From owner-freebsd-usb@FreeBSD.ORG Sun Aug 15 19:31:48 2010 Return-Path: Delivered-To: freebsd-usb@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id BF5761065674; Sun, 15 Aug 2010 19:31:48 +0000 (UTC) (envelope-from prt@prt.org) Received: from smtp5.uk.umis.net (smtp5.uk.umis.net [217.65.166.40]) by mx1.freebsd.org (Postfix) with ESMTP id 86C7C8FC13; Sun, 15 Aug 2010 19:31:48 +0000 (UTC) Received: from kate.prtsystems.ltd.uk ([217.65.165.35]) by smtp5.uk.umis.net with esmtpsa (TLSv1:AES256-SHA:256) (Exim 4.63 (FreeBSD)) (envelope-from ) id 1OkiOu-000F3v-Tt; Sun, 15 Aug 2010 18:57:44 +0000 Message-ID: <4C6838A8.6040907@prt.org> Date: Sun, 15 Aug 2010 19:57:44 +0100 From: Paul Thornton User-Agent: Thunderbird 2.0.0.24 (Windows/20100228) MIME-Version: 1.0 To: Hans Petter Selasky References: <201008151947.04963.hselasky@c2i.net> In-Reply-To: <201008151947.04963.hselasky@c2i.net> X-Enigmail-Version: 0.96.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Cc: freebsd-gnats-submit@freebsd.org, ed@freebsd.org, freebsd-usb@freebsd.org Subject: Re: usb/149675: uftdi doesn't react to break properly X-BeenThere: freebsd-usb@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: FreeBSD support for USB List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 15 Aug 2010 19:31:48 -0000 Hi, Hans Petter Selasky wrote: > I believe the following patch will fix your problem. Please apply and rebuild > kernel / ucom module. That has made a dramatic difference - things now look much better. Thank you very much for that. I'm still seeing a slight problem though. In my test setup, I have a PC (Windows but that shouldn't matter!) running an appplication which transmits a break followed by the 513 data bytes - and this repeats continually at near line-rate at 250kbaud. Pre-patching, FreeBSD was unable to keep data byte 1 in location 1 at all - it appeared randomly wherever it happened to be due to the lack of break clearing the input buffer, and as the data dump of the buffer refreshed, the values moved around as the breaks were turning up as datab. After I've applied the patch, it starts off well - data byte 1 is in location 1, but every 5-10 seconds there is a shift by a byte suggesting that an extra byte has been received. Is there any way that a break could still be creeping in to the data stream periodically? Under Linux, the code runs properly and there is no movement of the data bytes, so I'm reasonably confident that it isn't a bug of mine. Paul.