perf: Fix perf_enable_on_exec() event scheduling

There are two problems with the current perf_enable_on_exec() event
scheduling:

  - the newly enabled events will be immediately scheduled
    irrespective of their ctx event list order.

  - there's a hole in the ctx->lock between scheduling the events
    out and putting them back on.

Esp. the latter issue is a real problem because a hole in event
scheduling leaves the thing in an observable inconsistent state,
confusing things.

Fix both issues by first doing the enable iteration and at the end,
when there are newly enabled events, reschedule the ctx in one go.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

1 files changed, 27 insertions, 20 deletions

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 751538c..0679e73 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -2036,7 +2036,8 @@ static void add_event_to_ctx(struct perf_event *event,
 	event->tstamp_stopped = tstamp;
 }
 
-static void task_ctx_sched_out(struct perf_event_context *ctx);
+static void task_ctx_sched_out(struct perf_cpu_context *cpuctx,
+			       struct perf_event_context *ctx);
 static void
 ctx_sched_in(struct perf_event_context *ctx,
 	     struct perf_cpu_context *cpuctx,
@@ -2067,6 +2068,17 @@ static void ___perf_install_in_context(void *info)
 	add_event_to_ctx(event, ctx);
 }
 
+static void ctx_resched(struct perf_cpu_context *cpuctx,
+			struct perf_event_context *task_ctx)
+{
+	perf_pmu_disable(cpuctx->ctx.pmu);
+	if (task_ctx)
+		task_ctx_sched_out(cpuctx, task_ctx);
+	cpu_ctx_sched_out(cpuctx, EVENT_ALL);
+	perf_event_sched_in(cpuctx, task_ctx, current);
+	perf_pmu_enable(cpuctx->ctx.pmu);
+}
+
 /*
  * Cross CPU call to install and enable a performance event
  *
@@ -2087,7 +2099,7 @@ static int  __perf_install_in_context(void *info)
 	 * If there was an active task_ctx schedule it out.
 	 */
 	if (task_ctx)
-		task_ctx_sched_out(task_ctx);
+		task_ctx_sched_out(cpuctx, task_ctx);
 
 	/*
 	 * If the context we're installing events in is not the
@@ -2629,10 +2641,9 @@ void __perf_event_task_sched_out(struct task_struct *task,
 		perf_cgroup_sched_out(task, next);
 }
 
-static void task_ctx_sched_out(struct perf_event_context *ctx)
+static void task_ctx_sched_out(struct perf_cpu_context *cpuctx,
+			       struct perf_event_context *ctx)
 {
-	struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);
-
 	if (!cpuctx->task_ctx)
 		return;
 
@@ -3096,34 +3107,30 @@ static int event_enable_on_exec(struct perf_event *event,
 static void perf_event_enable_on_exec(int ctxn)
 {
 	struct perf_event_context *ctx, *clone_ctx = NULL;
+	struct perf_cpu_context *cpuctx;
 	struct perf_event *event;
 	unsigned long flags;
 	int enabled = 0;
-	int ret;
 
 	local_irq_save(flags);
 	ctx = current->perf_event_ctxp[ctxn];
 	if (!ctx || !ctx->nr_events)
 		goto out;
 
-	raw_spin_lock(&ctx->lock);
-	task_ctx_sched_out(ctx);
-
-	list_for_each_entry(event, &ctx->event_list, event_entry) {
-		ret = event_enable_on_exec(event, ctx);
-		if (ret)
-			enabled = 1;
-	}
+	cpuctx = __get_cpu_context(ctx);
+	perf_ctx_lock(cpuctx, ctx);
+	list_for_each_entry(event, &ctx->event_list, event_entry)
+		enabled |= event_enable_on_exec(event, ctx);
 
 	/*
-	 * Unclone this context if we enabled any event.
+	 * Unclone and reschedule this context if we enabled any event.
 	 */
-	if (enabled)
+	if (enabled) {
 		clone_ctx = unclone_ctx(ctx);
+		ctx_resched(cpuctx, ctx);
+	}
+	perf_ctx_unlock(cpuctx, ctx);
 
-	raw_spin_unlock(&ctx->lock);
-
-	perf_event_context_sched_in(ctx, ctx->task);
 out:
 	local_irq_restore(flags);
 
@@ -8737,7 +8744,7 @@ static void perf_event_exit_task_context(struct task_struct *child, int ctxn)
 	 * incremented the context's refcount before we do put_ctx below.
 	 */
 	raw_spin_lock(&child_ctx->lock);
-	task_ctx_sched_out(child_ctx);
+	task_ctx_sched_out(__get_cpu_context(child_ctx), child_ctx);
 	child->perf_event_ctxp[ctxn] = NULL;
 
 	/*