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

Pull s390 updates from Martin Schwidefsky:
 "This patch set contains the main portion of the changes for 3.18 in
  regard to the s390 architecture.  It is a bit bigger than usual,
  mainly because of a new driver and the vector extension patches.

  The interesting bits are:
   - Quite a bit of work on the tracing front.  Uprobes is enabled and
     the ftrace code is reworked to get some of the lost performance
     back if CONFIG_FTRACE is enabled.
   - To improve boot time with CONFIG_DEBIG_PAGEALLOC, support for the
     IPTE range facility is added.
   - The rwlock code is re-factored to improve writer fairness and to be
     able to use the interlocked-access instructions.
   - The kernel part for the support of the vector extension is added.
   - The device driver to access the CD/DVD on the HMC is added, this
     will hopefully come in handy to improve the installation process.
   - Add support for control-unit initiated reconfiguration.
   - The crypto device driver is enhanced to enable the additional AP
     domains and to allow the new crypto hardware to be used.
   - Bug fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (39 commits)
  s390/ftrace: simplify enabling/disabling of ftrace_graph_caller
  s390/ftrace: remove 31 bit ftrace support
  s390/kdump: add support for vector extension
  s390/disassembler: add vector instructions
  s390: add support for vector extension
  s390/zcrypt: Toleration of new crypto hardware
  s390/idle: consolidate idle functions and definitions
  s390/nohz: use a per-cpu flag for arch_needs_cpu
  s390/vtime: do not reset idle data on CPU hotplug
  s390/dasd: add support for control unit initiated reconfiguration
  s390/dasd: fix infinite loop during format
  s390/mm: make use of ipte range facility
  s390/setup: correct 4-level kernel page table detection
  s390/topology: call set_sched_topology early
  s390/uprobes: architecture backend for uprobes
  s390/uprobes: common library for kprobes and uprobes
  s390/rwlock: use the interlocked-access facility 1 instructions
  s390/rwlock: improve writer fairness
  s390/rwlock: remove interrupt-enabling rwlock variant.
  s390/mm: remove change bit override support
  ...

27 files changed, 368 insertions, 105 deletions

diff --git a/arch/s390/include/asm/barrier.h b/arch/s390/include/asm/barrier.h
index 19ff956b752..b5dce6544d7 100644
--- a/arch/s390/include/asm/barrier.h
+++ b/arch/s390/include/asm/barrier.h
@@ -15,11 +15,13 @@
 
 #ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
 /* Fast-BCR without checkpoint synchronization */
-#define mb() do {  asm volatile("bcr 14,0" : : : "memory"); } while (0)
+#define __ASM_BARRIER "bcr 14,0\n"
 #else
-#define mb() do {  asm volatile("bcr 15,0" : : : "memory"); } while (0)
+#define __ASM_BARRIER "bcr 15,0\n"
 #endif
 
+#define mb() do {  asm volatile(__ASM_BARRIER : : : "memory"); } while (0)
+
 #define rmb()				mb()
 #define wmb()				mb()
 #define read_barrier_depends()		do { } while(0)
diff --git a/arch/s390/include/asm/cputime.h b/arch/s390/include/asm/cputime.h
index 3001887f94b..f8c19698485 100644
--- a/arch/s390/include/asm/cputime.h
+++ b/arch/s390/include/asm/cputime.h
@@ -8,8 +8,6 @@
 #define _S390_CPUTIME_H
 
 #include <linux/types.h>
-#include <linux/percpu.h>
-#include <linux/spinlock.h>
 #include <asm/div64.h>
 
 
@@ -167,28 +165,8 @@ static inline clock_t cputime64_to_clock_t(cputime64_t cputime)
 	return clock;
 }
 
-struct s390_idle_data {
-	int nohz_delay;
-	unsigned int sequence;
-	unsigned long long idle_count;
-	unsigned long long idle_time;
-	unsigned long long clock_idle_enter;
-	unsigned long long clock_idle_exit;
-	unsigned long long timer_idle_enter;
-	unsigned long long timer_idle_exit;
-};
+cputime64_t arch_cpu_idle_time(int cpu);
 
-DECLARE_PER_CPU(struct s390_idle_data, s390_idle);
-
-cputime64_t s390_get_idle_time(int cpu);
-
-#define arch_idle_time(cpu) s390_get_idle_time(cpu)
-
-static inline int s390_nohz_delay(int cpu)
-{
-	return __get_cpu_var(s390_idle).nohz_delay != 0;
-}
-
-#define arch_needs_cpu(cpu) s390_nohz_delay(cpu)
+#define arch_idle_time(cpu) arch_cpu_idle_time(cpu)
 
 #endif /* _S390_CPUTIME_H */
diff --git a/arch/s390/include/asm/dis.h b/arch/s390/include/asm/dis.h
index 04a83f5773c..60323c21938 100644
--- a/arch/s390/include/asm/dis.h
+++ b/arch/s390/include/asm/dis.h
@@ -13,12 +13,13 @@
 #define OPERAND_FPR	0x2	/* Operand printed as %fx */
 #define OPERAND_AR	0x4	/* Operand printed as %ax */
 #define OPERAND_CR	0x8	/* Operand printed as %cx */
-#define OPERAND_DISP	0x10	/* Operand printed as displacement */
-#define OPERAND_BASE	0x20	/* Operand printed as base register */
-#define OPERAND_INDEX	0x40	/* Operand printed as index register */
-#define OPERAND_PCREL	0x80	/* Operand printed as pc-relative symbol */
-#define OPERAND_SIGNED	0x100	/* Operand printed as signed value */
-#define OPERAND_LENGTH	0x200	/* Operand printed as length (+1) */
+#define OPERAND_VR	0x10	/* Operand printed as %vx */
+#define OPERAND_DISP	0x20	/* Operand printed as displacement */
+#define OPERAND_BASE	0x40	/* Operand printed as base register */
+#define OPERAND_INDEX	0x80	/* Operand printed as index register */
+#define OPERAND_PCREL	0x100	/* Operand printed as pc-relative symbol */
+#define OPERAND_SIGNED	0x200	/* Operand printed as signed value */
+#define OPERAND_LENGTH	0x400	/* Operand printed as length (+1) */
 
 
 struct s390_operand {
diff --git a/arch/s390/include/asm/elf.h b/arch/s390/include/asm/elf.h
index 78f4f8711d5..f6e43d39e3d 100644
--- a/arch/s390/include/asm/elf.h
+++ b/arch/s390/include/asm/elf.h
@@ -102,6 +102,7 @@
 #define HWCAP_S390_ETF3EH	256
 #define HWCAP_S390_HIGH_GPRS	512
 #define HWCAP_S390_TE		1024
+#define HWCAP_S390_VXRS		2048
 
