diff --git a/arch/arm/include/asm/assembler.h b/arch/arm/include/asm/assembler.h
index 15f8a092b700..44912cd5da13 100644
--- a/arch/arm/include/asm/assembler.h
+++ b/arch/arm/include/asm/assembler.h
@@ -74,23 +74,56 @@
  * Enable and disable interrupts
  */
 #if __LINUX_ARM_ARCH__ >= 6
-	.macro	disable_irq
+	.macro	disable_irq_notrace
 	cpsid	i
 	.endm
 
-	.macro	enable_irq
+	.macro	enable_irq_notrace
 	cpsie	i
 	.endm
 #else
-	.macro	disable_irq
+	.macro	disable_irq_notrace
 	msr	cpsr_c, #PSR_I_BIT | SVC_MODE
 	.endm
 
-	.macro	enable_irq
+	.macro	enable_irq_notrace
 	msr	cpsr_c, #SVC_MODE
 	.endm
 #endif
 
+	.macro asm_trace_hardirqs_off
+#if defined(CONFIG_TRACE_IRQFLAGS)
+	stmdb   sp!, {r0-r3, ip, lr}
+	bl	trace_hardirqs_off
+	ldmia	sp!, {r0-r3, ip, lr}
+#endif
+	.endm
+
+	.macro asm_trace_hardirqs_on_cond, cond
+#if defined(CONFIG_TRACE_IRQFLAGS)
+	/*
+	 * actually the registers should be pushed and pop'd conditionally, but
+	 * after bl the flags are certainly clobbered
+	 */
+	stmdb   sp!, {r0-r3, ip, lr}
+	bl\cond	trace_hardirqs_on
+	ldmia	sp!, {r0-r3, ip, lr}
+#endif
+	.endm
+
+	.macro asm_trace_hardirqs_on
+	asm_trace_hardirqs_on_cond al
+	.endm
+
+	.macro disable_irq
+	disable_irq_notrace
+	asm_trace_hardirqs_off
+	.endm
+
+	.macro enable_irq
+	asm_trace_hardirqs_on
+	enable_irq_notrace
+	.endm
 /*
  * Save the current IRQ state and disable IRQs.  Note that this macro
  * assumes FIQs are enabled, and that the processor is in SVC mode.
@@ -104,10 +137,16 @@
  * Restore interrupt state previously stored in a register.  We don't
  * guarantee that this will preserve the flags.
  */
-	.macro	restore_irqs, oldcpsr
+	.macro	restore_irqs_notrace, oldcpsr
 	msr	cpsr_c, \oldcpsr
 	.endm
 
+	.macro restore_irqs, oldcpsr
+	tst	\oldcpsr, #PSR_I_BIT
+	asm_trace_hardirqs_on_cond eq
+	restore_irqs_notrace \oldcpsr
+	.endm
+
 #define USER(x...)				\
 9999:	x;					\
 	.section __ex_table,"a";		\
diff --git a/arch/arm/kernel/entry-armv.S b/arch/arm/kernel/entry-armv.S
index fc8af43c5000..792abd0dfae1 100644
--- a/arch/arm/kernel/entry-armv.S
+++ b/arch/arm/kernel/entry-armv.S
@@ -151,6 +151,8 @@ ENDPROC(__und_invalid)
 	@  r4 - orig_r0 (see pt_regs definition in ptrace.h)
 	@
 	stmia	r5, {r0 - r4}
+
+	asm_trace_hardirqs_off
 	.endm
 
 	.align	5
@@ -206,9 +208,6 @@ ENDPROC(__dabt_svc)
 __irq_svc:
 	svc_entry
 
-#ifdef CONFIG_TRACE_IRQFLAGS
-	bl	trace_hardirqs_off
-#endif
 #ifdef CONFIG_PREEMPT
 	get_thread_info tsk
 	ldr	r8, [tsk, #TI_PREEMPT]		@ get preempt count
@@ -383,6 +382,8 @@ ENDPROC(__pabt_svc)
 	@ Clear FP to mark the first stack frame
 	@
 	zero_fp
+
+	asm_trace_hardirqs_off
 	.endm
 
 	.macro	kuser_cmpxchg_check
@@ -437,9 +438,6 @@ __irq_usr:
 	usr_entry
 	kuser_cmpxchg_check
 
-#ifdef CONFIG_TRACE_IRQFLAGS
-	bl	trace_hardirqs_off
-#endif
 	get_thread_info tsk
 #ifdef CONFIG_PREEMPT
 	ldr	r8, [tsk, #TI_PREEMPT]		@ get preempt count
diff --git a/arch/arm/lib/bitops.h b/arch/arm/lib/bitops.h
index c7f2627385e7..d42252918bfb 100644
--- a/arch/arm/lib/bitops.h
+++ b/arch/arm/lib/bitops.h
@@ -60,8 +60,8 @@
 	tst	r2, r0, lsl r3
 	\instr	r2, r2, r0, lsl r3
 	\store	r2, [r1]
-	restore_irqs ip
 	moveq	r0, #0
+	restore_irqs ip
 	mov	pc, lr
 	.endm
 #endif