aboutsummaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
authorJack Lloyd <[email protected]>2018-03-20 17:01:50 -0400
committerJack Lloyd <[email protected]>2018-03-20 18:27:12 -0400
commit8a7559e4f8ad179802ef482131faa5dfc0a074ad (patch)
tree3b1ca0dd05d1e7a164d2e3319e42ad11aafc030e /src
parenteb5bc1b1ea573c44efe58b18ed07a82db88029fb (diff)
Add some helpers for handling carries in Ed25519
Diffstat (limited to 'src')
-rw-r--r--src/lib/pubkey/ed25519/ed25519_fe.cpp724
-rw-r--r--src/lib/pubkey/ed25519/ed25519_internal.h34
-rw-r--r--src/lib/pubkey/ed25519/sc_muladd.cpp373
-rw-r--r--src/lib/pubkey/ed25519/sc_reduce.cpp253
4 files changed, 491 insertions, 893 deletions
diff --git a/src/lib/pubkey/ed25519/ed25519_fe.cpp b/src/lib/pubkey/ed25519/ed25519_fe.cpp
index 5aa515e9b..bea9c504b 100644
--- a/src/lib/pubkey/ed25519/ed25519_fe.cpp
+++ b/src/lib/pubkey/ed25519/ed25519_fe.cpp
@@ -114,142 +114,144 @@ With tighter constraints on inputs can squeeze carries into int32.
//static
FE_25519 FE_25519::mul(const FE_25519& f, const FE_25519& g)
{
- int32_t f0 = f[0];
- int32_t f1 = f[1];
- int32_t f2 = f[2];
- int32_t f3 = f[3];
- int32_t f4 = f[4];
- int32_t f5 = f[5];
- int32_t f6 = f[6];
- int32_t f7 = f[7];
- int32_t f8 = f[8];
- int32_t f9 = f[9];
+ const int32_t f0 = f[0];
+ const int32_t f1 = f[1];
+ const int32_t f2 = f[2];
+ const int32_t f3 = f[3];
+ const int32_t f4 = f[4];
+ const int32_t f5 = f[5];
+ const int32_t f6 = f[6];
+ const int32_t f7 = f[7];
+ const int32_t f8 = f[8];
+ const int32_t f9 = f[9];
+
+ const int32_t g0 = g[0];
+ const int32_t g1 = g[1];
+ const int32_t g2 = g[2];
+ const int32_t g3 = g[3];
+ const int32_t g4 = g[4];
+ const int32_t g5 = g[5];
+ const int32_t g6 = g[6];
+ const int32_t g7 = g[7];
+ const int32_t g8 = g[8];
+ const int32_t g9 = g[9];
+
+ const int32_t g1_19 = 19 * g1; /* 1.959375*2^29 */
+ const int32_t g2_19 = 19 * g2; /* 1.959375*2^30; still ok */
+ const int32_t g3_19 = 19 * g3;
+ const int32_t g4_19 = 19 * g4;
+ const int32_t g5_19 = 19 * g5;
+ const int32_t g6_19 = 19 * g6;
+ const int32_t g7_19 = 19 * g7;
+ const int32_t g8_19 = 19 * g8;
+ const int32_t g9_19 = 19 * g9;
+ const int32_t f1_2 = 2 * f1;
+ const int32_t f3_2 = 2 * f3;
+ const int32_t f5_2 = 2 * f5;
+ const int32_t f7_2 = 2 * f7;
+ const int32_t f9_2 = 2 * f9;
+
+ const int64_t f0g0 = f0 * static_cast<int64_t>(g0);
+ const int64_t f0g1 = f0 * static_cast<int64_t>(g1);
+ const int64_t f0g2 = f0 * static_cast<int64_t>(g2);
+ const int64_t f0g3 = f0 * static_cast<int64_t>(g3);
+ const int64_t f0g4 = f0 * static_cast<int64_t>(g4);
+ const int64_t f0g5 = f0 * static_cast<int64_t>(g5);
+ const int64_t f0g6 = f0 * static_cast<int64_t>(g6);
+ const int64_t f0g7 = f0 * static_cast<int64_t>(g7);
+ const int64_t f0g8 = f0 * static_cast<int64_t>(g8);
+ const int64_t f0g9 = f0 * static_cast<int64_t>(g9);
+ const int64_t f1g0 = f1 * static_cast<int64_t>(g0);
+ const int64_t f1g1_2 = f1_2 * static_cast<int64_t>(g1);
+ const int64_t f1g2 = f1 * static_cast<int64_t>(g2);
+ const int64_t f1g3_2 = f1_2 * static_cast<int64_t>(g3);
+ const int64_t f1g4 = f1 * static_cast<int64_t>(g4);
+ const int64_t f1g5_2 = f1_2 * static_cast<int64_t>(g5);
+ const int64_t f1g6 = f1 * static_cast<int64_t>(g6);
+ const int64_t f1g7_2 = f1_2 * static_cast<int64_t>(g7);
+ const int64_t f1g8 = f1 * static_cast<int64_t>(g8);
+ const int64_t f1g9_38 = f1_2 * static_cast<int64_t>(g9_19);
+ const int64_t f2g0 = f2 * static_cast<int64_t>(g0);
+ const int64_t f2g1 = f2 * static_cast<int64_t>(g1);
+ const int64_t f2g2 = f2 * static_cast<int64_t>(g2);
+ const int64_t f2g3 = f2 * static_cast<int64_t>(g3);
+ const int64_t f2g4 = f2 * static_cast<int64_t>(g4);
+ const int64_t f2g5 = f2 * static_cast<int64_t>(g5);
+ const int64_t f2g6 = f2 * static_cast<int64_t>(g6);
+ const int64_t f2g7 = f2 * static_cast<int64_t>(g7);
+ const int64_t f2g8_19 = f2 * static_cast<int64_t>(g8_19);
+ const int64_t f2g9_19 = f2 * static_cast<int64_t>(g9_19);
+ const int64_t f3g0 = f3 * static_cast<int64_t>(g0);
+ const int64_t f3g1_2 = f3_2 * static_cast<int64_t>(g1);
+ const int64_t f3g2 = f3 * static_cast<int64_t>(g2);
+ const int64_t f3g3_2 = f3_2 * static_cast<int64_t>(g3);
+ const int64_t f3g4 = f3 * static_cast<int64_t>(g4);
+ const int64_t f3g5_2 = f3_2 * static_cast<int64_t>(g5);
+ const int64_t f3g6 = f3 * static_cast<int64_t>(g6);
+ const int64_t f3g7_38 = f3_2 * static_cast<int64_t>(g7_19);
+ const int64_t f3g8_19 = f3 * static_cast<int64_t>(g8_19);
+ const int64_t f3g9_38 = f3_2 * static_cast<int64_t>(g9_19);
+ const int64_t f4g0 = f4 * static_cast<int64_t>(g0);
+ const int64_t f4g1 = f4 * static_cast<int64_t>(g1);
+ const int64_t f4g2 = f4 * static_cast<int64_t>(g2);
+ const int64_t f4g3 = f4 * static_cast<int64_t>(g3);
+ const int64_t f4g4 = f4 * static_cast<int64_t>(g4);
+ const int64_t f4g5 = f4 * static_cast<int64_t>(g5);
+ const int64_t f4g6_19 = f4 * static_cast<int64_t>(g6_19);
+ const int64_t f4g7_19 = f4 * static_cast<int64_t>(g7_19);
+ const int64_t f4g8_19 = f4 * static_cast<int64_t>(g8_19);
+ const int64_t f4g9_19 = f4 * static_cast<int64_t>(g9_19);
+ const int64_t f5g0 = f5 * static_cast<int64_t>(g0);
+ const int64_t f5g1_2 = f5_2 * static_cast<int64_t>(g1);
+ const int64_t f5g2 = f5 * static_cast<int64_t>(g2);
+ const int64_t f5g3_2 = f5_2 * static_cast<int64_t>(g3);
+ const int64_t f5g4 = f5 * static_cast<int64_t>(g4);
+ const int64_t f5g5_38 = f5_2 * static_cast<int64_t>(g5_19);
+ const int64_t f5g6_19 = f5 * static_cast<int64_t>(g6_19);
+ const int64_t f5g7_38 = f5_2 * static_cast<int64_t>(g7_19);
+ const int64_t f5g8_19 = f5 * static_cast<int64_t>(g8_19);
+ const int64_t f5g9_38 = f5_2 * static_cast<int64_t>(g9_19);
+ const int64_t f6g0 = f6 * static_cast<int64_t>(g0);
+ const int64_t f6g1 = f6 * static_cast<int64_t>(g1);
+ const int64_t f6g2 = f6 * static_cast<int64_t>(g2);
+ const int64_t f6g3 = f6 * static_cast<int64_t>(g3);
+ const int64_t f6g4_19 = f6 * static_cast<int64_t>(g4_19);
+ const int64_t f6g5_19 = f6 * static_cast<int64_t>(g5_19);
+ const int64_t f6g6_19 = f6 * static_cast<int64_t>(g6_19);
+ const int64_t f6g7_19 = f6 * static_cast<int64_t>(g7_19);
+ const int64_t f6g8_19 = f6 * static_cast<int64_t>(g8_19);
+ const int64_t f6g9_19 = f6 * static_cast<int64_t>(g9_19);