 /*
  * These are used to set parameters in the core dumps.
@@ -225,6 +226,6 @@ int arch_setup_additional_pages(struct linux_binprm *, int);
 extern unsigned long arch_randomize_brk(struct mm_struct *mm);
 #define arch_randomize_brk arch_randomize_brk
 
-void *fill_cpu_elf_notes(void *ptr, struct save_area *sa);
+void *fill_cpu_elf_notes(void *ptr, struct save_area *sa, __vector128 *vxrs);
 
 #endif
diff --git a/arch/s390/include/asm/ftrace.h b/arch/s390/include/asm/ftrace.h
index bf246dae136..3aef8afec33 100644
--- a/arch/s390/include/asm/ftrace.h
+++ b/arch/s390/include/asm/ftrace.h
@@ -4,6 +4,7 @@
 #ifndef __ASSEMBLY__
 
 extern void _mcount(void);
+extern char ftrace_graph_caller_end;
 
 struct dyn_arch_ftrace { };
 
@@ -17,10 +18,8 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr)
 
 #endif /* __ASSEMBLY__ */
 
-#ifdef CONFIG_64BIT
-#define MCOUNT_INSN_SIZE  12
-#else
-#define MCOUNT_INSN_SIZE  22
-#endif
+#define MCOUNT_INSN_SIZE  18
+
+#define ARCH_SUPPORTS_FTRACE_OPS 1
 
 #endif /* _ASM_S390_FTRACE_H */
diff --git a/arch/s390/include/asm/idle.h b/arch/s390/include/asm/idle.h
new file mode 100644
index 00000000000..6af037f574b
--- /dev/null
+++ b/arch/s390/include/asm/idle.h
@@ -0,0 +1,26 @@
+/*
+ *  Copyright IBM Corp. 2014
+ *
+ *  Author: Martin Schwidefsky <schwidefsky@de.ibm.com>
+ */
+
+#ifndef _S390_IDLE_H
+#define _S390_IDLE_H
+
+#include <linux/types.h>
+#include <linux/device.h>
+
+struct s390_idle_data {
+	unsigned int sequence;
+	unsigned long long idle_count;
+	unsigned long long idle_time;
+	unsigned long long clock_idle_enter;
+	unsigned long long clock_idle_exit;
+	unsigned long long timer_idle_enter;
+	unsigned long long timer_idle_exit;
+};
+
+extern struct device_attribute dev_attr_idle_count;
+extern struct device_attribute dev_attr_idle_time_us;
+
+#endif /* _S390_IDLE_H */
diff --git a/arch/s390/include/asm/ipl.h b/arch/s390/include/asm/ipl.h
index c81661e756a..ece606c2ee8 100644
--- a/arch/s390/include/asm/ipl.h
+++ b/arch/s390/include/asm/ipl.h
@@ -89,12 +89,12 @@ extern u32 ipl_flags;
 extern u32 dump_prefix_page;
 
 struct dump_save_areas {
-	struct save_area **areas;
+	struct save_area_ext **areas;
 	int count;
 };
 
 extern struct dump_save_areas dump_save_areas;
-struct save_area *dump_save_area_create(int cpu);
+struct save_area_ext *dump_save_area_create(int cpu);
 
 extern void do_reipl(void);
 extern void do_halt(void);
diff --git a/arch/s390/include/asm/irq.h b/arch/s390/include/asm/irq.h
index c4dd400a279..e787cc1bff8 100644
--- a/arch/s390/include/asm/irq.h
+++ b/arch/s390/include/asm/irq.h
@@ -51,6 +51,7 @@ enum interruption_class {
 	IRQEXT_CMS,
 	IRQEXT_CMC,
 	IRQEXT_CMR,
+	IRQEXT_FTP,
 	IRQIO_CIO,
 	IRQIO_QAI,
 	IRQIO_DAS,
diff --git a/arch/s390/include/asm/kprobes.h b/arch/s390/include/asm/kprobes.h
index 4176dfe0fba..98629173ce3 100644
--- a/arch/s390/include/asm/kprobes.h
+++ b/arch/s390/include/asm/kprobes.h
@@ -84,6 +84,10 @@ int kprobe_fault_handler(struct pt_regs *regs, int trapnr);
 int kprobe_exceptions_notify(struct notifier_block *self,
 	unsigned long val, void *data);
 
+int probe_is_prohibited_opcode(u16 *insn);
+int probe_get_fixup_type(u16 *insn);
+int probe_is_insn_relative_long(u16 *insn);
+
 #define flush_insn_slot(p)	do { } while (0)
 
 #endif	/* _ASM_S390_KPROBES_H */
diff --git a/arch/s390/include/asm/lowcore.h b/arch/s390/include/asm/lowcore.h
index 4349197ab9d..6cc51fe8441 100644
--- a/arch/s390/include/asm/lowcore.h
+++ b/arch/s390/include/asm/lowcore.h
@@ -11,6 +11,7 @@
 #include <linux/types.h>
 #include <asm/ptrace.h>
 #include <asm/cpu.h>
+#include <asm/types.h>
 
 #ifdef CONFIG_32BIT
 
@@ -31,6 +32,11 @@ struct save_area {
 	u32	ctrl_regs[16];
 } __packed;
 
