From owner-freebsd-embedded@FreeBSD.ORG Mon Dec 19 06:44:51 2011 Return-Path: Delivered-To: freebsd-embedded@FreeBSD.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 65A721065675; Mon, 19 Dec 2011 06:44:51 +0000 (UTC) (envelope-from imp@bsdimp.com) Received: from harmony.bsdimp.com (bsdimp.com [199.45.160.85]) by mx1.freebsd.org (Postfix) with ESMTP id 1382F8FC13; Mon, 19 Dec 2011 06:44:50 +0000 (UTC) Received: from [10.0.0.63] (63.imp.bsdimp.com [10.0.0.63]) (authenticated bits=0) by harmony.bsdimp.com (8.14.4/8.14.3) with ESMTP id pBJ6gLj5041873 (version=TLSv1/SSLv3 cipher=DHE-DSS-AES128-SHA bits=128 verify=NO); Sun, 18 Dec 2011 23:42:23 -0700 (MST) (envelope-from imp@bsdimp.com) Mime-Version: 1.0 (Apple Message framework v1084) Content-Type: text/plain; charset=us-ascii From: Warner Losh In-Reply-To: Date: Sun, 18 Dec 2011 23:42:21 -0700 Content-Transfer-Encoding: quoted-printable Message-Id: <2CBD8651-E132-49DC-A082-37A8F5C626EA@bsdimp.com> References: <0F6CC18F-6973-42A2-AC03-F01BF59458AE@lassitu.de> <1100F70E-9DA9-4163-AC9A-423ECE5AA9A3@lassitu.de> <18CABB46-9B9A-41CB-8742-6723C5FF4D67@lassitu.de> To: Adrian Chadd X-Mailer: Apple Mail (2.1084) X-Greylist: Sender succeeded SMTP AUTH, not delayed by milter-greylist-4.0.1 (harmony.bsdimp.com [10.0.0.6]); Sun, 18 Dec 2011 23:42:23 -0700 (MST) Cc: Oleksandr Tymoshenko , "freebsd-embedded@freebsd.org" Subject: Re: Updated switch/glue patch? X-BeenThere: freebsd-embedded@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Dedicated and Embedded Systems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 19 Dec 2011 06:44:51 -0000 On Dec 18, 2011, at 9:20 PM, Adrian Chadd wrote: > On 18 December 2011 16:29, Warner Losh wrote: >=20 >> What is it without WITNESS? The performance numbers with WITNESS are = generally uninteresting for a deployed system. >=20 > It's still ridiculously CPU intensive. :( It turns out trying to do > those locks three times one each GPIO IO line fiddle is a bit over the > top. That's the real problem. WITNESS is a known pigdog on low-end systems, = so measurements with it tend to skew the results a tad too much. On the = other hand, it did magnify the problem to the point it was fixed :) The insane locking should be fixed. Warner