diff --git a/sched/irq/irq_csection.c b/sched/irq/irq_csection.c
index a688b8b44c..bc32f37f8e 100644
--- a/sched/irq/irq_csection.c
+++ b/sched/irq/irq_csection.c
@@ -615,7 +615,7 @@ bool irq_cpu_locked(int cpu)
       return false;
     }
 
-#ifdef CONFIG_ARCH_HAVE_FETCHADD
+#if defined(CONFIG_ARCH_HAVE_FETCHADD) && !defined(CONFIG_ARCH_GLOBAL_IRQDISABLE)
   /* If the global lockcount has been incremented then simply return true */
 
   if (g_global_lockcount > 0)
diff --git a/sched/sched/sched.h b/sched/sched/sched.h
index 0c1c681e9e..dca36d1f6c 100644
--- a/sched/sched/sched.h
+++ b/sched/sched/sched.h
@@ -374,7 +374,7 @@ extern volatile cpu_set_t g_cpu_lockset SP_SECTION;
 
 extern volatile spinlock_t g_cpu_tasklistlock SP_SECTION;
 
-#ifdef CONFIG_ARCH_HAVE_FETCHADD
+#if defined(CONFIG_ARCH_HAVE_FETCHADD) && !defined(CONFIG_ARCH_GLOBAL_IRQDISABLE)
 /* This is part of the sched_lock() logic to handle atomic operations when
  * locking the scheduler.
  */
@@ -451,7 +451,7 @@ int  sched_cpu_pause(FAR struct tcb_s *tcb);
 irqstate_t sched_tasklist_lock(void);
 void sched_tasklist_unlock(irqstate_t lock);
 
-#ifdef CONFIG_ARCH_HAVE_FETCHADD
+#if defined(CONFIG_ARCH_HAVE_FETCHADD) && !defined(CONFIG_ARCH_GLOBAL_IRQDISABLE)
 #  define sched_islocked_global() \
      (spin_islocked(&g_cpu_schedlock) || g_global_lockcount > 0)
 #else
diff --git a/sched/sched/sched_lock.c b/sched/sched/sched_lock.c
index 3722e4a3a1..1f2d3b9a15 100644
--- a/sched/sched/sched_lock.c
+++ b/sched/sched/sched_lock.c
@@ -43,6 +43,9 @@
 #include <sched.h>
 #include <assert.h>
 
+#include <arch/irq.h>
+
+#include <nuttx/irq.h>
 #include <nuttx/arch.h>
 #include <nuttx/sched_note.h>
 
@@ -116,7 +119,7 @@ volatile spinlock_t g_cpu_schedlock SP_SECTION = SP_UNLOCKED;
 volatile spinlock_t g_cpu_locksetlock SP_SECTION;
 volatile cpu_set_t g_cpu_lockset SP_SECTION;
 
-#ifdef CONFIG_ARCH_HAVE_FETCHADD
+#if defined(CONFIG_ARCH_HAVE_FETCHADD) && !defined(CONFIG_ARCH_GLOBAL_IRQDISABLE)
 /* This is part of the sched_lock() logic to handle atomic operations when
  * locking the scheduler.
  */
@@ -152,15 +155,30 @@ volatile int16_t g_global_lockcount;
 int sched_lock(void)
 {
   FAR struct tcb_s *rtcb;
+#if defined(CONFIG_ARCH_GLOBAL_IRQDISABLE)
+  irqstate_t flags;
+#endif
   int cpu;
 
-  /* The following operation is non-atomic unless CONFIG_ARCH_HAVE_FETCHADD
-   * defined.
+  /* The following operation is non-atomic unless CONFIG_ARCH_GLOBAL_IRQDISABLE
+   * or CONFIG_ARCH_HAVE_FETCHADD is defined.
    */
 
-#ifdef CONFIG_ARCH_HAVE_FETCHADD
-   DEBUGASSERT((uint16_t)g_global_lockcount < INT16_MAX); /* Not atomic! */
-   (void)up_fetchadd16(&g_global_lockcount, 1);
+#if defined(CONFIG_ARCH_GLOBAL_IRQDISABLE)
+  /* If the CPU supports suppression of interprocessor interrupts, then simple
+   * disabling interrupts will provide sufficient protection for the following
+   * operation.
+   */
+
+  flags = up_irq_save();
+
+#elif defined(CONFIG_ARCH_HAVE_FETCHADD)
+  /* If the CPU supports an atomic fetch add operation, then we can use the
+   * global lockcount to assure that the following operation is atomic.
+   */
+
+  DEBUGASSERT((uint16_t)g_global_lockcount < INT16_MAX); /* Not atomic! */
+  (void)up_fetchadd16(&g_global_lockcount, 1);
 #endif
 
   /* This operation is save if CONFIG_ARCH_HAVE_FETCHADD is defined.  NOTE
@@ -234,9 +252,11 @@ int sched_lock(void)
                              TSTATE_TASK_PENDING);
     }
 
-#ifdef CONFIG_ARCH_HAVE_FETCHADD
-   DEBUGASSERT(g_global_lockcount > 0);
-   (void)up_fetchsub16(&g_global_lockcount, 1);
+#if defined(CONFIG_ARCH_GLOBAL_IRQDISABLE)
+  up_irq_restore(flags);
+#elif defined(CONFIG_ARCH_HAVE_FETCHADD)
+  DEBUGASSERT(g_global_lockcount > 0);
+  (void)up_fetchsub16(&g_global_lockcount, 1);
 #endif
 
   return OK;