+struct save_area_ext {
+	struct save_area	sa;
+	__vector128		vx_regs[32];
+};
+
 struct _lowcore {
 	psw_t	restart_psw;			/* 0x0000 */
 	psw_t	restart_old_psw;		/* 0x0008 */
@@ -183,6 +189,11 @@ struct save_area {
 	u64	ctrl_regs[16];
 } __packed;
 
+struct save_area_ext {
+	struct save_area	sa;
+	__vector128		vx_regs[32];
+};
+
 struct _lowcore {
 	__u8	pad_0x0000[0x0014-0x0000];	/* 0x0000 */
 	__u32	ipl_parmblock_ptr;		/* 0x0014 */
@@ -310,7 +321,10 @@ struct _lowcore {
 
 	/* Extended facility list */
 	__u64	stfle_fac_list[32];		/* 0x0f00 */
-	__u8	pad_0x1000[0x11b8-0x1000];	/* 0x1000 */
+	__u8	pad_0x1000[0x11b0-0x1000];	/* 0x1000 */
+
+	/* Pointer to vector register save area */
+	__u64	vector_save_area_addr;		/* 0x11b0 */
 
 	/* 64 bit extparam used for pfault/diag 250: defined by architecture */
 	__u64	ext_params2;			/* 0x11B8 */
@@ -334,9 +348,10 @@ struct _lowcore {
 
 	/* Transaction abort diagnostic block */
 	__u8	pgm_tdb[256];			/* 0x1800 */
+	__u8	pad_0x1900[0x1c00-0x1900];	/* 0x1900 */
 
-	/* align to the top of the prefix area */
-	__u8	pad_0x1900[0x2000-0x1900];	/* 0x1900 */
+	/* Software defined save area for vector registers */
+	__u8	vector_save_area[1024];		/* 0x1c00 */
 } __packed;
 
 #endif /* CONFIG_32BIT */
diff --git a/arch/s390/include/asm/nmi.h b/arch/s390/include/asm/nmi.h
index 35f8ec18561..3027a5a72b7 100644
--- a/arch/s390/include/asm/nmi.h
+++ b/arch/s390/include/asm/nmi.h
@@ -38,7 +38,7 @@ struct mci {
 	__u32 pm :  1; /* 22 psw program mask and cc validity */
 	__u32 ia :  1; /* 23 psw instruction address validity */
 	__u32 fa :  1; /* 24 failing storage address validity */
-	__u32	 :  1; /* 25 */
+	__u32 vr :  1; /* 25 vector register validity */
 	__u32 ec :  1; /* 26 external damage code validity */
 	__u32 fp :  1; /* 27 floating point register validity */
 	__u32 gr :  1; /* 28 general register validity */
diff --git a/arch/s390/include/asm/pgtable.h b/arch/s390/include/asm/pgtable.h
index b7054356cc9..57c882761de 100644
--- a/arch/s390/include/asm/pgtable.h
+++ b/arch/s390/include/asm/pgtable.h
@@ -217,7 +217,6 @@ extern unsigned long MODULES_END;
  */
 
 /* Hardware bits in the page table entry */
-#define _PAGE_CO	0x100		/* HW Change-bit override */
 #define _PAGE_PROTECT	0x200		/* HW read-only bit  */
 #define _PAGE_INVALID	0x400		/* HW invalid bit    */
 #define _PAGE_LARGE	0x800		/* Bit to mark a large pte */
@@ -234,8 +233,8 @@ extern unsigned long MODULES_END;
 #define __HAVE_ARCH_PTE_SPECIAL
 
 /* Set of bits not changed in pte_modify */
-#define _PAGE_CHG_MASK		(PAGE_MASK | _PAGE_SPECIAL | _PAGE_CO | \
-				 _PAGE_DIRTY | _PAGE_YOUNG)
+#define _PAGE_CHG_MASK		(PAGE_MASK | _PAGE_SPECIAL | _PAGE_DIRTY | \
+				 _PAGE_YOUNG)
 
 /*
  * handle_pte_fault uses pte_present, pte_none and pte_file to find out the
@@ -354,7 +353,6 @@ extern unsigned long MODULES_END;
 
 #define _REGION3_ENTRY_LARGE	0x400	/* RTTE-format control, large page  */
 #define _REGION3_ENTRY_RO	0x200	/* page protection bit		    */
-#define _REGION3_ENTRY_CO	0x100	/* change-recording override	    */
 
 /* Bits in the segment table entry */
 #define _SEGMENT_ENTRY_BITS	0xfffffffffffffe33UL
@@ -371,7 +369,6 @@ extern unsigned long MODULES_END;
 #define _SEGMENT_ENTRY_YOUNG	0x1000	/* SW segment young bit */
 #define _SEGMENT_ENTRY_SPLIT	0x0800	/* THP splitting bit */
 #define _SEGMENT_ENTRY_LARGE	0x0400	/* STE-format control, large page */
-#define _SEGMENT_ENTRY_CO	0x0100	/* change-recording override   */
 #define _SEGMENT_ENTRY_READ	0x0002	/* SW segment read bit */
 #define _SEGMENT_ENTRY_WRITE	0x0001	/* SW segment write bit */
 
@@ -873,8 +870,6 @@ static inline void set_pte_at(struct mm_struct *mm, unsigned long addr,
 		pgste = pgste_set_pte(ptep, pgste, entry);
 		pgste_set_unlock(ptep, pgste);
 	} else {
-		if (!(pte_val(entry) & _PAGE_INVALID) && MACHINE_HAS_EDAT1)
-			pte_val(entry) |= _PAGE_CO;
 		*ptep = entry;
 	}
 }
@@ -1044,6 +1039,22 @@ static inline void __ptep_ipte_local(unsigned long address, pte_t *ptep)
 		: "=m" (*ptep) : "m" (*ptep), "a" (pto), "a" (address));
 }
 
