#endif
#include <assert.h>
-typedef struct { u64 hi,lo; } u128;
-
#if defined(BSWAP4) && defined(STRICT_ALIGNMENT)
/* redefine, because alignment is ensured */
#undef GETU32
} \
} while(0)
-#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. 8 is effectively reserved for testing purposes.
* - 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.
*/
-#define TABLE_BITS 4
-
#if TABLE_BITS==8
static void gcm_init_8bit(u128 Htable[256], u64 H[2])
}
}
-static void gcm_gmult_8bit(u64 Xi[2], u128 Htable[256])
+static void gcm_gmult_8bit(u64 Xi[2], const u128 Htable[256])
{
u128 Z = { 0, 0};
const u8 *xi = (const u8 *)Xi+15;
size_t rem, n = *xi;
const union { long one; char little; } is_endian = {1};
+ __fips_constseg
static const size_t rem_8bit[256] = {
PACK(0x0000), PACK(0x01C2), PACK(0x0384), PACK(0x0246),
PACK(0x0708), PACK(0x06CA), PACK(0x048C), PACK(0x054E),
}
#ifndef GHASH_ASM
+__fips_constseg
static const size_t rem_4bit[16] = {
PACK(0x0000), PACK(0x1C20), PACK(0x3840), PACK(0x2460),
PACK(0x7080), PACK(0x6CA0), PACK(0x48C0), PACK(0x54E0),
*/
u128 Hshr4[16]; /* Htable shifted right by 4 bits */
u8 Hshl4[16]; /* Htable shifted left by 4 bits */
+ __fips_constseg
static const unsigned short rem_8bit[256] = {
0x0000, 0x01C2, 0x0384, 0x0246, 0x0708, 0x06CA, 0x048C, 0x054E,
0x0E10, 0x0FD2, 0x0D94, 0x0C56, 0x0918, 0x08DA, 0x0A9C, 0x0B5E,
#endif
-struct gcm128_context {
- /* Following 6 names follow names in GCM specification */
- union { u64 u[2]; u32 d[4]; u8 c[16]; } Yi,EKi,EK0,
- Xi,H,len;
- /* Pre-computed table used by gcm_gmult_* */
-#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;
-};
-
-#if TABLE_BITS==4 && defined(GHASH_ASM) && !defined(I386_ONLY) && \
+#if TABLE_BITS==4 && (defined(GHASH_ASM) || defined(OPENSSL_CPUID_OBJ))
+# if !defined(I386_ONLY) && \
(defined(__i386) || defined(__i386__) || \
defined(__x86_64) || defined(__x86_64__) || \
defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64))
-# define GHASH_ASM_IAX
+# define GHASH_ASM_X86_OR_64
+# define GCM_FUNCREF_4BIT
extern unsigned int OPENSSL_ia32cap_P[2];
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
+#if defined(__i386) || defined(__i386__) || defined(_M_IX86)
+# define gcm_init_avx gcm_init_clmul
+# define gcm_gmult_avx gcm_gmult_clmul
+# define gcm_ghash_avx gcm_ghash_clmul
+#else
+void gcm_init_avx(u128 Htable[16],const u64 Xi[2]);
+void gcm_gmult_avx(u64 Xi[2],const u128 Htable[16]);
+void gcm_ghash_avx(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len);
+#endif
+
+# 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) || defined(__aarch64__)
+# include "arm_arch.h"
+# if __ARM_ARCH__>=7
+# define GHASH_ASM_ARM
+# define GCM_FUNCREF_4BIT
+# define PMULL_CAPABLE (OPENSSL_armcap_P & ARMV8_PMULL)
