From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from outmx009.isp.belgacom.be (outmx009.isp.belgacom.be [195.238.3.4]) by ozlabs.org (Postfix) with ESMTP id BB9FB2BDB5 for ; Tue, 4 Jan 2005 03:47:26 +1100 (EST) Received: from outmx009.isp.belgacom.be (localhost [127.0.0.1]) by outmx009.isp.belgacom.be (8.12.11/8.12.11/Skynet-OUT-2.22) with ESMTP id j03GlCr1006197 for ; Mon, 3 Jan 2005 17:47:13 +0100 (envelope-from ) Message-ID: <41D97734.3070800@246tNt.com> Date: Mon, 03 Jan 2005 17:47:48 +0100 From: Sylvain Munaut MIME-Version: 1.0 To: Mark Chambers References: <200412301246.48870.david.jander@protonic.nl> <061001c4eeaf$0b5e1860$0301a8c0@chuck2> <41D7269E.1070906@246tNt.com> <003c01c4f19d$98a16510$0301a8c0@chuck2> In-Reply-To: <003c01c4f19d$98a16510$0301a8c0@chuck2> Content-Type: text/plain; charset=us-ascii; format=flowed Cc: linuxppc-embedded@ozlabs.org Subject: Re: MPC5200 PCI byte-swapping List-Id: Linux on Embedded PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , >Thanks. I had an older manual that didn't spell it out so >clearly. > >I've been trying to interpret the PCI sections for some other >82xx family parts, and it appears that they do NOT do this >byte lane swapping, so this make the 5200 non-standard in >this regard, which is unfortunate. If I'm understanding this >right, one would have to have different drivers for a PCI >device on a 5200 and an 8270, for instance. > > Mmmm ... Still I use the intel eepro driver without problems or modifications. As long as the driver uses the proper readl/writel that should do it or am I mistaken ? I have a FPGA mounted with pci interface, I'll try to see what happens on the bus >Also, I note that when doing simple block reads from pre- >fetchable PCI space, it appears the 5200 does not prefetch, >but does each read individually. This is using stock ELDK >u-boot and 2.4.24 so I haven't yet determined if it's a >configuration matter, (or ruled out target disconnects) >but I'm suspecting that you can't get burst mode from the >5200 without using DMA. > > Last time I checked, the 2.4 from denx didn't create a pci window for prefetchable memory so prefetch mem zone were mapped as non-prefetchable, so no burst for sure. Sylvain