+static inline void __ptep_ipte_range(unsigned long address, int nr, 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
+	/* Invalidate a range of ptes + global TLB flush of the ptes */
+	do {
+		asm volatile(
+			"	.insn rrf,0xb2210000,%2,%0,%1,0"
+			: "+a" (address), "+a" (nr) : "a" (pto) : "memory");
+	} while (nr != 255);
+}
+
 static inline void ptep_flush_direct(struct mm_struct *mm,
 				     unsigned long address, pte_t *ptep)
 {
diff --git a/arch/s390/include/asm/processor.h b/arch/s390/include/asm/processor.h
index e568fc8a725..d559bdb03d1 100644
--- a/arch/s390/include/asm/processor.h
+++ b/arch/s390/include/asm/processor.h
@@ -13,9 +13,11 @@
 
 #define CIF_MCCK_PENDING	0	/* machine check handling is pending */
 #define CIF_ASCE		1	/* user asce needs fixup / uaccess */
+#define CIF_NOHZ_DELAY		2	/* delay HZ disable for a tick */
 
 #define _CIF_MCCK_PENDING	(1<<CIF_MCCK_PENDING)
 #define _CIF_ASCE		(1<<CIF_ASCE)
+#define _CIF_NOHZ_DELAY		(1<<CIF_NOHZ_DELAY)
 
 
 #ifndef __ASSEMBLY__
@@ -43,6 +45,8 @@ static inline int test_cpu_flag(int flag)
 	return !!(S390_lowcore.cpu_flags & (1U << flag));
 }
 
+#define arch_needs_cpu() test_cpu_flag(CIF_NOHZ_DELAY)
+
 /*
  * Default implementation of macro that returns current
  * instruction pointer ("program counter").
@@ -113,6 +117,7 @@ struct thread_struct {
 	int ri_signum;
 #ifdef CONFIG_64BIT
 	unsigned char trap_tdb[256];	/* Transaction abort diagnose block */
+	__vector128 *vxrs;		/* Vector register save area */
 #endif
 };
 
@@ -285,7 +290,12 @@ static inline unsigned long __rewind_psw(psw_t psw, unsigned long ilc)
 	return (psw.addr - ilc) & mask;
 #endif
 }
- 
+
+/*
+ * Function to stop a processor until the next interrupt occurs
+ */
+void enabled_wait(void);
+
 /*
  * Function to drop a processor into disabled wait state
  */
diff --git a/arch/s390/include/asm/ptrace.h b/arch/s390/include/asm/ptrace.h
index 55d69dd7473..be317feff7a 100644
--- a/arch/s390/include/asm/ptrace.h
+++ b/arch/s390/include/asm/ptrace.h
@@ -161,6 +161,12 @@ static inline long regs_return_value(struct pt_regs *regs)
 	return regs->gprs[2];
 }
 
+static inline void instruction_pointer_set(struct pt_regs *regs,
+					   unsigned long val)
+{
+	regs->psw.addr = val | PSW_ADDR_AMODE;
+}
+
 int regs_query_register_offset(const char *name);
 const char *regs_query_register_name(unsigned int offset);
 unsigned long regs_get_register(struct pt_regs *regs, unsigned int offset);
diff --git a/arch/s390/include/asm/setup.h b/arch/s390/include/asm/setup.h
index 089a49814c5..7736fdd7259 100644
--- a/arch/s390/include/asm/setup.h
+++ b/arch/s390/include/asm/setup.h
@@ -55,8 +55,8 @@ extern void detect_memory_memblock(void);
 #define MACHINE_FLAG_LPP	(1UL << 13)
 #define MACHINE_FLAG_TOPOLOGY	(1UL << 14)
 #define MACHINE_FLAG_TE		(1UL << 15)
-#define MACHINE_FLAG_RRBM	(1UL << 16)
 #define MACHINE_FLAG_TLB_LC	(1UL << 17)
+#define MACHINE_FLAG_VX		(1UL << 18)
 
 #define MACHINE_IS_VM		(S390_lowcore.machine_flags & MACHINE_FLAG_VM)
 #define MACHINE_IS_KVM		(S390_lowcore.machine_flags & MACHINE_FLAG_KVM)
@@ -78,8 +78,8 @@ extern void detect_memory_memblock(void);
 #define MACHINE_HAS_LPP		(0)
 #define MACHINE_HAS_TOPOLOGY	(0)
 #define MACHINE_HAS_TE		(0)
-#define MACHINE_HAS_RRBM	(0)
 #define MACHINE_HAS_TLB_LC	(0)
+#define MACHINE_HAS_VX		(0)
 #else /* CONFIG_64BIT */
 #define MACHINE_HAS_IEEE	(1)
 #define MACHINE_HAS_CSP		(1)
@@ -91,8 +91,8 @@ extern void detect_memory_memblock(void);
 #define MACHINE_HAS_LPP		(S390_lowcore.machine_flags & MACHINE_FLAG_LPP)
 #define MACHINE_HAS_TOPOLOGY	(S390_lowcore.machine_flags & MACHINE_FLAG_TOPOLOGY)
 #define MACHINE_HAS_TE		(S390_lowcore.machine_flags & MACHINE_FLAG_TE)
-#define MACHINE_HAS_RRBM	(S390_lowcore.machine_flags & MACHINE_FLAG_RRBM)
 #define MACHINE_HAS_TLB_LC	(S390_lowcore.machine_flags & MACHINE_FLAG_TLB_LC)
+#define MACHINE_HAS_VX		(S390_lowcore.machine_flags & MACHINE_FLAG_VX)
 #endif /* CONFIG_64BIT */
 
 /*
diff --git a/arch/s390/include/asm/sigp.h b/arch/s390/include/asm/sigp.h
index bf9c823d402..49576115dbb 100644
--- a/arch/s390/include/asm/sigp.h
+++ b/arch/s390/include/asm/sigp.h
@@ -15,6 +15,7 @@
 #define SIGP_SET_ARCHITECTURE	     18
 #define SIGP_COND_EMERGENCY_SIGNAL   19
 #define SIGP_SENSE_RUNNING	     21
+#define SIGP_STORE_ADDITIONAL_STATUS 23
 
 /* SIGP condition codes */
 #define SIGP_CC_ORDER_CODE_ACCEPTED 0
@@ -33,9 +34,10 @@
 
