From: Matthias Fuchs Date: Mon, 2 Nov 2009 15:22:16 +0100 Subject: [U-Boot] [RFC 3/5] CAN device driver for the SJA1000 In-Reply-To: <4AEED588.7060903@grandegger.com> References: <1257075217-26623-1-git-send-email-wg@grandegger.com> <200911021302.20662.matthias.fuchs@esd.eu> <4AEED588.7060903@grandegger.com> Message-ID: <200911021522.16256.matthias.fuchs@esd.eu> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de Hi Wolfgang, of course I can think of situations where some simple CAN mechanism might be helpful (e.g. simple hardware testing). But do we really need this inside a bootloader? Surely not for a production build. But please keep on hacking! On Monday 02 November 2009 13:50, Wolfgang Grandegger wrote: > Matthias Fuchs wrote: > > Hi Wolfgang, > > > > this patch conflicts with my simple SJA header posted some days ago > > > > http://lists.denx.de/pipermail/u-boot/2009-October/063097.html > > > > together with a fix for two of our boards - which has not much > > to do with CAN. WD asked me to use a C struct to access the SJA1000. > > > > http://lists.denx.de/pipermail/u-boot/2009-October/062902.html > > > > So where does this bring us? Either we want to use C structs for everything > > or decide it from patch to patch :-( > > Then it should be changed, of course. This patch is far from being > accepted and for the moment it's an implementation detail. I'm > especially interested to hear if such a generic CAN interface would > serve your purposes as well, as you require access to the SJA1000 somehow. I just need to bit bang around in the OCR register. So no need for a full blown and flash consuming CAN implementation. Of course I could life with your register access style. Especially because it makes the code more common with Socket-CAN files which prevents us from rewriting fully functional code ;-) Matthias