+ const int64_t f7g0 = f7 * static_cast<int64_t>(g0);
+ const int64_t f7g1_2 = f7_2 * static_cast<int64_t>(g1);
+ const int64_t f7g2 = f7 * static_cast<int64_t>(g2);
+ const int64_t f7g3_38 = f7_2 * static_cast<int64_t>(g3_19);
+ const int64_t f7g4_19 = f7 * static_cast<int64_t>(g4_19);
+ const int64_t f7g5_38 = f7_2 * static_cast<int64_t>(g5_19);
+ const int64_t f7g6_19 = f7 * static_cast<int64_t>(g6_19);
+ const int64_t f7g7_38 = f7_2 * static_cast<int64_t>(g7_19);
+ const int64_t f7g8_19 = f7 * static_cast<int64_t>(g8_19);
+ const int64_t f7g9_38 = f7_2 * static_cast<int64_t>(g9_19);
+ const int64_t f8g0 = f8 * static_cast<int64_t>(g0);
+ const int64_t f8g1 = f8 * static_cast<int64_t>(g1);
+ const int64_t f8g2_19 = f8 * static_cast<int64_t>(g2_19);
+ const int64_t f8g3_19 = f8 * static_cast<int64_t>(g3_19);
+ const int64_t f8g4_19 = f8 * static_cast<int64_t>(g4_19);
+ const int64_t f8g5_19 = f8 * static_cast<int64_t>(g5_19);
+ const int64_t f8g6_19 = f8 * static_cast<int64_t>(g6_19);
+ const int64_t f8g7_19 = f8 * static_cast<int64_t>(g7_19);
+ const int64_t f8g8_19 = f8 * static_cast<int64_t>(g8_19);
+ const int64_t f8g9_19 = f8 * static_cast<int64_t>(g9_19);
+ const int64_t f9g0 = f9 * static_cast<int64_t>(g0);
+ const int64_t f9g1_38 = f9_2 * static_cast<int64_t>(g1_19);
+ const int64_t f9g2_19 = f9 * static_cast<int64_t>(g2_19);
+ const int64_t f9g3_38 = f9_2 * static_cast<int64_t>(g3_19);
+ const int64_t f9g4_19 = f9 * static_cast<int64_t>(g4_19);
+ const int64_t f9g5_38 = f9_2 * static_cast<int64_t>(g5_19);
+ const int64_t f9g6_19 = f9 * static_cast<int64_t>(g6_19);
+ const int64_t f9g7_38 = f9_2 * static_cast<int64_t>(g7_19);
+ const int64_t f9g8_19 = f9 * static_cast<int64_t>(g8_19);
+ const int64_t f9g9_38 = f9_2 * static_cast<int64_t>(g9_19);
- int32_t g0 = g[0];
- int32_t g1 = g[1];
- int32_t g2 = g[2];
- int32_t g3 = g[3];
- int32_t g4 = g[4];
- int32_t g5 = g[5];
- int32_t g6 = g[6];
- int32_t g7 = g[7];
- int32_t g8 = g[8];
- int32_t g9 = g[9];
-
- int32_t g1_19 = 19 * g1; /* 1.959375*2^29 */
- int32_t g2_19 = 19 * g2; /* 1.959375*2^30; still ok */
- int32_t g3_19 = 19 * g3;
- int32_t g4_19 = 19 * g4;
- int32_t g5_19 = 19 * g5;
- int32_t g6_19 = 19 * g6;
- int32_t g7_19 = 19 * g7;
- int32_t g8_19 = 19 * g8;
- int32_t g9_19 = 19 * g9;
- int32_t f1_2 = 2 * f1;
- int32_t f3_2 = 2 * f3;
- int32_t f5_2 = 2 * f5;
- int32_t f7_2 = 2 * f7;
- int32_t f9_2 = 2 * f9;
- int64_t f0g0 = f0 * static_cast<int64_t>(g0);
- int64_t f0g1 = f0 * static_cast<int64_t>(g1);
- int64_t f0g2 = f0 * static_cast<int64_t>(g2);
- int64_t f0g3 = f0 * static_cast<int64_t>(g3);
- int64_t f0g4 = f0 * static_cast<int64_t>(g4);
- int64_t f0g5 = f0 * static_cast<int64_t>(g5);
- int64_t f0g6 = f0 * static_cast<int64_t>(g6);
- int64_t f0g7 = f0 * static_cast<int64_t>(g7);
- int64_t f0g8 = f0 * static_cast<int64_t>(g8);
- int64_t f0g9 = f0 * static_cast<int64_t>(g9);
- int64_t f1g0 = f1 * static_cast<int64_t>(g0);
- int64_t f1g1_2 = f1_2 * static_cast<int64_t>(g1);
- int64_t f1g2 = f1 * static_cast<int64_t>(g2);
- int64_t f1g3_2 = f1_2 * static_cast<int64_t>(g3);
- int64_t f1g4 = f1 * static_cast<int64_t>(g4);
- int64_t f1g5_2 = f1_2 * static_cast<int64_t>(g5);
- int64_t f1g6 = f1 * static_cast<int64_t>(g6);
- int64_t f1g7_2 = f1_2 * static_cast<int64_t>(g7);
- int64_t f1g8 = f1 * static_cast<int64_t>(g8);
- int64_t f1g9_38 = f1_2 * static_cast<int64_t>(g9_19);
- int64_t f2g0 = f2 * static_cast<int64_t>(g0);
- int64_t f2g1 = f2 * static_cast<int64_t>(g1);
- int64_t f2g2 = f2 * static_cast<int64_t>(g2);
- int64_t f2g3 = f2 * static_cast<int64_t>(g3);
- int64_t f2g4 = f2 * static_cast<int64_t>(g4);
- int64_t f2g5 = f2 * static_cast<int64_t>(g5);
- int64_t f2g6 = f2 * static_cast<int64_t>(g6);
- int64_t f2g7 = f2 * static_cast<int64_t>(g7);
- int64_t f2g8_19 = f2 * static_cast<int64_t>(g8_19);
- int64_t f2g9_19 = f2 * static_cast<int64_t>(g9_19);
- int64_t f3g0 = f3 * static_cast<int64_t>(g0);
- int64_t f3g1_2 = f3_2 * static_cast<int64_t>(g1);
- int64_t f3g2 = f3 * static_cast<int64_t>(g2);
- int64_t f3g3_2 = f3_2 * static_cast<int64_t>(g3);
- int64_t f3g4 = f3 * static_cast<int64_t>(g4);
- int64_t f3g5_2 = f3_2 * static_cast<int64_t>(g5);
- int64_t f3g6 = f3 * static_cast<int64_t>(g6);
- int64_t f3g7_38 = f3_2 * static_cast<int64_t>(g7_19);
- int64_t f3g8_19 = f3 * static_cast<int64_t>(g8_19);
- int64_t f3g9_38 = f3_2 * static_cast<int64_t>(g9_19);
- int64_t f4g0 = f4 * static_cast<int64_t>(g0);
- int64_t f4g1 = f4 * static_cast<int64_t>(g1);
- int64_t f4g2 = f4 * static_cast<int64_t>(g2);
- int64_t f4g3 = f4 * static_cast<int64_t>(g3);
- int64_t f4g4 = f4 * static_cast<int64_t>(g4);
- int64_t f4g5 = f4 * static_cast<int64_t>(g5);
- int64_t f4g6_19 = f4 * static_cast<int64_t>(g6_19);
- int64_t f4g7_19 = f4 * static_cast<int64_t>(g7_19);
- int64_t f4g8_19 = f4 * static_cast<int64_t>(g8_19);
- int64_t f4g9_19 = f4 * static_cast<int64_t>(g9_19);
- int64_t f5g0 = f5 * static_cast<int64_t>(g0);
- int64_t f5g1_2 = f5_2 * static_cast<int64_t>(g1);
- int64_t f5g2 = f5 * static_cast<int64_t>(g2);
- int64_t f5g3_2 = f5_2 * static_cast<int64_t>(g3);
- int64_t f5g4 = f5 * static_cast<int64_t>(g4);
- int64_t f5g5_38 = f5_2 * static_cast<int64_t>(g5_19);
- int64_t f5g6_19 = f5 * static_cast<int64_t>(g6_19);
- int64_t f5g7_38 = f5_2 * static_cast<int64_t>(g7_19);
- int64_t f5g8_19 = f5 * static_cast<int64_t>(g8_19);
- int64_t f5g9_38 = f5_2 * static_cast<int64_t>(g9_19);
- int64_t f6g0 = f6 * static_cast<int64_t>(g0);
- int64_t f6g1 = f6 * static_cast<int64_t>(g1);
- int64_t f6g2 = f6 * static_cast<int64_t>(g2);
- int64_t f6g3 = f6 * static_cast<int64_t>(g3);
- int64_t f6g4_19 = f6 * static_cast<int64_t>(g4_19);
- int64_t f6g5_19 = f6 * static_cast<int64_t>(g5_19);
- int64_t f6g6_19 = f6 * static_cast<int64_t>(g6_19);
- int64_t f6g7_19 = f6 * static_cast<int64_t>(g7_19);
- int64_t f6g8_19 = f6 * static_cast<int64_t>(g8_19);
- int64_t f6g9_19 = f6 * static_cast<int64_t>(g9_19);
- int64_t f7g0 = f7 * static_cast<int64_t>(g0);
- int64_t f7g1_2 = f7_2 * static_cast<int64_t>(g1);
- int64_t f7g2 = f7 * static_cast<int64_t>(g2);
- int64_t f7g3_38 = f7_2 * static_cast<int64_t>(g3_19);
- int64_t f7g4_19 = f7 * static_cast<int64_t>(g4_19);
- int64_t f7g5_38 = f7_2 * static_cast<int64_t>(g5_19);
