perf counters: hw driver API

Impact: restructure code, introduce hw_ops driver abstraction

Introduce this abstraction to handle counter details:

 struct hw_perf_counter_ops {
	void (*hw_perf_counter_enable)	(struct perf_counter *counter);
	void (*hw_perf_counter_disable)	(struct perf_counter *counter);
	void (*hw_perf_counter_read)	(struct perf_counter *counter);
 };

This will be useful to support assymetric hw details, and it will also
be useful to implement "software counters". (Counters that count kernel
managed sw events such as pagefaults, context-switches, wall-clock time
or task-local time.)

Signed-off-by: Ingo Molnar <mingo@elte.hu>

arch/x86/kernel/cpu/perf_counter.c

@@ -56,7 +56,7 @@ const int max_intel_perfmon_events = ARRAY_SIZE(intel_perfmon_event_map);
 /*
  * Setup the hardware configuration for a given hw_event_type
  */
-int hw_perf_counter_init(struct perf_counter *counter)
+static int __hw_perf_counter_init(struct perf_counter *counter)
 {
 	struct perf_counter_hw_event *hw_event = &counter->hw_event;
 	struct hw_perf_counter *hwc = &counter->hw;
@@ -135,7 +135,7 @@ u64 hw_perf_disable_all(void)
 EXPORT_SYMBOL_GPL(hw_perf_disable_all);
 
 static inline void
-__hw_perf_counter_disable(struct hw_perf_counter *hwc, unsigned int idx)
+__x86_perf_counter_disable(struct hw_perf_counter *hwc, unsigned int idx)
 {
 	wrmsr(hwc->config_base + idx, hwc->config, 0);
 }
@@ -149,13 +149,13 @@ static void __hw_perf_counter_set_period(struct hw_perf_counter *hwc, int idx)
 	wrmsr(hwc->counter_base + idx, hwc->next_count, 0);
 }
 
-static void __hw_perf_counter_enable(struct hw_perf_counter *hwc, int idx)
+static void __x86_perf_counter_enable(struct hw_perf_counter *hwc, int idx)
 {
 	wrmsr(hwc->config_base + idx,
 	      hwc->config | ARCH_PERFMON_EVENTSEL0_ENABLE, 0);
 }
 
-void hw_perf_counter_enable(struct perf_counter *counter)
+static void x86_perf_counter_enable(struct perf_counter *counter)
 {
 	struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters);
 	struct hw_perf_counter *hwc = &counter->hw;
@@ -170,12 +170,12 @@ void hw_perf_counter_enable(struct perf_counter *counter)
 
 	perf_counters_lapic_init(hwc->nmi);
 
-	__hw_perf_counter_disable(hwc, idx);
+	__x86_perf_counter_disable(hwc, idx);
 
 	cpuc->counters[idx] = counter;
 
 	__hw_perf_counter_set_period(hwc, idx);
-	__hw_perf_counter_enable(hwc, idx);
+	__x86_perf_counter_enable(hwc, idx);
 }
 
 #ifdef CONFIG_X86_64
@@ -282,20 +282,20 @@ void perf_counter_print_debug(void)
 	local_irq_enable();
 }
 
-void hw_perf_counter_disable(struct perf_counter *counter)
+static void x86_perf_counter_disable(struct perf_counter *counter)
 {
 	struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters);
 	struct hw_perf_counter *hwc = &counter->hw;
 	unsigned int idx = hwc->idx;
 
-	__hw_perf_counter_disable(hwc, idx);
+	__x86_perf_counter_disable(hwc, idx);
 
 	clear_bit(idx, cpuc->used);
 	cpuc->counters[idx] = NULL;
 	__hw_perf_save_counter(counter, hwc, idx);
 }
 
-void hw_perf_counter_read(struct perf_counter *counter)
+static void x86_perf_counter_read(struct perf_counter *counter)
 {
 	struct hw_perf_counter *hwc = &counter->hw;
 	unsigned long addr = hwc->counter_base + hwc->idx;
@@ -342,7 +342,7 @@ static void perf_save_and_restart(struct perf_counter *counter)
 	__hw_perf_counter_set_period(hwc, idx);
 
 	if (pmc_ctrl & ARCH_PERFMON_EVENTSEL0_ENABLE)
-		__hw_perf_counter_enable(hwc, idx);
+		__x86_perf_counter_enable(hwc, idx);
 }
 
 static void
@@ -572,3 +572,20 @@ void __init init_hw_perf_counters(void)
 
 	perf_counters_initialized = true;
 }
+
+static struct hw_perf_counter_ops x86_perf_counter_ops = {
+	.hw_perf_counter_enable		= x86_perf_counter_enable,
+	.hw_perf_counter_disable	= x86_perf_counter_disable,
+	.hw_perf_counter_read		= x86_perf_counter_read,
+};
+
+struct hw_perf_counter_ops *hw_perf_counter_init(struct perf_counter *counter)
+{
+	int err;
+
+	err = __hw_perf_counter_init(counter);
+	if (err)
+		return NULL;
+
+	return &x86_perf_counter_ops;
+}

include/linux/perf_counter.h

@@ -113,6 +113,17 @@ struct perf_data {
 	u8				data[PERF_DATA_BUFLEN];
 };
 
