From mboxrd@z Thu Jan 1 00:00:00 1970 From: cavokz@gmail.com (Domenico Andreoli) Date: Tue, 23 Jul 2013 01:52:38 +0200 Subject: [PATCH 1/5] ARM: bcm4760: Add platform infrastructure In-Reply-To: <201307211400.21295.arnd@arndb.de> References: <20130721002320.730568671@gmail.com> <201307211009.31789.arnd@arndb.de> <201307211400.21295.arnd@arndb.de> Message-ID: <20130722235238.GC6665@glitch> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Sun, Jul 21, 2013 at 02:00:21PM +0200, Arnd Bergmann wrote: > On Sunday 21 July 2013, Domenico Andreoli wrote: > > >bonus points if you use soc_device_register(); > > > > > >What is the system_rev variable actually used for? > > > > I use it to show something sensible in /proc/cpuinfo, to report > > whether the soc is a 4760 or 4761 and the silicon revision. > > > > If the proper way to deal with this has so much boilerplate as > > you say, I think I'll drop it altogether. I'd like to print the info > > somewhere in the bootlog anyway. > > > > btw this could well be one of the accesses requiring the io > > mapping above. > > If it's just for informational purposes, soc_device_register() sounds like > the correct place to put that information, you can register a couple > of strings there that get put into sysfs at a known location for the > SoC, and keeps the cpuinfo for the actual CPU core. > > Arnd I've implemented the registration using soc_device_register() but I've postponed to a later stage, these registers are in block I need to add later also for the clocks. This showed to be also the only user of the early io mapping, which is now gone. I'll respin the patchset as soon as you suggest what to do with the clock-frequency of the system timer. BRs, Domenico