On 14 Jan 2002, Eric W. Biederman wrote:

> Well I don't think openboot/openfirmware has text based tables but
> I may be wrong there.  My honest opinion is that we are probably best
> off defining our own table structures, and doing the corresponding kernel
> patches.  That way we don't have other people's legacy junk getting
> in the way :)

Maybe my memory is too fuzzy. Have you used a SPARC recently? I have not.
But my old memory says that all the devices on a sparc were defined in
textual strings in the forth boot prom. It seemed that linux had code to
parse this in the sparc tree. I will now walk across the hall and try my
friend's mac ... I can't get it into an OF prompt so who knows.

Actually though it used to be that if you walked the / tree in
sunos/solaris you saw a textual representation of all the machine
resources. This directory tree was yanked directly from the Open Firmware
strings. It might be worth a look.

I hate to bring this up as I know you've been working very hard on your
tables, but at the same time I figure we should take a look at it.

That's also not to say that the Open Firmware tables are the right way.
It's more a question which Werner asked that got me to thinking: should
they be textual tables? Also I've been using Plan 9 a lot in the last year
and have begun to like text-based things like that.

I turned on some debug and found that we're getting through memory, the
regions look good, and I get at least to:  hardware_setup and finish it
OK.

ron

Reply via email to