Merge branch 'perf/core' into perf/uprobes

Merge in latest upstream (and the latest perf development tree),
to prepare for tooling changes, and also to pick up v3.4 MM
changes that the uprobes code needs to take care of.

Signed-off-by: Ingo Molnar <mingo@kernel.org>

23 files changed, 242 insertions, 201 deletions

diff --git a/arch/tile/include/arch/spr_def.h b/arch/tile/include/arch/spr_def.h
index f548efeb2de3..d6ba449b5363 100644
--- a/arch/tile/include/arch/spr_def.h
+++ b/arch/tile/include/arch/spr_def.h
@@ -60,8 +60,8 @@
 	_concat4(SPR_IPI_EVENT_, CONFIG_KERNEL_PL,,)
 #define SPR_IPI_EVENT_RESET_K \
 	_concat4(SPR_IPI_EVENT_RESET_, CONFIG_KERNEL_PL,,)
-#define SPR_IPI_MASK_SET_K \
-	_concat4(SPR_IPI_MASK_SET_, CONFIG_KERNEL_PL,,)
+#define SPR_IPI_EVENT_SET_K \
+	_concat4(SPR_IPI_EVENT_SET_, CONFIG_KERNEL_PL,,)
 #define INT_IPI_K \
 	_concat4(INT_IPI_, CONFIG_KERNEL_PL,,)
 
diff --git a/arch/tile/include/asm/atomic.h b/arch/tile/include/asm/atomic.h
index 921dbeb8a70c..f2461429a4a4 100644
--- a/arch/tile/include/asm/atomic.h
+++ b/arch/tile/include/asm/atomic.h
@@ -17,10 +17,12 @@
 #ifndef _ASM_TILE_ATOMIC_H
 #define _ASM_TILE_ATOMIC_H
 
+#include <asm/cmpxchg.h>
+
 #ifndef __ASSEMBLY__
 
 #include <linux/compiler.h>
-#include <asm/system.h>
+#include <linux/types.h>
 
 #define ATOMIC_INIT(i)	{ (i) }
 
@@ -121,54 +123,6 @@ static inline int atomic_read(const atomic_t *v)
  */
 #define atomic_add_negative(i, v)	(atomic_add_return((i), (v)) < 0)
 
-/* Nonexistent functions intended to cause link errors. */
-extern unsigned long __xchg_called_with_bad_pointer(void);
-extern unsigned long __cmpxchg_called_with_bad_pointer(void);
-
-#define xchg(ptr, x)							\
-	({								\
-		typeof(*(ptr)) __x;					\
-		switch (sizeof(*(ptr))) {				\
-		case 4:							\
-			__x = (typeof(__x))(typeof(__x-__x))atomic_xchg( \
-				(atomic_t *)(ptr),			\
-				(u32)(typeof((x)-(x)))(x));		\
-			break;						\
-		case 8:							\
-			__x = (typeof(__x))(typeof(__x-__x))atomic64_xchg( \
-				(atomic64_t *)(ptr),			\
-				(u64)(typeof((x)-(x)))(x));		\
-			break;						\
-		default:						\
-			__xchg_called_with_bad_pointer();		\
-		}							\
-		__x;							\
-	})
-
-#define cmpxchg(ptr, o, n)						\
-	({								\
-		typeof(*(ptr)) __x;					\
-		switch (sizeof(*(ptr))) {				\
-		case 4:							\
-			__x = (typeof(__x))(typeof(__x-__x))atomic_cmpxchg( \
-				(atomic_t *)(ptr),			\
-				(u32)(typeof((o)-(o)))(o),		\
-				(u32)(typeof((n)-(n)))(n));		\
-			break;						\
-		case 8:							\
-			__x = (typeof(__x))(typeof(__x-__x))atomic64_cmpxchg( \
-				(atomic64_t *)(ptr),			\
-				(u64)(typeof((o)-(o)))(o),		\
-				(u64)(typeof((n)-(n)))(n));		\
-			break;						\
-		default:						\
-			__cmpxchg_called_with_bad_pointer();		\
-		}							\
-		__x;							\
-	})
-
-#define tas(ptr) (xchg((ptr), 1))
-
 #endif /* __ASSEMBLY__ */
 
