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

Pull s390 updates from Martin Schwidefsky:
 "Just a random collection of bug-fixes and cleanups, nothing new in
  this merge request."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (46 commits)
  s390/ap: Fix wrong or missing comments
  s390/ap: move receive callback to message struct
  s390/dasd: re-prioritize partition detection message
  s390/qeth: reshuffle initialization
  s390/qeth: cleanup drv attr usage
  s390/claw: cleanup drv attr usage
  s390/lcs: cleanup drv attr usage
  s390/ctc: cleanup drv attr usage
  s390/ccwgroup: remove ccwgroup_create_from_string
  s390/qeth: stop using struct ccwgroup driver for discipline callbacks
  s390/qeth: switch to ccwgroup_create_dev
  s390/claw: switch to ccwgroup_create_dev
  s390/lcs: switch to ccwgroup_create_dev
  s390/ctcm: switch to ccwgroup_create_dev
  s390/ccwgroup: exploit ccwdev_by_dev_id
  s390/ccwgroup: introduce ccwgroup_create_dev
  s390: fix race on TIF_MCCK_PENDING
  s390/barrier: make use of fast-bcr facility
  s390/barrier: cleanup barrier functions
  s390/claw: remove "eieio" calls
  ...

54 files changed, 592 insertions, 1127 deletions

diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index 9015060919a..b42f2866594 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -217,7 +217,7 @@ config COMPAT
 	def_bool y
 	prompt "Kernel support for 31 bit emulation"
 	depends on 64BIT
-	select COMPAT_BINFMT_ELF
+	select COMPAT_BINFMT_ELF if BINFMT_ELF
 	select ARCH_WANT_OLD_COMPAT_IPC
 	help
 	  Select this option if you want to enable your system kernel to
@@ -234,6 +234,25 @@ config KEYS_COMPAT
 config AUDIT_ARCH
 	def_bool y
 
+config HAVE_MARCH_Z900_FEATURES
+	def_bool n
+
+config HAVE_MARCH_Z990_FEATURES
+	def_bool n
+	select HAVE_MARCH_Z900_FEATURES
+
+config HAVE_MARCH_Z9_109_FEATURES
+	def_bool n
+	select HAVE_MARCH_Z990_FEATURES
+
+config HAVE_MARCH_Z10_FEATURES
+	def_bool n
+	select HAVE_MARCH_Z9_109_FEATURES
+
+config HAVE_MARCH_Z196_FEATURES
+	def_bool n
+	select HAVE_MARCH_Z10_FEATURES
+
 comment "Code generation options"
 
 choice
@@ -249,6 +268,7 @@ config MARCH_G5
 
 config MARCH_Z900
 	bool "IBM zSeries model z800 and z900"
+	select HAVE_MARCH_Z900_FEATURES if 64BIT
 	help
 	  Select this to enable optimizations for model z800/z900 (2064 and
 	  2066 series). This will enable some optimizations that are not
@@ -256,6 +276,7 @@ config MARCH_Z900
 
 config MARCH_Z990
 	bool "IBM zSeries model z890 and z990"
+	select HAVE_MARCH_Z990_FEATURES if 64BIT
 	help
 	  Select this to enable optimizations for model z890/z990 (2084 and
 	  2086 series). The kernel will be slightly faster but will not work
@@ -263,6 +284,7 @@ config MARCH_Z990
 
 config MARCH_Z9_109
 	bool "IBM System z9"
+	select HAVE_MARCH_Z9_109_FEATURES if 64BIT
 	help
 	  Select this to enable optimizations for IBM System z9 (2094 and
 	  2096 series). The kernel will be slightly faster but will not work
@@ -270,6 +292,7 @@ config MARCH_Z9_109
 
