Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

Pull second set of s390 patches from Martin Schwidefsky:
 "The second part of Heikos uaccess rework, the page table walker for
  uaccess is now a thing of the past (yay!)

  The code change to fix the theoretical TLB flush problem allows us to
  add a TLB flush optimization for zEC12, this machine has new
  instructions that allow to do CPU local TLB flushes for single pages
  and for all pages of a specific address space.

  Plus the usual bug fixing and some more cleanup"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/uaccess: rework uaccess code - fix locking issues
  s390/mm,tlb: optimize TLB flushing for zEC12
  s390/mm,tlb: safeguard against speculative TLB creation
  s390/irq: Use defines for external interruption codes
  s390/irq: Add defines for external interruption codes
  s390/sclp: add timeout for queued requests
  kvm/s390: also set guest pages back to stable on kexec/kdump
  lcs: Add missing destroy_timer_on_stack()
  s390/tape: Add missing destroy_timer_on_stack()
  s390/tape: Use del_timer_sync()
  s390/3270: fix crash with multiple reset device requests
  s390/bitops,atomic: add missing memory barriers
  s390/zcrypt: add length check for aligned data to avoid overflow in msg-type 6

47 files changed, 1032 insertions, 979 deletions

diff --git a/arch/s390/include/asm/atomic.h b/arch/s390/include/asm/atomic.h
index fa9aaf7144b7..1d4706114a45 100644
--- a/arch/s390/include/asm/atomic.h
+++ b/arch/s390/include/asm/atomic.h
@@ -15,23 +15,29 @@
 
 #include <linux/compiler.h>
 #include <linux/types.h>
+#include <asm/barrier.h>
 #include <asm/cmpxchg.h>
 
 #define ATOMIC_INIT(i)  { (i) }
 
+#define __ATOMIC_NO_BARRIER	"\n"
+
 #ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
 
 #define __ATOMIC_OR	"lao"
 #define __ATOMIC_AND	"lan"
 #define __ATOMIC_ADD	"laa"
+#define __ATOMIC_BARRIER "bcr	14,0\n"
 
-#define __ATOMIC_LOOP(ptr, op_val, op_string)				\
+#define __ATOMIC_LOOP(ptr, op_val, op_string, __barrier)		\
 ({									\
 	int old_val;							\
 									\
 	typecheck(atomic_t *, ptr);					\
 	asm volatile(							\
+		__barrier						\
 		op_string "	%0,%2,%1\n"				\
+		__barrier						\
 		: "=d" (old_val), "+Q" ((ptr)->counter)			\
 		: "d" (op_val)						\
 		: "cc", "memory");					\
@@ -43,8 +49,9 @@
 #define __ATOMIC_OR	"or"
 #define __ATOMIC_AND	"nr"
 #define __ATOMIC_ADD	"ar"
+#define __ATOMIC_BARRIER "\n"
 
-#define __ATOMIC_LOOP(ptr, op_val, op_string)				\
+#define __ATOMIC_LOOP(ptr, op_val, op_string, __barrier)		\
 ({									\
 	int old_val, new_val;						\
 									\
@@ -82,7 +89,7 @@ static inline void atomic_set(atomic_t *v, int i)
 
 static inline int atomic_add_return(int i, atomic_t *v)
 {
-	return __ATOMIC_LOOP(v, i, __ATOMIC_ADD) + i;
+	return __ATOMIC_LOOP(v, i, __ATOMIC_ADD, __ATOMIC_BARRIER) + i;
 }
 
 static inline void atomic_add(int i, atomic_t *v)
@@ -94,12 +101,10 @@ static inline void atomic_add(int i, atomic_t *v)
 			: "+Q" (v->counter)
 			: "i" (i)
 			: "cc", "memory");
-	} else {
-		atomic_add_return(i, v);
+		return;
 	}
-#else
-	atomic_add_return(i, v);
 #endif
+	__ATOMIC_LOOP(v, i, __ATOMIC_ADD, __ATOMIC_NO_BARRIER);
 }
 
