Merge branch 'master' of https://github.com/guanzhi/GmSSL

CMakeLists.txt

@@ -91,21 +91,6 @@ if (NOT MSVC)
 	endif()
 endif()
 
-if (APPLE)
-	include(CheckCSourceCompiles)
-	set(GMSSL_CMAKE_REQUIRED_LIBRARIES_SAVE "${CMAKE_REQUIRED_LIBRARIES}")
-	set(CMAKE_REQUIRED_LIBRARIES "-framework OpenCL")
-	check_c_source_compiles(
-		"#include <OpenCL/OpenCL.h>
-		int main(void) { cl_uint n = 0; clGetPlatformIDs(0, 0, &n); return 0; }"
-		GMSSL_HAVE_MACOS_OPENCL)
-	set(CMAKE_REQUIRED_LIBRARIES "${GMSSL_CMAKE_REQUIRED_LIBRARIES_SAVE}")
-	if (GMSSL_HAVE_MACOS_OPENCL)
-		set(GMSSL_DEFAULT_ENABLE_LMS_CL ON)
-		set(GMSSL_DEFAULT_ENABLE_XMSS_CL ON)
-	endif()
-endif()
-
 option(ENABLE_TEST_SPEED "Enable test speed" OFF)
 option(ENABLE_SLOW_TEST "Enable slow tests" OFF)
 
@@ -165,7 +150,11 @@ option(ENABLE_GHASH "Enable standalone GHASH command and test" ON)
 option(ENABLE_SKF "Enable SKF module" ON)
 option(ENABLE_SDF "Enable SDF module" ON)
 
+if (APPLE)
 	option(ENABLE_ASM_UNDERSCORE_PREFIX "Add prefix `_` to assembly symbols" ON)
+else()
+	option(ENABLE_ASM_UNDERSCORE_PREFIX "Add prefix `_` to assembly symbols" OFF)
+endif()
 
 option(ENABLE_TLS "Enable TLS and TLCP protocol support" ON)
 option(ENABLE_QUIC "Enable QUIC support" ON)

cmake/tool_pqc.cmake

@@ -21,9 +21,9 @@ endif()
 
 if(ENABLE_XMSS)
 	gmssl_signature_roundtrip(tool_xmss xmsskeygen xmsssign xmssverify
-		KEYGEN_ARGS -xmss_type XMSS_SHA2_10_256)
+		KEYGEN_ARGS -xmss_type XMSS_SM3_10_256)
 	gmssl_signature_roundtrip(tool_xmssmt xmssmtkeygen xmssmtsign xmssmtverify
-		KEYGEN_ARGS -xmssmt_type XMSSMT_SHA2_20_2_256)
+		KEYGEN_ARGS -xmssmt_type XMSSMT_SM3_20_2_256)
 endif()
 
 if(ENABLE_SPHINCS)

src/secp256r1.c

@@ -1526,6 +1526,8 @@ int secp256r1_ecdsa_signature_from_der(SECP256R1_ECDSA_SIGNATURE *sig, const uin
 	const uint8_t *d;
 	const uint8_t *r;
 	const uint8_t *s;
+	uint8_t rbuf[32] = {0};
+	uint8_t sbuf[32] = {0};
 	size_t dlen, rlen, slen;
 
 	if ((ret = asn1_sequence_from_der(&d, &dlen, in, inlen)) != 1) {
@@ -1541,8 +1543,10 @@ int secp256r1_ecdsa_signature_from_der(SECP256R1_ECDSA_SIGNATURE *sig, const uin
 		return -1;
 	}
 
