From c0248c96631f38f02d58762fc018e316843acac8 Mon Sep 17 00:00:00 2001 From: Mark Rutland Date: Mon, 5 Feb 2018 16:41:56 +0000 Subject: arm_pmu: kill arm_pmu_platdata Now that we have no platforms passing platform data to the arm_pmu code, we can get rid of the platdata and associated hooks, paving the way for rework of our IRQ handling. Signed-off-by: Mark Rutland Cc: Will Deacon Signed-off-by: Will Deacon --- include/linux/perf/arm_pmu.h | 17 ----------------- 1 file changed, 17 deletions(-) (limited to 'include') diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h index af0f44effd44..712764b35c6a 100644 --- a/include/linux/perf/arm_pmu.h +++ b/include/linux/perf/arm_pmu.h @@ -17,23 +17,6 @@ #include #include -/* - * struct arm_pmu_platdata - ARM PMU platform data - * - * @handle_irq: an optional handler which will be called from the - * interrupt and passed the address of the low level handler, - * and can be used to implement any platform specific handling - * before or after calling it. - * - * @irq_flags: if non-zero, these flags will be passed to request_irq - * when requesting interrupts for this PMU device. - */ -struct arm_pmu_platdata { - irqreturn_t (*handle_irq)(int irq, void *dev, - irq_handler_t pmu_handler); - unsigned long irq_flags; -}; - #ifdef CONFIG_ARM_PMU /* -- cgit v1.2.3 From d3d5aac206b4e9e569a22fe1811c909dde17587c Mon Sep 17 00:00:00 2001 From: Mark Rutland Date: Mon, 5 Feb 2018 16:41:57 +0000 Subject: arm_pmu: fold platform helpers into platform code The armpmu_{request,free}_irqs() helpers are only used by arm_pmu_platform.c, so let's fold them in and make them static. Signed-off-by: Mark Rutland Cc: Will Deacon Signed-off-by: Will Deacon --- drivers/perf/arm_pmu.c | 21 --------------------- drivers/perf/arm_pmu_platform.c | 21 +++++++++++++++++++++ include/linux/perf/arm_pmu.h | 2 -- 3 files changed, 21 insertions(+), 23 deletions(-) (limited to 'include') diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c index 82b09d1cb42c..373dfd7d8a1d 100644 --- a/drivers/perf/arm_pmu.c +++ b/drivers/perf/arm_pmu.c @@ -534,14 +534,6 @@ void armpmu_free_irq(struct arm_pmu *armpmu, int cpu) free_irq(irq, per_cpu_ptr(&hw_events->percpu_pmu, cpu)); } -void armpmu_free_irqs(struct arm_pmu *armpmu) -{ - int cpu; - - for_each_cpu(cpu, &armpmu->supported_cpus) - armpmu_free_irq(armpmu, cpu); -} - int armpmu_request_irq(struct arm_pmu *armpmu, int cpu) { int err = 0; @@ -593,19 +585,6 @@ err_out: return err; } -int armpmu_request_irqs(struct arm_pmu *armpmu) -{ - int cpu, err; - - for_each_cpu(cpu, &armpmu->supported_cpus) { - err = armpmu_request_irq(armpmu, cpu); - if (err) - break; - } - - return err; -} - static int armpmu_get_cpu_irq(struct arm_pmu *pmu, int cpu) { struct pmu_hw_events __percpu *hw_events = pmu->hw_events; diff --git a/drivers/perf/arm_pmu_platform.c b/drivers/perf/arm_pmu_platform.c index 46501cc79fd7..244558cfdbce 100644 --- a/drivers/perf/arm_pmu_platform.c +++ b/drivers/perf/arm_pmu_platform.c @@ -164,6 +164,27 @@ static int pmu_parse_irqs(struct arm_pmu *pmu) return 0; } +static int armpmu_request_irqs(struct arm_pmu *armpmu) +{ + int cpu, err; + + for_each_cpu(cpu, &armpmu->supported_cpus) { + err = armpmu_request_irq(armpmu, cpu); + if (err) + break; + } + + return err; +} + +static void armpmu_free_irqs(struct arm_pmu *armpmu) +{ + int cpu; + + for_each_cpu(cpu, &armpmu->supported_cpus) + armpmu_free_irq(armpmu, cpu); +} + int arm_pmu_device_probe(struct platform_device *pdev, const struct of_device_id *of_table, const struct pmu_probe_info *probe_table) diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h index 712764b35c6a..899bc7ef0881 100644 --- a/include/linux/perf/arm_pmu.h +++ b/include/linux/perf/arm_pmu.h @@ -159,8 +159,6 @@ static inline int arm_pmu_acpi_probe(armpmu_init_fn init_fn) { return 0; } struct arm_pmu *armpmu_alloc(void); void armpmu_free(struct arm_pmu *pmu); int armpmu_register(struct arm_pmu *pmu); -int armpmu_request_irqs(struct arm_pmu *armpmu); -void armpmu_free_irqs(struct arm_pmu *armpmu); int armpmu_request_irq(struct arm_pmu *armpmu, int cpu); void armpmu_free_irq(struct arm_pmu *armpmu, int cpu); -- cgit v1.2.3 From 0dc1a1851af1d593eee248b94c1277c7c7ccbbce Mon Sep 17 00:00:00 2001 From: Mark Rutland Date: Mon, 5 Feb 2018 16:41:58 +0000 Subject: arm_pmu: add armpmu_alloc_atomic() In ACPI systems, we don't know the makeup of CPUs until we hotplug them on, and thus have to allocate the PMU datastructures at hotplug time. Thus, we must use GFP_ATOMIC allocations. Let's add an armpmu_alloc_atomic() that we can use in this case. Signed-off-by: Mark Rutland Cc: Will Deacon Signed-off-by: Will Deacon --- drivers/perf/arm_pmu.c | 17 ++++++++++++++--- drivers/perf/arm_pmu_acpi.c | 2 +- include/linux/perf/arm_pmu.h | 1 + 3 files changed, 16 insertions(+), 4 deletions(-) (limited to 'include') diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c index 373dfd7d8a1d..4f73c5e8d623 100644 --- a/drivers/perf/arm_pmu.c +++ b/drivers/perf/arm_pmu.c @@ -760,18 +760,18 @@ static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu) &cpu_pmu->node); } -struct arm_pmu *armpmu_alloc(void) +static struct arm_pmu *__armpmu_alloc(gfp_t flags) { struct arm_pmu *pmu; int cpu; - pmu = kzalloc(sizeof(*pmu), GFP_KERNEL); + pmu = kzalloc(sizeof(*pmu), flags); if (!pmu) { pr_info("failed to allocate PMU device!\n"); goto out; } - pmu->hw_events = alloc_percpu(struct pmu_hw_events); + pmu->hw_events = alloc_percpu_gfp(struct pmu_hw_events, flags); if (!pmu->hw_events) { pr_info("failed to allocate per-cpu PMU data.\n"); goto out_free_pmu; @@ -817,6 +817,17 @@ out: return NULL; } +struct arm_pmu *armpmu_alloc(void) +{ + return __armpmu_alloc(GFP_KERNEL); +} + +struct arm_pmu *armpmu_alloc_atomic(void) +{ + return __armpmu_alloc(GFP_ATOMIC); +} + + void armpmu_free(struct arm_pmu *pmu) { free_percpu(pmu->hw_events); diff --git a/drivers/perf/arm_pmu_acpi.c b/drivers/perf/arm_pmu_acpi.c index 705f1a390e31..30c5f2bbce59 100644 --- a/drivers/perf/arm_pmu_acpi.c +++ b/drivers/perf/arm_pmu_acpi.c @@ -127,7 +127,7 @@ static struct arm_pmu *arm_pmu_acpi_find_alloc_pmu(void) return pmu; } - pmu = armpmu_alloc(); + pmu = armpmu_alloc_atomic(); if (!pmu) { pr_warn("Unable to allocate PMU for CPU%d\n", smp_processor_id()); diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h index 899bc7ef0881..1f8bb83ef42f 100644 --- a/include/linux/perf/arm_pmu.h +++ b/include/linux/perf/arm_pmu.h @@ -157,6 +157,7 @@ static inline int arm_pmu_acpi_probe(armpmu_init_fn init_fn) { return 0; } /* Internal functions only for core arm_pmu code */ struct arm_pmu *armpmu_alloc(void); +struct arm_pmu *armpmu_alloc_atomic(void); void