perf: Remove the nmi parameter from the swevent and overflow interface

The nmi parameter indicated if we could do wakeups from the current
context, if not, we would set some state and self-IPI and let the
resulting interrupt do the wakeup.

For the various event classes:

  - hardware: nmi=0; PMI is in fact an NMI or we run irq_work_run from
    the PMI-tail (ARM etc.)
  - tracepoint: nmi=0; since tracepoint could be from NMI context.
  - software: nmi=[0,1]; some, like the schedule thing cannot
    perform wakeups, and hence need 0.

As one can see, there is very little nmi=1 usage, and the down-side of
not using it is that on some platforms some software events can have a
jiffy delay in wakeup (when arch_irq_work_raise isn't implemented).

The up-side however is that we can remove the nmi parameter and save a
bunch of conditionals in fast paths.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Michael Cree <mcree@orcon.net.nz>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Deng-Cheng Zhu <dengcheng.zhu@gmail.com>
Cc: Anton Blanchard <anton@samba.org>
Cc: Eric B Munson <emunson@mgebm.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Don Zickus <dzickus@redhat.com>
Link: http://lkml.kernel.org/n/tip-agjev8eu666tvknpb3iaj0fg@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>

5 files changed, 33 insertions, 45 deletions

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 270e32f9fc0..dbd1ca75bd3 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -3972,7 +3972,7 @@ void perf_prepare_sample(struct perf_event_header *header,
 	}
 }
 
-static void perf_event_output(struct perf_event *event, int nmi,
+static void perf_event_output(struct perf_event *event,
 				struct perf_sample_data *data,
 				struct pt_regs *regs)
 {
@@ -3984,7 +3984,7 @@ static void perf_event_output(struct perf_event *event, int nmi,
 
 	perf_prepare_sample(&header, data, event, regs);
 
-	if (perf_output_begin(&handle, event, header.size, nmi, 1))
+	if (perf_output_begin(&handle, event, header.size, 1))
 		goto exit;
 
 	perf_output_sample(&handle, &header, data, event);
@@ -4024,7 +4024,7 @@ perf_event_read_event(struct perf_event *event,
 	int ret;
 
 	perf_event_header__init_id(&read_event.header, &sample, event);
-	ret = perf_output_begin(&handle, event, read_event.header.size, 0, 0);
+	ret = perf_output_begin(&handle, event, read_event.header.size, 0);
 	if (ret)
 		return;
 
@@ -4067,7 +4067,7 @@ static void perf_event_task_output(struct perf_event *event,
 	perf_event_header__init_id(&task_event->event_id.header, &sample, event);
 
 	ret = perf_output_begin(&handle, event,
-				task_event->event_id.header.size, 0, 0);
+				task_event->event_id.header.size, 0);
 	if (ret)
 		goto out;
 
@@ -4204,7 +4204,7 @@ static void perf_event_comm_output(struct perf_event *event,
 
 	perf_event_header__init_id(&comm_event->event_id.header, &sample, event);
 	ret = perf_output_begin(&handle, event,
-				comm_event->event_id.header.size, 0, 0);
+				comm_event->event_id.header.size, 0);
 
 	if (ret)
 		goto out;
@@ -4351,7 +4351,7 @@ static void perf_event_mmap_output(struct perf_event *event,
 
 	perf_event_header__init_id(&mmap_event->event_id.header, &sample, event);
 	ret = perf_output_begin(&handle, event,
-				mmap_event->event_id.header.size, 0, 0);
+				mmap_event->event_id.header.size, 0);
 	if (ret)
 		goto out;
 
@@ -4546,7 +4546,7 @@ static void perf_log_throttle(struct perf_event *event, int enable)
 	perf_event_header__init_id(&throttle_event.header, &sample, event);
 
 	ret = perf_output_begin(&handle, event,
-				throttle_event.header.size, 1, 0);
+				throttle_event.header.size, 0);
 	if (ret)
 		return;
 
@@ -4559,7 +4559,7 @@ static void perf_log_throttle(struct perf_event *event, int enable)
  * Generic event overflow handling, sampling.
  */
 
