diff --git a/arch/arm/src/arm/arm_assert.c b/arch/arm/src/arm/arm_assert.c
index e4f445bb8069e..4957af560b9eb 100644
--- a/arch/arm/src/arm/arm_assert.c
+++ b/arch/arm/src/arm/arm_assert.c
@@ -210,18 +210,12 @@ static void up_dumpstate(void)
    * stack?
    */
 
-  if (sp <= istackbase && sp > istackbase - istacksize)
+  if (sp < istackbase && sp > istackbase - istacksize)
     {
       /* Yes.. dump the interrupt stack */
 
+      _alert("Interrupt Stack\n", sp);
       up_stackdump(sp, istackbase);
-
-      /* Extract the user stack pointer which should lie
-       * at the base of the interrupt stack.
-       */
-
-      sp = g_intstackbase;
-      _alert("sp:     %08x\n", sp);
     }
   else if (CURRENT_REGS)
     {
@@ -229,6 +223,17 @@ static void up_dumpstate(void)
       up_stackdump(istackbase - istacksize, istackbase);
     }
 
+  /* Extract the user stack pointer if we are in an interrupt handler.
+   * If we are not in an interrupt handler.  Then sp is the user stack
+   * pointer (and the above range check should have failed).
+   */
+
+  if (CURRENT_REGS)
+    {
+      sp = CURRENT_REGS[REG_R13];
+      _alert("User sp: %08x\n", sp);
+    }
+
   /* Show user stack info */
 
   _alert("User stack:\n");
diff --git a/arch/arm/src/arm/arm_vectors.S b/arch/arm/src/arm/arm_vectors.S
index c03f091edad5e..9606131192af4 100644
--- a/arch/arm/src/arm/arm_vectors.S
+++ b/arch/arm/src/arm/arm_vectors.S
@@ -132,7 +132,7 @@ arm_vectorirq:
 
 #if CONFIG_ARCH_INTERRUPTSTACK > 3
 	ldr	sp, .Lirqstackbase	/* SP = interrupt stack base */
-	str	r0, [sp]		/* Save the user stack pointer */
+	str	r0, [sp, #-4]!		/* Save the xcp address at SP-4 then update SP */
 	bl	arm_decodeirq		/* Call the handler */
 	ldr	sp, [sp]		/* Restore the user stack pointer */
 #else
@@ -432,13 +432,13 @@ arm_vectorfiq:
 
 #if CONFIG_ARCH_INTERRUPTSTACK > 3
 	.bss
-	.align	4
+	.balign	4
 	.globl	g_intstackalloc
 	.type	g_intstackalloc, object
 	.globl	g_intstackbase
 	.type	g_intstackbase, object
 g_intstackalloc:
-	.skip	((CONFIG_ARCH_INTERRUPTSTACK & ~3) - 4)
+	.skip	(CONFIG_ARCH_INTERRUPTSTACK & ~3)
 g_intstackbase:
 	.skip	4
 	.size	g_intstackbase, 4
diff --git a/arch/arm/src/armv6-m/arm_assert.c b/arch/arm/src/armv6-m/arm_assert.c
index 382a039a3943d..11c97ba1e72c4 100644
--- a/arch/arm/src/armv6-m/arm_assert.c
+++ b/arch/arm/src/armv6-m/arm_assert.c
@@ -247,7 +247,7 @@ static void up_dumpstate(void)
    * stack?
    */
 
-  if (sp <= istackbase && sp > istackbase - istacksize)
+  if (sp < istackbase && sp > istackbase - istacksize)
     {
       /* Yes.. dump the interrupt stack */
 
diff --git a/arch/arm/src/armv6-m/arm_exception.S b/arch/arm/src/armv6-m/arm_exception.S
index c70c57a2eb969..ecc6c04df29ea 100644
--- a/arch/arm/src/armv6-m/arm_exception.S
+++ b/arch/arm/src/armv6-m/arm_exception.S
@@ -270,7 +270,7 @@ exception_common:
 	.bss
 	.global	g_intstackalloc
 	.global g_intstackbase
-	.align	4
+	.balign	4
 g_intstackalloc:
 	.skip	(CONFIG_ARCH_INTERRUPTSTACK & ~3)
 g_intstackbase:
diff --git a/arch/arm/src/armv7-a/arm_assert.c b/arch/arm/src/armv7-a/arm_assert.c
index 07a2decbf7f97..015aa83578267 100644
--- a/arch/arm/src/armv7-a/arm_assert.c
+++ b/arch/arm/src/armv7-a/arm_assert.c
@@ -252,7 +252,7 @@ static void up_dumpstate(void)
   if (rtcb->xcp.kstack)
     {
       kstackbase = (uint32_t)rtcb->xcp.kstack +
-                   CONFIG_ARCH_KERNEL_STACKSIZE - 4;
+                   CONFIG_ARCH_KERNEL_STACKSIZE;
 
