Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!usc!samsung!emory!mephisto!udel!mmdf
From: eo@ansa.co.uk (Ed Oskiewicz)
Newsgroups: comp.os.minix
Subject: (ST/1.5.10) Increasing buffer pool size causes problems
Message-ID: <22078@nigel.udel.EDU>
Date: 15 Jun 90 10:35:36 GMT
Sender: mmdf@udel.EDU
Lines: 29

So it went like this, 

I thought I'd run without a ram disk, by putting root onto a winchester
partition and reclaiming some of the extra space for an increased buffer pool.

I changed the parameters in boot.h and recompiled minix, booted from the new
system and it worked fine (my winchester partition was what I had previously
used as the ram disk template). Then I increased NR_BUFS to 160 (from 30) and
NR_BUF_HASH to 256 (from 32), recompiled everything and tried again. This time
total disaster, after booting lots of garbage was written to the screen and a
panic ensued.

I assume the problem is either that the buffer pool is too large and that
something breaks at 32, 64 or 128K OR that minix has got too large and that
init is too far away from something. Has anyone seen this problem before? if
so could you save me some work by pointing me at the answer.

To facilitate using this new organization I have written a shutdown(8) which
uses gawk and ps to kill off processes and sync and umount discs. I will post
it in the next week or so.

Ed Oskiewicz

email:	eo@ansa.co.uk (...ukc!ansa!eo)		ANSA,
tel:	+44 223 323 010				Poseidon House,
fax:	+44 223 359 779				Castle Park,
						Cambridge,
						United Kingdom CB3 0RD
---