 #ifndef __ASSEMBLY__
 
-static inline int __pcpu_sigp(u16 addr, u8 order, u32 parm, u32 *status)
+static inline int __pcpu_sigp(u16 addr, u8 order, unsigned long parm,
+			      u32 *status)
 {
-	register unsigned int reg1 asm ("1") = parm;
+	register unsigned long reg1 asm ("1") = parm;
 	int cc;
 
 	asm volatile(
diff --git a/arch/s390/include/asm/smp.h b/arch/s390/include/asm/smp.h
index 4f1307962a9..762d4f88af5 100644
--- a/arch/s390/include/asm/smp.h
+++ b/arch/s390/include/asm/smp.h
@@ -29,7 +29,6 @@ extern int smp_find_processor_id(u16 address);
 extern int smp_store_status(int cpu);
 extern int smp_vcpu_scheduled(int cpu);
 extern void smp_yield_cpu(int cpu);
-extern void smp_yield(void);
 extern void smp_cpu_set_polarization(int cpu, int val);
 extern int smp_cpu_get_polarization(int cpu);
 extern void smp_fill_possible_mask(void);
@@ -50,7 +49,6 @@ static inline int smp_find_processor_id(u16 address) { return 0; }
 static inline int smp_store_status(int cpu) { return 0; }
 static inline int smp_vcpu_scheduled(int cpu) { return 1; }
 static inline void smp_yield_cpu(int cpu) { }
-static inline void smp_yield(void) { }
 static inline void smp_fill_possible_mask(void) { }
 
 #endif /* CONFIG_SMP */
diff --git a/arch/s390/include/asm/spinlock.h b/arch/s390/include/asm/spinlock.h
index 96879f7ad6d..d6bdf906caa 100644
--- a/arch/s390/include/asm/spinlock.h
+++ b/arch/s390/include/asm/spinlock.h
@@ -37,11 +37,17 @@ _raw_compare_and_swap(unsigned int *lock, unsigned int old, unsigned int new)
  * (the type definitions are in asm/spinlock_types.h)
  */
 
+void arch_lock_relax(unsigned int cpu);
+
 void arch_spin_lock_wait(arch_spinlock_t *);
 int arch_spin_trylock_retry(arch_spinlock_t *);
-void arch_spin_relax(arch_spinlock_t *);
 void arch_spin_lock_wait_flags(arch_spinlock_t *, unsigned long flags);
 
+static inline void arch_spin_relax(arch_spinlock_t *lock)
+{
+	arch_lock_relax(lock->lock);
+}
+
 static inline u32 arch_spin_lockval(int cpu)
 {
 	return ~cpu;
@@ -64,11 +70,6 @@ static inline int arch_spin_trylock_once(arch_spinlock_t *lp)
 		      _raw_compare_and_swap(&lp->lock, 0, SPINLOCK_LOCKVAL));
 }
 
-static inline int arch_spin_tryrelease_once(arch_spinlock_t *lp)
-{
-	return _raw_compare_and_swap(&lp->lock, SPINLOCK_LOCKVAL, 0);
-}
-
 static inline void arch_spin_lock(arch_spinlock_t *lp)
 {
 	if (!arch_spin_trylock_once(lp))
@@ -91,7 +92,13 @@ static inline int arch_spin_trylock(arch_spinlock_t *lp)
 
 static inline void arch_spin_unlock(arch_spinlock_t *lp)
 {
-	arch_spin_tryrelease_once(lp);
+	typecheck(unsigned int, lp->lock);
+	asm volatile(
+		__ASM_BARRIER
+		"st	%1,%0\n"
+		: "+Q" (lp->lock)
+		: "d" (0)
+		: "cc", "memory");
 }
 
 static inline void arch_spin_unlock_wait(arch_spinlock_t *lock)
@@ -123,13 +130,12 @@ static inline void arch_spin_unlock_wait(arch_spinlock_t *lock)
  */
 #define arch_write_can_lock(x) ((x)->lock == 0)
 
-extern void _raw_read_lock_wait(arch_rwlock_t *lp);
-extern void _raw_read_lock_wait_flags(arch_rwlock_t *lp, unsigned long flags);
 extern int _raw_read_trylock_retry(arch_rwlock_t *lp);
-extern void _raw_write_lock_wait(arch_rwlock_t *lp);
-extern void _raw_write_lock_wait_flags(arch_rwlock_t *lp, unsigned long flags);
 extern int _raw_write_trylock_retry(arch_rwlock_t *lp);
 
+#define arch_read_lock_flags(lock, flags) arch_read_lock(lock)
+#define arch_write_lock_flags(lock, flags) arch_write_lock(lock)
+
 static inline int arch_read_trylock_once(arch_rwlock_t *rw)
 {
 	unsigned int old = ACCESS_ONCE(rw->lock);
@@ -144,16 +150,82 @@ static inline int arch_write_trylock_once(arch_rwlock_t *rw)
 		      _raw_compare_and_swap(&rw->lock, 0, 0x80000000));
 }
 
+#ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
+
+#define __RAW_OP_OR	"lao"
+#define __RAW_OP_AND	"lan"
+#define __RAW_OP_ADD	"laa"
+
+#define __RAW_LOCK(ptr, op_val, op_string)		\
+({							\
+	unsigned int old_val;				\
+							\
+	typecheck(unsigned int *, ptr);			\
+	asm volatile(					\
+		op_string "	%0,%2,%1\n"		\
+		"bcr	14,0\n"				\
+		: "=d" (old_val), "+Q" (*ptr)		\
+		: "d" (op_val)				\
+		: "cc", "memory");			\
+	old_val;					\
+})
+
+#define __RAW_UNLOCK(ptr, op_val, op_string)		\
+({							\
+	unsigned int old_val;				\
+							\
+	typecheck(unsigned int *, ptr);			\
+	asm volatile(					\
+		"bcr	14,0\n"				\
+		op_string "	%0,%2,%1\n"		\
+		: "=d" (old_val), "+Q" (*ptr)		\
+		: "d" (op_val)				\
+		: "cc", "memory");			\
+	old_val;					\
+})
+
+extern void _raw_read_lock_wait(arch_rwlock_t *lp);
+extern void _raw_write_lock_wait(arch_rwlock_t *lp, unsigned int prev);
+
 static inline void arch_read_lock(arch_rwlock_t *rw)
 {
-	if (!arch_read_trylock_once(rw))
+	unsigned int old;
+
+	old = __RAW_LOCK(&rw->lock, 1, __RAW_OP_ADD);
+	if ((int) old < 0)
 		_raw_read_lock_wait(rw);
 }
 
