From mboxrd@z Thu Jan 1 00:00:00 1970 From: Simon Kagstrom Date: Fri, 10 Jul 2009 09:24:52 +0200 Subject: [U-Boot] [PATCH 1/4]: arm: Kirkwood: Set MAC address during registration for kirkwood egiga In-Reply-To: <4A56D52A.8010401@gmail.com> References: <20090708130134.4b100399@marrow.netinsight.se> <20090708130222.7b3381ae@marrow.netinsight.se> <4A56D52A.8010401@gmail.com> Message-ID: <20090710092452.492dfafe@marrow.netinsight.se> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de On Thu, 09 Jul 2009 22:44:10 -0700 Ben Warren wrote: > Simon Kagstrom wrote: > > This patch sets the MAC address during registration in addition to > > during device init. Since U-boot might not access the ethernet device, > > Linux might end up with the MAC address unset. > > > I think this violates U-boot policy of only touching hardware if it's > used, but Wolfgang can say for sure. In general, initialize() functions > should just set up data structures and register the device. There's a > long history of MAC-address issues with ARM chips and Linux that I've > happily stayed clear of so can't claim to be an expert here. OK, I've tried looking around at how other boards do it, and at least the most similar (mv6436x_eth_initialize() for other Marvell boards) do it the same way as well as some others (ax88180.c). Should Linux otherwise read the passed U-boot environment and set it up by itself? // Simon