grpc 1.35.0 → 1.47.0

data/third_party/boringssl-with-bazel/src/crypto/fipsmodule/md4/md4.c

@@ -60,6 +60,7 @@
 #include <string.h>
 
 #include "../../internal.h"
+#include "../digest/md32_common.h"
 
 
 uint8_t *MD4(const uint8_t *data, size_t len, uint8_t out[MD4_DIGEST_LENGTH]) {
@@ -71,7 +72,7 @@ uint8_t *MD4(const uint8_t *data, size_t len, uint8_t out[MD4_DIGEST_LENGTH]) {
   return out;
 }
 
-// Implemented from RFC1186 The MD4 Message-Digest Algorithm.
+// Implemented from RFC 1186 The MD4 Message-Digest Algorithm.
 
 int MD4_Init(MD4_CTX *md4) {
   OPENSSL_memset(md4, 0, sizeof(MD4_CTX));
@@ -84,29 +85,26 @@ int MD4_Init(MD4_CTX *md4) {
 
 void md4_block_data_order(uint32_t *state, const uint8_t *data, size_t num);
 
-#define DATA_ORDER_IS_LITTLE_ENDIAN
+void MD4_Transform(MD4_CTX *c, const uint8_t data[MD4_CBLOCK]) {
+  md4_block_data_order(c->h, data, 1);
+}
 
-#define HASH_CTX MD4_CTX
-#define HASH_CBLOCK 64
-#define HASH_DIGEST_LENGTH 16
-#define HASH_UPDATE MD4_Update
-#define HASH_TRANSFORM MD4_Transform
-#define HASH_FINAL MD4_Final
-#define HASH_MAKE_STRING(c, s) \
-  do {                         \
-    uint32_t ll;               \
-    ll = (c)->h[0];            \
-    HOST_l2c(ll, (s));         \
-    ll = (c)->h[1];            \
-    HOST_l2c(ll, (s));         \
-    ll = (c)->h[2];            \
-    HOST_l2c(ll, (s));         \
-    ll = (c)->h[3];            \
-    HOST_l2c(ll, (s));         \
-  } while (0)
-#define HASH_BLOCK_DATA_ORDER md4_block_data_order
+int MD4_Update(MD4_CTX *c, const void *data, size_t len) {
+  crypto_md32_update(&md4_block_data_order, c->h, c->data, MD4_CBLOCK, &c->num,
+                     &c->Nh, &c->Nl, data, len);
+  return 1;
+}
 
-#include "../digest/md32_common.h"
+int MD4_Final(uint8_t out[MD4_DIGEST_LENGTH], MD4_CTX *c) {
+  crypto_md32_final(&md4_block_data_order, c->h, c->data, MD4_CBLOCK, &c->num,
+                    c->Nh, c->Nl, /*is_big_endian=*/0);
+
+  CRYPTO_store_u32_le(out, c->h[0]);
+  CRYPTO_store_u32_le(out + 4, c->h[1]);
+  CRYPTO_store_u32_le(out + 8, c->h[2]);
+  CRYPTO_store_u32_le(out + 12, c->h[3]);
+  return 1;
+}
 
 // As pointed out by Wei Dai <weidai@eskimo.com>, the above can be
 // simplified to the code below.  Wei attributes these optimizations
@@ -115,28 +113,26 @@ void md4_block_data_order(uint32_t *state, const uint8_t *data, size_t num);
 #define G(b, c, d) (((b) & (c)) | ((b) & (d)) | ((c) & (d)))
 #define H(b, c, d) ((b) ^ (c) ^ (d))
 
-#define ROTATE(a, n) (((a) << (n)) | ((a) >> (32 - (n))))
-
 #define R0(a, b, c, d, k, s, t)            \
   do {                                     \
     (a) += ((k) + (t) + F((b), (c), (d))); \
-    (a) = ROTATE(a, s);                    \
+    (a) = CRYPTO_rotl_u32(a, s);           \
   } while (0)
 
 #define R1(a, b, c, d, k, s, t)            \
   do {                                     \
     (a) += ((k) + (t) + G((b), (c), (d))); \
-    (a) = ROTATE(a, s);                    \
+    (a) = CRYPTO_rotl_u32(a, s);           \
   } while (0)
 