- int64_t f7g6_19 = f7 * static_cast<int64_t>(g6_19);
- int64_t f7g7_38 = f7_2 * static_cast<int64_t>(g7_19);
- int64_t f7g8_19 = f7 * static_cast<int64_t>(g8_19);
- int64_t f7g9_38 = f7_2 * static_cast<int64_t>(g9_19);
- int64_t f8g0 = f8 * static_cast<int64_t>(g0);
- int64_t f8g1 = f8 * static_cast<int64_t>(g1);
- int64_t f8g2_19 = f8 * static_cast<int64_t>(g2_19);
- int64_t f8g3_19 = f8 * static_cast<int64_t>(g3_19);
- int64_t f8g4_19 = f8 * static_cast<int64_t>(g4_19);
- int64_t f8g5_19 = f8 * static_cast<int64_t>(g5_19);
- int64_t f8g6_19 = f8 * static_cast<int64_t>(g6_19);
- int64_t f8g7_19 = f8 * static_cast<int64_t>(g7_19);
- int64_t f8g8_19 = f8 * static_cast<int64_t>(g8_19);
- int64_t f8g9_19 = f8 * static_cast<int64_t>(g9_19);
- int64_t f9g0 = f9 * static_cast<int64_t>(g0);
- int64_t f9g1_38 = f9_2 * static_cast<int64_t>(g1_19);
- int64_t f9g2_19 = f9 * static_cast<int64_t>(g2_19);
- int64_t f9g3_38 = f9_2 * static_cast<int64_t>(g3_19);
- int64_t f9g4_19 = f9 * static_cast<int64_t>(g4_19);
- int64_t f9g5_38 = f9_2 * static_cast<int64_t>(g5_19);
- int64_t f9g6_19 = f9 * static_cast<int64_t>(g6_19);
- int64_t f9g7_38 = f9_2 * static_cast<int64_t>(g7_19);
- int64_t f9g8_19 = f9 * static_cast<int64_t>(g8_19);
- int64_t f9g9_38 = f9_2 * static_cast<int64_t>(g9_19);
int64_t h0 = f0g0+f1g9_38+f2g8_19+f3g7_38+f4g6_19+f5g5_38+f6g4_19+f7g3_38+f8g2_19+f9g1_38;
int64_t h1 = f0g1+f1g0 +f2g9_19+f3g8_19+f4g7_19+f5g6_19+f6g5_19+f7g4_19+f8g3_19+f9g2_19;
int64_t h2 = f0g2+f1g1_2 +f2g0 +f3g9_38+f4g8_19+f5g7_38+f6g6_19+f7g5_38+f8g4_19+f9g3_38;
@@ -260,16 +262,6 @@ FE_25519 FE_25519::mul(const FE_25519& f, const FE_25519& g)
int64_t h7 = f0g7+f1g6 +f2g5 +f3g4 +f4g3 +f5g2 +f6g1 +f7g0 +f8g9_19+f9g8_19;
int64_t h8 = f0g8+f1g7_2 +f2g6 +f3g5_2 +f4g4 +f5g3_2 +f6g2 +f7g1_2 +f8g0 +f9g9_38;
int64_t h9 = f0g9+f1g8 +f2g7 +f3g6 +f4g5 +f5g4 +f6g3 +f7g2 +f8g1 +f9g0 ;
- int64_t carry0;
- int64_t carry1;
- int64_t carry2;
- int64_t carry3;
- int64_t carry4;
- int64_t carry5;
- int64_t carry6;
- int64_t carry7;
- int64_t carry8;
- int64_t carry9;
/*
|h0| <= (1.65*1.65*2^52*(1+19+19+19+19)+1.65*1.65*2^50*(38+38+38+38+38))
@@ -277,74 +269,49 @@ FE_25519 FE_25519::mul(const FE_25519& f, const FE_25519& g)
|h1| <= (1.65*1.65*2^51*(1+1+19+19+19+19+19+19+19+19))
i.e. |h1| <= 1.7*2^59; narrower ranges for h3, h5, h7, h9
*/
- const int64_t X24 = (1 << 24);
- const int64_t X25 = (1 << 25);
- const int64_t X26 = (1 << 26);
+ carry<26>(h0, h1);
+ carry<26>(h4, h5);
- carry0 = (h0 + X25) >> 26;
- h1 += carry0;
- h0 -= carry0 * X26;
- carry4 = (h4 + X25) >> 26;
- h5 += carry4;
- h4 -= carry4 * X26;
/* |h0| <= 2^25 */
/* |h4| <= 2^25 */
/* |h1| <= 1.71*2^59 */
/* |h5| <= 1.71*2^59 */
- carry1 = (h1 + X24) >> 25;
- h2 += carry1;
- h1 -= carry1 * X25;
- carry5 = (h5 + X24) >> 25;
- h6 += carry5;
- h5 -= carry5 * X25;
+ carry<25>(h1, h2);
+ carry<25>(h5, h6);
+
/* |h1| <= 2^24; from now on fits into int32 */
/* |h5| <= 2^24; from now on fits into int32 */
/* |h2| <= 1.41*2^60 */
/* |h6| <= 1.41*2^60 */
- carry2 = (h2 + X25) >> 26;
- h3 += carry2;
- h2 -= carry2 * X26;
- carry6 = (h6 + X25) >> 26;
- h7 += carry6;
- h6 -= carry6 * X26;
+ carry<26>(h2, h3);
+ carry<26>(h6, h7);
/* |h2| <= 2^25; from now on fits into int32 unchanged */
/* |h6| <= 2^25; from now on fits into int32 unchanged */
/* |h3| <= 1.71*2^59 */
/* |h7| <= 1.71*2^59 */
- carry3 = (h3 + X24) >> 25;
- h4 += carry3;
- h3 -= carry3 * X25;
- carry7 = (h7 + X24) >> 25;
- h8 += carry7;
- h7 -= carry7 * X25;
+ carry<25>(h3, h4);
+ carry<25>(h7, h8);
/* |h3| <= 2^24; from now on fits into int32 unchanged */
/* |h7| <= 2^24; from now on fits into int32 unchanged */
/* |h4| <= 1.72*2^34 */
/* |h8| <= 1.41*2^60 */
- carry4 = (h4 + X25) >> 26;
- h5 += carry4;
- h4 -= carry4 * X26;
- carry8 = (h8 + X25) >> 26;
- h9 += carry8;
- h8 -= carry8 * X26;
+ carry<26>(h4, h5);
+ carry<26>(h8, h9);
/* |h4| <= 2^25; from now on fits into int32 unchanged */
/* |h8| <= 2^25; from now on fits into int32 unchanged */
/* |h5| <= 1.01*2^24 */
/* |h9| <= 1.71*2^59 */
- carry9 = (h9 + X24) >> 25;
- h0 += carry9 * 19;
- h9 -= carry9 * X25;
+ carry<25, 19>(h9, h0);
+
/* |h9| <= 2^24; from now on fits into int32 unchanged */
/* |h0| <= 1.1*2^39 */
- carry0 = (h0 + X25) >> 26;
- h1 += carry0;
- h0 -= carry0 * X26;
+ carry<26>(h0, h1);
/* |h0| <= 2^25; from now on fits into int32 unchanged */
/* |h1| <= 1.01*2^24 */
@@ -369,10 +336,6 @@ See fe_mul.c for discussion of implementation strategy.
//static
FE_25519 FE_25519::sqr_iter(const FE_25519& f, size_t iter)
{
- const int64_t X24 = (1 << 24);
- const int64_t X25 = (1 << 25);
- const int64_t X26 = (1 << 26);
-
int32_t f0 = f[0];
int32_t f1 = f[1];
int32_t f2 = f[2];
@@ -467,58 +430,20 @@ FE_25519 FE_25519::sqr_iter(const FE_25519& f, size_t iter)
int64_t h8 = f0f8_2+f1f7_4 +f2f6_2 +f3f5_4 +f4f4 +f9f9_38;
int64_t h9 = f0f9_2+f1f8_2 +f2f7_2 +f3f6_2 +f4f5_2;
- int64_t carry0;
- int64_t carry1;
- int64_t carry2;
- int64_t carry3;
- int64_t carry4;
- int64_t carry5;
- int64_t carry6;
- int64_t carry7;
- int64_t carry8;
- int64_t carry9;
-
- carry0 = (h0 + X25) >> 26;
- h1 += carry0;
- h0 -= carry0 * X26;
- carry4 = (h4 + X25) >> 26;
- h5 += carry4;
- h4 -= carry4 * X26;
-
- carry1 = (h1 + X24) >> 25;
- h2 += carry1;
- h1 -= carry1 * X25;
- carry5 = (h5 + X24) >> 25;
- h6 += carry5;
- h5 -= carry5 * X25;
-
- carry2 = (h2 + X25) >> 26;
- h3 += carry2;
- h2 -= carry2 * X26;
- carry6 = (h6 + X25) >> 26;
- h7 += carry6;
- h6 -= carry6 * X26;
-
- carry3 = (h3 + X24) >> 25;
- h4 += carry3;
- h3 -= carry3 * X25;
- carry7 = (h7 + X24) >> 25;
- h8 += carry7;
- h7 -= carry7 * X25;
-
- carry4 = (h4 + X25) >> 26;
- h5 += carry4;
- h4 -= carry4 * X26;
- carry8 = (h8 + X25) >> 26;
- h9 += carry8;
- h8 -= carry8 * X26;
-
- carry9 = (h9 + X24) >> 25;
- h0 += carry9 * 19;
- h9 -= carry9 * X25;
- carry0 = (h0 + X25) >> 26;
- h1 += carry0;
- h0 -= carry0 * X26;
+ carry<26>(h0, h1);
+ carry<26>(h4, h5);
+ carry<25>(h1, h2);
+ carry<25>(h5, h6);
+ carry<26>(h2, h3);
+ carry<26>(h6, h7);
+
+ carry<25>(h3, h4);
+ carry<25>(h7, h8);
+
+ carry<26>(h4, h5);
+ carry<26>(h8, h9);
+ carry<25,19>(h9, h0);
+ carry<26>(h0, h1);
f0 = h0;
f1 = h1;
@@ -553,88 +478,85 @@ See fe_mul.c for discussion of implementation strategy.
