mirror of
https://github.com/cesanta/mongoose.git
synced 2024-11-27 12:49:01 +08:00
Add wolfSSL support
- Added Makefile wolfSSL build option with `SSL=WOLFSSL` - Fixed missing BIO symbol errors - Rename `fe` to `mg_fe` as it collides with `fe` defined in `wolfssl/wolfcrypt/fe_operations.h` - byte variable name shadows the wolfSSL byte type Tested with `make test` Signed-off-by: Juliusz Sosinowicz <juliusz@wolfssl.com>
This commit is contained in:
parent
323ba9d90c
commit
dc09851ac8
53
mongoose.c
53
mongoose.c
@ -4147,10 +4147,10 @@ static size_t encode_varint(uint8_t *buf, size_t value) {
|
||||
size_t len = 0;
|
||||
|
||||
do {
|
||||
uint8_t byte = (uint8_t) (value % 128);
|
||||
uint8_t b = (uint8_t) (value % 128);
|
||||
value /= 128;
|
||||
if (value > 0) byte |= 0x80;
|
||||
buf[len++] = byte;
|
||||
if (value > 0) b |= 0x80;
|
||||
buf[len++] = b;
|
||||
} while (value > 0);
|
||||
|
||||
return len;
|
||||
@ -11040,7 +11040,9 @@ static long mg_bio_ctrl(BIO *b, int cmd, long larg, void *pargs) {
|
||||
if (cmd == BIO_CTRL_PUSH) ret = 1;
|
||||
if (cmd == BIO_CTRL_POP) ret = 1;
|
||||
if (cmd == BIO_CTRL_FLUSH) ret = 1;
|
||||
#ifndef OPENSSL_IS_WOLFSSL
|
||||
if (cmd == BIO_C_SET_NBIO) ret = 1;
|
||||
#endif
|
||||
// MG_DEBUG(("%d -> %ld", cmd, ret));
|
||||
(void) b, (void) cmd, (void) larg, (void) pargs;
|
||||
return ret;
|
||||
@ -11132,7 +11134,7 @@ void mg_tls_init(struct mg_connection *c, const struct mg_tls_opts *opts) {
|
||||
}
|
||||
|
||||
SSL_set_mode(tls->ssl, SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER);
|
||||
#if OPENSSL_VERSION_NUMBER > 0x10002000L
|
||||
#if OPENSSL_VERSION_NUMBER > 0x10002000L && !defined(OPENSSL_IS_WOLFSSL)
|
||||
(void) SSL_set_ecdh_auto(tls->ssl, 1);
|
||||
#endif
|
||||
#if OPENSSL_VERSION_NUMBER >= 0x10100000L
|
||||
@ -11143,8 +11145,11 @@ void mg_tls_init(struct mg_connection *c, const struct mg_tls_opts *opts) {
|
||||
free(s);
|
||||
}
|
||||
#endif
|
||||
|
||||
#ifndef OPENSSL_IS_WOLFSSL
|
||||
tls->bm = BIO_meth_new(BIO_get_new_index() | BIO_TYPE_SOURCE_SINK, "bio_mg");
|
||||
#else
|
||||
tls->bm = BIO_meth_new(0, "bio_mg");
|
||||
#endif
|
||||
BIO_meth_set_write(tls->bm, mg_bio_write);
|
||||
BIO_meth_set_read(tls->bm, mg_bio_read);
|
||||
BIO_meth_set_ctrl(tls->bm, mg_bio_ctrl);
|
||||
@ -14468,7 +14473,7 @@ typedef uint64_t dlimb_t;
|
||||
typedef int64_t sdlimb_t;
|
||||
|
||||
#define NLIMBS (256 / X25519_WBITS)
|
||||
typedef limb_t fe[NLIMBS];
|
||||
typedef limb_t mg_fe[NLIMBS];
|
||||
|
||||
static limb_t umaal(limb_t *carry, limb_t acc, limb_t mand, limb_t mier) {
|
||||
dlimb_t tmp = (dlimb_t) mand * mier + acc + *carry;
|
||||
@ -14493,7 +14498,7 @@ static limb_t adc0(limb_t *carry, limb_t acc) {
|
||||
// - Invariant: result of propagate is < 2^255 + 1 word
|
||||
// - In particular, always less than 2p.
