diff --git a/arch/powerpc/include/asm/cputime.h b/arch/powerpc/include/asm/cputime.h
index 607559ab271ff98b45de1f11964416bebaf32841..6c840ceab8201cc191f4699d15561c62b2e0c9c2 100644
--- a/arch/powerpc/include/asm/cputime.h
+++ b/arch/powerpc/include/asm/cputime.h
@@ -32,6 +32,8 @@ static inline void setup_cputime_one_jiffy(void) { }
 typedef u64 __nocast cputime_t;
 typedef u64 __nocast cputime64_t;
 
+#define cmpxchg_cputime(ptr, old, new) cmpxchg(ptr, old, new)
+
 #ifdef __KERNEL__
 
 /*
diff --git a/arch/s390/include/asm/cputime.h b/arch/s390/include/asm/cputime.h
index f65bd36345194db88b71bca1f3740cca8744fa33..3001887f94b7a7c31ca01051db3f3cf9643e3ad0 100644
--- a/arch/s390/include/asm/cputime.h
+++ b/arch/s390/include/asm/cputime.h
@@ -18,6 +18,8 @@
 typedef unsigned long long __nocast cputime_t;
 typedef unsigned long long __nocast cputime64_t;
 
+#define cmpxchg_cputime(ptr, old, new) cmpxchg64(ptr, old, new)
+
 static inline unsigned long __div(unsigned long long n, unsigned long base)
 {
 #ifndef CONFIG_64BIT
diff --git a/include/asm-generic/cputime_jiffies.h b/include/asm-generic/cputime_jiffies.h
index d5cb78f539861afdc82eae345ee4c8148fb2448e..fe386fc6e85e27f3c4a1b99262fe023db6a31fc3 100644
--- a/include/asm-generic/cputime_jiffies.h
+++ b/include/asm-generic/cputime_jiffies.h
@@ -3,6 +3,8 @@
 
 typedef unsigned long __nocast cputime_t;
 
+#define cmpxchg_cputime(ptr, old, new) cmpxchg(ptr, old, new)
+
 #define cputime_one_jiffy		jiffies_to_cputime(1)
 #define cputime_to_jiffies(__ct)	(__force unsigned long)(__ct)
 #define cputime_to_scaled(__ct)		(__ct)
diff --git a/include/asm-generic/cputime_nsecs.h b/include/asm-generic/cputime_nsecs.h
index 4e817606c5494126a31ae50e58a6e76242d89f07..0419485891f2a8eb6125f154f52d7b16c5fb05b4 100644
--- a/include/asm-generic/cputime_nsecs.h
+++ b/include/asm-generic/cputime_nsecs.h
@@ -21,6 +21,8 @@
 typedef u64 __nocast cputime_t;
 typedef u64 __nocast cputime64_t;
 
+#define cmpxchg_cputime(ptr, old, new) cmpxchg64(ptr, old, new)
+
 #define cputime_one_jiffy		jiffies_to_cputime(1)
 
 #define cputime_div(__ct, divisor)  div_u64((__force u64)__ct, divisor)
diff --git a/kernel/sched/cputime.c b/kernel/sched/cputime.c
index 64492dff8a819fc56ac37c656ac6831c1209d58b..8394b1ee600c38ba6e9144a6326369b6ef0cdacd 100644
--- a/kernel/sched/cputime.c
+++ b/kernel/sched/cputime.c
@@ -554,6 +554,23 @@ static cputime_t scale_stime(u64 stime, u64 rtime, u64 total)
 	return (__force cputime_t) scaled;
 }
 
+/*
+ * Atomically advance counter to the new value. Interrupts, vcpu
+ * scheduling, and scaling inaccuracies can cause cputime_advance
+ * to be occasionally called with a new value smaller than counter.
+ * Let's enforce atomicity.
+ *
+ * Normally a caller will only go through this loop once, or not
+ * at all in case a previous caller updated counter the same jiffy.
+ */
+static void cputime_advance(cputime_t *counter, cputime_t new)
+{
+	cputime_t old;
+
+	while (new > (old = ACCESS_ONCE(*counter)))
+		cmpxchg_cputime(counter, old, new);
+}
+
 /*
  * Adjust tick based cputime random precision against scheduler
  * runtime accounting.
@@ -599,16 +616,8 @@ static void cputime_adjust(struct task_cputime *curr,
 		utime = rtime - stime;
 	}
 
-	/*
-	 * If the tick based count grows faster than the scheduler one,
-	 * the result of the scaling may go backward.
-	 * Let's enforce monotonicity.
-	 * Atomic exchange protects against concurrent cputime_adjust().
-	 */
-	while (stime > (rtime = ACCESS_ONCE(prev->stime)))
-		cmpxchg(&prev->stime, rtime, stime);
-	while (utime > (rtime = ACCESS_ONCE(prev->utime)))
-		cmpxchg(&prev->utime, rtime, utime);
+	cputime_advance(&prev->stime, stime);
+	cputime_advance(&prev->utime, utime);
 
 out:
 	*ut = prev->utime;