Adapt ctmul64.c to work with polyval.c.

diff --git a/src/ext/polyval/ctmul64.c b/src/ext/polyval/ctmul64.c
index a46f16fee977f6102abea7f7bcdf169a013c3e8e..c2205953a380f08b0e2abd5c1df0168b303260ad 100644 (file)
--- a/src/ext/polyval/ctmul64.c
+++ b/src/ext/polyval/ctmul64.c
@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2016 Thomas Pornin <pornin@bolet.org>
  *
- * Permission is hereby granted, free of charge, to any person obtaining 
+ * Permission is hereby granted, free of charge, to any person obtaining
  * a copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
  * without limitation the rights to use, copy, modify, merge, publish,
@@ -9,12 +9,12 @@
  * permit persons to whom the Software is furnished to do so, subject to
  * the following conditions:
  *
- * The above copyright notice and this permission notice shall be 
+ * The above copyright notice and this permission notice shall be
  * included in all copies or substantial portions of the Software.
  *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND 
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
  * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
  * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
@@ -22,8 +22,6 @@
  * SOFTWARE.
  */
 
-#include "inner.h"
-
 /*
  * This is the 64-bit variant of br_ghash_ctmul32(), with 64-bit operands
  * and bit reversal of 64-bit words.
@@ -73,46 +71,28 @@ rev64(uint64_t x)
 #undef RMS
 }
 
-/* see bearssl_ghash.h */
-void
-br_ghash_ctmul64(void *y, const void *h, const void *data, size_t len)
+
+static void
+pv_mul_y_h(polyval_t *pv)
 {
-       const unsigned char *buf, *hb;
-       unsigned char *yb;
        uint64_t y0, y1;
        uint64_t h0, h1, h2, h0r, h1r, h2r;
 
-       buf = data;
-       yb = y;
-       hb = h;
-       y1 = br_dec64be(yb);
-       y0 = br_dec64be(yb + 8);
-       h1 = br_dec64be(hb);
-       h0 = br_dec64be(hb + 8);
-       h0r = rev64(h0);
-       h1r = rev64(h1);
+       y0 = pv->y.lo;
+       y1 = pv->y.hi;
+       h0 = pv->h.lo;
+       h1 = pv->h.hi;
+       // TODO(nm) does it actually make sense for us to precompute this?
+       h0r = pv->hr.lo;
+       h1r = pv->hr.hi;
+
        h2 = h0 ^ h1;
        h2r = h0r ^ h1r;
-       while (len > 0) {
-               const unsigned char *src;
-               unsigned char tmp[16];
+       {
                uint64_t y0r, y1r, y2, y2r;
                uint64_t z0, z1, z2, z0h, z1h, z2h;
                uint64_t v0, v1, v2, v3;
 
-               if (len >= 16) {
-                       src = buf;
-                       buf += 16;
-                       len -= 16;
-               } else {
-                       memcpy(tmp, buf, len);
-                       memset(tmp + len, 0, (sizeof tmp) - len);
-                       src = tmp;
-                       len = 0;
-               }
-               y1 ^= br_dec64be(src);
-               y0 ^= br_dec64be(src + 8);
-
                y0r = rev64(y0);
                y1r = rev64(y1);
                y2 = y0 ^ y1;
@@ -135,20 +115,20 @@ br_ghash_ctmul64(void *y, const void *h, const void *data, size_t len)
                v2 = z1 ^ z2h;
                v3 = z1h;
 
+#if 0
+               // This step is GHASH only.
                v3 = (v3 << 1) | (v2 >> 63);
                v2 = (v2 << 1) | (v1 >> 63);
                v1 = (v1 << 1) | (v0 >> 63);
                v0 = (v0 << 1);
+#endif
 
                v2 ^= v0 ^ (v0 >> 1) ^ (v0 >> 2) ^ (v0 >> 7);
                v1 ^= (v0 << 63) ^ (v0 << 62) ^ (v0 << 57);
                v3 ^= v1 ^ (v1 >> 1) ^ (v1 >> 2) ^ (v1 >> 7);
                v2 ^= (v1 << 63) ^ (v1 << 62) ^ (v1 << 57);
 
-               y0 = v2;
-               y1 = v3;
+               pv->y.lo = v2;
+               pv->y.hi = v3;
        }
-
-       br_enc64be(yb, y1);
-       br_enc64be(yb + 8, y0);
 }

diff --git a/src/ext/polyval/polyval.c b/src/ext/polyval/polyval.c
index c74d0330a140b798cd86735d28a91c9cf9794a4d..163d10d8c9537149e5ce215aec655309411ee478 100644 (file)
--- a/src/ext/polyval/polyval.c
+++ b/src/ext/polyval/polyval.c
@@ -20,7 +20,11 @@ static inline void pv_xor(polyval_t *, u128);
 static inline void pv_init_extra(polyval_t *pv);
 
 /* Functions which we expect our multiply implementation to declare. */
-static inline void pv_mul_y_h(polyval_t *);
+/**
+ * Within the polyval struct, perform "y *= h".
+ */
+static
+void pv_mul_y_h(polyval_t *);
 
 #ifdef WORDS_BIG_ENDIAN
 #ifdef __GNUC__
@@ -48,6 +52,9 @@ static inline uint64_t bswap64(uint64_t v)
 #endif
 
 #ifdef PV_USE_CTMUL64
+
+#include "ext/polyval/ctmul64.c"
+
 static inline u128
 u128_from_bytes(const uint8_t *bytes)
 {
@@ -75,17 +82,11 @@ pv_xor(polyval_t *pv, u128 val)
 static inline void
 pv_init_extra(polyval_t *pv)
 {
-
+  pv->hr.lo = rev64(pv->h.lo);
+  pv->hr.hi = rev64(pv->h.hi);
 }
-static inline void pv_mul_y_h(polyval_t *pv)
-{
-
-}
-
-// #include "ext/polyval/ctmul64.c"
 #endif
 
-
 void
 polyval_init(polyval_t *pv, const uint8_t *key)
 {