 #ifndef __tilegx__
diff --git a/arch/tile/include/asm/atomic_32.h b/arch/tile/include/asm/atomic_32.h
index c03349e0ca9f..54d1da826f93 100644
--- a/arch/tile/include/asm/atomic_32.h
+++ b/arch/tile/include/asm/atomic_32.h
@@ -17,6 +17,7 @@
 #ifndef _ASM_TILE_ATOMIC_32_H
 #define _ASM_TILE_ATOMIC_32_H
 
+#include <asm/barrier.h>
 #include <arch/chip.h>
 
 #ifndef __ASSEMBLY__
@@ -199,7 +200,7 @@ static inline u64 atomic64_add_return(u64 i, atomic64_t *v)
  * @u: ...unless v is equal to u.
  *
  * Atomically adds @a to @v, so long as @v was not already @u.
- * Returns the old value of @v.
+ * Returns non-zero if @v was not @u, and zero otherwise.
  */
 static inline u64 atomic64_add_unless(atomic64_t *v, u64 a, u64 u)
 {
diff --git a/arch/tile/include/asm/atomic_64.h b/arch/tile/include/asm/atomic_64.h
index 27fe667fddfe..f4500c688ffa 100644
--- a/arch/tile/include/asm/atomic_64.h
+++ b/arch/tile/include/asm/atomic_64.h
@@ -19,6 +19,7 @@
 
 #ifndef __ASSEMBLY__
 
+#include <asm/barrier.h>
 #include <arch/spr_def.h>
 
 /* First, the 32-bit atomic ops that are "real" on our 64-bit platform. */
diff --git a/arch/tile/include/asm/system.h b/arch/tile/include/asm/barrier.h
index 23d1842f4839..990a217a0b72 100644
--- a/arch/tile/include/asm/system.h
+++ b/arch/tile/include/asm/barrier.h
@@ -12,20 +12,15 @@
  *   more details.
  */
 
-#ifndef _ASM_TILE_SYSTEM_H
-#define _ASM_TILE_SYSTEM_H
+#ifndef _ASM_TILE_BARRIER_H
+#define _ASM_TILE_BARRIER_H
 
 #ifndef __ASSEMBLY__
 
 #include <linux/types.h>
-#include <linux/irqflags.h>
-
-/* NOTE: we can't include <linux/ptrace.h> due to #include dependencies. */
-#include <asm/ptrace.h>
-
 #include <arch/chip.h>
-#include <arch/sim_def.h>
 #include <arch/spr_def.h>
+#include <asm/timex.h>
 
 /*
  * read_barrier_depends - Flush all pending reads that subsequents reads
@@ -78,17 +73,10 @@
  * as Alpha, "y" could be set to 3 and "x" to 0.  Use rmb()
  * in cases like this where there are no data dependencies.
  */
-
 #define read_barrier_depends()	do { } while (0)
 
 #define __sync()	__insn_mf()
 
-#if CHIP_HAS_SPLIT_CYCLE()
-#define get_cycles_low() __insn_mfspr(SPR_CYCLE_LOW)
-#else
-#define get_cycles_low() __insn_mfspr(SPR_CYCLE)   /* just get all 64 bits */
-#endif
-
 #if !CHIP_HAS_MF_WAITS_FOR_VICTIMS()
 #include <hv/syscall_public.h>
 /*
@@ -156,106 +144,5 @@ mb_incoherent(void)
 #define set_mb(var, value) \
 	do { var = value; mb(); } while (0)
 
