From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from server318.com (server318.com [216.35.196.92]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTPS id E0FBEDDFB7 for ; Tue, 19 May 2009 01:01:03 +1000 (EST) Message-id: <4A117823-3B-3CBB2B73@nukeman.tgi-sci.com> To: From: Dimiter Popoff Subject: Re: FEC & SDMA (bestcomm) interaction on the 5200 Date: Mon, 18 May 2009 17:59:47 +0200 List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , > From: Grant Likely > Date: Fri, 15 May 2009 08:52:19 -0600 > Subject: Re: FEC & SDMA (bestcomm) interaction on the 5200 > To: Dimiter Popoff > Cc: linuxppc-dev@ozlabs.org > > On Fri, May 15, 2009 at 2:36 AM, Dimiter Popoff wrote: > > Hi people, > > > > I am porting my (not linux) OS to the 5200. I went all the > > way understanding how the SDMA works and programming what > > I needed for it so things are now pretty stable in terms > > of disk I/O and system memory -> PCI (offscreen window > > buffers -> PCI display framebuffer). > > > > =A0And I wanted to make use of the Ethernet ("FEC", as they have > > it), hoping it would take me a few days (not my first one). > > > > =A0I am out of luck - it is practically undocumented. There is some > > talk of receive buffers which the FEC should see empty etc., > > how on earth is that supposed to happen when it has no bus > > master capability at all? It relies on the SDMA for its bus > > activity - but in what format does it expect to get these > > buffer related data? > > =A0Or (more likely, at least hopefully so) this talk in the > > MPC5200BUM (and MPC5200UM, for that) is just nonsense and > > the FEC simply puts in the FIFO incoming packets, writing as > > a last .l the receive frame status word? If so, I could > > easily handle that with the SDMA. > > =A0Same question about transmitting. I can only hope/assume > > that the FEC will just send what it is handed through the > > FIFO by the SDMA, the very first .l being the frame control > > word, and ending when the TFD from the data drd1a or whatever > > makes it through? > > It's been a long time since I've been in the bowels of the FEC driver, > but I believe you are correct. > > g. > > -- = > > Grant Likely, B.Sc., P.Eng. > Secret Lab Technologies Ltd. > Thanks for the input, Grant. It matches the memories of a guy at Frrecale who has had to deal with it ass well, so I guess I am on the right path - It won't be without trying this or that out but I am well enough positioned to start. Dimiter ------------------------------------------------------ Dimiter Popoff Transgalactic Instruments http://www.tgi-sci.com ------------------------------------------------------ http://www.flickr.com/photos/didi_tgi/sets/72157600228621276/