diff --git a/arch/tile/include/asm/atomic.h b/arch/tile/include/asm/atomic.h
index d385eaadece7a68fe603eeb625b3caedb3c4c345..70979846076332bf7771d8517afdcc3415518b17 100644
--- a/arch/tile/include/asm/atomic.h
+++ b/arch/tile/include/asm/atomic.h
@@ -166,7 +166,7 @@ static inline int atomic_cmpxchg(atomic_t *v, int o, int n)
  *
  * Atomically sets @v to @i and returns old @v
  */
-static inline u64 atomic64_xchg(atomic64_t *v, u64 n)
+static inline long long atomic64_xchg(atomic64_t *v, long long n)
 {
 	return xchg64(&v->counter, n);
 }
@@ -180,7 +180,8 @@ static inline u64 atomic64_xchg(atomic64_t *v, u64 n)
  * Atomically checks if @v holds @o and replaces it with @n if so.
  * Returns the old value at @v.
  */
-static inline u64 atomic64_cmpxchg(atomic64_t *v, u64 o, u64 n)
+static inline long long atomic64_cmpxchg(atomic64_t *v, long long o,
+					long long n)
 {
 	return cmpxchg64(&v->counter, o, n);
 }
diff --git a/arch/tile/include/asm/atomic_32.h b/arch/tile/include/asm/atomic_32.h
index 0d0395b1b1529d454f772ebb61e240d7fb41f52a..1ad4a1f7d42b8aa47eadbbb4cdcf2e50dc630fd0 100644
--- a/arch/tile/include/asm/atomic_32.h
+++ b/arch/tile/include/asm/atomic_32.h
@@ -80,7 +80,7 @@ static inline void atomic_set(atomic_t *v, int n)
 /* A 64bit atomic type */
 
 typedef struct {
-	u64 __aligned(8) counter;
+	long long counter;
 } atomic64_t;
 
 #define ATOMIC64_INIT(val) { (val) }
@@ -91,14 +91,14 @@ typedef struct {
  *
  * Atomically reads the value of @v.
  */
-static inline u64 atomic64_read(const atomic64_t *v)
+static inline long long atomic64_read(const atomic64_t *v)
 {
 	/*
 	 * Requires an atomic op to read both 32-bit parts consistently.
 	 * Casting away const is safe since the atomic support routines
 	 * do not write to memory if the value has not been modified.
 	 */
-	return _atomic64_xchg_add((u64 *)&v->counter, 0);
+	return _atomic64_xchg_add((long long *)&v->counter, 0);
 }
 
 /**
@@ -108,7 +108,7 @@ static inline u64 atomic64_read(const atomic64_t *v)
  *
  * Atomically adds @i to @v.
  */
-static inline void atomic64_add(u64 i, atomic64_t *v)
+static inline void atomic64_add(long long i, atomic64_t *v)
 {
 	_atomic64_xchg_add(&v->counter, i);
 }
@@ -120,7 +120,7 @@ static inline void atomic64_add(u64 i, atomic64_t *v)
  *
  * Atomically adds @i to @v and returns @i + @v
  */
-static inline u64 atomic64_add_return(u64 i, atomic64_t *v)
+static inline long long atomic64_add_return(long long i, atomic64_t *v)
 {
 	smp_mb();  /* barrier for proper semantics */
 	return _atomic64_xchg_add(&v->counter, i) + i;
@@ -135,7 +135,8 @@ static inline u64 atomic64_add_return(u64 i, atomic64_t *v)
  * Atomically adds @a to @v, so long as @v was not already @u.
  * Returns non-zero if @v was not @u, and zero otherwise.
  */
-static inline u64 atomic64_add_unless(atomic64_t *v, u64 a, u64 u)
+static inline long long atomic64_add_unless(atomic64_t *v, long long a,
+					long long u)
 {
 	smp_mb();  /* barrier for proper semantics */
 	return _atomic64_xchg_add_unless(&v->counter, a, u) != u;
@@ -151,7 +152,7 @@ static inline u64 atomic64_add_unless(atomic64_t *v, u64 a, u64 u)
  * atomic64_set() can't be just a raw store, since it would be lost if it
  * fell between the load and store of one of the other atomic ops.
  */
-static inline void atomic64_set(atomic64_t *v, u64 n)
+static inline void atomic64_set(atomic64_t *v, long long n)
 {
 	_atomic64_xchg(&v->counter, n);
 }
@@ -236,11 +237,13 @@ extern struct __get_user __atomic_xchg_add_unless(volatile int *p,
 extern struct __get_user __atomic_or(volatile int *p, int *lock, int n);
 extern struct __get_user __atomic_andn(volatile int *p, int *lock, int n);
 extern struct __get_user __atomic_xor(volatile int *p, int *lock, int n);
-extern u64 __atomic64_cmpxchg(volatile u64 *p, int *lock, u64 o, u64 n);
-extern u64 __atomic64_xchg(volatile u64 *p, int *lock, u64 n);
-extern u64 __atomic64_xchg_add(volatile u64 *p, int *lock, u64 n);
-extern u64 __atomic64_xchg_add_unless(volatile u64 *p,
-				      int *lock, u64 o, u64 n);
+extern long long __atomic64_cmpxchg(volatile long long *p, int *lock,
+					long long o, long long n);
+extern long long __atomic64_xchg(volatile long long *p, int *lock, long long n);
+extern long long __atomic64_xchg_add(volatile long long *p, int *lock,
+					long long n);
+extern long long __atomic64_xchg_add_unless(volatile long long *p,
+					int *lock, long long o, long long n);
 