 config MARCH_Z10
 	bool "IBM System z10"
+	select HAVE_MARCH_Z10_FEATURES if 64BIT
 	help
 	  Select this to enable optimizations for IBM System z10 (2097 and
 	  2098 series). The kernel will be slightly faster but will not work
@@ -277,6 +300,7 @@ config MARCH_Z10
 
 config MARCH_Z196
 	bool "IBM zEnterprise 114 and 196"
+	select HAVE_MARCH_Z196_FEATURES if 64BIT
 	help
 	  Select this to enable optimizations for IBM zEnterprise 114 and 196
 	  (2818 and 2817 series). The kernel will be slightly faster but will
@@ -406,33 +430,6 @@ config CHSC_SCH
 
 comment "Misc"
 
-config IPL
-	def_bool y
-	prompt "Builtin IPL record support"
-	help
-	  If you want to use the produced kernel to IPL directly from a
-	  device, you have to merge a bootsector specific to the device
-	  into the first bytes of the kernel. You will have to select the
-	  IPL device.
-
-choice
-	prompt "IPL method generated into head.S"
-	depends on IPL
-	default IPL_VM
-	help
-	  Select "tape" if you want to IPL the image from a Tape.
-
-	  Select "vm_reader" if you are running under VM/ESA and want
-	  to IPL the image from the emulated card reader.
-
-config IPL_TAPE
-	bool "tape"
-
-config IPL_VM
-	bool "vm_reader"
-
-endchoice
-
 source "fs/Kconfig.binfmt"
 
 config FORCE_MAX_ZONEORDER
@@ -569,7 +566,7 @@ config KEXEC
 
 config CRASH_DUMP
 	bool "kernel crash dumps"
-	depends on 64BIT
+	depends on 64BIT && SMP
 	select KEXEC
 	help
 	  Generate crash dump after being started by kexec.
diff --git a/arch/s390/boot/.gitignore b/arch/s390/boot/.gitignore
new file mode 100644
index 00000000000..017d5912ad2
--- /dev/null
+++ b/arch/s390/boot/.gitignore
@@ -0,0 +1,2 @@
+image
+bzImage
diff --git a/arch/s390/boot/compressed/.gitignore b/arch/s390/boot/compressed/.gitignore
new file mode 100644
index 00000000000..ae06b9b4c02
--- /dev/null
+++ b/arch/s390/boot/compressed/.gitignore
@@ -0,0 +1,3 @@
+sizes.h
+vmlinux
+vmlinux.lds
diff --git a/arch/s390/defconfig b/arch/s390/defconfig
index 1957a9dd256..37d2bf26796 100644
--- a/arch/s390/defconfig
+++ b/arch/s390/defconfig
@@ -155,7 +155,6 @@ CONFIG_CRYPTO_BLOWFISH=m
 CONFIG_CRYPTO_CAMELLIA=m
 CONFIG_CRYPTO_CAST5=m
 CONFIG_CRYPTO_CAST6=m
-CONFIG_CRYPTO_DES=m
 CONFIG_CRYPTO_FCRYPT=m
 CONFIG_CRYPTO_KHAZAD=m
 CONFIG_CRYPTO_SALSA20=m
diff --git a/arch/s390/include/asm/barrier.h b/arch/s390/include/asm/barrier.h
index 451273ad4d3..10a50880294 100644
--- a/arch/s390/include/asm/barrier.h
+++ b/arch/s390/include/asm/barrier.h
@@ -11,25 +11,28 @@
  * Force strict CPU ordering.
  * And yes, this is required on UP too when we're talking
  * to devices.
- *
- * This is very similar to the ppc eieio/sync instruction in that is
- * does a checkpoint syncronisation & makes sure that 
- * all memory ops have completed wrt other CPU's ( see 7-15 POP  DJB ).
  */
 
