3 files changed, 86 insertions, 37 deletions

diff --git a/Cryptlib/OpenSSL/crypto/bn/bn_lib.c b/Cryptlib/OpenSSL/crypto/bn/bn_lib.c
index 32a8fbaf..b66f5075 100755
--- a/Cryptlib/OpenSSL/crypto/bn/bn_lib.c
+++ b/Cryptlib/OpenSSL/crypto/bn/bn_lib.c
@@ -824,3 +824,55 @@ int bn_cmp_part_words(const BN_ULONG *a, const BN_ULONG *b,
 		}
 	return bn_cmp_words(a,b,cl);
 	}
+
+/* 
+ * Constant-time conditional swap of a and b.  
+ * a and b are swapped if condition is not 0.  The code assumes that at most one bit of condition is set.
+ * nwords is the number of words to swap.  The code assumes that at least nwords are allocated in both a and b,
+ * and that no more than nwords are used by either a or b.
+ * a and b cannot be the same number
+ */
+void BN_consttime_swap(BN_ULONG condition, BIGNUM *a, BIGNUM *b, int nwords)
+	{
+	BN_ULONG t;
+	int i;
+
+	bn_wcheck_size(a, nwords);
+	bn_wcheck_size(b, nwords);
+
+	assert(a != b);
+	assert((condition & (condition - 1)) == 0);
+	assert(sizeof(BN_ULONG) >= sizeof(int));
+
+	condition = ((condition - 1) >> (BN_BITS2 - 1)) - 1;
+
+	t = (a->top^b->top) & condition;
+	a->top ^= t;
+	b->top ^= t;
+
+#define BN_CONSTTIME_SWAP(ind) \
+	do { \
+		t = (a->d[ind] ^ b->d[ind]) & condition; \
+		a->d[ind] ^= t; \
+		b->d[ind] ^= t; \
+	} while (0)
+
+
+	switch (nwords) {
+	default:
+		for (i = 10; i < nwords; i++) 
+			BN_CONSTTIME_SWAP(i);
+		/* Fallthrough */
+	case 10: BN_CONSTTIME_SWAP(9); /* Fallthrough */
+	case 9: BN_CONSTTIME_SWAP(8); /* Fallthrough */
+	case 8: BN_CONSTTIME_SWAP(7); /* Fallthrough */
+	case 7: BN_CONSTTIME_SWAP(6); /* Fallthrough */
+	case 6: BN_CONSTTIME_SWAP(5); /* Fallthrough */
+	case 5: BN_CONSTTIME_SWAP(4); /* Fallthrough */
+	case 4: BN_CONSTTIME_SWAP(3); /* Fallthrough */
+	case 3: BN_CONSTTIME_SWAP(2); /* Fallthrough */
+	case 2: BN_CONSTTIME_SWAP(1); /* Fallthrough */
+	case 1: BN_CONSTTIME_SWAP(0);
+	}
+#undef BN_CONSTTIME_SWAP
+}
diff --git a/Cryptlib/OpenSSL/crypto/bn/bn_mont.c b/Cryptlib/OpenSSL/crypto/bn/bn_mont.c
index 4799b152..27cafb1f 100755
--- a/Cryptlib/OpenSSL/crypto/bn/bn_mont.c
+++ b/Cryptlib/OpenSSL/crypto/bn/bn_mont.c
@@ -701,32 +701,38 @@ BN_MONT_CTX *BN_MONT_CTX_copy(BN_MONT_CTX *to, BN_MONT_CTX *from)
 BN_MONT_CTX *BN_MONT_CTX_set_locked(BN_MONT_CTX **pmont, int lock,
 					const BIGNUM *mod, BN_CTX *ctx)
 	{
-	int got_write_lock = 0;
 	BN_MONT_CTX *ret;
 
 	CRYPTO_r_lock(lock);
-	if (!*pmont)
+	ret = *pmont;
+	CRYPTO_r_unlock(lock);
+	if (ret)
+		return ret;
+
+	/* We don't want to serialise globally while doing our lazy-init math in
+	 * BN_MONT_CTX_set. That punishes threads that are doing independent
+	 * things. Instead, punish the case where more than one thread tries to
+	 * lazy-init the same 'pmont', by having each do the lazy-init math work
+	 * independently and only use the one from the thread that wins the race
+	 * (the losers throw away the work they've done). */
+	ret = BN_MONT_CTX_new();
+	if (!ret)
+		return NULL;
+	if (!BN_MONT_CTX_set(ret, mod, ctx))
 		{
-		CRYPTO_r_unlock(lock);
-		CRYPTO_w_lock(lock);
-		got_write_lock = 1;
+		BN_MONT_CTX_free(ret);
+		return NULL;
+		}
 
-		if (!*pmont)
-			{
-			ret = BN_MONT_CTX_new();
-			if (ret && !BN_MONT_CTX_set(ret, mod, ctx))
-				BN_MONT_CTX_free(ret);
-			else
-				*pmont = ret;
-			}
+	/* The locked compare-and-set, after the local work is done. */
+	CRYPTO_w_lock(lock);
+	if (*pmont)
+		{
+		BN_MONT_CTX_free(ret);
+		ret = *pmont;
 		}
-	
-	ret = *pmont;
-	
-	if (got_write_lock)
-		CRYPTO_w_unlock(lock);
 	else
-		CRYPTO_r_unlock(lock);
-		
+		*pmont = ret;
+	CRYPTO_w_unlock(lock);
 	return ret;
 	}
diff --git a/Cryptlib/OpenSSL/crypto/bn/bn_word.c b/Cryptlib/OpenSSL/crypto/bn/bn_word.c
index ee7b87c4..de83a15b 100755
--- a/Cryptlib/OpenSSL/crypto/bn/bn_word.c
+++ b/Cryptlib/OpenSSL/crypto/bn/bn_word.c
@@ -144,26 +144,17 @@ int BN_add_word(BIGNUM *a, BN_ULONG w)
 			a->neg=!(a->neg);
 		return(i);
 		}
-	/* Only expand (and risk failing) if it's possibly necessary */
-	if (((BN_ULONG)(a->d[a->top - 1] + 1) == 0) &&
-			(bn_wexpand(a,a->top+1) == NULL))
-		return(0);
-	i=0;
-	for (;;)
+	for (i=0;w!=0 && i<a->top;i++)
 		{
-		if (i >= a->top)
-			l=w;
-		else
-			l=(a->d[i]+w)&BN_MASK2;
-		a->d[i]=l;
-		if (w > l)
-			w=1;
-		else
-			break;
-		i++;
+		a->d[i] = l = (a->d[i]+w)&BN_MASK2;
+		w = (w>l)?1:0;
 		}
-	if (i >= a->top)
+	if (w && i==a->top)
+		{
+		if (bn_wexpand(a,a->top+1) == NULL) return 0;
 		a->top++;
+		a->d[i]=w;
+		}
 	bn_check_top(a);
 	return(1);
 	}