|
||||
// - Also, output x >= min(x,19)
|
||||
static void propagate(fe x, limb_t over) {
|
||||
static void propagate(mg_fe x, limb_t over) {
|
||||
unsigned i;
|
||||
limb_t carry;
|
||||
over = x[NLIMBS - 1] >> (X25519_WBITS - 1) | over << 1;
|
||||
@ -14505,7 +14510,7 @@ static void propagate(fe x, limb_t over) {
|
||||
}
|
||||
}
|
||||
|
||||
static void add(fe out, const fe a, const fe b) {
|
||||
static void add(mg_fe out, const mg_fe a, const mg_fe b) {
|
||||
unsigned i;
|
||||
limb_t carry = 0;
|
||||
for (i = 0; i < NLIMBS; i++) {
|
||||
@ -14514,7 +14519,7 @@ static void add(fe out, const fe a, const fe b) {
|
||||
propagate(out, carry);
|
||||
}
|
||||
|
||||
static void sub(fe out, const fe a, const fe b) {
|
||||
static void sub(mg_fe out, const mg_fe a, const mg_fe b) {
|
||||
unsigned i;
|
||||
sdlimb_t carry = -38;
|
||||
for (i = 0; i < NLIMBS; i++) {
|
||||
@ -14525,9 +14530,9 @@ static void sub(fe out, const fe a, const fe b) {
|
||||
propagate(out, (limb_t) (1 + carry));
|
||||
}
|
||||
|
||||
// `b` can contain less than 8 limbs, thus we use `limb_t *` instead of `fe`
|
||||
// `b` can contain less than 8 limbs, thus we use `limb_t *` instead of `mg_fe`
|
||||
// to avoid build warnings
|
||||
static void mul(fe out, const fe a, const limb_t *b, unsigned nb) {
|
||||
static void mul(mg_fe out, const mg_fe a, const limb_t *b, unsigned nb) {
|
||||
limb_t accum[2 * NLIMBS] = {0};
|
||||
unsigned i, j;
|
||||
|
||||
@ -14550,13 +14555,13 @@ static void mul(fe out, const fe a, const limb_t *b, unsigned nb) {
|
||||
propagate(out, carry2);
|
||||
}
|
||||
|
||||
static void sqr(fe out, const fe a) {
|
||||
static void sqr(mg_fe out, const mg_fe a) {
|
||||
mul(out, a, a, NLIMBS);
|
||||
}
|
||||
static void mul1(fe out, const fe a) {
|
||||
static void mul1(mg_fe out, const mg_fe a) {
|
||||
mul(out, a, out, NLIMBS);
|
||||
}
|
||||
static void sqr1(fe a) {
|
||||
static void sqr1(mg_fe a) {
|
||||
mul1(a, a);
|
||||
}
|
||||
|
||||
@ -14573,7 +14578,7 @@ static void condswap(limb_t a[2 * NLIMBS], limb_t b[2 * NLIMBS],
|
||||
// Canonicalize a field element x, reducing it to the least residue which is
|
||||
// congruent to it mod 2^255-19
|
||||
// - Precondition: x < 2^255 + 1 word
|
||||
static limb_t canon(fe x) {
|
||||
static limb_t canon(mg_fe x) {
|
||||
// First, add 19.
|
||||
unsigned i;
|
||||
limb_t carry0 = 19;
|
||||
@ -14604,7 +14609,7 @@ static limb_t canon(fe x) {
|
||||
|
||||
static const limb_t a24[1] = {121665};
|
||||
|
||||
static void ladder_part1(fe xs[5]) {
|
||||
static void ladder_part1(mg_fe xs[5]) {
|
||||
limb_t *x2 = xs[0], *z2 = xs[1], *x3 = xs[2], *z3 = xs[3], *t1 = xs[4];
|
||||
add(t1, x2, z2); // t1 = A
|
||||
sub(z2, x2, z2); // z2 = B
|
||||
@ -14621,7 +14626,7 @@ static void ladder_part1(fe xs[5]) {
|
||||
add(z2, z2, t1); // z2 = E*a24 + AA
|
||||
}
|
||||
|
||||
static void ladder_part2(fe xs[5], const fe x1) {
|
||||
static void ladder_part2(mg_fe xs[5], const mg_fe x1) {
|
||||
limb_t *x2 = xs[0], *z2 = xs[1], *x3 = xs[2], *z3 = xs[3], *t1 = xs[4];
|
||||
sqr1(z3); // z3 = (DA-CB)^2
|
||||
mul1(z3, x1); // z3 = x1 * (DA-CB)^2
|
||||
@ -14631,13 +14636,13 @@ static void ladder_part2(fe xs[5], const fe x1) {
|
||||
mul1(x2, t1); // x2 = AA*BB
|
||||
}
|
||||
|
||||
static void x25519_core(fe xs[5], const uint8_t scalar[X25519_BYTES],
|
||||
static void x25519_core(mg_fe xs[5], const uint8_t scalar[X25519_BYTES],
|
||||
const uint8_t *x1, int clamp) {
|
||||
int i;
|
||||
fe x1_limbs;
|
||||
mg_fe x1_limbs;
|
||||
limb_t swap = 0;
|
||||
limb_t *x2 = xs[0], *x3 = xs[2], *z3 = xs[3];
|
||||
memset(xs, 0, 4 * sizeof(fe));
|
||||
memset(xs, 0, 4 * sizeof(mg_fe));
|
||||
x2[0] = z3[0] = 1;
|
||||
for (i = 0; i < NLIMBS; i++) {
|
||||
x3[i] = x1_limbs[i] =
|
||||
@ -14668,7 +14673,7 @@ static void x25519_core(fe xs[5], const uint8_t scalar[X25519_BYTES],
|
||||
int mg_tls_x25519(uint8_t out[X25519_BYTES], const uint8_t scalar[X25519_BYTES],
|
||||
const uint8_t x1[X25519_BYTES], int clamp) {
|
||||
int i, ret;
|
||||
fe xs[5], out_limbs;
|
||||
mg_fe xs[5], out_limbs;
|
||||
limb_t *x2, *z2, *z3, *prev;
|
||||
static const struct {
|
||||
uint8_t a, c, n;
|
||||
@ -14865,9 +14870,9 @@ uint32_t mg_crc32(uint32_t crc, const char *buf, size_t len) {
|
||||
0x9B64C2B0, 0x86D3D2D4, 0xA00AE278, 0xBDBDF21C};
|
||||
crc = ~crc;
|
||||
while (len--) {
|
||||
uint8_t byte = *(uint8_t *) buf++;
|
||||
crc = crclut[(crc ^ byte) & 0x0F] ^ (crc >> 4);
|
||||
crc = crclut[(crc ^ (byte >> 4)) & 0x0F] ^ (crc >> 4);
|
||||
uint8_t b = *(uint8_t *) buf++;
|
||||
crc = crclut[(crc ^ b) & 0x0F] ^ (crc >> 4);
|
||||
crc = crclut[(crc ^ (b >> 4)) & 0x0F] ^ (crc >> 4);