+# if defined(__arm__) || defined(__arm)
+# define NEON_CAPABLE (OPENSSL_armcap_P & ARMV7_NEON)
+# endif
+void gcm_init_neon(u128 Htable[16],const u64 Xi[2]);
+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);
+void gcm_init_v8(u128 Htable[16],const u64 Xi[2]);
+void gcm_gmult_v8(u64 Xi[2],const u128 Htable[16]);
+void gcm_ghash_v8(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len);
+# endif
+# elif defined(__sparc__) || defined(__sparc)
+# include "sparc_arch.h"
+# define GHASH_ASM_SPARC
+# define GCM_FUNCREF_4BIT
+extern unsigned int OPENSSL_sparcv9cap_P[];
+void gcm_init_vis3(u128 Htable[16],const u64 Xi[2]);
+void gcm_gmult_vis3(u64 Xi[2],const u128 Htable[16]);
+void gcm_ghash_vis3(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len);
+#elif defined(OPENSSL_CPUID_OBJ) && (defined(__powerpc__) || defined(__ppc__) || defined(_ARCH_PPC))
+# define GHASH_ASM_PPC
+# define GCM_FUNCREF_4BIT
+extern unsigned int OPENSSL_ppccap_P[];
+void gcm_init_p8(u128 Htable[16],const u64 Xi[2]);
+void gcm_gmult_p8(u64 Xi[2],const u128 Htable[16]);
+void gcm_ghash_p8(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len);
# endif
+#endif
+#ifdef GCM_FUNCREF_4BIT
# undef GCM_MUL
-# define GCM_MUL(ctx,Xi) (*((ctx)->gmult))(ctx->Xi.u,ctx->Htable)
-# undef GHASH
-# define GHASH(ctx,in,len) (*((ctx)->ghash))((ctx)->Xi.u,(ctx)->Htable,in,len)
+# 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)
#if TABLE_BITS==8
gcm_init_8bit(ctx->Htable,ctx->H.u);
#elif TABLE_BITS==4
-# if defined(GHASH_ASM_IAX) /* both x86 and x86_64 */
- if (OPENSSL_ia32cap_P[1]&(1<<1)) {
- gcm_init_clmul(ctx->Htable,ctx->H.u);
- ctx->gmult = gcm_gmult_clmul;
- ctx->ghash = gcm_ghash_clmul;
+# if defined(GHASH_ASM_X86_OR_64)
+# if !defined(GHASH_ASM_X86) || defined(OPENSSL_IA32_SSE2)
+ if (OPENSSL_ia32cap_P[0]&(1<<24) && /* check FXSR bit */
+ OPENSSL_ia32cap_P[1]&(1<<1) ) { /* check PCLMULQDQ bit */
+ if (((OPENSSL_ia32cap_P[1]>>22)&0x41)==0x41) { /* AVX+MOVBE */
+ gcm_init_avx(ctx->Htable,ctx->H.u);
+ ctx->gmult = gcm_gmult_avx;
+ ctx->ghash = gcm_ghash_avx;
+ } else {
+ 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 (OPENSSL_ia32cap_P[0]&(1<<23)) {
+# if defined(OPENSSL_IA32_SSE2)
+ if (OPENSSL_ia32cap_P[0]&(1<<25)) { /* check SSE bit */
+# else
+ if (OPENSSL_ia32cap_P[0]&(1<<23)) { /* check MMX bit */
+# endif
ctx->gmult = gcm_gmult_4bit_mmx;
ctx->ghash = gcm_ghash_4bit_mmx;
} else {
ctx->gmult = gcm_gmult_4bit;
ctx->ghash = gcm_ghash_4bit;
# endif
+# elif defined(GHASH_ASM_ARM)
+# ifdef PMULL_CAPABLE
+ if (PMULL_CAPABLE) {
+ gcm_init_v8(ctx->Htable,ctx->H.u);
+ ctx->gmult = gcm_gmult_v8;
+ ctx->ghash = gcm_ghash_v8;
+ } else
+# endif
+# ifdef NEON_CAPABLE
+ if (NEON_CAPABLE) {
+ gcm_init_neon(ctx->Htable,ctx->H.u);
+ ctx->gmult = gcm_gmult_neon;