+struct perf_counter;
+
+/**
+ * struct hw_perf_counter_ops - performance counter hw ops
+ */
+struct hw_perf_counter_ops {
+	void (*hw_perf_counter_enable)	(struct perf_counter *counter);
+	void (*hw_perf_counter_disable)	(struct perf_counter *counter);
+	void (*hw_perf_counter_read)	(struct perf_counter *counter);
+};
+
 /**
  * struct perf_counter - performance counter kernel representation:
  */
@@ -120,6 +131,7 @@ struct perf_counter {
 	struct list_head		list_entry;
 	struct list_head		sibling_list;
 	struct perf_counter		*group_leader;
+	struct hw_perf_counter_ops	*hw_ops;
 
 	int				active;
 #if BITS_PER_LONG == 64
@@ -185,6 +197,9 @@ struct perf_cpu_context {
 extern int perf_max_counters;
 
 #ifdef CONFIG_PERF_COUNTERS
+extern struct hw_perf_counter_ops *
+hw_perf_counter_init(struct perf_counter *counter);
+
 extern void perf_counter_task_sched_in(struct task_struct *task, int cpu);
 extern void perf_counter_task_sched_out(struct task_struct *task, int cpu);
 extern void perf_counter_task_tick(struct task_struct *task, int cpu);

kernel/perf_counter.c

@@ -37,18 +37,15 @@ static DEFINE_MUTEX(perf_resource_mutex);
 /*
  * Architecture provided APIs - weak aliases:
  */
-
-int __weak hw_perf_counter_init(struct perf_counter *counter)
+extern __weak struct hw_perf_counter_ops *
+hw_perf_counter_init(struct perf_counter *counter)
 {
-	return -EINVAL;
+	return ERR_PTR(-EINVAL);
 }
 
-void __weak hw_perf_counter_enable(struct perf_counter *counter)	 { }
-void __weak hw_perf_counter_disable(struct perf_counter *counter)	 { }
-void __weak hw_perf_counter_read(struct perf_counter *counter)		 { }
-void __weak hw_perf_disable_all(void) { }
-void __weak hw_perf_enable_all(void) { }
-void __weak hw_perf_counter_setup(void) { }
+void __weak hw_perf_disable_all(void)	 { }
+void __weak hw_perf_enable_all(void)	 { }
+void __weak hw_perf_counter_setup(void)	 { }
 
 #if BITS_PER_LONG == 64
 
@@ -146,7 +143,7 @@ static void __perf_counter_remove_from_context(void *info)
 	spin_lock(&ctx->lock);
 
 	if (counter->active) {
-		hw_perf_counter_disable(counter);
+		counter->hw_ops->hw_perf_counter_disable(counter);
 		counter->active = 0;
 		ctx->nr_active--;
 		cpuctx->active_oncpu--;
@@ -257,7 +254,7 @@ static void __perf_install_in_context(void *info)
 	ctx->nr_counters++;
 
 	if (cpuctx->active_oncpu < perf_max_counters) {
-		hw_perf_counter_enable(counter);
+		counter->hw_ops->hw_perf_counter_enable(counter);
 		counter->active = 1;
 		counter->oncpu = cpu;
 		ctx->nr_active++;
@@ -333,7 +330,7 @@ counter_sched_out(struct perf_counter *counter,
 	if (!counter->active)
 		return;
 
-	hw_perf_counter_disable(counter);
+	counter->hw_ops->hw_perf_counter_disable(counter);
 	counter->active	=  0;
 	counter->oncpu	= -1;
 
@@ -392,7 +389,7 @@ counter_sched_in(struct perf_counter *counter,
 		 struct perf_counter_context *ctx,
 		 int cpu)
 {
-	hw_perf_counter_enable(counter);
+	counter->hw_ops->hw_perf_counter_enable(counter);
 	counter->active = 1;
 	counter->oncpu = cpu;	/* TODO: put 'cpu' into cpuctx->cpu */
 
@@ -509,7 +506,9 @@ void perf_counter_init_task(struct task_struct *task)
  */
 static void __hw_perf_counter_read(void *info)
 {
-	hw_perf_counter_read(info);
+	struct perf_counter *counter = info;
+
+	counter->hw_ops->hw_perf_counter_read(counter);
 }
 
 static u64 perf_counter_read(struct perf_counter *counter)
@@ -816,8 +815,10 @@ perf_counter_alloc(struct perf_counter_hw_event *hw_event,
 		   int cpu,
 		   struct perf_counter *group_leader)
 {
-	struct perf_counter *counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	struct hw_perf_counter_ops *hw_ops;
+	struct perf_counter *counter;
 
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
 	if (!counter)
 		return NULL;
 
@@ -839,6 +840,14 @@ perf_counter_alloc(struct perf_counter_hw_event *hw_event,
 	counter->hw_event		= *hw_event;
 	counter->wakeup_pending		= 0;
 	counter->group_leader		= group_leader;
+	counter->hw_ops			= NULL;
+
+	hw_ops = hw_perf_counter_init(counter);
+	if (!hw_ops) {
+		kfree(counter);
+		return NULL;
+	}
+	counter->hw_ops = hw_ops;
 
 	return counter;
 }
@@ -908,10 +917,6 @@ asmlinkage int sys_perf_counter_open(
 	if (!counter)
 		goto err_put_context;
 
-	ret = hw_perf_counter_init(counter);
-	if (ret)
-		goto err_free_put_context;
-
 	perf_install_in_context(ctx, counter, cpu);
 
 	ret = anon_inode_getfd("[perf_counter]", &perf_fops, counter, 0);
@@ -927,8 +932,6 @@ err_remove_free_put_context:
 	mutex_lock(&counter->mutex);
 	perf_counter_remove_from_context(counter);
 	mutex_unlock(&counter->mutex);
-
-err_free_put_context:
 	kfree(counter);
 
 err_put_context: