diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c
index f98c4076a170..1b43086b097a 100644
--- a/arch/x86/kernel/ftrace.c
+++ b/arch/x86/kernel/ftrace.c
@@ -476,7 +476,10 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr)
 				&return_to_handler;
 
 	/* Nmi's are currently unsupported */
-	if (atomic_read(&in_nmi))
+	if (unlikely(atomic_read(&in_nmi)))
+		return;
+
+	if (unlikely(atomic_read(&current->tracing_graph_pause)))
 		return;
 
 	/*
diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index 449fa8e9e34f..11cac81eed08 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h
@@ -401,6 +401,16 @@ static inline int task_curr_ret_stack(struct task_struct *t)
 {
 	return t->curr_ret_stack;
 }
+
+static inline void pause_graph_tracing(void)
+{
+	atomic_inc(&current->tracing_graph_pause);
+}
+
+static inline void unpause_graph_tracing(void)
+{
+	atomic_dec(&current->tracing_graph_pause);
+}
 #else
 
 #define __notrace_funcgraph
@@ -412,6 +422,9 @@ static inline int task_curr_ret_stack(struct task_struct *tsk)
 {
 	return -1;
 }
+
+static inline void pause_graph_tracing(void) { }
+static inline void unpause_graph_tracing(void) { }
 #endif
 
 #ifdef CONFIG_TRACING
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 4c152e0acc9e..4b81fc5f7731 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1379,6 +1379,8 @@ struct task_struct {
 	 * because of depth overrun.
 	 */
 	atomic_t trace_overrun;
+	/* Pause for the tracing */
+	atomic_t tracing_graph_pause;
 #endif
 #ifdef CONFIG_TRACING
 	/* state flags for use by tracers */
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 2971fe48f55e..a12f80efceaa 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -1998,6 +1998,7 @@ static int alloc_retstack_tasklist(struct ftrace_ret_stack **ret_stack_list)
 			/* Make sure IRQs see the -1 first: */
 			barrier();
 			t->ret_stack = ret_stack_list[start++];
+			atomic_set(&t->tracing_graph_pause, 0);
 			atomic_set(&t->trace_overrun, 0);
 		}
 	} while_each_thread(g, t);
@@ -2077,6 +2078,7 @@ void ftrace_graph_init_task(struct task_struct *t)
 		if (!t->ret_stack)
 			return;
 		t->curr_ret_stack = -1;
+		atomic_set(&t->tracing_graph_pause, 0);
 		atomic_set(&t->trace_overrun, 0);
 	} else
 		t->ret_stack = NULL;
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 33549537f30f..0b8659bd5ad2 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -3590,14 +3590,7 @@ static __init int tracer_init_debugfs(void)
 
 int trace_vprintk(unsigned long ip, int depth, const char *fmt, va_list args)
 {
-	/*
-	 * Raw Spinlock because a normal spinlock would be traced here
-	 * and append an irrelevant couple spin_lock_irqsave/
-	 * spin_unlock_irqrestore traced by ftrace around this
-	 * TRACE_PRINTK trace.
-	 */
-	static raw_spinlock_t trace_buf_lock =
-				(raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED;
+	static DEFINE_SPINLOCK(trace_buf_lock);
 	static char trace_buf[TRACE_BUF_SIZE];
 
 	struct ring_buffer_event *event;
@@ -3618,8 +3611,8 @@ int trace_vprintk(unsigned long ip, int depth, const char *fmt, va_list args)
 	if (unlikely(atomic_read(&data->disabled)))
 		goto out;
 
-	local_irq_save(flags);
-	__raw_spin_lock(&trace_buf_lock);
+	pause_graph_tracing();
+	spin_lock_irqsave(&trace_buf_lock, irq_flags);
 	len = vsnprintf(trace_buf, TRACE_BUF_SIZE, fmt, args);
 
 	len = min(len, TRACE_BUF_SIZE-1);
@@ -3640,9 +3633,8 @@ int trace_vprintk(unsigned long ip, int depth, const char *fmt, va_list args)
 	ring_buffer_unlock_commit(tr->buffer, event, irq_flags);
 
  out_unlock:
-	__raw_spin_unlock(&trace_buf_lock);
-	local_irq_restore(flags);
-
+	spin_unlock_irqrestore(&trace_buf_lock, irq_flags);
+	unpause_graph_tracing();
  out:
 	preempt_enable_notrace();