armpmu_free(struct arm_pmu *pmu); int armpmu_register(struct arm_pmu *pmu); int armpmu_request_irq(struct arm_pmu *armpmu, int cpu); -- cgit v1.2.3 From 84b4be57ae17f8c0b3c1d8629e10f23910838fd7 Mon Sep 17 00:00:00 2001 From: Mark Rutland Date: Tue, 12 Dec 2017 16:56:06 +0000 Subject: arm_pmu: note IRQs and PMUs per-cpu To support ACPI systems, we need to request IRQs before we know the associated PMU, and thus we need some percpu variable that the IRQ handler can find the PMU from. As we're going to request IRQs without the PMU, we can't rely on the arm_pmu::active_irqs mask, and similarly need to track requested IRQs with a percpu variable. Signed-off-by: Mark Rutland [will: made armpmu_count_irq_users static] Signed-off-by: Will Deacon --- drivers/perf/arm_pmu.c | 69 +++++++++++++++++++++++++++++++++----------- include/linux/perf/arm_pmu.h | 1 - 2 files changed, 52 insertions(+), 18 deletions(-) (limited to 'include') diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c index 72118e6f9122..2b2af35db1b6 100644 --- a/drivers/perf/arm_pmu.c +++ b/drivers/perf/arm_pmu.c @@ -25,6 +25,9 @@ #include +static DEFINE_PER_CPU(struct arm_pmu *, cpu_armpmu); +static DEFINE_PER_CPU(int, cpu_irq); + static int armpmu_map_cache_event(const unsigned (*cache_map) [PERF_COUNT_HW_CACHE_MAX] @@ -332,6 +335,8 @@ static irqreturn_t armpmu_dispatch_irq(int irq, void *dev) * dereference. */ armpmu = *(void **)dev; + if (WARN_ON_ONCE(!armpmu)) + return IRQ_NONE; start_clock = sched_clock(); ret = armpmu->handle_irq(irq, armpmu); @@ -517,29 +522,45 @@ int perf_num_counters(void) } EXPORT_SYMBOL_GPL(perf_num_counters); -void armpmu_free_irq(struct arm_pmu *armpmu, int cpu) +static int armpmu_count_irq_users(const int irq) { - struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; - int irq = per_cpu(hw_events->irq, cpu); + int cpu, count = 0; - if (!cpumask_test_and_clear_cpu(cpu, &armpmu->active_irqs)) - return; + for_each_possible_cpu(cpu) { + if (per_cpu(cpu_irq, cpu) == irq) + count++; + } + + return count; +} - if (irq_is_percpu_devid(irq)) { - free_percpu_irq(irq, &hw_events->percpu_pmu); - cpumask_clear(&armpmu->active_irqs); +void armpmu_free_cpu_irq(int irq, int cpu) +{ + if (per_cpu(cpu_irq, cpu) == 0) return; - } + if (WARN_ON(irq != per_cpu(cpu_irq, cpu))) + return; + + if (!irq_is_percpu_devid(irq)) + free_irq(irq, per_cpu_ptr(&cpu_armpmu, cpu)); + else if (armpmu_count_irq_users(irq) == 1) + free_percpu_irq(irq, &cpu_armpmu); - free_irq(irq, per_cpu_ptr(&hw_events->percpu_pmu, cpu)); + per_cpu(cpu_irq, cpu) = 0; } -int armpmu_request_irq(struct arm_pmu *armpmu, int cpu) +void armpmu_free_irq(struct arm_pmu *armpmu, int cpu) { - int err = 0; struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; - const irq_handler_t handler = armpmu_dispatch_irq; int irq = per_cpu(hw_events->irq, cpu); + + armpmu_free_cpu_irq(irq, cpu); +} + +int armpmu_request_cpu_irq(int irq, int cpu) +{ + int err = 0; + const irq_handler_t handler = armpmu_dispatch_irq; if (!irq) return 0; @@ -560,16 +581,16 @@ int armpmu_request_irq(struct arm_pmu *armpmu, int cpu) irq_set_status_flags(irq, IRQ_NOAUTOEN); err = request_irq(irq, handler, irq_flags, "arm-pmu", - per_cpu_ptr(&hw_events->percpu_pmu, cpu)); - } else if (cpumask_empty(&armpmu->active_irqs)) { + per_cpu_ptr(&cpu_armpmu, cpu)); + } else if (armpmu_count_irq_users(irq) == 0) { err = request_percpu_irq(irq, handler, "arm-pmu", - &hw_events->percpu_pmu); + &cpu_armpmu); } if (err) goto err_out; - cpumask_set_cpu(cpu, &armpmu->active_irqs); + per_cpu(cpu_irq, cpu) = irq; return 0; err_out: @@ -577,6 +598,16 @@ err_out: return err; } +int armpmu_request_irq(struct arm_pmu *armpmu, int cpu) +{ + struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; + int irq = per_cpu(hw_events->irq, cpu); + if (!irq) + return 0; + + return armpmu_request_cpu_irq(irq, cpu); +} + static int armpmu_get_cpu_irq(struct arm_pmu *pmu, int cpu) { struct pmu_hw_events __percpu *hw_events = pmu->hw_events; @@ -599,6 +630,8 @@ static int arm_perf_starting_cpu(unsigned int cpu, struct hlist_node *node) if (pmu->reset) pmu->reset(pmu); + per_cpu(cpu_armpmu, cpu) = pmu; + irq = armpmu_get_cpu_irq(pmu, cpu); if (irq) { if (irq_is_percpu_devid(irq)) @@ -626,6 +659,8 @@ static int arm_perf_teardown_cpu(unsigned int cpu, struct hlist_node *node) disable_irq(irq); } + per_cpu(cpu_armpmu, cpu) = NULL; + return 0; } diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h index 1f8bb83ef42f..feec9e7e85db 100644 --- a/include/linux/perf/arm_pmu.h +++ b/include/linux/perf/arm_pmu.h @@ -75,7 +75,6 @@ enum armpmu_attr_groups { struct arm_pmu { struct pmu pmu; - cpumask_t active_irqs; cpumask_t supported_cpus; char *name; irqreturn_t (*handle_irq)(int irq_num, void *dev); -- cgit v1.2.3 From 167e61438da0664cab87c825a6c0cb83510d578e Mon Sep 17 00:00:00 2001 From: Mark Rutland Date: Mon, 9 Oct 2017 17:09:05 +0100 Subject: arm_pmu: acpi: request IRQs up-front We can't request IRQs in atomic context, so for ACPI systems we'll have to request them up-front, and later associate them with CPUs. This patch reorganises the arm_pmu code to do so. As we no longer have the arm_pmu structure at probe time, a number of prototypes need to be adjusted, requiring changes to the common arm_pmu code and arm_pmu platform code. Signed-off-by: Mark Rutland Cc: Will Deacon Signed-off-by: Will Deacon --- drivers/perf/arm_pmu.c | 22 ++-------------------- drivers/perf/arm_pmu_acpi.c | 19 ++++++------------- drivers/perf/arm_pmu_platform.c | 15 ++++++++++++--- include/linux/perf/arm_pmu.h | 5 +++-- 4 files changed, 23 insertions(+), 38 deletions(-) (limited to 'include') diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c index 2b2af35db1b6..0c2ed11c0603 100644 --- a/drivers/perf/arm_pmu.c +++ b/drivers/perf/arm_pmu.c @@ -534,7 +534,7 @@ static int armpmu_count_irq_users(const int irq) return count; } -void armpmu_free_cpu_irq(int irq, int cpu) +void armpmu_free_irq(int irq, int cpu) { if (per_cpu(cpu_irq, cpu) == 0) return; @@ -549,15 +549,7 @@ void armpmu_free_cpu_irq(int irq, int cpu) per_cpu(cpu_irq, cpu) = 0; } -void armpmu_free_irq(struct arm_pmu *armpmu, int cpu) -{ - struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; - int irq = per_cpu(hw_events->irq, cpu); - - armpmu_free_cpu_irq(irq, cpu); -} - -int armpmu_request_cpu_irq(int irq, int cpu) +int armpmu_request_irq(int irq, int cpu) { int err = 0; const irq_handler_t handler = armpmu_dispatch_irq; @@ -598,16 +590,6 @@ err_out: return err; } -int armpmu_request_irq(struct arm_pmu *armpmu, int cpu) -{ - struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; - int irq = per_cpu(hw_events->irq, cpu); - if (!irq) - return 0; - - return armpmu_request_cpu_irq(irq, cpu); -} - static int armpmu_get_cpu_irq(struct arm_pmu *pmu, int cpu) { struct pmu_hw_events __percpu *hw_events = pmu->hw_events; diff --git a/drivers/perf/arm_pmu_acpi.c b/drivers/perf/arm_pmu_acpi.c index 09a1a36cff57..0f197516d708 100644 --- a/drivers/perf/arm_pmu_acpi.c +++ b/drivers/perf/arm_pmu_acpi.c @@ -89,7 +89,13 @@ static int arm_pmu_acpi_parse_irqs(void) pr_warn("No ACPI PMU IRQ for CPU%d\n", cpu); } + /* + * Log and request the IRQ so the core arm_pmu code can manage + * it. We'll have to sanity-check IRQs later when we associate + * them with their PMUs. + */ per_cpu(pmu_irqs, cpu) = irq; + armpmu_request_irq(irq, cpu); } return 0; @@ -204,14 +210,6 @@ static int arm_pmu_acpi_cpu_starting(unsigned int cpu) cpumask_set_cpu(cpu, &pmu->supported_cpus); - /* - * Log and request the IRQ so the core arm_pmu code can manage it. In - * some situations (e.g. mismatched PPIs), we may fail to request the - * IRQ. However, it may be too late for us to do anything about it. - * The common ARM PMU code will log a warning in this case. - */ - armpmu_request_irq(pmu, cpu); - /* * Ideally, we'd probe the PMU here when we find the first matching * CPU. We can't do that for several reasons; see the comment in @@ -281,11 +279,6 @@ static int arm_pmu_acpi_init(void) if (acpi_disabled) return 0; - /* - * We can't request IRQs yet, since we don't know the cookie value - * until we know which CPUs share the same logical PMU. We'll handle - * that in arm_pmu_acpi_cpu_starting(). - */ ret = arm_pmu_acpi_parse_irqs(); if (ret) return ret; diff --git a/drivers/perf/arm_pmu_platform.c b/drivers/perf/arm_pmu_platform.c index 1dc3c1f574e0..7729eda5909d 100644 --- a/drivers/perf/arm_pmu_platform.c +++ b/drivers/perf/arm_pmu_platform.c @@ -159,10 +159,15 @@ static int pmu_parse_irqs(struct arm_pmu *pmu) static int armpmu_request_irqs(struct arm_pmu *armpmu) { + struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; int cpu, err; for_each_cpu(cpu, &armpmu->supported_cpus) { - err = armpmu_request_irq(armpmu, cpu); + int irq = per_cpu(hw_events->irq, cpu); + if (!irq) + continue; + + err = armpmu_request_irq(irq, cpu); if (err) break; } @@ -173,9 +178,13 @@ static int armpmu_request_irqs(struct arm_pmu *armpmu) static void armpmu_free_irqs(struct arm_pmu *armpmu) { int cpu; + struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; - for_each_cpu(cpu, &armpmu->supported_cpus) - armpmu_free_irq(armpmu, cpu); + for_each_cpu(cpu, &armpmu->supported_cpus) { + int irq = per_cpu(hw_events->irq, cpu); + + armpmu_free_irq(irq, cpu); + } } int arm_pmu_device_probe(struct platform_device *pdev, diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h index feec9e7e85db..40036a57d072 100644 --- a/include/linux/perf/arm_pmu.h +++ b/include/linux/perf/arm_pmu.h @@ -14,6 +14,7 @@ #include #include +#include #include #include @@ -159,8 +160,8 @@ struct arm_pmu *armpmu_alloc(void); struct arm_pmu *armpmu_alloc_atomic(void); void armpmu_free(struct arm_pmu *pmu); int armpmu_register(struct arm_pmu *pmu); -int armpmu_request_irq(struct arm_pmu *armpmu, int cpu); -void armpmu_free_irq(struct arm_pmu *armpmu, int cpu); +int armpmu_request_irq(int irq, int cpu); +void armpmu_free_irq(int irq, int cpu); #define ARMV8_PMU_PDEV_NAME "armv8-pmu" -- cgit v1.2.3