-/*
- * Pause the DMA engine and static network before task switching.
- */
-#define prepare_arch_switch(next) _prepare_arch_switch(next)
-void _prepare_arch_switch(struct task_struct *next);
-
-
-/*
- * switch_to(n) should switch tasks to task nr n, first
- * checking that n isn't the current task, in which case it does nothing.
- * The number of callee-saved registers saved on the kernel stack
- * is defined here for use in copy_thread() and must agree with __switch_to().
- */
-#endif /* !__ASSEMBLY__ */
-#define CALLEE_SAVED_FIRST_REG 30
-#define CALLEE_SAVED_REGS_COUNT 24   /* r30 to r52, plus an empty to align */
-#ifndef __ASSEMBLY__
-struct task_struct;
-#define switch_to(prev, next, last) ((last) = _switch_to((prev), (next)))
-extern struct task_struct *_switch_to(struct task_struct *prev,
-				      struct task_struct *next);
-
-/* Helper function for _switch_to(). */
-extern struct task_struct *__switch_to(struct task_struct *prev,
-				       struct task_struct *next,
-				       unsigned long new_system_save_k_0);
-
-/* Address that switched-away from tasks are at. */
-extern unsigned long get_switch_to_pc(void);
-
-/*
- * On SMP systems, when the scheduler does migration-cost autodetection,
- * it needs a way to flush as much of the CPU's caches as possible:
- *
- * TODO: fill this in!
- */
-static inline void sched_cacheflush(void)
-{
-}
-
-#define arch_align_stack(x) (x)
-
-/*
- * Is the kernel doing fixups of unaligned accesses?  If <0, no kernel
- * intervention occurs and SIGBUS is delivered with no data address
- * info.  If 0, the kernel single-steps the instruction to discover
- * the data address to provide with the SIGBUS.  If 1, the kernel does
- * a fixup.
- */
-extern int unaligned_fixup;
-
-/* Is the kernel printing on each unaligned fixup? */
-extern int unaligned_printk;
-
-/* Number of unaligned fixups performed */
-extern unsigned int unaligned_fixup_count;
-
-/* Init-time routine to do tile-specific per-cpu setup. */
-void setup_cpu(int boot);
-
-/* User-level DMA management functions */
-void grant_dma_mpls(void);
-void restrict_dma_mpls(void);
-
-#ifdef CONFIG_HARDWALL
-/* User-level network management functions */
-void reset_network_state(void);
-void grant_network_mpls(void);
-void restrict_network_mpls(void);
-int hardwall_deactivate(struct task_struct *task);
-
-/* Hook hardwall code into changes in affinity. */
-#define arch_set_cpus_allowed(p, new_mask) do { \
-	if (p->thread.hardwall && !cpumask_equal(&p->cpus_allowed, new_mask)) \
-		hardwall_deactivate(p); \
-} while (0)
-#endif
-
-/*
- * Kernel threads can check to see if they need to migrate their
- * stack whenever they return from a context switch; for user
- * threads, we defer until they are returning to user-space.
- */
-#define finish_arch_switch(prev) do {                                     \
-	if (unlikely((prev)->state == TASK_DEAD))                         \
-		__insn_mtspr(SPR_SIM_CONTROL, SIM_CONTROL_OS_EXIT |       \
-			((prev)->pid << _SIM_CONTROL_OPERATOR_BITS));     \
-	__insn_mtspr(SPR_SIM_CONTROL, SIM_CONTROL_OS_SWITCH |             \
-		(current->pid << _SIM_CONTROL_OPERATOR_BITS));            \
-	if (current->mm == NULL && !kstack_hash &&                        \
-	    current_thread_info()->homecache_cpu != smp_processor_id())   \
-		homecache_migrate_kthread();                              \
-} while (0)
-
-/* Support function for forking a new task. */
-void ret_from_fork(void);
-
-/* Called from ret_from_fork() when a new process starts up. */
-struct task_struct *sim_notify_fork(struct task_struct *prev);
-
 #endif /* !__ASSEMBLY__ */
-
-#endif /* _ASM_TILE_SYSTEM_H */
+#endif /* _ASM_TILE_BARRIER_H */
diff --git a/arch/tile/include/asm/bitops_32.h b/arch/tile/include/asm/bitops_32.h
index 571b118bfd9b..ddc4c1efde43 100644
--- a/arch/tile/include/asm/bitops_32.h
+++ b/arch/tile/include/asm/bitops_32.h
@@ -17,7 +17,6 @@
 