 #define R2(a, b, c, d, k, s, t)            \
   do {                                     \
     (a) += ((k) + (t) + H((b), (c), (d))); \
-    (a) = ROTATE(a, s);                    \
+    (a) = CRYPTO_rotl_u32(a, s);           \
   } while (0)
 
 void md4_block_data_order(uint32_t *state, const uint8_t *data, size_t num) {
-  uint32_t A, B, C, D, l;
+  uint32_t A, B, C, D;
   uint32_t X0, X1, X2, X3, X4, X5, X6, X7, X8, X9, X10, X11, X12, X13, X14, X15;
 
   A = state[0];
@@ -145,53 +141,53 @@ void md4_block_data_order(uint32_t *state, const uint8_t *data, size_t num) {
   D = state[3];
 
   for (; num--;) {
-    HOST_c2l(data, l);
-    X0 = l;
-    HOST_c2l(data, l);
-    X1 = l;
+    X0 = CRYPTO_load_u32_le(data);
+    data += 4;
+    X1 = CRYPTO_load_u32_le(data);
+    data += 4;
     // Round 0
     R0(A, B, C, D, X0, 3, 0);
-    HOST_c2l(data, l);
-    X2 = l;
+    X2 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(D, A, B, C, X1, 7, 0);
-    HOST_c2l(data, l);
-    X3 = l;
+    X3 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(C, D, A, B, X2, 11, 0);
-    HOST_c2l(data, l);
-    X4 = l;
+    X4 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(B, C, D, A, X3, 19, 0);
-    HOST_c2l(data, l);
-    X5 = l;
+    X5 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(A, B, C, D, X4, 3, 0);
-    HOST_c2l(data, l);
-    X6 = l;
+    X6 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(D, A, B, C, X5, 7, 0);
-    HOST_c2l(data, l);
-    X7 = l;
+    X7 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(C, D, A, B, X6, 11, 0);
-    HOST_c2l(data, l);
-    X8 = l;
+    X8 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(B, C, D, A, X7, 19, 0);
-    HOST_c2l(data, l);
-    X9 = l;
+    X9 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(A, B, C, D, X8, 3, 0);
-    HOST_c2l(data, l);
-    X10 = l;
+    X10 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(D, A, B, C, X9, 7, 0);
-    HOST_c2l(data, l);
-    X11 = l;
+    X11 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(C, D, A, B, X10, 11, 0);
-    HOST_c2l(data, l);
-    X12 = l;
+    X12 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(B, C, D, A, X11, 19, 0);
-    HOST_c2l(data, l);
-    X13 = l;
+    X13 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(A, B, C, D, X12, 3, 0);
-    HOST_c2l(data, l);
-    X14 = l;
+    X14 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(D, A, B, C, X13, 7, 0);
-    HOST_c2l(data, l);
-    X15 = l;
+    X15 = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(C, D, A, B, X14, 11, 0);
     R0(B, C, D, A, X15, 19, 0);
     // Round 1
@@ -236,21 +232,9 @@ void md4_block_data_order(uint32_t *state, const uint8_t *data, size_t num) {
   }
 }
 
-#undef DATA_ORDER_IS_LITTLE_ENDIAN
-#undef HASH_CTX
-#undef HASH_CBLOCK
-#undef HASH_DIGEST_LENGTH
-#undef HASH_UPDATE
-#undef HASH_TRANSFORM
-#undef HASH_FINAL
-#undef HASH_MAKE_STRING
-#undef HASH_BLOCK_DATA_ORDER
 #undef F
 #undef G
 #undef H
-#undef ROTATE
 #undef R0
 #undef R1
 #undef R2
-#undef HOST_c2l
-#undef HOST_l2c

data/third_party/boringssl-with-bazel/src/crypto/fipsmodule/md5/md5.c

@@ -60,8 +60,9 @@
 
 #include <openssl/mem.h>
 
-#include "internal.h"
 #include "../../internal.h"
+#include "../digest/md32_common.h"
+#include "internal.h"
 
 
 uint8_t *MD5(const uint8_t *data, size_t len, uint8_t out[MD5_DIGEST_LENGTH]) {
@@ -89,30 +90,26 @@ static void md5_block_data_order(uint32_t *state, const uint8_t *data,
                                  size_t num);
 #endif
 
+void MD5_Transform(MD5_CTX *c, const uint8_t data[MD5_CBLOCK]) {
+  md5_block_data_order(c->h, data, 1);
+}
 
