perf/core: Fix event_function_local()

Vincent reported triggering the WARN_ON_ONCE() in event_function_local().

While thinking through cases I noticed that by using event_function()
directly, we miss the inactive case usually handled by
event_function_call().

Therefore construct a blend of event_function_call() and
event_function() that handles the cases relevant to
event_function_local().

Reported-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org # 4.5+
Fixes: fae3fde65138 ("perf: Collapse and fix event_function_call() users")
Signed-off-by: Ingo Molnar <mingo@kernel.org>

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 1903b8f..6e454bf 100644 (file)
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -242,18 +242,6 @@ unlock:
        return ret;
 }
 
-static void event_function_local(struct perf_event *event, event_f func, void *data)
-{
-       struct event_function_struct efs = {
-               .event = event,
-               .func = func,
-               .data = data,
-       };
-
-       int ret = event_function(&efs);
-       WARN_ON_ONCE(ret);
-}
-
 static void event_function_call(struct perf_event *event, event_f func, void *data)
 {
        struct perf_event_context *ctx = event->ctx;
@@ -303,6 +291,54 @@ again:
        raw_spin_unlock_irq(&ctx->lock);
 }
 
+/*
+ * Similar to event_function_call() + event_function(), but hard assumes IRQs
+ * are already disabled and we're on the right CPU.
+ */
+static void event_function_local(struct perf_event *event, event_f func, void *data)
+{
+       struct perf_event_context *ctx = event->ctx;
+       struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);
+       struct task_struct *task = READ_ONCE(ctx->task);
+       struct perf_event_context *task_ctx = NULL;
+
+       WARN_ON_ONCE(!irqs_disabled());
+
+       if (task) {
+               if (task == TASK_TOMBSTONE)
+                       return;
+
+               task_ctx = ctx;
+       }
+
+       perf_ctx_lock(cpuctx, task_ctx);
+
+       task = ctx->task;
+       if (task == TASK_TOMBSTONE)
+               goto unlock;
+
+       if (task) {
+               /*
+                * We must be either inactive or active and the right task,
+                * otherwise we're screwed, since we cannot IPI to somewhere
+                * else.
+                */
+               if (ctx->is_active) {
+                       if (WARN_ON_ONCE(task != current))
+                               goto unlock;
+
+                       if (WARN_ON_ONCE(cpuctx->task_ctx != ctx))
+                               goto unlock;
+               }
+       } else {
+               WARN_ON_ONCE(&cpuctx->ctx != ctx);
+       }
+
+       func(event, cpuctx, ctx, data);
+unlock:
+       perf_ctx_unlock(cpuctx, task_ctx);
+}
+
 #define PERF_FLAG_ALL (PERF_FLAG_FD_NO_GROUP |\
                       PERF_FLAG_FD_OUTPUT  |\
                       PERF_FLAG_PID_CGROUP |\