       _alert("Kernel stack:\n");
       _alert("  base: %08x\n", kstackbase);
@@ -265,24 +265,10 @@ static void up_dumpstate(void)
 
   if (sp > istackbase - istacksize && sp < istackbase)
     {
-      uint32_t *stackbase;
-
       /* Yes.. dump the interrupt stack */
 
       _alert("Interrupt Stack\n", sp);
       up_stackdump(sp, istackbase);
-
-      /* Extract the user stack pointer which should lie
-       * at the base of the interrupt stack.
-       */
-
-#ifdef CONFIG_SMP
-      stackbase = (uint32_t *)arm_intstack_base();
-#else
-      stackbase = (uint32_t *)&g_intstackbase;
-#endif
-      sp        = *stackbase;
-      _alert("User sp: %08x\n", sp);
     }
   else if (CURRENT_REGS)
     {
@@ -291,6 +277,17 @@ static void up_dumpstate(void)
     }
 #endif
 
+  /* Extract the user stack pointer if we are in an interrupt handler.
+   * If we are not in an interrupt handler.  Then sp is the user stack
+   * pointer (and the above range check should have failed).
+   */
+
+  if (CURRENT_REGS)
+    {
+      sp = CURRENT_REGS[REG_R13];
+      _alert("User sp: %08x\n", sp);
+    }
+
   /* Dump the user stack if the stack pointer lies within the allocated user
    * stack memory.
    */
diff --git a/arch/arm/src/armv7-a/arm_cpuidlestack.c b/arch/arm/src/armv7-a/arm_cpuidlestack.c
index d389622d63ca3..1acdeae7177ca 100644
--- a/arch/arm/src/armv7-a/arm_cpuidlestack.c
+++ b/arch/arm/src/armv7-a/arm_cpuidlestack.c
@@ -133,10 +133,9 @@ int up_cpu_idlestack(int cpu, FAR struct tcb_s *tcb, size_t stack_size)
 
   /* Get the top of the stack */
 
-
   stack_alloc          = (uintptr_t)g_cpu_stackalloc[cpu];
   DEBUGASSERT(stack_alloc != 0 && STACK_ISALIGNED(stack_alloc));
-  top_of_stack         = stack_alloc + SMP_STACK_TOP;
+  top_of_stack         = stack_alloc + SMP_STACK_SIZE;
 
   tcb->adj_stack_size  = SMP_STACK_SIZE;
   tcb->stack_alloc_ptr = (FAR uint32_t *)stack_alloc;
diff --git a/arch/arm/src/armv7-a/arm_vectors.S b/arch/arm/src/armv7-a/arm_vectors.S
index 6df8755c4c946..e0559048164dc 100644
--- a/arch/arm/src/armv7-a/arm_vectors.S
+++ b/arch/arm/src/armv7-a/arm_vectors.S
@@ -261,7 +261,7 @@ arm_vectorirq:
 	/* Call arm_decodeirq() on the interrupt stack */
 
 	setirqstack r1, r3				/* SP = IRQ stack top */
-	str		r0, [sp]				/* Save the user stack pointer */
+	str		r0, [sp, #-4]!			/* Save the xcp address at SP-4 then update SP */
 	mov		r4, sp					/* Save the SP in a preserved register */
 	bic		sp, sp, #7				/* Force 8-byte alignment */
 	bl		arm_decodeirq			/* Call the handler */
@@ -1004,7 +1004,7 @@ arm_vectorfiq:
 
 #if CONFIG_ARCH_INTERRUPTSTACK > 7
 	setfiqstack r1, r4				/* SP = FIQ stack top */
-	str		r0, [sp]				/* Save the user stack pointer */
+	str		r0, [sp, #-4]!			/* Save the xcp address at SP-4 then update SP */
 	mov		r4, sp					/* Save the SP in a preserved register */
 	bic		sp, sp, #7				/* Force 8-byte alignment */
 	bl		arm_decodefiq			/* Call the handler */
@@ -1072,7 +1072,7 @@ arm_vectorfiq:
 
 #if !defined(CONFIG_SMP) && CONFIG_ARCH_INTERRUPTSTACK > 7
 	.bss
-	.align	4
+	.balign	8
 
