Optimize and cleanup 32-bit and 64-bit endianess transforms

* cipher/bithelp.h (bswap32, bswap64, le_bswap32, be_bswap32)
(le_bswap64, be_bswap64): New.
* cipher/bufhelp.h (buf_get_be32, buf_get_le32, buf_put_le32)
(buf_put_be32, buf_get_be64, buf_get_le64, buf_put_be64)
(buf_put_le64): New.
* cipher/blowfish.c (do_encrypt_block, do_decrypt_block): Use new
endian conversion helpers.
(do_bf_setkey): Turn endian specific code to generic.
* cipher/camellia.c (GETU32, PUTU32): Use new endian conversion
helpers.
* cipher/cast5.c (rol): Remove, use rol from bithelp.
(F1, F2, F3): Fix to use rol from bithelp.
(do_encrypt_block, do_decrypt_block, do_cast_setkey): Use new endian
conversion helpers.
* cipher/des.c (READ_64BIT_DATA, WRITE_64BIT_DATA): Ditto.
* cipher/md4.c (transform, md4_final): Ditto.
* cipher/md5.c (transform, md5_final): Ditto.
* cipher/rmd160.c (transform, rmd160_final): Ditto.
* cipher/salsa20.c (LE_SWAP32, LE_READ_UINT32): Ditto.
* cipher/scrypt.c (READ_UINT64, LE_READ_UINT64, LE_SWAP32): Ditto.
* cipher/seed.c (GETU32, PUTU32): Ditto.
* cipher/serpent.c (byte_swap_32): Remove.
(serpent_key_prepare, serpent_encrypt_internal)
(serpent_decrypt_internal): Use new endian conversion helpers.
* cipher/sha1.c (transform, sha1_final): Ditto.
* cipher/sha256.c (transform, sha256_final): Ditto.
* cipher/sha512.c (__transform, sha512_final): Ditto.
* cipher/stribog.c (transform, stribog_final): Ditto.
* cipher/tiger.c (transform, tiger_final): Ditto.
* cipher/twofish.c (INPACK, OUTUNPACK): Ditto.
* cipher/whirlpool.c (buffer_to_block, block_to_buffer): Ditto.
* configure.ac (gcry_cv_have_builtin_bswap32): Check for compiler
provided __builtin_bswap32.
(gcry_cv_have_builtin_bswap64): Check for compiler provided
__builtin_bswap64.
--

Patch add helper functions that provide conversions to/from integers and
buffers of different endianess. Benefits are code cleanup and optimization
for architectures that have byte-swaping instructions and/or can do fast
unaligned memory accesses.

Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>

1 files changed, 143 insertions, 0 deletions

diff --git a/cipher/bufhelp.h b/cipher/bufhelp.h
index d829cf19..c637dac9 100644
--- a/cipher/bufhelp.h
+++ b/cipher/bufhelp.h
@@ -20,6 +20,8 @@
 #ifndef G10_BUFHELP_H
 #define G10_BUFHELP_H
 
+#include <config.h>
+
 #ifdef HAVE_STDINT_H
 # include <stdint.h> /* uintptr_t */
 #elif defined(HAVE_INTTYPES_H)
@@ -28,6 +30,8 @@
 /* In this case, uintptr_t is provided by config.h. */
 #endif
 
+#include "bithelp.h"
+
 
 #if defined(__i386__) || defined(__x86_64__) || \
     (defined(__arm__) && defined(__ARM_FEATURE_UNALIGNED))
@@ -179,4 +183,143 @@ do_bytes:
     }
 }
 