 /* Return failure from the atomic wrappers. */
 struct __get_user __atomic_bad_address(int __user *addr);
diff --git a/arch/tile/include/asm/cmpxchg.h b/arch/tile/include/asm/cmpxchg.h
index 4001d5eab4bb7f1fb59e6e62c924bd71b4b10e2f..0ccda3c425be0d3b19a27c13b68ad6b6eaa37525 100644
--- a/arch/tile/include/asm/cmpxchg.h
+++ b/arch/tile/include/asm/cmpxchg.h
@@ -35,10 +35,10 @@ int _atomic_xchg(int *ptr, int n);
 int _atomic_xchg_add(int *v, int i);
 int _atomic_xchg_add_unless(int *v, int a, int u);
 int _atomic_cmpxchg(int *ptr, int o, int n);
-u64 _atomic64_xchg(u64 *v, u64 n);
-u64 _atomic64_xchg_add(u64 *v, u64 i);
-u64 _atomic64_xchg_add_unless(u64 *v, u64 a, u64 u);
-u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n);
+long long _atomic64_xchg(long long *v, long long n);
+long long _atomic64_xchg_add(long long *v, long long i);
+long long _atomic64_xchg_add_unless(long long *v, long long a, long long u);
+long long _atomic64_cmpxchg(long long *v, long long o, long long n);
 
 #define xchg(ptr, n)							\
 	({								\
@@ -53,7 +53,8 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n);
 		if (sizeof(*(ptr)) != 4)				\
 			__cmpxchg_called_with_bad_pointer();		\
 		smp_mb();						\
-		(typeof(*(ptr)))_atomic_cmpxchg((int *)ptr, (int)o, (int)n); \
+		(typeof(*(ptr)))_atomic_cmpxchg((int *)ptr, (int)o,	\
+						(int)n);		\
 	})
 
 #define xchg64(ptr, n)							\
@@ -61,7 +62,8 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n);
 		if (sizeof(*(ptr)) != 8)				\
 			__xchg_called_with_bad_pointer();		\
 		smp_mb();						\
-		(typeof(*(ptr)))_atomic64_xchg((u64 *)(ptr), (u64)(n));	\
+		(typeof(*(ptr)))_atomic64_xchg((long long *)(ptr),	\
+						(long long)(n));	\
 	})
 
 #define cmpxchg64(ptr, o, n)						\
@@ -69,7 +71,8 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n);
 		if (sizeof(*(ptr)) != 8)				\
 			__cmpxchg_called_with_bad_pointer();		\
 		smp_mb();						\
-		(typeof(*(ptr)))_atomic64_cmpxchg((u64 *)ptr, (u64)o, (u64)n); \
+		(typeof(*(ptr)))_atomic64_cmpxchg((long long *)ptr,	\
+					(long long)o, (long long)n);	\
 	})
 