//static
FE_25519 FE_25519::sqr2(const FE_25519& f)
{
- const int64_t X24 = (1 << 24);
- const int64_t X25 = (1 << 25);
- const int64_t X26 = (1 << 26);
+ const int32_t f0 = f[0];
+ const int32_t f1 = f[1];
+ const int32_t f2 = f[2];
+ const int32_t f3 = f[3];
+ const int32_t f4 = f[4];
+ const int32_t f5 = f[5];
+ const int32_t f6 = f[6];
+ const int32_t f7 = f[7];
+ const int32_t f8 = f[8];
+ const int32_t f9 = f[9];
+ const int32_t f0_2 = 2 * f0;
+ const int32_t f1_2 = 2 * f1;
+ const int32_t f2_2 = 2 * f2;
+ const int32_t f3_2 = 2 * f3;
+ const int32_t f4_2 = 2 * f4;
+ const int32_t f5_2 = 2 * f5;
+ const int32_t f6_2 = 2 * f6;
+ const int32_t f7_2 = 2 * f7;
+ const int32_t f5_38 = 38 * f5; /* 1.959375*2^30 */
+ const int32_t f6_19 = 19 * f6; /* 1.959375*2^30 */
+ const int32_t f7_38 = 38 * f7; /* 1.959375*2^30 */
+ const int32_t f8_19 = 19 * f8; /* 1.959375*2^30 */
+ const int32_t f9_38 = 38 * f9; /* 1.959375*2^30 */
+ const int64_t f0f0 = f0 * static_cast<int64_t>(f0);
+ const int64_t f0f1_2 = f0_2 * static_cast<int64_t>(f1);
+ const int64_t f0f2_2 = f0_2 * static_cast<int64_t>(f2);
+ const int64_t f0f3_2 = f0_2 * static_cast<int64_t>(f3);
+ const int64_t f0f4_2 = f0_2 * static_cast<int64_t>(f4);
+ const int64_t f0f5_2 = f0_2 * static_cast<int64_t>(f5);
+ const int64_t f0f6_2 = f0_2 * static_cast<int64_t>(f6);
+ const int64_t f0f7_2 = f0_2 * static_cast<int64_t>(f7);
+ const int64_t f0f8_2 = f0_2 * static_cast<int64_t>(f8);
+ const int64_t f0f9_2 = f0_2 * static_cast<int64_t>(f9);
+ const int64_t f1f1_2 = f1_2 * static_cast<int64_t>(f1);
+ const int64_t f1f2_2 = f1_2 * static_cast<int64_t>(f2);
+ const int64_t f1f3_4 = f1_2 * static_cast<int64_t>(f3_2);
+ const int64_t f1f4_2 = f1_2 * static_cast<int64_t>(f4);
+ const int64_t f1f5_4 = f1_2 * static_cast<int64_t>(f5_2);
+ const int64_t f1f6_2 = f1_2 * static_cast<int64_t>(f6);
+ const int64_t f1f7_4 = f1_2 * static_cast<int64_t>(f7_2);
+ const int64_t f1f8_2 = f1_2 * static_cast<int64_t>(f8);
+ const int64_t f1f9_76 = f1_2 * static_cast<int64_t>(f9_38);
+ const int64_t f2f2 = f2 * static_cast<int64_t>(f2);
+ const int64_t f2f3_2 = f2_2 * static_cast<int64_t>(f3);
+ const int64_t f2f4_2 = f2_2 * static_cast<int64_t>(f4);
+ const int64_t f2f5_2 = f2_2 * static_cast<int64_t>(f5);
+ const int64_t f2f6_2 = f2_2 * static_cast<int64_t>(f6);
+ const int64_t f2f7_2 = f2_2 * static_cast<int64_t>(f7);
+ const int64_t f2f8_38 = f2_2 * static_cast<int64_t>(f8_19);
+ const int64_t f2f9_38 = f2 * static_cast<int64_t>(f9_38);
+ const int64_t f3f3_2 = f3_2 * static_cast<int64_t>(f3);
+ const int64_t f3f4_2 = f3_2 * static_cast<int64_t>(f4);
+ const int64_t f3f5_4 = f3_2 * static_cast<int64_t>(f5_2);
+ const int64_t f3f6_2 = f3_2 * static_cast<int64_t>(f6);
+ const int64_t f3f7_76 = f3_2 * static_cast<int64_t>(f7_38);
+ const int64_t f3f8_38 = f3_2 * static_cast<int64_t>(f8_19);
+ const int64_t f3f9_76 = f3_2 * static_cast<int64_t>(f9_38);
+ const int64_t f4f4 = f4 * static_cast<int64_t>(f4);
+ const int64_t f4f5_2 = f4_2 * static_cast<int64_t>(f5);
+ const int64_t f4f6_38 = f4_2 * static_cast<int64_t>(f6_19);
+ const int64_t f4f7_38 = f4 * static_cast<int64_t>(f7_38);
+ const int64_t f4f8_38 = f4_2 * static_cast<int64_t>(f8_19);
+ const int64_t f4f9_38 = f4 * static_cast<int64_t>(f9_38);
+ const int64_t f5f5_38 = f5 * static_cast<int64_t>(f5_38);
+ const int64_t f5f6_38 = f5_2 * static_cast<int64_t>(f6_19);
+ const int64_t f5f7_76 = f5_2 * static_cast<int64_t>(f7_38);
+ const int64_t f5f8_38 = f5_2 * static_cast<int64_t>(f8_19);
+ const int64_t f5f9_76 = f5_2 * static_cast<int64_t>(f9_38);
+ const int64_t f6f6_19 = f6 * static_cast<int64_t>(f6_19);
+ const int64_t f6f7_38 = f6 * static_cast<int64_t>(f7_38);
+ const int64_t f6f8_38 = f6_2 * static_cast<int64_t>(f8_19);
+ const int64_t f6f9_38 = f6 * static_cast<int64_t>(f9_38);
+ const int64_t f7f7_38 = f7 * static_cast<int64_t>(f7_38);
+ const int64_t f7f8_38 = f7_2 * static_cast<int64_t>(f8_19);
+ const int64_t f7f9_76 = f7_2 * static_cast<int64_t>(f9_38);
+ const int64_t f8f8_19 = f8 * static_cast<int64_t>(f8_19);
+ const int64_t f8f9_38 = f8 * static_cast<int64_t>(f9_38);
+ const int64_t f9f9_38 = f9 * static_cast<int64_t>(f9_38);
- int32_t f0 = f[0];
- int32_t f1 = f[1];
- int32_t f2 = f[2];
- int32_t f3 = f[3];
- int32_t f4 = f[4];
- int32_t f5 = f[5];
- int32_t f6 = f[6];
- int32_t f7 = f[7];
- int32_t f8 = f[8];
- int32_t f9 = f[9];
- int32_t f0_2 = 2 * f0;
- int32_t f1_2 = 2 * f1;
- int32_t f2_2 = 2 * f2;
- int32_t f3_2 = 2 * f3;
- int32_t f4_2 = 2 * f4;
- int32_t f5_2 = 2 * f5;
- int32_t f6_2 = 2 * f6;
- int32_t f7_2 = 2 * f7;
- int32_t f5_38 = 38 * f5; /* 1.959375*2^30 */
- int32_t f6_19 = 19 * f6; /* 1.959375*2^30 */
- int32_t f7_38 = 38 * f7; /* 1.959375*2^30 */
- int32_t f8_19 = 19 * f8; /* 1.959375*2^30 */
- int32_t f9_38 = 38 * f9; /* 1.959375*2^30 */
- int64_t f0f0 = f0 * static_cast<int64_t>(f0);
- int64_t f0f1_2 = f0_2 * static_cast<int64_t>(f1);
- int64_t f0f2_2 = f0_2 * static_cast<int64_t>(f2);
- int64_t f0f3_2 = f0_2 * static_cast<int64_t>(f3);
- int64_t f0f4_2 = f0_2 * static_cast<int64_t>(f4);
- int64_t f0f5_2 = f0_2 * static_cast<int64_t>(f5);
- int64_t f0f6_2 = f0_2 * static_cast<int64_t>(f6);
- int64_t f0f7_2 = f0_2 * static_cast<int64_t>(f7);
- int64_t f0f8_2 = f0_2 * static_cast<int64_t>(f8);
- int64_t f0f9_2 = f0_2 * static_cast<int64_t>(f9);
- int64_t f1f1_2 = f1_2 * static_cast<int64_t>(f1);
- int64_t f1f2_2 = f1_2 * static_cast<int64_t>(f2);
- int64_t f1f3_4 = f1_2 * static_cast<int64_t>(f3_2);
- int64_t f1f4_2 = f1_2 * static_cast<int64_t>(f4);
- int64_t f1f5_4 = f1_2 * static_cast<int64_t>(f5_2);
- int64_t f1f6_2 = f1_2 * static_cast<int64_t>(f6);
- int64_t f1f7_4 = f1_2 * static_cast<int64_t>(f7_2);
- int64_t f1f8_2 = f1_2 * static_cast<int64_t>(f8);
- int64_t f1f9_76 = f1_2 * static_cast<int64_t>(f9_38);
- int64_t f2f2 = f2 * static_cast<int64_t>(f2);
- int64_t f2f3_2 = f2_2 * static_cast<int64_t>(f3);
- int64_t f2f4_2 = f2_2 * static_cast<int64_t>(f4);
- int64_t f2f5_2 = f2_2 * static_cast<int64_t>(f5);
- int64_t f2f6_2 = f2_2 * static_cast<int64_t>(f6);
- int64_t f2f7_2 = f2_2 * static_cast<int64_t>(f7);
- int64_t f2f8_38 = f2_2 * static_cast<int64_t>(f8_19);
- int64_t f2f9_38 = f2 * static_cast<int64_t>(f9_38);
- int64_t f3f3_2 = f3_2 * static_cast<int64_t>(f3);
- int64_t f3f4_2 = f3_2 * static_cast<int64_t>(f4);
- int64_t f3f5_4 = f3_2 * static_cast<int64_t>(f5_2);
- int64_t f3f6_2 = f3_2 * static_cast<int64_t>(f6);
- int64_t f3f7_76 = f3_2 * static_cast<int64_t>(f7_38);
- int64_t f3f8_38 = f3_2 * static_cast<int64_t>(f8_19);
- int64_t f3f9_76 = f3_2 * static_cast<int64_t>(f9_38);
- int64_t f4f4 = f4 * static_cast<int64_t>(f4);
- int64_t f4f5_2 = f4_2 * static_cast<int64_t>(f5);
- int64_t f4f6_38 = f4_2 * static_cast<int64_t>(f6_19);
