On Mon, 8 Jun 1998, David A. Gatwood wrote:

> On Mon, 8 Jun 1998, Nathan Raymond wrote:
> 
> > On Mon, 8 Jun 1998, Erik E. Fair wrote:
> > 
> > > The "ofpcc" is for systems with correct and complete OpenFirmware (IEEE
> > > 1275), a category that, alas, does not include any Apple products. After
> > > all, that sort of thing might make it easier to boot an OS other than
> > > MacOS...
> > 
> > PCI PowerMacs *do* have OpenFirmware - Apple had to impliment OF for PCI
> > support.
> > 
> > OF is how LinuxPPC works - MkLinux uses a booter like NetBSD, but LinuxPPC
> > does NOT need the MacOS at all to boot (and it only works on PCI PowerMacs
> > of course).  You have to break into OF early at startup and change some
> > variables at the command line to boot into LinuxPPC.  No reason why NetBSD
> > PPC couldn't do the same.
> 
> I think the key here was correct and complete.

Point taken.

> For instance, the G3
> PowerMacs, unless the recent models have fixed the problem, can't boot
> from their internal IDE drives via OpenFirmware because of an OF bug.  I
> read in the MkLinux sources that... either the 6400 or 6500, I believe...
> didn't return certain information about the IDE drives... to which Michael
> Burg's comment was fudge it.  I'd almost bet that they cut corners in
> other places as well, and that other bugs exist....  :-|

However, Apple has simply made it more difficult - not impossible - to
boot other operating systems with some of their OF implimentations.

>From <http://www.linuxppc.org/help/install_help/>

     Older machines, such as the 7200, 7300, 7500, 8200, 8500, etc.
     usually have version 1.0.5 of Apple's Open Firmware implementation.
     Newer machines (e.g. the 6400, etc.) and most clones have version
     2.0. There are some problems with the older version (1.0.5):
     
     * It doesn't have a driver for the screen display on the 7200, 8200
     and clones (e.g. the PowerCenter 150). If you don't have a terminal
     or emulator to connect to the serial port, you can still set
     input-device to 'kbd' so that you can issue commands to OF's user
     interface, as long as you don't mind typing blind.
     
     * It doesn't define a 'screen' alias, so you have to know the OF
     device name for your screen. On the 7300/7500/8500, this is
     '/chaos/control' (or alternatively 'vci/@B', which amounts to the
     same thing). It does have a driver for the screen display on these
     machines, but the driver often doesn't initialize the screen
     successfully.
     
     * The OF variables get reset to default values every time MacOS
     starts up.
     
     * In a device name such as 'scsi/sd@1:2', the number after the
     colon is the partition number. If you put a non-zero partition
     number to specify a partition explicitly, OF will fall over when it
     tries to open the device. Putting zero (which means the first
     bootable partition) works OK.
     
     These problems don't seem to be present in version 2 of OF.
     However, it seems that version 2 tends to allocate virtual
     addresses for the frame buffer starting at 0, which causes
     problems. The workaround for this is to set the load-base variable
     to 100000 (the 1MB point). This seems to be necessary particularly
     on the 6400 and similar machines (5400, 5500, etc.).
     
     Also, on the PowerBook 3400 (with OF version 2.0.1), to boot off
     the internal IDE hard disk, it seems to be necessary to give a
     boot-device value such as 'ata0/ata-disk@0,0:8'. Note the ',0' and
     the explicit partition number; the partition number is hexadecimal,
     so partition 10 would be 'ata0/ata-disk@0,0:a'.


Don't seem like big hurdles to me...

--
Nathan Raymond
http://www.everythingmac.com