-device /bandit/gc/mesh/sd@0:0
> 
> That horribly broke things.  A few useful things you might want to add to
> the install docs:

yes see above about ritual goat sacrifice.  

> command-option-p-r at boot resets the PRAM and NVRAM (useful if you totally
> screw it up)

i will add a note about this, ill see about potato..

> If you have an old-world Mac, read www.netbsd.org's macppc System Disk
> Tutorial (http://www.netbsd.org/Ports/macppc/SystemDisk-tutorial/).  Since
> the OF on oldworlds is buggy (to be polite) you should really run Apple's
> OF updater first.  Also, you may need to patch the OF updater (manually, see
> http://www.netbsd.org/Ports/macppc/SystemDisk-tutorial/of105patch.html )
> if you want to boot off a non-Apple disk.  I needed some extra cruft in my
> nvramrc to retry booting several times since the disk I was installing to
> takes a few seconds to spin up.

this is one of the unfortunate reasons why the `make system bootable'
step is never going to be that reliable, we can't install these
patches ourselves (1 they are non-free, 2 we don't know how it detects
the exact machine revision)

> If you don't see the OF on the screen, try the modem port at 38400 bps, 8N1.
> 
> If you've got an oldworld Mac between the 7300-8500 (possibly earlier as well)
> you'll probably need the patched system disk (see URLs above) to get the
> console OF to work reliably.  Apparently OF tries to display before the video
> timings are all set...

actually i think it sets them to something rediculous that no monitor
supports. 

> And last (but not least), there's not enough space in the nvramrc for both
> the disk spinup patch and the console/OF patch.  Choose wisely.

heh

> > that MIGHT work, oldworld OF requires a lot of witchcraft to make
> > boot most of the time... =20
> 
> No kidding!
> 
> So it boots now, and life is beautiful.  Next question: is there any way to
> write out an nvramrc from Linux?  I've tried using nvsetenv, but writing
> out 40 lines of FORTH code, complete with carriage returns, doesn't seem to
> work.

nvsetenv nvramrc > nvramrc

that will only output the contents of nvramrc and redirect it to a
file. in thoery:

nvsetenv nvramrc "$(cat nvramrc)" 

would restore it, but that is untested.  

-- 
Ethan Benson
http://www.alaska.net/~erbenson/

Attachment: pgp_N8O7KIkvr.pgp
Description: PGP signature


Reply to: