From 378d4ce7552df580e3ddd89c2faa9f8c5086d646 Mon Sep 17 00:00:00 2001 From: Uros Majstorovic Date: Wed, 2 Feb 2022 06:25:38 +0100 Subject: renamed crypto -> ext --- ext/libressl/crypto/modes/Makefile | 14 + ext/libressl/crypto/modes/cbc128.c | 202 +++ ext/libressl/crypto/modes/ccm128.c | 441 ++++++ ext/libressl/crypto/modes/cfb128.c | 234 ++++ ext/libressl/crypto/modes/ctr128.c | 251 ++++ ext/libressl/crypto/modes/cts128.c | 267 ++++ ext/libressl/crypto/modes/gcm128.c | 1566 ++++++++++++++++++++++ ext/libressl/crypto/modes/ghash-elf-armv4.S | 412 ++++++ ext/libressl/crypto/modes/ghash-elf-x86_64.S | 1030 ++++++++++++++ ext/libressl/crypto/modes/ghash-macosx-x86_64.S | 1027 ++++++++++++++ ext/libressl/crypto/modes/ghash-masm-x86_64.S | 1256 +++++++++++++++++ ext/libressl/crypto/modes/ghash-mingw64-x86_64.S | 1175 ++++++++++++++++ ext/libressl/crypto/modes/modes_lcl.h | 113 ++ ext/libressl/crypto/modes/ofb128.c | 119 ++ ext/libressl/crypto/modes/xts128.c | 185 +++ 15 files changed, 8292 insertions(+) create mode 100644 ext/libressl/crypto/modes/Makefile create mode 100644 ext/libressl/crypto/modes/cbc128.c create mode 100644 ext/libressl/crypto/modes/ccm128.c create mode 100644 ext/libressl/crypto/modes/cfb128.c create mode 100644 ext/libressl/crypto/modes/ctr128.c create mode 100644 ext/libressl/crypto/modes/cts128.c create mode 100644 ext/libressl/crypto/modes/gcm128.c create mode 100644 ext/libressl/crypto/modes/ghash-elf-armv4.S create mode 100644 ext/libressl/crypto/modes/ghash-elf-x86_64.S create mode 100644 ext/libressl/crypto/modes/ghash-macosx-x86_64.S create mode 100644 ext/libressl/crypto/modes/ghash-masm-x86_64.S create mode 100644 ext/libressl/crypto/modes/ghash-mingw64-x86_64.S create mode 100644 ext/libressl/crypto/modes/modes_lcl.h create mode 100644 ext/libressl/crypto/modes/ofb128.c create mode 100644 ext/libressl/crypto/modes/xts128.c (limited to 'ext/libressl/crypto/modes') diff --git a/ext/libressl/crypto/modes/Makefile b/ext/libressl/crypto/modes/Makefile new file mode 100644 index 0000000..aeba042 --- /dev/null +++ b/ext/libressl/crypto/modes/Makefile @@ -0,0 +1,14 @@ +include ../../ssl_common.mk +CFLAGS += -D__BEGIN_HIDDEN_DECLS= -D__END_HIDDEN_DECLS= + +obj = cbc128.o ccm128.o cfb128.o ctr128.o cts128.o gcm128.o ofb128.o xts128.o + + +all: $(obj) +dep: all + +%.o: %.c + $(CC) $(CFLAGS) -c $< + +clean: + rm -f *.o *.a diff --git a/ext/libressl/crypto/modes/cbc128.c b/ext/libressl/crypto/modes/cbc128.c new file mode 100644 index 0000000..7502a48 --- /dev/null +++ b/ext/libressl/crypto/modes/cbc128.c @@ -0,0 +1,202 @@ +/* $OpenBSD: cbc128.c,v 1.4 2015/02/10 09:46:30 miod Exp $ */ +/* ==================================================================== + * Copyright (c) 2008 The OpenSSL Project. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * + * 3. All advertising materials mentioning features or use of this + * software must display the following acknowledgment: + * "This product includes software developed by the OpenSSL Project + * for use in the OpenSSL Toolkit. (http://www.openssl.org/)" + * + * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to + * endorse or promote products derived from this software without + * prior written permission. For written permission, please contact + * openssl-core@openssl.org. + * + * 5. Products derived from this software may not be called "OpenSSL" + * nor may "OpenSSL" appear in their names without prior written + * permission of the OpenSSL Project. + * + * 6. Redistributions of any form whatsoever must retain the following + * acknowledgment: + * "This product includes software developed by the OpenSSL Project + * for use in the OpenSSL Toolkit (http://www.openssl.org/)" + * + * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY + * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR + * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, + * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED + * OF THE POSSIBILITY OF SUCH DAMAGE. + * ==================================================================== + * + */ + +#include +#include "modes_lcl.h" +#include + +#ifndef MODES_DEBUG +# ifndef NDEBUG +# define NDEBUG +# endif +#endif + +#undef STRICT_ALIGNMENT +#ifdef __STRICT_ALIGNMENT +#define STRICT_ALIGNMENT 1 +#else +#define STRICT_ALIGNMENT 0 +#endif + +void CRYPTO_cbc128_encrypt(const unsigned char *in, unsigned char *out, + size_t len, const void *key, + unsigned char ivec[16], block128_f block) +{ + size_t n; + const unsigned char *iv = ivec; + +#if !defined(OPENSSL_SMALL_FOOTPRINT) + if (STRICT_ALIGNMENT && + ((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) { + while (len>=16) { + for(n=0; n<16; ++n) + out[n] = in[n] ^ iv[n]; + (*block)(out, out, key); + iv = out; + len -= 16; + in += 16; + out += 16; + } + } else { + while (len>=16) { + for(n=0; n<16; n+=sizeof(size_t)) + *(size_t*)(out+n) = + *(size_t*)(in+n) ^ *(size_t*)(iv+n); + (*block)(out, out, key); + iv = out; + len -= 16; + in += 16; + out += 16; + } + } +#endif + while (len) { + for(n=0; n<16 && n=16) { + (*block)(in, out, key); + for(n=0; n<16; ++n) + out[n] ^= iv[n]; + iv = in; + len -= 16; + in += 16; + out += 16; + } + } else if (16%sizeof(size_t) == 0) { /* always true */ + while (len>=16) { + size_t *out_t=(size_t *)out, *iv_t=(size_t *)iv; + + (*block)(in, out, key); + for(n=0; n<16/sizeof(size_t); n++) + out_t[n] ^= iv_t[n]; + iv = in; + len -= 16; + in += 16; + out += 16; + } + } + memcpy(ivec,iv,16); + } else { + if (STRICT_ALIGNMENT && + ((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) { + unsigned char c; + while (len>=16) { + (*block)(in, tmp.c, key); + for(n=0; n<16; ++n) { + c = in[n]; + out[n] = tmp.c[n] ^ ivec[n]; + ivec[n] = c; + } + len -= 16; + in += 16; + out += 16; + } + } else if (16%sizeof(size_t) == 0) { /* always true */ + while (len>=16) { + size_t c, *out_t=(size_t *)out, *ivec_t=(size_t *)ivec; + const size_t *in_t=(const size_t *)in; + + (*block)(in, tmp.c, key); + for(n=0; n<16/sizeof(size_t); n++) { + c = in_t[n]; + out_t[n] = tmp.t[n] ^ ivec_t[n]; + ivec_t[n] = c; + } + len -= 16; + in += 16; + out += 16; + } + } + } +#endif + while (len) { + unsigned char c; + (*block)(in, tmp.c, key); + for(n=0; n<16 && n +#include "modes_lcl.h" +#include + +#ifndef MODES_DEBUG +# ifndef NDEBUG +# define NDEBUG +# endif +#endif + +/* First you setup M and L parameters and pass the key schedule. + * This is called once per session setup... */ +void CRYPTO_ccm128_init(CCM128_CONTEXT *ctx, + unsigned int M,unsigned int L,void *key,block128_f block) +{ + memset(ctx->nonce.c,0,sizeof(ctx->nonce.c)); + ctx->nonce.c[0] = ((u8)(L-1)&7) | (u8)(((M-2)/2)&7)<<3; + ctx->blocks = 0; + ctx->block = block; + ctx->key = key; +} + +/* !!! Following interfaces are to be called *once* per packet !!! */ + +/* Then you setup per-message nonce and pass the length of the message */ +int CRYPTO_ccm128_setiv(CCM128_CONTEXT *ctx, + const unsigned char *nonce,size_t nlen,size_t mlen) +{ + unsigned int L = ctx->nonce.c[0]&7; /* the L parameter */ + + if (nlen<(14-L)) return -1; /* nonce is too short */ + + if (sizeof(mlen)==8 && L>=3) { + ctx->nonce.c[8] = (u8)(mlen>>(56%(sizeof(mlen)*8))); + ctx->nonce.c[9] = (u8)(mlen>>(48%(sizeof(mlen)*8))); + ctx->nonce.c[10] = (u8)(mlen>>(40%(sizeof(mlen)*8))); + ctx->nonce.c[11] = (u8)(mlen>>(32%(sizeof(mlen)*8))); + } + else + ctx->nonce.u[1] = 0; + + ctx->nonce.c[12] = (u8)(mlen>>24); + ctx->nonce.c[13] = (u8)(mlen>>16); + ctx->nonce.c[14] = (u8)(mlen>>8); + ctx->nonce.c[15] = (u8)mlen; + + ctx->nonce.c[0] &= ~0x40; /* clear Adata flag */ + memcpy(&ctx->nonce.c[1],nonce,14-L); + + return 0; +} + +/* Then you pass additional authentication data, this is optional */ +void CRYPTO_ccm128_aad(CCM128_CONTEXT *ctx, + const unsigned char *aad,size_t alen) +{ unsigned int i; + block128_f block = ctx->block; + + if (alen==0) return; + + ctx->nonce.c[0] |= 0x40; /* set Adata flag */ + (*block)(ctx->nonce.c,ctx->cmac.c,ctx->key), + ctx->blocks++; + + if (alen<(0x10000-0x100)) { + ctx->cmac.c[0] ^= (u8)(alen>>8); + ctx->cmac.c[1] ^= (u8)alen; + i=2; + } + else if (sizeof(alen)==8 && alen>=(size_t)1<<(32%(sizeof(alen)*8))) { + ctx->cmac.c[0] ^= 0xFF; + ctx->cmac.c[1] ^= 0xFF; + ctx->cmac.c[2] ^= (u8)(alen>>(56%(sizeof(alen)*8))); + ctx->cmac.c[3] ^= (u8)(alen>>(48%(sizeof(alen)*8))); + ctx->cmac.c[4] ^= (u8)(alen>>(40%(sizeof(alen)*8))); + ctx->cmac.c[5] ^= (u8)(alen>>(32%(sizeof(alen)*8))); + ctx->cmac.c[6] ^= (u8)(alen>>24); + ctx->cmac.c[7] ^= (u8)(alen>>16); + ctx->cmac.c[8] ^= (u8)(alen>>8); + ctx->cmac.c[9] ^= (u8)alen; + i=10; + } + else { + ctx->cmac.c[0] ^= 0xFF; + ctx->cmac.c[1] ^= 0xFE; + ctx->cmac.c[2] ^= (u8)(alen>>24); + ctx->cmac.c[3] ^= (u8)(alen>>16); + ctx->cmac.c[4] ^= (u8)(alen>>8); + ctx->cmac.c[5] ^= (u8)alen; + i=6; + } + + do { + for(;i<16 && alen;++i,++aad,--alen) + ctx->cmac.c[i] ^= *aad; + (*block)(ctx->cmac.c,ctx->cmac.c,ctx->key), + ctx->blocks++; + i=0; + } while (alen); +} + +/* Finally you encrypt or decrypt the message */ + +/* counter part of nonce may not be larger than L*8 bits, + * L is not larger than 8, therefore 64-bit counter... */ +static void ctr64_inc(unsigned char *counter) { + unsigned int n=8; + u8 c; + + counter += 8; + do { + --n; + c = counter[n]; + ++c; + counter[n] = c; + if (c) return; + } while (n); +} + +int CRYPTO_ccm128_encrypt(CCM128_CONTEXT *ctx, + const unsigned char *inp, unsigned char *out, + size_t len) +{ + size_t n; + unsigned int i,L; + unsigned char flags0 = ctx->nonce.c[0]; + block128_f block = ctx->block; + void * key = ctx->key; + union { u64 u[2]; u8 c[16]; } scratch; + + if (!(flags0&0x40)) + (*block)(ctx->nonce.c,ctx->cmac.c,key), + ctx->blocks++; + + ctx->nonce.c[0] = L = flags0&7; + for (n=0,i=15-L;i<15;++i) { + n |= ctx->nonce.c[i]; + ctx->nonce.c[i]=0; + n <<= 8; + } + n |= ctx->nonce.c[15]; /* reconstructed length */ + ctx->nonce.c[15]=1; + + if (n!=len) return -1; /* length mismatch */ + + ctx->blocks += ((len+15)>>3)|1; + if (ctx->blocks > (U64(1)<<61)) return -2; /* too much data */ + + while (len>=16) { +#ifdef __STRICT_ALIGNMENT + union { u64 u[2]; u8 c[16]; } temp; + + memcpy (temp.c,inp,16); + ctx->cmac.u[0] ^= temp.u[0]; + ctx->cmac.u[1] ^= temp.u[1]; +#else + ctx->cmac.u[0] ^= ((u64*)inp)[0]; + ctx->cmac.u[1] ^= ((u64*)inp)[1]; +#endif + (*block)(ctx->cmac.c,ctx->cmac.c,key); + (*block)(ctx->nonce.c,scratch.c,key); + ctr64_inc(ctx->nonce.c); +#ifdef __STRICT_ALIGNMENT + temp.u[0] ^= scratch.u[0]; + temp.u[1] ^= scratch.u[1]; + memcpy(out,temp.c,16); +#else + ((u64*)out)[0] = scratch.u[0]^((u64*)inp)[0]; + ((u64*)out)[1] = scratch.u[1]^((u64*)inp)[1]; +#endif + inp += 16; + out += 16; + len -= 16; + } + + if (len) { + for (i=0; icmac.c[i] ^= inp[i]; + (*block)(ctx->cmac.c,ctx->cmac.c,key); + (*block)(ctx->nonce.c,scratch.c,key); + for (i=0; inonce.c[i]=0; + + (*block)(ctx->nonce.c,scratch.c,key); + ctx->cmac.u[0] ^= scratch.u[0]; + ctx->cmac.u[1] ^= scratch.u[1]; + + ctx->nonce.c[0] = flags0; + + return 0; +} + +int CRYPTO_ccm128_decrypt(CCM128_CONTEXT *ctx, + const unsigned char *inp, unsigned char *out, + size_t len) +{ + size_t n; + unsigned int i,L; + unsigned char flags0 = ctx->nonce.c[0]; + block128_f block = ctx->block; + void * key = ctx->key; + union { u64 u[2]; u8 c[16]; } scratch; + + if (!(flags0&0x40)) + (*block)(ctx->nonce.c,ctx->cmac.c,key); + + ctx->nonce.c[0] = L = flags0&7; + for (n=0,i=15-L;i<15;++i) { + n |= ctx->nonce.c[i]; + ctx->nonce.c[i]=0; + n <<= 8; + } + n |= ctx->nonce.c[15]; /* reconstructed length */ + ctx->nonce.c[15]=1; + + if (n!=len) return -1; + + while (len>=16) { +#ifdef __STRICT_ALIGNMENT + union { u64 u[2]; u8 c[16]; } temp; +#endif + (*block)(ctx->nonce.c,scratch.c,key); + ctr64_inc(ctx->nonce.c); +#ifdef __STRICT_ALIGNMENT + memcpy (temp.c,inp,16); + ctx->cmac.u[0] ^= (scratch.u[0] ^= temp.u[0]); + ctx->cmac.u[1] ^= (scratch.u[1] ^= temp.u[1]); + memcpy (out,scratch.c,16); +#else + ctx->cmac.u[0] ^= (((u64*)out)[0] = scratch.u[0]^((u64*)inp)[0]); + ctx->cmac.u[1] ^= (((u64*)out)[1] = scratch.u[1]^((u64*)inp)[1]); +#endif + (*block)(ctx->cmac.c,ctx->cmac.c,key); + + inp += 16; + out += 16; + len -= 16; + } + + if (len) { + (*block)(ctx->nonce.c,scratch.c,key); + for (i=0; icmac.c[i] ^= (out[i] = scratch.c[i]^inp[i]); + (*block)(ctx->cmac.c,ctx->cmac.c,key); + } + + for (i=15-L;i<16;++i) + ctx->nonce.c[i]=0; + + (*block)(ctx->nonce.c,scratch.c,key); + ctx->cmac.u[0] ^= scratch.u[0]; + ctx->cmac.u[1] ^= scratch.u[1]; + + ctx->nonce.c[0] = flags0; + + return 0; +} + +static void ctr64_add (unsigned char *counter,size_t inc) +{ size_t n=8, val=0; + + counter += 8; + do { + --n; + val += counter[n] + (inc&0xff); + counter[n] = (unsigned char)val; + val >>= 8; /* carry bit */ + inc >>= 8; + } while(n && (inc || val)); +} + +int CRYPTO_ccm128_encrypt_ccm64(CCM128_CONTEXT *ctx, + const unsigned char *inp, unsigned char *out, + size_t len,ccm128_f stream) +{ + size_t n; + unsigned int i,L; + unsigned char flags0 = ctx->nonce.c[0]; + block128_f block = ctx->block; + void * key = ctx->key; + union { u64 u[2]; u8 c[16]; } scratch; + + if (!(flags0&0x40)) + (*block)(ctx->nonce.c,ctx->cmac.c,key), + ctx->blocks++; + + ctx->nonce.c[0] = L = flags0&7; + for (n=0,i=15-L;i<15;++i) { + n |= ctx->nonce.c[i]; + ctx->nonce.c[i]=0; + n <<= 8; + } + n |= ctx->nonce.c[15]; /* reconstructed length */ + ctx->nonce.c[15]=1; + + if (n!=len) return -1; /* length mismatch */ + + ctx->blocks += ((len+15)>>3)|1; + if (ctx->blocks > (U64(1)<<61)) return -2; /* too much data */ + + if ((n=len/16)) { + (*stream)(inp,out,n,key,ctx->nonce.c,ctx->cmac.c); + n *= 16; + inp += n; + out += n; + len -= n; + if (len) ctr64_add(ctx->nonce.c,n/16); + } + + if (len) { + for (i=0; icmac.c[i] ^= inp[i]; + (*block)(ctx->cmac.c,ctx->cmac.c,key); + (*block)(ctx->nonce.c,scratch.c,key); + for (i=0; inonce.c[i]=0; + + (*block)(ctx->nonce.c,scratch.c,key); + ctx->cmac.u[0] ^= scratch.u[0]; + ctx->cmac.u[1] ^= scratch.u[1]; + + ctx->nonce.c[0] = flags0; + + return 0; +} + +int CRYPTO_ccm128_decrypt_ccm64(CCM128_CONTEXT *ctx, + const unsigned char *inp, unsigned char *out, + size_t len,ccm128_f stream) +{ + size_t n; + unsigned int i,L; + unsigned char flags0 = ctx->nonce.c[0]; + block128_f block = ctx->block; + void * key = ctx->key; + union { u64 u[2]; u8 c[16]; } scratch; + + if (!(flags0&0x40)) + (*block)(ctx->nonce.c,ctx->cmac.c,key); + + ctx->nonce.c[0] = L = flags0&7; + for (n=0,i=15-L;i<15;++i) { + n |= ctx->nonce.c[i]; + ctx->nonce.c[i]=0; + n <<= 8; + } + n |= ctx->nonce.c[15]; /* reconstructed length */ + ctx->nonce.c[15]=1; + + if (n!=len) return -1; + + if ((n=len/16)) { + (*stream)(inp,out,n,key,ctx->nonce.c,ctx->cmac.c); + n *= 16; + inp += n; + out += n; + len -= n; + if (len) ctr64_add(ctx->nonce.c,n/16); + } + + if (len) { + (*block)(ctx->nonce.c,scratch.c,key); + for (i=0; icmac.c[i] ^= (out[i] = scratch.c[i]^inp[i]); + (*block)(ctx->cmac.c,ctx->cmac.c,key); + } + + for (i=15-L;i<16;++i) + ctx->nonce.c[i]=0; + + (*block)(ctx->nonce.c,scratch.c,key); + ctx->cmac.u[0] ^= scratch.u[0]; + ctx->cmac.u[1] ^= scratch.u[1]; + + ctx->nonce.c[0] = flags0; + + return 0; +} + +size_t CRYPTO_ccm128_tag(CCM128_CONTEXT *ctx,unsigned char *tag,size_t len) +{ unsigned int M = (ctx->nonce.c[0]>>3)&7; /* the M parameter */ + + M *= 2; M += 2; + if (len != M) return 0; + memcpy(tag,ctx->cmac.c,M); + return M; +} diff --git a/ext/libressl/crypto/modes/cfb128.c b/ext/libressl/crypto/modes/cfb128.c new file mode 100644 index 0000000..88bfbc4 --- /dev/null +++ b/ext/libressl/crypto/modes/cfb128.c @@ -0,0 +1,234 @@ +/* $OpenBSD: cfb128.c,v 1.4 2015/02/10 09:46:30 miod Exp $ */ +/* ==================================================================== + * Copyright (c) 2008 The OpenSSL Project. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * + * 3. All advertising materials mentioning features or use of this + * software must display the following acknowledgment: + * "This product includes software developed by the OpenSSL Project + * for use in the OpenSSL Toolkit. (http://www.openssl.org/)" + * + * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to + * endorse or promote products derived from this software without + * prior written permission. For written permission, please contact + * openssl-core@openssl.org. + * + * 5. Products derived from this software may not be called "OpenSSL" + * nor may "OpenSSL" appear in their names without prior written + * permission of the OpenSSL Project. + * + * 6. Redistributions of any form whatsoever must retain the following + * acknowledgment: + * "This product includes software developed by the OpenSSL Project + * for use in the OpenSSL Toolkit (http://www.openssl.org/)" + * + * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY + * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR + * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, + * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED + * OF THE POSSIBILITY OF SUCH DAMAGE. + * ==================================================================== + * + */ + +#include +#include "modes_lcl.h" +#include + +#ifndef MODES_DEBUG +# ifndef NDEBUG +# define NDEBUG +# endif +#endif + +/* The input and output encrypted as though 128bit cfb mode is being + * used. The extra state information to record how much of the + * 128bit block we have used is contained in *num; + */ +void CRYPTO_cfb128_encrypt(const unsigned char *in, unsigned char *out, + size_t len, const void *key, + unsigned char ivec[16], int *num, + int enc, block128_f block) +{ + unsigned int n; + size_t l = 0; + + n = *num; + + if (enc) { +#if !defined(OPENSSL_SMALL_FOOTPRINT) + if (16%sizeof(size_t) == 0) do { /* always true actually */ + while (n && len) { + *(out++) = ivec[n] ^= *(in++); + --len; + n = (n+1) % 16; + } +#ifdef __STRICT_ALIGNMENT + if (((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) + break; +#endif + while (len>=16) { + (*block)(ivec, ivec, key); + for (; n<16; n+=sizeof(size_t)) { + *(size_t*)(out+n) = + *(size_t*)(ivec+n) ^= *(size_t*)(in+n); + } + len -= 16; + out += 16; + in += 16; + n = 0; + } + if (len) { + (*block)(ivec, ivec, key); + while (len--) { + out[n] = ivec[n] ^= in[n]; + ++n; + } + } + *num = n; + return; + } while (0); + /* the rest would be commonly eliminated by x86* compiler */ +#endif + while (l=16) { + (*block)(ivec, ivec, key); + for (; n<16; n+=sizeof(size_t)) { + size_t t = *(size_t*)(in+n); + *(size_t*)(out+n) = *(size_t*)(ivec+n) ^ t; + *(size_t*)(ivec+n) = t; + } + len -= 16; + out += 16; + in += 16; + n = 0; + } + if (len) { + (*block)(ivec, ivec, key); + while (len--) { + unsigned char c; + out[n] = ivec[n] ^ (c = in[n]); ivec[n] = c; + ++n; + } + } + *num = n; + return; + } while (0); + /* the rest would be commonly eliminated by x86* compiler */ +#endif + while (l128) return; + + /* fill in the first half of the new IV with the current IV */ + memcpy(ovec,ivec,16); + /* construct the new IV */ + (*block)(ivec,ivec,key); + num = (nbits+7)/8; + if (enc) /* encrypt the input */ + for(n=0 ; n < num ; ++n) + out[n] = (ovec[16+n] = in[n] ^ ivec[n]); + else /* decrypt the input */ + for(n=0 ; n < num ; ++n) + out[n] = (ovec[16+n] = in[n]) ^ ivec[n]; + /* shift ovec left... */ + rem = nbits%8; + num = nbits/8; + if(rem==0) + memcpy(ivec,ovec+num,16); + else + for(n=0 ; n < 16 ; ++n) + ivec[n] = ovec[n+num]<>(8-rem); + + /* it is not necessary to cleanse ovec, since the IV is not secret */ +} + +/* N.B. This expects the input to be packed, MS bit first */ +void CRYPTO_cfb128_1_encrypt(const unsigned char *in, unsigned char *out, + size_t bits, const void *key, + unsigned char ivec[16], int *num, + int enc, block128_f block) +{ + size_t n; + unsigned char c[1],d[1]; + + for(n=0 ; n> (unsigned int)(n%8)); + } +} + +void CRYPTO_cfb128_8_encrypt(const unsigned char *in, unsigned char *out, + size_t length, const void *key, + unsigned char ivec[16], int *num, + int enc, block128_f block) +{ + size_t n; + + for(n=0 ; n +#include "modes_lcl.h" +#include + +#ifndef MODES_DEBUG +# ifndef NDEBUG +# define NDEBUG +# endif +#endif +#include + +/* NOTE: the IV/counter CTR mode is big-endian. The code itself + * is endian-neutral. */ + +/* increment counter (128-bit int) by 1 */ +static void ctr128_inc(unsigned char *counter) { + u32 n=16; + u8 c; + + do { + --n; + c = counter[n]; + ++c; + counter[n] = c; + if (c) return; + } while (n); +} + +#if !defined(OPENSSL_SMALL_FOOTPRINT) +static void +ctr128_inc_aligned(unsigned char *counter) +{ +#if BYTE_ORDER == LITTLE_ENDIAN + ctr128_inc(counter); +#else + size_t *data, c, n; + data = (size_t *)counter; + n = 16 / sizeof(size_t); + do { + --n; + c = data[n]; + ++c; + data[n] = c; + if (c) + return; + } while (n); +#endif +} +#endif + +/* 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 contained in *num, and the + * encrypted counter is kept in ecount_buf. Both *num and + * ecount_buf must be initialised with zeros before the first + * call to CRYPTO_ctr128_encrypt(). + * + * This algorithm assumes that the counter is in the x lower bits + * of the IV (ivec), and that the application has full control over + * overflow and the rest of the IV. This implementation takes NO + * responsability for checking that the counter doesn't overflow + * into the rest of the IV when incremented. + */ +void CRYPTO_ctr128_encrypt(const unsigned char *in, unsigned char *out, + size_t len, const void *key, + unsigned char ivec[16], unsigned char ecount_buf[16], + unsigned int *num, block128_f block) +{ + unsigned int n; + size_t l=0; + + assert(*num < 16); + + n = *num; + +#if !defined(OPENSSL_SMALL_FOOTPRINT) + if (16%sizeof(size_t) == 0) do { /* always true actually */ + while (n && len) { + *(out++) = *(in++) ^ ecount_buf[n]; + --len; + n = (n+1) % 16; + } + +#ifdef __STRICT_ALIGNMENT + if (((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) + break; +#endif + while (len>=16) { + (*block)(ivec, ecount_buf, key); + ctr128_inc_aligned(ivec); + for (; n<16; n+=sizeof(size_t)) + *(size_t *)(out+n) = + *(size_t *)(in+n) ^ *(size_t *)(ecount_buf+n); + len -= 16; + out += 16; + in += 16; + n = 0; + } + if (len) { + (*block)(ivec, ecount_buf, key); + ctr128_inc_aligned(ivec); + while (len--) { + out[n] = in[n] ^ ecount_buf[n]; + ++n; + } + } + *num = n; + return; + } while(0); + /* the rest would be commonly eliminated by x86* compiler */ +#endif + while (l=16) { + size_t blocks = len/16; + /* + * 1<<28 is just a not-so-small yet not-so-large number... + * Below condition is practically never met, but it has to + * be checked for code correctness. + */ + if (sizeof(size_t)>sizeof(unsigned int) && blocks>(1U<<28)) + blocks = (1U<<28); + /* + * As (*func) operates on 32-bit counter, caller + * has to handle overflow. 'if' below detects the + * overflow, which is then handled by limiting the + * amount of blocks to the exact overflow point... + */ + ctr32 += (u32)blocks; + if (ctr32 < blocks) { + blocks -= ctr32; + ctr32 = 0; + } + (*func)(in,out,blocks,key,ivec); + /* (*ctr) does not update ivec, caller does: */ + PUTU32(ivec+12,ctr32); + /* ... overflow was detected, propogate carry. */ + if (ctr32 == 0) ctr96_inc(ivec); + blocks *= 16; + len -= blocks; + out += blocks; + in += blocks; + } + if (len) { + memset(ecount_buf,0,16); + (*func)(ecount_buf,ecount_buf,1,key,ivec); + ++ctr32; + PUTU32(ivec+12,ctr32); + if (ctr32 == 0) ctr96_inc(ivec); + while (len--) { + out[n] = in[n] ^ ecount_buf[n]; + ++n; + } + } + + *num=n; +} diff --git a/ext/libressl/crypto/modes/cts128.c b/ext/libressl/crypto/modes/cts128.c new file mode 100644 index 0000000..b2f7174 --- /dev/null +++ b/ext/libressl/crypto/modes/cts128.c @@ -0,0 +1,267 @@ +/* $OpenBSD: cts128.c,v 1.5 2015/07/19 18:27:26 miod Exp $ */ +/* ==================================================================== + * Copyright (c) 2008 The OpenSSL Project. All rights reserved. + * + * Rights for redistribution and usage in source and binary + * forms are granted according to the OpenSSL license. + */ + +#include +#include "modes_lcl.h" +#include + +#ifndef MODES_DEBUG +# ifndef NDEBUG +# define NDEBUG +# endif +#endif + +/* + * Trouble with Ciphertext Stealing, CTS, mode is that there is no + * common official specification, but couple of cipher/application + * specific ones: RFC2040 and RFC3962. Then there is 'Proposal to + * Extend CBC Mode By "Ciphertext Stealing"' at NIST site, which + * deviates from mentioned RFCs. Most notably it allows input to be + * of block length and it doesn't flip the order of the last two + * blocks. CTS is being discussed even in ECB context, but it's not + * adopted for any known application. This implementation provides + * two interfaces: one compliant with above mentioned RFCs and one + * compliant with the NIST proposal, both extending CBC mode. + */ + +size_t CRYPTO_cts128_encrypt_block(const unsigned char *in, unsigned char *out, + size_t len, const void *key, + unsigned char ivec[16], block128_f block) +{ size_t residue, n; + + if (len <= 16) return 0; + + if ((residue=len%16) == 0) residue = 16; + + len -= residue; + + CRYPTO_cbc128_encrypt(in,out,len,key,ivec,block); + + in += len; + out += len; + + for (n=0; n +#include "modes_lcl.h" +#include + +#ifndef MODES_DEBUG +# ifndef NDEBUG +# define NDEBUG +# endif +#endif + +#if defined(BSWAP4) && defined(__STRICT_ALIGNMENT) +/* redefine, because alignment is ensured */ +#undef GETU32 +#define GETU32(p) BSWAP4(*(const u32 *)(p)) +#undef PUTU32 +#define PUTU32(p,v) *(u32 *)(p) = BSWAP4(v) +#endif + +#define PACK(s) ((size_t)(s)<<(sizeof(size_t)*8-16)) +#define REDUCE1BIT(V) \ + do { \ + if (sizeof(size_t)==8) { \ + u64 T = U64(0xe100000000000000) & (0-(V.lo&1)); \ + V.lo = (V.hi<<63)|(V.lo>>1); \ + V.hi = (V.hi>>1 )^T; \ + } else { \ + u32 T = 0xe1000000U & (0-(u32)(V.lo&1)); \ + V.lo = (V.hi<<63)|(V.lo>>1); \ + V.hi = (V.hi>>1 )^((u64)T<<32); \ + } \ + } while(0) + +/* + * Even though permitted values for TABLE_BITS are 8, 4 and 1, it should + * never be set to 8. 8 is effectively reserved for testing purposes. + * TABLE_BITS>1 are lookup-table-driven implementations referred to as + * "Shoup's" in GCM specification. In other words OpenSSL does not cover + * whole spectrum of possible table driven implementations. Why? In + * non-"Shoup's" case memory access pattern is segmented in such manner, + * that it's trivial to see that cache timing information can reveal + * fair portion of intermediate hash value. Given that ciphertext is + * always available to attacker, it's possible for him to attempt to + * deduce secret parameter H and if successful, tamper with messages + * [which is nothing but trivial in CTR mode]. In "Shoup's" case it's + * not as trivial, but there is no reason to believe that it's resistant + * to cache-timing attack. And the thing about "8-bit" implementation is + * that it consumes 16 (sixteen) times more memory, 4KB per individual + * key + 1KB shared. Well, on pros side it should be twice as fast as + * "4-bit" version. And for gcc-generated x86[_64] code, "8-bit" version + * was observed to run ~75% faster, closer to 100% for commercial + * compilers... Yet "4-bit" procedure is preferred, because it's + * believed to provide better security-performance balance and adequate + * all-round performance. "All-round" refers to things like: + * + * - shorter setup time effectively improves overall timing for + * handling short messages; + * - larger table allocation can become unbearable because of VM + * subsystem penalties (for example on Windows large enough free + * results in VM working set trimming, meaning that consequent + * malloc would immediately incur working set expansion); + * - larger table has larger cache footprint, which can affect + * performance of other code paths (not necessarily even from same + * thread in Hyper-Threading world); + * + * Value of 1 is not appropriate for performance reasons. + */ +#if TABLE_BITS==8 + +static void gcm_init_8bit(u128 Htable[256], u64 H[2]) +{ + int i, j; + u128 V; + + Htable[0].hi = 0; + Htable[0].lo = 0; + V.hi = H[0]; + V.lo = H[1]; + + for (Htable[128]=V, i=64; i>0; i>>=1) { + REDUCE1BIT(V); + Htable[i] = V; + } + + for (i=2; i<256; i<<=1) { + u128 *Hi = Htable+i, H0 = *Hi; + for (j=1; j>8); + Z.hi = (Z.hi>>8); +#if SIZE_MAX == 0xffffffffffffffff + Z.hi ^= rem_8bit[rem]; +#else + Z.hi ^= (u64)rem_8bit[rem]<<32; +#endif + } + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP8 + Xi[0] = BSWAP8(Z.hi); + Xi[1] = BSWAP8(Z.lo); +#else + u8 *p = (u8 *)Xi; + u32 v; + v = (u32)(Z.hi>>32); PUTU32(p,v); + v = (u32)(Z.hi); PUTU32(p+4,v); + v = (u32)(Z.lo>>32); PUTU32(p+8,v); + v = (u32)(Z.lo); PUTU32(p+12,v); +#endif +#else /* BIG_ENDIAN */ + Xi[0] = Z.hi; + Xi[1] = Z.lo; +#endif +} +#define GCM_MUL(ctx,Xi) gcm_gmult_8bit(ctx->Xi.u,ctx->Htable) + +#elif TABLE_BITS==4 + +static void gcm_init_4bit(u128 Htable[16], u64 H[2]) +{ + u128 V; +#if defined(OPENSSL_SMALL_FOOTPRINT) + int i; +#endif + + Htable[0].hi = 0; + Htable[0].lo = 0; + V.hi = H[0]; + V.lo = H[1]; + +#if defined(OPENSSL_SMALL_FOOTPRINT) + for (Htable[8]=V, i=4; i>0; i>>=1) { + REDUCE1BIT(V); + Htable[i] = V; + } + + for (i=2; i<16; i<<=1) { + u128 *Hi = Htable+i; + int j; + for (V=*Hi, j=1; j>32; + Htable[j].lo = V.hi<<32|V.hi>>32; + } +#endif + } +#endif +} + +#ifndef GHASH_ASM +static const size_t rem_4bit[16] = { + PACK(0x0000), PACK(0x1C20), PACK(0x3840), PACK(0x2460), + PACK(0x7080), PACK(0x6CA0), PACK(0x48C0), PACK(0x54E0), + PACK(0xE100), PACK(0xFD20), PACK(0xD940), PACK(0xC560), + PACK(0x9180), PACK(0x8DA0), PACK(0xA9C0), PACK(0xB5E0) }; + +static void gcm_gmult_4bit(u64 Xi[2], const u128 Htable[16]) +{ + u128 Z; + int cnt = 15; + size_t rem, nlo, nhi; + + nlo = ((const u8 *)Xi)[15]; + nhi = nlo>>4; + nlo &= 0xf; + + Z.hi = Htable[nlo].hi; + Z.lo = Htable[nlo].lo; + + while (1) { + rem = (size_t)Z.lo&0xf; + Z.lo = (Z.hi<<60)|(Z.lo>>4); + Z.hi = (Z.hi>>4); +#if SIZE_MAX == 0xffffffffffffffff + Z.hi ^= rem_4bit[rem]; +#else + Z.hi ^= (u64)rem_4bit[rem]<<32; +#endif + Z.hi ^= Htable[nhi].hi; + Z.lo ^= Htable[nhi].lo; + + if (--cnt<0) break; + + nlo = ((const u8 *)Xi)[cnt]; + nhi = nlo>>4; + nlo &= 0xf; + + rem = (size_t)Z.lo&0xf; + Z.lo = (Z.hi<<60)|(Z.lo>>4); + Z.hi = (Z.hi>>4); +#if SIZE_MAX == 0xffffffffffffffff + Z.hi ^= rem_4bit[rem]; +#else + Z.hi ^= (u64)rem_4bit[rem]<<32; +#endif + Z.hi ^= Htable[nlo].hi; + Z.lo ^= Htable[nlo].lo; + } + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP8 + Xi[0] = BSWAP8(Z.hi); + Xi[1] = BSWAP8(Z.lo); +#else + u8 *p = (u8 *)Xi; + u32 v; + v = (u32)(Z.hi>>32); PUTU32(p,v); + v = (u32)(Z.hi); PUTU32(p+4,v); + v = (u32)(Z.lo>>32); PUTU32(p+8,v); + v = (u32)(Z.lo); PUTU32(p+12,v); +#endif +#else /* BIG_ENDIAN */ + Xi[0] = Z.hi; + Xi[1] = Z.lo; +#endif +} + +#if !defined(OPENSSL_SMALL_FOOTPRINT) +/* + * Streamed gcm_mult_4bit, see CRYPTO_gcm128_[en|de]crypt for + * details... Compiler-generated code doesn't seem to give any + * performance improvement, at least not on x86[_64]. It's here + * mostly as reference and a placeholder for possible future + * non-trivial optimization[s]... + */ +static void gcm_ghash_4bit(u64 Xi[2],const u128 Htable[16], + const u8 *inp,size_t len) +{ + u128 Z; + int cnt; + size_t rem, nlo, nhi; + +#if 1 + do { + cnt = 15; + nlo = ((const u8 *)Xi)[15]; + nlo ^= inp[15]; + nhi = nlo>>4; + nlo &= 0xf; + + Z.hi = Htable[nlo].hi; + Z.lo = Htable[nlo].lo; + + while (1) { + rem = (size_t)Z.lo&0xf; + Z.lo = (Z.hi<<60)|(Z.lo>>4); + Z.hi = (Z.hi>>4); +#if SIZE_MAX == 0xffffffffffffffff + Z.hi ^= rem_4bit[rem]; +#else + Z.hi ^= (u64)rem_4bit[rem]<<32; +#endif + Z.hi ^= Htable[nhi].hi; + Z.lo ^= Htable[nhi].lo; + + if (--cnt<0) break; + + nlo = ((const u8 *)Xi)[cnt]; + nlo ^= inp[cnt]; + nhi = nlo>>4; + nlo &= 0xf; + + rem = (size_t)Z.lo&0xf; + Z.lo = (Z.hi<<60)|(Z.lo>>4); + Z.hi = (Z.hi>>4); +#if SIZE_MAX == 0xffffffffffffffff + Z.hi ^= rem_4bit[rem]; +#else + Z.hi ^= (u64)rem_4bit[rem]<<32; +#endif + Z.hi ^= Htable[nlo].hi; + Z.lo ^= Htable[nlo].lo; + } +#else + /* + * Extra 256+16 bytes per-key plus 512 bytes shared tables + * [should] give ~50% improvement... One could have PACK()-ed + * the rem_8bit even here, but the priority is to minimize + * cache footprint... + */ + u128 Hshr4[16]; /* Htable shifted right by 4 bits */ + u8 Hshl4[16]; /* Htable shifted left by 4 bits */ + static const unsigned short rem_8bit[256] = { + 0x0000, 0x01C2, 0x0384, 0x0246, 0x0708, 0x06CA, 0x048C, 0x054E, + 0x0E10, 0x0FD2, 0x0D94, 0x0C56, 0x0918, 0x08DA, 0x0A9C, 0x0B5E, + 0x1C20, 0x1DE2, 0x1FA4, 0x1E66, 0x1B28, 0x1AEA, 0x18AC, 0x196E, + 0x1230, 0x13F2, 0x11B4, 0x1076, 0x1538, 0x14FA, 0x16BC, 0x177E, + 0x3840, 0x3982, 0x3BC4, 0x3A06, 0x3F48, 0x3E8A, 0x3CCC, 0x3D0E, + 0x3650, 0x3792, 0x35D4, 0x3416, 0x3158, 0x309A, 0x32DC, 0x331E, + 0x2460, 0x25A2, 0x27E4, 0x2626, 0x2368, 0x22AA, 0x20EC, 0x212E, + 0x2A70, 0x2BB2, 0x29F4, 0x2836, 0x2D78, 0x2CBA, 0x2EFC, 0x2F3E, + 0x7080, 0x7142, 0x7304, 0x72C6, 0x7788, 0x764A, 0x740C, 0x75CE, + 0x7E90, 0x7F52, 0x7D14, 0x7CD6, 0x7998, 0x785A, 0x7A1C, 0x7BDE, + 0x6CA0, 0x6D62, 0x6F24, 0x6EE6, 0x6BA8, 0x6A6A, 0x682C, 0x69EE, + 0x62B0, 0x6372, 0x6134, 0x60F6, 0x65B8, 0x647A, 0x663C, 0x67FE, + 0x48C0, 0x4902, 0x4B44, 0x4A86, 0x4FC8, 0x4E0A, 0x4C4C, 0x4D8E, + 0x46D0, 0x4712, 0x4554, 0x4496, 0x41D8, 0x401A, 0x425C, 0x439E, + 0x54E0, 0x5522, 0x5764, 0x56A6, 0x53E8, 0x522A, 0x506C, 0x51AE, + 0x5AF0, 0x5B32, 0x5974, 0x58B6, 0x5DF8, 0x5C3A, 0x5E7C, 0x5FBE, + 0xE100, 0xE0C2, 0xE284, 0xE346, 0xE608, 0xE7CA, 0xE58C, 0xE44E, + 0xEF10, 0xEED2, 0xEC94, 0xED56, 0xE818, 0xE9DA, 0xEB9C, 0xEA5E, + 0xFD20, 0xFCE2, 0xFEA4, 0xFF66, 0xFA28, 0xFBEA, 0xF9AC, 0xF86E, + 0xF330, 0xF2F2, 0xF0B4, 0xF176, 0xF438, 0xF5FA, 0xF7BC, 0xF67E, + 0xD940, 0xD882, 0xDAC4, 0xDB06, 0xDE48, 0xDF8A, 0xDDCC, 0xDC0E, + 0xD750, 0xD692, 0xD4D4, 0xD516, 0xD058, 0xD19A, 0xD3DC, 0xD21E, + 0xC560, 0xC4A2, 0xC6E4, 0xC726, 0xC268, 0xC3AA, 0xC1EC, 0xC02E, + 0xCB70, 0xCAB2, 0xC8F4, 0xC936, 0xCC78, 0xCDBA, 0xCFFC, 0xCE3E, + 0x9180, 0x9042, 0x9204, 0x93C6, 0x9688, 0x974A, 0x950C, 0x94CE, + 0x9F90, 0x9E52, 0x9C14, 0x9DD6, 0x9898, 0x995A, 0x9B1C, 0x9ADE, + 0x8DA0, 0x8C62, 0x8E24, 0x8FE6, 0x8AA8, 0x8B6A, 0x892C, 0x88EE, + 0x83B0, 0x8272, 0x8034, 0x81F6, 0x84B8, 0x857A, 0x873C, 0x86FE, + 0xA9C0, 0xA802, 0xAA44, 0xAB86, 0xAEC8, 0xAF0A, 0xAD4C, 0xAC8E, + 0xA7D0, 0xA612, 0xA454, 0xA596, 0xA0D8, 0xA11A, 0xA35C, 0xA29E, + 0xB5E0, 0xB422, 0xB664, 0xB7A6, 0xB2E8, 0xB32A, 0xB16C, 0xB0AE, + 0xBBF0, 0xBA32, 0xB874, 0xB9B6, 0xBCF8, 0xBD3A, 0xBF7C, 0xBEBE }; + /* + * This pre-processing phase slows down procedure by approximately + * same time