diff --git a/tools/firmware/hvmloader/util.h b/tools/firmware/hvmloader/util.h
index 31889de634..4f0baade0e 100644
--- a/tools/firmware/hvmloader/util.h
+++ b/tools/firmware/hvmloader/util.h
@@ -133,7 +133,7 @@ static inline void cpu_relax(void)
 #define barrier() asm volatile ( "" : : : "memory" )
 #define rmb()     barrier()
 #define wmb()     barrier()
-#define mb()      asm volatile ( "lock; addl $0,0(%%esp)" : : : "memory" )
+#define mb()      asm volatile ( "lock addl $0, -4(%%esp)" ::: "memory" )
 
 /*
  * Divide a 64-bit dividend by a 32-bit divisor.
diff --git a/tools/libs/ctrl/include/xenctrl.h b/tools/libs/ctrl/include/xenctrl.h
index ba70bec9c4..3796425e1e 100644
--- a/tools/libs/ctrl/include/xenctrl.h
+++ b/tools/libs/ctrl/include/xenctrl.h
@@ -68,11 +68,11 @@
 #define xen_barrier() asm volatile ( "" : : : "memory")
 
 #if defined(__i386__)
-#define xen_mb()  asm volatile ( "lock; addl $0,0(%%esp)" : : : "memory" )
+#define xen_mb()  asm volatile ( "lock addl $0, -4(%%esp)" ::: "memory" )
 #define xen_rmb() xen_barrier()
 #define xen_wmb() xen_barrier()
 #elif defined(__x86_64__)
-#define xen_mb()  asm volatile ( "mfence" : : : "memory")
+#define xen_mb()  asm volatile ( "lock addl $0, -32(%%rsp)" ::: "memory" )
 #define xen_rmb() xen_barrier()
 #define xen_wmb() xen_barrier()
 #elif defined(__arm__)
diff --git a/xen/include/asm-x86/system.h b/xen/include/asm-x86/system.h
index 45c183bd10..630909965e 100644
--- a/xen/include/asm-x86/system.h
+++ b/xen/include/asm-x86/system.h
@@ -220,7 +220,7 @@ static always_inline unsigned long __xadd(
  *
  * Refer to the vendor system programming manuals for further details.
  */
-#define smp_mb()        mb()
+#define smp_mb()        asm volatile ( "lock addl $0, -4(%%rsp)" ::: "memory" )
 #define smp_rmb()       barrier()
 #define smp_wmb()       barrier()