-	if (secp256r1_from_32bytes(sig->r, r) != 1
-		|| secp256r1_from_32bytes(sig->s, s) != 1) {
+	memcpy(rbuf + sizeof(rbuf) - rlen, r, rlen);
+	memcpy(sbuf + sizeof(sbuf) - slen, s, slen);
+	if (secp256r1_from_32bytes(sig->r, rbuf) != 1
+		|| secp256r1_from_32bytes(sig->s, sbuf) != 1) {
 		error_print();
 		return -1;
 	}

src/sm3_arm64.c

@@ -128,7 +128,7 @@ void sm3_compress_blocks(uint32_t digest[8], const uint8_t *data, size_t blocks)
 
 		for (j = 0; j < 16; j += 4) {
 			words = vld1q_u32((uint32_t *)data + j);
-			words = vrev32q_u8(words);
+			words = (uint32x4_t)vrev32q_u8((uint8x16_t)words);
 			vst1q_u32(W + j, words);
 		}
 

src/sm4_arm64.c

@@ -140,7 +140,7 @@ void sm4_encrypt(const SM4_KEY *key, const unsigned char in[16], unsigned char o
 	uint8x16x4_t S1 = vld1q_u8_x4(S + 64);
 	uint8x16x4_t S2 = vld1q_u8_x4(S + 128);
 	uint8x16x4_t S3 = vld1q_u8_x4(S + 192);
-	uint8x16_t vx;
+	uint32x4_t vx;
 	uint8x16_t vt;
 	uint32_t X0, X1, X2, X3, X4;
 	int i;
@@ -156,10 +156,10 @@ void sm4_encrypt(const SM4_KEY *key, const unsigned char in[16], unsigned char o
 
 		// const time X4 = S32(X4)
 		vx = vdupq_n_u32(X4);
-		vt = vqtbl4q_u8(S0, vx);
-		vt = vqtbx4q_u8(vt, S1, veorq_u8(vx, vdupq_n_u8(0x40)));
-		vt = vqtbx4q_u8(vt, S2, veorq_u8(vx, vdupq_n_u8(0x80)));
-		vx = vqtbx4q_u8(vt, S3, veorq_u8(vx, vdupq_n_u8(0xc0)));
+		vt = vqtbl4q_u8(S0, (uint8x16_t)vx);
+		vt = vqtbx4q_u8(vt, S1, veorq_u8((uint8x16_t)vx, vdupq_n_u8(0x40)));
+		vt = vqtbx4q_u8(vt, S2, veorq_u8((uint8x16_t)vx, vdupq_n_u8(0x80)));
+		vx = (uint32x4_t)vqtbx4q_u8(vt, S3, veorq_u8((uint8x16_t)vx, vdupq_n_u8(0xc0)));
 		X4 = vgetq_lane_u32(vx, 0);
 
 		X4 = X0 ^ L32(X4);
@@ -278,7 +278,7 @@ void sm4_ctr32_encrypt_4blocks(const SM4_KEY *key, uint8_t iv[16], const uint8_t
 
 	memcpy(buf, iv, 16);
 	ctr = vld1q_u32(buf);
-	ctr = vrev32q_u8(ctr);
+	ctr = (uint32x4_t)vrev32q_u8((uint8x16_t)ctr);
 
 	ctr0 = vdupq_n_u32(vgetq_lane_u32(ctr, 0));
 	ctr1 = vdupq_n_u32(vgetq_lane_u32(ctr, 1));
@@ -301,10 +301,10 @@ void sm4_ctr32_encrypt_4blocks(const SM4_KEY *key, uint8_t iv[16], const uint8_t
 			x4 = veorq_u32(veorq_u32(x1, x2), veorq_u32(x3, rk));
 
 			// X4 = SBOX(X4)
-			xt = vqtbl4q_u8(S0, x4);
-			xt = vqtbx4q_u8(xt, S1, veorq_u8(x4, vdupq_n_u8(0x40)));
-			xt = vqtbx4q_u8(xt, S2, veorq_u8(x4, vdupq_n_u8(0x80)));
-			x4 = vqtbx4q_u8(xt, S3, veorq_u8(x4, vdupq_n_u8(0xc0)));
+			xt = (uint32x4_t)vqtbl4q_u8(S0, (uint8x16_t)x4);
+			xt = (uint32x4_t)vqtbx4q_u8((uint8x16_t)xt, S1, veorq_u8((uint8x16_t)x4, vdupq_n_u8(0x40)));
+			xt = (uint32x4_t)vqtbx4q_u8((uint8x16_t)xt, S2, veorq_u8((uint8x16_t)x4, vdupq_n_u8(0x80)));
+			x4 = (uint32x4_t)vqtbx4q_u8((uint8x16_t)xt, S3, veorq_u8((uint8x16_t)x4, vdupq_n_u8(0xc0)));
 