-static int __perf_event_overflow(struct perf_event *event, int nmi,
+static int __perf_event_overflow(struct perf_event *event,
 				   int throttle, struct perf_sample_data *data,
 				   struct pt_regs *regs)
 {
@@ -4602,34 +4602,28 @@ static int __perf_event_overflow(struct perf_event *event, int nmi,
 	if (events && atomic_dec_and_test(&event->event_limit)) {
 		ret = 1;
 		event->pending_kill = POLL_HUP;
-		if (nmi) {
-			event->pending_disable = 1;
-			irq_work_queue(&event->pending);
-		} else
-			perf_event_disable(event);
+		event->pending_disable = 1;
+		irq_work_queue(&event->pending);
 	}
 
 	if (event->overflow_handler)
-		event->overflow_handler(event, nmi, data, regs);
+		event->overflow_handler(event, data, regs);
 	else
-		perf_event_output(event, nmi, data, regs);
+		perf_event_output(event, data, regs);
 
 	if (event->fasync && event->pending_kill) {
-		if (nmi) {
-			event->pending_wakeup = 1;
-			irq_work_queue(&event->pending);
-		} else
-			perf_event_wakeup(event);
+		event->pending_wakeup = 1;
+		irq_work_queue(&event->pending);
 	}
 
 	return ret;
 }
 
-int perf_event_overflow(struct perf_event *event, int nmi,
+int perf_event_overflow(struct perf_event *event,
 			  struct perf_sample_data *data,
 			  struct pt_regs *regs)
 {
-	return __perf_event_overflow(event, nmi, 1, data, regs);
+	return __perf_event_overflow(event, 1, data, regs);
 }
 
 /*
@@ -4678,7 +4672,7 @@ again:
 }
 
 static void perf_swevent_overflow(struct perf_event *event, u64 overflow,
-				    int nmi, struct perf_sample_data *data,
+				    struct perf_sample_data *data,
 				    struct pt_regs *regs)
 {
 	struct hw_perf_event *hwc = &event->hw;
@@ -4692,7 +4686,7 @@ static void perf_swevent_overflow(struct perf_event *event, u64 overflow,
 		return;
 
 	for (; overflow; overflow--) {
-		if (__perf_event_overflow(event, nmi, throttle,
+		if (__perf_event_overflow(event, throttle,
 					    data, regs)) {
 			/*
 			 * We inhibit the overflow from happening when
@@ -4705,7 +4699,7 @@ static void perf_swevent_overflow(struct perf_event *event, u64 overflow,
 }
 
 static void perf_swevent_event(struct perf_event *event, u64 nr,
-			       int nmi, struct perf_sample_data *data,
+			       struct perf_sample_data *data,
 			       struct pt_regs *regs)
 {
 	struct hw_perf_event *hwc = &event->hw;
@@ -4719,12 +4713,12 @@ static void perf_swevent_event(struct perf_event *event, u64 nr,
 		return;
 
 	if (nr == 1 && hwc->sample_period == 1 && !event->attr.freq)
-		return perf_swevent_overflow(event, 1, nmi, data, regs);
+		return perf_swevent_overflow(event, 1, data, regs);
 
 	if (local64_add_negative(nr, &hwc->period_left))
 		return;
 
-	perf_swevent_overflow(event, 0, nmi, data, regs);
+	perf_swevent_overflow(event, 0, data, regs);
 }
 
 static int perf_exclude_event(struct perf_event *event,
@@ -4812,7 +4806,7 @@ find_swevent_head(struct swevent_htable *swhash, struct perf_event *event)
 }
 
 static void do_perf_sw_event(enum perf_type_id type, u32 event_id,
-				    u64 nr, int nmi,
+				    u64 nr,
 				    struct perf_sample_data *data,
 				    struct pt_regs *regs)
 {
@@ -4828,7 +4822,7 @@ static void do_perf_sw_event(enum perf_type_id type, u32 event_id,
 
 	hlist_for_each_entry_rcu(event, node, head, hlist_entry) {
 		if (perf_swevent_match(event, type, event_id, data, regs))
-			perf_swevent_event(event, nr, nmi, data, regs);
+			perf_swevent_event(event, nr, data, regs);
 	}
 end:
 	rcu_read_unlock();
@@ -4849,8 +4843,7 @@ inline void perf_swevent_put_recursion_context(int rctx)
 	put_recursion_context(swhash->recursion, rctx);
 }
 
-void __perf_sw_event(u32 event_id, u64 nr, int nmi,
-			    struct pt_regs *regs, u64 addr)
+void __perf_sw_event(u32 event_id, u64 nr, struct pt_regs *regs, u64 addr)
 {
 	struct perf_sample_data data;
 	int rctx;
@@ -4862,7 +4855,7 @@ void __perf_sw_event(u32 event_id, u64 nr, int nmi,
 
 	perf_sample_data_init(&data, addr);
 