 #else
@@ -81,10 +84,11 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n);
 		switch (sizeof(*(ptr))) {				\
 		case 4:							\
 			__x = (typeof(__x))(unsigned long)		\
-				__insn_exch4((ptr), (u32)(unsigned long)(n)); \
+				__insn_exch4((ptr),			\
+					(u32)(unsigned long)(n));	\
 			break;						\
 		case 8:							\
-			__x = (typeof(__x))			\
+			__x = (typeof(__x))				\
 				__insn_exch((ptr), (unsigned long)(n));	\
 			break;						\
 		default:						\
@@ -103,10 +107,12 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n);
 		switch (sizeof(*(ptr))) {				\
 		case 4:							\
 			__x = (typeof(__x))(unsigned long)		\
-				__insn_cmpexch4((ptr), (u32)(unsigned long)(n)); \
+				__insn_cmpexch4((ptr),			\
+					(u32)(unsigned long)(n));	\
 			break;						\
 		case 8:							\
-			__x = (typeof(__x))__insn_cmpexch((ptr), (u64)(n)); \
+			__x = (typeof(__x))__insn_cmpexch((ptr),	\
+						(long long)(n));	\
 			break;						\
 		default:						\
 			__cmpxchg_called_with_bad_pointer();		\
diff --git a/arch/tile/include/asm/percpu.h b/arch/tile/include/asm/percpu.h
index 63294f5a8efbca7820c891232c7f79303213e653..4f7ae39fa2022c537770bd10e1cbb0e231249b53 100644
--- a/arch/tile/include/asm/percpu.h
+++ b/arch/tile/include/asm/percpu.h
@@ -15,9 +15,37 @@
 #ifndef _ASM_TILE_PERCPU_H
 #define _ASM_TILE_PERCPU_H
 
-register unsigned long __my_cpu_offset __asm__("tp");
-#define __my_cpu_offset __my_cpu_offset
-#define set_my_cpu_offset(tp) (__my_cpu_offset = (tp))
+register unsigned long my_cpu_offset_reg asm("tp");
+
+#ifdef CONFIG_PREEMPT
+/*
+ * For full preemption, we can't just use the register variable
+ * directly, since we need barrier() to hazard against it, causing the
+ * compiler to reload anything computed from a previous "tp" value.
+ * But we also don't want to use volatile asm, since we'd like the
+ * compiler to be able to cache the value across multiple percpu reads.
+ * So we use a fake stack read as a hazard against barrier().
+ * The 'U' constraint is like 'm' but disallows postincrement.
+ */
+static inline unsigned long __my_cpu_offset(void)
+{
+	unsigned long tp;
+	register unsigned long *sp asm("sp");
+	asm("move %0, tp" : "=r" (tp) : "U" (*sp));
+	return tp;
+}
+#define __my_cpu_offset __my_cpu_offset()
+#else
+/*
+ * We don't need to hazard against barrier() since "tp" doesn't ever
+ * change with PREEMPT_NONE, and with PREEMPT_VOLUNTARY it only
+ * changes at function call points, at which we are already re-reading
+ * the value of "tp" due to "my_cpu_offset_reg" being a global variable.
+ */
+#define __my_cpu_offset my_cpu_offset_reg
+#endif
+
+#define set_my_cpu_offset(tp) (my_cpu_offset_reg = (tp))
 