 #define atomic_add_negative(_i, _v)	(atomic_add_return(_i, _v) < 0)
@@ -115,12 +120,12 @@ static inline void atomic_add(int i, atomic_t *v)
 
 static inline void atomic_clear_mask(unsigned int mask, atomic_t *v)
 {
-	__ATOMIC_LOOP(v, ~mask, __ATOMIC_AND);
+	__ATOMIC_LOOP(v, ~mask, __ATOMIC_AND, __ATOMIC_NO_BARRIER);
 }
 
 static inline void atomic_set_mask(unsigned int mask, atomic_t *v)
 {
-	__ATOMIC_LOOP(v, mask, __ATOMIC_OR);
+	__ATOMIC_LOOP(v, mask, __ATOMIC_OR, __ATOMIC_NO_BARRIER);
 }
 
 #define atomic_xchg(v, new) (xchg(&((v)->counter), new))
@@ -157,19 +162,24 @@ static inline int __atomic_add_unless(atomic_t *v, int a, int u)
 
 #ifdef CONFIG_64BIT
 
+#define __ATOMIC64_NO_BARRIER	"\n"
+
 #ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
 
 #define __ATOMIC64_OR	"laog"
 #define __ATOMIC64_AND	"lang"
 #define __ATOMIC64_ADD	"laag"
+#define __ATOMIC64_BARRIER "bcr	14,0\n"
 
-#define __ATOMIC64_LOOP(ptr, op_val, op_string)				\
+#define __ATOMIC64_LOOP(ptr, op_val, op_string, __barrier)		\
 ({									\
 	long long old_val;						\
 									\
 	typecheck(atomic64_t *, ptr);					\
 	asm volatile(							\
+		__barrier						\
 		op_string "	%0,%2,%1\n"				\
+		__barrier						\
 		: "=d" (old_val), "+Q" ((ptr)->counter)			\
 		: "d" (op_val)						\
 		: "cc", "memory");					\
@@ -181,8 +191,9 @@ static inline int __atomic_add_unless(atomic_t *v, int a, int u)
 #define __ATOMIC64_OR	"ogr"
 #define __ATOMIC64_AND	"ngr"
 #define __ATOMIC64_ADD	"agr"
+#define __ATOMIC64_BARRIER "\n"
 
-#define __ATOMIC64_LOOP(ptr, op_val, op_string)				\
+#define __ATOMIC64_LOOP(ptr, op_val, op_string, __barrier)		\
 ({									\
 	long long old_val, new_val;					\
 									\
@@ -220,17 +231,32 @@ static inline void atomic64_set(atomic64_t *v, long long i)
 
 static inline long long atomic64_add_return(long long i, atomic64_t *v)
 {
-	return __ATOMIC64_LOOP(v, i, __ATOMIC64_ADD) + i;
+	return __ATOMIC64_LOOP(v, i, __ATOMIC64_ADD, __ATOMIC64_BARRIER) + i;
+}
+
+static inline void atomic64_add(long long i, atomic64_t *v)
+{
+#ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
+	if (__builtin_constant_p(i) && (i > -129) && (i < 128)) {
+		asm volatile(
+			"agsi	%0,%1\n"
+			: "+Q" (v->counter)
+			: "i" (i)
+			: "cc", "memory");
+		return;
+	}
+#endif
+	__ATOMIC64_LOOP(v, i, __ATOMIC64_ADD, __ATOMIC64_NO_BARRIER);
 }
 
 static inline void atomic64_clear_mask(unsigned long mask, atomic64_t *v)
 {
-	__ATOMIC64_LOOP(v, ~mask, __ATOMIC64_AND);
+	__ATOMIC64_LOOP(v, ~mask, __ATOMIC64_AND, __ATOMIC64_NO_BARRIER);
 }
 
 static inline void atomic64_set_mask(unsigned long mask, atomic64_t *v)
 {
-	__ATOMIC64_LOOP(v, mask, __ATOMIC64_OR);
+	__ATOMIC64_LOOP(v, mask, __ATOMIC64_OR, __ATOMIC64_NO_BARRIER);
 }
 
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
@@ -334,25 +360,13 @@ static inline void atomic64_clear_mask(unsigned long long mask, atomic64_t *v)
 	} while (atomic64_cmpxchg(v, old, new) != old);
 }
 