-#define DATA_ORDER_IS_LITTLE_ENDIAN
+int MD5_Update(MD5_CTX *c, const void *data, size_t len) {
+  crypto_md32_update(&md5_block_data_order, c->h, c->data, MD5_CBLOCK, &c->num,
+                     &c->Nh, &c->Nl, data, len);
+  return 1;
+}
 
-#define HASH_CTX MD5_CTX
-#define HASH_CBLOCK 64
-#define HASH_DIGEST_LENGTH 16
-#define HASH_UPDATE MD5_Update
-#define HASH_TRANSFORM MD5_Transform
-#define HASH_FINAL MD5_Final
-#define HASH_MAKE_STRING(c, s) \
-  do {                         \
-    uint32_t ll;               \
-    ll = (c)->h[0];            \
-    HOST_l2c(ll, (s));         \
-    ll = (c)->h[1];            \
-    HOST_l2c(ll, (s));         \
-    ll = (c)->h[2];            \
-    HOST_l2c(ll, (s));         \
-    ll = (c)->h[3];            \
-    HOST_l2c(ll, (s));         \
-  } while (0)
-#define HASH_BLOCK_DATA_ORDER md5_block_data_order
+int MD5_Final(uint8_t out[MD5_DIGEST_LENGTH], MD5_CTX *c) {
+  crypto_md32_final(&md5_block_data_order, c->h, c->data, MD5_CBLOCK, &c->num,
+                    c->Nh, c->Nl, /*is_big_endian=*/0);
 
-#include "../digest/md32_common.h"
+  CRYPTO_store_u32_le(out, c->h[0]);
+  CRYPTO_store_u32_le(out + 4, c->h[1]);
+  CRYPTO_store_u32_le(out + 8, c->h[2]);
+  CRYPTO_store_u32_le(out + 12, c->h[3]);
+  return 1;
+}
 
 // As pointed out by Wei Dai <weidai@eskimo.com>, the above can be
 // simplified to the code below.  Wei attributes these optimizations
@@ -122,33 +119,31 @@ static void md5_block_data_order(uint32_t *state, const uint8_t *data,
 #define H(b, c, d) ((b) ^ (c) ^ (d))
 #define I(b, c, d) (((~(d)) | (b)) ^ (c))
 
-#define ROTATE(a, n) (((a) << (n)) | ((a) >> (32 - (n))))
-
 #define R0(a, b, c, d, k, s, t)            \
   do {                                     \
     (a) += ((k) + (t) + F((b), (c), (d))); \
-    (a) = ROTATE(a, s);                    \
+    (a) = CRYPTO_rotl_u32(a, s);           \
     (a) += (b);                            \
   } while (0)
 
 #define R1(a, b, c, d, k, s, t)            \
   do {                                     \
     (a) += ((k) + (t) + G((b), (c), (d))); \
-    (a) = ROTATE(a, s);                    \
+    (a) = CRYPTO_rotl_u32(a, s);           \
     (a) += (b);                            \
   } while (0)
 
 #define R2(a, b, c, d, k, s, t)            \
   do {                                     \
     (a) += ((k) + (t) + H((b), (c), (d))); \
-    (a) = ROTATE(a, s);                    \
+    (a) = CRYPTO_rotl_u32(a, s);           \
     (a) += (b);                            \
   } while (0)
 
 #define R3(a, b, c, d, k, s, t)            \
   do {                                     \
     (a) += ((k) + (t) + I((b), (c), (d))); \
-    (a) = ROTATE(a, s);                    \
+    (a) = CRYPTO_rotl_u32(a, s);           \
     (a) += (b);                            \
   } while (0)
 