+ ctx->ghash = gcm_ghash_neon;
+ } else
+# endif
+ {
+ gcm_init_4bit(ctx->Htable,ctx->H.u);
+ ctx->gmult = gcm_gmult_4bit;
+ ctx->ghash = gcm_ghash_4bit;
+ }
+# elif defined(GHASH_ASM_SPARC)
+ if (OPENSSL_sparcv9cap_P[0] & SPARCV9_VIS3) {
+ gcm_init_vis3(ctx->Htable,ctx->H.u);
+ ctx->gmult = gcm_gmult_vis3;
+ ctx->ghash = gcm_ghash_vis3;
+ } else {
+ gcm_init_4bit(ctx->Htable,ctx->H.u);
+ ctx->gmult = gcm_gmult_4bit;
+ ctx->ghash = gcm_ghash_4bit;
+ }
+# elif defined(GHASH_ASM_PPC)
+ if (OPENSSL_ppccap_P[0] & (1<<2)) {
+ gcm_init_p8(ctx->Htable,ctx->H.u);
+ ctx->gmult = gcm_gmult_p8;
+ ctx->ghash = gcm_ghash_p8;
+ } 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
{
const union { long one; char little; } is_endian = {1};
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;
GCM_MUL(ctx,Yi);
if (is_endian.little)
+#ifdef BSWAP4
+ ctr = BSWAP4(ctx->Yi.d[3]);
+#else
ctr = GETU32(ctx->Yi.c+12);
+#endif
else
ctr = ctx->Yi.d[3];
}
(*ctx->block)(ctx->Yi.c,ctx->EK0.c,ctx->key);
++ctr;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
}
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;
const union { long one; char little; } is_endian = {1};
unsigned int n, ctr;
size_t i;
- u64 mlen = ctx->len.u[1];
+ 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
#if 0
n = (unsigned int)mlen%16; /* alternative to ctx->mres */
}
if (is_endian.little)
+#ifdef BSWAP4
+ ctr = BSWAP4(ctx->Yi.d[3]);
+#else
ctr = GETU32(ctx->Yi.c+12);
+#endif
else
ctr = ctx->Yi.d[3];
size_t j=GHASH_CHUNK;
while (j) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ 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 (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
- for (i=0; i<16; i+=sizeof(size_t))
- *(size_t *)(out+i) =
- *(size_t *)(in+i)^*(size_t *)(ctx->EKi.c+i);
+ 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;
size_t j=i;
while (len>=16) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ 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 (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
- for (i=0; i<16; i+=sizeof(size_t))
- *(size_t *)(out+i) =
- *(size_t *)(in+i)^*(size_t *)(ctx->EKi.c+i);
+ 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) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ 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 (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
- for (i=0; i<16; i+=sizeof(size_t))
- *(size_t *)(ctx->Xi.c+i) ^=
- *(size_t *)(out+i) =
- *(size_t *)(in+i)^*(size_t *)(ctx->EKi.c+i);
+ 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;
}
#endif
if (len) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ (*block)(ctx->Yi.c,ctx->EKi.c,key);
++ctr;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
while (len--) {
#endif
for (i=0;i<len;++i) {
