Subject: Re: -current compile problems for pmax.
To: Simon Burge <>
From: Castor Fu <>
List: port-pmax
Date: 01/15/1999 15:44:28
I had forgotten that there were a lot of changes in the pmax tree itself
as part of pmax-nisimura-wscons.   The pmax port is currently broken.  Sorry.

Since it'd be really good to get as much of this branch merged in as
possible, I'll continue merging bits from the pmax tree until I hear from 
Tohru, and until I get a kernel which compiles cleanly in my cross environment.
I'll try to minimize bringing in changes related to the wscons code.

Sorry about the confusion.


On Sat, 16 Jan 1999, Simon Burge wrote:

> I've got the following problems with -current (after your latest updates
> as of a few hours ago).
> + No definition of INT_MASK_REAL_DEV used in mips/mips/trap.c.  Should
>   this be MIPS_HARD_INT_MASK?

I believe the intent of this was to allow the mips3 cp0 count interrupt
to be handled separately from the hardware interrupt handler.  I had
assumed this code was already in pmax.  I will check it and the other 

> + _Lots_ of duplication (and warnings) between mips/include/cpuarch.h
>   and mips/include/cpuregs.h.
> + pmap_set_referenced() in mips/mips/pmap.c doesn't seem to be needed?

Not my code.  I'm quite willing to pitch it, but thought Tohru might
be in the middle of something.

> + mips_flushcache_allpvh() in mips/mips/pmap.c doesn't seem to be
>   needed?

This could be guarded by an ifdef.  It's definitely needed for the
machines without L2 caches which lack cache coherency.

> + Missing function mips1_clean_tlb()?  Called in
>   mips/mips/mips_machdep.c

Uhh, Tohru might have gc'd it.  I don't have any running mips1 machines.