-#endif /* CONFIG_64BIT */
-
 static inline void atomic64_add(long long i, atomic64_t *v)
 {
-#ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
-	if (__builtin_constant_p(i) && (i > -129) && (i < 128)) {
-		asm volatile(
-			"agsi	%0,%1\n"
-			: "+Q" (v->counter)
-			: "i" (i)
-			: "cc", "memory");
-	} else {
-		atomic64_add_return(i, v);
-	}
-#else
 	atomic64_add_return(i, v);
-#endif
 }
 
+#endif /* CONFIG_64BIT */
+
 static inline int atomic64_add_unless(atomic64_t *v, long long i, long long u)
 {
 	long long c, old;
diff --git a/arch/s390/include/asm/bitops.h b/arch/s390/include/asm/bitops.h
index ec5ef891db6b..520542477678 100644
--- a/arch/s390/include/asm/bitops.h
+++ b/arch/s390/include/asm/bitops.h
@@ -47,14 +47,18 @@
 
 #include <linux/typecheck.h>
 #include <linux/compiler.h>
+#include <asm/barrier.h>
+
+#define __BITOPS_NO_BARRIER	"\n"
 
 #ifndef CONFIG_64BIT
 
 #define __BITOPS_OR		"or"
 #define __BITOPS_AND		"nr"
 #define __BITOPS_XOR		"xr"
+#define __BITOPS_BARRIER	"\n"
 
-#define __BITOPS_LOOP(__addr, __val, __op_string)		\
+#define __BITOPS_LOOP(__addr, __val, __op_string, __barrier)	\
 ({								\
 	unsigned long __old, __new;				\
 								\
@@ -67,7 +71,7 @@
 		"	jl	0b"				\
 		: "=&d" (__old), "=&d" (__new), "+Q" (*(__addr))\
 		: "d" (__val)					\
-		: "cc");					\
+		: "cc", "memory");				\
 	__old;							\
 })
 
@@ -78,17 +82,20 @@
 #define __BITOPS_OR		"laog"
 #define __BITOPS_AND		"lang"
 #define __BITOPS_XOR		"laxg"
+#define __BITOPS_BARRIER	"bcr	14,0\n"
 
-#define __BITOPS_LOOP(__addr, __val, __op_string)		\
+#define __BITOPS_LOOP(__addr, __val, __op_string, __barrier)	\
 ({								\
 	unsigned long __old;					\
 								\
 	typecheck(unsigned long *, (__addr));			\
 	asm volatile(						\
+		__barrier					\
 		__op_string "	%0,%2,%1\n"			\
+		__barrier					\
 		: "=d" (__old),	"+Q" (*(__addr))		\
 		: "d" (__val)					\
-		: "cc");					\
+		: "cc", "memory");				\
 	__old;							\
 })
 
@@ -97,8 +104,9 @@
 #define __BITOPS_OR		"ogr"
 #define __BITOPS_AND		"ngr"
 #define __BITOPS_XOR		"xgr"
+#define __BITOPS_BARRIER	"\n"
 
-#define __BITOPS_LOOP(__addr, __val, __op_string)		\
+#define __BITOPS_LOOP(__addr, __val, __op_string, __barrier)	\
 ({								\
 	unsigned long __old, __new;				\
 								\
@@ -111,7 +119,7 @@
 		"	jl	0b"				\
 		: "=&d" (__old), "=&d" (__new), "+Q" (*(__addr))\
 		: "d" (__val)					\
-		: "cc");					\
+		: "cc", "memory");				\
 	__old;							\
 })
 
