From mboxrd@z Thu Jan 1 00:00:00 1970 From: arnd@arndb.de (Arnd Bergmann) Date: Tue, 26 Mar 2013 14:31:36 +0000 Subject: [PATCH 0/8] SIRF multiplatform support In-Reply-To: References: <1363779679-16880-1-git-send-email-arnd@arndb.de> <201303251133.31426.arnd@arndb.de> Message-ID: <201303261431.37027.arnd@arndb.de> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Tuesday 26 March 2013, Barry Song wrote: > everything is fine except that: > 1. Shawn Guo's uncompress debug patch is not in the tree, so the > uncompress debug is break. but that is not an issue Right. > 2. i want to change: > #define SIRFSOC_NUM_IRQS 128 > to: > #define SIRFSOC_NUM_IRQS 64 > the old 128 is suitable for marco, in fact prima2/atlas6 is 64. in > order that all SoCs can use the same irqs.h headfile, i pick the > biggest value. since now we have moved to linear irq, i want to fix it > back. > > then nr_irqs in DT_MACHINE_START(ATLAS6_DT/PRIMA2_DT should also take 64. Yes, sounds good. > 3. will we have similar codes between prima2 and marco for initilizing > timer tick? > CLOCKSOURCE_OF_DECLARE(sirfsoc_marco_timer, "sirf,marco-tick", > sirfsoc_of_timer_init ); > CLOCKSOURCE_OF_DECLARE(sirfsoc_prima2_timer, "sirf,prima2-tick", > sirfsoc_prima2_timer_init); > > will we change "CLOCKSOURCE_OF_DECLARE(sirfsoc_marco_timer, > "sirf,marco-tick", sirfsoc_of_timer_init )" to things like > "CLOCKSOURCE_OF_DECLARE(sirfsoc_marco_timer, "sirf,marco-tick", > sirfsoc_marco_timer_init )" as prima2 does and drop > > anyway, this is a trivial issue. Yes, it seemed less invasive to do it without that change for marco. Since these changes are now just cosmetic, how about I pull the branch I have into the arm-soc tree and you do any further changes you'd like on top? It may also be a good point in time now to rename the directory to mach-sirf or mach-sirfsoc now as you suggested for an earlier release, since we have moved a lot of code out of it now. Any patches against the drivers that are getting moved will have to change anyway. Arnd