|
||||
}
|
||||
return ~crc;
|
||||
}
|
||||
|
@ -80,10 +80,10 @@ static size_t encode_varint(uint8_t *buf, size_t value) {
|
||||
size_t len = 0;
|
||||
|
||||
do {
|
||||
uint8_t byte = (uint8_t) (value % 128);
|
||||
uint8_t b = (uint8_t) (value % 128);
|
||||
value /= 128;
|
||||
if (value > 0) byte |= 0x80;
|
||||
buf[len++] = byte;
|
||||
if (value > 0) b |= 0x80;
|
||||
buf[len++] = b;
|
||||
} while (value > 0);
|
||||
|
||||
return len;
|
||||
|
@ -67,7 +67,9 @@ static long mg_bio_ctrl(BIO *b, int cmd, long larg, void *pargs) {
|
||||
if (cmd == BIO_CTRL_PUSH) ret = 1;
|
||||
if (cmd == BIO_CTRL_POP) ret = 1;
|
||||
if (cmd == BIO_CTRL_FLUSH) ret = 1;
|
||||
#ifndef OPENSSL_IS_WOLFSSL
|
||||
if (cmd == BIO_C_SET_NBIO) ret = 1;
|
||||
#endif
|
||||
// MG_DEBUG(("%d -> %ld", cmd, ret));
|
||||
(void) b, (void) cmd, (void) larg, (void) pargs;
|
||||
return ret;
|
||||
@ -159,7 +161,7 @@ void mg_tls_init(struct mg_connection *c, const struct mg_tls_opts *opts) {
|
||||
}
|
||||
|
||||
SSL_set_mode(tls->ssl, SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER);
|
||||
#if OPENSSL_VERSION_NUMBER > 0x10002000L
|
||||
#if OPENSSL_VERSION_NUMBER > 0x10002000L && !defined(OPENSSL_IS_WOLFSSL)
|
||||
(void) SSL_set_ecdh_auto(tls->ssl, 1);
|
||||
#endif
|
||||
#if OPENSSL_VERSION_NUMBER >= 0x10100000L
|
||||
@ -170,8 +172,11 @@ void mg_tls_init(struct mg_connection *c, const struct mg_tls_opts *opts) {
|
||||
free(s);
|
||||
}
|
||||
#endif
|
||||
|
||||
#ifndef OPENSSL_IS_WOLFSSL
|
||||
tls->bm = BIO_meth_new(BIO_get_new_index() | BIO_TYPE_SOURCE_SINK, "bio_mg");
|
||||
#else
|
||||
tls->bm = BIO_meth_new(0, "bio_mg");
|
||||
#endif
|
||||
BIO_meth_set_write(tls->bm, mg_bio_write);
|
||||
BIO_meth_set_read(tls->bm, mg_bio_read);
|
||||
BIO_meth_set_ctrl(tls->bm, mg_bio_ctrl);
|
||||
|
@ -16,7 +16,7 @@ typedef uint64_t dlimb_t;
|
||||
typedef int64_t sdlimb_t;
|
||||
|
||||
#define NLIMBS (256 / X25519_WBITS)
|
||||
typedef limb_t fe[NLIMBS];
|
||||
typedef limb_t mg_fe[NLIMBS];
|
||||
|
||||
static limb_t umaal(limb_t *carry, limb_t acc, limb_t mand, limb_t mier) {
|
||||
dlimb_t tmp = (dlimb_t) mand * mier + acc + *carry;
|
||||
@ -41,7 +41,7 @@ static limb_t adc0(limb_t *carry, limb_t acc) {
|
||||
// - Invariant: result of propagate is < 2^255 + 1 word
|
||||
// - In particular, always less than 2p.
