From owner-freebsd-hackers Wed Nov 8 10:31:42 2000 Delivered-To: freebsd-hackers@freebsd.org Received: from prism.flugsvamp.com (cb58709-a.mdsn1.wi.home.com [24.17.241.9]) by hub.freebsd.org (Postfix) with ESMTP id CD45637B479; Wed, 8 Nov 2000 10:31:36 -0800 (PST) Received: (from jlemon@localhost) by prism.flugsvamp.com (8.11.0/8.11.0) id eA8ISUj13784; Wed, 8 Nov 2000 12:28:30 -0600 (CST) (envelope-from jlemon) Date: Wed, 8 Nov 2000 12:28:30 -0600 (CST) From: Jonathan Lemon Message-Id: <200011081828.eA8ISUj13784@prism.flugsvamp.com> To: asmodai@FreeBSD.ORG, hackers@FreeBSD.ORG Subject: Re: PSE/PAE support X-Newsgroups: local.mail.freebsd-hackers In-Reply-To: Organization: Cc: Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG In article you write: >Reading some documents and sources I came to the following conclusion: > >We support 4 Kbyte pages and 4 Mbyte pages, but we do not support 2 >Mbyte pages. On IA-32 at least. > >Is there a reason to? > >I could understand that some people would prefer 2 Mbyte pages in some >situations. But looking at pmap.c I see we only test CPUID_PSE and >don't use CPUID_PAE anywhere. My initial feeling is that it wouldn't win us anything. We use a two-level page table, with the first level covering a 4MB page, and the second level containing 1024 pte's, each covering 4K. Since each pte takes up 4 bytes, by using a 4MB page, we eliminate an entire page of pte's at the second level. With a 2M page, we would still need a second level page entry, plus some code to differentiate between 4K entries and a "half-page" 2M entry. But I'm by no means a vm hacker, so I may be wrong here. -- Jonathan To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message