-#define eieio()	asm volatile("bcr 15,0" : : : "memory")
-#define SYNC_OTHER_CORES(x)   eieio()
-#define mb()    eieio()
-#define rmb()   eieio()
-#define wmb()   eieio()
-#define read_barrier_depends() do { } while(0)
-#define smp_mb()       mb()
-#define smp_rmb()      rmb()
-#define smp_wmb()      wmb()
-#define smp_read_barrier_depends()    read_barrier_depends()
-#define smp_mb__before_clear_bit()     smp_mb()
-#define smp_mb__after_clear_bit()      smp_mb()
+static inline void mb(void)
+{
+#ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
+	/* Fast-BCR without checkpoint synchronization */
+	asm volatile("bcr 14,0" : : : "memory");
+#else
+	asm volatile("bcr 15,0" : : : "memory");
+#endif
+}
+
+#define rmb()				mb()
+#define wmb()				mb()
+#define read_barrier_depends()		do { } while(0)
+#define smp_mb()			mb()
+#define smp_rmb()			rmb()
+#define smp_wmb()			wmb()
+#define smp_read_barrier_depends()	read_barrier_depends()
+#define smp_mb__before_clear_bit()	smp_mb()
+#define smp_mb__after_clear_bit()	smp_mb()
 
-#define set_mb(var, value)      do { var = value; mb(); } while (0)
+#define set_mb(var, value)		do { var = value; mb(); } while (0)
 
 #endif /* __ASM_BARRIER_H */
diff --git a/arch/s390/include/asm/ccwgroup.h b/arch/s390/include/asm/ccwgroup.h
index f2ea2c56a7e..f2ef34f6d6e 100644
--- a/arch/s390/include/asm/ccwgroup.h
+++ b/arch/s390/include/asm/ccwgroup.h
@@ -29,9 +29,7 @@ struct ccwgroup_device {
 
 /**
  * struct ccwgroup_driver - driver for ccw group devices
- * @max_slaves: maximum number of slave devices
- * @driver_id: unique id
- * @probe: function called on probe
+ * @setup: function called during device creation to setup the device
  * @remove: function called on remove
  * @set_online: function called when device is set online
  * @set_offline: function called when device is set offline
@@ -44,10 +42,7 @@ struct ccwgroup_device {
  * @driver: embedded driver structure
  */
 struct ccwgroup_driver {
-	int max_slaves;
-	unsigned long driver_id;
-
-	int (*probe) (struct ccwgroup_device *);
+	int (*setup) (struct ccwgroup_device *);
 	void (*remove) (struct ccwgroup_device *);
 	int (*set_online) (struct ccwgroup_device *);
 	int (*set_offline) (struct ccwgroup_device *);
@@ -63,9 +58,8 @@ struct ccwgroup_driver {
 
 extern int  ccwgroup_driver_register   (struct ccwgroup_driver *cdriver);
 extern void ccwgroup_driver_unregister (struct ccwgroup_driver *cdriver);
-int ccwgroup_create_from_string(struct device *root, unsigned int creator_id,
-				struct ccw_driver *cdrv, int num_devices,
-				const char *buf);
+int ccwgroup_create_dev(struct device *root, struct ccwgroup_driver *gdrv,
+			int num_devices, const char *buf);
 
 extern int ccwgroup_probe_ccwdev(struct ccw_device *cdev);
 extern void ccwgroup_remove_ccwdev(struct ccw_device *cdev);
diff --git a/arch/s390/include/asm/io.h b/arch/s390/include/asm/io.h
index b7ff6afc3ca..27216d31799 100644
--- a/arch/s390/include/asm/io.h
+++ b/arch/s390/include/asm/io.h
@@ -38,11 +38,8 @@ static inline void * phys_to_virt(unsigned long address)
 	return (void *) address;
 }
 
-/*
- * Convert a physical pointer to a virtual kernel pointer for /dev/mem
- * access
- */
-#define xlate_dev_mem_ptr(p)	__va(p)
+void *xlate_dev_mem_ptr(unsigned long phys);
+void unxlate_dev_mem_ptr(unsigned long phys, void *addr);
 
