From owner-freebsd-mips@FreeBSD.ORG Thu Nov 15 18:25:28 2012 Return-Path: Delivered-To: mips@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id B179DBF0; Thu, 15 Nov 2012 18:25:28 +0000 (UTC) (envelope-from alc@rice.edu) Received: from mh3.mail.rice.edu (mh3.mail.rice.edu [128.42.199.10]) by mx1.freebsd.org (Postfix) with ESMTP id 82D2C8FC19; Thu, 15 Nov 2012 18:25:28 +0000 (UTC) Received: from mh3.mail.rice.edu (localhost.localdomain [127.0.0.1]) by mh3.mail.rice.edu (Postfix) with ESMTP id 74A83401E6; Thu, 15 Nov 2012 12:25:22 -0600 (CST) Received: from mh3.mail.rice.edu (localhost.localdomain [127.0.0.1]) by mh3.mail.rice.edu (Postfix) with ESMTP id 73A9E401CB; Thu, 15 Nov 2012 12:25:22 -0600 (CST) X-Virus-Scanned: by amavis-2.7.0 at mh3.mail.rice.edu, auth channel Received: from mh3.mail.rice.edu ([127.0.0.1]) by mh3.mail.rice.edu (mh3.mail.rice.edu [127.0.0.1]) (amavis, port 10026) with ESMTP id czahMvILN02y; Thu, 15 Nov 2012 12:25:22 -0600 (CST) Received: from adsl-216-63-78-18.dsl.hstntx.swbell.net (adsl-216-63-78-18.dsl.hstntx.swbell.net [216.63.78.18]) (using TLSv1 with cipher RC4-MD5 (128/128 bits)) (No client certificate requested) (Authenticated sender: alc) by mh3.mail.rice.edu (Postfix) with ESMTPSA id E37F84018B; Thu, 15 Nov 2012 12:25:21 -0600 (CST) Message-ID: <50A53391.4080909@rice.edu> Date: Thu, 15 Nov 2012 12:25:21 -0600 From: Alan Cox User-Agent: Mozilla/5.0 (X11; FreeBSD i386; rv:16.0) Gecko/20121111 Thunderbird/16.0.2 MIME-Version: 1.0 To: mips@freebsd.org Subject: ZERO_REGION_SIZE Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: "Jayachandran C." X-BeenThere: freebsd-mips@freebsd.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: Porting FreeBSD to MIPS List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 15 Nov 2012 18:25:28 -0000 Given the possibility of L1 cache aliasing on some (many?) MIPS processors, I think that you, i.e., MIPS users, should evaluate the performance effects of the following change: Index: mips/include/vmparam.h =================================================================== --- mips/include/vmparam.h (revision 243091) +++ mips/include/vmparam.h (working copy) @@ -190,6 +190,6 @@ */ #define VM_NFREEORDER 9 -#define ZERO_REGION_SIZE (64 * 1024) /* 64KB */ +#define ZERO_REGION_SIZE 4096 #endif /* !_MACHINE_VMPARAM_H_ */ You can quantify the impact with a simple test like: dd if=/dev/zero of=/dev/null bs=<64 or 128>k count= If possible, please use a kernel without WITNESS or INVARIANTS. Alan P.S. I would also be curious how setting this to 8192 would perform.