 	.globl	g_intstackalloc
 	.type	g_intstackalloc, object
@@ -1080,7 +1080,7 @@ arm_vectorfiq:
 	.type	g_intstackbase, object
 
 g_intstackalloc:
-	.skip	((CONFIG_ARCH_INTERRUPTSTACK & ~7) - 4)
+	.skip	((CONFIG_ARCH_INTERRUPTSTACK + 4) & ~7)
 g_intstackbase:
 	.skip	4
 	.size	g_intstackbase, 4
@@ -1096,7 +1096,7 @@ g_intstackbase:
 	.type	g_fiqstackbase, object
 
 g_fiqstackalloc:
-	.skip	((CONFIG_ARCH_INTERRUPTSTACK & ~7) - 4)
+	.skip	((CONFIG_ARCH_INTERRUPTSTACK + 4) & ~7)
 g_fiqstackbase:
 	.skip	4
 	.size	g_fiqstackbase, 4
diff --git a/arch/arm/src/armv7-a/smp.h b/arch/arm/src/armv7-a/smp.h
index 7f3e37539b61f..e168ce680556d 100644
--- a/arch/arm/src/armv7-a/smp.h
+++ b/arch/arm/src/armv7-a/smp.h
@@ -58,7 +58,6 @@
 #define SMP_STACK_MASK       7
 #define SMP_STACK_SIZE       ((CONFIG_SMP_IDLETHREAD_STACKSIZE + 7) & ~7)
 #define SMP_STACK_WORDS      (SMP_STACK_SIZE >> 2)
-#define SMP_STACK_TOP        (SMP_STACK_SIZE - 8)
 
 /****************************************************************************
  * Public Data
@@ -122,9 +121,9 @@ void __cpu3_start(void);
  *
  * Description:
  *   Continues the C-level initialization started by the assembly language
- *   __cpu[n]_start function.  At a minimum, this function needs to initialize
- *   interrupt handling and, perhaps, wait on WFI for arm_cpu_start() to
- *   issue an SGI.
+ *   __cpu[n]_start function.  At a minimum, this function needs to
+ *   initialize interrupt handling and, perhaps, wait on WFI for
+ *   arm_cpu_start() to issue an SGI.
  *
  *   This function must be provided by the each ARMv7-A MCU and implement
  *   MCU-specific initialization logic.
diff --git a/arch/arm/src/armv7-m/arm_assert.c b/arch/arm/src/armv7-m/arm_assert.c
index 382297f202bdb..c1a29cdd60b2e 100644
--- a/arch/arm/src/armv7-m/arm_assert.c
+++ b/arch/arm/src/armv7-m/arm_assert.c
@@ -244,7 +244,7 @@ static void up_dumpstate(void)
    * stack?
    */
 
-  if (sp <= istackbase && sp > istackbase - istacksize)
+  if (sp < istackbase && sp > istackbase - istacksize)
     {
       /* Yes.. dump the interrupt stack */
 
diff --git a/arch/arm/src/armv7-m/gnu/arm_exception.S b/arch/arm/src/armv7-m/gnu/arm_exception.S
index b9580eeec5732..34e5031264c39 100644
--- a/arch/arm/src/armv7-m/gnu/arm_exception.S
+++ b/arch/arm/src/armv7-m/gnu/arm_exception.S
@@ -67,7 +67,7 @@
    * no privileged task has run.
    */
 
-#  if defined(CONFIG_BUILD_PROTECTED) && CONFIG_ARCH_INTERRUPTSTACK < 4
+#  if defined(CONFIG_BUILD_PROTECTED) && CONFIG_ARCH_INTERRUPTSTACK < 8
 #    error Interrupt stack must be used with high priority interrupts in kernel mode
 #  endif
 
@@ -197,7 +197,7 @@ exception_common:
 	 * here prohibits nested interrupts without some additional logic!
 	 */
 
-	setintstack	r2, r3
+	setintstack	r2, r3				/* SP = IRQ stack top */
 
 #else
 	/* Otherwise, we will re-use the interrupted thread's stack.  That may
@@ -321,7 +321,7 @@ exception_common:
 	.bss
 	.global	g_intstackalloc
 	.global	g_intstackbase
-	.align	8
+	.balign	8
 g_intstackalloc:
 	.skip	((CONFIG_ARCH_INTERRUPTSTACK + 4) & ~7)
 g_intstackbase:
diff --git a/arch/arm/src/armv7-m/gnu/arm_lazyexception.S b/arch/arm/src/armv7-m/gnu/arm_lazyexception.S
index 57ee25096ff8b..9d15f901eeb1a 100644
--- a/arch/arm/src/armv7-m/gnu/arm_lazyexception.S
+++ b/arch/arm/src/armv7-m/gnu/arm_lazyexception.S
@@ -50,7 +50,7 @@
    * nested interrupt, the interrupt stack if no privileged task has run.
    */
 