 /*
  * Convert a virtual cached pointer to an uncached pointer
diff --git a/arch/s390/include/asm/qdio.h b/arch/s390/include/asm/qdio.h
index d75c8e78f7e..f039d86adf6 100644
--- a/arch/s390/include/asm/qdio.h
+++ b/arch/s390/include/asm/qdio.h
@@ -258,11 +258,6 @@ struct slsb {
 	u8 val[QDIO_MAX_BUFFERS_PER_Q];
 } __attribute__ ((packed, aligned(256)));
 
-#define CHSC_AC2_MULTI_BUFFER_AVAILABLE	0x0080
-#define CHSC_AC2_MULTI_BUFFER_ENABLED	0x0040
-#define CHSC_AC2_DATA_DIV_AVAILABLE	0x0010
-#define CHSC_AC2_DATA_DIV_ENABLED	0x0002
-
 /**
  * struct qdio_outbuf_state - SBAL related asynchronous operation information
  *   (for communication with upper layer programs)
@@ -293,6 +288,8 @@ struct qdio_outbuf_state {
 #define AC1_SC_QEBSM_AVAILABLE		0x02	/* available for subchannel */
 #define AC1_SC_QEBSM_ENABLED		0x01	/* enabled for subchannel */
 
+#define CHSC_AC2_MULTI_BUFFER_AVAILABLE	0x0080
+#define CHSC_AC2_MULTI_BUFFER_ENABLED	0x0040
 #define CHSC_AC2_DATA_DIV_AVAILABLE	0x0010
 #define CHSC_AC2_DATA_DIV_ENABLED	0x0002
 
@@ -328,11 +325,13 @@ typedef void qdio_handler_t(struct ccw_device *, unsigned int, int,
 			    int, int, unsigned long);
 
 /* qdio errors reported to the upper-layer program */
-#define QDIO_ERROR_SIGA_TARGET			0x02
-#define QDIO_ERROR_SIGA_ACCESS_EXCEPTION	0x10
-#define QDIO_ERROR_SIGA_BUSY			0x20
-#define QDIO_ERROR_ACTIVATE_CHECK_CONDITION	0x40
-#define QDIO_ERROR_SLSB_STATE			0x80
+#define QDIO_ERROR_ACTIVATE			0x0001
+#define QDIO_ERROR_GET_BUF_STATE		0x0002
+#define QDIO_ERROR_SET_BUF_STATE		0x0004
+#define QDIO_ERROR_SLSB_STATE			0x0100
+
+#define QDIO_ERROR_FATAL			0x00ff
+#define QDIO_ERROR_TEMPORARY			0xff00
 
 /* for qdio_cleanup */
 #define QDIO_FLAG_CLEANUP_USING_CLEAR		0x01
diff --git a/arch/s390/include/asm/setup.h b/arch/s390/include/asm/setup.h
index b21e46e5d4b..7244e1f6412 100644
--- a/arch/s390/include/asm/setup.h
+++ b/arch/s390/include/asm/setup.h
@@ -82,7 +82,6 @@ extern unsigned int user_mode;
 #define MACHINE_FLAG_LPAR	(1UL << 12)
 #define MACHINE_FLAG_SPP	(1UL << 13)
 #define MACHINE_FLAG_TOPOLOGY	(1UL << 14)
-#define MACHINE_FLAG_STCKF	(1UL << 15)
 
