Reorganise intreadwrite.h

This changes intreadwrite.h to support per-arch implementations of the
various macros allowing us to take advantage of special instructions
or other properties the compiler does not know about.

Originally committed as revision 18600 to svn://svn.ffmpeg.org/ffmpeg/trunk

libavutil/intreadwrite.h

@@ -23,119 +23,88 @@
 #include "config.h"
 #include "bswap.h"
 
-#ifdef __GNUC__
+/*
+ * Arch-specific headers can provide any combination of
+ * AV_[RW][BLN](16|32|64) macros.  Preprocessor symbols must be
+ * defined, even if these are implemented as inline functions.
+ */
+
+
+/*
+ * Define AV_[RW]N helper macros to simplify definitions not provided
+ * by per-arch headers.
+ */
+
+#if   defined(__GNUC__)
 
 struct unaligned_64 { uint64_t l; } __attribute__((packed));
 struct unaligned_32 { uint32_t l; } __attribute__((packed));
 struct unaligned_16 { uint16_t l; } __attribute__((packed));
 
-#define AV_RN16(a) (((const struct unaligned_16 *) (a))->l)
-#define AV_RN32(a) (((const struct unaligned_32 *) (a))->l)
-#define AV_RN64(a) (((const struct unaligned_64 *) (a))->l)
-
-#define AV_WN16(a, b) (((struct unaligned_16 *) (a))->l) = (b)
-#define AV_WN32(a, b) (((struct unaligned_32 *) (a))->l) = (b)
-#define AV_WN64(a, b) (((struct unaligned_64 *) (a))->l) = (b)
+#   define AV_RN(s, p) (((const struct unaligned_##s *) (p))->l)
+#   define AV_WN(s, p, v) (((struct unaligned_##s *) (p))->l) = (v)
 
 #elif defined(__DECC)
 
-#define AV_RN16(a) (*((const __unaligned uint16_t*)(a)))
-#define AV_RN32(a) (*((const __unaligned uint32_t*)(a)))
-#define AV_RN64(a) (*((const __unaligned uint64_t*)(a)))
-
-#define AV_WN16(a, b) *((__unaligned uint16_t*)(a)) = (b)
-#define AV_WN32(a, b) *((__unaligned uint32_t*)(a)) = (b)
-#define AV_WN64(a, b) *((__unaligned uint64_t*)(a)) = (b)
-
-#else
-
-#define AV_RN16(a) (*((const uint16_t*)(a)))
-#define AV_RN32(a) (*((const uint32_t*)(a)))
-#define AV_RN64(a) (*((const uint64_t*)(a)))
-
-#define AV_WN16(a, b) *((uint16_t*)(a)) = (b)
-#define AV_WN32(a, b) *((uint32_t*)(a)) = (b)
-#define AV_WN64(a, b) *((uint64_t*)(a)) = (b)
-
-#endif /* !__GNUC__ */
-
-/* endian macros */
-#define AV_RB8(x)     (((const uint8_t*)(x))[0])
-#define AV_WB8(p, d)  do { ((uint8_t*)(p))[0] = (d); } while(0)
-
-#define AV_RL8(x)     AV_RB8(x)
-#define AV_WL8(p, d)  AV_WB8(p, d)
-
-#if HAVE_FAST_UNALIGNED
-# ifdef WORDS_BIGENDIAN
-#  define AV_RB16(x)    AV_RN16(x)
-#  define AV_WB16(p, d) AV_WN16(p, d)
-
-#  define AV_RL16(x)    bswap_16(AV_RN16(x))
-#  define AV_WL16(p, d) AV_WN16(p, bswap_16(d))
-
-#  define AV_RB32(x)    AV_RN32(x)
-#  define AV_WB32(p, d) AV_WN32(p, d)
+#   define AV_RN(s, p) (*((const __unaligned uint##s##_t*)(p)))
+#   define AV_WN(s, p, v) *((__unaligned uint##s##_t*)(p)) = (v)
 
-#  define AV_RL32(x)    bswap_32(AV_RN32(x))
-#  define AV_WL32(p, d) AV_WN32(p, bswap_32(d))
+#elif HAVE_FAST_UNALIGNED
 
-#  define AV_RB64(x)    AV_RN64(x)
-#  define AV_WB64(p, d) AV_WN64(p, d)
+#   define AV_RN(s, p) (*((const uint##s##_t*)(p)))
+#   define AV_WN(s, p, v) *((uint##s##_t*)(p)) = (v)
 