+
+#ifndef BUFHELP_FAST_UNALIGNED_ACCESS
+
+/* Functions for loading and storing unaligned u32 values of different
+   endianness.  */
+static inline u32 buf_get_be32(const void *_buf)
+{
+  const byte *in = _buf;
+  return ((u32)in[0] << 24) | ((u32)in[1] << 16) | \
+         ((u32)in[2] << 8) | (u32)in[3];
+}
+
+static inline u32 buf_get_le32(const void *_buf)
+{
+  const byte *in = _buf;
+  return ((u32)in[3] << 24) | ((u32)in[2] << 16) | \
+         ((u32)in[1] << 8) | (u32)in[0];
+}
+
+static inline void buf_put_be32(void *_buf, u32 val)
+{
+  byte *out = _buf;
+  out[0] = val >> 24;
+  out[1] = val >> 16;
+  out[2] = val >> 8;
+  out[3] = val;
+}
+
+static inline void buf_put_le32(void *_buf, u32 val)
+{
+  byte *out = _buf;
+  out[3] = val >> 24;
+  out[2] = val >> 16;
+  out[1] = val >> 8;
+  out[0] = val;
+}
+
+#ifdef HAVE_U64_TYPEDEF
+/* Functions for loading and storing unaligned u64 values of different
+   endianness.  */
+static inline u64 buf_get_be64(const void *_buf)
+{
+  const byte *in = _buf;
+  return ((u64)in[0] << 56) | ((u64)in[1] << 48) | \
+         ((u64)in[2] << 40) | ((u64)in[3] << 32) | \
+         ((u64)in[4] << 24) | ((u64)in[5] << 16) | \
+         ((u64)in[6] << 8) | (u64)in[7];
+}
+
+static inline u64 buf_get_le64(const void *_buf)
+{
+  const byte *in = _buf;
+  return ((u64)in[7] << 56) | ((u64)in[6] << 48) | \
+         ((u64)in[5] << 40) | ((u64)in[4] << 32) | \
+         ((u64)in[3] << 24) | ((u64)in[2] << 16) | \
+         ((u64)in[1] << 8) | (u64)in[0];
+}
+
+static inline void buf_put_be64(void *_buf, u64 val)
+{
+  byte *out = _buf;
+  out[0] = val >> 56;
+  out[1] = val >> 48;
+  out[2] = val >> 40;
+  out[3] = val >> 32;
+  out[4] = val >> 24;
+  out[5] = val >> 16;
+  out[6] = val >> 8;
+  out[7] = val;
+}
+
+static inline void buf_put_le64(void *_buf, u64 val)
+{
+  byte *out = _buf;
+  out[7] = val >> 56;
+  out[6] = val >> 48;
+  out[5] = val >> 40;
+  out[4] = val >> 32;
+  out[3] = val >> 24;
+  out[2] = val >> 16;
+  out[1] = val >> 8;
+  out[0] = val;
+}
+#endif /*HAVE_U64_TYPEDEF*/
+
+#else /*BUFHELP_FAST_UNALIGNED_ACCESS*/
+
+/* Functions for loading and storing unaligned u32 values of different
+   endianness.  */
+static inline u32 buf_get_be32(const void *_buf)
+{
+  return be_bswap32(*(const u32 *)_buf);
+}
+
+static inline u32 buf_get_le32(const void *_buf)
+{
+  return le_bswap32(*(const u32 *)_buf);
+}
+
+static inline void buf_put_be32(void *_buf, u32 val)
+{
+  u32 *out = _buf;
+  *out = be_bswap32(val);
+}
+
+static inline void buf_put_le32(void *_buf, u32 val)
+{
+  u32 *out = _buf;
+  *out = le_bswap32(val);
+}
+
+#ifdef HAVE_U64_TYPEDEF
+/* Functions for loading and storing unaligned u64 values of different
+   endianness.  */
+static inline u64 buf_get_be64(const void *_buf)
+{
+  return be_bswap64(*(const u64 *)_buf);
+}
+
+static inline u64 buf_get_le64(const void *_buf)
+{
+  return le_bswap64(*(const u64 *)_buf);
+}
+
+static inline void buf_put_be64(void *_buf, u64 val)
+{
+  u64 *out = _buf;
+  *out = be_bswap64(val);
+}
+
+static inline void buf_put_le64(void *_buf, u64 val)
+{
+  u64 *out = _buf;
+  *out = le_bswap64(val);
+}
+#endif /*HAVE_U64_TYPEDEF*/
+
+#endif /*BUFHELP_FAST_UNALIGNED_ACCESS*/
+
 #endif /*G10_BITHELP_H*/