From mboxrd@z Thu Jan 1 00:00:00 1970 From: will.deacon@arm.com (Will Deacon) Date: Mon, 8 Mar 2010 12:00:02 -0000 Subject: [PATCH] ARM: Add Versatile Express CA9x4 processor support In-Reply-To: <20100305204307.GB4885@n2100.arm.linux.org.uk> References: <002101cabc5c$81fa5440$85eefcc0$@deacon@arm.com> <20100305204307.GB4885@n2100.arm.linux.org.uk> Message-ID: <000a01cabeb6$e2681e00$a7385a00$@deacon@arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hello Russell, > On Fri, Mar 05, 2010 at 12:08:02PM -0000, Will Deacon wrote: > > The only problem with adding the PMU interrupts here is that > > they are not visible to kernel/pmu.c, which includes . > > Adding a conditional #include worked, but it's not > > very nice. Would it be possible to include the tile IRQs in the > > include/mach/irqs.h? > > Given that this platform can have multiple different tiles fitted, I > don't think it's correct to publicise the interrupt numbers globally. > > Currently, we have the situation where 99.9% of the tile dependencies > are entirely separated from the motherboard - the only exception being > the base of the motherboard IRQ numbers, and the base addresses of > the motherboard CS regions. > > It would be better if there was some way to allow board support to > register their individual details with the PMU code. The fact that > the PMU interrupts on CA9x4 are 92-95 does not mean that on another > Versatile Express tile, they will be the same - and in that case if > you select the interrupts through ifdefs, you're tied to only one > platform inspite of the rest of the kernel image supporting multiple > platforms. You're right. Adding a PMU IRQ registration mechanism might also help to solve the problem of registering other event sources that a board may have. I'll try and put something together this week as an RFC. Cheers, Will