@@ -158,7 +153,7 @@ static void md5_block_data_order(uint32_t *state, const uint8_t *data,
 #endif
 static void md5_block_data_order(uint32_t *state, const uint8_t *data,
                                  size_t num) {
-  uint32_t A, B, C, D, l;
+  uint32_t A, B, C, D;
   uint32_t XX0, XX1, XX2, XX3, XX4, XX5, XX6, XX7, XX8, XX9, XX10, XX11, XX12,
       XX13, XX14, XX15;
 #define X(i) XX##i
@@ -169,53 +164,53 @@ static void md5_block_data_order(uint32_t *state, const uint8_t *data,
   D = state[3];
 
   for (; num--;) {
-    HOST_c2l(data, l);
-    X(0) = l;
-    HOST_c2l(data, l);
-    X(1) = l;
+    X(0) = CRYPTO_load_u32_le(data);
+    data += 4;
+    X(1) = CRYPTO_load_u32_le(data);
+    data += 4;
     // Round 0
     R0(A, B, C, D, X(0), 7, 0xd76aa478L);
-    HOST_c2l(data, l);
-    X(2) = l;
+    X(2) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(D, A, B, C, X(1), 12, 0xe8c7b756L);
-    HOST_c2l(data, l);
-    X(3) = l;
+    X(3) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(C, D, A, B, X(2), 17, 0x242070dbL);
-    HOST_c2l(data, l);
-    X(4) = l;
+    X(4) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(B, C, D, A, X(3), 22, 0xc1bdceeeL);
-    HOST_c2l(data, l);
-    X(5) = l;
+    X(5) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(A, B, C, D, X(4), 7, 0xf57c0fafL);
-    HOST_c2l(data, l);
-    X(6) = l;
+    X(6) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(D, A, B, C, X(5), 12, 0x4787c62aL);
-    HOST_c2l(data, l);
-    X(7) = l;
+    X(7) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(C, D, A, B, X(6), 17, 0xa8304613L);
-    HOST_c2l(data, l);
-    X(8) = l;
+    X(8) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(B, C, D, A, X(7), 22, 0xfd469501L);
-    HOST_c2l(data, l);
-    X(9) = l;
+    X(9) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(A, B, C, D, X(8), 7, 0x698098d8L);
-    HOST_c2l(data, l);
-    X(10) = l;
+    X(10) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(D, A, B, C, X(9), 12, 0x8b44f7afL);
-    HOST_c2l(data, l);
-    X(11) = l;
+    X(11) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(C, D, A, B, X(10), 17, 0xffff5bb1L);
-    HOST_c2l(data, l);
-    X(12) = l;
+    X(12) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(B, C, D, A, X(11), 22, 0x895cd7beL);
-    HOST_c2l(data, l);
-    X(13) = l;
+    X(13) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(A, B, C, D, X(12), 7, 0x6b901122L);
-    HOST_c2l(data, l);
-    X(14) = l;
+    X(14) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(D, A, B, C, X(13), 12, 0xfd987193L);
-    HOST_c2l(data, l);
-    X(15) = l;
+    X(15) = CRYPTO_load_u32_le(data);
+    data += 4;
     R0(C, D, A, B, X(14), 17, 0xa679438eL);
     R0(B, C, D, A, X(15), 22, 0x49b40821L);
     // Round 1
@@ -279,23 +274,11 @@ static void md5_block_data_order(uint32_t *state, const uint8_t *data,
 #undef X
 #endif
 
-#undef DATA_ORDER_IS_LITTLE_ENDIAN
-#undef HASH_CTX
-#undef HASH_CBLOCK
-#undef HASH_DIGEST_LENGTH
-#undef HASH_UPDATE
-#undef HASH_TRANSFORM
-#undef HASH_FINAL
-#undef HASH_MAKE_STRING
-#undef HASH_BLOCK_DATA_ORDER
 #undef F
 #undef G
 #undef H
 #undef I
-#undef ROTATE
 #undef R0
 #undef R1
 #undef R2
 #undef R3
-#undef HOST_c2l
-#undef HOST_l2c

data/third_party/boringssl-with-bazel/src/crypto/fipsmodule/modes/cbc.c

@@ -52,20 +52,25 @@
 #include <openssl/type_check.h>
 
 #include "internal.h"
+#include "../../internal.h"
 
 
 void CRYPTO_cbc128_encrypt(const uint8_t *in, uint8_t *out, size_t len,
                            const AES_KEY *key, uint8_t ivec[16],
                            block128_f block) {
-  size_t n;
-  const uint8_t *iv = ivec;
-
   assert(key != NULL && ivec != NULL);
-  assert(len == 0 || (in != NULL && out != NULL));
+  if (len == 0) {
+    // Avoid |ivec| == |iv| in the |memcpy| below, which is not legal in C.
+    return;
+  }
 
