From f313ca4daefa8aee7837f545a4b43795bd692789 Mon Sep 17 00:00:00 2001
From: Ray Strode <rstrode@redhat.com>
Date: Tue, 23 Aug 2016 16:53:39 -0400
Subject: [PATCH] lib: Add full barrier before writing data_tail

The kernel says here http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/kernel/events/ring_buffer.c?id=7a1dcf6ad#n61 :

	 * Since the mmap() consumer (userspace) can run on a different CPU:
	 *
	 *   kernel				user
	 *
	 *   if (LOAD ->data_tail) {		LOAD ->data_head
	 *			(A)		smp_rmb()	(C)
	 *	STORE $data			LOAD $data
	 *	smp_wmb()	(B)		smp_mb()	(D)
	 *	STORE ->data_head		STORE ->data_tail
	 *   }
	 *
	 * Where A pairs with D, and B pairs with C.
	 *
	 * In our case (A) is a control dependency that separates the load of
	 * the ->data_tail and the stores of $data. In case ->data_tail
	 * indicates there is no room in the buffer to store $data we do not.
	 *
	 * D needs to be a full barrier since it separates the data READ
	 * from the tail WRITE.
	 *
	 * For B a WMB is sufficient since it separates two WRITEs, and for C
	 * an RMB is sufficient since it separates two READs.
	 *
	 * See perf_output_begin().

So I'm pretty sure we need a full barrier before writing out data_tail.
---
 lib/sp-perf-counter.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/sp-perf-counter.c b/lib/sp-perf-counter.c
index bd7315b3..d41bc9ea 100644
--- a/lib/sp-perf-counter.c
+++ b/lib/sp-perf-counter.c
@@ -291,6 +291,8 @@ sp_perf_counter_flush (SpPerfCounter     *self,
     }
 
   info->tail = tail;
+
+  atomic_thread_fence (memory_order_seq_cst);
   info->map->data_tail = tail;
 }