@@ -149,12 +157,12 @@ static inline void set_bit(unsigned long nr, volatile unsigned long *ptr)
 			"oi	%0,%b1\n"
 			: "+Q" (*caddr)
 			: "i" (1 << (nr & 7))
-			: "cc");
+			: "cc", "memory");
 		return;
 	}
 #endif
 	mask = 1UL << (nr & (BITS_PER_LONG - 1));
-	__BITOPS_LOOP(addr, mask, __BITOPS_OR);
+	__BITOPS_LOOP(addr, mask, __BITOPS_OR, __BITOPS_NO_BARRIER);
 }
 
 static inline void clear_bit(unsigned long nr, volatile unsigned long *ptr)
@@ -170,12 +178,12 @@ static inline void clear_bit(unsigned long nr, volatile unsigned long *ptr)
 			"ni	%0,%b1\n"
 			: "+Q" (*caddr)
 			: "i" (~(1 << (nr & 7)))
-			: "cc");
+			: "cc", "memory");
 		return;
 	}
 #endif
 	mask = ~(1UL << (nr & (BITS_PER_LONG - 1)));
-	__BITOPS_LOOP(addr, mask, __BITOPS_AND);
+	__BITOPS_LOOP(addr, mask, __BITOPS_AND, __BITOPS_NO_BARRIER);
 }
 
 static inline void change_bit(unsigned long nr, volatile unsigned long *ptr)
@@ -191,12 +199,12 @@ static inline void change_bit(unsigned long nr, volatile unsigned long *ptr)
 			"xi	%0,%b1\n"
 			: "+Q" (*caddr)
 			: "i" (1 << (nr & 7))
-			: "cc");
+			: "cc", "memory");
 		return;
 	}
 #endif
 	mask = 1UL << (nr & (BITS_PER_LONG - 1));
-	__BITOPS_LOOP(addr, mask, __BITOPS_XOR);
+	__BITOPS_LOOP(addr, mask, __BITOPS_XOR, __BITOPS_NO_BARRIER);
 }
 
 static inline int
@@ -206,8 +214,7 @@ test_and_set_bit(unsigned long nr, volatile unsigned long *ptr)
 	unsigned long old, mask;
 
 	mask = 1UL << (nr & (BITS_PER_LONG - 1));
-	old = __BITOPS_LOOP(addr, mask, __BITOPS_OR);
-	barrier();
+	old = __BITOPS_LOOP(addr, mask, __BITOPS_OR, __BITOPS_BARRIER);
 	return (old & mask) != 0;
 }
 
@@ -218,8 +225,7 @@ test_and_clear_bit(unsigned long nr, volatile unsigned long *ptr)
 	unsigned long old, mask;
 
 	mask = ~(1UL << (nr & (BITS_PER_LONG - 1)));
-	old = __BITOPS_LOOP(addr, mask, __BITOPS_AND);
-	barrier();
+	old = __BITOPS_LOOP(addr, mask, __BITOPS_AND, __BITOPS_BARRIER);
 	return (old & ~mask) != 0;
 }
 
@@ -230,8 +236,7 @@ test_and_change_bit(unsigned long nr, volatile unsigned long *ptr)
 	unsigned long old, mask;
 
 	mask = 1UL << (nr & (BITS_PER_LONG - 1));
-	old = __BITOPS_LOOP(addr, mask, __BITOPS_XOR);
-	barrier();
+	old = __BITOPS_LOOP(addr, mask, __BITOPS_XOR, __BITOPS_BARRIER);
 	return (old & mask) != 0;
 }
 
diff --git a/arch/s390/include/asm/futex.h b/arch/s390/include/asm/futex.h
index fda46bd38c99..69cf5b5eddc9 100644
--- a/arch/s390/include/asm/futex.h
+++ b/arch/s390/include/asm/futex.h
@@ -1,12 +1,25 @@
 #ifndef _ASM_S390_FUTEX_H
 #define _ASM_S390_FUTEX_H
 
-#include <linux/futex.h>
 #include <linux/uaccess.h>
+#include <linux/futex.h>
+#include <asm/mmu_context.h>
 #include <asm/errno.h>
 
