*BSD News Article 34293


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!yarrina.connect.com.au!werple.apana.org.au!troll.apana.org.au!phil
From: phil@troll.apana.org.au (Phil Homewood)
Newsgroups: comp.os.386bsd.questions
Subject: Re: FreeBSD SoundBlaster support
Date: 14 Aug 1994 06:35:55 GMT
Organization: - The APANA News Hub Brisbane - Australia
Lines: 26
Distribution: world
Message-ID: <32ke0b$dvp@troll.apana.org.au>
References: <321lar$k58@qualcomm.com> <323cbg$mo9@agate.berkeley.edu> <323g2o$93c@sandra.teleport.com> <32g39d$ped$2@heifetz.msen.com>
NNTP-Posting-Host: diamond.apana.org.au
X-Newsreader: TIN [version 1.2 PL2]

Operator (aka Mark Taylor) (root@cybernet.com) wrote:
: I think that the 640Kbyte limit is on the DATA+TEXT segment size. 
> There is some
: stuff in the kernel that is neither (?).  I've got a 1.1.5.1 kernel which
: has SoundBlaster, ethernet, SCSI, and NFS, etc..., support which is
: 670679 bytes long.

The default load address for 1.1.5.1 has changed to 1MB I think, 
thus eliminating the ugly 640K limitation. The kernel just needs to be
able to load into contiguous address space (I believe) to function
correctly.

Back before 1.1.5.1, the kernel defaulted to loading low, which meant
you couldn't load a kernel >640K unless you hacked the kernel makefile
to tell it to load high :)

My kernel, which has been up and running reliably (23 days uptime now ;)
has the following output from 'size':

text	data	bss	dec	hex
487424	32768	145472	665664	a2840

This is just on 650K, and works like a dream.

Phil.
"640K will be enough - nobody will ever need that much!"