[PATCH] Use optimised byteswap again

Looks like Al's patch to sparsify the byteorder code prevented use of the
assembler-optimised code.  Not that I can blame him -- three underscores
are C-coded and two underscores are assembly-optimised.

include/linux/byteorder/big_endian.h

@@ -27,12 +27,12 @@
 #define __constant_be32_to_cpu(x) ((__force __u32)(__be32)(x))
 #define __constant_cpu_to_be16(x) ((__force __be16)(__u16)(x))
 #define __constant_be16_to_cpu(x) ((__force __u16)(__be16)(x))
-#define __cpu_to_le64(x) ((__force __le64)___swab64((x)))
-#define __le64_to_cpu(x) ___swab64((__force __u64)(__le64)(x))
-#define __cpu_to_le32(x) ((__force __le32)___swab32((x)))
-#define __le32_to_cpu(x) ___swab32((__force __u32)(__le32)(x))
-#define __cpu_to_le16(x) ((__force __le16)___swab16((x)))
-#define __le16_to_cpu(x) ___swab16((__force __u16)(__le16)(x))
+#define __cpu_to_le64(x) ((__force __le64)__swab64((x)))
+#define __le64_to_cpu(x) __swab64((__force __u64)(__le64)(x))
+#define __cpu_to_le32(x) ((__force __le32)__swab32((x)))
+#define __le32_to_cpu(x) __swab32((__force __u32)(__le32)(x))
+#define __cpu_to_le16(x) ((__force __le16)__swab16((x)))
+#define __le16_to_cpu(x) __swab16((__force __u16)(__le16)(x))
 #define __cpu_to_be64(x) ((__force __be64)(__u64)(x))
 #define __be64_to_cpu(x) ((__force __u64)(__be64)(x))
 #define __cpu_to_be32(x) ((__force __be32)(__u32)(x))

include/linux/byteorder/little_endian.h

@@ -33,12 +33,12 @@
 #define __le32_to_cpu(x) ((__force __u32)(__le32)(x))
 #define __cpu_to_le16(x) ((__force __le16)(__u16)(x))
 #define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
-#define __cpu_to_be64(x) ((__force __be64)___swab64((x)))
-#define __be64_to_cpu(x) ___swab64((__force __u64)(__be64)(x))
-#define __cpu_to_be32(x) ((__force __be32)___swab32((x)))
-#define __be32_to_cpu(x) ___swab32((__force __u32)(__be32)(x))
-#define __cpu_to_be16(x) ((__force __be16)___swab16((x)))
-#define __be16_to_cpu(x) ___swab16((__force __u16)(__be16)(x))
+#define __cpu_to_be64(x) ((__force __be64)__swab64((x)))
+#define __be64_to_cpu(x) __swab64((__force __u64)(__be64)(x))
+#define __cpu_to_be32(x) ((__force __be32)__swab32((x)))
+#define __be32_to_cpu(x) __swab32((__force __u32)(__be32)(x))
+#define __cpu_to_be16(x) ((__force __be16)__swab16((x)))
+#define __be16_to_cpu(x) __swab16((__force __u16)(__be16)(x))
 
 static inline __le64 __cpu_to_le64p(const __u64 *p)
 {