diff --git a/arch/i386/lib/delay.c b/arch/i386/lib/delay.c
index f6edb11364df..66595edcd7b6 100644
--- a/arch/i386/lib/delay.c
+++ b/arch/i386/lib/delay.c
@@ -12,6 +12,7 @@
 
 #include <linux/module.h>
 #include <linux/sched.h>
+#include <linux/preempt.h>
 #include <linux/delay.h>
 
 #include <asm/processor.h>
@@ -42,11 +43,13 @@ static void delay_tsc(unsigned long loops)
 {
 	unsigned long bclock, now;
 
+	preempt_disable();		/* TSC's are per-cpu */
 	rdtscl(bclock);
 	do {
 		rep_nop();
 		rdtscl(now);
 	} while ((now-bclock) < loops);
+	preempt_enable();
 }
 
 /*
diff --git a/arch/x86_64/lib/delay.c b/arch/x86_64/lib/delay.c
index 2dbebd308347..4d3f1f641336 100644
--- a/arch/x86_64/lib/delay.c
+++ b/arch/x86_64/lib/delay.c
@@ -10,7 +10,9 @@
 
 #include <linux/module.h>
 #include <linux/sched.h>
+#include <linux/preempt.h>
 #include <linux/delay.h>
+
 #include <asm/delay.h>
 #include <asm/msr.h>
 
@@ -27,14 +29,15 @@ int read_current_timer(unsigned long *timer_value)
 void __delay(unsigned long loops)
 {
 	unsigned bclock, now;
-	
+
+	preempt_disable();		/* TSC's are pre-cpu */
 	rdtscl(bclock);
-	do
-	{
+	do {
 		rep_nop(); 
 		rdtscl(now);
 	}
-	while((now-bclock) < loops);
+	while ((now-bclock) < loops);
+	preempt_enable();
 }
 EXPORT_SYMBOL(__delay);