-int futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr, u32 oldval, u32 newval);
-int __futex_atomic_op_inuser(int op, u32 __user *uaddr, int oparg, int *old);
+#define __futex_atomic_op(insn, ret, oldval, newval, uaddr, oparg)	\
+	asm volatile(							\
+		"   sacf  256\n"					\
+		"0: l     %1,0(%6)\n"					\
+		"1:"insn						\
+		"2: cs    %1,%2,0(%6)\n"				\
+		"3: jl    1b\n"						\
+		"   lhi   %0,0\n"					\
+		"4: sacf  768\n"					\
+		EX_TABLE(0b,4b) EX_TABLE(2b,4b) EX_TABLE(3b,4b)		\
+		: "=d" (ret), "=&d" (oldval), "=&d" (newval),		\
+		  "=m" (*uaddr)						\
+		: "0" (-EFAULT), "d" (oparg), "a" (uaddr),		\
+		  "m" (*uaddr) : "cc");
 
 static inline int futex_atomic_op_inuser(int encoded_op, u32 __user *uaddr)
 {
@@ -14,13 +27,37 @@ static inline int futex_atomic_op_inuser(int encoded_op, u32 __user *uaddr)
 	int cmp = (encoded_op >> 24) & 15;
 	int oparg = (encoded_op << 8) >> 20;
 	int cmparg = (encoded_op << 20) >> 20;
-	int oldval, ret;
+	int oldval = 0, newval, ret;
 
+	update_primary_asce(current);
 	if (encoded_op & (FUTEX_OP_OPARG_SHIFT << 28))
 		oparg = 1 << oparg;
 
 	pagefault_disable();
-	ret = __futex_atomic_op_inuser(op, uaddr, oparg, &oldval);
+	switch (op) {
+	case FUTEX_OP_SET:
+		__futex_atomic_op("lr %2,%5\n",
+				  ret, oldval, newval, uaddr, oparg);
+		break;
+	case FUTEX_OP_ADD:
+		__futex_atomic_op("lr %2,%1\nar %2,%5\n",
+				  ret, oldval, newval, uaddr, oparg);
+		break;
+	case FUTEX_OP_OR:
+		__futex_atomic_op("lr %2,%1\nor %2,%5\n",
+				  ret, oldval, newval, uaddr, oparg);
+		break;
+	case FUTEX_OP_ANDN:
+		__futex_atomic_op("lr %2,%1\nnr %2,%5\n",
+				  ret, oldval, newval, uaddr, oparg);
+		break;
+	case FUTEX_OP_XOR:
+		__futex_atomic_op("lr %2,%1\nxr %2,%5\n",
+				  ret, oldval, newval, uaddr, oparg);
+		break;
+	default:
+		ret = -ENOSYS;
+	}
 	pagefault_enable();
 
 	if (!ret) {
@@ -37,4 +74,23 @@ static inline int futex_atomic_op_inuser(int encoded_op, u32 __user *uaddr)
 	return ret;
 }
 
+static inline int futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr,
+						u32 oldval, u32 newval)
+{
+	int ret;
+
+	update_primary_asce(current);
+	asm volatile(
+		"   sacf 256\n"
+		"0: cs   %1,%4,0(%5)\n"
+		"1: la   %0,0\n"
+		"2: sacf 768\n"
+		EX_TABLE(0b,2b) EX_TABLE(1b,2b)
+		: "=d" (ret), "+d" (oldval), "=m" (*uaddr)
+		: "0" (-EFAULT), "d" (newval), "a" (uaddr), "m" (*uaddr)
+		: "cc", "memory");
+	*uval = oldval;
+	return ret;
+}
+
 #endif /* _ASM_S390_FUTEX_H */
diff --git a/arch/s390/include/asm/irq.h b/arch/s390/include/asm/irq.h
index 35f0faab5361..c4dd400a2791 100644
--- a/arch/s390/include/asm/irq.h
+++ b/arch/s390/include/asm/irq.h
@@ -16,6 +16,20 @@
 /* This number is used when no interrupt has been assigned */
 #define NO_IRQ		0
 
+/* External interruption codes */
+#define EXT_IRQ_INTERRUPT_KEY	0x0040
+#define EXT_IRQ_CLK_COMP	0x1004
+#define EXT_IRQ_CPU_TIMER	0x1005
+#define EXT_IRQ_WARNING_TRACK	0x1007
+#define EXT_IRQ_MALFUNC_ALERT	0x1200
+#define EXT_IRQ_EMERGENCY_SIG	0x1201
+#define EXT_IRQ_EXTERNAL_CALL	0x1202
+#define EXT_IRQ_TIMING_ALERT	0x1406
+#define EXT_IRQ_MEASURE_ALERT	0x1407
+#define EXT_IRQ_SERVICE_SIG	0x2401
+#define EXT_IRQ_CP_SERVICE	0x2603
+#define EXT_IRQ_IUCV		0x4000
+
 #ifndef __ASSEMBLY__
 