 #define MACHINE_IS_VM		(S390_lowcore.machine_flags & MACHINE_FLAG_VM)
 #define MACHINE_IS_KVM		(S390_lowcore.machine_flags & MACHINE_FLAG_KVM)
@@ -101,7 +100,6 @@ extern unsigned int user_mode;
 #define MACHINE_HAS_PFMF	(0)
 #define MACHINE_HAS_SPP		(0)
 #define MACHINE_HAS_TOPOLOGY	(0)
-#define MACHINE_HAS_STCKF	(0)
 #else /* __s390x__ */
 #define MACHINE_HAS_IEEE	(1)
 #define MACHINE_HAS_CSP		(1)
@@ -113,7 +111,6 @@ extern unsigned int user_mode;
 #define MACHINE_HAS_PFMF	(S390_lowcore.machine_flags & MACHINE_FLAG_PFMF)
 #define MACHINE_HAS_SPP		(S390_lowcore.machine_flags & MACHINE_FLAG_SPP)
 #define MACHINE_HAS_TOPOLOGY	(S390_lowcore.machine_flags & MACHINE_FLAG_TOPOLOGY)
-#define MACHINE_HAS_STCKF	(S390_lowcore.machine_flags & MACHINE_FLAG_STCKF)
 #endif /* __s390x__ */
 
 #define ZFCPDUMP_HSA_SIZE	(32UL<<20)
diff --git a/arch/s390/include/asm/thread_info.h b/arch/s390/include/asm/thread_info.h
index a73038155e0..003b04edcff 100644
--- a/arch/s390/include/asm/thread_info.h
+++ b/arch/s390/include/asm/thread_info.h
@@ -95,7 +95,6 @@ static inline struct thread_info *current_thread_info(void)
 #define TIF_SYSCALL_AUDIT	9	/* syscall auditing active */
 #define TIF_SECCOMP		10	/* secure computing */
 #define TIF_SYSCALL_TRACEPOINT	11	/* syscall tracepoint instrumentation */
-#define TIF_SIE			12	/* guest execution active */
 #define TIF_POLLING_NRFLAG	16	/* true if poll_idle() is polling
 					   TIF_NEED_RESCHED */
 #define TIF_31BIT		17	/* 32bit process */
@@ -114,7 +113,6 @@ static inline struct thread_info *current_thread_info(void)
 #define _TIF_SYSCALL_AUDIT	(1<<TIF_SYSCALL_AUDIT)
 #define _TIF_SECCOMP		(1<<TIF_SECCOMP)
 #define _TIF_SYSCALL_TRACEPOINT	(1<<TIF_SYSCALL_TRACEPOINT)
-#define _TIF_SIE		(1<<TIF_SIE)
 #define _TIF_POLLING_NRFLAG	(1<<TIF_POLLING_NRFLAG)
 #define _TIF_31BIT		(1<<TIF_31BIT)
 #define _TIF_SINGLE_STEP	(1<<TIF_SINGLE_STEP)
diff --git a/arch/s390/include/asm/timex.h b/arch/s390/include/asm/timex.h
index c447a27a7fd..239ece9e53c 100644
--- a/arch/s390/include/asm/timex.h
+++ b/arch/s390/include/asm/timex.h
@@ -73,11 +73,15 @@ static inline void local_tick_enable(unsigned long long comp)
 
 typedef unsigned long long cycles_t;
 
-static inline unsigned long long get_clock (void)
+static inline unsigned long long get_clock(void)
 {
 	unsigned long long clk;
 
+#ifdef CONFIG_HAVE_MARCH_Z9_109_FEATURES
+	asm volatile(".insn s,0xb27c0000,%0" : "=Q" (clk) : : "cc");
+#else
 	asm volatile("stck %0" : "=Q" (clk) : : "cc");
+#endif
 	return clk;
 }
 
@@ -86,17 +90,6 @@ static inline void get_clock_ext(char *clk)
 	asm volatile("stcke %0" : "=Q" (*clk) : : "cc");
 }
 