-#  if defined(CONFIG_BUILD_PROTECTED) && CONFIG_ARCH_INTERRUPTSTACK < 4
+#  if defined(CONFIG_BUILD_PROTECTED) && CONFIG_ARCH_INTERRUPTSTACK < 8
 #    error Interrupt stack must be used with high priority interrupts in kernel mode
 #  endif
 
@@ -192,7 +192,7 @@ exception_common:
 	 * here prohibits nested interrupts without some additional logic!
 	 */
 
-	setintstack	r2, r3
+	setintstack	r2, r3				/* SP = IRQ stack top */
 
 #else
 	/* Otherwise, we will re-use the interrupted thread's stack.  That may
@@ -340,7 +340,7 @@ exception_common:
 	.bss
 	.global	g_intstackalloc
 	.global	g_intstackbase
-	.align	8
+	.balign	8
 g_intstackalloc:
 	.skip	((CONFIG_ARCH_INTERRUPTSTACK + 4) & ~7)
 g_intstackbase:
diff --git a/arch/arm/src/armv7-r/arm_assert.c b/arch/arm/src/armv7-r/arm_assert.c
index c91e2ca1d33bc..ed516c750bee1 100644
--- a/arch/arm/src/armv7-r/arm_assert.c
+++ b/arch/arm/src/armv7-r/arm_assert.c
@@ -195,7 +195,7 @@ static void up_dumpstate(void)
   uint32_t sp   = arm_getsp();
   uint32_t ustackbase;
   uint32_t ustacksize;
-#if CONFIG_ARCH_INTERRUPTSTACK > 3
+#if CONFIG_ARCH_INTERRUPTSTACK > 7
   uint32_t istackbase;
   uint32_t istacksize;
 #endif
@@ -214,11 +214,11 @@ static void up_dumpstate(void)
 
   _alert("Current sp: %08x\n", sp);
 
-#if CONFIG_ARCH_INTERRUPTSTACK > 3
+#if CONFIG_ARCH_INTERRUPTSTACK > 7
   /* Get the limits on the interrupt stack memory */
 
   istackbase = (uint32_t)&g_intstackbase;
-  istacksize = (CONFIG_ARCH_INTERRUPTSTACK & ~3);
+  istacksize = (CONFIG_ARCH_INTERRUPTSTACK & ~7);
 
   /* Show interrupt stack info */
 
@@ -253,7 +253,7 @@ static void up_dumpstate(void)
     }
 #endif
 
-#if CONFIG_ARCH_INTERRUPTSTACK > 3
+#if CONFIG_ARCH_INTERRUPTSTACK > 7
   /* Does the current stack pointer lie within the interrupt stack? */
 
   if (sp > istackbase - istacksize && sp < istackbase)
@@ -262,13 +262,6 @@ static void up_dumpstate(void)
 
       _alert("Interrupt Stack\n", sp);
       up_stackdump(sp, istackbase);
-
-      /* Extract the user stack pointer which should lie
-       * at the base of the interrupt stack.
-       */
-
-      sp = g_intstackbase;
-      _alert("User sp: %08x\n", sp);
     }
   else if (CURRENT_REGS)
     {
@@ -277,6 +270,17 @@ static void up_dumpstate(void)
     }
 #endif
 
+  /* Extract the user stack pointer if we are in an interrupt handler.
+   * If we are not in an interrupt handler.  Then sp is the user stack
+   * pointer (and the above range check should have failed).
+   */
+
+  if (CURRENT_REGS)
+    {
+      sp = CURRENT_REGS[REG_R13];
+      _alert("User sp: %08x\n", sp);
+    }
+
   /* Dump the user stack if the stack pointer lies within the allocated user
    * stack memory.
    */
diff --git a/arch/arm/src/armv7-r/arm_vectors.S b/arch/arm/src/armv7-r/arm_vectors.S
index bea7c927bcebf..f3064ad219987 100644
--- a/arch/arm/src/armv7-r/arm_vectors.S
+++ b/arch/arm/src/armv7-r/arm_vectors.S
@@ -179,9 +179,9 @@ arm_vectorirq:
 	mov		fp, #0					/* Init frame pointer */
 	mov		r0, sp					/* Get r0=xcp */
 