 #include <linux/compiler.h>
 #include <linux/atomic.h>
-#include <asm/system.h>
 
 /* Tile-specific routines to support <asm/bitops.h>. */
 unsigned long _atomic_or(volatile unsigned long *p, unsigned long mask);
diff --git a/arch/tile/include/asm/bitops_64.h b/arch/tile/include/asm/bitops_64.h
index e9c8e381ee0e..60b87ee54fb8 100644
--- a/arch/tile/include/asm/bitops_64.h
+++ b/arch/tile/include/asm/bitops_64.h
@@ -17,7 +17,6 @@
 
 #include <linux/compiler.h>
 #include <linux/atomic.h>
-#include <asm/system.h>
 
 /* See <asm/bitops.h> for API comments. */
 
@@ -39,10 +38,10 @@ static inline void clear_bit(unsigned nr, volatile unsigned long *addr)
 
 static inline void change_bit(unsigned nr, volatile unsigned long *addr)
 {
-	unsigned long old, mask = (1UL << (nr % BITS_PER_LONG));
-	long guess, oldval;
+	unsigned long mask = (1UL << (nr % BITS_PER_LONG));
+	unsigned long guess, oldval;
 	addr += nr / BITS_PER_LONG;
-	old = *addr;
+	oldval = *addr;
 	do {
 		guess = oldval;
 		oldval = atomic64_cmpxchg((atomic64_t *)addr,
@@ -86,7 +85,7 @@ static inline int test_and_change_bit(unsigned nr,
 				      volatile unsigned long *addr)
 {
 	unsigned long mask = (1UL << (nr % BITS_PER_LONG));
-	long guess, oldval = *addr;
+	unsigned long guess, oldval;
 	addr += nr / BITS_PER_LONG;
 	oldval = *addr;
 	do {
diff --git a/arch/tile/include/asm/cacheflush.h b/arch/tile/include/asm/cacheflush.h
index e925f4bb498f..0fc63c488edf 100644
--- a/arch/tile/include/asm/cacheflush.h
+++ b/arch/tile/include/asm/cacheflush.h
@@ -20,7 +20,6 @@
 /* Keep includes the same across arches.  */
 #include <linux/mm.h>
 #include <linux/cache.h>
-#include <asm/system.h>
 #include <arch/icache.h>
 
