From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.mdivac.com (adsl-68-253-104-89.dsl.bcvloh.ameritech.net [68.253.104.89]) by ozlabs.org (Postfix) with ESMTP id 343CD67BDC for ; Fri, 4 Aug 2006 04:53:58 +1000 (EST) Message-ID: <44D24643.1090208@mdivac.com> Date: Thu, 03 Aug 2006 14:53:55 -0400 From: Frank D Lombardo MIME-Version: 1.0 To: dhlii@comcast.net Subject: Re: Xilinx TEMAC References: <44CE4C1D.6050103@datadesigncorp.net> <44CE5EA5.3030306@gmail.com> <44CF89B5.7080906@datadesigncorp.net> <44D07B38.5060700@gmail.com> <44D1991A.2070208@dlasys.net> In-Reply-To: <44D1991A.2070208@dlasys.net> Content-Type: text/plain; charset=us-ascii; format=flowed Cc: linuxppc-embedded List-Id: Linux on Embedded PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , David H. Lynch Jr. wrote: > Ameet Patil wrote: > >> > Did you happen to get an Ethernet mac going on the virtex2 or >> > virtex4 with the 2.6 kernel? >> >> Nope! :-( >> Rumour has it that Montavista is soon going to release the source for >> 2.6 kernel to run on Xilinx platforms. >> >> >> > I have 2 different TEMAC's working under 2.6 with a V4. > > The MontaVista driver that has been posted on the list that I think > is from MontaVista for the PLB TEMAC with very minor modifications to > get it to work at speeds other than GigaBit. > > A driver I wrote for the LocalLink TEMAC that works limpingly. I > have been unable to figure out how to get a receive interrupt from the > LocalLink TEMAC so receives are polled and performance is poor. > I have pretty much abandoned work on it - my client wants the > smallest possible FPGA footprint BUTanother OS I am porting requires > interrupt driven IO and we are looking for a single minimal hardware > implimentation for all platforms. > > Anyway, if anyone wants the LL TEMAC in its current state I can post it. > Currently it sends fine. It drops about 50% of all received packets > - but that is probably debugging overhead and poll rate tuning. > > > > > > I would be interested looking at the LL TEMAC patch. Hopefully I can pick up where you left off. We would require an interrupt driven TEMAC