as it makes each loop spin faster. In other words + * single block performance is approximately same as straightforward + * "4-bit" implementation, and then it goes only faster... + */ + for (cnt=0; cnt<16; ++cnt) { + Z.hi = Htable[cnt].hi; + Z.lo = Htable[cnt].lo; + Hshr4[cnt].lo = (Z.hi<<60)|(Z.lo>>4); + Hshr4[cnt].hi = (Z.hi>>4); + Hshl4[cnt] = (u8)(Z.lo<<4); + } + + do { + for (Z.lo=0, Z.hi=0, cnt=15; cnt; --cnt) { + nlo = ((const u8 *)Xi)[cnt]; + nlo ^= inp[cnt]; + nhi = nlo>>4; + nlo &= 0xf; + + Z.hi ^= Htable[nlo].hi; + Z.lo ^= Htable[nlo].lo; + + rem = (size_t)Z.lo&0xff; + + Z.lo = (Z.hi<<56)|(Z.lo>>8); + Z.hi = (Z.hi>>8); + + Z.hi ^= Hshr4[nhi].hi; + Z.lo ^= Hshr4[nhi].lo; + Z.hi ^= (u64)rem_8bit[rem^Hshl4[nhi]]<<48; + } + + nlo = ((const u8 *)Xi)[0]; + nlo ^= inp[0]; + nhi = nlo>>4; + nlo &= 0xf; + + Z.hi ^= Htable[nlo].hi; + Z.lo ^= Htable[nlo].lo; + + rem = (size_t)Z.lo&0xf; + + Z.lo = (Z.hi<<60)|(Z.lo>>4); + Z.hi = (Z.hi>>4); + + Z.hi ^= Htable[nhi].hi; + Z.lo ^= Htable[nhi].lo; + Z.hi ^= ((u64)rem_8bit[rem<<4])<<48; +#endif + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP8 + Xi[0] = BSWAP8(Z.hi); + Xi[1] = BSWAP8(Z.lo); +#else + u8 *p = (u8 *)Xi; + u32 v; + v = (u32)(Z.hi>>32); PUTU32(p,v); + v = (u32)(Z.hi); PUTU32(p+4,v); + v = (u32)(Z.lo>>32); PUTU32(p+8,v); + v = (u32)(Z.lo); PUTU32(p+12,v); +#endif +#else /* BIG_ENDIAN */ + Xi[0] = Z.hi; + Xi[1] = Z.lo; +#endif + } while (inp+=16, len-=16); +} +#endif +#else +void gcm_gmult_4bit(u64 Xi[2],const u128 Htable[16]); +void gcm_ghash_4bit(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); +#endif + +#define GCM_MUL(ctx,Xi) gcm_gmult_4bit(ctx->Xi.u,ctx->Htable) +#if defined(GHASH_ASM) || !defined(OPENSSL_SMALL_FOOTPRINT) +#define GHASH(ctx,in,len) gcm_ghash_4bit((ctx)->Xi.u,(ctx)->Htable,in,len) +/* GHASH_CHUNK is "stride parameter" missioned to mitigate cache + * trashing effect. In other words idea is to hash data while it's + * still in L1 cache after encryption pass... */ +#define GHASH_CHUNK (3*1024) +#endif + +#else /* TABLE_BITS */ + +static void gcm_gmult_1bit(u64 Xi[2],const u64 H[2]) +{ + u128 V,Z = { 0,0 }; + long X; + int i,j; + const long *xi = (const long *)Xi; + + V.hi = H[0]; /* H is in host byte order, no byte swapping */ + V.lo = H[1]; + + for (j=0; j<16/sizeof(long); ++j) { +#if BYTE_ORDER == LITTLE_ENDIAN +#if SIZE_MAX == 0xffffffffffffffff +#ifdef BSWAP8 + X = (long)(BSWAP8(xi[j])); +#else + const u8 *p = (const u8 *)(xi+j); + X = (long)((u64)GETU32(p)<<32|GETU32(p+4)); +#endif +#else + const u8 *p = (const u8 *)(xi+j); + X = (long)GETU32(p); +#endif +#else /* BIG_ENDIAN */ + X = xi[j]; +#endif + + for (i=0; i<8*sizeof(long); ++i, X<<=1) { + u64 M = (u64)(X>>(8*sizeof(long)-1)); + Z.hi ^= V.hi&M; + Z.lo ^= V.lo&M; + + REDUCE1BIT(V); + } + } + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP8 + Xi[0] = BSWAP8(Z.hi); + Xi[1] = BSWAP8(Z.lo); +#else + u8 *p = (u8 *)Xi; + u32 v; + v = (u32)(Z.hi>>32); PUTU32(p,v); + v = (u32)(Z.hi); PUTU32(p+4,v); + v = (u32)(Z.lo>>32); PUTU32(p+8,v); + v = (u32)(Z.lo); PUTU32(p+12,v); +#endif +#else /* BIG_ENDIAN */ + Xi[0] = Z.hi; + Xi[1] = Z.lo; +#endif +} +#define GCM_MUL(ctx,Xi) gcm_gmult_1bit(ctx->Xi.u,ctx->H.u) + +#endif + +#if defined(GHASH_ASM) && \ + (defined(__i386) || defined(__i386__) || \ + defined(__x86_64) || defined(__x86_64__) || \ + defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64)) +#include "x86_arch.h" +#endif + +#if TABLE_BITS==4 && defined(GHASH_ASM) +# if (defined(__i386) || defined(__i386__) || \ + defined(__x86_64) || defined(__x86_64__) || \ + defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64)) +# define GHASH_ASM_X86_OR_64 +# define GCM_FUNCREF_4BIT + +void gcm_init_clmul(u128 Htable[16],const u64 Xi[2]); +void gcm_gmult_clmul(u64 Xi[2],const u128 Htable[16]); +void gcm_ghash_clmul(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); + +# if defined(__i386) || defined(__i386__) || defined(_M_IX86) +# define GHASH_ASM_X86 +void gcm_gmult_4bit_mmx(u64 Xi[2],const u128 Htable[16]); +void gcm_ghash_4bit_mmx(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); + +void gcm_gmult_4bit_x86(u64 Xi[2],const u128 Htable[16]); +void gcm_ghash_4bit_x86(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); +# endif +# elif defined(__arm__) || defined(__arm) +# include "arm_arch.h" +# if __ARM_ARCH__>=7 && !defined(__STRICT_ALIGNMENT) +# define GHASH_ASM_ARM +# define GCM_FUNCREF_4BIT +void gcm_gmult_neon(u64 Xi[2],const u128 Htable[16]); +void gcm_ghash_neon(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); +# endif +# endif +#endif + +#ifdef GCM_FUNCREF_4BIT +# undef GCM_MUL +# define GCM_MUL(ctx,Xi) (*gcm_gmult_p)(ctx->Xi.u,ctx->Htable) +# ifdef GHASH +# undef GHASH +# define GHASH(ctx,in,len) (*gcm_ghash_p)(ctx->Xi.u,ctx->Htable,in,len) +# endif +#endif + +void CRYPTO_gcm128_init(GCM128_CONTEXT *ctx,void *key,block128_f block) +{ + memset(ctx,0,sizeof(*ctx)); + ctx->block = block; + ctx->key = key; + + (*block)(ctx->H.c,ctx->H.c,key); + +#if BYTE_ORDER == LITTLE_ENDIAN + /* H is stored in host byte order */ +#ifdef BSWAP8 + ctx->H.u[0] = BSWAP8(ctx->H.u[0]); + ctx->H.u[1] = BSWAP8(ctx->H.u[1]); +#else + u8 *p = ctx->H.c; + u64 hi,lo; + hi = (u64)GETU32(p) <<32|GETU32(p+4); + lo = (u64)GETU32(p+8)<<32|GETU32(p+12); + ctx->H.u[0] = hi; + ctx->H.u[1] = lo; +#endif +#endif + +#if TABLE_BITS==8 + gcm_init_8bit(ctx->Htable,ctx->H.u); +#elif TABLE_BITS==4 +# if defined(GHASH_ASM_X86_OR_64) +# if !defined(GHASH_ASM_X86) || defined(OPENSSL_IA32_SSE2) + /* check FXSR and PCLMULQDQ bits */ + if ((OPENSSL_cpu_caps() & (CPUCAP_MASK_FXSR | CPUCAP_MASK_PCLMUL)) == + (CPUCAP_MASK_FXSR | CPUCAP_MASK_PCLMUL)) { + gcm_init_clmul(ctx->Htable,ctx->H.u); + ctx->gmult = gcm_gmult_clmul; + ctx->ghash = gcm_ghash_clmul; + return; + } +# endif + gcm_init_4bit(ctx->Htable,ctx->H.u); +# if defined(GHASH_ASM_X86) /* x86 only */ +# if defined(OPENSSL_IA32_SSE2) + if (OPENSSL_cpu_caps() & CPUCAP_MASK_SSE) { /* check SSE bit */ +# else + if (OPENSSL_cpu_caps() & CPUCAP_MASK_MMX) { /* check MMX bit */ +# endif + ctx->gmult = gcm_gmult_4bit_mmx; + ctx->ghash = gcm_ghash_4bit_mmx; + } else { + ctx->gmult = gcm_gmult_4bit_x86; + ctx->ghash = gcm_ghash_4bit_x86; + } +# else + ctx->gmult = gcm_gmult_4bit; + ctx->ghash = gcm_ghash_4bit; +# endif +# elif defined(GHASH_ASM_ARM) + if (OPENSSL_armcap_P & ARMV7_NEON) { + ctx->gmult = gcm_gmult_neon; + ctx->ghash = gcm_ghash_neon; + } else { + gcm_init_4bit(ctx->Htable,ctx->H.u); + ctx->gmult = gcm_gmult_4bit; + ctx->ghash = gcm_ghash_4bit; + } +# else + gcm_init_4bit(ctx->Htable,ctx->H.u); +# endif +#endif +} + +void CRYPTO_gcm128_setiv(GCM128_CONTEXT *ctx,const unsigned char *iv,size_t len) +{ + unsigned int ctr; +#ifdef GCM_FUNCREF_4BIT + void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; +#endif + + ctx->Yi.u[0] = 0; + ctx->Yi.u[1] = 0; + ctx->Xi.u[0] = 0; + ctx->Xi.u[1] = 0; + ctx->len.u[0] = 0; /* AAD length */ + ctx->len.u[1] = 0; /* message length */ + ctx->ares = 0; + ctx->mres = 0; + + if (len==12) { + memcpy(ctx->Yi.c,iv,12); + ctx->Yi.c[15]=1; + ctr=1; + } + else { + size_t i; + u64 len0 = len; + + while (len>=16) { + for (i=0; i<16; ++i) ctx->Yi.c[i] ^= iv[i]; + GCM_MUL(ctx,Yi); + iv += 16; + len -= 16; + } + if (len) { + for (i=0; iYi.c[i] ^= iv[i]; + GCM_MUL(ctx,Yi); + } + len0 <<= 3; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP8 + ctx->Yi.u[1] ^= BSWAP8(len0); +#else + ctx->Yi.c[8] ^= (u8)(len0>>56); + ctx->Yi.c[9] ^= (u8)(len0>>48); + ctx->Yi.c[10] ^= (u8)(len0>>40); + ctx->Yi.c[11] ^= (u8)(len0>>32); + ctx->Yi.c[12] ^= (u8)(len0>>24); + ctx->Yi.c[13] ^= (u8)(len0>>16); + ctx->Yi.c[14] ^= (u8)(len0>>8); + ctx->Yi.c[15] ^= (u8)(len0); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.u[1] ^= len0; +#endif + + GCM_MUL(ctx,Yi); + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctr = BSWAP4(ctx->Yi.d[3]); +#else + ctr = GETU32(ctx->Yi.c+12); +#endif +#else /* BIG_ENDIAN */ + ctr = ctx->Yi.d[3]; +#endif + } + + (*ctx->block)(ctx->Yi.c,ctx->EK0.c,ctx->key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif +} + +int CRYPTO_gcm128_aad(GCM128_CONTEXT *ctx,const unsigned char *aad,size_t len) +{ + size_t i; + unsigned int n; + u64 alen = ctx->len.u[0]; +#ifdef GCM_FUNCREF_4BIT + void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; +# ifdef GHASH + void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], + const u8 *inp,size_t len) = ctx->ghash; +# endif +#endif + + if (ctx->len.u[1]) return -2; + + alen += len; + if (alen>(U64(1)<<61) || (sizeof(len)==8 && alenlen.u[0] = alen; + + n = ctx->ares; + if (n) { + while (n && len) { + ctx->Xi.c[n] ^= *(aad++); + --len; + n = (n+1)%16; + } + if (n==0) GCM_MUL(ctx,Xi); + else { + ctx->ares = n; + return 0; + } + } + +#ifdef GHASH + if ((i = (len&(size_t)-16))) { + GHASH(ctx,aad,i); + aad += i; + len -= i; + } +#else + while (len>=16) { + for (i=0; i<16; ++i) ctx->Xi.c[i] ^= aad[i]; + GCM_MUL(ctx,Xi); + aad += 16; + len -= 16; + } +#endif + if (len) { + n = (unsigned int)len; + for (i=0; iXi.c[i] ^= aad[i]; + } + + ctx->ares = n; + return 0; +} + +int CRYPTO_gcm128_encrypt(GCM128_CONTEXT *ctx, + const unsigned char *in, unsigned char *out, + size_t len) +{ + unsigned int n, ctr; + size_t i; + u64 mlen = ctx->len.u[1]; + block128_f block = ctx->block; + void *key = ctx->key; +#ifdef GCM_FUNCREF_4BIT + void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; +# ifdef GHASH + void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], + const u8 *inp,size_t len) = ctx->ghash; +# endif +#endif + + mlen += len; + if (mlen>((U64(1)<<36)-32) || (sizeof(len)==8 && mlenlen.u[1] = mlen; + + if (ctx->ares) { + /* First call to encrypt finalizes GHASH(AAD) */ + GCM_MUL(ctx,Xi); + ctx->ares = 0; + } + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctr = BSWAP4(ctx->Yi.d[3]); +#else + ctr = GETU32(ctx->Yi.c+12); +#endif +#else /* BIG_ENDIAN */ + ctr = ctx->Yi.d[3]; +#endif + + n = ctx->mres; +#if !defined(OPENSSL_SMALL_FOOTPRINT) + if (16%sizeof(size_t) == 0) do { /* always true actually */ + if (n) { + while (n && len) { + ctx->Xi.c[n] ^= *(out++) = *(in++)^ctx->EKi.c[n]; + --len; + n = (n+1)%16; + } + if (n==0) GCM_MUL(ctx,Xi); + else { + ctx->mres = n; + return 0; + } + } +#ifdef __STRICT_ALIGNMENT + if (((size_t)in|(size_t)out)%sizeof(size_t) != 0) + break; +#endif +#if defined(GHASH) && defined(GHASH_CHUNK) + while (len>=GHASH_CHUNK) { + size_t j=GHASH_CHUNK; + + while (j) { + size_t *out_t=(size_t *)out; + const size_t *in_t=(const size_t *)in; + + (*block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + for (i=0; i<16/sizeof(size_t); ++i) + out_t[i] = in_t[i] ^ ctx->EKi.t[i]; + out += 16; + in += 16; + j -= 16; + } + GHASH(ctx,out-GHASH_CHUNK,GHASH_CHUNK); + len -= GHASH_CHUNK; + } + if ((i = (len&(size_t)-16))) { + size_t j=i; + + while (len>=16) { + size_t *out_t=(size_t *)out; + const size_t *in_t=(const size_t *)in; + + (*block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + for (i=0; i<16/sizeof(size_t); ++i) + out_t[i] = in_t[i] ^ ctx->EKi.t[i]; + out += 16; + in += 16; + len -= 16; + } + GHASH(ctx,out-j,j); + } +#else + while (len>=16) { + size_t *out_t=(size_t *)out; + const size_t *in_t=(const size_t *)in; + + (*block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + for (i=0; i<16/sizeof(size_t); ++i) + ctx->Xi.t[i] ^= + out_t[i] = in_t[i]^ctx->EKi.t[i]; + GCM_MUL(ctx,Xi); + out += 16; + in += 16; + len -= 16; + } +#endif + if (len) { + (*block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + while (len--) { + ctx->Xi.c[n] ^= out[n] = in[n]^ctx->EKi.c[n]; + ++n; + } + } + + ctx->mres = n; + return 0; + } while(0); +#endif + for (i=0;iYi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + } + ctx->Xi.c[n] ^= out[i] = in[i]^ctx->EKi.c[n]; + n = (n+1)%16; + if (n==0) + GCM_MUL(ctx,Xi); + } + + ctx->mres = n; + return 0; +} + +int CRYPTO_gcm128_decrypt(GCM128_CONTEXT *ctx, + const unsigned char *in, unsigned char *out, + size_t len) +{ + unsigned int n, ctr; + size_t i; + u64 mlen = ctx->len.u[1]; + block128_f block = ctx->block; + void *key = ctx->key; +#ifdef GCM_FUNCREF_4BIT + void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; +# ifdef GHASH + void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], + const u8 *inp,size_t len) = ctx->ghash; +# endif +#endif + + mlen += len; + if (mlen>((U64(1)<<36)-32) || (sizeof(len)==8 && mlenlen.u[1] = mlen; + + if (ctx->ares) { + /* First call to decrypt finalizes GHASH(AAD) */ + GCM_MUL(ctx,Xi); + ctx->ares = 0; + } + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctr = BSWAP4(ctx->Yi.d[3]); +#else + ctr = GETU32(ctx->Yi.c+12); +#endif +#else /* BIG_ENDIAN */ + ctr = ctx->Yi.d[3]; +#endif + + n = ctx->mres; +#if !defined(OPENSSL_SMALL_FOOTPRINT) + if (16%sizeof(size_t) == 0) do { /* always true actually */ + if (n) { + while (n && len) { + u8 c = *(in++); + *(out++) = c^ctx->EKi.c[n]; + ctx->Xi.c[n] ^= c; + --len; + n = (n+1)%16; + } + if (n==0) GCM_MUL (ctx,Xi); + else { + ctx->mres = n; + return 0; + } + } +#ifdef __STRICT_ALIGNMENT + if (((size_t)in|(size_t)out)%sizeof(size_t) != 0) + break; +#endif +#if defined(GHASH) && defined(GHASH_CHUNK) + while (len>=GHASH_CHUNK) { + size_t j=GHASH_CHUNK; + + GHASH(ctx,in,GHASH_CHUNK); + while (j) { + size_t *out_t=(size_t *)out; + const size_t *in_t=(const size_t *)in; + + (*block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + for (i=0; i<16/sizeof(size_t); ++i) + out_t[i] = in_t[i]^ctx->EKi.t[i]; + out += 16; + in += 16; + j -= 16; + } + len -= GHASH_CHUNK; + } + if ((i = (len&(size_t)-16))) { + GHASH(ctx,in,i); + while (len>=16) { + size_t *out_t=(size_t *)out; + const size_t *in_t=(const size_t *)in; + + (*block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + for (i=0; i<16/sizeof(size_t); ++i) + out_t[i] = in_t[i]^ctx->EKi.t[i]; + out += 16; + in += 16; + len -= 16; + } + } +#else + while (len>=16) { + size_t *out_t=(size_t *)out; + const size_t *in_t=(const size_t *)in; + + (*block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + for (i=0; i<16/sizeof(size_t); ++i) { + size_t c = in[i]; + out[i] = c^ctx->EKi.t[i]; + ctx->Xi.t[i] ^= c; + } + GCM_MUL(ctx,Xi); + out += 16; + in += 16; + len -= 16; + } +#endif + if (len) { + (*block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + while (len--) { + u8 c = in[n]; + ctx->Xi.c[n] ^= c; + out[n] = c^ctx->EKi.c[n]; + ++n; + } + } + + ctx->mres = n; + return 0; + } while(0); +#endif + for (i=0;iYi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + } + c = in[i]; + out[i] = c^ctx->EKi.c[n]; + ctx->Xi.c[n] ^= c; + n = (n+1)%16; + if (n==0) + GCM_MUL(ctx,Xi); + } + + ctx->mres = n; + return 0; +} + +int CRYPTO_gcm128_encrypt_ctr32(GCM128_CONTEXT *ctx, + const unsigned char *in, unsigned char *out, + size_t len, ctr128_f stream) +{ + unsigned int n, ctr; + size_t i; + u64 mlen = ctx->len.u[1]; + void *key = ctx->key; +#ifdef GCM_FUNCREF_4BIT + void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; +# ifdef GHASH + void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], + const u8 *inp,size_t len) = ctx->ghash; +# endif +#endif + + mlen += len; + if (mlen>((U64(1)<<36)-32) || (sizeof(len)==8 && mlenlen.u[1] = mlen; + + if (ctx->ares) { + /* First call to encrypt finalizes GHASH(AAD) */ + GCM_MUL(ctx,Xi); + ctx->ares = 0; + } + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctr = BSWAP4(ctx->Yi.d[3]); +#else + ctr = GETU32(ctx->Yi.c+12); +#endif +#else /* BIG_ENDIAN */ + ctr = ctx->Yi.d[3]; +#endif + + n = ctx->mres; + if (n) { + while (n && len) { + ctx->Xi.c[n] ^= *(out++) = *(in++)^ctx->EKi.c[n]; + --len; + n = (n+1)%16; + } + if (n==0) GCM_MUL(ctx,Xi); + else { + ctx->mres = n; + return 0; + } + } +#if defined(GHASH) && !defined(OPENSSL_SMALL_FOOTPRINT) + while (len>=GHASH_CHUNK) { + (*stream)(in,out,GHASH_CHUNK/16,key,ctx->Yi.c); + ctr += GHASH_CHUNK/16; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + GHASH(ctx,out,GHASH_CHUNK); + out += GHASH_CHUNK; + in += GHASH_CHUNK; + len -= GHASH_CHUNK; + } +#endif + if ((i = (len&(size_t)-16))) { + size_t j=i/16; + + (*stream)(in,out,j,key,ctx->Yi.c); + ctr += (unsigned int)j; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + in += i; + len -= i; +#if defined(GHASH) + GHASH(ctx,out,i); + out += i; +#else + while (j--) { + for (i=0;i<16;++i) ctx->Xi.c[i] ^= out[i]; + GCM_MUL(ctx,Xi); + out += 16; + } +#endif + } + if (len) { + (*ctx->block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + while (len--) { + ctx->Xi.c[n] ^= out[n] = in[n]^ctx->EKi.c[n]; + ++n; + } + } + + ctx->mres = n; + return 0; +} + +int CRYPTO_gcm128_decrypt_ctr32(GCM128_CONTEXT *ctx, + const unsigned char *in, unsigned char *out, + size_t len,ctr128_f stream) +{ + unsigned int n, ctr; + size_t i; + u64 mlen = ctx->len.u[1]; + void *key = ctx->key; +#ifdef GCM_FUNCREF_4BIT + void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; +# ifdef GHASH + void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], + const u8 *inp,size_t len) = ctx->ghash; +# endif +#endif + + mlen += len; + if (mlen>((U64(1)<<36)-32) || (sizeof(len)==8 && mlenlen.u[1] = mlen; + + if (ctx->ares) { + /* First call to decrypt finalizes GHASH(AAD) */ + GCM_MUL(ctx,Xi); + ctx->ares = 0; + } + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctr = BSWAP4(ctx->Yi.d[3]); +#else + ctr = GETU32(ctx->Yi.c+12); +#endif +#else /* BIG_ENDIAN */ + ctr = ctx->Yi.d[3]; +#endif + + n = ctx->mres; + if (n) { + while (n && len) { + u8 c = *(in++); + *(out++) = c^ctx->EKi.c[n]; + ctx->Xi.c[n] ^= c; + --len; + n = (n+1)%16; + } + if (n==0) GCM_MUL (ctx,Xi); + else { + ctx->mres = n; + return 0; + } + } +#if defined(GHASH) && !defined(OPENSSL_SMALL_FOOTPRINT) + while (len>=GHASH_CHUNK) { + GHASH(ctx,in,GHASH_CHUNK); + (*stream)(in,out,GHASH_CHUNK/16,key,ctx->Yi.c); + ctr += GHASH_CHUNK/16; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + out += GHASH_CHUNK; + in += GHASH_CHUNK; + len -= GHASH_CHUNK; + } +#endif + if ((i = (len&(size_t)-16))) { + size_t j=i/16; + +#if defined(GHASH) + GHASH(ctx,in,i); +#else + while (j--) { + size_t k; + for (k=0;k<16;++k) ctx->Xi.c[k] ^= in[k]; + GCM_MUL(ctx,Xi); + in += 16; + } + j = i/16; + in -= i; +#endif + (*stream)(in,out,j,key,ctx->Yi.c); + ctr += (unsigned int)j; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + out += i; + in += i; + len -= i; + } + if (len) { + (*ctx->block)(ctx->Yi.c,ctx->EKi.c,key); + ++ctr; +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +#else + PUTU32(ctx->Yi.c+12,ctr); +#endif +#else /* BIG_ENDIAN */ + ctx->Yi.d[3] = ctr; +#endif + while (len--) { + u8 c = in[n]; + ctx->Xi.c[n] ^= c; + out[n] = c^ctx->EKi.c[n]; + ++n; + } + } + + ctx->mres = n; + return 0; +} + +int CRYPTO_gcm128_finish(GCM128_CONTEXT *ctx,const unsigned char *tag, + size_t len) +{ + u64 alen = ctx->len.u[0]<<3; + u64 clen = ctx->len.u[1]<<3; +#ifdef GCM_FUNCREF_4BIT + void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; +#endif + + if (ctx->mres || ctx->ares) + GCM_MUL(ctx,Xi); + +#if BYTE_ORDER == LITTLE_ENDIAN +#ifdef BSWAP8 + alen = BSWAP8(alen); + clen = BSWAP8(clen); +#else + { + u8 *p = ctx->len.c; + + ctx->len.u[0] = alen; + ctx->len.u[1] = clen; + + alen = (u64)GETU32(p) <<32|GETU32(p+4); + clen = (u64)GETU32(p+8)<<32|GETU32(p+12); + } +#endif +#endif + + ctx->Xi.u[0] ^= alen; + ctx->Xi.u[1] ^= clen; + GCM_MUL(ctx,Xi); + + ctx->Xi.u[0] ^= ctx->EK0.u[0]; + ctx->Xi.u[1] ^= ctx->EK0.u[1]; + + if (tag && len<=sizeof(ctx->Xi)) + return memcmp(ctx->Xi.c,tag,len); + else + return -1; +} + +void CRYPTO_gcm128_tag(GCM128_CONTEXT *ctx, unsigned char *tag, size_t len) +{ + CRYPTO_gcm128_finish(ctx, NULL, 0); + memcpy(tag, ctx->Xi.c, len<=sizeof(ctx->Xi.c)?len:sizeof(ctx->Xi.c)); +} + +#if 0 + +GCM128_CONTEXT *CRYPTO_gcm128_new(void *key, block128_f block) +{ + GCM128_CONTEXT *ret; + + if ((ret = malloc(sizeof(GCM128_CONTEXT)))) + CRYPTO_gcm128_init(ret,key,block); + + return ret; +} + +void CRYPTO_gcm128_release(GCM128_CONTEXT *ctx) +{ + freezero(ctx, sizeof(*ctx)); +} + +#endif diff --git a/ext/libressl/crypto/modes/ghash-elf-armv4.S b/ext/libressl/crypto/modes/ghash-elf-armv4.S new file mode 100644 index 0000000..af42593 --- /dev/null +++ b/ext/libressl/crypto/modes/ghash-elf-armv4.S @@ -0,0 +1,412 @@ +#include "arm_arch.h" + +.text +.syntax unified +.code 32 + +.type rem_4bit,%object +.align 5 +rem_4bit: +.short 0x0000,0x1C20,0x3840,0x2460 +.short 0x7080,0x6CA0,0x48C0,0x54E0 +.short 0xE100,0xFD20,0xD940,0xC560 +.short 0x9180,0x8DA0,0xA9C0,0xB5E0 +.size rem_4bit,.-rem_4bit + +.type rem_4bit_get,%function +rem_4bit_get: + sub r2,pc,#8 + sub r2,r2,#32 @ &rem_4bit + b .Lrem_4bit_got + nop +.size rem_4bit_get,.-rem_4bit_get + +.global gcm_ghash_4bit +.type gcm_ghash_4bit,%function +gcm_ghash_4bit: + sub r12,pc,#8 + add r3,r2,r3 @ r3 to point at the end + stmdb sp!,{r3-r11,lr} @ save r3/end too + sub r12,r12,#48 @ &rem_4bit + + ldmia r12,{r4-r11} @ copy rem_4bit ... + stmdb sp!,{r4-r11} @ ... to stack + + ldrb r12,[r2,#15] + ldrb r14,[r0,#15] +.Louter: + eor r12,r12,r14 + and r14,r12,#0xf0 + and r12,r12,#0x0f + mov r3,#14 + + add r7,r1,r12,lsl#4 + ldmia r7,{r4-r7} @ load Htbl[nlo] + add r11,r1,r14 + ldrb r12,[r2,#14] + + and r14,r4,#0xf @ rem + ldmia r11,{r8-r11} @ load Htbl[nhi] + add r14,r14,r14 + eor r4,r8,r4,lsr#4 + ldrh r8,[sp,r14] @ rem_4bit[rem] + eor r4,r4,r5,lsl#28 + ldrb r14,[r0,#14] + eor r5,r9,r5,lsr#4 + eor r5,r5,r6,lsl#28 + eor r6,r10,r6,lsr#4 + eor r6,r6,r7,lsl#28 + eor r7,r11,r7,lsr#4 + eor r12,r12,r14 + and r14,r12,#0xf0 + and r12,r12,#0x0f + eor r7,r7,r8,lsl#16 + +.Linner: + add r11,r1,r12,lsl#4 + and r12,r4,#0xf @ rem + subs r3,r3,#1 + add r12,r12,r12 + ldmia r11,{r8-r11} @ load Htbl[nlo] + eor r4,r8,r4,lsr#4 + eor r4,r4,r5,lsl#28 + eor r5,r9,r5,lsr#4 + eor r5,r5,r6,lsl#28 + ldrh r8,[sp,r12] @ rem_4bit[rem] + eor r6,r10,r6,lsr#4 + ldrbpl r12,[r2,r3] + eor r6,r6,r7,lsl#28 + eor r7,r11,r7,lsr#4 + + add r11,r1,r14 + and r14,r4,#0xf @ rem + eor r7,r7,r8,lsl#16 @ ^= rem_4bit[rem] + add r14,r14,r14 + ldmia r11,{r8-r11} @ load Htbl[nhi] + eor r4,r8,r4,lsr#4 + ldrbpl r8,[r0,r3] + eor r4,r4,r5,lsl#28 + eor r5,r9,r5,lsr#4 + ldrh r9,[sp,r14] + eor r5,r5,r6,lsl#28 + eor r6,r10,r6,lsr#4 + eor r6,r6,r7,lsl#28 + eorpl r12,r12,r8 + eor r7,r11,r7,lsr#4 + andpl r14,r12,#0xf0 + andpl r12,r12,#0x0f + eor r7,r7,r9,lsl#16 @ ^= rem_4bit[rem] + bpl .Linner + + ldr r3,[sp,#32] @ re-load r3/end + add r2,r2,#16 + mov r14,r4 +#if __ARM_ARCH__>=7 && defined(__ARMEL__) + rev r4,r4 + str r4,[r0,#12] +#elif defined(__ARMEB__) + str r4,[r0,#12] +#else + mov r9,r4,lsr#8 + strb r4,[r0,#12+3] + mov r10,r4,lsr#16 + strb r9,[r0,#12+2] + mov r11,r4,lsr#24 + strb r10,[r0,#12+1] + strb r11,[r0,#12] +#endif + cmp r2,r3 +#if __ARM_ARCH__>=7 && defined(__ARMEL__) + rev r5,r5 + str r5,[r0,#8] +#elif defined(__ARMEB__) + str r5,[r0,#8] +#else + mov r9,r5,lsr#8 + strb r5,[r0,#8+3] + mov r10,r5,lsr#16 + strb r9,[r0,#8+2] + mov r11,r5,lsr#24 + strb r10,[r0,#8+1] + strb r11,[r0,#8] +#endif + ldrbne r12,[r2,#15] +#if __ARM_ARCH__>=7 && defined(__ARMEL__) + rev r6,r6 + str r6,[r0,#4] +#elif defined(__ARMEB__) + str r6,[r0,#4] +#else + mov r9,r6,lsr#8 + strb r6,[r0,#4+3] + mov r10,r6,lsr#16 + strb r9,[r0,#4+2] + mov r11,r6,lsr#24 + strb r10,[r0,#4+1] + strb r11,[r0,#4] +#endif + +#if __ARM_ARCH__>=7 && defined(__ARMEL__) + rev r7,r7 + str r7,[r0,#0] +#elif defined(__ARMEB__) + str r7,[r0,#0] +#else + mov r9,r7,lsr#8 + strb r7,[r0,#0+3] + mov r10,r7,lsr#16 + strb r9,[r0,#0+2] + mov r11,r7,lsr#24 + strb r10,[r0,#0+1] + strb r11,[r0,#0] +#endif + + bne .Louter + + add sp,sp,#36 +#if __ARM_ARCH__>=5 + ldmia sp!,{r4-r11,pc} +#else + ldmia sp!,{r4-r11,lr} + tst lr,#1 + moveq pc,lr @ be binary compatible with V4, yet + .word 0xe12fff1e @ interoperable with Thumb ISA:-) +#endif +.size gcm_ghash_4bit,.-gcm_ghash_4bit + +.global gcm_gmult_4bit +.type gcm_gmult_4bit,%function +gcm_gmult_4bit: + stmdb sp!,{r4-r11,lr} + ldrb r12,[r0,#15] + b rem_4bit_get +.Lrem_4bit_got: + and r14,r12,#0xf0 + and r12,r12,#0x0f + mov r3,#14 + + add r7,r1,r12,lsl#4 + ldmia r7,{r4-r7} @ load Htbl[nlo] + ldrb r12,[r0,#14] + + add r11,r1,r14 + and r14,r4,#0xf @ rem + ldmia r11,{r8-r11} @ load Htbl[nhi] + add r14,r14,r14 + eor r4,r8,r4,lsr#4 + ldrh r8,[r2,r14] @ rem_4bit[rem] + eor r4,r4,r5,lsl#28 + eor r5,r9,r5,lsr#4 + eor r5,r5,r6,lsl#28 + eor r6,r10,r6,lsr#4 + eor r6,r6,r7,lsl#28 + eor r7,r11,r7,lsr#4 + and r14,r12,#0xf0 + eor r7,r7,r8,lsl#16 + and r12,r12,#0x0f + +.Loop: + add r11,r1,r12,lsl#4 + and r12,r4,#0xf @ rem + subs r3,r3,#1 + add r12,r12,r12 + ldmia r11,{r8-r11} @ load Htbl[nlo] + eor r4,r8,r4,lsr#4 + eor r4,r4,r5,lsl#28 + eor r5,r9,r5,lsr#4 + eor r5,r5,r6,lsl#28 + ldrh r8,[r2,r12] @ rem_4bit[rem] + eor r6,r10,r6,lsr#4 + ldrbpl r12,[r0,r3] + eor r6,r6,r7,lsl#28 + eor r7,r11,r7,lsr#4 + + add r11,r1,r14 + and r14,r4,#0xf @ rem + eor r7,r7,r8,lsl#16 @ ^= rem_4bit[rem] + add r14,r14,r14 + ldmia r11,{r8-r11} @ load Htbl[nhi] + eor r4,r8,r4,lsr#4 + eor r4,r4,r5,lsl#28 + eor r5,r9,r5,lsr#4 + ldrh r8,[r2,r14] @ rem_4bit[rem] + eor r5,r5,r6,lsl#28 + eor r6,r10,r6,lsr#4 + eor r6,r6,r7,lsl#28 + eor r7,r11,r7,lsr#4 + andpl r14,r12,#0xf0 + andpl r12,r12,#0x0f + eor r7,r7,r8,lsl#16 @ ^= rem_4bit[rem] + bpl .Loop +#if __ARM_ARCH__>=7 && defined(__ARMEL__) + rev r4,r4 + str r4,[r0,#12] +#elif defined(__ARMEB__) + str r4,[r0,#12] +#else + mov r9,r4,lsr#8 + strb r4,[r0,#12+3] + mov r10,r4,lsr#16 + strb r9,[r0,#12+2] + mov r11,r4,lsr#24 + strb r10,[r0,#12+1] + strb r11,[r0,#12] +#endif + +#if __ARM_ARCH__>=7 && defined(__ARMEL__) + rev r5,r5 + str r5,[r0,#8] +#elif defined(__ARMEB__) + str r5,[r0,#8] +#else + mov r9,r5,lsr#8 + strb r5,[r0,#8+3] + mov r10,r5,lsr#16 + strb r9,[r0,#8+2] + mov r11,r5,lsr#24 + strb r10,[r0,#8+1] + strb r11,[r0,#8] +#endif + +#if __ARM_ARCH__>=7 && defined(__ARMEL__) + rev r6,r6 + str r6,[r0,#4] +#elif defined(__ARMEB__) + str r6,[r0,#4] +#else + mov r9,r6,lsr#8 + strb r6,[r0,#4+3] + mov r10,r6,lsr#16 + strb r9,[r0,#4+2] + mov r11,r6,lsr#24 + strb r10,[r0,#4+1] + strb r11,[r0,#4] +#endif + +#if __ARM_ARCH__>=7 && defined(__ARMEL__) + rev r7,r7 + str r7,[r0,#0] +#elif defined(__ARMEB__) + str r7,[r0,#0] +#else + mov r9,r7,lsr#8 + strb r7,[r0,#0+3] + mov r10,r7,lsr#16 + strb r9,[r0,#0+2] + mov r11,r7,lsr#24 + strb r10,[r0,#0+1] + strb r11,[r0,#0] +#endif + +#if __ARM_ARCH__>=5 + ldmia sp!,{r4-r11,pc} +#else + ldmia sp!,{r4-r11,lr} + tst lr,#1 + moveq pc,lr @ be binary compatible with V4, yet + .word 0xe12fff1e @ interoperable with Thumb ISA:-) +#endif +.size gcm_gmult_4bit,.-gcm_gmult_4bit +#if __ARM_ARCH__>=7 && !defined(__STRICT_ALIGNMENT) +.fpu neon + +.global gcm_gmult_neon +.type gcm_gmult_neon,%function +.align 4 +gcm_gmult_neon: + sub r1,#16 @ point at H in GCM128_CTX + vld1.64 d29,[r0,:64]!@ load Xi + vmov.i32 d5,#0xe1 @ our irreducible polynomial + vld1.64 d28,[r0,:64]! + vshr.u64 d5,#32 + vldmia r1,{d0-d1} @ load H + veor q12,q12 +#ifdef __ARMEL__ + vrev64.8 q14,q14 +#endif + veor q13,q13 + veor q11,q11 + mov r1,#16 + veor q10,q10 + mov r3,#16 + veor d2,d2 + vdup.8 d4,d28[0] @ broadcast lowest byte + b .Linner_neon +.size gcm_gmult_neon,.-gcm_gmult_neon + +.global gcm_ghash_neon +.type gcm_ghash_neon,%function +.align 4 +gcm_ghash_neon: + vld1.64 d21,[r0,:64]! @ load Xi + vmov.i32 d5,#0xe1 @ our irreducible polynomial + vld1.64 d20,[r0,:64]! + vshr.u64 d5,#32 + vldmia r0,{d0-d1} @ load H + veor q12,q12 + nop +#ifdef __ARMEL__ + vrev64.8 q10,q10 +#endif +.Louter_neon: + vld1.64 d29,[r2]! @ load inp + veor q13,q13 + vld1.64 d28,[r2]! + veor q11,q11 + mov r1,#16 +#ifdef __ARMEL__ + vrev64.8 q14,q14 +#endif + veor d2,d2 + veor q14,q10 @ inp^=Xi + veor q10,q10 + vdup.8 d4,d28[0] @ broadcast lowest byte +.Linner_neon: + subs r1,r1,#1 + vmull.p8 q9,d1,d4 @ H.lo·Xi[i] + vmull.p8 q8,d0,d4 @ H.hi·Xi[i] + vext.8 q14,q12,#1 @ IN>>=8 + + veor q10,q13 @ modulo-scheduled part + vshl.i64 d22,#48 + vdup.8 d4,d28[0] @ broadcast lowest byte + veor d3,d18,d20 + + veor d21,d22 + vuzp.8 q9,q8 + vsli.8 d2,d3,#1 @ compose the "carry" byte + vext.8 q10,q12,#1 @ Z>>=8 + + vmull.p8 q11,d2,d5 @ "carry"·0xe1 + vshr.u8 d2,d3,#7 @ save Z's bottom bit + vext.8 q13,q9,q12,#1 @ Qlo>>=8 + veor q10,q8 + bne .Linner_neon + + veor q10,q13 @ modulo-scheduled artefact + vshl.i64 d22,#48 + veor d21,d22 + + @ finalization, normalize Z:Zo + vand d2,d5 @ suffices to mask the bit + vshr.u64 d3,d20,#63 + vshl.i64 q10,#1 + subs r3,#16 + vorr q10,q1 @ Z=Z:Zo<<1 + bne .Louter_neon + +#ifdef __ARMEL__ + vrev64.8 q10,q10 +#endif + sub r0,#16 + vst1.64 d21,[r0,:64]! @ write out Xi + vst1.64 d20,[r0,:64] + + .word 0xe12fff1e +.size gcm_ghash_neon,.-gcm_ghash_neon +#endif +.asciz "GHASH for ARMv4/NEON, CRYPTOGAMS by " +.align 2 +#if defined(HAVE_GNU_STACK) +.section .note.GNU-stack,"",%progbits +#endif diff --git a/ext/libressl/crypto/modes/ghash-elf-x86_64.S b/ext/libressl/crypto/modes/ghash-elf-x86_64.S new file mode 100644 index 0000000..5f31626 --- /dev/null +++ b/ext/libressl/crypto/modes/ghash-elf-x86_64.S @@ -0,0 +1,1030 @@ +#include "x86_arch.h" +.text + +.globl gcm_gmult_4bit +.type gcm_gmult_4bit,@function +.align 16 +gcm_gmult_4bit: + pushq %rbx + pushq %rbp + pushq %r12 +.Lgmult_prologue: + + movzbq 15(%rdi),%r8 + leaq .Lrem_4bit(%rip),%r11 + xorq %rax,%rax + xorq %rbx,%rbx + movb %r8b,%al + movb %r8b,%bl + shlb $4,%al + movq $14,%rcx + movq 8(%rsi,%rax,1),%r8 + movq (%rsi,%rax,1),%r9 + andb $240,%bl + movq %r8,%rdx + jmp .Loop1 + +.align 16 +.Loop1: + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + movb (%rdi,%rcx,1),%al + shrq $4,%r9 + xorq 8(%rsi,%rbx,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rbx,1),%r9 + movb %al,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + shlb $4,%al + xorq %r10,%r8 + decq %rcx + js .Lbreak1 + + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rax,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rax,1),%r9 + andb $240,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + xorq %r10,%r8 + jmp .Loop1 + +.align 16 +.Lbreak1: + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rax,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rax,1),%r9 + andb $240,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + xorq %r10,%r8 + + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rbx,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rbx,1),%r9 + xorq %r10,%r8 + xorq (%r11,%rdx,8),%r9 + + bswapq %r8 + bswapq %r9 + movq %r8,8(%rdi) + movq %r9,(%rdi) + + movq 16(%rsp),%rbx + leaq 24(%rsp),%rsp +.Lgmult_epilogue: + retq +.size gcm_gmult_4bit,.-gcm_gmult_4bit +.globl gcm_ghash_4bit +.type gcm_ghash_4bit,@function +.align 16 +gcm_ghash_4bit: + pushq %rbx + pushq %rbp + pushq %r12 + pushq %r13 + pushq %r14 + pushq %r15 + subq $280,%rsp +.Lghash_prologue: + movq %rdx,%r14 + movq %rcx,%r15 + subq $-128,%rsi + leaq 16+128(%rsp),%rbp + xorl %edx,%edx + movq 0+0-128(%rsi),%r8 + movq 0+8-128(%rsi),%rax + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq 16+0-128(%rsi),%r9 + shlb $4,%dl + movq 16+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,0(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,0(%rbp) + movq 32+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,0-128(%rbp) + movq 32+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,1(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,8(%rbp) + movq 48+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,8-128(%rbp) + movq 48+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,2(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,16(%rbp) + movq 64+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,16-128(%rbp) + movq 64+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,3(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,24(%rbp) + movq 80+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,24-128(%rbp) + movq 80+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,4(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,32(%rbp) + movq 96+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,32-128(%rbp) + movq 96+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,5(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,40(%rbp) + movq 112+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,40-128(%rbp) + movq 112+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,6(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,48(%rbp) + movq 128+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,48-128(%rbp) + movq 128+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,7(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,56(%rbp) + movq 144+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,56-128(%rbp) + movq 144+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,8(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,64(%rbp) + movq 160+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,64-128(%rbp) + movq 160+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,9(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,72(%rbp) + movq 176+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,72-128(%rbp) + movq 176+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,10(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,80(%rbp) + movq 192+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,80-128(%rbp) + movq 192+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,11(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,88(%rbp) + movq 208+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,88-128(%rbp) + movq 208+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,12(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,96(%rbp) + movq 224+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,96-128(%rbp) + movq 224+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,13(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,104(%rbp) + movq 240+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,104-128(%rbp) + movq 240+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,14(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,112(%rbp) + shlb $4,%dl + movq %rax,112-128(%rbp) + shlq $60,%r10 + movb %dl,15(%rsp) + orq %r10,%rbx + movq %r9,120(%rbp) + movq %rbx,120-128(%rbp) + addq $-128,%rsi + movq 8(%rdi),%r8 + movq 0(%rdi),%r9 + addq %r14,%r15 + leaq .Lrem_8bit(%rip),%r11 + jmp .Louter_loop +.align 16 +.Louter_loop: + xorq (%r14),%r9 + movq 8(%r14),%rdx + leaq 16(%r14),%r14 + xorq %r8,%rdx + movq %r9,(%rdi) + movq %rdx,8(%rdi) + shrq $32,%rdx + xorq %rax,%rax + roll $8,%edx + movb %dl,%al + movzbl %dl,%ebx + shlb $4,%al + shrl $4,%ebx + roll $8,%edx + movq 8(%rsi,%rax,1),%r8 + movq (%rsi,%rax,1),%r9 + movb %dl,%al + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + xorq %r8,%r12 + movq %r9,%r10 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 8(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 4(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 0(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + andl $240,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl -4(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + movzwq (%r11,%r12,2),%r12 + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + shlq $48,%r12 + xorq %r10,%r8 + xorq %r12,%r9 + movzbq %r8b,%r13 + shrq $4,%r8 + movq %r9,%r10 + shlb $4,%r13b + shrq $4,%r9 + xorq 8(%rsi,%rcx,1),%r8 + movzwq (%r11,%r13,2),%r13 + shlq $60,%r10 + xorq (%rsi,%rcx,1),%r9 + xorq %r10,%r8 + shlq $48,%r13 + bswapq %r8 + xorq %r13,%r9 + bswapq %r9 + cmpq %r15,%r14 + jb .Louter_loop + movq %r8,8(%rdi) + movq %r9,(%rdi) + + leaq 280(%rsp),%rsi + movq 0(%rsi),%r15 + movq 8(%rsi),%r14 + movq 16(%rsi),%r13 + movq 24(%rsi),%r12 + movq 32(%rsi),%rbp + movq 40(%rsi),%rbx + leaq 48(%rsi),%rsp +.Lghash_epilogue: + retq +.size gcm_ghash_4bit,.