 /* Caches are physically-indexed and so don't need special treatment */
@@ -152,4 +151,14 @@ static inline void finv_buffer_local(void *buffer, size_t size)
  */
 void finv_buffer_remote(void *buffer, size_t size, int hfh);
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible:
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 #endif /* _ASM_TILE_CACHEFLUSH_H */
diff --git a/arch/tile/include/asm/cmpxchg.h b/arch/tile/include/asm/cmpxchg.h
new file mode 100644
index 000000000000..276f067e3640
--- /dev/null
+++ b/arch/tile/include/asm/cmpxchg.h
@@ -0,0 +1,73 @@
+/*
+ * cmpxchg.h -- forked from asm/atomic.h with this copyright:
+ *
+ * Copyright 2010 Tilera Corporation. All Rights Reserved.
+ *
+ *   This program is free software; you can redistribute it and/or
+ *   modify it under the terms of the GNU General Public License
+ *   as published by the Free Software Foundation, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ *
+ */
+
+#ifndef _ASM_TILE_CMPXCHG_H
+#define _ASM_TILE_CMPXCHG_H
+
+#ifndef __ASSEMBLY__
+
+/* Nonexistent functions intended to cause link errors. */
+extern unsigned long __xchg_called_with_bad_pointer(void);
+extern unsigned long __cmpxchg_called_with_bad_pointer(void);
+
+#define xchg(ptr, x)							\
+	({								\
+		typeof(*(ptr)) __x;					\
+		switch (sizeof(*(ptr))) {				\
+		case 4:							\
+			__x = (typeof(__x))(typeof(__x-__x))atomic_xchg( \
+				(atomic_t *)(ptr),			\
+				(u32)(typeof((x)-(x)))(x));		\
+			break;						\
+		case 8:							\
+			__x = (typeof(__x))(typeof(__x-__x))atomic64_xchg( \
+				(atomic64_t *)(ptr),			\
+				(u64)(typeof((x)-(x)))(x));		\
+			break;						\
+		default:						\
+			__xchg_called_with_bad_pointer();		\
+		}							\
+		__x;							\
+	})
+
+#define cmpxchg(ptr, o, n)						\
+	({								\
+		typeof(*(ptr)) __x;					\
+		switch (sizeof(*(ptr))) {				\
+		case 4:							\
+			__x = (typeof(__x))(typeof(__x-__x))atomic_cmpxchg( \
+				(atomic_t *)(ptr),			\
+				(u32)(typeof((o)-(o)))(o),		\
+				(u32)(typeof((n)-(n)))(n));		\
+			break;						\
+		case 8:							\
+			__x = (typeof(__x))(typeof(__x-__x))atomic64_cmpxchg( \
+				(atomic64_t *)(ptr),			\
+				(u64)(typeof((o)-(o)))(o),		\
+				(u64)(typeof((n)-(n)))(n));		\
+			break;						\
+		default:						\
+			__cmpxchg_called_with_bad_pointer();		\
+		}							\
+		__x;							\
+	})
+
+#define tas(ptr) (xchg((ptr), 1))
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* _ASM_TILE_CMPXCHG_H */
diff --git a/arch/tile/include/asm/compat.h b/arch/tile/include/asm/compat.h
index bf95f55b82b0..4b4b28969a65 100644
--- a/arch/tile/include/asm/compat.h
+++ b/arch/tile/include/asm/compat.h
@@ -242,17 +242,6 @@ long compat_sys_fallocate(int fd, int mode,
 long compat_sys_sched_rr_get_interval(compat_pid_t pid,
 				      struct compat_timespec __user *interval);
 
-/* Versions of compat functions that differ from generic Linux. */
-struct compat_msgbuf;
-long tile_compat_sys_msgsnd(int msqid,
-			    struct compat_msgbuf __user *msgp,
-			    size_t msgsz, int msgflg);
-long tile_compat_sys_msgrcv(int msqid,
-			    struct compat_msgbuf __user *msgp,
-			    size_t msgsz, long msgtyp, int msgflg);
-long tile_compat_sys_ptrace(compat_long_t request, compat_long_t pid,
-			    compat_long_t addr, compat_long_t data);
-
 /* Tilera Linux syscalls that don't have "compat" versions. */
 #define compat_sys_flush_cache sys_flush_cache
 
diff --git a/arch/tile/include/asm/exec.h b/arch/tile/include/asm/exec.h
new file mode 100644
index 000000000000..a714e1950867
--- /dev/null
+++ b/arch/tile/include/asm/exec.h
@@ -0,0 +1,20 @@
+/*
+ * Copyright 2010 Tilera Corporation. All Rights Reserved.
+ *
+ *   This program is free software; you can redistribute it and/or
+ *   modify it under the terms of the GNU General Public License
+ *   as published by the Free Software Foundation, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+#ifndef _ASM_TILE_EXEC_H
+#define _ASM_TILE_EXEC_H
+
+#define arch_align_stack(x) (x)
+
+#endif /* _ASM_TILE_EXEC_H */
diff --git a/arch/tile/include/asm/highmem.h b/arch/tile/include/asm/highmem.h
index b2a6c5de79ab..fc8429a31c85 100644
--- a/arch/tile/include/asm/highmem.h
+++ b/arch/tile/include/asm/highmem.h
@@ -59,7 +59,7 @@ void *kmap_fix_kpte(struct page *page, int finished);
 /* This macro is used only in map_new_virtual() to map "page". */
 #define kmap_prot page_to_kpgprot(page)
 