-static inline unsigned long long get_clock_fast(void)
-{
-	unsigned long long clk;
-
-	if (MACHINE_HAS_STCKF)
-		asm volatile(".insn	s,0xb27c0000,%0" : "=Q" (clk) : : "cc");
-	else
-		clk = get_clock();
-	return clk;
-}
-
 static inline unsigned long long get_clock_xt(void)
 {
 	unsigned char clk[16];
diff --git a/arch/s390/kernel/.gitignore b/arch/s390/kernel/.gitignore
new file mode 100644
index 00000000000..c5f676c3c22
--- /dev/null
+++ b/arch/s390/kernel/.gitignore
@@ -0,0 +1 @@
+vmlinux.lds
diff --git a/arch/s390/kernel/compat_signal.c b/arch/s390/kernel/compat_signal.c
index 28040fd5e8a..377c096ca4a 100644
--- a/arch/s390/kernel/compat_signal.c
+++ b/arch/s390/kernel/compat_signal.c
@@ -437,13 +437,6 @@ get_sigframe(struct k_sigaction *ka, struct pt_regs * regs, size_t frame_size)
 			sp = current->sas_ss_sp + current->sas_ss_size;
 	}
 
-	/* This is the legacy signal stack switching. */
-	else if (!user_mode(regs) &&
-		 !(ka->sa.sa_flags & SA_RESTORER) &&
-		 ka->sa.sa_restorer) {
-		sp = (unsigned long) ka->sa.sa_restorer;
-	}
-
 	return (void __user *)((sp - frame_size) & -8ul);
 }
 
diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c
index 9475e682727..d84181f1f5e 100644
--- a/arch/s390/kernel/early.c
+++ b/arch/s390/kernel/early.c
@@ -374,8 +374,6 @@ static __init void detect_machine_facilities(void)
 		S390_lowcore.machine_flags |= MACHINE_FLAG_MVCOS;
 	if (test_facility(40))
 		S390_lowcore.machine_flags |= MACHINE_FLAG_SPP;
-	if (test_facility(25))
-		S390_lowcore.machine_flags |= MACHINE_FLAG_STCKF;
 #endif
 }
 
diff --git a/arch/s390/kernel/entry.S b/arch/s390/kernel/entry.S
index 74ee563fe62..1ae93b573d7 100644
--- a/arch/s390/kernel/entry.S
+++ b/arch/s390/kernel/entry.S
@@ -145,22 +145,23 @@ STACK_SIZE  = 1 << STACK_SHIFT
  *  gpr2 = prev
  */
 ENTRY(__switch_to)
+	stm	%r6,%r15,__SF_GPRS(%r15)	# store gprs of prev task
+	st	%r15,__THREAD_ksp(%r2)		# store kernel stack of prev
 	l	%r4,__THREAD_info(%r2)		# get thread_info of prev
 	l	%r5,__THREAD_info(%r3)		# get thread_info of next
+	lr	%r15,%r5
+	ahi	%r15,STACK_SIZE			# end of kernel stack of next
+	st	%r3,__LC_CURRENT		# store task struct of next
+	st	%r5,__LC_THREAD_INFO		# store thread info of next
+	st	%r15,__LC_KERNEL_STACK		# store end of kernel stack
+	lctl	%c4,%c4,__TASK_pid(%r3)		# load pid to control reg. 4
+	mvc	__LC_CURRENT_PID(4,%r0),__TASK_pid(%r3)	# store pid of next
+	l	%r15,__THREAD_ksp(%r3)		# load kernel stack of next
 	tm	__TI_flags+3(%r4),_TIF_MCCK_PENDING # machine check pending?
 	jz	0f
 	ni	__TI_flags+3(%r4),255-_TIF_MCCK_PENDING	# clear flag in prev
 	oi	__TI_flags+3(%r5),_TIF_MCCK_PENDING	# set it in next
-0:	stm	%r6,%r15,__SF_GPRS(%r15)	# store gprs of prev task
-	st	%r15,__THREAD_ksp(%r2)		# store kernel stack of prev
-	l	%r15,__THREAD_ksp(%r3)		# load kernel stack of next
-	lctl	%c4,%c4,__TASK_pid(%r3)		# load pid to control reg. 4
-	lm	%r6,%r15,__SF_GPRS(%r15)	# load gprs of next task
-	st	%r3,__LC_CURRENT		# store task struct of next
-	mvc	__LC_CURRENT_PID(4,%r0),__TASK_pid(%r3)	# store pid of next
-	st	%r5,__LC_THREAD_INFO		# store thread info of next
-	ahi	%r5,STACK_SIZE			# end of kernel stack of next
-	st	%r5,__LC_KERNEL_STACK		# store end of kernel stack
+0:	lm	%r6,%r15,__SF_GPRS(%r15)	# load gprs of next task
 	br	%r14
 