if (n==0) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ (*block)(ctx->Yi.c,ctx->EKi.c,key);
++ctr;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
}
const union { long one; char little; } is_endian = {1};
unsigned int n, ctr;
size_t i;
- u64 mlen = ctx->len.u[1];
+ 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 && mlen<len))
}
if (is_endian.little)
+#ifdef BSWAP4
+ ctr = BSWAP4(ctx->Yi.d[3]);
+#else
ctr = GETU32(ctx->Yi.c+12);
+#endif
else
ctr = ctx->Yi.d[3];
GHASH(ctx,in,GHASH_CHUNK);
while (j) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ 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 (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
- for (i=0; i<16; i+=sizeof(size_t))
- *(size_t *)(out+i) =
- *(size_t *)(in+i)^*(size_t *)(ctx->EKi.c+i);
+ 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;
if ((i = (len&(size_t)-16))) {
GHASH(ctx,in,i);
while (len>=16) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ 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 (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
- for (i=0; i<16; i+=sizeof(size_t))
- *(size_t *)(out+i) =
- *(size_t *)(in+i)^*(size_t *)(ctx->EKi.c+i);
+ 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) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ 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 (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
- for (i=0; i<16; i+=sizeof(size_t)) {
- size_t c = *(size_t *)(in+i);
- *(size_t *)(out+i) = c^*(size_t *)(ctx->EKi.c+i);
- *(size_t *)(ctx->Xi.c+i) ^= c;
+ 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;
}
#endif
if (len) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ (*block)(ctx->Yi.c,ctx->EKi.c,key);
++ctr;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
while (len--) {
for (i=0;i<len;++i) {
u8 c;
if (n==0) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ (*block)(ctx->Yi.c,ctx->EKi.c,key);
++ctr;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
}
const union { long one; char little; } is_endian = {1};
unsigned int n, ctr;
size_t i;
- u64 mlen = ctx->len.u[1];
+ 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 && mlen<len))
}
if (is_endian.little)
+#ifdef BSWAP4
+ ctr = BSWAP4(ctx->Yi.d[3]);
+#else
ctr = GETU32(ctx->Yi.c+12);
+#endif
else
ctr = ctx->Yi.d[3];
}
#if defined(GHASH) && !defined(OPENSSL_SMALL_FOOTPRINT)
while (len>=GHASH_CHUNK) {
- (*stream)(in,out,GHASH_CHUNK/16,ctx->key,ctx->Yi.c);
+ (*stream)(in,out,GHASH_CHUNK/16,key,ctx->Yi.c);
ctr += GHASH_CHUNK/16;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
GHASH(ctx,out,GHASH_CHUNK);
if ((i = (len&(size_t)-16))) {
size_t j=i/16;
- (*stream)(in,out,j,ctx->key,ctx->Yi.c);
+ (*stream)(in,out,j,key,ctx->Yi.c);
ctr += (unsigned int)j;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