- int64_t f4f7_38 = f4 * static_cast<int64_t>(f7_38);
- int64_t f4f8_38 = f4_2 * static_cast<int64_t>(f8_19);
- int64_t f4f9_38 = f4 * static_cast<int64_t>(f9_38);
- int64_t f5f5_38 = f5 * static_cast<int64_t>(f5_38);
- int64_t f5f6_38 = f5_2 * static_cast<int64_t>(f6_19);
- int64_t f5f7_76 = f5_2 * static_cast<int64_t>(f7_38);
- int64_t f5f8_38 = f5_2 * static_cast<int64_t>(f8_19);
- int64_t f5f9_76 = f5_2 * static_cast<int64_t>(f9_38);
- int64_t f6f6_19 = f6 * static_cast<int64_t>(f6_19);
- int64_t f6f7_38 = f6 * static_cast<int64_t>(f7_38);
- int64_t f6f8_38 = f6_2 * static_cast<int64_t>(f8_19);
- int64_t f6f9_38 = f6 * static_cast<int64_t>(f9_38);
- int64_t f7f7_38 = f7 * static_cast<int64_t>(f7_38);
- int64_t f7f8_38 = f7_2 * static_cast<int64_t>(f8_19);
- int64_t f7f9_76 = f7_2 * static_cast<int64_t>(f9_38);
- int64_t f8f8_19 = f8 * static_cast<int64_t>(f8_19);
- int64_t f8f9_38 = f8 * static_cast<int64_t>(f9_38);
- int64_t f9f9_38 = f9 * static_cast<int64_t>(f9_38);
int64_t h0 = f0f0 +f1f9_76+f2f8_38+f3f7_76+f4f6_38+f5f5_38;
int64_t h1 = f0f1_2+f2f9_38+f3f8_38+f4f7_38+f5f6_38;
int64_t h2 = f0f2_2+f1f1_2 +f3f9_76+f4f8_38+f5f7_76+f6f6_19;
@@ -645,16 +567,6 @@ FE_25519 FE_25519::sqr2(const FE_25519& f)
int64_t h7 = f0f7_2+f1f6_2 +f2f5_2 +f3f4_2 +f8f9_38;
int64_t h8 = f0f8_2+f1f7_4 +f2f6_2 +f3f5_4 +f4f4 +f9f9_38;
int64_t h9 = f0f9_2+f1f8_2 +f2f7_2 +f3f6_2 +f4f5_2;
- int64_t carry0;
- int64_t carry1;
- int64_t carry2;
- int64_t carry3;
- int64_t carry4;
- int64_t carry5;
- int64_t carry6;
- int64_t carry7;
- int64_t carry8;
- int64_t carry9;
h0 += h0;
h1 += h1;
@@ -667,48 +579,21 @@ FE_25519 FE_25519::sqr2(const FE_25519& f)
h8 += h8;
h9 += h9;
- carry0 = (h0 + X25) >> 26;
- h1 += carry0;
- h0 -= carry0 * X26;
- carry4 = (h4 + X25) >> 26;
- h5 += carry4;
- h4 -= carry4 * X26;
-
- carry1 = (h1 + X24) >> 25;
- h2 += carry1;
- h1 -= carry1 * X25;
- carry5 = (h5 + X24) >> 25;
- h6 += carry5;
- h5 -= carry5 * X25;
-
- carry2 = (h2 + X25) >> 26;
- h3 += carry2;
- h2 -= carry2 * X26;
- carry6 = (h6 + X25) >> 26;
- h7 += carry6;
- h6 -= carry6 * X26;
-
- carry3 = (h3 + X24) >> 25;
- h4 += carry3;
- h3 -= carry3 * X25;
- carry7 = (h7 + X24) >> 25;
- h8 += carry7;
- h7 -= carry7 * X25;
-
- carry4 = (h4 + X25) >> 26;
- h5 += carry4;
- h4 -= carry4 * X26;
- carry8 = (h8 + X25) >> 26;
- h9 += carry8;
- h8 -= carry8 * X26;
-
- carry9 = (h9 + X24) >> 25;
- h0 += carry9 * 19;
- h9 -= carry9 * X25;
+ carry<26>(h0, h1);
+ carry<26>(h4, h5);
+
+ carry<25>(h1, h2);
+ carry<25>(h5, h6);
+
+ carry<26>(h2, h3);
+ carry<26>(h6, h7);
- carry0 = (h0 + X25) >> 26;
- h1 += carry0;
- h0 -= carry0 * X26;
+ carry<25>(h3, h4);
+ carry<25>(h7, h8);
+ carry<26>(h4, h5);
+ carry<26>(h8, h9);
+ carry<25,19>(h9, h0);
+ carry<26>(h0, h1);
return FE_25519(h0, h1, h2, h3, h4, h5, h6, h7, h8, h9);
}
@@ -719,10 +604,6 @@ Ignores top bit of h.
void FE_25519::from_bytes(const uint8_t s[32])
{
- const int64_t X24 = (1 << 24);
- const int64_t X25 = (1 << 25);
- const int64_t X26 = (1 << 26);
-
int64_t h0 = load_4(s);
int64_t h1 = load_3(s + 4) << 6;
int64_t h2 = load_3(s + 7) << 5;
@@ -734,37 +615,17 @@ void FE_25519::from_bytes(const uint8_t s[32])
int64_t h8 = load_3(s + 26) << 4;
int64_t h9 = (load_3(s + 29) & 0x7fffff) << 2;
- const int64_t carry9 = (h9 + X24) >> 25;
- h0 += carry9 * 19;
- h9 -= carry9 * X25;
- const int64_t carry1 = (h1 + X24) >> 25;
- h2 += carry1;
- h1 -= carry1 * X25;
- const int64_t carry3 = (h3 + X24) >> 25;
- h4 += carry3;
- h3 -= carry3 * X25;
- const int64_t carry5 = (h5 + X24) >> 25;
- h6 += carry5;
- h5 -= carry5 * X25;
- const int64_t carry7 = (h7 + X24) >> 25;
- h8 += carry7;
- h7 -= carry7 * X25;
-
- const int64_t carry0 = (h0 + X25) >> 26;
- h1 += carry0;
- h0 -= carry0 * X26;
- const int64_t carry2 = (h2 + X25) >> 26;
- h3 += carry2;
- h2 -= carry2 * X26;
- const int64_t carry4 = (h4 + X25) >> 26;
- h5 += carry4;
- h4 -= carry4 * X26;
- const int64_t carry6 = (h6 + X25) >> 26;
- h7 += carry6;
- h6 -= carry6 * X26;
- const int64_t carry8 = (h8 + X25) >> 26;
- h9 += carry8;
- h8 -= carry8 * X26;
+ carry<25,19>(h9, h0);
+ carry<25>(h1, h2);
+ carry<25>(h3, h4);
+ carry<25>(h5, h6);
+ carry<25>(h7, h8);
+
+ carry<26>(h0, h1);
+ carry<26>(h2, h3);
+ carry<26>(h4, h5);
+ carry<26>(h6, h7);
+ carry<26>(h8, h9);
m_fe[0] = h0;
m_fe[1] = h1;
@@ -819,16 +680,6 @@ void FE_25519::to_bytes(uint8_t s[32]) const
int32_t h8 = m_fe[8];
int32_t h9 = m_fe[9];
int32_t q;
- int32_t carry0;
- int32_t carry1;
- int32_t carry2;
- int32_t carry3;
- int32_t carry4;
- int32_t carry5;
- int32_t carry6;
- int32_t carry7;
- int32_t carry8;
- int32_t carry9;
q = (19 * h9 + ((static_cast<int32_t>(1) << 24))) >> 25;
q = (h0 + q) >> 26;
@@ -846,34 +697,17 @@ void FE_25519::to_bytes(uint8_t s[32]) const
h0 += 19 * q;
/* Goal: Output h-2^255 q, which is between 0 and 2^255-20. */
- carry0 = h0 >> 26;
- h1 += carry0;
- h0 -= carry0 * X26;
- carry1 = h1 >> 25;
- h2 += carry1;
- h1 -= carry1 * X25;
- carry2 = h2 >> 26;
- h3 += carry2;
- h2 -= carry2 * X26;
- carry3 = h3 >> 25;
- h4 += carry3;
- h3 -= carry3 * X25;
- carry4 = h4 >> 26;
- h5 += carry4;
- h4 -= carry4 * X26;
- carry5 = h5 >> 25;
- h6 += carry5;
- h5 -= carry5 * X25;
- carry6 = h6 >> 26;
- h7 += carry6;
- h6 -= carry6 * X26;
- carry7 = h7 >> 25;
- h8 += carry7;
- h7 -= carry7 * X25;
- carry8 = h8 >> 26;
- h9 += carry8;
- h8 -= carry8 * X26;
- carry9 = h9 >> 25;
+ carry0<26>(h0, h1);
+ carry0<25>(h1, h2);
+ carry0<26>(h2, h3);
+ carry0<25>(h3, h4);
+ carry0<26>(h4, h5);
+ carry0<25>(h5, h6);
+ carry0<26>(h6, h7);
+ carry0<25>(h7, h8);
+ carry0<26>(h8, h9);
+
+ int32_t carry9 = h9 >> 25;
h9 -= carry9 * X25;
/* h10 = carry9 */
diff --git a/src/lib/pubkey/ed25519/ed25519_internal.h b/src/lib/pubkey/ed25519/ed25519_internal.h
index cb75e8540..0efeee6d7 100644
--- a/src/lib/pubkey/ed25519/ed25519_internal.h
+++ b/src/lib/pubkey/ed25519/ed25519_internal.h
@@ -28,6 +28,40 @@ inline uint64_t load_4(const uint8_t* in)
return load_le<uint32_t>(in, 0);
}
+template<size_t S, int64_t MUL=1>
+inline void carry(int64_t& h0, int64_t& h1)
+ {
+ static_assert(S > 0 && S < 64, "Shift in range");
+
+ const int64_t X1 = (static_cast<int64_t>(1) << S);
+ const int64_t X2 = (static_cast<int64_t>(1) << (S - 1));
+ int64_t c = (h0 + X2) >> S;
+ h1 += c * MUL;
+ h0 -= c * X1;
+ }
+
+template<size_t S>
+inline void carry0(int64_t& h0, int64_t& h1)
+ {
+ static_assert(S > 0 && S < 64, "Shift in range");
+
+ const int64_t X1 = (static_cast<int64_t>(1) << S);
+ int64_t c = h0 >> S;
+ h1 += c;
+ h0 -= c * X1;
+ }
+
+template<size_t S>
+inline void carry0(int32_t& h0, int32_t& h1)
+ {
+ static_assert(S > 0 && S < 32, "Shift in range");
+
+ const int32_t X1 = (static_cast<int64_t>(1) << S);
+ int32_t c = h0 >> S;
+ h1 += c;
+ h0 -= c * X1;
+ }
+
/*
ge means group element.