-static inline void arch_read_lock_flags(arch_rwlock_t *rw, unsigned long flags)
+static inline void arch_read_unlock(arch_rwlock_t *rw)
+{
+	__RAW_UNLOCK(&rw->lock, -1, __RAW_OP_ADD);
+}
+
+static inline void arch_write_lock(arch_rwlock_t *rw)
+{
+	unsigned int old;
+
+	old = __RAW_LOCK(&rw->lock, 0x80000000, __RAW_OP_OR);
+	if (old != 0)
+		_raw_write_lock_wait(rw, old);
+	rw->owner = SPINLOCK_LOCKVAL;
+}
+
+static inline void arch_write_unlock(arch_rwlock_t *rw)
+{
+	rw->owner = 0;
+	__RAW_UNLOCK(&rw->lock, 0x7fffffff, __RAW_OP_AND);
+}
+
+#else /* CONFIG_HAVE_MARCH_Z196_FEATURES */
+
+extern void _raw_read_lock_wait(arch_rwlock_t *lp);
+extern void _raw_write_lock_wait(arch_rwlock_t *lp);
+
+static inline void arch_read_lock(arch_rwlock_t *rw)
 {
 	if (!arch_read_trylock_once(rw))
-		_raw_read_lock_wait_flags(rw, flags);
+		_raw_read_lock_wait(rw);
 }
 
 static inline void arch_read_unlock(arch_rwlock_t *rw)
@@ -169,19 +241,24 @@ static inline void arch_write_lock(arch_rwlock_t *rw)
 {
 	if (!arch_write_trylock_once(rw))
 		_raw_write_lock_wait(rw);
-}
-
-static inline void arch_write_lock_flags(arch_rwlock_t *rw, unsigned long flags)
-{
-	if (!arch_write_trylock_once(rw))
-		_raw_write_lock_wait_flags(rw, flags);
+	rw->owner = SPINLOCK_LOCKVAL;
 }
 
 static inline void arch_write_unlock(arch_rwlock_t *rw)
 {
-	_raw_compare_and_swap(&rw->lock, 0x80000000, 0);
+	typecheck(unsigned int, rw->lock);
+
+	rw->owner = 0;
+	asm volatile(
+		__ASM_BARRIER
+		"st	%1,%0\n"
+		: "+Q" (rw->lock)
+		: "d" (0)
+		: "cc", "memory");
 }
 
+#endif /* CONFIG_HAVE_MARCH_Z196_FEATURES */
+
 static inline int arch_read_trylock(arch_rwlock_t *rw)
 {
 	if (!arch_read_trylock_once(rw))
@@ -191,12 +268,20 @@ static inline int arch_read_trylock(arch_rwlock_t *rw)
 
 static inline int arch_write_trylock(arch_rwlock_t *rw)
 {
-	if (!arch_write_trylock_once(rw))
-		return _raw_write_trylock_retry(rw);
+	if (!arch_write_trylock_once(rw) && !_raw_write_trylock_retry(rw))
+		return 0;
+	rw->owner = SPINLOCK_LOCKVAL;
 	return 1;
 }
 
-#define arch_read_relax(lock)	cpu_relax()
-#define arch_write_relax(lock)	cpu_relax()
+static inline void arch_read_relax(arch_rwlock_t *rw)
+{
+	arch_lock_relax(rw->owner);
+}
+
+static inline void arch_write_relax(arch_rwlock_t *rw)
+{
+	arch_lock_relax(rw->owner);
+}
 
 #endif /* __ASM_SPINLOCK_H */
diff --git a/arch/s390/include/asm/spinlock_types.h b/arch/s390/include/asm/spinlock_types.h
index b2cd6ff7c2c..d84b6939237 100644
--- a/arch/s390/include/asm/spinlock_types.h
+++ b/arch/s390/include/asm/spinlock_types.h
@@ -13,6 +13,7 @@ typedef struct {
 
 typedef struct {
 	unsigned int lock;
+	unsigned int owner;
 } arch_rwlock_t;
 
 #define __ARCH_RW_LOCK_UNLOCKED		{ 0 }
diff --git a/arch/s390/include/asm/switch_to.h b/arch/s390/include/asm/switch_to.h
index 18ea9e3f814..2542a7e4c8b 100644
--- a/arch/s390/include/asm/switch_to.h
+++ b/arch/s390/include/asm/switch_to.h
@@ -103,6 +103,61 @@ static inline void restore_fp_regs(freg_t *fprs)
 	asm volatile("ld 15,%0" : : "Q" (fprs[15]));
 }
 