-#  define AV_RL64(x)    bswap_64(AV_RN64(x))
-#  define AV_WL64(p, d) AV_WN64(p, bswap_64(d))
-# else /* WORDS_BIGENDIAN */
-#  define AV_RB16(x)    bswap_16(AV_RN16(x))
-#  define AV_WB16(p, d) AV_WN16(p, bswap_16(d))
-
-#  define AV_RL16(x)    AV_RN16(x)
-#  define AV_WL16(p, d) AV_WN16(p, d)
-
-#  define AV_RB32(x)    bswap_32(AV_RN32(x))
-#  define AV_WB32(p, d) AV_WN32(p, bswap_32(d))
-
-#  define AV_RL32(x)    AV_RN32(x)
-#  define AV_WL32(p, d) AV_WN32(p, d)
-
-#  define AV_RB64(x)    bswap_64(AV_RN64(x))
-#  define AV_WB64(p, d) AV_WN64(p, bswap_64(d))
+#else
 
-#  define AV_RL64(x)    AV_RN64(x)
-#  define AV_WL64(p, d) AV_WN64(p, d)
-# endif
-#else /* HAVE_FAST_UNALIGNED */
-#define AV_RB16(x)  ((((const uint8_t*)(x))[0] << 8) | ((const uint8_t*)(x))[1])
+#ifndef AV_RB16
+#define AV_RB16(x)  ((((const uint8_t*)(x))[0] << 8) | \
+                      ((const uint8_t*)(x))[1])
+#endif
+#ifndef AV_WB16
 #define AV_WB16(p, d) do { \
                     ((uint8_t*)(p))[1] = (d); \
                     ((uint8_t*)(p))[0] = (d)>>8; } while(0)
+#endif
 
+#ifndef AV_RL16
 #define AV_RL16(x)  ((((const uint8_t*)(x))[1] << 8) | \
                       ((const uint8_t*)(x))[0])
+#endif
+#ifndef AV_WL16
 #define AV_WL16(p, d) do { \
                     ((uint8_t*)(p))[0] = (d); \
                     ((uint8_t*)(p))[1] = (d)>>8; } while(0)
+#endif
 
+#ifndef AV_RB32
 #define AV_RB32(x)  ((((const uint8_t*)(x))[0] << 24) | \
                      (((const uint8_t*)(x))[1] << 16) | \
                      (((const uint8_t*)(x))[2] <<  8) | \
                       ((const uint8_t*)(x))[3])
+#endif
+#ifndef AV_WB32
 #define AV_WB32(p, d) do { \
                     ((uint8_t*)(p))[3] = (d); \
                     ((uint8_t*)(p))[2] = (d)>>8; \
                     ((uint8_t*)(p))[1] = (d)>>16; \
                     ((uint8_t*)(p))[0] = (d)>>24; } while(0)
+#endif
 
+#ifndef AV_RL32
 #define AV_RL32(x) ((((const uint8_t*)(x))[3] << 24) | \
                     (((const uint8_t*)(x))[2] << 16) | \
                     (((const uint8_t*)(x))[1] <<  8) | \
                      ((const uint8_t*)(x))[0])
+#endif
+#ifndef AV_WL32
 #define AV_WL32(p, d) do { \
                     ((uint8_t*)(p))[0] = (d); \
                     ((uint8_t*)(p))[1] = (d)>>8; \
                     ((uint8_t*)(p))[2] = (d)>>16; \
                     ((uint8_t*)(p))[3] = (d)>>24; } while(0)
+#endif
 
+#ifndef AV_RB64
 #define AV_RB64(x)  (((uint64_t)((const uint8_t*)(x))[0] << 56) | \
                      ((uint64_t)((const uint8_t*)(x))[1] << 48) | \
                      ((uint64_t)((const uint8_t*)(x))[2] << 40) | \
@@ -144,6 +113,8 @@ struct unaligned_16 { uint16_t l; } __attribute__((packed));
                      ((uint64_t)((const uint8_t*)(x))[5] << 16) | \
                      ((uint64_t)((const uint8_t*)(x))[6] <<  8) | \
                       (uint64_t)((const uint8_t*)(x))[7])
+#endif
+#ifndef AV_WB64
 #define AV_WB64(p, d) do { \
                     ((uint8_t*)(p))[7] = (d);     \
                     ((uint8_t*)(p))[6] = (d)>>8;  \
@@ -153,7 +124,9 @@ struct unaligned_16 { uint16_t l; } __attribute__((packed));
                     ((uint8_t*)(p))[2] = (d)>>40; \
                     ((uint8_t*)(p))[1] = (d)>>48; \
                     ((uint8_t*)(p))[0] = (d)>>56; } while(0)
+#endif
 