in += i;
#endif
}
if (len) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ (*ctx->block)(ctx->Yi.c,ctx->EKi.c,key);
++ctr;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
while (len--) {
const union { long one; char little; } is_endian = {1};
unsigned int n, ctr;
size_t i;
- u64 mlen = ctx->len.u[1];
+ 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 && mlen<len))
}
if (is_endian.little)
+#ifdef BSWAP4
+ ctr = BSWAP4(ctx->Yi.d[3]);
+#else
ctr = GETU32(ctx->Yi.c+12);
+#endif
else
ctr = ctx->Yi.d[3];
#if defined(GHASH) && !defined(OPENSSL_SMALL_FOOTPRINT)
while (len>=GHASH_CHUNK) {
GHASH(ctx,in,GHASH_CHUNK);
- (*stream)(in,out,GHASH_CHUNK/16,ctx->key,ctx->Yi.c);
+ (*stream)(in,out,GHASH_CHUNK/16,key,ctx->Yi.c);
ctr += GHASH_CHUNK/16;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
out += GHASH_CHUNK;
j = i/16;
in -= i;
#endif
- (*stream)(in,out,j,ctx->key,ctx->Yi.c);
+ (*stream)(in,out,j,key,ctx->Yi.c);
ctr += (unsigned int)j;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
out += i;
len -= i;
}
if (len) {
- (*ctx->block)(ctx->Yi.c,ctx->EKi.c,ctx->key);
+ (*ctx->block)(ctx->Yi.c,ctx->EKi.c,key);
++ctr;
if (is_endian.little)
+#ifdef BSWAP4
+ ctx->Yi.d[3] = BSWAP4(ctr);
+#else
PUTU32(ctx->Yi.c+12,ctr);
+#endif
else
ctx->Yi.d[3] = ctr;
while (len--) {
const union { long one; char little; } is_endian = {1};
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)
+ if (ctx->mres || ctx->ares)
GCM_MUL(ctx,Xi);
if (is_endian.little) {
/* Test Case 5 */
#define K5 K4
#define P5 P4
-static const u8 A5[]= {0xfe,0xed,0xfa,0xce,0xde,0xad,0xbe,0xef,0xfe,0xed,0xfa,0xce,0xde,0xad,0xbe,0xef,
- 0xab,0xad,0xda,0xd2},
- IV5[]= {0xca,0xfe,0xba,0xbe,0xfa,0xce,0xdb,0xad},
+#define A5 A4
+static const u8 IV5[]= {0xca,0xfe,0xba,0xbe,0xfa,0xce,0xdb,0xad},
C5[]= {0x61,0x35,0x3b,0x4c,0x28,0x06,0x93,0x4a,0x77,0x7f,0xf5,0x1f,0xa2,0x2a,0x47,0x55,
0x69,0x9b,0x2a,0x71,0x4f,0xcd,0xc6,0xf8,0x37,0x66,0xe5,0xf9,0x7b,0x6c,0x74,0x23,
0x73,0x80,0x69,0x00,0xe4,0x9f,0x24,0xb2,0x2b,0x09,0x75,0x44,0xd4,0x89,0x6b,0x42,
0xa2,0x41,0x89,0x97,0x20,0x0e,0xf8,0x2e,0x44,0xae,0x7e,0x3f},
T18[]= {0xa4,0x4a,0x82,0x66,0xee,0x1c,0x8e,0xb0,0xc8,0xb5,0xd4,0xcf,0x5a,0xe9,0xf1,0x9a};
+/* Test Case 19 */
+#define K19 K1
+#define P19 P1
+#define IV19 IV1
+#define C19 C1
+static const u8 A19[]= {0xd9,0x31,0x32,0x25,0xf8,0x84,0x06,0xe5,0xa5,0x59,0x09,0xc5,0xaf,0xf5,0x26,0x9a,
+ 0x86,0xa7,0xa9,0x53,0x15,0x34,0xf7,0xda,0x2e,0x4c,0x30,0x3d,0x8a,0x31,0x8a,0x72,
+ 0x1c,0x3c,0x0c,0x95,0x95,0x68,0x09,0x53,0x2f,0xcf,0x0e,0x24,0x49,0xa6,0xb5,0x25,
+ 0xb1,0x6a,0xed,0xf5,0xaa,0x0d,0xe6,0x57,0xba,0x63,0x7b,0x39,0x1a,0xaf,0xd2,0x55,
+ 0x52,0x2d,0xc1,0xf0,0x99,0x56,0x7d,0x07,0xf4,0x7f,0x37,0xa3,0x2a,0x84,0x42,0x7d,
