Andrew, First, great work! On Mon, Sep 15, 2014 at 04:51:03PM -0700, Andrew Bresticker wrote: > The current MIPS GIC driver and the platform code using it are rather > ugly and could use a good cleanup before adding device-tree support [0]. > This major issues addressed in this series are converting the GIC (and > platforms using it) to use IRQ domains and properly mapping interrupts > through the GIC instead of using it transparently. For part 2 I plan > on: updating the driver to use proper iomem accessors, cleaning up and > moving the GIC clocksource driver to drivers/clocksource/, adding DT > support, and possibly converting the GIC driver to use generic irqchip. > > Patches 1-16 are cleanups for the existing GIC driver and prepare platforms > using it for the switch to IRQ domains and using the GIC in a non-transparent > way. > > Patches 17-24 convert the GIC driver to use IRQ domains and updates the > platforms using it to properly map GIC interrupts instead of using the static > routing tables to make the GIC appear transparent. > > I've tested this series on Malta and, with additional patches, on the > DT-enabled Danube platform. Unfortunately I do not have SEAD-3 hardware, > so that has only been compile tested. Compile tested on all other affected > architectures (ath79, ralink, lantiq). > > [0] https://lkml.org/lkml/2014/9/5/542 > > Andrew Bresticker (24): ... > MIPS: Move GIC to drivers/irqchip/ > irqchip: mips-gic: Implement generic irq_ack/irq_eoi callbacks > irqchip: mips-gic: Implement irq_set_type callback > irqchip: mips-gic: Fix gic_set_affinity() return value > irqchip: mips-gic: Use IRQ domains > irqchip: mips-gic: Stop using per-platform mapping tables > irqchip: mips-gic: Probe for number of external interrupts > irqchip: mips-gic: Use separate edge/level irq_chips > irqchip: mips-gic: Support local interrupts > irqchip: mips-gic: Remove unnecessary globals ... > drivers/irqchip/Kconfig | 4 + > drivers/irqchip/Makefile | 1 + > drivers/irqchip/irq-mips-gic.c | 597 +++++++++++++++++++++ It would be too much of a pia to have this go through irqchip, so Acked-by: Jason Cooper <jason@xxxxxxxxxxxxxx> Please make sure to add the Tested-by's, etc before merging through mips. thx, Jason.