 			// X4 = L(X4)
 			xt = veorq_u32(x4, vrolq_n_u32(x4,  2));
@@ -326,16 +326,16 @@ void sm4_ctr32_encrypt_4blocks(const SM4_KEY *key, uint8_t iv[16], const uint8_t
 		x01 = vzipq_u32(x02.val[0], x13.val[0]);
 		x23 = vzipq_u32(x02.val[1], x13.val[1]);
 
-		x0 = vrev32q_u8(x01.val[0]);
+		x0 = (uint32x4_t)vrev32q_u8((uint8x16_t)x01.val[0]);
 		vst1q_u32(buf, x0);
 
-		x1 = vrev32q_u8(x01.val[1]);
+		x1 = (uint32x4_t)vrev32q_u8((uint8x16_t)x01.val[1]);
 		vst1q_u32(buf + 4, x1);
 
-		x2 = vrev32q_u8(x23.val[0]);
+		x2 = (uint32x4_t)vrev32q_u8((uint8x16_t)x23.val[0]);
 		vst1q_u32(buf + 8, x2);
 
-		x3 = vrev32q_u8(x23.val[1]);
+		x3 = (uint32x4_t)vrev32q_u8((uint8x16_t)x23.val[1]);
 		vst1q_u32(buf + 12, x3);
 
 		// xor with plaintext

src/tls13.c

@@ -5874,9 +5874,9 @@ int tls13_recv_certificate_request(TLS_CONNECT *conn)
 	int signed_certificate_timestamp = 0;
 
 	int common_sig_algs[4];
-	size_t common_sig_algs_cnt;
+	size_t common_sig_algs_cnt = 0;
 	int common_sig_algs_cert[4];
-	size_t common_sig_algs_cert_cnt;
+	size_t common_sig_algs_cert_cnt = 0;
 	const uint8_t *ca_names = NULL;
 	size_t ca_names_len = 0;
 	const uint8_t *filters = NULL;

tests/secp384r1test.c

@@ -75,6 +75,8 @@ static int test_secp384r1(void)
 {
 	secp384r1_t a;
 	secp384r1_t b;
+	secp384r1_t p;
+	secp384r1_t n;
 	uint8_t buf[48];
 
 	if (secp384r1_set_zero(a) != 1 || !secp384r1_is_zero(a)) {
@@ -101,9 +103,12 @@ static int test_secp384r1(void)
 		error_print();
 		return -1;
 	}
-	if (secp384r1_check_hex(SECP384R1_P, secp384r1_p) != 1
-		|| secp384r1_check_hex(SECP384R1_B, secp384r1_b) != 1
-		|| secp384r1_check_hex(SECP384R1_N, secp384r1_n) != 1) {
+	if (secp384r1_from_hex(p, secp384r1_p) != 1
+		|| secp384r1_from_hex(b, secp384r1_b) != 1
+		|| secp384r1_from_hex(n, secp384r1_n) != 1
+		|| secp384r1_check_hex(p, secp384r1_p) != 1
+		|| secp384r1_check_hex(b, secp384r1_b) != 1
+		|| secp384r1_check_hex(n, secp384r1_n) != 1) {
 		error_print();
 		return -1;
 	}
@@ -232,6 +237,7 @@ static int test_secp384r1_point(void)
 	SECP384R1_POINT Q;
 	SECP384R1_POINT R;
 	secp384r1_t k;
+	secp384r1_t n;
 	uint8_t octets[97];
 
 	if (secp384r1_point_set_xy(&P, SECP384R1_POINT_G.X, SECP384R1_POINT_G.Y) != 1
@@ -261,7 +267,8 @@ static int test_secp384r1_point(void)
 		error_print();
 		return -1;
 	}
-	if (secp384r1_point_mul_generator(&R, SECP384R1_N) != 1
+	if (secp384r1_from_hex(n, secp384r1_n) != 1
+		|| secp384r1_point_mul_generator(&R, n) != 1
 		|| !secp384r1_point_is_at_infinity(&R)) {
 		error_print();
 		return -1;