Index: linux-2.6.17-rc6-mm2/include/asm-x86_64/local.h
===================================================================
--- linux-2.6.17-rc6-mm2.orig/include/asm-x86_64/local.h	2006-06-12 09:08:24.284596533 -0700
+++ linux-2.6.17-rc6-mm2/include/asm-x86_64/local.h	2006-06-12 09:14:52.730280444 -0700
@@ -45,29 +45,34 @@ static inline void local_sub(long i, loc
 		:"ir" (i), "m" (v->counter));
 }
 
-/* On x86-64 these are better than the atomic variants on SMP kernels
-   because they dont use a lock prefix. */
+/*
+ * Use these for per-cpu local_t variables: on some archs they are
+ * much more efficient than these naive implementations.  Note they take
+ * a variable (eg. mystruct.foo), not an address.
+ *
+ * This could be done better if we moved the per cpu data directly
+ * after GS.
+ */
+#define cpu_local_read(v)	local_read(&per_cpu(v, raw_smp_processor_id()))
+#define cpu_local_set(v, i)	local_set(&per_cpu(v, raw_smp_processor_id()), (i))
+#define cpu_local_inc(v)	local_inc(&per_cpu(v, raw_smp_processor_id()))
+#define cpu_local_dec(v)	local_dec(&per_cpu(v, raw_smp_processor_id()))
+#define cpu_local_add(i, v)	local_add((i), &per_cpu(v, raw_smp_processor_id()))
+#define cpu_local_sub(i, v)	local_sub((i), &per_cpu(v, raw_smp_processor_id()))
+
+/*
+ * Non-atomic increments, ie. preemption disabled and won't be touched
+ * in interrupt, etc.  Some archs can optimize this case well.
+ */
 #define __local_inc(l)		local_inc(l)
 #define __local_dec(l)		local_dec(l)
 #define __local_add(i,l)	local_add((i),(l))
 #define __local_sub(i,l)	local_sub((i),(l))
 
-/* Use these for per-cpu local_t variables: on some archs they are
- * much more efficient than these naive implementations.  Note they take
- * a variable, not an address.
- */
-#define cpu_local_read(v)	local_read(&__get_cpu_var(v))
-#define cpu_local_set(v, i)	local_set(&__get_cpu_var(v), (i))
-#define cpu_local_inc(v)	local_inc(&__get_cpu_var(v))
-#define cpu_local_dec(v)	local_dec(&__get_cpu_var(v))
-#define cpu_local_add(i, v)	local_add((i), &__get_cpu_var(v))
-#define cpu_local_sub(i, v)	local_sub((i), &__get_cpu_var(v))
-
 #define __cpu_local_inc(v)	cpu_local_inc(v)
 #define __cpu_local_dec(v)	cpu_local_dec(v)
-#define __cpu_local_add(i, v)	cpu_local_add((i), (v))
-#define __cpu_local_sub(i, v)	cpu_local_sub((i), (v))
-
-#endif /* _ARCH_I386_LOCAL_H */
+#define __cpu_local_add(i, v)	cpu_local_add(i, v)
+#define __cpu_local_sub(i, v)	cpu_local_sub(i, v)
 
+#endif /* _ARCH_X8664_LOCAL_H */
 
Index: linux-2.6.17-rc6-mm2/include/asm-i386/local.h
===================================================================
--- linux-2.6.17-rc6-mm2.orig/include/asm-i386/local.h	2006-06-05 17:57:02.000000000 -0700
+++ linux-2.6.17-rc6-mm2/include/asm-i386/local.h	2006-06-12 09:13:57.534479835 -0700
@@ -45,26 +45,26 @@ static __inline__ void local_sub(long i,
 		:"ir" (i), "m" (v->counter));
 }
 
-/* On x86, these are no better than the atomic variants. */
+#define cpu_local_read(v)	local_read(&per_cpu(v, raw_smp_processor_id()))
+#define cpu_local_set(v, i)	local_set(&per_cpu(v, raw_smp_processor_id()), (i))
+#define cpu_local_inc(v)	local_inc(&per_cpu(v, raw_smp_processor_id()))
+#define cpu_local_dec(v)	local_dec(&per_cpu(v, raw_smp_processor_id()))
+#define cpu_local_add(i, v)	local_add((i), &per_cpu(v, raw_smp_processor_id()))
+#define cpu_local_sub(i, v)	local_sub((i), &per_cpu(v, raw_smp_processor_id()))
+
+/*
+ * Non-atomic increments, ie. preemption disabled and won't be touched
+ * in interrupt, etc.
+ */
 #define __local_inc(l)		local_inc(l)
 #define __local_dec(l)		local_dec(l)
 #define __local_add(i,l)	local_add((i),(l))
 #define __local_sub(i,l)	local_sub((i),(l))
 
-/* Use these for per-cpu local_t variables: on some archs they are
- * much more efficient than these naive implementations.  Note they take
- * a variable, not an address.
- */
-#define cpu_local_read(v)	local_read(&__get_cpu_var(v))
-#define cpu_local_set(v, i)	local_set(&__get_cpu_var(v), (i))
-#define cpu_local_inc(v)	local_inc(&__get_cpu_var(v))
-#define cpu_local_dec(v)	local_dec(&__get_cpu_var(v))
-#define cpu_local_add(i, v)	local_add((i), &__get_cpu_var(v))
-#define cpu_local_sub(i, v)	local_sub((i), &__get_cpu_var(v))
-
 #define __cpu_local_inc(v)	cpu_local_inc(v)
 #define __cpu_local_dec(v)	cpu_local_dec(v)
-#define __cpu_local_add(i, v)	cpu_local_add((i), (v))
-#define __cpu_local_sub(i, v)	cpu_local_sub((i), (v))
+#define __cpu_local_add(i, v)	cpu_local_add(i, v)
+#define __cpu_local_sub(i, v)	cpu_local_sub(i, v)
 
 #endif /* _ARCH_I386_LOCAL_H */
+