 #include <linux/hardirq.h>
@@ -77,8 +91,8 @@ struct ext_code {
 
 typedef void (*ext_int_handler_t)(struct ext_code, unsigned int, unsigned long);
 
-int register_external_interrupt(u16 code, ext_int_handler_t handler);
-int unregister_external_interrupt(u16 code, ext_int_handler_t handler);
+int register_external_irq(u16 code, ext_int_handler_t handler);
+int unregister_external_irq(u16 code, ext_int_handler_t handler);
 
 enum irq_subclass {
 	IRQ_SUBCLASS_MEASUREMENT_ALERT = 5,
diff --git a/arch/s390/include/asm/mmu.h b/arch/s390/include/asm/mmu.h
index ff132ac64ddd..f77695a82f64 100644
--- a/arch/s390/include/asm/mmu.h
+++ b/arch/s390/include/asm/mmu.h
@@ -1,9 +1,11 @@
 #ifndef __MMU_H
 #define __MMU_H
 
+#include <linux/cpumask.h>
 #include <linux/errno.h>
 
 typedef struct {
+	cpumask_t cpu_attach_mask;
 	atomic_t attach_count;
 	unsigned int flush_mm;
 	spinlock_t list_lock;
diff --git a/arch/s390/include/asm/mmu_context.h b/arch/s390/include/asm/mmu_context.h
index 38149b63dc44..71be346d0e3c 100644
--- a/arch/s390/include/asm/mmu_context.h
+++ b/arch/s390/include/asm/mmu_context.h
@@ -15,6 +15,7 @@
 static inline int init_new_context(struct task_struct *tsk,
 				   struct mm_struct *mm)
 {
+	cpumask_clear(&mm->context.cpu_attach_mask);
 	atomic_set(&mm->context.attach_count, 0);
 	mm->context.flush_mm = 0;
 	mm->context.asce_bits = _ASCE_TABLE_LENGTH | _ASCE_USER_BITS;
@@ -29,41 +30,61 @@ static inline int init_new_context(struct task_struct *tsk,
 
 #define destroy_context(mm)             do { } while (0)
 
-#ifndef CONFIG_64BIT
-#define LCTL_OPCODE "lctl"
-#else
-#define LCTL_OPCODE "lctlg"
-#endif
-
-static inline void update_mm(struct mm_struct *mm, struct task_struct *tsk)
+static inline void update_user_asce(struct mm_struct *mm, int load_primary)
 {
 	pgd_t *pgd = mm->pgd;
 
 	S390_lowcore.user_asce = mm->context.asce_bits | __pa(pgd);
-	/* Load primary space page table origin. */
-	asm volatile(LCTL_OPCODE" 1,1,%0\n" : : "m" (S390_lowcore.user_asce));
+	if (load_primary)
+		__ctl_load(S390_lowcore.user_asce, 1, 1);
 	set_fs(current->thread.mm_segment);
 }
 
+static inline void clear_user_asce(struct mm_struct *mm, int load_primary)
+{
+	S390_lowcore.user_asce = S390_lowcore.kernel_asce;
+
+	if (load_primary)
+		__ctl_load(S390_lowcore.user_asce, 1, 1);
+	__ctl_load(S390_lowcore.user_asce, 7, 7);
+}
+
+static inline void update_primary_asce(struct task_struct *tsk)
+{
+	unsigned long asce;
+
+	__ctl_store(asce, 1, 1);
+	if (asce != S390_lowcore.kernel_asce)
+		__ctl_load(S390_lowcore.kernel_asce, 1, 1);
+	set_tsk_thread_flag(tsk, TIF_ASCE);
+}
+
 static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next,
 			     struct task_struct *tsk)
 {
 	int cpu = smp_processor_id();
 
