From owner-freebsd-hackers Tue Oct 31 4:32:44 2000 Delivered-To: freebsd-hackers@freebsd.org Received: from hda.hda.com (host65.hda.com [63.104.68.65]) by hub.freebsd.org (Postfix) with ESMTP id 1AC7037B4C5 for ; Tue, 31 Oct 2000 04:32:39 -0800 (PST) Received: (from dufault@localhost) by hda.hda.com (8.9.3/8.9.3) id HAA31834; Tue, 31 Oct 2000 07:35:31 -0500 (EST) (envelope-from dufault) From: Peter Dufault Message-Id: <200010311235.HAA31834@hda.hda.com> Subject: Re: Filesystem holes In-Reply-To: <200010310907.e9V97mk17233@earth.backplane.com> from Matt Dillon at "Oct 31, 2000 01:07:48 am" To: Matt Dillon Date: Tue, 31 Oct 2000 07:34:16 -0500 (EST) Cc: Terry Lambert , Ryan Thompson , freebsd-hackers@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL61 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > ... Sparse matrixes are the big math problem > that benefit, but only because the solution to a sparse matrix problem > is not even close to random so the sparse matrix winds up still being > sparse all the way to the end of the solution. I use them for bus simulations, which also are permanently sparse. It would be nice to free up the regions when I "remove" a virtual board, but in a check through POSIX I could find nothing defined to behave that way either for mapped files or mapped memory objects. Also, a write from any process would repopulate the region which I really wouldn't want but I don't see that level of control over mapping between unrelated processes (Now I start thinking about MAPFIXED to a specified virtual address and implementing funny tricks but I don't have time to work on that). In my case I'd be better off with shared memory objects that aren't persistent but appear in the name space so that I don't accidentally start copying a virtual bus file when the programs exit improperly. In the sparse matrix calculations with no checkpointing or need to appear in a name space I'd think the best thing would be to use VM with the matrix initially mapped to a copy on write zero page. I guess you can't do that without mmap because of swap allocation. Peter -- Peter Dufault (dufault@hda.com) Realtime development, Machine control, HD Associates, Inc. Fail-Safe systems, Agency approval To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message