Subject: Re: RAM not detected properly
To: None <thorpej@nas.nasa.gov>
From: Andrew Brown <codewarrior@daemon.org>
List: port-i386
Date: 11/15/1996 15:00:19
> From: Jason Thorpe
>
>On Sat, 9 Nov 1996 22:53:42 +0530 
> "HARAWAT.IN.ORACLE.COM" <HARAWAT.IN.ORACLE.COM.ofcmail@in.oracle.com> wrote:
>
> > 	I have a compaq deskro machine which was initially having 16 Mb of 
> > RAM, I upgraded the RAM to 32 Mb, now the machine while coming up shows
> > 32 Mb 
> > of RAM, but the netbsd1.1 kernel still uses only 16 Mb of RAM, any ideas
> > what 
> > could be the possible reasons for this, from where the kernel picks the
> > amount 
> > of RAM in  the machine. 
>
>Some BIOSes don't report RAM correctly.  NetBSD 1.1 didn't have an easy
>way of setting this, but NetBSD 1.2 does; use the EXTMEM_SIZE kernel
>option:
>
>options		EXTMEM_SIZE=(31*1024)
>
>that accounts for 640k + hole + 31M of extended memory.

now this makes some sense to me but this option (as far as i have seen,
disregarding people saying "do this!") is not very well documented at all.
can someone put together some notes about what this number should be and
why, what that number that netbsd prints when it bootstraps means (example
from my machine which has 16M),

   Nov 12 19:32:28 noc /netbsd: real mem  = 16384000
   Nov 12 19:32:28 noc /netbsd: avail mem = 14016512

what to do with NKPED, NKMEMCLUSTERS, and NMBCLUSTERS and why and what they
affect?  i've looked around at NKPDE and i see it used in only one place
and i'm not well-enough versed in the kernel code to go "oh!  i see...".

i'm trying to upgrade machines from 64M (works fine on 1.1 and 1.2) to
96 or 128 and i can't get rid of both the ptdi panic or the vm_fault
panic.

-- 
|-----< "CODE WARRIOR" >-----|
andrew@echonyc.com (TheMan)        * "ah!  i see you have the internet
codewarrior@daemon.org                               that goes *ping*!"
warfare@graffiti.com      * "information is power -- share the wealth."