+	update_primary_asce(tsk);
 	if (prev == next)
 		return;
+	if (MACHINE_HAS_TLB_LC)
+		cpumask_set_cpu(cpu, &next->context.cpu_attach_mask);
 	if (atomic_inc_return(&next->context.attach_count) >> 16) {
-		/* Delay update_mm until all TLB flushes are done. */
+		/* Delay update_user_asce until all TLB flushes are done. */
 		set_tsk_thread_flag(tsk, TIF_TLB_WAIT);
+		/* Clear old ASCE by loading the kernel ASCE. */
+		clear_user_asce(next, 0);
 	} else {
 		cpumask_set_cpu(cpu, mm_cpumask(next));
-		update_mm(next, tsk);
+		update_user_asce(next, 0);
 		if (next->context.flush_mm)
 			/* Flush pending TLBs */
 			__tlb_flush_mm(next);
 	}
 	atomic_dec(&prev->context.attach_count);
 	WARN_ON(atomic_read(&prev->context.attach_count) < 0);
+	if (MACHINE_HAS_TLB_LC)
+		cpumask_clear_cpu(cpu, &prev->context.cpu_attach_mask);
 }
 
 #define finish_arch_post_lock_switch finish_arch_post_lock_switch
@@ -80,7 +101,7 @@ static inline void finish_arch_post_lock_switch(void)
 		cpu_relax();
 
 	cpumask_set_cpu(smp_processor_id(), mm_cpumask(mm));
-	update_mm(mm, tsk);
+	update_user_asce(mm, 0);
 	if (mm->context.flush_mm)
 		__tlb_flush_mm(mm);
 	preempt_enable();
diff --git a/arch/s390/include/asm/pgtable.h b/arch/s390/include/asm/pgtable.h
index 50a75d96f939..12f75313e086 100644
--- a/arch/s390/include/asm/pgtable.h
+++ b/arch/s390/include/asm/pgtable.h
@@ -1070,12 +1070,35 @@ static inline void __ptep_ipte(unsigned long address, pte_t *ptep)
 		: "=m" (*ptep) : "m" (*ptep), "a" (pto), "a" (address));
 }
 
+static inline void __ptep_ipte_local(unsigned long address, pte_t *ptep)
+{
+	unsigned long pto = (unsigned long) ptep;
+
+#ifndef CONFIG_64BIT
+	/* pto in ESA mode must point to the start of the segment table */
+	pto &= 0x7ffffc00;
+#endif
+	/* Invalidation + local TLB flush for the pte */
+	asm volatile(
+		"	.insn rrf,0xb2210000,%2,%3,0,1"
+		: "=m" (*ptep) : "m" (*ptep), "a" (pto), "a" (address));
+}
+
 static inline void ptep_flush_direct(struct mm_struct *mm,
 				     unsigned long address, pte_t *ptep)
 {
+	int active, count;
+
 	if (pte_val(*ptep) & _PAGE_INVALID)
 		return;
-	__ptep_ipte(address, ptep);
+	active = (mm == current->active_mm) ? 1 : 0;
+	count = atomic_add_return(0x10000, &mm->context.attach_count);
+	if (MACHINE_HAS_TLB_LC && (count & 0xffff) <= active &&
+	    cpumask_equal(mm_cpumask(mm), cpumask_of(smp_processor_id())))
+		__ptep_ipte_local(address, ptep);
+	else
+		__ptep_ipte(address, ptep);
+	atomic_sub(0x10000, &mm->context.attach_count);
 }
 
 static inline void ptep_flush_lazy(struct mm_struct *mm,
@@ -1384,35 +1407,6 @@ static inline pmd_t *pmd_offset(pud_t *pud, unsigned long address)
 #define pte_offset_map(pmd, address) pte_offset_kernel(pmd, address)
 #define pte_unmap(pte) do { } while (0)
 