|
||||
// - Also, output x >= min(x,19)
|
||||
static void propagate(fe x, limb_t over) {
|
||||
static void propagate(mg_fe x, limb_t over) {
|
||||
unsigned i;
|
||||
limb_t carry;
|
||||
over = x[NLIMBS - 1] >> (X25519_WBITS - 1) | over << 1;
|
||||
@ -53,7 +53,7 @@ static void propagate(fe x, limb_t over) {
|
||||
}
|
||||
}
|
||||
|
||||
static void add(fe out, const fe a, const fe b) {
|
||||
static void add(mg_fe out, const mg_fe a, const mg_fe b) {
|
||||
unsigned i;
|
||||
limb_t carry = 0;
|
||||
for (i = 0; i < NLIMBS; i++) {
|
||||
@ -62,7 +62,7 @@ static void add(fe out, const fe a, const fe b) {
|
||||
propagate(out, carry);
|
||||
}
|
||||
|
||||
static void sub(fe out, const fe a, const fe b) {
|
||||
static void sub(mg_fe out, const mg_fe a, const mg_fe b) {
|
||||
unsigned i;
|
||||
sdlimb_t carry = -38;
|
||||
for (i = 0; i < NLIMBS; i++) {
|
||||
@ -73,9 +73,9 @@ static void sub(fe out, const fe a, const fe b) {
|
||||
propagate(out, (limb_t) (1 + carry));
|
||||
}
|
||||
|
||||
// `b` can contain less than 8 limbs, thus we use `limb_t *` instead of `fe`
|
||||
// `b` can contain less than 8 limbs, thus we use `limb_t *` instead of `mg_fe`
|
||||
// to avoid build warnings
|
||||
static void mul(fe out, const fe a, const limb_t *b, unsigned nb) {
|
||||
static void mul(mg_fe out, const mg_fe a, const limb_t *b, unsigned nb) {
|
||||
limb_t accum[2 * NLIMBS] = {0};
|
||||
unsigned i, j;
|
||||
|
||||
@ -98,13 +98,13 @@ static void mul(fe out, const fe a, const limb_t *b, unsigned nb) {
|
||||
propagate(out, carry2);
|
||||
}
|
||||
|
||||
static void sqr(fe out, const fe a) {
|
||||
static void sqr(mg_fe out, const mg_fe a) {
|
||||
mul(out, a, a, NLIMBS);
|
||||
}
|
||||
static void mul1(fe out, const fe a) {
|
||||
static void mul1(mg_fe out, const mg_fe a) {
|
||||
mul(out, a, out, NLIMBS);
|
||||
}
|
||||
static void sqr1(fe a) {
|
||||
static void sqr1(mg_fe a) {
|
||||
mul1(a, a);
|
||||
}
|
||||
|
||||
@ -121,7 +121,7 @@ static void condswap(limb_t a[2 * NLIMBS], limb_t b[2 * NLIMBS],
|
||||
// Canonicalize a field element x, reducing it to the least residue which is
|
||||
// congruent to it mod 2^255-19
|
||||
// - Precondition: x < 2^255 + 1 word
|
||||
static limb_t canon(fe x) {
|
||||
static limb_t canon(mg_fe x) {
|
||||
// First, add 19.
