No subject

[postmaster@bharat.ColumbiaSC.NCR.COM postmaster@bharat.ColumbiaSC.NCR.COM]

>From root Thu Mar 14 14:00 EST 1996 remote from bharat
>From 9fans Thu Mar 14 13:29:49 0500 1996 remote from cse.psu.edu
Received: from cse.psu.edu by bharat.ColumbiaSC.NCR.COM; Thu, 14 Mar 96 14:00 EST
Received: by ncrcae.ColumbiaSC.NCR.COM; 14 Mar 96 13:59:58 EST
Received: from cse.psu.edu by casey.ColumbiaSC.NCR.COM; Thu, 14 Mar 96 13:58 EST
Received: from ncrgw1.UUCP (uucp@localhost) by ncrhub4.attgis.com (8.7.3/8.7.3) with UUCP id NAA04358 for casey.columbiasc.ncr.com!karth; Thu, 14 Mar 1996 13:58:09 -0500 (EST)
Received: by ncrgw1.ATTGIS.COM; 14 Mar 96 13:42:46 EST
Received: by colossus.cse.psu.edu id <78688>; Thu, 14 Mar 1996 13:30:19 -0500
Received: from krystal.com ([205.230.227.129]) by colossus.cse.psu.edu with SMTP id <78695>; Thu, 14 Mar 1996 13:30:02 -0500
Received: by krystal.com (8.7.3/8.6.9) id MAA27301; Thu, 14 Mar 1996 12:29:49 -0600 (CST)
From: cse.psu.edu!9fans 
Date: Thu, 14 Mar 1996 13:29:49 -0500
Message-Id: <199603141829.MAA27301@krystal.com>
To: cse.psu.edu!9fans  
>From: bsdi.com!prb (Paul Borman)
Subject: re: cant start 9pcfs
Content-Length: 1416
Content-Type: text
Auto-Forwarded-From: casey!karth
Auto-Forward-Count: 1
Sender: owner-9fans@cse.psu.edu
Precedence: bulk
Reply-To: 9fans@cse.psu.edu

> >>This is most certainly due to a memory sizing problem (heaven knows
> >>we have made the rounds with this one with BSD/OS...)  For now I
> 
> are the bytes in the CMOS that give the extended memory size not standard,
> or do some machines not set them correctly?

Well, we have found that many (all?) Dell machines think that the CMOS
can never report more than 16MB, so it doesn't, even if you have more
than 16MB of memory.  Then there are machines which we can't accurately
probe for how much memory there is, but the CMOS is right!  There are
also weird caching affect (i.e. you can read/write memory beyond the
end of memory as long as it stays in cache) as well as new and unique
ways to remap addresses that are beyond the end of memory.  To give you
and idea, here are the various parameters one can set in their boot.default
file for BSD/OS:

	-basemen mem	Assume this much base memory (memory below 1M)
	-cmosmem	Limit memory search to the mount given by the CMOS
	-extended mem	Limit the amount of memory to check
	-memsize mem	Just assume this it the amount of memory
	-noflushcache	Don't flush the cache while sizing memory

Each one of these is needed for at least one machine that we have had
problems with.  I should note that we can probe the proper amount of
memory on *most* machines, but...

Finding out how much memory you have is certainly a black art.

				-Paul Borman
				 prb@bsdi.com