 #include <asm-generic/percpu.h>
 
diff --git a/arch/tile/kernel/hardwall.c b/arch/tile/kernel/hardwall.c
index df27a1fd94a310a759612a5c6706b1011a94f787..531f4c365351119eeb249904cc01fd60307931b4 100644
--- a/arch/tile/kernel/hardwall.c
+++ b/arch/tile/kernel/hardwall.c
@@ -66,7 +66,7 @@ static struct hardwall_type hardwall_types[] = {
 		0,
 		"udn",
 		LIST_HEAD_INIT(hardwall_types[HARDWALL_UDN].list),
-		__SPIN_LOCK_INITIALIZER(hardwall_types[HARDWALL_UDN].lock),
+		__SPIN_LOCK_UNLOCKED(hardwall_types[HARDWALL_UDN].lock),
 		NULL
 	},
 #ifndef __tilepro__
@@ -77,7 +77,7 @@ static struct hardwall_type hardwall_types[] = {
 		1,  /* disabled pending hypervisor support */
 		"idn",
 		LIST_HEAD_INIT(hardwall_types[HARDWALL_IDN].list),
-		__SPIN_LOCK_INITIALIZER(hardwall_types[HARDWALL_IDN].lock),
+		__SPIN_LOCK_UNLOCKED(hardwall_types[HARDWALL_IDN].lock),
 		NULL
 	},
 	{  /* access to user-space IPI */
@@ -87,7 +87,7 @@ static struct hardwall_type hardwall_types[] = {
 		0,
 		"ipi",
 		LIST_HEAD_INIT(hardwall_types[HARDWALL_IPI].list),
-		__SPIN_LOCK_INITIALIZER(hardwall_types[HARDWALL_IPI].lock),
+		__SPIN_LOCK_UNLOCKED(hardwall_types[HARDWALL_IPI].lock),
 		NULL
 	},
 #endif
diff --git a/arch/tile/kernel/intvec_32.S b/arch/tile/kernel/intvec_32.S
index 088d5c141e681084ce030165f303d15f4b69876f..2cbe6d5dd6b04db3ea071fb12c3dbb1866dc818e 100644
--- a/arch/tile/kernel/intvec_32.S
+++ b/arch/tile/kernel/intvec_32.S
@@ -815,6 +815,9 @@ STD_ENTRY(interrupt_return)
 	}
 	bzt     r28, 1f
 	bnz     r29, 1f
+	/* Disable interrupts explicitly for preemption. */
+	IRQ_DISABLE(r20,r21)
+	TRACE_IRQS_OFF
 	jal     preempt_schedule_irq
 	FEEDBACK_REENTER(interrupt_return)
 1:
diff --git a/arch/tile/kernel/intvec_64.S b/arch/tile/kernel/intvec_64.S
index ec755d3f373467ebe271dd1743cc6f56b6f9d7da..b8fc497f24370c0d7c17536664cb19c40923e5bf 100644
--- a/arch/tile/kernel/intvec_64.S
+++ b/arch/tile/kernel/intvec_64.S
@@ -841,6 +841,9 @@ STD_ENTRY(interrupt_return)
 	}
 	beqzt   r28, 1f
 	bnez    r29, 1f
+	/* Disable interrupts explicitly for preemption. */
+	IRQ_DISABLE(r20,r21)
+	TRACE_IRQS_OFF
 	jal     preempt_schedule_irq
 	FEEDBACK_REENTER(interrupt_return)
 1:
diff --git a/arch/tile/kernel/stack.c b/arch/tile/kernel/stack.c
index 362284af3afd31ab39081447b6f1295a866c0ab9..c93977a62116dfecd12a74376e307c89c7875b91 100644
--- a/arch/tile/kernel/stack.c
+++ b/arch/tile/kernel/stack.c
@@ -23,6 +23,7 @@
 #include <linux/mmzone.h>
 #include <linux/dcache.h>
 #include <linux/fs.h>
+#include <linux/string.h>
 #include <asm/backtrace.h>
 #include <asm/page.h>
 #include <asm/ucontext.h>
@@ -332,21 +333,18 @@ static void describe_addr(struct KBacktraceIterator *kbt,
 	}
 
 	if (vma->vm_file) {
-		char *s;
 		p = d_path(&vma->vm_file->f_path, buf, bufsize);
 		if (IS_ERR(p))
 			p = "?";
-		s = strrchr(p, '/');
-		if (s)
-			p = s+1;
+		name = kbasename(p);
 	} else {
-		p = "anon";
+		name = "anon";
 	}
 
 	/* Generate a string description of the vma info. */
-	namelen = strlen(p);
+	namelen = strlen(name);
 	remaining = (bufsize - 1) - namelen;
-	memmove(buf, p, namelen);
+	memmove(buf, name, namelen);
 	snprintf(buf + namelen, remaining, "[%lx+%lx] ",
 		 vma->vm_start, vma->vm_end - vma->vm_start);
 }
diff --git a/arch/tile/lib/atomic_32.c b/arch/tile/lib/atomic_32.c
index 759efa337be88ebaf72cd1047151a2489a7745cb..c89b211fd9e7c093ed26e932432b329974003a04 100644
--- a/arch/tile/lib/atomic_32.c
+++ b/arch/tile/lib/atomic_32.c
@@ -107,19 +107,19 @@ unsigned long _atomic_xor(volatile unsigned long *p, unsigned long mask)
 EXPORT_SYMBOL(_atomic_xor);
 
 
-u64 _atomic64_xchg(u64 *v, u64 n)
+long long _atomic64_xchg(long long *v, long long n)
 {
 	return __atomic64_xchg(v, __atomic_setup(v), n);
 }
 EXPORT_SYMBOL(_atomic64_xchg);
 
-u64 _atomic64_xchg_add(u64 *v, u64 i)
+long long _atomic64_xchg_add(long long *v, long long i)
 {
 	return __atomic64_xchg_add(v, __atomic_setup(v), i);
 }
 EXPORT_SYMBOL(_atomic64_xchg_add);
 
-u64 _atomic64_xchg_add_unless(u64 *v, u64 a, u64 u)
+long long _atomic64_xchg_add_unless(long long *v, long long a, long long u)
 {
 	/*
 	 * Note: argument order is switched here since it is easier
@@ -130,7 +130,7 @@ u64 _atomic64_xchg_add_unless(u64 *v, u64 a, u64 u)
 }
 EXPORT_SYMBOL(_atomic64_xchg_add_unless);
 
-u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n)
+long long _atomic64_cmpxchg(long long *v, long long o, long long n)
 {
 	return __atomic64_cmpxchg(v, __atomic_setup(v), o, n);
 }