-#if CONFIG_ARCH_INTERRUPTSTACK > 3
+#if CONFIG_ARCH_INTERRUPTSTACK > 7
 	ldr		sp, .Lirqstackbase		/* SP = interrupt stack base */
-	str		r0, [sp]				/* Save the user stack pointer */
+	str		r0, [sp, #-4]!			/* Save the xcp address at SP-4 then update SP */
 	mov		r4, sp					/* Save the SP in a preserved register */
 	bic		sp, sp, #7				/* Force 8-byte alignment */
 	bl		arm_decodeirq			/* Call the handler */
@@ -232,7 +232,7 @@ arm_vectorirq:
 
 .Lirqtmp:
 	.word	g_irqtmp
-#if CONFIG_ARCH_INTERRUPTSTACK > 3
+#if CONFIG_ARCH_INTERRUPTSTACK > 7
 .Lirqstackbase:
 	.word	g_intstackbase
 #endif
@@ -890,9 +890,9 @@ arm_vectorfiq:
 	mov		fp, #0					/* Init frame pointer */
 	mov		r0, sp					/* Get r0=xcp */
 
-#if CONFIG_ARCH_INTERRUPTSTACK > 3
+#if CONFIG_ARCH_INTERRUPTSTACK > 7
 	ldr		sp, .Lfiqstackbase		/* SP = interrupt stack base */
-	str		r0, [sp]				/* Save the user stack pointer */
+	str		r0, [sp, #-4]!			/* Save the xcp address at SP-4 then update SP */
 	mov		r4, sp					/* Save the SP in a preserved register */
 	bic		sp, sp, #7				/* Force 8-byte alignment */
 	bl		arm_decodefiq			/* Call the handler */
@@ -943,7 +943,7 @@ arm_vectorfiq:
 
 .Lfiqtmp:
 	.word	g_fiqtmp
-#if CONFIG_ARCH_INTERRUPTSTACK > 3
+#if CONFIG_ARCH_INTERRUPTSTACK > 7
 .Lfiqstackbase:
 	.word	g_intstackbase
 #endif
@@ -957,9 +957,9 @@ arm_vectorfiq:
  *  Name: g_intstackalloc/g_intstackbase
  ************************************************************************************/
 
-#if CONFIG_ARCH_INTERRUPTSTACK > 3
+#if CONFIG_ARCH_INTERRUPTSTACK > 7
 	.bss
-	.align	4
+	.balign	8
 
 	.globl	g_intstackalloc
 	.type	g_intstackalloc, object
@@ -967,11 +967,11 @@ arm_vectorfiq:
 	.type	g_intstackbase, object
 
 g_intstackalloc:
-	.skip	((CONFIG_ARCH_INTERRUPTSTACK & ~3) - 4)
+	.skip	((CONFIG_ARCH_INTERRUPTSTACK + 4) & ~7)
 g_intstackbase:
 	.skip	4
 	.size	g_intstackbase, 4
-	.size	g_intstackalloc, (CONFIG_ARCH_INTERRUPTSTACK & ~3)
+	.size	g_intstackalloc, (CONFIG_ARCH_INTERRUPTSTACK & ~7)
 
-#endif /* CONFIG_ARCH_INTERRUPTSTACK > 3 */
+#endif /* CONFIG_ARCH_INTERRUPTSTACK > 7 */
 	.end
diff --git a/arch/arm/src/armv8-m/arm_assert.c b/arch/arm/src/armv8-m/arm_assert.c
index ac0b9b7215113..bc6cb3a1d9356 100644
--- a/arch/arm/src/armv8-m/arm_assert.c
+++ b/arch/arm/src/armv8-m/arm_assert.c
@@ -244,7 +244,7 @@ static void up_dumpstate(void)
    * stack?
    */
 
-  if (sp <= istackbase && sp > istackbase - istacksize)
+  if (sp < istackbase && sp > istackbase - istacksize)
     {
       /* Yes.. dump the interrupt stack */
 
diff --git a/arch/arm/src/armv8-m/arm_exception.S b/arch/arm/src/armv8-m/arm_exception.S
index c7dc5af6e097e..9a3c434e0a92d 100644
--- a/arch/arm/src/armv8-m/arm_exception.S
+++ b/arch/arm/src/armv8-m/arm_exception.S
@@ -67,7 +67,7 @@
    * no privileged task has run.
    */
 