-	do_perf_sw_event(PERF_TYPE_SOFTWARE, event_id, nr, nmi, &data, regs);
+	do_perf_sw_event(PERF_TYPE_SOFTWARE, event_id, nr, &data, regs);
 
 	perf_swevent_put_recursion_context(rctx);
 	preempt_enable_notrace();
@@ -5110,7 +5103,7 @@ void perf_tp_event(u64 addr, u64 count, void *record, int entry_size,
 
 	hlist_for_each_entry_rcu(event, node, head, hlist_entry) {
 		if (perf_tp_event_match(event, &data, regs))
-			perf_swevent_event(event, count, 1, &data, regs);
+			perf_swevent_event(event, count, &data, regs);
 	}
 
 	perf_swevent_put_recursion_context(rctx);
@@ -5203,7 +5196,7 @@ void perf_bp_event(struct perf_event *bp, void *data)
 	perf_sample_data_init(&sample, bp->attr.bp_addr);
 
 	if (!bp->hw.state && !perf_exclude_event(bp, regs))
-		perf_swevent_event(bp, 1, 1, &sample, regs);
+		perf_swevent_event(bp, 1, &sample, regs);
 }
 #endif
 
@@ -5232,7 +5225,7 @@ static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer)
 
 	if (regs && !perf_exclude_event(event, regs)) {
 		if (!(event->attr.exclude_idle && current->pid == 0))
-			if (perf_event_overflow(event, 0, &data, regs))
+			if (perf_event_overflow(event, &data, regs))
 				ret = HRTIMER_NORESTART;
 	}
 
diff --git a/kernel/events/internal.h b/kernel/events/internal.h
index 114f27f3a62..09097dd8116 100644
--- a/kernel/events/internal.h
+++ b/kernel/events/internal.h
@@ -27,7 +27,6 @@ struct ring_buffer {
 	void				*data_pages[0];
 };
 
-
 extern void rb_free(struct ring_buffer *rb);
 extern struct ring_buffer *
 rb_alloc(int nr_pages, long watermark, int cpu, int flags);
diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
index fc2701c9920..8b3b73630fa 100644
--- a/kernel/events/ring_buffer.c
+++ b/kernel/events/ring_buffer.c
@@ -38,11 +38,8 @@ static void perf_output_wakeup(struct perf_output_handle *handle)
 {
 	atomic_set(&handle->rb->poll, POLL_IN);
 
-	if (handle->nmi) {
-		handle->event->pending_wakeup = 1;
-		irq_work_queue(&handle->event->pending);
-	} else
-		perf_event_wakeup(handle->event);
+	handle->event->pending_wakeup = 1;
+	irq_work_queue(&handle->event->pending);
 }
 
 /*
@@ -102,7 +99,7 @@ out:
 
 int perf_output_begin(struct perf_output_handle *handle,
 		      struct perf_event *event, unsigned int size,
-		      int nmi, int sample)
+		      int sample)
 {
 	struct ring_buffer *rb;
 	unsigned long tail, offset, head;
@@ -127,7 +124,6 @@ int perf_output_begin(struct perf_output_handle *handle,
 
 	handle->rb	= rb;
 	handle->event	= event;
-	handle->nmi	= nmi;
 	handle->sample	= sample;
 
 	if (!rb->nr_pages)
diff --git a/kernel/sched.c b/kernel/sched.c
index 3f2e502d609..d08d110b897 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -2220,7 +2220,7 @@ void set_task_cpu(struct task_struct *p, unsigned int new_cpu)
 
 	if (task_cpu(p) != new_cpu) {
 		p->se.nr_migrations++;
-		perf_sw_event(PERF_COUNT_SW_CPU_MIGRATIONS, 1, 1, NULL, 0);
+		perf_sw_event(PERF_COUNT_SW_CPU_MIGRATIONS, 1, NULL, 0);
 	}
 
 	__set_task_cpu(p, new_cpu);
diff --git a/kernel/watchdog.c b/kernel/watchdog.c
index 752b75ba662..a6708e677a0 100644
--- a/kernel/watchdog.c
+++ b/kernel/watchdog.c
@@ -211,7 +211,7 @@ static struct perf_event_attr wd_hw_attr = {
 };
 
 /* Callback function for perf event subsystem */
-static void watchdog_overflow_callback(struct perf_event *event, int nmi,
+static void watchdog_overflow_callback(struct perf_event *event,
 		 struct perf_sample_data *data,
 		 struct pt_regs *regs)
 {