diff --git a/src/lib/pubkey/ed25519/sc_muladd.cpp b/src/lib/pubkey/ed25519/sc_muladd.cpp
index 83863eeb4..948fdcc86 100644
--- a/src/lib/pubkey/ed25519/sc_muladd.cpp
+++ b/src/lib/pubkey/ed25519/sc_muladd.cpp
@@ -28,41 +28,43 @@ void sc_muladd(uint8_t* s, const uint8_t* a, const uint8_t* b, const uint8_t* c)
const int64_t X20 = (1 << 20);
const int64_t X21 = (1 << 21);
- int64_t a0 = 2097151 & load_3(a);
- int64_t a1 = 2097151 & (load_4(a + 2) >> 5);
- int64_t a2 = 2097151 & (load_3(a + 5) >> 2);
- int64_t a3 = 2097151 & (load_4(a + 7) >> 7);
- int64_t a4 = 2097151 & (load_4(a + 10) >> 4);
- int64_t a5 = 2097151 & (load_3(a + 13) >> 1);
- int64_t a6 = 2097151 & (load_4(a + 15) >> 6);
- int64_t a7 = 2097151 & (load_3(a + 18) >> 3);
- int64_t a8 = 2097151 & load_3(a + 21);
- int64_t a9 = 2097151 & (load_4(a + 23) >> 5);
- int64_t a10 = 2097151 & (load_3(a + 26) >> 2);
+ const int32_t MASK = 0x1fffff;
+
+ int64_t a0 = MASK & load_3(a);
+ int64_t a1 = MASK & (load_4(a + 2) >> 5);
+ int64_t a2 = MASK & (load_3(a + 5) >> 2);
+ int64_t a3 = MASK & (load_4(a + 7) >> 7);
+ int64_t a4 = MASK & (load_4(a + 10) >> 4);
+ int64_t a5 = MASK & (load_3(a + 13) >> 1);
+ int64_t a6 = MASK & (load_4(a + 15) >> 6);
+ int64_t a7 = MASK & (load_3(a + 18) >> 3);
+ int64_t a8 = MASK & load_3(a + 21);
+ int64_t a9 = MASK & (load_4(a + 23) >> 5);
+ int64_t a10 = MASK & (load_3(a + 26) >> 2);
int64_t a11 = (load_4(a + 28) >> 7);
- int64_t b0 = 2097151 & load_3(b);
- int64_t b1 = 2097151 & (load_4(b + 2) >> 5);
- int64_t b2 = 2097151 & (load_3(b + 5) >> 2);
- int64_t b3 = 2097151 & (load_4(b + 7) >> 7);
- int64_t b4 = 2097151 & (load_4(b + 10) >> 4);
- int64_t b5 = 2097151 & (load_3(b + 13) >> 1);
- int64_t b6 = 2097151 & (load_4(b + 15) >> 6);
- int64_t b7 = 2097151 & (load_3(b + 18) >> 3);
- int64_t b8 = 2097151 & load_3(b + 21);
- int64_t b9 = 2097151 & (load_4(b + 23) >> 5);
- int64_t b10 = 2097151 & (load_3(b + 26) >> 2);
+ int64_t b0 = MASK & load_3(b);
+ int64_t b1 = MASK & (load_4(b + 2) >> 5);
+ int64_t b2 = MASK & (load_3(b + 5) >> 2);
+ int64_t b3 = MASK & (load_4(b + 7) >> 7);
+ int64_t b4 = MASK & (load_4(b + 10) >> 4);
+ int64_t b5 = MASK & (load_3(b + 13) >> 1);
+ int64_t b6 = MASK & (load_4(b + 15) >> 6);
+ int64_t b7 = MASK & (load_3(b + 18) >> 3);
+ int64_t b8 = MASK & load_3(b + 21);
+ int64_t b9 = MASK & (load_4(b + 23) >> 5);
+ int64_t b10 = MASK & (load_3(b + 26) >> 2);
int64_t b11 = (load_4(b + 28) >> 7);
- int64_t c0 = 2097151 & load_3(c);
- int64_t c1 = 2097151 & (load_4(c + 2) >> 5);
- int64_t c2 = 2097151 & (load_3(c + 5) >> 2);
- int64_t c3 = 2097151 & (load_4(c + 7) >> 7);
- int64_t c4 = 2097151 & (load_4(c + 10) >> 4);
- int64_t c5 = 2097151 & (load_3(c + 13) >> 1);
- int64_t c6 = 2097151 & (load_4(c + 15) >> 6);
- int64_t c7 = 2097151 & (load_3(c + 18) >> 3);
- int64_t c8 = 2097151 & load_3(c + 21);
- int64_t c9 = 2097151 & (load_4(c + 23) >> 5);
- int64_t c10 = 2097151 & (load_3(c + 26) >> 2);
+ int64_t c0 = MASK & load_3(c);
+ int64_t c1 = MASK & (load_4(c + 2) >> 5);
+ int64_t c2 = MASK & (load_3(c + 5) >> 2);
+ int64_t c3 = MASK & (load_4(c + 7) >> 7);
+ int64_t c4 = MASK & (load_4(c + 10) >> 4);
+ int64_t c5 = MASK & (load_3(c + 13) >> 1);
+ int64_t c6 = MASK & (load_4(c + 15) >> 6);
+ int64_t c7 = MASK & (load_3(c + 18) >> 3);
+ int64_t c8 = MASK & load_3(c + 21);
+ int64_t c9 = MASK & (load_4(c + 23) >> 5);
+ int64_t c10 = MASK & (load_3(c + 26) >> 2);
int64_t c11 = (load_4(c + 28) >> 7);
int64_t s0;
int64_t s1;
@@ -88,29 +90,6 @@ void sc_muladd(uint8_t* s, const uint8_t* a, const uint8_t* b, const uint8_t* c)
int64_t s21;
int64_t s22;
int64_t s23;
- int64_t carry0;
- int64_t carry1;
- int64_t carry2;
- int64_t carry3;
- int64_t carry4;
- int64_t carry5;
- int64_t carry6;
- int64_t carry7;
- int64_t carry8;
- int64_t carry9;
- int64_t carry10;
- int64_t carry11;
- int64_t carry12;
- int64_t carry13;
- int64_t carry14;
- int64_t carry15;
- int64_t carry16;
- int64_t carry17;
- int64_t carry18;
- int64_t carry19;
- int64_t carry20;
- int64_t carry21;
- int64_t carry22;
s0 = c0 + a0*b0;
s1 = c1 + a0*b1 + a1*b0;
@@ -137,76 +116,30 @@ void sc_muladd(uint8_t* s, const uint8_t* a, const uint8_t* b, const uint8_t* c)
s22 = a11*b11;
s23 = 0;
- carry0 = (s0 + X20) >> 21;
- s1 += carry0;
- s0 -= carry0 * X21;
- carry2 = (s2 + X20) >> 21;
- s3 += carry2;
- s2 -= carry2 * X21;
- carry4 = (s4 + X20) >> 21;
- s5 += carry4;
- s4 -= carry4 * X21;
- carry6 = (s6 + X20) >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry8 = (s8 + X20) >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry10 = (s10 + X20) >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
- carry12 = (s12 + X20) >> 21;
- s13 += carry12;
- s12 -= carry12 * X21;
- carry14 = (s14 + X20) >> 21;
- s15 += carry14;
- s14 -= carry14 * X21;
- carry16 = (s16 + X20) >> 21;
- s17 += carry16;
- s16 -= carry16 * X21;
- carry18 = (s18 + X20) >> 21;
- s19 += carry18;
- s18 -= carry18 * X21;
- carry20 = (s20 + X20) >> 21;
- s21 += carry20;
- s20 -= carry20 * X21;
- carry22 = (s22 + X20) >> 21;
- s23 += carry22;
- s22 -= carry22 * X21;
-
- carry1 = (s1 + X20) >> 21;
- s2 += carry1;
- s1 -= carry1 * X21;
- carry3 = (s3 + X20) >> 21;
- s4 += carry3;
- s3 -= carry3 * X21;
- carry5 = (s5 + X20) >> 21;
- s6 += carry5;
- s5 -= carry5 * X21;
- carry7 = (s7 + X20) >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry9 = (s9 + X20) >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry11 = (s11 + X20) >> 21;
- s12 += carry11;
- s11 -= carry11 * X21;
- carry13 = (s13 + X20) >> 21;
- s14 += carry13;
- s13 -= carry13 * X21;
- carry15 = (s15 + X20) >> 21;
- s16 += carry15;
- s15 -= carry15 * X21;
- carry17 = (s17 + X20) >> 21;
- s18 += carry17;
- s17 -= carry17 * X21;
- carry19 = (s19 + X20) >> 21;
- s20 += carry19;
- s19 -= carry19 * X21;
- carry21 = (s21 + X20) >> 21;
- s22 += carry21;
- s21 -= carry21 * X21;
+ carry<21>(s0, s1);
+ carry<21>(s2, s3);
+ carry<21>(s4, s5);
+ carry<21>(s6, s7);
+ carry<21>(s8, s9);
+ carry<21>(s10, s11);
+ carry<21>(s12, s13);
+ carry<21>(s14, s15);
+ carry<21>(s16, s17);
+ carry<21>(s18, s19);
+ carry<21>(s20, s21);
+ carry<21>(s22, s23);
+
+ carry<21>(s1, s2);
+ carry<21>(s3, s4);
