Hey,

> > the kernel was compiled with optimization flags (-maltivec is included
> > just by using the option above) and reports the cpu as:
> >
> > cpu0 at mainbus0: 7400 (Revision 2.7), ID 0 (primary)
> > cpu0: HID0 8294c0a4<EMCP,ECLK,DOZE,DPM,EIEC,ICE,DCE,SGE,BTIC,BHT>
> > cpu0: 409.10 MHz, no-parity 1MB WB L2 cache (PB SRAM) at 2:1 ratio
> 
> I'm still wondering why it probes as ~410MHz.
me too, results with the powerlogix software deviate +-1 mhz...

 
> > - I don't know if I have a backside cache. To be honest I don't even
> > know exactly what a backside cache is. I thought it was essentially
> > 1:2 cache on the cpu or the cpu doughterboard.
> 
> You do and it essentially means 'cache on a separate bus' - as opposed
> to ' cache between CPU and mainboard'.


should I try to make a kernel with cache enable options but no backside?
could that be useful? From what you say I gather it would be stupid,
since I have to set up the correct type of cache.

 
> > - powerlogix states that speculative addressing could be a problem on
> > non-g3/g4 computers which get such a processor. But My question is
> > "why" ?
> 
> No idea, I think it's enables here. ( Phase5 G3 card though )
> See arch/powerpc/oea/cpu_subr.c for more flags to play with.

the question is if you or other 604/604e (not ev) owners have that
enabled when they upgraded to g3/g4.
 
> Are there any switches on the card?

I didn't see any.

> > the powerlogic utility reports the L2CR bytesin hex that it will use.
> > Could it be useful to get those? netbsd should come up with the same
> > values?
> 
> Most definitely and yes I think so. It would at least tell us what MacOS
> uses.

0xB9080000

I wonder to what netbsd sets them with my settings...

ah I compiled another kernel (very tedious, I need to boot your old 2.99
kernel, of which I don't have the configuration, which is slow  of
course had has broken mc0) specifying -mcpu-7400 and -O3 in mk.conf
(instead of passing them on the command line)
	- the kernel is smaller than my previous one
	- X11 works with the standard X server and has no color bandings!
	- it seem a tad more stable, but the ssytem freezed after a cvs update,
a bit of compile and X usage.
	- ppc kernels freeze in X just freeze, like good ol' mac! they don't
drop back into console like sparcs.

I gather something is flakey :) think at the X stuff!

-R