+  assert(in != NULL && out != NULL);
+  size_t n;
+  const uint8_t *iv = ivec;
   while (len >= 16) {
-    for (n = 0; n < 16; n += sizeof(size_t)) {
-      store_word_le(out + n, load_word_le(in + n) ^ load_word_le(iv + n));
+    for (n = 0; n < 16; n += sizeof(crypto_word_t)) {
+      CRYPTO_store_word_le(
+          out + n, CRYPTO_load_word_le(in + n) ^ CRYPTO_load_word_le(iv + n));
     }
     (*block)(out, out, key);
     iv = out;
@@ -97,30 +102,36 @@ void CRYPTO_cbc128_encrypt(const uint8_t *in, uint8_t *out, size_t len,
 void CRYPTO_cbc128_decrypt(const uint8_t *in, uint8_t *out, size_t len,
                            const AES_KEY *key, uint8_t ivec[16],
                            block128_f block) {
-  size_t n;
-  union {
-    size_t t[16 / sizeof(size_t)];
-    uint8_t c[16];
-  } tmp;
-
   assert(key != NULL && ivec != NULL);
-  assert(len == 0 || (in != NULL && out != NULL));
+  if (len == 0) {
+    // Avoid |ivec| == |iv| in the |memcpy| below, which is not legal in C.
+    return;
+  }
+
+  assert(in != NULL && out != NULL);
 
   const uintptr_t inptr = (uintptr_t) in;
   const uintptr_t outptr = (uintptr_t) out;
   // If |in| and |out| alias, |in| must be ahead.
   assert(inptr >= outptr || inptr + len <= outptr);
 
+  size_t n;
+  union {
+    crypto_word_t t[16 / sizeof(crypto_word_t)];
+    uint8_t c[16];
+  } tmp;
+
   if ((inptr >= 32 && outptr <= inptr - 32) || inptr < outptr) {
     // If |out| is at least two blocks behind |in| or completely disjoint, there
     // is no need to decrypt to a temporary block.
-    OPENSSL_STATIC_ASSERT(16 % sizeof(size_t) == 0,
+    OPENSSL_STATIC_ASSERT(16 % sizeof(crypto_word_t) == 0,
                           "block cannot be evenly divided into words");
     const uint8_t *iv = ivec;
     while (len >= 16) {
       (*block)(in, out, key);
-      for (n = 0; n < 16; n += sizeof(size_t)) {
-        store_word_le(out + n, load_word_le(out + n) ^ load_word_le(iv + n));
+      for (n = 0; n < 16; n += sizeof(crypto_word_t)) {
+        CRYPTO_store_word_le(out + n, CRYPTO_load_word_le(out + n) ^
+                                          CRYPTO_load_word_le(iv + n));
       }
       iv = in;
       len -= 16;
@@ -129,16 +140,16 @@ void CRYPTO_cbc128_decrypt(const uint8_t *in, uint8_t *out, size_t len,
     }
     OPENSSL_memcpy(ivec, iv, 16);
   } else {
-    OPENSSL_STATIC_ASSERT(16 % sizeof(size_t) == 0,
+    OPENSSL_STATIC_ASSERT(16 % sizeof(crypto_word_t) == 0,
                           "block cannot be evenly divided into words");
 
     while (len >= 16) {
       (*block)(in, tmp.c, key);
-      for (n = 0; n < 16; n += sizeof(size_t)) {
-        size_t c = load_word_le(in + n);
-        store_word_le(out + n,
-                      tmp.t[n / sizeof(size_t)] ^ load_word_le(ivec + n));
-        store_word_le(ivec + n, c);
+      for (n = 0; n < 16; n += sizeof(crypto_word_t)) {
+        crypto_word_t c = CRYPTO_load_word_le(in + n);
+        CRYPTO_store_word_le(out + n, tmp.t[n / sizeof(crypto_word_t)] ^
+                                          CRYPTO_load_word_le(ivec + n));
+        CRYPTO_store_word_le(ivec + n, c);
       }
       len -= 16;
       in += 16;