+ carry<21>(s5, s6);
+ carry<21>(s7, s8);
+ carry<21>(s9, s10);
+ carry<21>(s11, s12);
+ carry<21>(s13, s14);
+ carry<21>(s15, s16);
+ carry<21>(s17, s18);
+ carry<21>(s19, s20);
+ carry<21>(s21, s22);
s11 += s23 * 666643;
s12 += s23 * 470296;
@@ -256,40 +189,18 @@ void sc_muladd(uint8_t* s, const uint8_t* a, const uint8_t* b, const uint8_t* c)
s11 -= s18 * 683901;
s18 = 0;
- carry6 = (s6 + X20) >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry8 = (s8 + X20) >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry10 = (s10 + X20) >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
- carry12 = (s12 + X20) >> 21;
- s13 += carry12;
- s12 -= carry12 * X21;
- carry14 = (s14 + X20) >> 21;
- s15 += carry14;
- s14 -= carry14 * X21;
- carry16 = (s16 + X20) >> 21;
- s17 += carry16;
- s16 -= carry16 * X21;
-
- carry7 = (s7 + X20) >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry9 = (s9 + X20) >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry11 = (s11 + X20) >> 21;
- s12 += carry11;
- s11 -= carry11 * X21;
- carry13 = (s13 + X20) >> 21;
- s14 += carry13;
- s13 -= carry13 * X21;
- carry15 = (s15 + X20) >> 21;
- s16 += carry15;
- s15 -= carry15 * X21;
+ carry<21>(s6, s7);
+ carry<21>(s8, s9);
+ carry<21>(s10, s11);
+ carry<21>(s12, s13);
+ carry<21>(s14, s15);
+ carry<21>(s16, s17);
+
+ carry<21>(s7, s8);
+ carry<21>(s9, s10);
+ carry<21>(s11, s12);
+ carry<21>(s13, s14);
+ carry<21>(s15, s16);
s5 += s17 * 666643;
s6 += s17 * 470296;
@@ -339,43 +250,19 @@ void sc_muladd(uint8_t* s, const uint8_t* a, const uint8_t* b, const uint8_t* c)
s5 -= s12 * 683901;
s12 = 0;
- carry0 = (s0 + X20) >> 21;
- s1 += carry0;
- s0 -= carry0 * X21;
- carry2 = (s2 + X20) >> 21;
- s3 += carry2;
- s2 -= carry2 * X21;
- carry4 = (s4 + X20) >> 21;
- s5 += carry4;
- s4 -= carry4 * X21;
- carry6 = (s6 + X20) >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry8 = (s8 + X20) >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry10 = (s10 + X20) >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
-
- carry1 = (s1 + X20) >> 21;
- s2 += carry1;
- s1 -= carry1 * X21;
- carry3 = (s3 + X20) >> 21;
- s4 += carry3;
- s3 -= carry3 * X21;
- carry5 = (s5 + X20) >> 21;
- s6 += carry5;
- s5 -= carry5 * X21;
- carry7 = (s7 + X20) >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry9 = (s9 + X20) >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry11 = (s11 + X20) >> 21;
- s12 += carry11;
- s11 -= carry11 * X21;
+ carry<21>(s0, s1);
+ carry<21>(s2, s3);
+ carry<21>(s4, s5);
+ carry<21>(s6, s7);
+ carry<21>(s8, s9);
+ carry<21>(s10, s11);
+
+ carry<21>(s1, s2);
+ carry<21>(s3, s4);
+ carry<21>(s5, s6);
+ carry<21>(s7, s8);
+ carry<21>(s9, s10);
+ carry<21>(s11, s12);
s0 += s12 * 666643;
s1 += s12 * 470296;
@@ -385,42 +272,18 @@ void sc_muladd(uint8_t* s, const uint8_t* a, const uint8_t* b, const uint8_t* c)
s5 -= s12 * 683901;
s12 = 0;
- carry0 = s0 >> 21;
- s1 += carry0;
- s0 -= carry0 * X21;
- carry1 = s1 >> 21;
- s2 += carry1;
- s1 -= carry1 * X21;
- carry2 = s2 >> 21;
- s3 += carry2;
- s2 -= carry2 * X21;
- carry3 = s3 >> 21;
- s4 += carry3;
- s3 -= carry3 * X21;
- carry4 = s4 >> 21;
- s5 += carry4;
- s4 -= carry4 * X21;
- carry5 = s5 >> 21;
- s6 += carry5;
- s5 -= carry5 * X21;
- carry6 = s6 >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry7 = s7 >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry8 = s8 >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry9 = s9 >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry10 = s10 >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
- carry11 = s11 >> 21;
- s12 += carry11;
- s11 -= carry11 * X21;
+ carry<21>(s0, s1);
+ carry<21>(s1, s2);
+ carry<21>(s2, s3);
+ carry<21>(s3, s4);
+ carry<21>(s4, s5);
+ carry<21>(s5, s6);
+ carry<21>(s6, s7);
+ carry<21>(s7, s8);
+ carry<21>(s8, s9);
+ carry<21>(s9, s10);
+ carry<21>(s10, s11);
+ carry0<21>(s11, s12);
s0 += s12 * 666643;
s1 += s12 * 470296;
@@ -430,39 +293,17 @@ void sc_muladd(uint8_t* s, const uint8_t* a, const uint8_t* b, const uint8_t* c)
s5 -= s12 * 683901;
s12 = 0;
- carry0 = s0 >> 21;
- s1 += carry0;
- s0 -= carry0 * X21;
- carry1 = s1 >> 21;
- s2 += carry1;
- s1 -= carry1 * X21;
- carry2 = s2 >> 21;
- s3 += carry2;
- s2 -= carry2 * X21;
- carry3 = s3 >> 21;
- s4 += carry3;
- s3 -= carry3 * X21;
- carry4 = s4 >> 21;
- s5 += carry4;
- s4 -= carry4 * X21;
- carry5 = s5 >> 21;
- s6 += carry5;
- s5 -= carry5 * X21;
- carry6 = s6 >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry7 = s7 >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry8 = s8 >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry9 = s9 >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry10 = s10 >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
+ carry0<21>(s0, s1);
+ carry0<21>(s1, s2);
+ carry0<21>(s2, s3);
+ carry0<21>(s3, s4);
+ carry0<21>(s4, s5);
+ carry0<21>(s5, s6);
+ carry0<21>(s6, s7);
+ carry0<21>(s7, s8);
+ carry0<21>(s8, s9);
+ carry0<21>(s9, s10);
+ carry0<21>(s10, s11);
s[0] = s0 >> 0;
s[1] = s0 >> 8;
diff --git a/src/lib/pubkey/ed25519/sc_reduce.cpp b/src/lib/pubkey/ed25519/sc_reduce.cpp
index 5765c7dac..ec8f94955 100644
--- a/src/lib/pubkey/ed25519/sc_reduce.cpp
+++ b/src/lib/pubkey/ed25519/sc_reduce.cpp
@@ -24,51 +24,32 @@ Output:
void sc_reduce(uint8_t* s)
{
- const int64_t X20 = (1 << 20);
- const int64_t X21 = (1 << 21);
-
-
- int64_t s0 = 2097151 & load_3(s);
- int64_t s1 = 2097151 & (load_4(s + 2) >> 5);
- int64_t s2 = 2097151 & (load_3(s + 5) >> 2);
- int64_t s3 = 2097151 & (load_4(s + 7) >> 7);
- int64_t s4 = 2097151 & (load_4(s + 10) >> 4);
- int64_t s5 = 2097151 & (load_3(s + 13) >> 1);
- int64_t s6 = 2097151 & (load_4(s + 15) >> 6);
- int64_t s7 = 2097151 & (load_3(s + 18) >> 3);
- int64_t s8 = 2097151 & load_3(s + 21);
- int64_t s9 = 2097151 & (load_4(s + 23) >> 5);
- int64_t s10 = 2097151 & (load_3(s + 26) >> 2);
- int64_t s11 = 2097151 & (load_4(s + 28) >> 7);
- int64_t s12 = 2097151 & (load_4(s + 31) >> 4);
- int64_t s13 = 2097151 & (load_3(s + 34) >> 1);
- int64_t s14 = 2097151 & (load_4(s + 36) >> 6);
- int64_t s15 = 2097151 & (load_3(s + 39) >> 3);
- int64_t s16 = 2097151 & load_3(s + 42);
- int64_t s17 = 2097151 & (load_4(s + 44) >> 5);
