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, 7 insertions, 38 deletions

diff --git a/cipher/rmd160.c b/cipher/rmd160.c
index 7f143dfa..d156e61d 100644
--- a/cipher/rmd160.c
+++ b/cipher/rmd160.c
@@ -28,6 +28,7 @@
 #include "cipher.h" /* Only used for the rmd160_hash_buffer() prototype. */
 
 #include "bithelp.h"
+#include "bufhelp.h"
 
 /*********************************
  * RIPEMD-160 is not patented, see (as of 25.10.97)
@@ -170,32 +171,11 @@ transform ( void *ctx, const unsigned char *data )
   RMD160_CONTEXT *hd = ctx;
   register u32 a,b,c,d,e;
   u32 aa,bb,cc,dd,ee,t;
-#ifdef WORDS_BIGENDIAN
   u32 x[16];
-  {
-    int i;
-    byte *p2;
-    const byte *p1;
-    for (i=0, p1=data, p2=(byte*)x; i < 16; i++, p2 += 4 )
-      {
-        p2[3] = *p1++;
-        p2[2] = *p1++;
-        p2[1] = *p1++;
-        p2[0] = *p1++;
-      }
-  }
-#else
-  /* This version is better because it is always aligned;
-   * The performance penalty on a 586-100 is about 6% which
-   * is acceptable - because the data is more local it might
-   * also be possible that this is faster on some machines.
-   * This function (when compiled with -02 on gcc 2.7.2)
-   * executes on a 586-100 (39.73 bogomips) at about 1900kb/sec;
-   * [measured with a 4MB data and "gpgm --print-md rmd160"] */
-  u32 x[16];
-  memcpy( x, data, 64 );
-#endif
+  int i;
 
+  for ( i = 0; i < 16; i++ )
+    x[i] = buf_get_le32(data + i * 4);
 
 #define K0  0x00000000
 #define K1  0x5A827999
@@ -469,24 +449,13 @@ rmd160_final( void *context )
       memset(hd->bctx.buf, 0, 56 ); /* fill next block with zeroes */
     }
   /* append the 64 bit count */
-  hd->bctx.buf[56] = lsb	   ;
-  hd->bctx.buf[57] = lsb >>  8;
-  hd->bctx.buf[58] = lsb >> 16;
-  hd->bctx.buf[59] = lsb >> 24;
-  hd->bctx.buf[60] = msb	   ;
-  hd->bctx.buf[61] = msb >>  8;
-  hd->bctx.buf[62] = msb >> 16;
-  hd->bctx.buf[63] = msb >> 24;
+  buf_put_le32(hd->bctx.buf + 56, lsb);
+  buf_put_le32(hd->bctx.buf + 60, msb);
   burn = transform( hd, hd->bctx.buf );
   _gcry_burn_stack (burn);
 
   p = hd->bctx.buf;
-#ifdef WORDS_BIGENDIAN
-#define X(a) do { *p++ = hd->h##a	   ; *p++ = hd->h##a >> 8;	\
-	          *p++ = hd->h##a >> 16; *p++ = hd->h##a >> 24; } while(0)
-#else /* little endian */
-#define X(a) do { *(u32*)p = hd->h##a ; p += 4; } while(0)
-#endif
+#define X(a) do { *(u32*)p = le_bswap32(hd->h##a) ; p += 4; } while(0)
   X(0);
   X(1);
   X(2);