|
||||
unsigned i;
|
||||
limb_t carry0 = 19;
|
||||
@ -152,7 +152,7 @@ static limb_t canon(fe x) {
|
||||
|
||||
static const limb_t a24[1] = {121665};
|
||||
|
||||
static void ladder_part1(fe xs[5]) {
|
||||
static void ladder_part1(mg_fe xs[5]) {
|
||||
limb_t *x2 = xs[0], *z2 = xs[1], *x3 = xs[2], *z3 = xs[3], *t1 = xs[4];
|
||||
add(t1, x2, z2); // t1 = A
|
||||
sub(z2, x2, z2); // z2 = B
|
||||
@ -169,7 +169,7 @@ static void ladder_part1(fe xs[5]) {
|
||||
add(z2, z2, t1); // z2 = E*a24 + AA
|
||||
}
|
||||
|
||||
static void ladder_part2(fe xs[5], const fe x1) {
|
||||
static void ladder_part2(mg_fe xs[5], const mg_fe x1) {
|
||||
limb_t *x2 = xs[0], *z2 = xs[1], *x3 = xs[2], *z3 = xs[3], *t1 = xs[4];
|
||||
sqr1(z3); // z3 = (DA-CB)^2
|
||||
mul1(z3, x1); // z3 = x1 * (DA-CB)^2
|
||||
@ -179,13 +179,13 @@ static void ladder_part2(fe xs[5], const fe x1) {
|
||||
mul1(x2, t1); // x2 = AA*BB
|
||||
}
|
||||
|
||||
static void x25519_core(fe xs[5], const uint8_t scalar[X25519_BYTES],
|
||||
static void x25519_core(mg_fe xs[5], const uint8_t scalar[X25519_BYTES],
|
||||
const uint8_t *x1, int clamp) {
|
||||
int i;
|
||||
fe x1_limbs;
|
||||
mg_fe x1_limbs;
|
||||
limb_t swap = 0;
|
||||
limb_t *x2 = xs[0], *x3 = xs[2], *z3 = xs[3];
|
||||
memset(xs, 0, 4 * sizeof(fe));
|
||||
memset(xs, 0, 4 * sizeof(mg_fe));
|
||||
x2[0] = z3[0] = 1;
|
||||
for (i = 0; i < NLIMBS; i++) {
|
||||
x3[i] = x1_limbs[i] =
|
||||
@ -216,7 +216,7 @@ static void x25519_core(fe xs[5], const uint8_t scalar[X25519_BYTES],
|
||||
int mg_tls_x25519(uint8_t out[X25519_BYTES], const uint8_t scalar[X25519_BYTES],
|
||||
const uint8_t x1[X25519_BYTES], int clamp) {
|
||||
int i, ret;
|
||||
fe xs[5], out_limbs;
|
||||
mg_fe xs[5], out_limbs;
|
||||
limb_t *x2, *z2, *z3, *prev;
|
||||
static const struct {
|
||||
uint8_t a, c, n;
|
||||
|
@ -63,9 +63,9 @@ uint32_t mg_crc32(uint32_t crc, const char *buf, size_t len) {
|
||||
0x9B64C2B0, 0x86D3D2D4, 0xA00AE278, 0xBDBDF21C};
|
||||
crc = ~crc;
|
||||
while (len--) {
|
||||
uint8_t byte = *(uint8_t *) buf++;
|
||||
crc = crclut[(crc ^ byte) & 0x0F] ^ (crc >> 4);
|
||||
crc = crclut[(crc ^ (byte >> 4)) & 0x0F] ^ (crc >> 4);
|
||||
uint8_t b = *(uint8_t *) buf++;
|
||||
crc = crclut[(crc ^ b) & 0x0F] ^ (crc >> 4);
|
||||
crc = crclut[(crc ^ (b >> 4)) & 0x0F] ^ (crc >> 4);
|
||||
}
|
||||
return ~crc;
|
||||
}
|
||||
|
@ -53,6 +53,12 @@ CFLAGS += -DMG_TLS=MG_TLS_OPENSSL -I$(OPENSSL)/include
|
||||
LDFLAGS += -L$(OPENSSL)/lib -lssl -lcrypto
|
||||
endif
|
||||
|
||||
ifeq "$(SSL)" "WOLFSSL"
|
||||
WOLFSSL ?= /usr/local
|
||||
CFLAGS += -DMG_TLS=MG_TLS_OPENSSL -I$(WOLFSSL)/include -I$(WOLFSSL)/include/wolfssl -DEXTERNAL_OPTS_OPENVPN
|
||||
LDFLAGS += -L$(WOLFSSL)/lib -lwolfssl
|
||||
endif
|
||||
|
||||
ifeq "$(SSL)" "BUILTIN"
|
||||
CFLAGS += -DMG_TLS=MG_TLS_BUILTIN
|
||||
endif
|
||||
|
Loading…
Reference in New Issue
Block a user