diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c
index 054461e..158bd31 100644
--- a/arch/sparc64/kernel/setup.c
+++ b/arch/sparc64/kernel/setup.c
@@ -542,6 +542,8 @@
 	}
 #endif
 
+	smp_setup_cpu_possible_map();
+
 	paging_init();
 }
 
diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c
index 1fb6323..1f7ad8a 100644
--- a/arch/sparc64/kernel/smp.c
+++ b/arch/sparc64/kernel/smp.c
@@ -1079,18 +1079,12 @@
 	return 0;
 }
 
+/* Constrain the number of cpus to max_cpus.  */
 void __init smp_prepare_cpus(unsigned int max_cpus)
 {
-	int instance, mid;
-
-	instance = 0;
-	while (!cpu_find_by_instance(instance, NULL, &mid)) {
-		if (mid < max_cpus)
-			cpu_set(mid, phys_cpu_present_map);
-		instance++;
-	}
-
 	if (num_possible_cpus() > max_cpus) {
+		int instance, mid;
+
 		instance = 0;
 		while (!cpu_find_by_instance(instance, NULL, &mid)) {
 			if (mid != boot_cpu_id) {
@@ -1105,6 +1099,22 @@
 	smp_store_cpu_info(boot_cpu_id);
 }
 
+/* Set this up early so that things like the scheduler can init
+ * properly.  We use the same cpu mask for both the present and
+ * possible cpu map.
+ */
+void __init smp_setup_cpu_possible_map(void)
+{
+	int instance, mid;
+
+	instance = 0;
+	while (!cpu_find_by_instance(instance, NULL, &mid)) {
+		if (mid < NR_CPUS)
+			cpu_set(mid, phys_cpu_present_map);
+		instance++;
+	}
+}
+
 void __devinit smp_prepare_boot_cpu(void)
 {
 	if (hard_smp_processor_id() >= NR_CPUS) {
diff --git a/include/asm-sparc64/futex.h b/include/asm-sparc64/futex.h
index 6a332a9..0caf601 100644
--- a/include/asm-sparc64/futex.h
+++ b/include/asm-sparc64/futex.h
@@ -1,6 +1,86 @@
-#ifndef _ASM_FUTEX_H
-#define _ASM_FUTEX_H
+#ifndef _SPARC64_FUTEX_H
+#define _SPARC64_FUTEX_H
 
-#include <asm-generic/futex.h>
+#include <linux/futex.h>
+#include <asm/errno.h>
+#include <asm/system.h>
+#include <asm/uaccess.h>
 
-#endif
+#define __futex_cas_op(insn, ret, oldval, uaddr, oparg)	\
+	__asm__ __volatile__(				\
+	"\n1:	lduwa	[%3] %%asi, %2\n"		\
+	"	" insn "\n"				\
+	"2:	casa	[%3] %%asi, %2, %1\n"		\
+	"	cmp	%2, %1\n"			\
+	"	bne,pn	%%icc, 1b\n"			\
+	"	 mov	0, %0\n"			\
+	"3:\n"						\
+	"	.section .fixup,#alloc,#execinstr\n"	\
+	"	.align	4\n"				\
+	"4:	ba	3b\n"				\
+	"	 mov	%5, %0\n"			\
+	"	.previous\n"				\
+	"	.section __ex_table,#alloc\n"		\
+	"	.align	4\n"				\
+	"	.word	1b, 4b\n"			\
+	"	.word	2b, 4b\n"			\
+	"	.previous\n"				\
+	: "=&r" (ret), "=&r" (oldval), "=&r" (tem)	\
+	: "r" (uaddr), "r" (oparg), "i" (-EFAULT)	\
+	: "memory")
+
+static inline int futex_atomic_op_inuser(int encoded_op, int __user *uaddr)
+{
+	int op = (encoded_op >> 28) & 7;
+	int cmp = (encoded_op >> 24) & 15;
+	int oparg = (encoded_op << 8) >> 20;
+	int cmparg = (encoded_op << 20) >> 20;
+	int oldval = 0, ret, tem;
+
+	if (unlikely(!access_ok(VERIFY_WRITE, uaddr, sizeof(int))))
+		return -EFAULT;
+	if (unlikely((((unsigned long) uaddr) & 0x3UL)))
+		return -EINVAL;
+
+	if (encoded_op & (FUTEX_OP_OPARG_SHIFT << 28))
+		oparg = 1 << oparg;
+
+	inc_preempt_count();
+
+	switch (op) {
+	case FUTEX_OP_SET:
+		__futex_cas_op("mov\t%4, %1", ret, oldval, uaddr, oparg);
+		break;
+	case FUTEX_OP_ADD:
+		__futex_cas_op("add\t%2, %4, %1", ret, oldval, uaddr, oparg);
+		break;
+	case FUTEX_OP_OR:
+		__futex_cas_op("or\t%2, %4, %1", ret, oldval, uaddr, oparg);
+		break;
+	case FUTEX_OP_ANDN:
+		__futex_cas_op("and\t%2, %4, %1", ret, oldval, uaddr, oparg);
+		break;
+	case FUTEX_OP_XOR:
+		__futex_cas_op("xor\t%2, %4, %1", ret, oldval, uaddr, oparg);
+		break;
+	default:
+		ret = -ENOSYS;
+	}
+
+	dec_preempt_count();
+
+	if (!ret) {
+		switch (cmp) {
+		case FUTEX_OP_CMP_EQ: ret = (oldval == cmparg); break;
+		case FUTEX_OP_CMP_NE: ret = (oldval != cmparg); break;
+		case FUTEX_OP_CMP_LT: ret = (oldval < cmparg); break;
+		case FUTEX_OP_CMP_GE: ret = (oldval >= cmparg); break;
+		case FUTEX_OP_CMP_LE: ret = (oldval <= cmparg); break;
+		case FUTEX_OP_CMP_GT: ret = (oldval > cmparg); break;
+		default: ret = -ENOSYS;
+		}
+	}
+	return ret;
+}
+
+#endif /* !(_SPARC64_FUTEX_H) */
diff --git a/include/asm-sparc64/smp.h b/include/asm-sparc64/smp.h
index 110a2de..473edb2 100644
--- a/include/asm-sparc64/smp.h
+++ b/include/asm-sparc64/smp.h
@@ -66,8 +66,14 @@
 
 #define raw_smp_processor_id() (current_thread_info()->cpu)
 
+extern void smp_setup_cpu_possible_map(void);
+
 #endif /* !(__ASSEMBLY__) */
 
+#else
+
+#define smp_setup_cpu_possible_map() do { } while (0)
+
 #endif /* !(CONFIG_SMP) */
 
 #define NO_PROC_ID		0xFF