-void *__kmap_atomic(struct page *page);
+void *kmap_atomic(struct page *page);
 void __kunmap_atomic(void *kvaddr);
 void *kmap_atomic_pfn(unsigned long pfn);
 void *kmap_atomic_prot_pfn(unsigned long pfn, pgprot_t prot);
diff --git a/arch/tile/include/asm/irq.h b/arch/tile/include/asm/irq.h
index f80f8ceabc67..33cff9a3058b 100644
--- a/arch/tile/include/asm/irq.h
+++ b/arch/tile/include/asm/irq.h
@@ -21,7 +21,7 @@
 #define NR_IRQS 32
 
 /* IRQ numbers used for linux IPIs. */
-#define IRQ_RESCHEDULE 1
+#define IRQ_RESCHEDULE 0
 
 #define irq_canonicalize(irq)   (irq)
 
diff --git a/arch/tile/include/asm/pgtable.h b/arch/tile/include/asm/pgtable.h
index 1a20b7ef8ea2..67490910774d 100644
--- a/arch/tile/include/asm/pgtable.h
+++ b/arch/tile/include/asm/pgtable.h
@@ -29,7 +29,6 @@
 #include <linux/spinlock.h>
 #include <asm/processor.h>
 #include <asm/fixmap.h>
-#include <asm/system.h>
 
 struct mm_struct;
 struct vm_area_struct;
diff --git a/arch/tile/include/asm/setup.h b/arch/tile/include/asm/setup.h
index 7caf0f36b030..e58613e0752f 100644
--- a/arch/tile/include/asm/setup.h
+++ b/arch/tile/include/asm/setup.h
@@ -31,6 +31,28 @@ void early_panic(const char *fmt, ...);
 void warn_early_printk(void);
 void __init disable_early_printk(void);
 
+/* Init-time routine to do tile-specific per-cpu setup. */
+void setup_cpu(int boot);
+
+/* User-level DMA management functions */
+void grant_dma_mpls(void);
+void restrict_dma_mpls(void);
+
+#ifdef CONFIG_HARDWALL
+/* User-level network management functions */
+void reset_network_state(void);
+void grant_network_mpls(void);
+void restrict_network_mpls(void);
+struct task_struct;
+int hardwall_deactivate(struct task_struct *task);
+
+/* Hook hardwall code into changes in affinity. */
+#define arch_set_cpus_allowed(p, new_mask) do { \
+	if (p->thread.hardwall && !cpumask_equal(&p->cpus_allowed, new_mask)) \
+		hardwall_deactivate(p); \
+} while (0)
+#endif
+
 #endif /* __KERNEL__ */
 
 #endif /* _ASM_TILE_SETUP_H */
diff --git a/arch/tile/include/asm/smp.h b/arch/tile/include/asm/smp.h
index 532124ae4b12..1aa759aeb5b3 100644
--- a/arch/tile/include/asm/smp.h
+++ b/arch/tile/include/asm/smp.h
@@ -43,10 +43,6 @@ void evaluate_message(int tag);
 /* Boot a secondary cpu */
 void online_secondary(void);
 
-/* Call a function on a specified set of CPUs (may include this one). */
-extern void on_each_cpu_mask(const struct cpumask *mask,
-			     void (*func)(void *), void *info, bool wait);
-
 /* Topology of the supervisor tile grid, and coordinates of boot processor */
 extern HV_Topology smp_topology;
 
@@ -91,9 +87,6 @@ void print_disabled_cpus(void);
 
