From owner-cvs-all Fri Aug 2 20:12:44 2002 Delivered-To: cvs-all@freebsd.org Received: from mx1.FreeBSD.org (mx1.FreeBSD.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 117C537B401; Fri, 2 Aug 2002 20:12:38 -0700 (PDT) Received: from canning.wemm.org (canning.wemm.org [192.203.228.65]) by mx1.FreeBSD.org (Postfix) with ESMTP id 8C86F43E42; Fri, 2 Aug 2002 20:12:37 -0700 (PDT) (envelope-from peter@wemm.org) Received: from wemm.org (localhost [127.0.0.1]) by canning.wemm.org (Postfix) with ESMTP id 78CDA2A7D6; Fri, 2 Aug 2002 20:12:37 -0700 (PDT) (envelope-from peter@wemm.org) X-Mailer: exmh version 2.5 07/13/2001 with nmh-1.0.4 To: David Greenman-Lawrence Cc: cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/i386/i386 pmap.c In-Reply-To: <20020802181752.C54567@nexus.root.com> Date: Fri, 02 Aug 2002 20:12:37 -0700 From: Peter Wemm Message-Id: <20020803031237.78CDA2A7D6@canning.wemm.org> Sender: owner-cvs-all@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG David Greenman-Lawrence wrote: > >peter 2002/08/02 18:02:37 PDT > > > > Modified files: > > sys/i386/i386 pmap.c > > Log: > > Take advantage of the fact that there is a small 1MB direct mapped region > > on x86 in between KERNBASE and the kernel load address. pmap_mapdev() > > can return pointers to this for devices operating in the isa "hole". > > Actually the mapping of the low 1MB of physical address space was a Jolitz > abortion which I had hoped to someday remove. > > -DG Indeed. I do not particularly like it but it seemed fair game for this well contained case. It isn't exposed outside of pmap.c and is dead simple to turn back off again. Mapping this 1MB block of memory at KERNBASE and then also adding it to the general page pool does make me a bit nervous. On a similar topic, we have a couple of weird TLB things happening on certain ppro and early pentiumII steppings that look suspiciously related to one of the errata that boils down to 'Do Not point a 4MB page at physical address zero.' While we are not doing exactly what is described there, I have noticed that the wierd(TM) things happen to happen on the same steppings so far. Admittedly it was from a very small sample. I'd like to be able to configure the kernel to load at 4MB instead of 1MB and use whole 4MB pages for the kernel and the initial data, instead of effectively losing 1MB from the 4MB tlb slot. That requires a cleanup of a fair few places, including some drivers that "know" the relationship between the isa hole and KERNBASE. Even vm86 "knows" more than is good for it. Cheers, -Peter -- Peter Wemm - peter@wemm.org; peter@FreeBSD.org; peter@yahoo-inc.com "All of this is for nothing if we don't go to the stars" - JMS/B5 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message