-gcm_ghash_4bit +.globl gcm_init_clmul +.type gcm_init_clmul,@function +.align 16 +gcm_init_clmul: + movdqu (%rsi),%xmm2 + pshufd $78,%xmm2,%xmm2 + + + pshufd $255,%xmm2,%xmm4 + movdqa %xmm2,%xmm3 + psllq $1,%xmm2 + pxor %xmm5,%xmm5 + psrlq $63,%xmm3 + pcmpgtd %xmm4,%xmm5 + pslldq $8,%xmm3 + por %xmm3,%xmm2 + + + pand .L0x1c2_polynomial(%rip),%xmm5 + pxor %xmm5,%xmm2 + + + movdqa %xmm2,%xmm0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + movdqu %xmm2,(%rdi) + movdqu %xmm0,16(%rdi) + retq +.size gcm_init_clmul,.-gcm_init_clmul +.globl gcm_gmult_clmul +.type gcm_gmult_clmul,@function +.align 16 +gcm_gmult_clmul: + movdqu (%rdi),%xmm0 + movdqa .Lbswap_mask(%rip),%xmm5 + movdqu (%rsi),%xmm2 +.byte 102,15,56,0,197 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 +.byte 102,15,56,0,197 + movdqu %xmm0,(%rdi) + retq +.size gcm_gmult_clmul,.-gcm_gmult_clmul +.globl gcm_ghash_clmul +.type gcm_ghash_clmul,@function +.align 16 +gcm_ghash_clmul: + movdqa .Lbswap_mask(%rip),%xmm5 + + movdqu (%rdi),%xmm0 + movdqu (%rsi),%xmm2 +.byte 102,15,56,0,197 + + subq $16,%rcx + jz .Lodd_tail + + movdqu 16(%rsi),%xmm8 + + + + + + movdqu (%rdx),%xmm3 + movdqu 16(%rdx),%xmm6 +.byte 102,15,56,0,221 +.byte 102,15,56,0,245 + pxor %xmm3,%xmm0 + movdqa %xmm6,%xmm7 + pshufd $78,%xmm6,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm6,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,242,0 +.byte 102,15,58,68,250,17 +.byte 102,15,58,68,220,0 + pxor %xmm6,%xmm3 + pxor %xmm7,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm7 + pxor %xmm4,%xmm6 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm8,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm8,%xmm4 + + leaq 32(%rdx),%rdx + subq $32,%rcx + jbe .Leven_tail + +.Lmod_loop: +.byte 102,65,15,58,68,192,0 +.byte 102,65,15,58,68,200,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + movdqu (%rdx),%xmm3 + pxor %xmm6,%xmm0 + pxor %xmm7,%xmm1 + + movdqu 16(%rdx),%xmm6 +.byte 102,15,56,0,221 +.byte 102,15,56,0,245 + + movdqa %xmm6,%xmm7 + pshufd $78,%xmm6,%xmm9 + pshufd $78,%xmm2,%xmm10 + pxor %xmm6,%xmm9 + pxor %xmm2,%xmm10 + pxor %xmm3,%xmm1 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 +.byte 102,15,58,68,242,0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + +.byte 102,15,58,68,250,17 + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + +.byte 102,69,15,58,68,202,0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm8,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm8,%xmm4 + + pxor %xmm6,%xmm9 + pxor %xmm7,%xmm9 + movdqa %xmm9,%xmm10 + psrldq $8,%xmm9 + pslldq $8,%xmm10 + pxor %xmm9,%xmm7 + pxor %xmm10,%xmm6 + + leaq 32(%rdx),%rdx + subq $32,%rcx + ja .Lmod_loop + +.Leven_tail: +.byte 102,65,15,58,68,192,0 +.byte 102,65,15,58,68,200,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + pxor %xmm6,%xmm0 + pxor %xmm7,%xmm1 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + testq %rcx,%rcx + jnz .Ldone + +.Lodd_tail: + movdqu (%rdx),%xmm3 +.byte 102,15,56,0,221 + pxor %xmm3,%xmm0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 +.Ldone: +.byte 102,15,56,0,197 + movdqu %xmm0,(%rdi) + retq +.LSEH_end_gcm_ghash_clmul: +.size gcm_ghash_clmul,.-gcm_ghash_clmul +.align 64 +.Lbswap_mask: +.byte 15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0 +.L0x1c2_polynomial: +.byte 1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xc2 +.align 64 +.type .Lrem_4bit,@object +.Lrem_4bit: +.long 0,0,0,471859200,0,943718400,0,610271232 +.long 0,1887436800,0,1822425088,0,1220542464,0,1423966208 +.long 0,3774873600,0,4246732800,0,3644850176,0,3311403008 +.long 0,2441084928,0,2376073216,0,2847932416,0,3051356160 +.type .Lrem_8bit,@object +.Lrem_8bit: +.value 0x0000,0x01C2,0x0384,0x0246,0x0708,0x06CA,0x048C,0x054E +.value 0x0E10,0x0FD2,0x0D94,0x0C56,0x0918,0x08DA,0x0A9C,0x0B5E +.value 0x1C20,0x1DE2,0x1FA4,0x1E66,0x1B28,0x1AEA,0x18AC,0x196E +.value 0x1230,0x13F2,0x11B4,0x1076,0x1538,0x14FA,0x16BC,0x177E +.value 0x3840,0x3982,0x3BC4,0x3A06,0x3F48,0x3E8A,0x3CCC,0x3D0E +.value 0x3650,0x3792,0x35D4,0x3416,0x3158,0x309A,0x32DC,0x331E +.value 0x2460,0x25A2,0x27E4,0x2626,0x2368,0x22AA,0x20EC,0x212E +.value 0x2A70,0x2BB2,0x29F4,0x2836,0x2D78,0x2CBA,0x2EFC,0x2F3E +.value 0x7080,0x7142,0x7304,0x72C6,0x7788,0x764A,0x740C,0x75CE +.value 0x7E90,0x7F52,0x7D14,0x7CD6,0x7998,0x785A,0x7A1C,0x7BDE +.value 0x6CA0,0x6D62,0x6F24,0x6EE6,0x6BA8,0x6A6A,0x682C,0x69EE +.value 0x62B0,0x6372,0x6134,0x60F6,0x65B8,0x647A,0x663C,0x67FE +.value 0x48C0,0x4902,0x4B44,0x4A86,0x4FC8,0x4E0A,0x4C4C,0x4D8E +.value 0x46D0,0x4712,0x4554,0x4496,0x41D8,0x401A,0x425C,0x439E +.value 0x54E0,0x5522,0x5764,0x56A6,0x53E8,0x522A,0x506C,0x51AE +.value 0x5AF0,0x5B32,0x5974,0x58B6,0x5DF8,0x5C3A,0x5E7C,0x5FBE +.value 0xE100,0xE0C2,0xE284,0xE346,0xE608,0xE7CA,0xE58C,0xE44E +.value 0xEF10,0xEED2,0xEC94,0xED56,0xE818,0xE9DA,0xEB9C,0xEA5E +.value 0xFD20,0xFCE2,0xFEA4,0xFF66,0xFA28,0xFBEA,0xF9AC,0xF86E +.value 0xF330,0xF2F2,0xF0B4,0xF176,0xF438,0xF5FA,0xF7BC,0xF67E +.value 0xD940,0xD882,0xDAC4,0xDB06,0xDE48,0xDF8A,0xDDCC,0xDC0E +.value 0xD750,0xD692,0xD4D4,0xD516,0xD058,0xD19A,0xD3DC,0xD21E +.value 0xC560,0xC4A2,0xC6E4,0xC726,0xC268,0xC3AA,0xC1EC,0xC02E +.value 0xCB70,0xCAB2,0xC8F4,0xC936,0xCC78,0xCDBA,0xCFFC,0xCE3E +.value 0x9180,0x9042,0x9204,0x93C6,0x9688,0x974A,0x950C,0x94CE +.value 0x9F90,0x9E52,0x9C14,0x9DD6,0x9898,0x995A,0x9B1C,0x9ADE +.value 0x8DA0,0x8C62,0x8E24,0x8FE6,0x8AA8,0x8B6A,0x892C,0x88EE +.value 0x83B0,0x8272,0x8034,0x81F6,0x84B8,0x857A,0x873C,0x86FE +.value 0xA9C0,0xA802,0xAA44,0xAB86,0xAEC8,0xAF0A,0xAD4C,0xAC8E +.value 0xA7D0,0xA612,0xA454,0xA596,0xA0D8,0xA11A,0xA35C,0xA29E +.value 0xB5E0,0xB422,0xB664,0xB7A6,0xB2E8,0xB32A,0xB16C,0xB0AE +.value 0xBBF0,0xBA32,0xB874,0xB9B6,0xBCF8,0xBD3A,0xBF7C,0xBEBE + +.byte 71,72,65,83,72,32,102,111,114,32,120,56,54,95,54,52,44,32,67,82,89,80,84,79,71,65,77,83,32,98,121,32,60,97,112,112,114,111,64,111,112,101,110,115,115,108,46,111,114,103,62,0 +.align 64 +#if defined(HAVE_GNU_STACK) +.section .note.GNU-stack,"",%progbits +#endif diff --git a/ext/libressl/crypto/modes/ghash-macosx-x86_64.S b/ext/libressl/crypto/modes/ghash-macosx-x86_64.S new file mode 100644 index 0000000..e6840a7 --- /dev/null +++ b/ext/libressl/crypto/modes/ghash-macosx-x86_64.S @@ -0,0 +1,1027 @@ +#include "x86_arch.h" +.text + +.globl _gcm_gmult_4bit + +.p2align 4 +_gcm_gmult_4bit: + pushq %rbx + pushq %rbp + pushq %r12 +L$gmult_prologue: + + movzbq 15(%rdi),%r8 + leaq L$rem_4bit(%rip),%r11 + xorq %rax,%rax + xorq %rbx,%rbx + movb %r8b,%al + movb %r8b,%bl + shlb $4,%al + movq $14,%rcx + movq 8(%rsi,%rax,1),%r8 + movq (%rsi,%rax,1),%r9 + andb $240,%bl + movq %r8,%rdx + jmp L$oop1 + +.p2align 4 +L$oop1: + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + movb (%rdi,%rcx,1),%al + shrq $4,%r9 + xorq 8(%rsi,%rbx,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rbx,1),%r9 + movb %al,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + shlb $4,%al + xorq %r10,%r8 + decq %rcx + js L$break1 + + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rax,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rax,1),%r9 + andb $240,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + xorq %r10,%r8 + jmp L$oop1 + +.p2align 4 +L$break1: + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rax,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rax,1),%r9 + andb $240,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + xorq %r10,%r8 + + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rbx,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rbx,1),%r9 + xorq %r10,%r8 + xorq (%r11,%rdx,8),%r9 + + bswapq %r8 + bswapq %r9 + movq %r8,8(%rdi) + movq %r9,(%rdi) + + movq 16(%rsp),%rbx + leaq 24(%rsp),%rsp +L$gmult_epilogue: + retq + +.globl _gcm_ghash_4bit + +.p2align 4 +_gcm_ghash_4bit: + pushq %rbx + pushq %rbp + pushq %r12 + pushq %r13 + pushq %r14 + pushq %r15 + subq $280,%rsp +L$ghash_prologue: + movq %rdx,%r14 + movq %rcx,%r15 + subq $-128,%rsi + leaq 16+128(%rsp),%rbp + xorl %edx,%edx + movq 0+0-128(%rsi),%r8 + movq 0+8-128(%rsi),%rax + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq 16+0-128(%rsi),%r9 + shlb $4,%dl + movq 16+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,0(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,0(%rbp) + movq 32+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,0-128(%rbp) + movq 32+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,1(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,8(%rbp) + movq 48+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,8-128(%rbp) + movq 48+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,2(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,16(%rbp) + movq 64+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,16-128(%rbp) + movq 64+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,3(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,24(%rbp) + movq 80+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,24-128(%rbp) + movq 80+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,4(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,32(%rbp) + movq 96+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,32-128(%rbp) + movq 96+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,5(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,40(%rbp) + movq 112+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,40-128(%rbp) + movq 112+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,6(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,48(%rbp) + movq 128+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,48-128(%rbp) + movq 128+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,7(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,56(%rbp) + movq 144+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,56-128(%rbp) + movq 144+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,8(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,64(%rbp) + movq 160+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,64-128(%rbp) + movq 160+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,9(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,72(%rbp) + movq 176+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,72-128(%rbp) + movq 176+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,10(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,80(%rbp) + movq 192+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,80-128(%rbp) + movq 192+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,11(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,88(%rbp) + movq 208+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,88-128(%rbp) + movq 208+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,12(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,96(%rbp) + movq 224+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,96-128(%rbp) + movq 224+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,13(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,104(%rbp) + movq 240+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,104-128(%rbp) + movq 240+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,14(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,112(%rbp) + shlb $4,%dl + movq %rax,112-128(%rbp) + shlq $60,%r10 + movb %dl,15(%rsp) + orq %r10,%rbx + movq %r9,120(%rbp) + movq %rbx,120-128(%rbp) + addq $-128,%rsi + movq 8(%rdi),%r8 + movq 0(%rdi),%r9 + addq %r14,%r15 + leaq L$rem_8bit(%rip),%r11 + jmp L$outer_loop +.p2align 4 +L$outer_loop: + xorq (%r14),%r9 + movq 8(%r14),%rdx + leaq 16(%r14),%r14 + xorq %r8,%rdx + movq %r9,(%rdi) + movq %rdx,8(%rdi) + shrq $32,%rdx + xorq %rax,%rax + roll $8,%edx + movb %dl,%al + movzbl %dl,%ebx + shlb $4,%al + shrl $4,%ebx + roll $8,%edx + movq 8(%rsi,%rax,1),%r8 + movq (%rsi,%rax,1),%r9 + movb %dl,%al + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + xorq %r8,%r12 + movq %r9,%r10 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 8(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 4(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 0(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + andl $240,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl -4(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + movzwq (%r11,%r12,2),%r12 + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + shlq $48,%r12 + xorq %r10,%r8 + xorq %r12,%r9 + movzbq %r8b,%r13 + shrq $4,%r8 + movq %r9,%r10 + shlb $4,%r13b + shrq $4,%r9 + xorq 8(%rsi,%rcx,1),%r8 + movzwq (%r11,%r13,2),%r13 + shlq $60,%r10 + xorq (%rsi,%rcx,1),%r9 + xorq %r10,%r8 + shlq $48,%r13 + bswapq %r8 + xorq %r13,%r9 + bswapq %r9 + cmpq %r15,%r14 + jb L$outer_loop + movq %r8,8(%rdi) + movq %r9,(%rdi) + + leaq 280(%rsp),%rsi + movq 0(%rsi),%r15 + movq 8(%rsi),%r14 + movq 16(%rsi),%r13 + movq 24(%rsi),%r12 + movq 32(%rsi),%rbp + movq 40(%rsi),%rbx + leaq 48(%rsi),%rsp +L$ghash_epilogue: + retq + +.globl _gcm_init_clmul + +.p2align 4 +_gcm_init_clmul: + movdqu (%rsi),%xmm2 + pshufd $78,%xmm2,%xmm2 + + + pshufd $255,%xmm2,%xmm4 + movdqa %xmm2,%xmm3 + psllq $1,%xmm2 + pxor %xmm5,%xmm5 + psrlq $63,%xmm3 + pcmpgtd %xmm4,%xmm5 + pslldq $8,%xmm3 + por %xmm3,%xmm2 + + + pand L$0x1c2_polynomial(%rip),%xmm5 + pxor %xmm5,%xmm2 + + + movdqa %xmm2,%xmm0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + movdqu %xmm2,(%rdi) + movdqu %xmm0,16(%rdi) + retq + +.globl _gcm_gmult_clmul + +.p2align 4 +_gcm_gmult_clmul: + movdqu (%rdi),%xmm0 + movdqa L$bswap_mask(%rip),%xmm5 + movdqu (%rsi),%xmm2 +.byte 102,15,56,0,197 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 +.byte 102,15,56,0,197 + movdqu %xmm0,(%rdi) + retq + +.globl _gcm_ghash_clmul + +.p2align 4 +_gcm_ghash_clmul: + movdqa L$bswap_mask(%rip),%xmm5 + + movdqu (%rdi),%xmm0 + movdqu (%rsi),%xmm2 +.byte 102,15,56,0,197 + + subq $16,%rcx + jz L$odd_tail + + movdqu 16(%rsi),%xmm8 + + + + + + movdqu (%rdx),%xmm3 + movdqu 16(%rdx),%xmm6 +.byte 102,15,56,0,221 +.byte 102,15,56,0,245 + pxor %xmm3,%xmm0 + movdqa %xmm6,%xmm7 + pshufd $78,%xmm6,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm6,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,242,0 +.byte 102,15,58,68,250,17 +.byte 102,15,58,68,220,0 + pxor %xmm6,%xmm3 + pxor %xmm7,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm7 + pxor %xmm4,%xmm6 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm8,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm8,%xmm4 + + leaq 32(%rdx),%rdx + subq $32,%rcx + jbe L$even_tail + +L$mod_loop: +.byte 102,65,15,58,68,192,0 +.byte 102,65,15,58,68,200,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + movdqu (%rdx),%xmm3 + pxor %xmm6,%xmm0 + pxor %xmm7,%xmm1 + + movdqu 16(%rdx),%xmm6 +.byte 102,15,56,0,221 +.byte 102,15,56,0,245 + + movdqa %xmm6,%xmm7 + pshufd $78,%xmm6,%xmm9 + pshufd $78,%xmm2,%xmm10 + pxor %xmm6,%xmm9 + pxor %xmm2,%xmm10 + pxor %xmm3,%xmm1 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 +.byte 102,15,58,68,242,0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + +.byte 102,15,58,68,250,17 + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + +.byte 102,69,15,58,68,202,0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm8,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm8,%xmm4 + + pxor %xmm6,%xmm9 + pxor %xmm7,%xmm9 + movdqa %xmm9,%xmm10 + psrldq $8,%xmm9 + pslldq $8,%xmm10 + pxor %xmm9,%xmm7 + pxor %xmm10,%xmm6 + + leaq 32(%rdx),%rdx + subq $32,%rcx + ja L$mod_loop + +L$even_tail: +.byte 102,65,15,58,68,192,0 +.byte 102,65,15,58,68,200,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + pxor %xmm6,%xmm0 + pxor %xmm7,%xmm1 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + testq %rcx,%rcx + jnz L$done + +L$odd_tail: + movdqu (%rdx),%xmm3 +.byte 102,15,56,0,221 + pxor %xmm3,%xmm0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 +L$done: +.byte 102,15,56,0,197 + movdqu %xmm0,(%rdi) + retq +L$SEH_end_gcm_ghash_clmul: + +.p2align 6 +L$bswap_mask: +.byte 15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0 +L$0x1c2_polynomial: +.byte 1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xc2 +.p2align 6 + +L$rem_4bit: +.long 0,0,0,471859200,0,943718400,0,610271232 +.long 0,1887436800,0,1822425088,0,1220542464,0,1423966208 +.long 0,3774873600,0,4246732800,0,3644850176,0,3311403008 +.long 0,2441084928,0,2376073216,0,2847932416,0,3051356160 + +L$rem_8bit: +.value 0x0000,0x01C2,0x0384,0x0246,0x0708,0x06CA,0x048C,0x054E +.value 0x0E10,0x0FD2,0x0D94,0x0C56,0x0918,0x08DA,0x0A9C,0x0B5E +.value 0x1C20,0x1DE2,0x1FA4,0x1E66,0x1B28,0x1AEA,0x18AC,0x196E +.value 0x1230,0x13F2,0x11B4,0x1076,0x1538,0x14FA,0x16BC,0x177E +.value 0x3840,0x3982,0x3BC4,0x3A06,0x3F48,0x3E8A,0x3CCC,0x3D0E +.value 0x3650,0x3792,0x35D4,0x3416,0x3158,0x309A,0x32DC,0x331E +.value 0x2460,0x25A2,0x27E4,0x2626,0x2368,0x22AA,0x20EC,0x212E +.value 0x2A70,0x2BB2,0x29F4,0x2836,0x2D78,0x2CBA,0x2EFC,0x2F3E +.value 0x7080,0x7142,0x7304,0x72C6,0x7788,0x764A,0x740C,0x75CE +.value 0x7E90,0x7F52,0x7D14,0x7CD6,0x7998,0x785A,0x7A1C,0x7BDE +.value 0x6CA0,0x6D62,0x6F24,0x6EE6,0x6BA8,0x6A6A,0x682C,0x69EE +.value 0x62B0,0x6372,0x6134,0x60F6,0x65B8,0x647A,0x663C,0x67FE +.value 0x48C0,0x4902,0x4B44,0x4A86,0x4FC8,0x4E0A,0x4C4C,0x4D8E +.value 0x46D0,0x4712,0x4554,0x4496,0x41D8,0x401A,0x425C,0x439E +.value 0x54E0,0x5522,0x5764,0x56A6,0x53E8,0x522A,0x506C,0x51AE +.value 0x5AF0,0x5B32,0x5974,0x58B6,0x5DF8,0x5C3A,0x5E7C,0x5FBE +.value 0xE100,0xE0C2,0xE284,0xE346,0xE608,0xE7CA,0xE58C,0xE44E +.value 0xEF10,0xEED2,0xEC94,0xED56,0xE818,0xE9DA,0xEB9C,0xEA5E +.value 0xFD20,0xFCE2,0xFEA4,0xFF66,0xFA28,0xFBEA,0xF9AC,0xF86E +.value 0xF330,0xF2F2,0xF0B4,0xF176,0xF438,0xF5FA,0xF7BC,0xF67E +.value 0xD940,0xD882,0xDAC4,0xDB06,0xDE48,0xDF8A,0xDDCC,0xDC0E +.value 0xD750,0xD692,0xD4D4,0xD516,0xD058,0xD19A,0xD3DC,0xD21E +.value 0xC560,0xC4A2,0xC6E4,0xC726,0xC268,0xC3AA,0xC1EC,0xC02E +.value 0xCB70,0xCAB2,0xC8F4,0xC936,0xCC78,0xCDBA,0xCFFC,0xCE3E +.value 0x9180,0x9042,0x9204,0x93C6,0x9688,0x974A,0x950C,0x94CE +.value 0x9F90,0x9E52,0x9C14,0x9DD6,0x9898,0x995A,0x9B1C,0x9ADE +.value 0x8DA0,0x8C62,0x8E24,0x8FE6,0x8AA8,0x8B6A,0x892C,0x88EE +.value 0x83B0,0x8272,0x8034,0x81F6,0x84B8,0x857A,0x873C,0x86FE +.value 0xA9C0,0xA802,0xAA44,0xAB86,0xAEC8,0xAF0A,0xAD4C,0xAC8E +.value 0xA7D0,0xA612,0xA454,0xA596,0xA0D8,0xA11A,0xA35C,0xA29E +.value 0xB5E0,0xB422,0xB664,0xB7A6,0xB2E8,0xB32A,0xB16C,0xB0AE +.value 0xBBF0,0xBA32,0xB874,0xB9B6,0xBCF8,0xBD3A,0xBF7C,0xBEBE + +.byte 71,72,65,83,72,32,102,111,114,32,120,56,54,95,54,52,44,32,67,82,89,80,84,79,71,65,77,83,32,98,121,32,60,97,112,112,114,111,64,111,112,101,110,115,115,108,46,111,114,103,62,0 +.p2align 6 diff --git a/ext/libressl/crypto/modes/ghash-masm-x86_64.S b/ext/libressl/crypto/modes/ghash-masm-x86_64.S new file mode 100644 index 0000000..ffdc1b5 --- /dev/null +++ b/ext/libressl/crypto/modes/ghash-masm-x86_64.S @@ -0,0 +1,1256 @@ +; 1 "crypto/modes/ghash-masm-x86_64.S.tmp" +; 1 "" 1 +; 1 "" 3 +; 340 "" 3 +; 1 "" 1 +; 1 "" 2 +; 1 "crypto/modes/ghash-masm-x86_64.S.tmp" 2 +OPTION DOTNAME + +; 1 "./crypto/x86_arch.h" 1 + + +; 16 "./crypto/x86_arch.h" + + + + + + + + + +; 40 "./crypto/x86_arch.h" + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +; 3 "crypto/modes/ghash-masm-x86_64.S.tmp" 2 +.text$ SEGMENT ALIGN(64) 'CODE' + +PUBLIC gcm_gmult_4bit + +ALIGN 16 +gcm_gmult_4bit PROC PUBLIC + mov QWORD PTR[8+rsp],rdi ;WIN64 prologue + mov QWORD PTR[16+rsp],rsi + mov rax,rsp +$L$SEH_begin_gcm_gmult_4bit:: + mov rdi,rcx + mov rsi,rdx + + + push rbx + push rbp + push r12 +$L$gmult_prologue:: + + movzx r8,BYTE PTR[15+rdi] + lea r11,QWORD PTR[$L$rem_4bit] + xor rax,rax + xor rbx,rbx + mov al,r8b + mov bl,r8b + shl al,4 + mov rcx,14 + mov r8,QWORD PTR[8+rax*1+rsi] + mov r9,QWORD PTR[rax*1+rsi] + and bl,0f0h + mov rdx,r8 + jmp $L$oop1 + +ALIGN 16 +$L$oop1:: + shr r8,4 + and rdx,0fh + mov r10,r9 + mov al,BYTE PTR[rcx*1+rdi] + shr r9,4 + xor r8,QWORD PTR[8+rbx*1+rsi] + shl r10,60 + xor r9,QWORD PTR[rbx*1+rsi] + mov bl,al + xor r9,QWORD PTR[rdx*8+r11] + mov rdx,r8 + shl al,4 + xor r8,r10 + dec rcx + js $L$break1 + + shr r8,4 + and rdx,0fh + mov r10,r9 + shr r9,4 + xor r8,QWORD PTR[8+rax*1+rsi] + shl r10,60 + xor r9,QWORD PTR[rax*1+rsi] + and bl,0f0h + xor r9,QWORD PTR[rdx*8+r11] + mov rdx,r8 + xor r8,r10 + jmp $L$oop1 + +ALIGN 16 +$L$break1:: + shr r8,4 + and rdx,0fh + mov r10,r9 + shr r9,4 + xor r8,QWORD PTR[8+rax*1+rsi] + shl r10,60 + xor r9,QWORD PTR[rax*1+rsi] + and bl,0f0h + xor r9,QWORD PTR[rdx*8+r11] + mov rdx,r8 + xor r8,r10 + + shr r8,4 + and rdx,0fh + mov r10,r9 + shr r9,4 + xor r8,QWORD PTR[8+rbx*1+rsi] + shl r10,60 + xor r9,QWORD PTR[rbx*1+rsi] + xor r8,r10 + xor r9,QWORD PTR[rdx*8+r11] + + bswap r8 + bswap r9 + mov QWORD PTR[8+rdi],r8 + mov QWORD PTR[rdi],r9 + + mov rbx,QWORD PTR[16+rsp] + lea rsp,QWORD PTR[24+rsp] +$L$gmult_epilogue:: + mov rdi,QWORD PTR[8+rsp] ;WIN64 epilogue + mov rsi,QWORD PTR[16+rsp] + DB 0F3h,0C3h ;repret +$L$SEH_end_gcm_gmult_4bit:: +gcm_gmult_4bit ENDP +PUBLIC gcm_ghash_4bit + +ALIGN 16 +gcm_ghash_4bit PROC PUBLIC + mov QWORD PTR[8+rsp],rdi ;WIN64 prologue + mov QWORD PTR[16+rsp],rsi + mov rax,rsp +$L$SEH_begin_gcm_ghash_4bit:: + mov rdi,rcx + mov rsi,rdx + mov rdx,r8 + mov rcx,r9 + + + push rbx + push rbp + push r12 + push r13 + push r14 + push r15 + sub rsp,280 +$L$ghash_prologue:: + mov r14,rdx + mov r15,rcx + sub rsi,-128 + lea rbp,QWORD PTR[((16+128))+rsp] + xor edx,edx + mov r8,QWORD PTR[((0+0-128))+rsi] + mov rax,QWORD PTR[((0+8-128))+rsi] + mov dl,al + shr rax,4 + mov r10,r8 + shr r8,4 + mov r9,QWORD PTR[((16+0-128))+rsi] + shl dl,4 + mov rbx,QWORD PTR[((16+8-128))+rsi] + shl r10,60 + mov BYTE PTR[rsp],dl + or rax,r10 + mov dl,bl + shr rbx,4 + mov r10,r9 + shr r9,4 + mov QWORD PTR[rbp],r8 + mov r8,QWORD PTR[((32+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((0-128))+rbp],rax + mov rax,QWORD PTR[((32+8-128))+rsi] + shl r10,60 + mov BYTE PTR[1+rsp],dl + or rbx,r10 + mov dl,al + shr rax,4 + mov r10,r8 + shr r8,4 + mov QWORD PTR[8+rbp],r9 + mov r9,QWORD PTR[((48+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((8-128))+rbp],rbx + mov rbx,QWORD PTR[((48+8-128))+rsi] + shl r10,60 + mov BYTE PTR[2+rsp],dl + or rax,r10 + mov dl,bl + shr rbx,4 + mov r10,r9 + shr r9,4 + mov QWORD PTR[16+rbp],r8 + mov r8,QWORD PTR[((64+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((16-128))+rbp],rax + mov rax,QWORD PTR[((64+8-128))+rsi] + shl r10,60 + mov BYTE PTR[3+rsp],dl + or rbx,r10 + mov dl,al + shr rax,4 + mov r10,r8 + shr r8,4 + mov QWORD PTR[24+rbp],r9 + mov r9,QWORD PTR[((80+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((24-128))+rbp],rbx + mov rbx,QWORD PTR[((80+8-128))+rsi] + shl r10,60 + mov BYTE PTR[4+rsp],dl + or rax,r10 + mov dl,bl + shr rbx,4 + mov r10,r9 + shr r9,4 + mov QWORD PTR[32+rbp],r8 + mov r8,QWORD PTR[((96+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((32-128))+rbp],rax + mov rax,QWORD PTR[((96+8-128))+rsi] + shl r10,60 + mov BYTE PTR[5+rsp],dl + or rbx,r10 + mov dl,al + shr rax,4 + mov r10,r8 + shr r8,4 + mov QWORD PTR[40+rbp],r9 + mov r9,QWORD PTR[((112+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((40-128))+rbp],rbx + mov rbx,QWORD PTR[((112+8-128))+rsi] + shl r10,60 + mov BYTE PTR[6+rsp],dl + or rax,r10 + mov dl,bl + shr rbx,4 + mov r10,r9 + shr r9,4 + mov QWORD PTR[48+rbp],r8 + mov r8,QWORD PTR[((128+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((48-128))+rbp],rax + mov rax,QWORD PTR[((128+8-128))+rsi] + shl r10,60 + mov BYTE PTR[7+rsp],dl + or rbx,r10 + mov dl,al + shr rax,4 + mov r10,r8 + shr r8,4 + mov QWORD PTR[56+rbp],r9 + mov r9,QWORD PTR[((144+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((56-128))+rbp],rbx + mov rbx,QWORD PTR[((144+8-128))+rsi] + shl r10,60 + mov BYTE PTR[8+rsp],dl + or rax,r10 + mov dl,bl + shr rbx,4 + mov r10,r9 + shr r9,4 + mov QWORD PTR[64+rbp],r8 + mov r8,QWORD PTR[((160+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((64-128))+rbp],rax + mov rax,QWORD PTR[((160+8-128))+rsi] + shl r10,60 + mov BYTE PTR[9+rsp],dl + or rbx,r10 + mov dl,al + shr rax,4 + mov r10,r8 + shr r8,4 + mov QWORD PTR[72+rbp],r9 + mov r9,QWORD PTR[((176+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((72-128))+rbp],rbx + mov rbx,QWORD PTR[((176+8-128))+rsi] + shl r10,60 + mov BYTE PTR[10+rsp],dl + or rax,r10 + mov dl,bl + shr rbx,4 + mov r10,r9 + shr r9,4 + mov QWORD PTR[80+rbp],r8 + mov r8,QWORD PTR[((192+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((80-128))+rbp],rax + mov rax,QWORD PTR[((192+8-128))+rsi] + shl r10,60 + mov BYTE PTR[11+rsp],dl + or rbx,r10 + mov dl,al + shr rax,4 + mov r10,r8 + shr r8,4 + mov QWORD PTR[88+rbp],r9 + mov r9,QWORD PTR[((208+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((88-128))+rbp],rbx + mov rbx,QWORD PTR[((208+8-128))+rsi] + shl r10,60 + mov BYTE PTR[12+rsp],dl + or rax,r10 + mov dl,bl + shr rbx,4 + mov r10,r9 + shr r9,4 + mov QWORD PTR[96+rbp],r8 + mov r8,QWORD PTR[((224+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((96-128))+rbp],rax + mov rax,QWORD PTR[((224+8-128))+rsi] + shl r10,60 + mov BYTE PTR[13+rsp],dl + or rbx,r10 + mov dl,al + shr rax,4 + mov r10,r8 + shr r8,4 + mov QWORD PTR[104+rbp],r9 + mov r9,QWORD PTR[((240+0-128))+rsi] + shl dl,4 + mov QWORD PTR[((104-128))+rbp],rbx + mov rbx,QWORD PTR[((240+8-128))+rsi] + shl r10,60 + mov BYTE PTR[14+rsp],dl + or rax,r10 + mov dl,bl + shr rbx,4 + mov r10,r9 + shr r9,4 + mov QWORD PTR[112+rbp],r8 + shl dl,4 + mov QWORD PTR[((112-128))+rbp],rax + shl r10,60 + mov BYTE PTR[15+rsp],dl + or rbx,r10 + mov QWORD PTR[120+rbp],r9 + mov QWORD PTR[((120-128))+rbp],rbx + add rsi,-128 + mov r8,QWORD PTR[8+rdi] + mov r9,QWORD PTR[rdi] + add r15,r14 + lea r11,QWORD PTR[$L$rem_8bit] + jmp $L$outer_loop +ALIGN 16 +$L$outer_loop:: + xor r9,QWORD PTR[r14] + mov rdx,QWORD PTR[8+r14] + lea r14,QWORD PTR[16+r14] + xor rdx,r8 + mov QWORD PTR[rdi],r9 + mov QWORD PTR[8+rdi],rdx + shr rdx,32 + xor rax,rax + rol edx,8 + mov al,dl + movzx ebx,dl + shl al,4 + shr ebx,4 + rol edx,8 + mov r8,QWORD PTR[8+rax*1+rsi] + mov r9,QWORD PTR[rax*1+rsi] + mov al,dl + movzx ecx,dl + shl al,4 + movzx r12,BYTE PTR[rbx*1+rsp] + shr ecx,4 + xor r12,r8 + mov r10,r9 + shr r8,8 + movzx r12,r12b + shr r9,8 + xor r8,QWORD PTR[((-128))+rbx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rbx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r12,WORD PTR[r12*2+r11] + movzx ebx,dl + shl al,4 + movzx r13,BYTE PTR[rcx*1+rsp] + shr ebx,4 + shl r12,48 + xor r13,r8 + mov r10,r9 + xor r9,r12 + shr r8,8 + movzx r13,r13b + shr r9,8 + xor r8,QWORD PTR[((-128))+rcx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rcx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r13,WORD PTR[r13*2+r11] + movzx ecx,dl + shl al,4 + movzx r12,BYTE PTR[rbx*1+rsp] + shr ecx,4 + shl r13,48 + xor r12,r8 + mov r10,r9 + xor r9,r13 + shr r8,8 + movzx r12,r12b + mov edx,DWORD PTR[8+rdi] + shr r9,8 + xor r8,QWORD PTR[((-128))+rbx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rbx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r12,WORD PTR[r12*2+r11] + movzx ebx,dl + shl al,4 + movzx r13,BYTE PTR[rcx*1+rsp] + shr ebx,4 + shl r12,48 + xor r13,r8 + mov r10,r9 + xor r9,r12 + shr r8,8 + movzx r13,r13b + shr r9,8 + xor r8,QWORD PTR[((-128))+rcx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rcx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r13,WORD PTR[r13*2+r11] + movzx ecx,dl + shl al,4 + movzx r12,BYTE PTR[rbx*1+rsp] + shr ecx,4 + shl r13,48 + xor r12,r8 + mov r10,r9 + xor r9,r13 + shr r8,8 + movzx r12,r12b + shr r9,8 + xor r8,QWORD PTR[((-128))+rbx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rbx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r12,WORD PTR[r12*2+r11] + movzx ebx,dl + shl al,4 + movzx r13,BYTE PTR[rcx*1+rsp] + shr ebx,4 + shl r12,48 + xor r13,r8 + mov r10,r9 + xor r9,r12 + shr r8,8 + movzx r13,r13b + shr r9,8 + xor r8,QWORD PTR[((-128))+rcx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rcx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r13,WORD PTR[r13*2+r11] + movzx ecx,dl + shl al,4 + movzx r12,BYTE PTR[rbx*1+rsp] + shr ecx,4 + shl r13,48 + xor r12,r8 + mov r10,r9 + xor r9,r13 + shr r8,8 + movzx r12,r12b + mov edx,DWORD PTR[4+rdi] + shr r9,8 + xor r8,QWORD PTR[((-128))+rbx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rbx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r12,WORD PTR[r12*2+r11] + movzx ebx,dl + shl al,4 + movzx r13,BYTE PTR[rcx*1+rsp] + shr ebx,4 + shl r12,48 + xor r13,r8 + mov r10,r9 + xor r9,r12 + shr r8,8 + movzx r13,r13b + shr r9,8 + xor r8,QWORD PTR[((-128))+rcx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rcx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r13,WORD PTR[r13*2+r11] + movzx ecx,dl + shl al,4 + movzx r12,BYTE PTR[rbx*1+rsp] + shr ecx,4 + shl r13,48 + xor r12,r8 + mov r10,r9 + xor r9,r13 + shr r8,8 + movzx r12,r12b + shr r9,8 + xor r8,QWORD PTR[((-128))+rbx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rbx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r12,WORD PTR[r12*2+r11] + movzx ebx,dl + shl al,4 + movzx r13,BYTE PTR[rcx*1+rsp] + shr ebx,4 + shl r12,48 + xor r13,r8 + mov r10,r9 + xor r9,r12 + shr r8,8 + movzx r13,r13b + shr r9,8 + xor r8,QWORD PTR[((-128))+rcx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rcx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r13,WORD PTR[r13*2+r11] + movzx ecx,dl + shl al,4 + movzx r12,BYTE PTR[rbx*1+rsp] + shr ecx,4 + shl r13,48 + xor r12,r8 + mov r10,r9 + xor r9,r13 + shr r8,8 + movzx r12,r12b + mov edx,DWORD PTR[rdi] + shr r9,8 + xor r8,QWORD PTR[((-128))+rbx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rbx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r12,WORD PTR[r12*2+r11] + movzx ebx,dl + shl al,4 + movzx r13,BYTE PTR[rcx*1+rsp] + shr ebx,4 + shl r12,48 + xor r13,r8 + mov r10,r9 + xor r9,r12 + shr r8,8 + movzx r13,r13b + shr r9,8 + xor r8,QWORD PTR[((-128))+rcx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rcx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r13,WORD PTR[r13*2+r11] + movzx ecx,dl + shl al,4 + movzx r12,BYTE PTR[rbx*1+rsp] + shr ecx,4 + shl r13,48 + xor r12,r8 + mov r10,r9 + xor r9,r13 + shr r8,8 + movzx r12,r12b + shr r9,8 + xor r8,QWORD PTR[((-128))+rbx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rbx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r12,WORD PTR[r12*2+r11] + movzx ebx,dl + shl al,4 + movzx r13,BYTE PTR[rcx*1+rsp] + shr ebx,4 + shl r12,48 + xor r13,r8 + mov r10,r9 + xor r9,r12 + shr r8,8 + movzx r13,r13b + shr r9,8 + xor r8,QWORD PTR[((-128))+rcx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rcx*8+rbp] + rol edx,8 + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + mov al,dl + xor r8,r10 + movzx r13,WORD PTR[r13*2+r11] + movzx ecx,dl + shl al,4 + movzx r12,BYTE PTR[rbx*1+rsp] + and ecx,240 + shl r13,48 + xor r12,r8 + mov r10,r9 + xor r9,r13 + shr r8,8 + movzx r12,r12b + mov edx,DWORD PTR[((-4))+rdi] + shr r9,8 + xor r8,QWORD PTR[((-128))+rbx*8+rbp] + shl r10,56 + xor r9,QWORD PTR[rbx*8+rbp] + movzx r12,WORD PTR[r12*2+r11] + xor r8,QWORD PTR[8+rax*1+rsi] + xor r9,QWORD PTR[rax*1+rsi] + shl r12,48 + xor r8,r10 + xor r9,r12 + movzx r13,r8b + shr r8,4 + mov r10,r9 + shl r13b,4 + shr r9,4 + xor r8,QWORD PTR[8+rcx*1+rsi] + movzx r13,WORD PTR[r13*2+r11] + shl r10,60 + xor r9,QWORD PTR[rcx*1+rsi] + xor r8,r10 + shl r13,48 + bswap r8 + xor r9,r13 + bswap r9 + cmp r14,r15 + jb $L$outer_loop + mov QWORD PTR[8+rdi],r8 + mov QWORD PTR[rdi],r9 + + lea rsi,QWORD PTR[280+rsp] + mov r15,QWORD PTR[rsi] + mov r14,QWORD PTR[8+rsi] + mov r13,QWORD PTR[16+rsi] + mov r12,QWORD PTR[24+rsi] + mov rbp,QWORD PTR[32+rsi] + mov rbx,QWORD PTR[40+rsi] + lea rsp,QWORD PTR[48+rsi] +$L$ghash_epilogue:: + mov rdi,QWORD PTR[8+rsp] ;WIN64 epilogue + mov rsi,QWORD PTR[16+rsp] + DB 0F3h,0C3h ;repret +$L$SEH_end_gcm_ghash_4bit:: +gcm_ghash_4bit ENDP +PUBLIC gcm_init_clmul + +ALIGN 16 +gcm_init_clmul PROC PUBLIC + movdqu xmm2,XMMWORD PTR[rdx] + pshufd xmm2,xmm2,78 + + + pshufd xmm4,xmm2,255 + movdqa xmm3,xmm2 + psllq xmm2,1 + pxor xmm5,xmm5 + psrlq xmm3,63 + pcmpgtd xmm5,xmm4 + pslldq xmm3,8 + por xmm2,xmm3 + + + pand xmm5,XMMWORD PTR[$L$0x1c2_polynomial] + pxor xmm2,xmm5 + + + movdqa xmm0,xmm2 + movdqa xmm1,xmm0 + pshufd xmm3,xmm0,78 + pshufd xmm4,xmm2,78 + pxor xmm3,xmm0 + pxor xmm4,xmm2 +DB 102,15,58,68,194,0 +DB 102,15,58,68,202,17 +DB 102,15,58,68,220,0 + pxor xmm3,xmm0 + pxor xmm3,xmm1 + + movdqa xmm4,xmm3 + psrldq xmm3,8 + pslldq xmm4,8 + pxor xmm1,xmm3 + pxor xmm0,xmm4 + + movdqa xmm3,xmm0 + psllq xmm0,1 + pxor xmm0,xmm3 + psllq xmm0,5 + pxor xmm0,xmm3 + psllq xmm0,57 + movdqa xmm4,xmm0 + pslldq xmm0,8 + psrldq xmm4,8 + pxor xmm0,xmm3 + pxor xmm1,xmm4 + + + movdqa xmm4,xmm0 + psrlq xmm0,5 + pxor xmm0,xmm4 + psrlq xmm0,1 + pxor xmm0,xmm4 + pxor xmm4,xmm1 + psrlq xmm0,1 + pxor xmm0,xmm4 + movdqu XMMWORD PTR[rcx],xmm2 + movdqu XMMWORD PTR[16+rcx],xmm0 + DB 0F3h,0C3h ;repret +gcm_init_clmul ENDP +PUBLIC gcm_gmult_clmul + +ALIGN 16 +gcm_gmult_clmul PROC PUBLIC + movdqu xmm0,XMMWORD PTR[rcx] + movdqa xmm5,XMMWORD PTR[$L$bswap_mask] + movdqu xmm2,XMMWORD PTR[rdx] +DB 102,15,56,0,197 + movdqa xmm1,xmm0 + pshufd xmm3,xmm0,78 + pshufd xmm4,xmm2,78 + pxor xmm3,xmm0 + pxor xmm4,xmm2 +DB 102,15,58,68,194,0 +DB 102,15,58,68,202,17 +DB 102,15,58,68,220,0 + pxor xmm3,xmm0 + pxor xmm3,xmm1 + + movdqa xmm4,xmm3 + psrldq xmm3,8 + pslldq xmm4,8 + pxor xmm1,xmm3 + pxor xmm0,xmm4 + + movdqa xmm3,xmm0 + psllq xmm0,1 + pxor xmm0,xmm3 + psllq xmm0,5 + pxor xmm0,xmm3 + psllq xmm0,57 + movdqa xmm4,xmm0 + pslldq xmm0,8 + psrldq xmm4,8 + pxor xmm0,xmm3 + pxor xmm1,xmm4 + + + movdqa xmm4,xmm0 + psrlq xmm0,5 + pxor xmm0,xmm4 + psrlq xmm0,1 + pxor xmm0,xmm4 + pxor xmm4,xmm1 + psrlq xmm0,1 + pxor xmm0,xmm4 +DB 102,15,56,0,197 + movdqu XMMWORD PTR[rcx],xmm0 + DB 0F3h,0C3h ;repret +gcm_gmult_clmul ENDP +PUBLIC gcm_ghash_clmul + +ALIGN 16 +gcm_ghash_clmul PROC PUBLIC +$L$SEH_begin_gcm_ghash_clmul:: + +DB 048h,083h,0ech,058h +DB 00fh,029h,034h,024h +DB 00fh,029h,07ch,024h,010h +DB 044h,00fh,029h,044h,024h,020h +DB 044h,00fh,029h,04ch,024h,030h +DB 044h,00fh,029h,054h,024h,040h + movdqa xmm5,XMMWORD PTR[$L$bswap_mask] + + movdqu xmm0,XMMWORD PTR[rcx] + movdqu xmm2,XMMWORD PTR[rdx] +DB 102,15,56,0,197 + + sub r9,010h + jz $L$odd_tail + + movdqu xmm8,XMMWORD PTR[16+rdx] + + + + + + movdqu xmm3,XMMWORD PTR[r8] + movdqu xmm6,XMMWORD PTR[16+r8] +DB 102,15,56,0,221 +DB 102,15,56,0,245 + pxor xmm0,xmm3 + movdqa xmm7,xmm6 + pshufd xmm3,xmm6,78 + pshufd xmm4,xmm2,78 + pxor xmm3,xmm6 + pxor xmm4,xmm2 +DB 102,15,58,68,242,0 +DB 102,15,58,68,250,17 +DB 102,15,58,68,220,0 + pxor xmm3,xmm6 + pxor xmm3,xmm7 + + movdqa xmm4,xmm3 + psrldq xmm3,8 + pslldq xmm4,8 + pxor xmm7,xmm3 + pxor xmm6,xmm4 + movdqa xmm1,xmm0 + pshufd xmm3,xmm0,78 + pshufd xmm4,xmm8,78 + pxor xmm3,xmm0 + pxor xmm4,xmm8 + + lea r8,QWORD PTR[32+r8] + sub r9,020h + jbe $L$even_tail + +$L$mod_loop:: +DB 102,65,15,58,68,192,0 +DB 102,65,15,58,68,200,17 +DB 102,15,58,68,220,0 + pxor xmm3,xmm0 + pxor xmm3,xmm1 + + movdqa xmm4,xmm3 + psrldq xmm3,8 + pslldq xmm4,8 + pxor xmm1,xmm3 + pxor xmm0,xmm4 + movdqu xmm3,XMMWORD PTR[r8] + pxor xmm0,xmm6 + pxor xmm1,xmm7 + + movdqu xmm6,XMMWORD PTR[16+r8] +DB 102,15,56,0,221 +DB 102,15,56,0,245 + + movdqa xmm7,xmm6 + pshufd xmm9,xmm6,78 + pshufd xmm10,xmm2,78 + pxor xmm9,xmm6 + pxor xmm10,xmm2 + pxor xmm1,xmm3 + + movdqa xmm3,xmm0 + psllq xmm0,1 + pxor xmm0,xmm3 + psllq xmm0,5 + pxor xmm0,xmm3 +DB 102,15,58,68,242,0 + psllq xmm0,57 + movdqa xmm4,xmm0 + pslldq xmm0,8 + psrldq xmm4,8 + pxor xmm0,xmm3 + pxor xmm1,xmm4 + +DB 102,15,58,68,250,17 + movdqa xmm4,xmm0 + psrlq xmm0,5 + pxor xmm0,xmm4 + psrlq xmm0,1 + pxor xmm0,xmm4 + pxor xmm4,xmm1 + psrlq xmm0,1 + pxor xmm0,xmm4 + +DB 102,69,15,58,68,202,0 + movdqa xmm1,xmm0 + pshufd xmm3,xmm0,78 + pshufd xmm4,xmm8,78 + pxor xmm3,xmm0 + pxor xmm4,xmm8 + + pxor xmm9,xmm6 + pxor xmm9,xmm7 + movdqa xmm10,xmm9 + psrldq xmm9,8 + pslldq xmm10,8 + pxor xmm7,xmm9 + pxor xmm6,xmm10 + + lea r8,QWORD PTR[32+r8] + sub r9,020h + ja $L$mod_loop + +$L$even_tail:: +DB 102,65,15,58,68,192,0 +DB 102,65,15,58,68,200,17 +DB 102,15,58,68,220,0 + pxor xmm3,xmm0 + pxor xmm3,xmm1 + + movdqa xmm4,xmm3 + psrldq xmm3,8 + pslldq xmm4,8 + pxor xmm1,xmm3 + pxor xmm0,xmm4 + pxor xmm0,xmm6 + pxor xmm1,xmm7 + + movdqa xmm3,xmm0 + psllq xmm0,1 + pxor xmm0,xmm3 + psllq xmm0,5 + pxor xmm0,xmm3 + psllq xmm0,57 + movdqa xmm4,xmm0 + pslldq xmm0,8 + psrldq xmm4,8 + pxor xmm0,xmm3 + pxor xmm1,xmm4 + + + movdqa xmm4,xmm0 + psrlq xmm0,5 + pxor xmm0,xmm4 + psrlq xmm0,1 + pxor xmm0,xmm4 + pxor xmm4,xmm1 + psrlq xmm0,1 + pxor xmm0,xmm4 + test r9,r9 + jnz $L$done + +$L$odd_tail:: + movdqu xmm3,XMMWORD PTR[r8] +DB 102,15,56,0,221 + pxor xmm0,xmm3 + movdqa xmm1,xmm0 + pshufd xmm3,xmm0,78 + pshufd xmm4,xmm2,78 + pxor xmm3,xmm0 + pxor xmm4,xmm2 +DB 102,15,58,68,194,0 +DB 102,15,58,68,202,17 +DB 102,15,58,68,220,0 + pxor xmm3,xmm0 + pxor xmm3,xmm1 + + movdqa xmm4,xmm3 + psrldq xmm3,8 + pslldq xmm4,8 + pxor xmm1,xmm3 + pxor xmm0,xmm4 + + movdqa xmm3,xmm0 + psllq xmm0,1 + pxor xmm0,xmm3 + psllq xmm0,5 + pxor xmm0,xmm3 + psllq xmm0,57 + movdqa xmm4,xmm0 + pslldq xmm0,8 + psrldq xmm4,8 + pxor xmm0,xmm3 + pxor xmm1,xmm4 + + + movdqa xmm4,xmm0 + psrlq xmm0,5 + pxor xmm0,xmm4 + psrlq xmm0,1 + pxor xmm0,xmm4 + pxor xmm4,xmm1 + psrlq xmm0,1 + pxor xmm0,xmm4 +$L$done:: +DB 102,15,56,0,197 + movdqu XMMWORD PTR[rcx],xmm0 + movaps xmm6,XMMWORD PTR[rsp] + movaps xmm7,XMMWORD PTR[16+rsp] + movaps xmm8,XMMWORD PTR[32+rsp] + movaps xmm9,XMMWORD PTR[48+rsp] + movaps xmm10,XMMWORD PTR[64+rsp] + add rsp,058h + DB 0F3h,0C3h ;repret +$L$SEH_end_gcm_ghash_clmul:: +gcm_ghash_clmul ENDP +ALIGN 64 +$L$bswap_mask:: +DB 15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0 +$L$0x1c2_polynomial:: +DB 1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0c2h +ALIGN 64 + +$L$rem_4bit:: + DD 0,0,0,471859200,0,943718400,0,610271232 + DD 0,1887436800,0,1822425088,0,1220542464,0,1423966208 + DD 0,3774873600,0,4246732800,0,3644850176,0,3311403008 + DD 0,2441084928,0,2376073216,0,2847932416,0,3051356160 + +$L$rem_8bit:: + DW 00000h,001C2h,00384h,00246h,00708h,006CAh,0048Ch,0054Eh + DW 00E10h,00FD2h,00D94h,00C56h,00918h,008DAh,00A9Ch,00B5Eh + DW 01C20h,01DE2h,01FA4h,01E66h,01B28h,01AEAh,018ACh,0196Eh + DW 01230h,013F2h,011B4h,01076h,01538h,014FAh,016BCh,0177Eh + DW 03840h,03982h,03BC4h,03A06h,03F48h,03E8Ah,03CCCh,03D0Eh + DW 03650h,03792h,035D4h,03416h,03158h,0309Ah,032DCh,0331Eh + DW 02460h,025A2h,027E4h,02626h,02368h,022AAh,020ECh,0212Eh + DW 02A70h,02BB2h,029F4h,02836h,02D78h,02CBAh,02EFCh,02F3Eh + DW 07080h,07142h,07304h,072C6h,07788h,0764Ah,0740Ch,075CEh + DW 07E90h,07F52h,07D14h,07CD6h,07998h,0785Ah,07A1Ch,07BDEh + DW 06CA0h,06D62h,06F24h,06EE6h,06BA8h,06A6Ah,0682Ch,069EEh + DW 062B0h,06372h,06134h,060F6h,065B8h,0647Ah,0663Ch,067FEh + DW 048C0h,04902h,04B44h,04A86h,04FC8h,04E0Ah,04C4Ch,04D8Eh + DW 046D0h,04712h,04554h,04496h,041D8h,0401Ah,0425Ch,0439Eh + DW 054E0h,05522h,05764h,056A6h,053E8h,0522Ah,0506Ch,051AEh + DW 05AF0h,05B32h,05974h,058B6h,05DF8h,05C3Ah,05E7Ch,05FBEh + DW 0E100h,0E0C2h,0E284h,0E346h,0E608h,0E7CAh,0E58Ch,0E44Eh + DW 0EF10h,0EED2h,0EC94h,0ED56h,0E818h,0E9DAh,0EB9Ch,0EA5Eh + DW 0FD20h,0FCE2h,0FEA4h,0FF66h,0FA28h,0FBEAh,0F9ACh,0F86Eh + DW 0F330h,0F2F2h,0F0B4h,0F176h,0F438h,0F5FAh,0F7BCh,0F67Eh + DW 0D940h,0D882h,0DAC4h,0DB06h,0DE48h,0DF8Ah,0DDCCh,0DC0Eh + DW 0D750h,0D692h,0D4D4h,0D516h,0D058h,0D19Ah,0D3DCh,0D21Eh + DW 0C560h,0C4A2h,0C6E4h,0C726h,0C268h,0C3AAh,0C1ECh,0C02Eh + DW 0CB70h,0CAB2h,0C8F4h,0C936h,0CC78h,0CDBAh,0CFFCh,0CE3Eh + DW 09180h,09042h,09204h,093C6h,09688h,0974Ah,0950Ch,094CEh + DW 09F90h,09E52h,09C14h,09DD6h,09898h,0995Ah,09B1Ch,09ADEh + DW 08DA0h,08C62h,08E24h,08FE6h,08AA8h,08B6Ah,0892Ch,088EEh + DW 083B0h,08272h,08034h,081F6h,084B8h,0857Ah,0873Ch,086FEh + DW 0A9C0h,0A802h,0AA44h,0AB86h,0AEC8h,0AF0Ah,0AD4Ch,0AC8Eh + DW 0A7D0h,0A612h,0A454h,0A596h,0A0D8h,0A11Ah,0A35Ch,0A29Eh + DW 0B5E0h,0B422h,0B664h,0B7A6h,0B2E8h,0B32Ah,0B16Ch,0B0AEh + DW 0BBF0h,0BA32h,0B874h,0B9B6h,0BCF8h,0BD3Ah,0BF7Ch,0BEBEh + +DB 71,72,65,83,72,32,102,111,114,32,120,56,54,95,54,52 +DB 44,32,67,82,89,80,84,79,71,65,77,83,32,98,121,32 +DB 60,97,112,112,114,111,64,111,112,101,110,115,115,108,46,111 +DB 114,103,62,0 +ALIGN 64 +EXTERN __imp_RtlVirtualUnwind:NEAR + +ALIGN 16 +se_handler PROC PRIVATE + push rsi + push rdi + push rbx + push rbp + push r12 + push r13 + push r14 + push r15 + pushfq + sub rsp,64 + + mov rax,QWORD PTR[120+r8] + mov rbx,QWORD PTR[248+r8] + + mov rsi,QWORD PTR[8+r9] + mov r11,QWORD PTR[56+r9] + + mov r10d,DWORD PTR[r11] + lea r10,QWORD PTR[r10*1+rsi] + cmp rbx,r10 + jb $L$in_prologue + + mov rax,QWORD PTR[152+r8] + + mov r10d,DWORD PTR[4+r11] + lea r10,QWORD PTR[r10*1+rsi] + cmp rbx,r10 + jae $L$in_prologue + + lea rax,QWORD PTR[24+rax] + + mov rbx,QWORD PTR[((-8))+rax] + mov rbp,QWORD PTR[((-16))+rax] + mov r12,QWORD PTR[((-24))+rax] + mov QWORD PTR[144+r8],rbx + mov QWORD PTR[160+r8],rbp + mov QWORD PTR[216+r8],r12 + +$L$in_prologue:: + mov rdi,QWORD PTR[8+rax] + mov rsi,QWORD PTR[16+rax] + mov QWORD PTR[152+r8],rax + mov QWORD PTR[168+r8],rsi + mov QWORD PTR[176+r8],rdi + + mov rdi,QWORD PTR[40+r9] + mov rsi,r8 + mov ecx,154 + DD 0a548f3fch + + mov rsi,r9 + xor rcx,rcx + mov rdx,QWORD PTR[8+rsi] + mov r8,QWORD PTR[rsi] + mov r9,QWORD PTR[16+rsi] + mov r10,QWORD PTR[40+rsi] + lea r11,QWORD PTR[56+rsi] + lea r12,QWORD PTR[24+rsi] + mov QWORD PTR[32+rsp],r10 + mov QWORD PTR[40+rsp],r11 + mov QWORD PTR[48+rsp],r12 + mov QWORD PTR[56+rsp],rcx + call QWORD PTR[__imp_RtlVirtualUnwind] + + mov eax,1 + add rsp,64 + popfq + pop r15 + pop r14 + pop r13 + pop r12 + pop rbp + pop rbx + pop rdi + pop rsi + DB 0F3h,0C3h ;repret +se_handler ENDP + +.text$ ENDS +.pdata SEGMENT READONLY ALIGN(4) +ALIGN 4 + DD imagerel $L$SEH_begin_gcm_gmult_4bit + DD imagerel $L$SEH_end_gcm_gmult_4bit + DD imagerel $L$SEH_info_gcm_gmult_4bit + + DD imagerel $L$SEH_begin_gcm_ghash_4bit + DD imagerel $L$SEH_end_gcm_ghash_4bit + DD imagerel $L$SEH_info_gcm_ghash_4bit + + DD imagerel $L$SEH_begin_gcm_ghash_clmul + DD imagerel $L$SEH_end_gcm_ghash_clmul + DD imagerel $L$SEH_info_gcm_ghash_clmul + +.pdata ENDS +.xdata SEGMENT READONLY ALIGN(8) +ALIGN 8 +$L$SEH_info_gcm_gmult_4bit:: +DB 9,0,0,0 + DD imagerel se_handler + DD imagerel $L$gmult_prologue,imagerel $L$gmult_epilogue +$L$SEH_info_gcm_ghash_4bit:: +DB 9,0,0,0 + DD imagerel se_handler + DD imagerel $L$ghash_prologue,imagerel $L$ghash_epilogue +$L$SEH_info_gcm_ghash_clmul:: +DB 001h,01fh,00bh,000h +DB 01fh,0a8h,004h,000h +DB 019h,098h,003h,000h +DB 013h,088h,002h,000h +DB 00dh,078h,001h,000h +DB 008h,068h,000h,000h +DB 004h,0a2h,000h,000h + +.xdata ENDS +END + diff --git a/ext/libressl/crypto/modes/ghash-mingw64-x86_64.S b/ext/libressl/crypto/modes/ghash-mingw64-x86_64.S new file mode 100644 index 0000000..cd0823b --- /dev/null +++ b/ext/libressl/crypto/modes/ghash-mingw64-x86_64.S @@ -0,0 +1,1175 @@ +#include "x86_arch.h" +.text + +.globl gcm_gmult_4bit +.def gcm_gmult_4bit; .scl 2; .type 32; .endef +.p2align 4 +gcm_gmult_4bit: + movq %rdi,8(%rsp) + movq %rsi,16(%rsp) + movq %rsp,%rax +.LSEH_begin_gcm_gmult_4bit: + movq %rcx,%rdi + movq %rdx,%rsi + + pushq %rbx + pushq %rbp + pushq %r12 +.Lgmult_prologue: + + movzbq 15(%rdi),%r8 + leaq .Lrem_4bit(%rip),%r11 + xorq %rax,%rax + xorq %rbx,%rbx + movb %r8b,%al + movb %r8b,%bl + shlb $4,%al + movq $14,%rcx + movq 8(%rsi,%rax,1),%r8 + movq (%rsi,%rax,1),%r9 + andb $240,%bl + movq %r8,%rdx + jmp .Loop1 + +.p2align 4 +.Loop1: + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + movb (%rdi,%rcx,1),%al + shrq $4,%r9 + xorq 8(%rsi,%rbx,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rbx,1),%r9 + movb %al,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + shlb $4,%al + xorq %r10,%r8 + decq %rcx + js .Lbreak1 + + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rax,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rax,1),%r9 + andb $240,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + xorq %r10,%r8 + jmp .Loop1 + +.p2align 4 +.Lbreak1: + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rax,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rax,1),%r9 + andb $240,%bl + xorq (%r11,%rdx,8),%r9 + movq %r8,%rdx + xorq %r10,%r8 + + shrq $4,%r8 + andq $15,%rdx + movq %r9,%r10 + shrq $4,%r9 + xorq 8(%rsi,%rbx,1),%r8 + shlq $60,%r10 + xorq (%rsi,%rbx,1),%r9 + xorq %r10,%r8 + xorq (%r11,%rdx,8),%r9 + + bswapq %r8 + bswapq %r9 + movq %r8,8(%rdi) + movq %r9,(%rdi) + + movq 16(%rsp),%rbx + leaq 24(%rsp),%rsp +.Lgmult_epilogue: + movq 8(%rsp),%rdi + movq 16(%rsp),%rsi + retq +.LSEH_end_gcm_gmult_4bit: +.globl gcm_ghash_4bit +.def gcm_ghash_4bit; .scl 2; .type 32; .endef +.p2align 4 +gcm_ghash_4bit: + movq %rdi,8(%rsp) + movq %rsi,16(%rsp) + movq %rsp,%rax +.LSEH_begin_gcm_ghash_4bit: + movq %rcx,%rdi + movq %rdx,%rsi + movq %r8,%rdx + movq %r9,%rcx + + pushq %rbx + pushq %rbp + pushq %r12 + pushq %r13 + pushq %r14 + pushq %r15 + subq $280,%rsp +.Lghash_prologue: + movq %rdx,%r14 + movq %rcx,%r15 + subq $-128,%rsi + leaq 16+128(%rsp),%rbp + xorl %edx,%edx + movq 0+0-128(%rsi),%r8 + movq 0+8-128(%rsi),%rax + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq 16+0-128(%rsi),%r9 + shlb $4,%dl + movq 16+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,0(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,0(%rbp) + movq 32+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,0-128(%rbp) + movq 32+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,1(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,8(%rbp) + movq 48+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,8-128(%rbp) + movq 48+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,2(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,16(%rbp) + movq 64+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,16-128(%rbp) + movq 64+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,3(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,24(%rbp) + movq 80+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,24-128(%rbp) + movq 80+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,4(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,32(%rbp) + movq 96+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,32-128(%rbp) + movq 96+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,5(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,40(%rbp) + movq 112+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,40-128(%rbp) + movq 112+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,6(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,48(%rbp) + movq 128+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,48-128(%rbp) + movq 128+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,7(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,56(%rbp) + movq 144+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,56-128(%rbp) + movq 144+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,8(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,64(%rbp) + movq 160+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,64-128(%rbp) + movq 160+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,9(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,72(%rbp) + movq 176+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,72-128(%rbp) + movq 176+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,10(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,80(%rbp) + movq 192+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,80-128(%rbp) + movq 192+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,11(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,88(%rbp) + movq 208+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,88-128(%rbp) + movq 208+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,12(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,96(%rbp) + movq 224+0-128(%rsi),%r8 + shlb $4,%dl + movq %rax,96-128(%rbp) + movq 224+8-128(%rsi),%rax + shlq $60,%r10 + movb %dl,13(%rsp) + orq %r10,%rbx + movb %al,%dl + shrq $4,%rax + movq %r8,%r10 + shrq $4,%r8 + movq %r9,104(%rbp) + movq 240+0-128(%rsi),%r9 + shlb $4,%dl + movq %rbx,104-128(%rbp) + movq 240+8-128(%rsi),%rbx + shlq $60,%r10 + movb %dl,14(%rsp) + orq %r10,%rax + movb %bl,%dl + shrq $4,%rbx + movq %r9,%r10 + shrq $4,%r9 + movq %r8,112(%rbp) + shlb $4,%dl + movq %rax,112-128(%rbp) + shlq $60,%r10 + movb %dl,15(%rsp) + orq %r10,%rbx + movq %r9,120(%rbp) + movq %rbx,120-128(%rbp) + addq $-128,%rsi + movq 8(%rdi),%r8 + movq 0(%rdi),%r9 + addq %r14,%r15 + leaq .Lrem_8bit(%rip),%r11 + jmp .Louter_loop +.p2align 4 +.Louter_loop: + xorq (%r14),%r9 + movq 8(%r14),%rdx + leaq 16(%r14),%r14 + xorq %r8,%rdx + movq %r9,(%rdi) + movq %rdx,8(%rdi) + shrq $32,%rdx + xorq %rax,%rax + roll $8,%edx + movb %dl,%al + movzbl %dl,%ebx + shlb $4,%al + shrl $4,%ebx + roll $8,%edx + movq 8(%rsi,%rax,1),%r8 + movq (%rsi,%rax,1),%r9 + movb %dl,%al + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + xorq %r8,%r12 + movq %r9,%r10 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 8(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 4(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl 0(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + shrl $4,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r12,2),%r12 + movzbl %dl,%ebx + shlb $4,%al + movzbq (%rsp,%rcx,1),%r13 + shrl $4,%ebx + shlq $48,%r12 + xorq %r8,%r13 + movq %r9,%r10 + xorq %r12,%r9 + shrq $8,%r8 + movzbq %r13b,%r13 + shrq $8,%r9 + xorq -128(%rbp,%rcx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rcx,8),%r9 + roll $8,%edx + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + movb %dl,%al + xorq %r10,%r8 + movzwq (%r11,%r13,2),%r13 + movzbl %dl,%ecx + shlb $4,%al + movzbq (%rsp,%rbx,1),%r12 + andl $240,%ecx + shlq $48,%r13 + xorq %r8,%r12 + movq %r9,%r10 + xorq %r13,%r9 + shrq $8,%r8 + movzbq %r12b,%r12 + movl -4(%rdi),%edx + shrq $8,%r9 + xorq -128(%rbp,%rbx,8),%r8 + shlq $56,%r10 + xorq (%rbp,%rbx,8),%r9 + movzwq (%r11,%r12,2),%r12 + xorq 8(%rsi,%rax,1),%r8 + xorq (%rsi,%rax,1),%r9 + shlq $48,%r12 + xorq %r10,%r8 + xorq %r12,%r9 + movzbq %r8b,%r13 + shrq $4,%r8 + movq %r9,%r10 + shlb $4,%r13b + shrq $4,%r9 + xorq 8(%rsi,%rcx,1),%r8 + movzwq (%r11,%r13,2),%r13 + shlq $60,%r10 + xorq (%rsi,%rcx,1),%r9 + xorq %r10,%r8 + shlq $48,%r13 + bswapq %r8 + xorq %r13,%r9 + bswapq %r9 + cmpq %r15,%r14 + jb .Louter_loop + movq %r8,8(%rdi) + movq %r9,(%rdi) + + leaq 280(%rsp),%rsi + movq 0(%rsi),%r15 + movq 8(%rsi),%r14 + movq 16(%rsi),%r13 + movq 24(%rsi),%r12 + movq 32(%rsi),%rbp + movq 40(%rsi),%rbx + leaq 48(%rsi),%rsp +.Lghash_epilogue: + movq 8(%rsp),%rdi + movq 16(%rsp),%rsi + retq +.LSEH_end_gcm_ghash_4bit: +.globl gcm_init_clmul +.def gcm_init_clmul; .scl 2; .type 32; .endef +.p2align 4 +gcm_init_clmul: + movdqu (%rdx),%xmm2 + pshufd $78,%xmm2,%xmm2 + + + pshufd $255,%xmm2,%xmm4 + movdqa %xmm2,%xmm3 + psllq $1,%xmm2 + pxor %xmm5,%xmm5 + psrlq $63,%xmm3 + pcmpgtd %xmm4,%xmm5 + pslldq $8,%xmm3 + por %xmm3,%xmm2 + + + pand .L0x1c2_polynomial(%rip),%xmm5 + pxor %xmm5,%xmm2 + + + movdqa %xmm2,%xmm0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + movdqu %xmm2,(%rcx) + movdqu %xmm0,16(%rcx) + retq + +.globl gcm_gmult_clmul +.def gcm_gmult_clmul; .scl 2; .type 32; .endef +.p2align 4 +gcm_gmult_clmul: + movdqu (%rcx),%xmm0 + movdqa .Lbswap_mask(%rip),%xmm5 + movdqu (%rdx),%xmm2 +.byte 102,15,56,0,197 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 +.byte 102,15,56,0,197 + movdqu %xmm0,(%rcx) + retq + +.globl gcm_ghash_clmul +.def gcm_ghash_clmul; .scl 2; .type 32; .endef +.p2align 4 +gcm_ghash_clmul: +.LSEH_begin_gcm_ghash_clmul: + +.byte 0x48,0x83,0xec,0x58 +.byte 0x0f,0x29,0x34,0x24 +.byte 0x0f,0x29,0x7c,0x24,0x10 +.byte 0x44,0x0f,0x29,0x44,0x24,0x20 +.byte 0x44,0x0f,0x29,0x4c,0x24,0x30 +.byte 0x44,0x0f,0x29,0x54,0x24,0x40 + movdqa .Lbswap_mask(%rip),%xmm5 + + movdqu (%rcx),%xmm0 + movdqu (%rdx),%xmm2 +.byte 102,15,56,0,197 + + subq $16,%r9 + jz .Lodd_tail + + movdqu 16(%rdx),%xmm8 + + + + + + movdqu (%r8),%xmm3 + movdqu 16(%r8),%xmm6 +.byte 102,15,56,0,221 +.byte 102,15,56,0,245 + pxor %xmm3,%xmm0 + movdqa %xmm6,%xmm7 + pshufd $78,%xmm6,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm6,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,242,0 +.byte 102,15,58,68,250,17 +.byte 102,15,58,68,220,0 + pxor %xmm6,%xmm3 + pxor %xmm7,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm7 + pxor %xmm4,%xmm6 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm8,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm8,%xmm4 + + leaq 32(%r8),%r8 + subq $32,%r9 + jbe .Leven_tail + +.Lmod_loop: +.byte 102,65,15,58,68,192,0 +.byte 102,65,15,58,68,200,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + movdqu (%r8),%xmm3 + pxor %xmm6,%xmm0 + pxor %xmm7,%xmm1 + + movdqu 16(%r8),%xmm6 +.byte 102,15,56,0,221 +.byte 102,15,56,0,245 + + movdqa %xmm6,%xmm7 + pshufd $78,%xmm6,%xmm9 + pshufd $78,%xmm2,%xmm10 + pxor %xmm6,%xmm9 + pxor %xmm2,%xmm10 + pxor %xmm3,%xmm1 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 +.byte 102,15,58,68,242,0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + +.byte 102,15,58,68,250,17 + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + +.byte 102,69,15,58,68,202,0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm8,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm8,%xmm4 + + pxor %xmm6,%xmm9 + pxor %xmm7,%xmm9 + movdqa %xmm9,%xmm10 + psrldq $8,%xmm9 + pslldq $8,%xmm10 + pxor %xmm9,%xmm7 + pxor %xmm10,%xmm6 + + leaq 32(%r8),%r8 + subq $32,%r9 + ja .Lmod_loop + +.Leven_tail: +.byte 102,65,15,58,68,192,0 +.byte 102,65,15,58,68,200,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + pxor %xmm6,%xmm0 + pxor %xmm7,%xmm1 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + testq %r9,%r9 + jnz .Ldone + +.Lodd_tail: + movdqu (%r8),%xmm3 +.byte 102,15,56,0,221 + pxor %xmm3,%xmm0 + movdqa %xmm0,%xmm1 + pshufd $78,%xmm0,%xmm3 + pshufd $78,%xmm2,%xmm4 + pxor %xmm0,%xmm3 + pxor %xmm2,%xmm4 +.byte 102,15,58,68,194,0 +.byte 102,15,58,68,202,17 +.byte 102,15,58,68,220,0 + pxor %xmm0,%xmm3 + pxor %xmm1,%xmm3 + + movdqa %xmm3,%xmm4 + psrldq $8,%xmm3 + pslldq $8,%xmm4 + pxor %xmm3,%xmm1 + pxor %xmm4,%xmm0 + + movdqa %xmm0,%xmm3 + psllq $1,%xmm0 + pxor %xmm3,%xmm0 + psllq $5,%xmm0 + pxor %xmm3,%xmm0 + psllq $57,%xmm0 + movdqa %xmm0,%xmm4 + pslldq $8,%xmm0 + psrldq $8,%xmm4 + pxor %xmm3,%xmm0 + pxor %xmm4,%xmm1 + + + movdqa %xmm0,%xmm4 + psrlq $5,%xmm0 + pxor %xmm4,%xmm0 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 + pxor %xmm1,%xmm4 + psrlq $1,%xmm0 + pxor %xmm4,%xmm0 +.Ldone: +.byte 102,15,56,0,197 + movdqu %xmm0,(%rcx) + movaps (%rsp),%xmm6 + movaps 16(%rsp),%xmm7 + movaps 32(%rsp),%xmm8 + movaps 48(%rsp),%xmm9 + movaps 64(%rsp),%xmm10 + addq $88,%rsp + retq +.LSEH_end_gcm_ghash_clmul: + +.p2align 6 +.Lbswap_mask: +.byte 15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0 +.L0x1c2_polynomial: +.byte 1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xc2 +.p2align 6 + +.Lrem_4bit: +.long 0,0,0,471859200,0,943718400,0,610271232 +.long 0,1887436800,0,1822425088,0,1220542464,0,1423966208 +.long 0,3774873600,0,4246732800,0,3644850176,0,3311403008 +.long 0,2441084928,0,2376073216,0,2847932416,0,3051356160 + +.Lrem_8bit: +.value 0x0000,0x01C2,0x0384,0x0246,0x0708,0x06CA,0x048C,0x054E +.value 0x0E10,0x0FD2,0x0D94,0x0C56,0x0918,0x08DA,0x0A9C,0x0B5E +.value 0x1C20,0x1DE2,0x1FA4,0x1E66,0x1B28,0x1AEA,0x18AC,0x196E +.value 0x1230,0x13F2,0x11B4,0x1076,0x1538,0x14FA,0x16BC,0x177E +.value 0x3840,0x3982,0x3BC4,0x3A06,0x3F48,0x3E8A,0x3CCC,0x3D0E +.value 0x3650,0x3792,0x35D4,0x3416,0x3158,0x309A,0x32DC,0x331E +.value 0x2460,0x25A2,0x27E4,0x2626,0x2368,0x22AA,0x20EC,0x212E +.value 0x2A70,0x2BB2,0x29F4,0x2836,0x2D78,0x2CBA,0x2EFC,0x2F3E +.value 0x7080,0x7142,0x7304,0x72C6,0x7788,0x764A,0x740C,0x75CE +.value 0x7E90,0x7F52,0x7D14,0x7CD6,0x7998,0x785A,0x7A1C,0x7BDE +.value 0x6CA0,0x6D62,0x6F24,0x6EE6,0x6BA8,0x6A6A,0x682C,0x69EE +.value 0x62B0,0x6372,0x6134,0x60F6,0x65B8,0x647A,0x663C,0x67FE +.value 0x48C0,0x4902,0x4B44,0x4A86,0x4FC8,0x4E0A,0x4C4C,0x4D8E +.value 0x46D0,0x4712,0x4554,0x4496,0x41D8,0x401A,0x425C,0x439E +.value 0x54E0,0x5522,0x5764,0x56A6,0x53E8,0x522A,0x506C,0x51AE +.value 0x5AF0,0x5B32,0x5974,0x58B6,0x5DF8,0x5C3A,0x5E7C,0x5FBE +.value 0xE100,0xE0C2,0xE284,0xE346,0xE608,0xE7CA,0xE58C,0xE44E +.value 0xEF10,0xEED2,0xEC94,0xED56,0xE818,0xE9DA,0xEB9C,0xEA5E +.value 0xFD20,0xFCE2,0xFEA4,0xFF66,0xFA28,0xFBEA,0xF9AC,0xF86E +.value 0xF330,0xF2F2,0xF0B4,0xF176,0xF438,0xF5FA,0xF7BC,0xF67E +.value 0xD940,0xD882,0xDAC4,0xDB06,0xDE48,0xDF8A,0xDDCC,0xDC0E +.value 0xD750,0xD692,0xD4D4,0xD516,0xD058,0xD19A,0xD3DC,0xD21E +.value 0xC560,0xC4A2,0xC6E4,0xC726,0xC268,0xC3AA,0xC1EC,0xC02E +.value 0xCB70,0xCAB2,0xC8F4,0xC936,0xCC78,0xCDBA,0xCFFC,0xCE3E +.value 0x9180,0x9042,0x9204,0x93C6,0x9688,0x974A,0x950C,0x94CE +.value 0x9F90,0x9E52,0x9C14,0x9DD6,0x9898,0x995A,0x9B1C,0x9ADE +.value 0x8DA0,0x8C62,0x8E24,0x8FE6,0x8AA8,0x8B6A,0x892C,0x88EE +.value 0x83B0,0x8272,0x8034,0x81F6,0x84B8,0x857A,0x873C,0x86FE +.value 0xA9C0,0xA802,0xAA44,0xAB86,0xAEC8,0xAF0A,0xAD4C,0xAC8E +.value 0xA7D0,0xA612,0xA454,0xA596,0xA0D8,0xA11A,0xA35C,0xA29E +.value 0xB5E0,0xB422,0xB664,0xB7A6,0xB2E8,0xB32A,0xB16C,0xB0AE +.value 0xBBF0,0xBA32,0xB874,0xB9B6,0xBCF8,0xBD3A,0xBF7C,0xBEBE + +.byte 71,72,65,83,72,32,102,111,114,32,120,56,54,95,54,52,44,32,67,82,89,80,84,79,71,65,77,83,32,98,121,32,60,97,112,112,114,111,64,111,112,101,110,115,115,108,46,111,114,103,62,0 +.p2align 6 + +.def se_handler; .scl 3; .type 32; .endef +.p2align 4 +se_handler: + pushq %rsi + pushq %rdi + pushq %rbx + pushq %rbp + pushq %r12 + pushq %r13 + pushq %r14 + pushq %r15 + pushfq + subq $64,%rsp + + movq 120(%r8),%rax + movq 248(%r8),%rbx + + movq 8(%r9),%rsi + movq 56(%r9),%r11 + + movl 0(%r11),%r10d + leaq (%rsi,%r10,1),%r10 + cmpq %r10,%rbx + jb .Lin_prologue + + movq 152(%r8),%rax + + movl 4(%r11),%r10d + leaq (%rsi,%r10,1),%r10 + cmpq %r10,%rbx + jae .Lin_prologue + + leaq 24(%rax),%rax + + movq -8(%rax),%rbx + movq -16(%rax),%rbp + movq -24(%rax),%r12 + movq %rbx,144(%r8) + movq %rbp,160(%r8) + movq %r12,216(%r8) + +.Lin_prologue: + movq 8(%rax),%rdi + movq 16(%rax),%rsi + movq %rax,152(%r8) + movq %rsi,168(%r8) + movq %rdi,176(%r8) + + movq 40(%r9),%rdi + movq %r8,%rsi + movl $154,%ecx +.long 0xa548f3fc + + movq %r9,%rsi + xorq %rcx,%rcx + movq 8(%rsi),%rdx + movq 0(%rsi),%r8 + movq 16(%rsi),%r9 + movq 40(%rsi),%r10 + leaq 56(%rsi),%r11 + leaq 24(%rsi),%r12 + movq %r10,32(%rsp) + movq %r11,40(%rsp) + movq %r12,48(%rsp) + movq %rcx,56(%rsp) + call *__imp_RtlVirtualUnwind(%rip) + + movl $1,%eax + addq $64,%rsp + popfq + popq %r15 + popq %r14 + popq %r13 + popq %r12 + popq %rbp + popq %rbx + popq %rdi + popq %rsi + retq + + +.section .pdata +.p2align 2 +.rva .LSEH_begin_gcm_gmult_4bit +.rva .LSEH_end_gcm_gmult_4bit +.rva .LSEH_info_gcm_gmult_4bit + +.rva .LSEH_begin_gcm_ghash_4bit +.rva .LSEH_end_gcm_ghash_4bit +.rva .LSEH_info_gcm_ghash_4bit + +.rva .LSEH_begin_gcm_ghash_clmul +.rva .LSEH_end_gcm_ghash_clmul +.rva .LSEH_info_gcm_ghash_clmul + +.section .xdata +.p2align 3 +.LSEH_info_gcm_gmult_4bit: +.byte 9,0,0,0 +.rva se_handler +.rva .Lgmult_prologue,.Lgmult_epilogue +.LSEH_info_gcm_ghash_4bit: +.byte 9,0,0,0 +.rva se_handler +.rva .Lghash_prologue,.Lghash_epilogue +.LSEH_info_gcm_ghash_clmul: +.byte 0x01,0x1f,0x0b,0x00 +.byte 0x1f,0xa8,0x04,0x00 +.byte 0x19,0x98,0x03,0x00 +.byte 0x13,0x88,0x02,0x00 +.byte 0x0d,0x78,0x01,0x00 +.byte 0x08,0x68,0x00,0x00 +.byte 0x04,0xa2,0x00,0x00 diff --git a/ext/libressl/crypto/modes/modes_lcl.h b/ext/libressl/crypto/modes/modes_lcl.h new file mode 100644 index 0000000..bfea189 --- /dev/null +++ b/ext/libressl/crypto/modes/modes_lcl.h @@ -0,0 +1,113 @@ +/* $OpenBSD: modes_lcl.h,v 1.10 2016/12/21 15:49:29 jsing Exp $ */ +/* ==================================================================== + * Copyright (c) 2010 The OpenSSL Project. All rights reserved. + * + * Redistribution and use is governed by OpenSSL license. + * ==================================================================== + */ + +#include + +#include + +#include + +__BEGIN_HIDDEN_DECLS + +#if defined(_LP64) +typedef long i64; +typedef unsigned long u64; +#define U64(C) C##UL +#else +typedef long long i64; +typedef unsigned long long u64; +#define U64(C) C##ULL +#endif + +typedef unsigned int u32; +typedef unsigned char u8; + +#if !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) +#if defined(__GNUC__) && __GNUC__>=2 +# if defined(__x86_64) || defined(__x86_64__) +# define BSWAP8(x) ({ u64 ret=(x); \ + asm ("bswapq %0" \ + : "+r"(ret)); ret; }) +# define BSWAP4(x) ({ u32 ret=(x); \ + asm ("bswapl %0" \ + : "+r"(ret)); ret; }) +# elif (defined(__i386) || defined(__i386__)) +# define BSWAP8(x) ({ u32 lo=(u64)(x)>>32,hi=(x); \ + asm ("bswapl %0; bswapl %1" \ + : "+r"(hi),"+r"(lo)); \ + (u64)hi<<32|lo; }) +# define BSWAP4(x) ({ u32 ret=(x); \ + asm ("bswapl %0" \ + : "+r"(ret)); ret; }) +# elif (defined(__arm__) || defined(__arm)) && !defined(__STRICT_ALIGNMENT) +# if (__ARM_ARCH >= 6) +# define BSWAP8(x) ({ u32 lo=(u64)(x)>>32,hi=(x); \ + asm ("rev %0,%0; rev %1,%1" \ + : "+r"(hi),"+r"(lo)); \ + (u64)hi<<32|lo; }) +# define BSWAP4(x) ({ u32 ret; \ + asm ("rev %0,%1" \ + : "=r"(ret) : "r"((u32)(x))); \ + ret; }) +# endif +# endif +#endif +#endif + +#if defined(BSWAP4) && !defined(__STRICT_ALIGNMENT) +#define GETU32(p) BSWAP4(*(const u32 *)(p)) +#define PUTU32(p,v) *(u32 *)(p) = BSWAP4(v) +#else +#define GETU32(p) ((u32)(p)[0]<<24|(u32)(p)[1]<<16|(u32)(p)[2]<<8|(u32)(p)[3]) +#define PUTU32(p,v) ((p)[0]=(u8)((v)>>24),(p)[1]=(u8)((v)>>16),(p)[2]=(u8)((v)>>8),(p)[3]=(u8)(v)) +#endif + +/* GCM definitions */ + +typedef struct { u64 hi,lo; } u128; + +#ifdef TABLE_BITS +#undef TABLE_BITS +#endif +/* + * Even though permitted values for TABLE_BITS are 8, 4 and 1, it should + * never be set to 8 [or 1]. For further information see gcm128.c. + */ +#define TABLE_BITS 4 + +struct gcm128_context { + /* Following 6 names follow names in GCM specification */ + union { u64 u[2]; u32 d[4]; u8 c[16]; size_t t[16/sizeof(size_t)]; } + Yi,EKi,EK0,len,Xi,H; + /* Relative position of Xi, H and pre-computed Htable is used + * in some assembler modules, i.e. don't change the order! */ +#if TABLE_BITS==8 + u128 Htable[256]; +#else + u128 Htable[16]; + void (*gmult)(u64 Xi[2],const u128 Htable[16]); + void (*ghash)(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); +#endif + unsigned int mres, ares; + block128_f block; + void *key; +}; + +struct xts128_context { + void *key1, *key2; + block128_f block1,block2; +}; + +struct ccm128_context { + union { u64 u[2]; u8 c[16]; } nonce, cmac; + u64 blocks; + block128_f block; + void *key; +}; + +__END_HIDDEN_DECLS diff --git a/ext/libressl/crypto/modes/ofb128.c b/ext/libressl/crypto/modes/ofb128.c new file mode 100644 index 0000000..c6ca67a --- /dev/null +++ b/ext/libressl/crypto/modes/ofb128.c @@ -0,0 +1,119 @@ +/* $OpenBSD: ofb128.c,v 1.4 2015/02/10 09:46:30 miod Exp $ */ +/* ==================================================================== + * Copyright (c) 2008 The OpenSSL Project. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * + * 3. All advertising materials mentioning features or use of this + * software must display the following acknowledgment: + * "This product includes software developed by the OpenSSL Project + * for use in the OpenSSL Toolkit. (http://www.openssl.org/)" + * + * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to + * endorse or promote products derived from this software without + * prior written permission. For written permission, please contact + * openssl-core@openssl.org. + * + * 5. Products derived from this software may not be called "OpenSSL" + * nor may "OpenSSL" appear in their names without prior written + * permission of the OpenSSL Project. + * + * 6. Redistributions of any form whatsoever must retain the following + * acknowledgment: + * "This product includes software developed by the OpenSSL Project + * for use in the OpenSSL Toolkit (http://www.openssl.org/)" + * + * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY + * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR + * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, + * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED + * OF THE POSSIBILITY OF SUCH DAMAGE. + * ==================================================================== + * + */ + +#include +#include "modes_lcl.h" +#include + +#ifndef MODES_DEBUG +# ifndef NDEBUG +# define NDEBUG +# endif +#endif + +/* The input and output encrypted as though 128bit ofb mode is being + * used. The extra state information to record how much of the + * 128bit block we have used is contained in *num; + */ +void CRYPTO_ofb128_encrypt(const unsigned char *in, unsigned char *out, + size_t len, const void *key, + unsigned char ivec[16], int *num, + block128_f block) +{ + unsigned int n; + size_t l=0; + + n = *num; + +#if !defined(OPENSSL_SMALL_FOOTPRINT) + if (16%sizeof(size_t) == 0) do { /* always true actually */ + while (n && len) { + *(out++) = *(in++) ^ ivec[n]; + --len; + n = (n+1) % 16; + } +#ifdef __STRICT_ALIGNMENT + if (((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) + break; +#endif + while (len>=16) { + (*block)(ivec, ivec, key); + for (; n<16; n+=sizeof(size_t)) + *(size_t*)(out+n) = + *(size_t*)(in+n) ^ *(size_t*)(ivec+n); + len -= 16; + out += 16; + in += 16; + n = 0; + } + if (len) { + (*block)(ivec, ivec, key); + while (len--) { + out[n] = in[n] ^ ivec[n]; + ++n; + } + } + *num = n; + return; + } while(0); + /* the rest would be commonly eliminated by x86* compiler */ +#endif + while (l +#include +#include "modes_lcl.h" +#include + +#ifndef MODES_DEBUG +# ifndef NDEBUG +# define NDEBUG +# endif +#endif + +int CRYPTO_xts128_encrypt(const XTS128_CONTEXT *ctx, const unsigned char iv[16], + const unsigned char *inp, unsigned char *out, + size_t len, int enc) +{ + union { u64 u[2]; u32 d[4]; u8 c[16]; } tweak, scratch; + unsigned int i; + + if (len<16) return -1; + + memcpy(tweak.c, iv, 16); + + (*ctx->block2)(tweak.c,tweak.c,ctx->key2); + + if (!enc && (len%16)) len-=16; + + while (len>=16) { +#ifdef __STRICT_ALIGNMENT + memcpy(scratch.c,inp,16); + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; +#else + scratch.u[0] = ((u64*)inp)[0]^tweak.u[0]; + scratch.u[1] = ((u64*)inp)[1]^tweak.u[1]; +#endif + (*ctx->block1)(scratch.c,scratch.c,ctx->key1); +#ifdef __STRICT_ALIGNMENT + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; + memcpy(out,scratch.c,16); +#else + ((u64*)out)[0] = scratch.u[0]^=tweak.u[0]; + ((u64*)out)[1] = scratch.u[1]^=tweak.u[1]; +#endif + inp += 16; + out += 16; + len -= 16; + + if (len==0) return 0; + +#if BYTE_ORDER == LITTLE_ENDIAN + unsigned int carry,res; + + res = 0x87&(((int)tweak.d[3])>>31); + carry = (unsigned int)(tweak.u[0]>>63); + tweak.u[0] = (tweak.u[0]<<1)^res; + tweak.u[1] = (tweak.u[1]<<1)|carry; +#else /* BIG_ENDIAN */ + size_t c; + + for (c=0,i=0;i<16;++i) { + /*+ substitutes for |, because c is 1 bit */ + c += ((size_t)tweak.c[i])<<1; + tweak.c[i] = (u8)c; + c = c>>8; + } + tweak.c[0] ^= (u8)(0x87&(0-c)); +#endif + } + if (enc) { + for (i=0;iblock1)(scratch.c,scratch.c,ctx->key1); + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; + memcpy(out-16,scratch.c,16); + } + else { + union { u64 u[2]; u8 c[16]; } tweak1; + +#if BYTE_ORDER == LITTLE_ENDIAN + unsigned int carry,res; + + res = 0x87&(((int)tweak.d[3])>>31); + carry = (unsigned int)(tweak.u[0]>>63); + tweak1.u[0] = (tweak.u[0]<<1)^res; + tweak1.u[1] = (tweak.u[1]<<1)|carry; +#else + size_t c; + + for (c=0,i=0;i<16;++i) { + /*+ substitutes for |, because c is 1 bit */ + c += ((size_t)tweak.c[i])<<1; + tweak1.c[i] = (u8)c; + c = c>>8; + } + tweak1.c[0] ^= (u8)(0x87&(0-c)); +#endif +#ifdef __STRICT_ALIGNMENT + memcpy(scratch.c,inp,16); + scratch.u[0] ^= tweak1.u[0]; + scratch.u[1] ^= tweak1.u[1]; +#else + scratch.u[0] = ((u64*)inp)[0]^tweak1.u[0]; + scratch.u[1] = ((u64*)inp)[1]^tweak1.u[1]; +#endif + (*ctx->block1)(scratch.c,scratch.c,ctx->key1); + scratch.u[0] ^= tweak1.u[0]; + scratch.u[1] ^= tweak1.u[1]; + + for (i=0;iblock1)(scratch.c,scratch.c,ctx->key1); +#ifdef __STRICT_ALIGNMENT + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; + memcpy (out,scratch.c,16); +#else + ((u64*)out)[0] = scratch.u[0]^tweak.u[0]; + ((u64*)out)[1] = scratch.u[1]^tweak.u[1]; +#endif + } + + return 0; +} -- cgit v1.2.3