 #else /* !CONFIG_SMP */
 
-#define on_each_cpu_mask(mask, func, info, wait)		\
-  do { if (cpumask_test_cpu(0, (mask))) func(info); } while (0)
-
 #define smp_master_cpu		0
 #define smp_height		1
 #define smp_width		1
diff --git a/arch/tile/include/asm/spinlock_32.h b/arch/tile/include/asm/spinlock_32.h
index a5e4208d34f9..c0a77b38d39a 100644
--- a/arch/tile/include/asm/spinlock_32.h
+++ b/arch/tile/include/asm/spinlock_32.h
@@ -19,7 +19,6 @@
 
 #include <linux/atomic.h>
 #include <asm/page.h>
-#include <asm/system.h>
 #include <linux/compiler.h>
 
 /*
diff --git a/arch/tile/include/asm/spinlock_64.h b/arch/tile/include/asm/spinlock_64.h
index 72be5904e020..5f8b6a095fd8 100644
--- a/arch/tile/include/asm/spinlock_64.h
+++ b/arch/tile/include/asm/spinlock_64.h
@@ -137,7 +137,7 @@ static inline void arch_read_unlock(arch_rwlock_t *rw)
 static inline void arch_write_unlock(arch_rwlock_t *rw)
 {
 	__insn_mf();
-	rw->lock = 0;
+	__insn_exch4(&rw->lock, 0);  /* Avoid waiting in the write buffer. */
 }
 
 static inline int arch_read_trylock(arch_rwlock_t *rw)
diff --git a/arch/tile/include/asm/stack.h b/arch/tile/include/asm/stack.h
index 4d97a2db932e..0e9d382a2d45 100644
--- a/arch/tile/include/asm/stack.h
+++ b/arch/tile/include/asm/stack.h
@@ -25,7 +25,6 @@
 struct KBacktraceIterator {
 	BacktraceIterator it;
 	struct task_struct *task;     /* task we are backtracing */
-	pte_t *pgtable;		      /* page table for user space access */
 	int end;		      /* iteration complete. */
 	int new_context;              /* new context is starting */
 	int profile;                  /* profiling, so stop on async intrpt */
diff --git a/arch/tile/include/asm/switch_to.h b/arch/tile/include/asm/switch_to.h
new file mode 100644
index 000000000000..1d48c5fee8b7
--- /dev/null
+++ b/arch/tile/include/asm/switch_to.h
@@ -0,0 +1,76 @@
+/*
+ * Copyright 2010 Tilera Corporation. All Rights Reserved.
+ *
+ *   This program is free software; you can redistribute it and/or
+ *   modify it under the terms of the GNU General Public License
+ *   as published by the Free Software Foundation, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+#ifndef _ASM_TILE_SWITCH_TO_H
+#define _ASM_TILE_SWITCH_TO_H
+
+#include <arch/sim_def.h>
+
+/*
+ * switch_to(n) should switch tasks to task nr n, first
+ * checking that n isn't the current task, in which case it does nothing.
+ * The number of callee-saved registers saved on the kernel stack
+ * is defined here for use in copy_thread() and must agree with __switch_to().
+ */
+#define CALLEE_SAVED_FIRST_REG 30
+#define CALLEE_SAVED_REGS_COUNT 24   /* r30 to r52, plus an empty to align */
+
+#ifndef __ASSEMBLY__
+
+struct task_struct;
+
+/*
+ * Pause the DMA engine and static network before task switching.
+ */
+#define prepare_arch_switch(next) _prepare_arch_switch(next)
+void _prepare_arch_switch(struct task_struct *next);
+
+struct task_struct;
+#define switch_to(prev, next, last) ((last) = _switch_to((prev), (next)))
+extern struct task_struct *_switch_to(struct task_struct *prev,
+				      struct task_struct *next);
+
+/* Helper function for _switch_to(). */
+extern struct task_struct *__switch_to(struct task_struct *prev,
+				       struct task_struct *next,
+				       unsigned long new_system_save_k_0);
+
+/* Address that switched-away from tasks are at. */
+extern unsigned long get_switch_to_pc(void);
+
+/*
+ * Kernel threads can check to see if they need to migrate their
+ * stack whenever they return from a context switch; for user
+ * threads, we defer until they are returning to user-space.
+ */
+#define finish_arch_switch(prev) do {                                     \
+	if (unlikely((prev)->state == TASK_DEAD))                         \
+		__insn_mtspr(SPR_SIM_CONTROL, SIM_CONTROL_OS_EXIT |       \
+			((prev)->pid << _SIM_CONTROL_OPERATOR_BITS));     \
+	__insn_mtspr(SPR_SIM_CONTROL, SIM_CONTROL_OS_SWITCH |             \
+		(current->pid << _SIM_CONTROL_OPERATOR_BITS));            \
+	if (current->mm == NULL && !kstack_hash &&                        \
+	    current_thread_info()->homecache_cpu != smp_processor_id())   \
+		homecache_migrate_kthread();                              \
+} while (0)
+
+/* Support function for forking a new task. */
+void ret_from_fork(void);
+
+/* Called from ret_from_fork() when a new process starts up. */
+struct task_struct *sim_notify_fork(struct task_struct *prev);
+
+#endif /* !__ASSEMBLY__ */
+
+#endif /* _ASM_TILE_SWITCH_TO_H */
diff --git a/arch/tile/include/asm/timex.h b/arch/tile/include/asm/timex.h
index 29921f0b86da..dc987d53e2a9 100644
--- a/arch/tile/include/asm/timex.h
+++ b/arch/tile/include/asm/timex.h
@@ -29,11 +29,13 @@ typedef unsigned long long cycles_t;
 