data/third_party/boringssl-with-bazel/src/crypto/fipsmodule/modes/cfb.c

@@ -72,10 +72,11 @@ void CRYPTO_cfb128_encrypt(const uint8_t *in, uint8_t *out, size_t len,
     }
     while (len >= 16) {
       (*block)(ivec, ivec, key);
-      for (; n < 16; n += sizeof(size_t)) {
-        size_t tmp = load_word_le(ivec + n) ^ load_word_le(in + n);
-        store_word_le(ivec + n, tmp);
-        store_word_le(out + n, tmp);
+      for (; n < 16; n += sizeof(crypto_word_t)) {
+        crypto_word_t tmp =
+            CRYPTO_load_word_le(ivec + n) ^ CRYPTO_load_word_le(in + n);
+        CRYPTO_store_word_le(ivec + n, tmp);
+        CRYPTO_store_word_le(out + n, tmp);
       }
       len -= 16;
       out += 16;
@@ -101,10 +102,10 @@ void CRYPTO_cfb128_encrypt(const uint8_t *in, uint8_t *out, size_t len,
     }
     while (len >= 16) {
       (*block)(ivec, ivec, key);
-      for (; n < 16; n += sizeof(size_t)) {
-        size_t t = load_word_le(in + n);
-        store_word_le(out + n, load_word_le(ivec + n) ^ t);
-        store_word_le(ivec + n, t);
+      for (; n < 16; n += sizeof(crypto_word_t)) {
+        crypto_word_t t = CRYPTO_load_word_le(in + n);
+        CRYPTO_store_word_le(out + n, CRYPTO_load_word_le(ivec + n) ^ t);
+        CRYPTO_store_word_le(ivec + n, t);
       }
       len -= 16;
       out += 16;

data/third_party/boringssl-with-bazel/src/crypto/fipsmodule/modes/ctr.c

@@ -52,6 +52,7 @@
 #include <string.h>
 
 #include "internal.h"
+#include "../../internal.h"
 
 
 // NOTE: the IV/counter CTR mode is big-endian.  The code itself
@@ -69,8 +70,8 @@ static void ctr128_inc(uint8_t *counter) {
   } while (n);
 }
 
-OPENSSL_STATIC_ASSERT(16 % sizeof(size_t) == 0,
-                      "block cannot be divided into size_t");
+OPENSSL_STATIC_ASSERT(16 % sizeof(crypto_word_t) == 0,
+                      "block cannot be divided into crypto_word_t");
 
 // The input encrypted as though 128bit counter mode is being used.  The extra
 // state information to record how much of the 128bit block we have used is
@@ -102,9 +103,9 @@ void CRYPTO_ctr128_encrypt(const uint8_t *in, uint8_t *out, size_t len,
   while (len >= 16) {
     (*block)(ivec, ecount_buf, key);
     ctr128_inc(ivec);
-    for (n = 0; n < 16; n += sizeof(size_t)) {
-      store_word_le(out + n,
-                    load_word_le(in + n) ^ load_word_le(ecount_buf + n));
+    for (n = 0; n < 16; n += sizeof(crypto_word_t)) {
+      CRYPTO_store_word_le(out + n, CRYPTO_load_word_le(in + n) ^
+                                        CRYPTO_load_word_le(ecount_buf + n));
     }
     len -= 16;
     out += 16;
@@ -152,7 +153,7 @@ void CRYPTO_ctr128_encrypt_ctr32(const uint8_t *in, uint8_t *out, size_t len,
     n = (n + 1) % 16;
   }
 
-  ctr32 = GETU32(ivec + 12);
+  ctr32 = CRYPTO_load_u32_be(ivec + 12);
   while (len >= 16) {
     size_t blocks = len / 16;
     // 1<<28 is just a not-so-small yet not-so-large number...
@@ -172,7 +173,7 @@ void CRYPTO_ctr128_encrypt_ctr32(const uint8_t *in, uint8_t *out, size_t len,
     }
     (*func)(in, out, blocks, key, ivec);
     // (*func) does not update ivec, caller does:
-    PUTU32(ivec + 12, ctr32);
+    CRYPTO_store_u32_be(ivec + 12, ctr32);
     // ... overflow was detected, propogate carry.
     if (ctr32 == 0) {
       ctr96_inc(ivec);
@@ -186,7 +187,7 @@ void CRYPTO_ctr128_encrypt_ctr32(const uint8_t *in, uint8_t *out, size_t len,
     OPENSSL_memset(ecount_buf, 0, 16);
     (*func)(ecount_buf, ecount_buf, 1, key, ivec);
     ++ctr32;
-    PUTU32(ivec + 12, ctr32);
+    CRYPTO_store_u32_be(ivec + 12, ctr32);
     if (ctr32 == 0) {
       ctr96_inc(ivec);
     }

data/third_party/boringssl-with-bazel/src/crypto/fipsmodule/modes/gcm.c

@@ -73,7 +73,7 @@ static const size_t kSizeTWithoutLower4Bits = (size_t) -16;
 