-#  if defined(CONFIG_BUILD_PROTECTED) && CONFIG_ARCH_INTERRUPTSTACK < 4
+#  if defined(CONFIG_BUILD_PROTECTED) && CONFIG_ARCH_INTERRUPTSTACK < 8
 #    error Interrupt stack must be used with high priority interrupts in kernel mode
 #  endif
 
@@ -355,7 +355,7 @@ exception_common:
 	.bss
 	.global	g_intstackalloc
 	.global	g_intstackbase
-	.align	8
+	.balign	8
 g_intstackalloc:
 	.skip	((CONFIG_ARCH_INTERRUPTSTACK + 4) & ~7)
 g_intstackbase:
diff --git a/arch/arm/src/armv8-m/arm_lazyexception.S b/arch/arm/src/armv8-m/arm_lazyexception.S
index ade51c1adbac1..fe89b3206331e 100644
--- a/arch/arm/src/armv8-m/arm_lazyexception.S
+++ b/arch/arm/src/armv8-m/arm_lazyexception.S
@@ -50,7 +50,7 @@
    * nested interrupt, the interrupt stack if no privileged task has run.
    */
 
-#  if defined(CONFIG_BUILD_PROTECTED) && CONFIG_ARCH_INTERRUPTSTACK < 4
+#  if defined(CONFIG_BUILD_PROTECTED) && CONFIG_ARCH_INTERRUPTSTACK < 8
 #    error Interrupt stack must be used with high priority interrupts in kernel mode
 #  endif
 
@@ -373,7 +373,7 @@ exception_common:
 	.bss
 	.global	g_intstackalloc
 	.global	g_intstackbase
-	.align	8
+	.balign	8
 g_intstackalloc:
 	.skip	((CONFIG_ARCH_INTERRUPTSTACK + 4) & ~7)
 g_intstackbase:
diff --git a/arch/arm/src/c5471/c5471_vectors.S b/arch/arm/src/c5471/c5471_vectors.S
index 95f4357a771a7..813e0534edc4a 100644
--- a/arch/arm/src/c5471/c5471_vectors.S
+++ b/arch/arm/src/c5471/c5471_vectors.S
@@ -162,7 +162,7 @@ arm_vectorirq:
 
 #if CONFIG_ARCH_INTERRUPTSTACK > 3
 	ldr	sp, .Lirqstackbase	/* SP = interrupt stack base */
-	str	r1, [sp]		/* Save the user stack pointer */
+	str	r1, [sp, #-4]!		/* Save the xcp address at SP-4 then update SP */
 	bl	arm_doirq		/* Call the handler */
 	ldr	sp, [sp]		/* Restore the user stack pointer */
 #else
@@ -471,13 +471,13 @@ arm_vectoraddrexcptn:
 
 #if CONFIG_ARCH_INTERRUPTSTACK > 3
 	.bss
-	.align	4
+	.balign	4
 	.global	g_intstackalloc
 	.global	g_intstackbase
 	.type	g_intstackalloc, object
 	.type	g_intstackbase, object
 g_intstackalloc:
-	.skip	((CONFIG_ARCH_INTERRUPTSTACK & ~3) - 4)
+	.skip	(CONFIG_ARCH_INTERRUPTSTACK & ~3)
 g_intstackbase:
 	.skip	4
 	.size	g_intstackbase, 4
diff --git a/arch/arm/src/cxd56xx/cxd56_irq.c b/arch/arm/src/cxd56xx/cxd56_irq.c
index f113dcef06f9e..d59872e61576e 100644
--- a/arch/arm/src/cxd56xx/cxd56_irq.c
+++ b/arch/arm/src/cxd56xx/cxd56_irq.c
@@ -112,17 +112,17 @@ static uint64_t g_intstack_alloc[INTSTACK_ALLOC >> 3];
 