+ 0x64,0x3a,0x8c,0xdc,0xbf,0xe5,0xc0,0xc9,0x75,0x98,0xa2,0xbd,0x25,0x55,0xd1,0xaa,
+ 0x8c,0xb0,0x8e,0x48,0x59,0x0d,0xbb,0x3d,0xa7,0xb0,0x8b,0x10,0x56,0x82,0x88,0x38,
+ 0xc5,0xf6,0x1e,0x63,0x93,0xba,0x7a,0x0a,0xbc,0xc9,0xf6,0x62,0x89,0x80,0x15,0xad},
+ T19[]= {0x5f,0xea,0x79,0x3a,0x2d,0x6f,0x97,0x4d,0x37,0xe6,0x8e,0x0c,0xb8,0xff,0x94,0x92};
+
+/* Test Case 20 */
+#define K20 K1
+#define A20 A1
+static const u8 IV20[64]={0xff,0xff,0xff,0xff}, /* this results in 0xff in counter LSB */
+ P20[288],
+ C20[]= {0x56,0xb3,0x37,0x3c,0xa9,0xef,0x6e,0x4a,0x2b,0x64,0xfe,0x1e,0x9a,0x17,0xb6,0x14,
+ 0x25,0xf1,0x0d,0x47,0xa7,0x5a,0x5f,0xce,0x13,0xef,0xc6,0xbc,0x78,0x4a,0xf2,0x4f,
+ 0x41,0x41,0xbd,0xd4,0x8c,0xf7,0xc7,0x70,0x88,0x7a,0xfd,0x57,0x3c,0xca,0x54,0x18,
+ 0xa9,0xae,0xff,0xcd,0x7c,0x5c,0xed,0xdf,0xc6,0xa7,0x83,0x97,0xb9,0xa8,0x5b,0x49,
+ 0x9d,0xa5,0x58,0x25,0x72,0x67,0xca,0xab,0x2a,0xd0,0xb2,0x3c,0xa4,0x76,0xa5,0x3c,
+ 0xb1,0x7f,0xb4,0x1c,0x4b,0x8b,0x47,0x5c,0xb4,0xf3,0xf7,0x16,0x50,0x94,0xc2,0x29,
+ 0xc9,0xe8,0xc4,0xdc,0x0a,0x2a,0x5f,0xf1,0x90,0x3e,0x50,0x15,0x11,0x22,0x13,0x76,
+ 0xa1,0xcd,0xb8,0x36,0x4c,0x50,0x61,0xa2,0x0c,0xae,0x74,0xbc,0x4a,0xcd,0x76,0xce,
+ 0xb0,0xab,0xc9,0xfd,0x32,0x17,0xef,0x9f,0x8c,0x90,0xbe,0x40,0x2d,0xdf,0x6d,0x86,
+ 0x97,0xf4,0xf8,0x80,0xdf,0xf1,0x5b,0xfb,0x7a,0x6b,0x28,0x24,0x1e,0xc8,0xfe,0x18,
+ 0x3c,0x2d,0x59,0xe3,0xf9,0xdf,0xff,0x65,0x3c,0x71,0x26,0xf0,0xac,0xb9,0xe6,0x42,
+ 0x11,0xf4,0x2b,0xae,0x12,0xaf,0x46,0x2b,0x10,0x70,0xbe,0xf1,0xab,0x5e,0x36,0x06,
+ 0x87,0x2c,0xa1,0x0d,0xee,0x15,0xb3,0x24,0x9b,0x1a,0x1b,0x95,0x8f,0x23,0x13,0x4c,
+ 0x4b,0xcc,0xb7,0xd0,0x32,0x00,0xbc,0xe4,0x20,0xa2,0xf8,0xeb,0x66,0xdc,0xf3,0x64,
+ 0x4d,0x14,0x23,0xc1,0xb5,0x69,0x90,0x03,0xc1,0x3e,0xce,0xf4,0xbf,0x38,0xa3,0xb6,
+ 0x0e,0xed,0xc3,0x40,0x33,0xba,0xc1,0x90,0x27,0x83,0xdc,0x6d,0x89,0xe2,0xe7,0x74,
+ 0x18,0x8a,0x43,0x9c,0x7e,0xbc,0xc0,0x67,0x2d,0xbd,0xa4,0xdd,0xcf,0xb2,0x79,0x46,
+ 0x13,0xb0,0xbe,0x41,0x31,0x5e,0xf7,0x78,0x70,0x8a,0x70,0xee,0x7d,0x75,0x16,0x5c},
+ T20[]= {0x8b,0x30,0x7f,0x6b,0x33,0x28,0x6d,0x0a,0xb0,0x26,0xa9,0xed,0x3f,0xe1,0xe8,0x5f};
+
#define TEST_CASE(n) do { \
u8 out[sizeof(P##n)]; \
AES_set_encrypt_key(K##n,sizeof(K##n)*8,&key); \
TEST_CASE(16);
TEST_CASE(17);
TEST_CASE(18);
+ TEST_CASE(19);
+ TEST_CASE(20);
#ifdef OPENSSL_CPUID_OBJ
{
ctr_t/(double)sizeof(buf),
(gcm_t-ctr_t)/(double)sizeof(buf));
#ifdef GHASH
- GHASH(&ctx,buf.c,sizeof(buf));
+ {
+ void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16],
+ const u8 *inp,size_t len) = ctx.ghash;
+
+ GHASH((&ctx),buf.c,sizeof(buf));
start = OPENSSL_rdtsc();
- for (i=0;i<100;++i) GHASH(&ctx,buf.c,sizeof(buf));
+ for (i=0;i<100;++i) GHASH((&ctx),buf.c,sizeof(buf));
gcm_t = OPENSSL_rdtsc() - start;
printf("%.2f\n",gcm_t/(double)sizeof(buf)/(double)i);
+ }
#endif
}
#endif