Subject: Re: MOL without OF
From: Samuel Rydh (samuel@ibrium.se)
Date: Fri Dec 08 2000 - 13:24:12 MST
On Fri, Dec 08, 2000 at 03:38:24PM +0100, Benjamin Herrenschmidt wrote:
> >Does this work theoretically? During all the time we have tried to get MOL
> >working on APUS we never thought about it, until mol.o for 2.4 complained
> >about find_path_device being unresolved.
>
> I don't have a firm answer to what MOL does with this function, but some
> time ago, Paul and I discussed about what to do with those OF functions
> on non-OF machines, and the result of the discussion was that everybody
> should have them, the device-tree just beeing empty on machines that
> don't support OF (so find_path_device would always return NULL).
>
The kernel OF tree is not used, with the following execptions:
- Some cpu properties are copied from the kernel tree to
the file loaded one (timebase-frequency, bus-frequency and
clock-frequency).
- The emulation/timer.c file uses the 'cpu-version' and
'timebase-frequency' properties to find out about the timebase
frequency.
- In the oldworld setting, the kernel OF tree is used to
verify the existence of a built-in ROM if no ROM image
is specified.
That is, the kernel OF tree is not used for anything vital
although the timer calibration should be fixed for best
performance. On my machine, MOL boots if I set the kernel
OF tree to NULL.
/Samuel
----------------------------------------------------------
E-mail <samuel@ibrium.se> WWW: <http://www.ibrium.se>
Phone/fax: (home) +46 8 4418431, (work) +46 8 7908470
----------------------------------------------------------
This archive was generated by hypermail 2a24 : Fri Dec 08 2000 - 17:12:44 MST