 const uint32_t g_cpu_intstack_top[CONFIG_SMP_NCPUS] =
 {
-  (uint32_t)g_intstack_alloc + INTSTACK_SIZE - 8,
+  (uint32_t)g_intstack_alloc + INTSTACK_SIZE,
 #if CONFIG_SMP_NCPUS > 1
-  (uint32_t)g_intstack_alloc + (2 * INTSTACK_SIZE) - 8,
+  (uint32_t)g_intstack_alloc + (2 * INTSTACK_SIZE),
 #if CONFIG_SMP_NCPUS > 2
-  (uint32_t)g_intstack_alloc + (3 * INTSTACK_SIZE) - 8,
+  (uint32_t)g_intstack_alloc + (3 * INTSTACK_SIZE),
 #if CONFIG_SMP_NCPUS > 3
-  (uint32_t)g_intstack_alloc + (4 * INTSTACK_SIZE) - 8,
+  (uint32_t)g_intstack_alloc + (4 * INTSTACK_SIZE),
 #if CONFIG_SMP_NCPUS > 4
-  (uint32_t)g_intstack_alloc + (5 * INTSTACK_SIZE) - 8,
+  (uint32_t)g_intstack_alloc + (5 * INTSTACK_SIZE),
 #if CONFIG_SMP_NCPUS > 5
-  (uint32_t)g_intstack_alloc + (6 * INTSTACK_SIZE) - 8,
+  (uint32_t)g_intstack_alloc + (6 * INTSTACK_SIZE),
 #endif /* CONFIG_SMP_NCPUS > 5 */
 #endif /* CONFIG_SMP_NCPUS > 4 */
 #endif /* CONFIG_SMP_NCPUS > 3 */
@@ -667,6 +667,6 @@ uintptr_t arm_intstack_base(void)
 #if defined(CONFIG_SMP) && CONFIG_ARCH_INTERRUPTSTACK > 7
 uintptr_t arm_intstack_alloc(void)
 {
-  return g_cpu_intstack_top[up_cpu_index()] - (INTSTACK_SIZE - 8);
+  return g_cpu_intstack_top[up_cpu_index()] - INTSTACK_SIZE;
 }
 #endif
diff --git a/arch/arm/src/imx6/imx_irq.c b/arch/arm/src/imx6/imx_irq.c
index 1f7738f9a042d..7d3b76033e576 100644
--- a/arch/arm/src/imx6/imx_irq.c
+++ b/arch/arm/src/imx6/imx_irq.c
@@ -88,15 +88,15 @@ uint64_t g_fiqstack_alloc[INTSTACK_ALLOC >> 3];
 
 uintptr_t g_irqstack_top[CONFIG_SMP_NCPUS] =
 {
-  (uintptr_t)g_irqstack_alloc + INTSTACK_SIZE - 8,
+  (uintptr_t)g_irqstack_alloc + INTSTACK_SIZE,
 #if CONFIG_SMP_NCPUS > 1
-  (uintptr_t)g_irqstack_alloc + (2 * INTSTACK_SIZE) - 8,
+  (uintptr_t)g_irqstack_alloc + (2 * INTSTACK_SIZE),
 #endif
 #if CONFIG_SMP_NCPUS > 2
-  (uintptr_t)g_irqstack_alloc + (3 * INTSTACK_SIZE) - 8,
+  (uintptr_t)g_irqstack_alloc + (3 * INTSTACK_SIZE),
 #endif
 #if CONFIG_SMP_NCPUS > 3
-  (uintptr_t)g_irqstack_alloc + (4 * INTSTACK_SIZE) - 8
+  (uintptr_t)g_irqstack_alloc + (4 * INTSTACK_SIZE)
 #endif
 };
 
@@ -215,6 +215,6 @@ uintptr_t arm_intstack_base(void)
 #if defined(CONFIG_SMP) && CONFIG_ARCH_INTERRUPTSTACK > 7
 uintptr_t arm_intstack_alloc(void)
 {
-  return g_irqstack_top[up_cpu_index()] - (INTSTACK_SIZE - 8);
+  return g_irqstack_top[up_cpu_index()] - INTSTACK_SIZE;
 }
 #endif
diff --git a/arch/arm/src/lc823450/lc823450_irq.c b/arch/arm/src/lc823450/lc823450_irq.c
index 4e0beaa0715b1..4583ae95f5e80 100644
--- a/arch/arm/src/lc823450/lc823450_irq.c
+++ b/arch/arm/src/lc823450/lc823450_irq.c
@@ -95,9 +95,9 @@ uint64_t g_intstack_alloc[INTSTACK_ALLOC >> 3];
 