-static inline void __pmd_idte(unsigned long address, pmd_t *pmdp)
-{
-	unsigned long sto = (unsigned long) pmdp -
-			    pmd_index(address) * sizeof(pmd_t);
-
-	if (!(pmd_val(*pmdp) & _SEGMENT_ENTRY_INVALID)) {
-		asm volatile(
-			"	.insn	rrf,0xb98e0000,%2,%3,0,0"
-			: "=m" (*pmdp)
-			: "m" (*pmdp), "a" (sto),
-			  "a" ((address & HPAGE_MASK))
-			: "cc"
-		);
-	}
-}
-
-static inline void __pmd_csp(pmd_t *pmdp)
-{
-	register unsigned long reg2 asm("2") = pmd_val(*pmdp);
-	register unsigned long reg3 asm("3") = pmd_val(*pmdp) |
-					       _SEGMENT_ENTRY_INVALID;
-	register unsigned long reg4 asm("4") = ((unsigned long) pmdp) + 5;
-
-	asm volatile(
-		"	csp %1,%3"
-		: "=m" (*pmdp)
-		: "d" (reg2), "d" (reg3), "d" (reg4), "m" (*pmdp) : "cc");
-}
-
 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLB_PAGE)
 static inline unsigned long massage_pgprot_pmd(pgprot_t pgprot)
 {
@@ -1481,18 +1475,80 @@ static inline pmd_t pmd_mkwrite(pmd_t pmd)
 }
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLB_PAGE */
 
+static inline void __pmdp_csp(pmd_t *pmdp)
+{
+	register unsigned long reg2 asm("2") = pmd_val(*pmdp);
+	register unsigned long reg3 asm("3") = pmd_val(*pmdp) |
+					       _SEGMENT_ENTRY_INVALID;
+	register unsigned long reg4 asm("4") = ((unsigned long) pmdp) + 5;
+
+	asm volatile(
+		"	csp %1,%3"
+		: "=m" (*pmdp)
+		: "d" (reg2), "d" (reg3), "d" (reg4), "m" (*pmdp) : "cc");
+}
+
+static inline void __pmdp_idte(unsigned long address, pmd_t *pmdp)
+{
+	unsigned long sto;
+
+	sto = (unsigned long) pmdp - pmd_index(address) * sizeof(pmd_t);
+	asm volatile(
+		"	.insn	rrf,0xb98e0000,%2,%3,0,0"
+		: "=m" (*pmdp)
+		: "m" (*pmdp), "a" (sto), "a" ((address & HPAGE_MASK))
+		: "cc" );
+}
+
+static inline void __pmdp_idte_local(unsigned long address, pmd_t *pmdp)
+{
+	unsigned long sto;
+
+	sto = (unsigned long) pmdp - pmd_index(address) * sizeof(pmd_t);
+	asm volatile(
+		"	.insn	rrf,0xb98e0000,%2,%3,0,1"
+		: "=m" (*pmdp)
+		: "m" (*pmdp), "a" (sto), "a" ((address & HPAGE_MASK))
+		: "cc" );
+}
+
+static inline void pmdp_flush_direct(struct mm_struct *mm,
+				     unsigned long address, pmd_t *pmdp)
+{
+	int active, count;
+
+	if (pmd_val(*pmdp) & _SEGMENT_ENTRY_INVALID)
+		return;
+	if (!MACHINE_HAS_IDTE) {
+		__pmdp_csp(pmdp);
+		return;
+	}
+	active = (mm == current->active_mm) ? 1 : 0;
+	count = atomic_add_return(0x10000, &mm->context.attach_count);
+	if (MACHINE_HAS_TLB_LC && (count & 0xffff) <= active &&
+	    cpumask_equal(mm_cpumask(mm), cpumask_of(smp_processor_id())))
+		__pmdp_idte_local(address, pmdp);
+	else
+		__pmdp_idte(address, pmdp);
+	atomic_sub(0x10000, &mm->context.attach_count);
+}
+
 static inline void pmdp_flush_lazy(struct mm_struct *mm,
 				   unsigned long address, pmd_t *pmdp)
 {
 	int active, count;
 
+	if (pmd_val(*pmdp) & _SEGMENT_ENTRY_INVALID)
+		return;
 	active = (mm == current->active_mm) ? 1 : 0;
 	count = atomic_add_return(0x1