 __critical_start:
diff --git a/arch/s390/kernel/entry64.S b/arch/s390/kernel/entry64.S
index 4e1c292fa7e..229fe1d0774 100644
--- a/arch/s390/kernel/entry64.S
+++ b/arch/s390/kernel/entry64.S
@@ -81,16 +81,14 @@ _TIF_EXIT_SIE = (_TIF_SIGPENDING | _TIF_NEED_RESCHED | _TIF_MCCK_PENDING)
 
 	.macro	HANDLE_SIE_INTERCEPT scratch
 #if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE)
-	tm	__TI_flags+6(%r12),_TIF_SIE>>8
-	jz	.+42
-	tm	__LC_MACHINE_FLAGS+6,0x20	# MACHINE_FLAG_SPP
-	jz	.+8
-	.insn	s,0xb2800000,BASED(.Lhost_id)	# set host id
+	tmhh	%r8,0x0001		# interrupting from user ?
+	jnz	.+42
 	lgr	\scratch,%r9
 	slg	\scratch,BASED(.Lsie_loop)
 	clg	\scratch,BASED(.Lsie_length)
-	jhe	.+10
+	jhe	.+22
 	lg	%r9,BASED(.Lsie_loop)
+	SPP	BASED(.Lhost_id)	# set host id
 #endif
 	.endm
 
@@ -148,6 +146,14 @@ _TIF_EXIT_SIE = (_TIF_SIGPENDING | _TIF_NEED_RESCHED | _TIF_MCCK_PENDING)
 	ssm	__LC_RETURN_PSW
 	.endm
 
+	.macro STCK savearea
+#ifdef CONFIG_HAVE_MARCH_Z9_109_FEATURES
+	.insn	s,0xb27c0000,\savearea		# store clock fast
+#else
+	.insn	s,0xb2050000,\savearea		# store clock
+#endif
+	.endm
+
 	.section .kprobes.text, "ax"
 
 /*
@@ -158,22 +164,23 @@ _TIF_EXIT_SIE = (_TIF_SIGPENDING | _TIF_NEED_RESCHED | _TIF_MCCK_PENDING)
  *  gpr2 = prev
  */
 ENTRY(__switch_to)
+	stmg	%r6,%r15,__SF_GPRS(%r15)	# store gprs of prev task
+	stg	%r15,__THREAD_ksp(%r2)		# store kernel stack of prev
 	lg	%r4,__THREAD_info(%r2)		# get thread_info of prev
 	lg	%r5,__THREAD_info(%r3)		# get thread_info of next
+	lgr	%r15,%r5
+	aghi	%r15,STACK_SIZE			# end of kernel stack of next
+	stg	%r3,__LC_CURRENT		# store task struct of next
+	stg	%r5,__LC_THREAD_INFO		# store thread info of next
+	stg	%r15,__LC_KERNEL_STACK		# store end of kernel stack
+	lctl	%c4,%c4,__TASK_pid(%r3)		# load pid to control reg. 4
+	mvc	__LC_CURRENT_PID+4(4,%r0),__TASK_pid(%r3) # store pid of next
+	lg	%r15,__THREAD_ksp(%r3)		# load kernel stack of next
 	tm	__TI_flags+7(%r4),_TIF_MCCK_PENDING # machine check pending?
 	jz	0f
 	ni	__TI_flags+7(%r4),255-_TIF_MCCK_PENDING	# clear flag in prev
 	oi	__TI_flags+7(%r5),_TIF_MCCK_PENDING	# set it in next
-0:	stmg	%r6,%r15,__SF_GPRS(%r15)	# store gprs of prev task
-	stg	%r15,__THREAD_ksp(%r2)		# store kernel stack of prev
-	lg	%r15,__THREAD_ksp(%r3)		# load kernel stack of next
-	lctl	%c4,%c4,__TASK_pid(%r3)		# load pid to control reg. 4
-	lmg	%r6,%r15,__SF_GPRS(%r15)	# load gprs of next task
-	stg	%r3,__LC_CURRENT		# store task struct of next
-	mvc	__LC_CURRENT_PID+4(4,%r0),__TASK_pid(%r3) # store pid of next
-	stg	%r5,__LC_THREAD_INFO		# store thread info of next
-	aghi	%r5,STACK_SIZE			# end of kernel stack of next
-	stg	%r5,__LC_KERNEL_STACK		# store end of kernel stack
+0:	lmg	%r6,%r15,__SF_GPRS(%r15)	# load gprs of next task
 	br	%r14
 