 const uint32_t g_cpu_intstack_top[CONFIG_SMP_NCPUS] =
 {
-  (uint32_t)g_intstack_alloc + INTSTACK_SIZE - 8,
+  (uint32_t)g_intstack_alloc + INTSTACK_SIZE,
 #if CONFIG_SMP_NCPUS > 1
-  (uint32_t)g_intstack_alloc + (2 * INTSTACK_SIZE) - 8,
+  (uint32_t)g_intstack_alloc + (2 * INTSTACK_SIZE),
 #endif /* CONFIG_SMP_NCPUS > 1 */
 };
 #endif
@@ -888,6 +888,6 @@ uintptr_t arm_intstack_base(void)
 #if defined(CONFIG_SMP) && CONFIG_ARCH_INTERRUPTSTACK > 7
 uintptr_t arm_intstack_alloc(void)
 {
-  return g_cpu_intstack_top[up_cpu_index()] - (INTSTACK_SIZE - 8);
+  return g_cpu_intstack_top[up_cpu_index()] - INTSTACK_SIZE;
 }
 #endif
diff --git a/boards/arm/imx6/sabre-6quad/configs/nsh/defconfig b/boards/arm/imx6/sabre-6quad/configs/nsh/defconfig
index c42ae64c05b59..30aa9a25a8b34 100644
--- a/boards/arm/imx6/sabre-6quad/configs/nsh/defconfig
+++ b/boards/arm/imx6/sabre-6quad/configs/nsh/defconfig
@@ -20,11 +20,12 @@ CONFIG_ARCH_STACKDUMP=y
 CONFIG_BOARD_LOOPSPERMSEC=99369
 CONFIG_BOOT_RUNFROMSDRAM=y
 CONFIG_BUILTIN=y
+CONFIG_DEBUG_FULLOPT=y
+CONFIG_DEBUG_SYMBOLS=y
 CONFIG_DEV_ZERO=y
 CONFIG_FS_PROCFS=y
 CONFIG_HAVE_CXX=y
 CONFIG_HAVE_CXXINITIALIZE=y
-CONFIG_HOST_WINDOWS=y
 CONFIG_IMX6_UART1=y
 CONFIG_IMX_DDR_SIZE=1073741824
 CONFIG_INTELHEX_BINARY=y
@@ -42,10 +43,12 @@ CONFIG_RAM_SIZE=1073741824
 CONFIG_RAM_START=0x10000000
 CONFIG_RAM_VSTART=0x10000000
 CONFIG_RAW_BINARY=y
+CONFIG_READLINE_CMD_HISTORY=y
 CONFIG_RR_INTERVAL=200
 CONFIG_SCHED_HPWORK=y
 CONFIG_SCHED_HPWORKPRIORITY=192
 CONFIG_SCHED_WAITPID=y
+CONFIG_STACK_COLORATION=y
 CONFIG_START_MONTH=3
 CONFIG_START_YEAR=2016
 CONFIG_SYMTAB_ORDEREDBYNAME=y
diff --git a/boards/arm/stm32/stm32f4discovery/configs/wifi/defconfig b/boards/arm/stm32/stm32f4discovery/configs/wifi/defconfig
index 54912f275dc5f..9f48660c75bb2 100644
--- a/boards/arm/stm32/stm32f4discovery/configs/wifi/defconfig
+++ b/boards/arm/stm32/stm32f4discovery/configs/wifi/defconfig
@@ -5,7 +5,6 @@
 # You can then do "make savedefconfig" to generate a new defconfig file that includes your
 # modifications.
 #
-# CONFIG_ARCH_FPU is not set
 # CONFIG_MMCSD_HAVE_CARDDETECT is not set
 # CONFIG_MMCSD_HAVE_WRITEPROTECT is not set
 # CONFIG_NSH_ARGCAT is not set
@@ -19,7 +18,9 @@ CONFIG_ARCH_BUTTONS=y
 CONFIG_ARCH_CHIP="stm32"
 CONFIG_ARCH_CHIP_STM32=y
 CONFIG_ARCH_CHIP_STM32F407VG=y
+CONFIG_ARCH_INTERRUPTSTACK=2048
 CONFIG_ARCH_STACKDUMP=y
+CONFIG_ARMV7M_LAZYFPU=y
 CONFIG_BOARDCTL_RESET=y
 CONFIG_BOARD_LOOPSPERMSEC=16717
 CONFIG_BUILTIN=y
@@ -101,6 +102,7 @@ CONFIG_SYSTEM_NSH_SYMTAB_ARRAYNAME="g_symtab"
 CONFIG_SYSTEM_NSH_SYMTAB_COUNTNAME="g_nsymbols"
 CONFIG_SYSTEM_NTPC=y
 CONFIG_TESTING_OSTEST=y
+CONFIG_TESTING_OSTEST_FPUSIZE=132
 CONFIG_USART2_RXBUFSIZE=128
 CONFIG_USART2_SERIAL_CONSOLE=y
 CONFIG_USART2_TXBUFSIZE=128