 #if defined(GHASH_ASM_X86_64) || defined(GHASH_ASM_X86)
 static inline void gcm_reduce_1bit(u128 *V) {
-  if (sizeof(size_t) == 8) {
+  if (sizeof(crypto_word_t) == 8) {
     uint64_t T = UINT64_C(0xe100000000000000) & (0 - (V->hi & 1));
     V->hi = (V->lo << 63) | (V->hi >> 1);
     V->lo = (V->lo >> 1) ^ T;
@@ -377,9 +377,10 @@ int CRYPTO_gcm128_encrypt(GCM128_CONTEXT *ctx, const AES_KEY *key,
       (*block)(ctx->Yi.c, ctx->EKi.c, key);
       ++ctr;
       ctx->Yi.d[3] = CRYPTO_bswap4(ctr);
-      for (size_t i = 0; i < 16; i += sizeof(size_t)) {
-        store_word_le(out + i,
-                      load_word_le(in + i) ^ ctx->EKi.t[i / sizeof(size_t)]);
+      for (size_t i = 0; i < 16; i += sizeof(crypto_word_t)) {
+        CRYPTO_store_word_le(out + i,
+                             CRYPTO_load_word_le(in + i) ^
+                                 ctx->EKi.t[i / sizeof(crypto_word_t)]);
       }
       out += 16;
       in += 16;
@@ -394,9 +395,10 @@ int CRYPTO_gcm128_encrypt(GCM128_CONTEXT *ctx, const AES_KEY *key,
       (*block)(ctx->Yi.c, ctx->EKi.c, key);
       ++ctr;
       ctx->Yi.d[3] = CRYPTO_bswap4(ctr);
-      for (size_t i = 0; i < 16; i += sizeof(size_t)) {
-        store_word_le(out + i,
-                      load_word_le(in + i) ^ ctx->EKi.t[i / sizeof(size_t)]);
+      for (size_t i = 0; i < 16; i += sizeof(crypto_word_t)) {
+        CRYPTO_store_word_le(out + i,
+                             CRYPTO_load_word_le(in + i) ^
+                                 ctx->EKi.t[i / sizeof(crypto_word_t)]);
       }
       out += 16;
       in += 16;
@@ -468,9 +470,10 @@ int CRYPTO_gcm128_decrypt(GCM128_CONTEXT *ctx, const AES_KEY *key,
       (*block)(ctx->Yi.c, ctx->EKi.c, key);
       ++ctr;
       ctx->Yi.d[3] = CRYPTO_bswap4(ctr);
-      for (size_t i = 0; i < 16; i += sizeof(size_t)) {
-        store_word_le(out + i,
-                      load_word_le(in + i) ^ ctx->EKi.t[i / sizeof(size_t)]);
+      for (size_t i = 0; i < 16; i += sizeof(crypto_word_t)) {
+        CRYPTO_store_word_le(out + i,
+                             CRYPTO_load_word_le(in + i) ^
+                                 ctx->EKi.t[i / sizeof(crypto_word_t)]);
       }
       out += 16;
       in += 16;
@@ -485,9 +488,10 @@ int CRYPTO_gcm128_decrypt(GCM128_CONTEXT *ctx, const AES_KEY *key,
       (*block)(ctx->Yi.c, ctx->EKi.c, key);
       ++ctr;
       ctx->Yi.d[3] = CRYPTO_bswap4(ctr);
-      for (size_t i = 0; i < 16; i += sizeof(size_t)) {
-        store_word_le(out + i,
-                      load_word_le(in + i) ^ ctx->EKi.t[i / sizeof(size_t)]);
+      for (size_t i = 0; i < 16; i += sizeof(crypto_word_t)) {
+        CRYPTO_store_word_le(out + i,
+                             CRYPTO_load_word_le(in + i) ^
+                                 ctx->EKi.t[i / sizeof(crypto_word_t)]);
       }
       out += 16;
       in += 16;