 __critical_start:
@@ -458,7 +465,7 @@ pgm_svcper:
  * IO interrupt handler routine
  */
 ENTRY(io_int_handler)
-	stck	__LC_INT_CLOCK
+	STCK	__LC_INT_CLOCK
 	stpt	__LC_ASYNC_ENTER_TIMER
 	stmg	%r8,%r15,__LC_SAVE_AREA_ASYNC
 	lg	%r10,__LC_LAST_BREAK
@@ -604,7 +611,7 @@ io_notify_resume:
  * External interrupt handler routine
  */
 ENTRY(ext_int_handler)
-	stck	__LC_INT_CLOCK
+	STCK	__LC_INT_CLOCK
 	stpt	__LC_ASYNC_ENTER_TIMER
 	stmg	%r8,%r15,__LC_SAVE_AREA_ASYNC
 	lg	%r10,__LC_LAST_BREAK
@@ -622,6 +629,7 @@ ext_skip:
 	mvc	__PT_R8(64,%r11),__LC_SAVE_AREA_ASYNC
 	stmg	%r8,%r9,__PT_PSW(%r11)
 	TRACE_IRQS_OFF
+	xc	__SF_BACKCHAIN(8,%r15),__SF_BACKCHAIN(%r15)
 	lghi	%r1,4096
 	lgr	%r2,%r11		# pass pointer to pt_regs
 	llgf	%r3,__LC_EXT_CPU_ADDR	# get cpu address + interruption code
@@ -638,7 +646,7 @@ ENTRY(psw_idle)
 	larl	%r1,psw_idle_lpsw+4
 	stg	%r1,__SF_EMPTY+8(%r15)
 	larl	%r1,.Lvtimer_max
-	stck	__IDLE_ENTER(%r2)
+	STCK	__IDLE_ENTER(%r2)
 	ltr	%r5,%r5
 	stpt	__VQ_IDLE_ENTER(%r3)
 	jz	psw_idle_lpsw
@@ -654,7 +662,7 @@ __critical_end:
  * Machine check handler routines
  */
 ENTRY(mcck_int_handler)
-	stck	__LC_MCCK_CLOCK
+	STCK	__LC_MCCK_CLOCK
 	la	%r1,4095		# revalidate r1
 	spt	__LC_CPU_TIMER_SAVE_AREA-4095(%r1)	# revalidate cpu timer
 	lmg	%r0,%r15,__LC_GPREGS_SAVE_AREA-4095(%r1)# revalidate gprs
@@ -967,7 +975,6 @@ ENTRY(sie64a)
 	xc	__SF_EMPTY+16(8,%r15),__SF_EMPTY+16(%r15) #