+#ifndef AV_RL64
 #define AV_RL64(x)  (((uint64_t)((const uint8_t*)(x))[7] << 56) | \
                      ((uint64_t)((const uint8_t*)(x))[6] << 48) | \
                      ((uint64_t)((const uint8_t*)(x))[5] << 40) | \
@@ -162,6 +135,8 @@ struct unaligned_16 { uint16_t l; } __attribute__((packed));
                      ((uint64_t)((const uint8_t*)(x))[2] << 16) | \
                      ((uint64_t)((const uint8_t*)(x))[1] <<  8) | \
                       (uint64_t)((const uint8_t*)(x))[0])
+#endif
+#ifndef AV_WL64
 #define AV_WL64(p, d) do { \
                     ((uint8_t*)(p))[0] = (d);     \
                     ((uint8_t*)(p))[1] = (d)>>8;  \
@@ -171,7 +146,101 @@ struct unaligned_16 { uint16_t l; } __attribute__((packed));
                     ((uint8_t*)(p))[5] = (d)>>40; \
                     ((uint8_t*)(p))[6] = (d)>>48; \
                     ((uint8_t*)(p))[7] = (d)>>56; } while(0)
-#endif  /* HAVE_FAST_UNALIGNED */
+#endif
+
+#ifdef WORDS_BIGENDIAN
+#   define AV_RN(s, p)    AV_RB##s(p)
+#   define AV_WN(s, p, v) AV_WB##s(p, v)
+#else
+#   define AV_RN(s, p)    AV_RL##s(p)
+#   define AV_WN(s, p, v) AV_WL##s(p, v)
+#endif
+
+#endif /* HAVE_FAST_UNALIGNED */
+
+#ifndef AV_RN16
+#   define AV_RN16(p) AV_RN(16, p)
+#endif
+
+#ifndef AV_RN32
+#   define AV_RN32(p) AV_RN(32, p)
+#endif
+
+#ifndef AV_RN64
+#   define AV_RN64(p) AV_RN(64, p)
+#endif
+
+#ifndef AV_WN16
+#   define AV_WN16(p, v) AV_WN(16, p, v)
+#endif
+
+#ifndef AV_WN32
+#   define AV_WN32(p, v) AV_WN(32, p, v)
+#endif
+
+#ifndef AV_WN64
+#   define AV_WN64(p, v) AV_WN(64, p, v)
+#endif
+
+#ifdef WORDS_BIGENDIAN
+#   define AV_RB(s, p)    AV_RN(s, p)
+#   define AV_WB(s, p, v) AV_WN(s, p, v)
+#   define AV_RL(s, p)    bswap_##s(AV_RN(s, p))
+#   define AV_WL(s, p, v) AV_WN(s, p, bswap_##s(v))
+#else
+#   define AV_RB(s, p)    bswap_##s(AV_RN(s, p))
+#   define AV_WB(s, p, v) AV_WN(s, p, bswap_##s(v))
+#   define AV_RL(s, p)    AV_RN(s, p)
+#   define AV_WL(s, p, v) AV_WN(s, p, v)
+#endif
+
+#define AV_RB8(x)     (((const uint8_t*)(x))[0])
+#define AV_WB8(p, d)  do { ((uint8_t*)(p))[0] = (d); } while(0)
+
+#define AV_RL8(x)     AV_RB8(x)
+#define AV_WL8(p, d)  AV_WB8(p, d)
+
+#ifndef AV_RB16
+#   define AV_RB16(p)    AV_RB(16, p)
+#endif
+#ifndef AV_WB16
+#   define AV_WB16(p, v) AV_WB(16, p, v)
+#endif
+
+#ifndef AV_RL16
+#   define AV_RL16(p)    AV_RL(16, p)
+#endif
+#ifndef AV_WL16
+#   define AV_WL16(p, v) AV_WL(16, p, v)
+#endif
+
+#ifndef AV_RB32
+#   define AV_RB32(p)    AV_RB(32, p)
+#endif
+#ifndef AV_WB32
+#   define AV_WB32(p, v) AV_WB(32, p, v)
+#endif
+
+#ifndef AV_RL32
+#   define AV_RL32(p)    AV_RL(32, p)
+#endif
+#ifndef AV_WL32
+#   define AV_WL32(p, v) AV_WL(32, p, v)
+#endif
+
+#ifndef AV_RB64
+#   define AV_RB64(p)    AV_RB(64, p)
+#endif
+#ifndef AV_WB64
+#   define AV_WB64(p, v) AV_WB(64, p, v)
+#endif
+
+#ifndef AV_RL64
+#   define AV_RL64(p)    AV_RL(64, p)
+#endif
+#ifndef AV_WL64
+#   define AV_WL64(p, v) AV_WL(64, p, v)
+#endif
 
 #define AV_RB24(x)  ((((const uint8_t*)(x))[0] << 16) | \
                      (((const uint8_t*)(x))[1] <<  8) | \