From owner-freebsd-questions Fri Oct 6 03:33:33 1995 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id DAA22423 for questions-outgoing; Fri, 6 Oct 1995 03:33:33 -0700 Received: from ref.tfs.com (ref.tfs.com [140.145.254.251]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id DAA22416 for ; Fri, 6 Oct 1995 03:33:31 -0700 Received: (from julian@localhost) by ref.tfs.com (8.6.11/8.6.9) id DAA07467; Fri, 6 Oct 1995 03:32:17 -0700 From: Julian Elischer Message-Id: <199510061032.DAA07467@ref.tfs.com> Subject: Re: kernel options for > 64mby To: davidg@Root.COM Date: Fri, 6 Oct 1995 03:32:16 -0700 (PDT) Cc: elh_fbsd@spnet.com, questions@freebsd.org, elh@spnet.com In-Reply-To: <199510061026.DAA00423@corbin.Root.COM> from "David Greenman" at Oct 6, 95 03:26:06 am X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 1314 Sender: owner-questions@freebsd.org Precedence: bulk > > >Under OSF I made it go hunting if it thought that there might be more.. > >basically, if the BIOS says 64MB (well I use >=63MB as I don't trust Bios's) > >I do memory tests upwards till I start getting failures.. > > > >The trick is to do this BEFORE YOU SWITCH ON PAGING > >so that you don't get pagefaults.. > > > >I could send you the code but it PROBABLY doesn't apply directly, > >and you'd probably do better implimenting it yourself once you've > >heard the idea of how I did it.. > > Whether or not you have paging enabled isn't a problem. The problem is > that some (Compaq) motherboards will happily generate an NMI if the parity > isn't read back correctly for non-existent memory. The only correct way to > do this is to get the correct size by asking the BIOS in the correct way > for it. We have code that will do this, but it bloats our bootblocks to the > point that they won't fit anymore. > > -DG > well if you get an NMI you can pretty safely assume it's past the end of memory then, as you just WROTE there as part of the ram test.. works like a charm here.. I did it before paging to avoid having to fiddle with pagefaults, (YUK) but I could handle ignoring NMI traps. (i.e. just set a flag saying it happenned and return) I'd say the correect way is to probe memory.. :)