+static inline void save_vx_regs(__vector128 *vxrs)
+{
+	typedef struct { __vector128 _[__NUM_VXRS]; } addrtype;
+
+	asm volatile(
+		"	la	1,%0\n"
+		"	.word	0xe70f,0x1000,0x003e\n"	/* vstm 0,15,0(1) */
+		"	.word	0xe70f,0x1100,0x0c3e\n"	/* vstm 16,31,256(1) */
+		: "=Q" (*(addrtype *) vxrs) : : "1");
+}
+
+static inline void save_vx_regs_safe(__vector128 *vxrs)
+{
+	unsigned long cr0, flags;
+
+	flags = arch_local_irq_save();
+	__ctl_store(cr0, 0, 0);
+	__ctl_set_bit(0, 17);
+	__ctl_set_bit(0, 18);
+	save_vx_regs(vxrs);
+	__ctl_load(cr0, 0, 0);
+	arch_local_irq_restore(flags);
+}
+
+static inline void restore_vx_regs(__vector128 *vxrs)
+{
+	typedef struct { __vector128 _[__NUM_VXRS]; } addrtype;
+
+	asm volatile(
+		"	la	1,%0\n"
+		"	.word	0xe70f,0x1000,0x0036\n"	/* vlm 0,15,0(1) */
+		"	.word	0xe70f,0x1100,0x0c36\n"	/* vlm 16,31,256(1) */
+		: : "Q" (*(addrtype *) vxrs) : "1");
+}
+
+static inline void save_fp_vx_regs(struct task_struct *task)
+{
+#ifdef CONFIG_64BIT
+	if (task->thread.vxrs)
+		save_vx_regs(task->thread.vxrs);
+	else
+#endif
+	save_fp_regs(task->thread.fp_regs.fprs);
+}
+
+static inline void restore_fp_vx_regs(struct task_struct *task)
+{
+#ifdef CONFIG_64BIT
+	if (task->thread.vxrs)
+		restore_vx_regs(task->thread.vxrs);
+	else
+#endif
+	restore_fp_regs(task->thread.fp_regs.fprs);
+}
+
 static inline void save_access_regs(unsigned int *acrs)
 {
 	typedef struct { int _[NUM_ACRS]; } acrstype;
@@ -120,16 +175,16 @@ static inline void restore_access_regs(unsigned int *acrs)
 #define switch_to(prev,next,last) do {					\
 	if (prev->mm) {							\
 		save_fp_ctl(&prev->thread.fp_regs.fpc);			\
-		save_fp_regs(prev->thread.fp_regs.fprs);		\
+		save_fp_vx_regs(prev);					\
 		save_access_regs(&prev->thread.acrs[0]);		\
 		save_ri_cb(prev->thread.ri_cb);				\
 	}								\
 	if (next->mm) {							\
+		update_cr_regs(next);					\
 		restore_fp_ctl(&next->thread.fp_regs.fpc);		\
-		restore_fp_regs(next->thread.fp_regs.fprs);		\
+		restore_fp_vx_regs(next);				\
 		restore_access_regs(&next->thread.acrs[0]);		\
 		restore_ri_cb(next->thread.ri_cb, prev->thread.ri_cb);	\
-		update_cr_regs(next);					\
 	}								\
 	prev = __switch_to(prev,next);					\
 } while (0)
diff --git a/arch/s390/include/asm/thread_info.h b/arch/s390/include/asm/thread_info.h
index b833e9c0bfb..4d62fd5b56e 100644
--- a/arch/s390/include/asm/thread_info.h
+++ b/arch/s390/include/asm/thread_info.h
@@ -84,11 +84,13 @@ static inline struct thread_info *current_thread_info(void)
 #define TIF_SYSCALL_AUDIT	4	/* syscall auditing active */
 #define TIF_SECCOMP		5	/* secure computing */
 #define TIF_SYSCALL_TRACEPOINT	6	/* syscall tracepoint instrumentation */
+#define TIF_UPROBE		7	/* breakpointed or single-stepping */
 #define TIF_31BIT		16	/* 32bit process */
 #define TIF_MEMDIE		17	/* is terminating due to OOM killer */
 #define TIF_RESTORE_SIGMASK	18	/* restore signal mask in do_signal() */
 #define TIF_SINGLE_STEP		19	/* This task is single stepped */
 #define TIF_BLOCK_STEP		20	/* This task is block stepped */
+#define TIF_UPROBE_SINGLESTEP	21	/* This task is uprobe single stepped */
 
 #define _TIF_NOTIFY_RESUME	(1<<TIF_NOTIFY_RESUME)
 #define _TIF_SIGPENDING		(1<<TIF_SIGPENDING)
@@ -97,6 +99,7 @@ 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_UPROBE		(1<<TIF_UPROBE)
 #define _TIF_31BIT		(1<<TIF_31BIT)
 #define _TIF_SINGLE_STEP	(1<<TIF_SINGLE_STEP)
 
diff --git a/arch/s390/include/asm/uprobes.h b/arch/s390/include/asm/uprobes.h
new file mode 100644
index 00000000000..1411dff7fea
--- /dev/null
+++ b/arch/s390/include/asm/uprobes.h
@@ -0,0 +1,42 @@
+/*
+ *    User-space Probes (UProbes) for s390
+ *
+ *    Copyright IBM Corp. 2014
+ *    Author(s): Jan Willeke,
+ */
+
+#ifndef _ASM_UPROBES_H
+#define _ASM_UPROBES_H
+
+#include <linux/notifier.h>
+
+typedef u16 uprobe_opcode_t;
+
+#define UPROBE_XOL_SLOT_BYTES	256 /* cache aligned */
+
+#define UPROBE_SWBP_INSN	0x0002
+#define UPROBE_SWBP_INSN_SIZE	2
+
+struct arch_uprobe {
+	union{
+		uprobe_opcode_t insn[3];
+		uprobe_opcode_t ixol[3];
+	};
+	unsigned int saved_per : 1;
+	unsigned int saved_int_code;
+};
+
+struct arch_uprobe_task {
+};
+
+int arch_uprobe_analyze_insn(struct arch_uprobe *aup, struct mm_struct *mm,
+			     unsigned long addr);
+int arch_uprobe_pre_xol(struct arch_uprobe *aup, struct pt_regs *regs);
+int arch_uprobe_post_xol(struct arch_uprobe *aup, struct pt_regs *regs);
+bool arch_uprobe_xol_was_trapped(struct task_struct *tsk);
+int arch_uprobe_exception_notify(struct notifier_block *self, unsigned long val,
+				 void *data);
+void arch_uprobe_abort_xol(struct arch_uprobe *ap, struct pt_regs *regs);
+unsigned long arch_uretprobe_hijack_return_addr(unsigned long trampoline,
+						struct pt_regs *regs);
+#endif	/* _ASM_UPROBES_H */
diff --git a/arch/s390/include/asm/vdso.h b/arch/s390/include/asm/vdso.h
index bc9746a7d47..a62526d0920 100644
--- a/arch/s390/include/asm/vdso.h
+++ b/arch/s390/include/asm/vdso.h
@@ -22,13 +22,17 @@ struct vdso_data {
 	__u64 xtime_tod_stamp;		/* TOD clock for xtime		0x08 */
 	__u64 xtime_clock_sec;		/* Kernel time			0x10 */
 	__u64 xtime_clock_nsec;		/*				0x18 */
-	__u64 wtom_clock_sec;		/* Wall to monotonic clock	0x20 */
-	__u64 wtom_clock_nsec;		/*				0x28 */
-	__u32 tz_minuteswest;		/* Minutes west of Greenwich	0x30 */
-	__u32 tz_dsttime;		/* Type of dst correction	0x34 */
-	__u32 ectg_available;		/* ECTG instruction present	0x38 */
-	__u32 tk_mult;			/* Mult. used for xtime_nsec	0x3c */
-	__u32 tk_shift;			/* Shift used for xtime_nsec	0x40 */
+	__u64 xtime_coarse_sec;		/* Coarse kernel time		0x20 */
+	__u64 xtime_coarse_nsec;	/*				0x28 */
+	__u64 wtom_clock_sec;		/* Wall to monotonic clock	0x30 */
+	__u64 wtom_clock_nsec;		/*				0x38 */
+	__u64 wtom_coarse_sec;		/* Coarse wall to monotonic	0x40 */
+	__u64 wtom_coarse_nsec;		/*				0x48 */
+	__u32 tz_minuteswest;		/* Minutes west of Greenwich	0x50 */
+	__u32 tz_dsttime;		/* Type of dst correction	0x54 */
+	__u32 ectg_available;		/* ECTG instruction present	0x58 */
+	__u32 tk_mult;			/* Mult. used for xtime_nsec	0x5c */
+	__u32 tk_shift;			/* Shift used for xtime_nsec	0x60 */
 };
 
