bitops: use builtins directly

[wimlib] / include / wimlib / unaligned.h

diff --git a/include/wimlib/unaligned.h b/include/wimlib/unaligned.h
index b8a3481b159592e273fd6f618c0ab503a86273a1..5c6aa33b75da3aa6d7601841e12eea17385fe878 100644 (file)
--- a/include/wimlib/unaligned.h
+++ b/include/wimlib/unaligned.h
@@ -83,6 +83,16 @@ get_unaligned_le32(const u8 *p)
                        ((u32)p[1] << 8) | p[0];
 }
 
+static forceinline u32
+get_unaligned_be32(const u8 *p)
+{
+       if (UNALIGNED_ACCESS_IS_FAST)
+               return be32_to_cpu(load_be32_unaligned(p));
+       else
+               return ((u32)p[0] << 24) | ((u32)p[1] << 16) |
+                       ((u32)p[2] << 8) | p[3];
+}
+
 static forceinline void
 put_unaligned_le16(u16 v, u8 *p)
 {
@@ -107,42 +117,17 @@ put_unaligned_le32(u32 v, u8 *p)
        }
 }
 
-/*
- * Given a 32-bit value that was loaded with the platform's native endianness,
- * return a 32-bit value whose high-order 8 bits are 0 and whose low-order 24
- * bits contain the first 3 bytes, arranged in octets in a platform-dependent
- * order, at the memory location from which the input 32-bit value was loaded.
- */
-static forceinline u32
-loaded_u32_to_u24(u32 v)
-{
-       if (CPU_IS_LITTLE_ENDIAN)
-               return v & 0xFFFFFF;
-       else
-               return v >> 8;
-}
-
-/*
- * Load the next 3 bytes from the memory location @p into the 24 low-order bits
- * of a 32-bit value.  The order in which the 3 bytes will be arranged as octets
- * in the 24 bits is platform-dependent.  At least LOAD_U24_REQUIRED_NBYTES
- * bytes must be available at @p; note that this may be more than 3.
- */
-static forceinline u32
-load_u24_unaligned(const u8 *p)
+static forceinline void
+put_unaligned_be32(u32 v, u8 *p)
 {
-#if UNALIGNED_ACCESS_IS_FAST
-#  define LOAD_U24_REQUIRED_NBYTES 4
-       return loaded_u32_to_u24(load_u32_unaligned(p));
-#else
-#  define LOAD_U24_REQUIRED_NBYTES 3
-#  if CPU_IS_BIG_ENDIAN
-       return ((u32)p[2] << 0) | ((u32)p[1] << 8) | ((u32)p[0] << 16);
-#  else
-       return ((u32)p[0] << 0) | ((u32)p[1] << 8) | ((u32)p[2] << 16);
-#  endif
-#endif
+       if (UNALIGNED_ACCESS_IS_FAST) {
+               store_be32_unaligned(cpu_to_be32(v), p);
+       } else {
+               p[0] = (u8)(v >> 24);
+               p[1] = (u8)(v >> 16);
+               p[2] = (u8)(v >> 8);
+               p[3] = (u8)(v >> 0);
+       }
 }
 
-
 #endif /* _WIMLIB_UNALIGNED_H */