SHA-1: Add SSSE3 implementation

* cipher/Makefile.am: Add 'sha1-ssse3-amd64.c'.
* cipher/sha1-ssse3-amd64.c: New.
* cipher/sha1.c (USE_SSSE3): New.
(SHA1_CONTEXT) [USE_SSSE3]: Add 'use_ssse3'.
(sha1_init) [USE_SSSE3]: Initialize 'use_ssse3'.
(transform): Rename to...
(_transform): this.
(transform): New.
* configure.ac [host=x86_64]: Add 'sha1-ssse3-amd64.lo'.
--

Patch adds SSSE3 implementation based on white paper "Improving the Performance
of the Secure Hash Algorithm (SHA-1)" at
 http://software.intel.com/en-us/articles/improving-the-performance-of-the-secure-hash-algorithm-1

Benchmarks:

cpu                Old        New        Diff
Intel i5-4570      9.02 c/B   5.22 c/B   1.72x
Intel i5-2450M     12.27 c/B  7.24 c/B   1.69x
Intel Core2 T8100  7.94 c/B   6.76 c/B   1.17x

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

1 files changed, 38 insertions, 1 deletions

diff --git a/cipher/sha1.c b/cipher/sha1.c
index 025b3aba..af57b192 100644
--- a/cipher/sha1.c
+++ b/cipher/sha1.c
@@ -43,6 +43,15 @@
 #include "hash-common.h"
 
 
+/* USE_SSSE3 indicates whether to compile with Intel SSSE3 code. */
+#undef USE_SSSE3
+#if defined(__x86_64__) && defined(HAVE_COMPATIBLE_GCC_AMD64_PLATFORM_AS) && \
+    defined(HAVE_GCC_INLINE_ASM_SSSE3) && \
+    defined(HAVE_INTEL_SYNTAX_PLATFORM_AS)
+# define USE_SSSE3 1
+#endif
+
+
 /* A macro to test whether P is properly aligned for an u32 type.
    Note that config.h provides a suitable replacement for uintptr_t if
    it does not exist in stdint.h.  */
@@ -56,6 +65,9 @@ typedef struct
 {
   gcry_md_block_ctx_t bctx;
   u32           h0,h1,h2,h3,h4;
+#ifdef USE_SSSE3
+  unsigned int use_ssse3:1;
+#endif
 } SHA1_CONTEXT;
 
 static unsigned int
@@ -78,6 +90,10 @@ sha1_init (void *context)
   hd->bctx.count = 0;
   hd->bctx.blocksize = 64;
   hd->bctx.bwrite = transform;
+
+#ifdef USE_SSSE3
+  hd->use_ssse3 = (_gcry_get_hw_features () & HWF_INTEL_SSSE3) != 0;
+#endif
 }
 
 
@@ -107,7 +123,7 @@ sha1_init (void *context)
  * Transform NBLOCKS of each 64 bytes (16 32-bit words) at DATA.
  */
 static unsigned int
-transform (void *ctx, const unsigned char *data)
+_transform (void *ctx, const unsigned char *data)
 {
   SHA1_CONTEXT *hd = ctx;
   const u32 *idata = (const void *)data;
@@ -217,6 +233,27 @@ transform (void *ctx, const unsigned char *data)
 }
 
 
+#ifdef USE_SSSE3
+unsigned int
+_gcry_sha1_transform_amd64_ssse3 (void *state, const unsigned char *data);
+#endif
+
+
+static unsigned int
+transform (void *ctx, const unsigned char *data)
+{
+  SHA1_CONTEXT *hd = ctx;
+
+#ifdef USE_SSSE3
+  if (hd->use_ssse3)
+    return _gcry_sha1_transform_amd64_ssse3 (&hd->h0, data)
+           + 4 * sizeof(void*);
+#endif
+
+  return _transform (hd, data);
+}
+
+
 /* The routine final terminates the computation and
  * returns the digest.
  * The handle is prepared for a new cycle, but adding bytes to the