s390/irqflags: optimize irq restore

The ssm instruction takes longer that stnsm/stosm as it is often
used to modify DAT and PER. We know that irqsave/irqrestore only
deals with external and I/O interrupts and we know that irqrestore
can transition only from disabled->disabled or disabled->enabled,
so we can use the faster stosm.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>

arch/s390/include/asm/irqflags.h

@@ -8,6 +8,8 @@
 
 #include <linux/types.h>
 
+#define ARCH_IRQ_ENABLED	(3UL << (BITS_PER_LONG - 8))
+
 /* store then OR system mask. */
 #define __arch_local_irq_stosm(__or)					\
 ({									\
@@ -54,14 +56,17 @@ static inline notrace void arch_local_irq_enable(void)
 	__arch_local_irq_stosm(0x03);
 }
 
+/* This only restores external and I/O interrupt state */
 static inline notrace void arch_local_irq_restore(unsigned long flags)
 {
-	__arch_local_irq_ssm(flags);
+	/* only disabled->disabled and disabled->enabled is valid */
+	if (flags & ARCH_IRQ_ENABLED)
+		arch_local_irq_enable();
 }
 
 static inline notrace bool arch_irqs_disabled_flags(unsigned long flags)
 {
-	return !(flags & (3UL << (BITS_PER_LONG - 8)));
+	return !(flags & ARCH_IRQ_ENABLED);
 }
 
 static inline notrace bool arch_irqs_disabled(void)

arch/s390/mm/init.c

@@ -98,7 +98,7 @@ void __init paging_init(void)
 	__ctl_load(S390_lowcore.kernel_asce, 1, 1);
 	__ctl_load(S390_lowcore.kernel_asce, 7, 7);
 	__ctl_load(S390_lowcore.kernel_asce, 13, 13);
-	arch_local_irq_restore(4UL << (BITS_PER_LONG - 8));
+	__arch_local_irq_stosm(0x04);
 
 	sparse_memory_present_with_active_regions(MAX_NUMNODES);
 	sparse_init();