 #if CHIP_HAS_SPLIT_CYCLE()
 cycles_t get_cycles(void);
+#define get_cycles_low() __insn_mfspr(SPR_CYCLE_LOW)
 #else
 static inline cycles_t get_cycles(void)
 {
 	return __insn_mfspr(SPR_CYCLE);
 }
+#define get_cycles_low() __insn_mfspr(SPR_CYCLE)   /* just get all 64 bits */
 #endif
 
 cycles_t get_clock_rate(void);
diff --git a/arch/tile/include/asm/traps.h b/arch/tile/include/asm/traps.h
index 5f20f920f932..e28c3df4176a 100644
--- a/arch/tile/include/asm/traps.h
+++ b/arch/tile/include/asm/traps.h
@@ -64,7 +64,11 @@ void do_breakpoint(struct pt_regs *, int fault_num);
 
 
 #ifdef __tilegx__
+/* kernel/single_step.c */
 void gx_singlestep_handle(struct pt_regs *, int fault_num);
+
+/* kernel/intvec_64.S */
+void fill_ra_stack(void);
 #endif
 
-#endif /* _ASM_TILE_SYSCALLS_H */
+#endif /* _ASM_TILE_TRAPS_H */
diff --git a/arch/tile/include/asm/unaligned.h b/arch/tile/include/asm/unaligned.h
index 137e2de5b102..37dfbe598872 100644
--- a/arch/tile/include/asm/unaligned.h
+++ b/arch/tile/include/asm/unaligned.h
@@ -21,4 +21,19 @@
 #define get_unaligned	__get_unaligned_le
 #define put_unaligned	__put_unaligned_le
 
+/*
+ * Is the kernel doing fixups of unaligned accesses?  If <0, no kernel
+ * intervention occurs and SIGBUS is delivered with no data address
+ * info.  If 0, the kernel single-steps the instruction to discover
+ * the data address to provide with the SIGBUS.  If 1, the kernel does
+ * a fixup.
+ */
+extern int unaligned_fixup;
+
+/* Is the kernel printing on each unaligned fixup? */
+extern int unaligned_printk;
+
+/* Number of unaligned fixups performed */
+extern unsigned int unaligned_fixup_count;
+
 #endif /* _ASM_TILE_UNALIGNED_H */