drivers/perf: arm_pmu: manage interrupts per-cpu
authorMark Rutland <mark.rutland@arm.com>
Fri, 10 Mar 2017 10:46:14 +0000 (10:46 +0000)
committerWill Deacon <will.deacon@arm.com>
Fri, 31 Mar 2017 17:20:02 +0000 (18:20 +0100)
commit7ed98e0168bd23d8ea3294e95254cc5b4000c948
tree7e8de90dca98d514bde6ec83674188f8cefca897
parent2681f0184276d7fc934b6866a5a267f5b3369f7d
drivers/perf: arm_pmu: manage interrupts per-cpu

When requesting or freeing interrupts, we use platform_get_irq() to find
relevant irqs, backing this up with additional information in an
optional irq_affinity table.

This means that our irq request and free paths are tied to a
platform_device, and our request path must jump through a number of
hoops in order to determine the required affinity of each interrupt.

Given that the affinity must be static, we can compute the affinity once
up-front at probe time, simplifying the irq request and free paths. By
recording interrupts in a per-cpu data structure, we simplify a few
paths, and permit a subsequent rework of the request and free paths.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
[will: rename local nr_irqs variable to avoid conflict with global]
Signed-off-by: Will Deacon <will.deacon@arm.com>
drivers/perf/arm_pmu.c
include/linux/perf/arm_pmu.h