From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <40EE9C47.7040302@9fs.org> Date: Fri, 9 Jul 2004 14:23:19 +0100 From: Nigel Roles User-Agent: Mozilla Thunderbird 0.7.1 (X11/20040626) MIME-Version: 1.0 To: Fans of the OS Plan 9 from Bell Labs <9fans@cse.psu.edu> Subject: Re: [9fans] icky specs References: <239b02548abf81a01e884444aa9482bf@coraid.com> In-Reply-To: <239b02548abf81a01e884444aa9482bf@coraid.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Topicbox-Message-UUID: bb27310c-eacd-11e9-9e20-41e7f4b1d025 Brantley Coile wrote: >Anyone have a pointer to where on the web I can >get a copy of the *CHI specs? > > Brantley > > > http://www.usb.org/developers/docs/ Well well, I took a cursory look at EHCI, and it's an Intel spec., which admits to fixing a lot of USB 1.1 host controller issues (i.e. those specific to UHCI). Most of the changes relate to reducing the overhead I/O load. This is achieved by allowing descriptors to reference more and bigger buffers. (UHCI or OHCI) reads in-memory descriptors to decide what to do for each endpoint. These are chained and there can be as many of them as the driver designer chooses. It always worried me that there must be a limit as to how many of these can be walked before the chip runs out of time to make a decision. Evidently, at 480mbps, it became a real issue.