diff --git a/tools/perf/bench/futex-hash.c b/tools/perf/bench/futex-hash.c
index 0999ac536d869c57b0dcb02505148698b9470bd5..8fc1782f90f529ef5c2c0575506607aa7a32b12c 100644
--- a/tools/perf/bench/futex-hash.c
+++ b/tools/perf/bench/futex-hash.c
@@ -8,8 +8,10 @@
  * many threads and futexes as possible.
  */
 
+/* For the CLR_() macros */
+#include <pthread.h>
+
 #include "../perf.h"
-#include "../util/util.h"
 #include "../util/stat.h"
 #include <subcmd/parse-options.h>
 #include "../util/header.h"
@@ -17,9 +19,7 @@
 #include "futex.h"
 
 #include <err.h>
-#include <stdlib.h>
 #include <sys/time.h>
-#include <pthread.h>
 
 static unsigned int nthreads = 0;
 static unsigned int nsecs    = 10;
diff --git a/tools/perf/bench/futex-lock-pi.c b/tools/perf/bench/futex-lock-pi.c
index 6952db65508abced08317b18e8262cc72881b32b..be95506cc0d153ad9bd759990391fc410ef82684 100644
--- a/tools/perf/bench/futex-lock-pi.c
+++ b/tools/perf/bench/futex-lock-pi.c
@@ -2,6 +2,9 @@
  * Copyright (C) 2015 Davidlohr Bueso.
  */
 
+/* For the CLR_() macros */
+#include <pthread.h>
+
 #include "../perf.h"
 #include "../util/util.h"
 #include "../util/stat.h"
@@ -13,7 +16,6 @@
 #include <err.h>
 #include <stdlib.h>
 #include <sys/time.h>
-#include <pthread.h>
 
 struct worker {
 	int tid;
diff --git a/tools/perf/bench/futex-requeue.c b/tools/perf/bench/futex-requeue.c
index 71823868301347a47fdc7d31723dd325426ac3a5..7d9712c8ff01a8d208b32215a2b15e875e84eb87 100644
--- a/tools/perf/bench/futex-requeue.c
+++ b/tools/perf/bench/futex-requeue.c
@@ -8,6 +8,9 @@
  * requeues without waking up any tasks -- thus mimicking a regular futex_wait.
  */
 
+/* For the CLR_() macros */
+#include <pthread.h>
+
 #include "../perf.h"
 #include "../util/util.h"
 #include "../util/stat.h"
@@ -19,7 +22,6 @@
 #include <err.h>
 #include <stdlib.h>
 #include <sys/time.h>
-#include <pthread.h>
 
 static u_int32_t futex1 = 0, futex2 = 0;
 
diff --git a/tools/perf/bench/futex-wake-parallel.c b/tools/perf/bench/futex-wake-parallel.c
index 91aaf2a1fa9050dae0a866e6a851777a87c37a0b..aca12284d8af2bc3f887aa959c61e3520f593c0c 100644
--- a/tools/perf/bench/futex-wake-parallel.c
+++ b/tools/perf/bench/futex-wake-parallel.c
@@ -7,6 +7,9 @@
  * it can be used to measure futex_wake() changes.
  */
 
+/* For the CLR_() macros */
+#include <pthread.h>
+
 #include "../perf.h"
 #include "../util/util.h"
 #include "../util/stat.h"
@@ -18,7 +21,6 @@
 #include <err.h>
 #include <stdlib.h>
 #include <sys/time.h>
-#include <pthread.h>
 
 struct thread_data {
 	pthread_t worker;
diff --git a/tools/perf/bench/futex-wake.c b/tools/perf/bench/futex-wake.c
index f416bd705f661f056da8e1424643922f1999600d..877e680b28a8c9139d77050f0c577cbe60ea7fc5 100644
--- a/tools/perf/bench/futex-wake.c
+++ b/tools/perf/bench/futex-wake.c
@@ -8,6 +8,9 @@
  * one or more tasks, and thus the waitqueue is never empty.
  */
 
+/* For the CLR_() macros */
+#include <pthread.h>
+
 #include "../perf.h"
 #include "../util/util.h"
 #include "../util/stat.h"
@@ -19,7 +22,6 @@
 #include <err.h>
 #include <stdlib.h>
 #include <sys/time.h>
-#include <pthread.h>
 
 /* all threads will block on the same futex */
 static u_int32_t futex1 = 0;
diff --git a/tools/perf/bench/numa.c b/tools/perf/bench/numa.c
index 7500d959d7eb0002e4d5da138e6b8365bc51db8e..f7f530081aa9421a0e01ae362fa02c3192c2896c 100644
--- a/tools/perf/bench/numa.c
+++ b/tools/perf/bench/numa.c
@@ -4,6 +4,9 @@
  * numa: Simulate NUMA-sensitive workload and measure their NUMA performance
  */
 
+/* For the CLR_() macros */
+#include <pthread.h>
+
 #include "../perf.h"
 #include "../builtin.h"
 #include "../util/util.h"
@@ -21,7 +24,6 @@
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
-#include <pthread.h>
 #include <sys/mman.h>
 #include <sys/time.h>
 #include <sys/resource.h>