 struct vdso_per_cpu_data {
diff --git a/arch/s390/include/asm/vtimer.h b/arch/s390/include/asm/vtimer.h
index bfe25d513ad..10a179af62d 100644
--- a/arch/s390/include/asm/vtimer.h
+++ b/arch/s390/include/asm/vtimer.h
@@ -28,6 +28,4 @@ extern int del_virt_timer(struct vtimer_list *timer);
 extern void init_cpu_vtimer(void);
 extern void vtime_init(void);
 
-extern void vtime_stop_cpu(void);
-
 #endif /* _ASM_S390_TIMER_H */
diff --git a/arch/s390/include/uapi/asm/sigcontext.h b/arch/s390/include/uapi/asm/sigcontext.h
index b30de9c01bb..5f0b8d7ddb0 100644
--- a/arch/s390/include/uapi/asm/sigcontext.h
+++ b/arch/s390/include/uapi/asm/sigcontext.h
@@ -7,10 +7,14 @@
 #define _ASM_S390_SIGCONTEXT_H
 
 #include <linux/compiler.h>
+#include <linux/types.h>
 
-#define __NUM_GPRS 16
-#define __NUM_FPRS 16
-#define __NUM_ACRS 16
+#define __NUM_GPRS		16
+#define __NUM_FPRS		16
+#define __NUM_ACRS		16
+#define __NUM_VXRS		32
+#define __NUM_VXRS_LOW		16
+#define __NUM_VXRS_HIGH		16
 
 #ifndef __s390x__
 
@@ -59,6 +63,16 @@ typedef struct
 	_s390_fp_regs     fpregs;
 } _sigregs;
 
+typedef struct
+{
+#ifndef __s390x__
+	unsigned long gprs_high[__NUM_GPRS];
+#endif
+	unsigned long long vxrs_low[__NUM_VXRS_LOW];
+	__vector128 vxrs_high[__NUM_VXRS_HIGH];
+	unsigned char __reserved[128];
+} _sigregs_ext;
+
 struct sigcontext
 {
 	unsigned long	oldmask[_SIGCONTEXT_NSIG_WORDS];
diff --git a/arch/s390/include/uapi/asm/types.h b/arch/s390/include/uapi/asm/types.h
index 038f2b9178a..3c3951e3415 100644
--- a/arch/s390/include/uapi/asm/types.h
+++ b/arch/s390/include/uapi/asm/types.h
@@ -17,6 +17,10 @@
 typedef unsigned long addr_t; 
 typedef __signed__ long saddr_t;
 
+typedef struct {
+	__u32 u[4];
+} __vector128;
+
 #endif /* __ASSEMBLY__ */
 
 #endif /* _UAPI_S390_TYPES_H */
diff --git a/arch/s390/include/uapi/asm/ucontext.h b/arch/s390/include/uapi/asm/ucontext.h
index 3e077b2a470..64a69aa5dde 100644
--- a/arch/s390/include/uapi/asm/ucontext.h
+++ b/arch/s390/include/uapi/asm/ucontext.h
@@ -7,10 +7,15 @@
 #ifndef _ASM_S390_UCONTEXT_H
 #define _ASM_S390_UCONTEXT_H
 
-#define UC_EXTENDED	0x00000001
-
-#ifndef __s390x__
+#define UC_GPRS_HIGH	1	/* uc_mcontext_ext has valid high gprs */
+#define UC_VXRS		2	/* uc_mcontext_ext has valid vector regs */
 
+/*
+ * The struct ucontext_extended describes how the registers are stored
+ * on a rt signal frame. Please note that the structure is not fixed,
+ * if new CPU registers are added to the user state the size of the
+ * struct ucontext_extended will increase.
+ */
 struct ucontext_extended {
 	unsigned long	  uc_flags;
 	struct ucontext  *uc_link;
@@ -19,11 +24,9 @@ struct ucontext_extended {
 	sigset_t	  uc_sigmask;
 	/* Allow for uc_sigmask growth.  Glibc uses a 1024-bit sigset_t.  */
 	unsigned char	  __unused[128 - sizeof(sigset_t)];
-	unsigned long	  uc_gprs_high[16];
+	_sigregs_ext	  uc_mcontext_ext;
 };
 
-#endif
-
 struct ucontext {
 	unsigned long	  uc_flags;
 	struct ucontext  *uc_link;