- int64_t s18 = 2097151 & (load_3(s + 47) >> 2);
- int64_t s19 = 2097151 & (load_4(s + 49) >> 7);
- int64_t s20 = 2097151 & (load_4(s + 52) >> 4);
- int64_t s21 = 2097151 & (load_3(s + 55) >> 1);
- int64_t s22 = 2097151 & (load_4(s + 57) >> 6);
+ const uint32_t MASK = 0x1fffff;
+
+ int64_t s0 = MASK & load_3(s);
+ int64_t s1 = MASK & (load_4(s + 2) >> 5);
+ int64_t s2 = MASK & (load_3(s + 5) >> 2);
+ int64_t s3 = MASK & (load_4(s + 7) >> 7);
+ int64_t s4 = MASK & (load_4(s + 10) >> 4);
+ int64_t s5 = MASK & (load_3(s + 13) >> 1);
+ int64_t s6 = MASK & (load_4(s + 15) >> 6);
+ int64_t s7 = MASK & (load_3(s + 18) >> 3);
+ int64_t s8 = MASK & load_3(s + 21);
+ int64_t s9 = MASK & (load_4(s + 23) >> 5);
+ int64_t s10 = MASK & (load_3(s + 26) >> 2);
+ int64_t s11 = MASK & (load_4(s + 28) >> 7);
+ int64_t s12 = MASK & (load_4(s + 31) >> 4);
+ int64_t s13 = MASK & (load_3(s + 34) >> 1);
+ int64_t s14 = MASK & (load_4(s + 36) >> 6);
+ int64_t s15 = MASK & (load_3(s + 39) >> 3);
+ int64_t s16 = MASK & load_3(s + 42);
+ int64_t s17 = MASK & (load_4(s + 44) >> 5);
+ int64_t s18 = MASK & (load_3(s + 47) >> 2);
+ int64_t s19 = MASK & (load_4(s + 49) >> 7);
+ int64_t s20 = MASK & (load_4(s + 52) >> 4);
+ int64_t s21 = MASK & (load_3(s + 55) >> 1);
+ int64_t s22 = MASK & (load_4(s + 57) >> 6);
int64_t s23 = (load_4(s + 60) >> 3);
- int64_t carry0;
- int64_t carry1;
- int64_t carry2;
- int64_t carry3;
- int64_t carry4;
- int64_t carry5;
- int64_t carry6;
- int64_t carry7;
- int64_t carry8;
- int64_t carry9;
- int64_t carry10;
- int64_t carry11;
- int64_t carry12;
- int64_t carry13;
- int64_t carry14;
- int64_t carry15;
- int64_t carry16;
s11 += s23 * 666643;
s12 += s23 * 470296;
@@ -118,40 +99,18 @@ void sc_reduce(uint8_t* s)
s11 -= s18 * 683901;
s18 = 0;
- carry6 = (s6 + X20) >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry8 = (s8 + X20) >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry10 = (s10 + X20) >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
- carry12 = (s12 + X20) >> 21;
- s13 += carry12;
- s12 -= carry12 * X21;
- carry14 = (s14 + X20) >> 21;
- s15 += carry14;
- s14 -= carry14 * X21;
- carry16 = (s16 + X20) >> 21;
- s17 += carry16;
- s16 -= carry16 * X21;
+ carry<21>(s6, s7);
+ carry<21>(s8, s9);
+ carry<21>(s10, s11);
+ carry<21>(s12, s13);
+ carry<21>(s14, s15);
+ carry<21>(s16, s17);
- carry7 = (s7 + X20) >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry9 = (s9 + X20) >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry11 = (s11 + X20) >> 21;
- s12 += carry11;
- s11 -= carry11 * X21;
- carry13 = (s13 + X20) >> 21;
- s14 += carry13;
- s13 -= carry13 * X21;
- carry15 = (s15 + X20) >> 21;
- s16 += carry15;
- s15 -= carry15 * X21;
+ carry<21>(s7, s8);
+ carry<21>(s9, s10);
+ carry<21>(s11, s12);
+ carry<21>(s13, s14);
+ carry<21>(s15, s16);
s5 += s17 * 666643;
s6 += s17 * 470296;
@@ -201,43 +160,19 @@ void sc_reduce(uint8_t* s)
s5 -= s12 * 683901;
s12 = 0;
- carry0 = (s0 + X20) >> 21;
- s1 += carry0;
- s0 -= carry0 * X21;
- carry2 = (s2 + X20) >> 21;
- s3 += carry2;
- s2 -= carry2 * X21;
- carry4 = (s4 + X20) >> 21;
- s5 += carry4;
- s4 -= carry4 * X21;
- carry6 = (s6 + X20) >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry8 = (s8 + X20) >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry10 = (s10 + X20) >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
+ carry<21>(s0, s1);
+ carry<21>(s2, s3);
+ carry<21>(s4, s5);
+ carry<21>(s6, s7);
+ carry<21>(s8, s9);
+ carry<21>(s10, s11);
- carry1 = (s1 + X20) >> 21;
- s2 += carry1;
- s1 -= carry1 * X21;
- carry3 = (s3 + X20) >> 21;
- s4 += carry3;
- s3 -= carry3 * X21;
- carry5 = (s5 + X20) >> 21;
- s6 += carry5;
- s5 -= carry5 * X21;
- carry7 = (s7 + X20) >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry9 = (s9 + X20) >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry11 = (s11 + X20) >> 21;
- s12 += carry11;
- s11 -= carry11 * X21;
+ carry<21>(s1, s2);
+ carry<21>(s3, s4);
+ carry<21>(s5, s6);
+ carry<21>(s7, s8);
+ carry<21>(s9, s10);
+ carry<21>(s11, s12);
s0 += s12 * 666643;
s1 += s12 * 470296;
@@ -247,42 +182,18 @@ void sc_reduce(uint8_t* s)
s5 -= s12 * 683901;
s12 = 0;
- carry0 = s0 >> 21;
- s1 += carry0;
- s0 -= carry0 * X21;
- carry1 = s1 >> 21;
- s2 += carry1;
- s1 -= carry1 * X21;
- carry2 = s2 >> 21;
- s3 += carry2;
- s2 -= carry2 * X21;
- carry3 = s3 >> 21;
- s4 += carry3;
- s3 -= carry3 * X21;
- carry4 = s4 >> 21;
- s5 += carry4;
- s4 -= carry4 * X21;
- carry5 = s5 >> 21;
- s6 += carry5;
- s5 -= carry5 * X21;
- carry6 = s6 >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry7 = s7 >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry8 = s8 >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry9 = s9 >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry10 = s10 >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
- carry11 = s11 >> 21;
- s12 += carry11;
- s11 -= carry11 * X21;
+ carry<21>(s0, s1);
+ carry<21>(s1, s2);
+ carry<21>(s2, s3);
+ carry<21>(s3, s4);
+ carry<21>(s4, s5);
+ carry<21>(s5, s6);
+ carry<21>(s6, s7);
+ carry<21>(s7, s8);
+ carry<21>(s8, s9);
+ carry<21>(s9, s10);
+ carry<21>(s10, s11);
+ carry0<21>(s11, s12);
s0 += s12 * 666643;
s1 += s12 * 470296;
@@ -292,39 +203,17 @@ void sc_reduce(uint8_t* s)
s5 -= s12 * 683901;
s12 = 0;
- carry0 = s0 >> 21;
- s1 += carry0;
- s0 -= carry0 * X21;
- carry1 = s1 >> 21;
- s2 += carry1;
- s1 -= carry1 * X21;
- carry2 = s2 >> 21;
- s3 += carry2;
- s2 -= carry2 * X21;
- carry3 = s3 >> 21;
- s4 += carry3;
- s3 -= carry3 * X21;
- carry4 = s4 >> 21;
- s5 += carry4;
- s4 -= carry4 * X21;
- carry5 = s5 >> 21;
- s6 += carry5;
- s5 -= carry5 * X21;
- carry6 = s6 >> 21;
- s7 += carry6;
- s6 -= carry6 * X21;
- carry7 = s7 >> 21;
- s8 += carry7;
- s7 -= carry7 * X21;
- carry8 = s8 >> 21;
- s9 += carry8;
- s8 -= carry8 * X21;
- carry9 = s9 >> 21;
- s10 += carry9;
- s9 -= carry9 * X21;
- carry10 = s10 >> 21;
- s11 += carry10;
- s10 -= carry10 * X21;
+ carry0<21>(s0, s1);
+ carry0<21>(s1, s2);
+ carry0<21>(s2, s3);
+ carry0<21>(s3, s4);
+ carry0<21>(s4, s5);
+ carry0<21>(s5, s6);
+ carry0<21>(s6, s7);
+ carry0<21>(s7, s8);
+ carry0<21>(s8, s9);
+ carry0<21>(s9, s10);
+ carry0<21>(s10, s11);
s[0] = s0 >> 0;
s[1] = s0 >> 8;