$code.=<<___;
################################################################################
-# void ecp_nistz256_select_w5(uint64_t *val, uint64_t *in_t, int index);
-.globl ecp_nistz256_select_w5
-.type ecp_nistz256_select_w5,\@abi-omnipotent
+# void ecp_nistz256_scatter_w5(uint64_t *val, uint64_t *in_t, int index);
+.globl ecp_nistz256_scatter_w5
+.type ecp_nistz256_scatter_w5,\@abi-omnipotent
.align 32
-ecp_nistz256_select_w5:
+ecp_nistz256_scatter_w5:
+ lea -3($index,$index,2), $index
+ movdqa 0x00($in_t), %xmm0
+ shl \$5, $index
+ movdqa 0x10($in_t), %xmm1
+ movdqa 0x20($in_t), %xmm2
+ movdqa 0x30($in_t), %xmm3
+ movdqa 0x40($in_t), %xmm4
+ movdqa 0x50($in_t), %xmm5
+ movdqa %xmm0, 0x00($val,$index)
+ movdqa %xmm1, 0x10($val,$index)
+ movdqa %xmm2, 0x20($val,$index)
+ movdqa %xmm3, 0x30($val,$index)
+ movdqa %xmm4, 0x40($val,$index)
+ movdqa %xmm5, 0x50($val,$index)
+
+ ret
+.size ecp_nistz256_scatter_w5,.-ecp_nistz256_scatter_w5
+
+################################################################################
+# void ecp_nistz256_gather_w5(uint64_t *val, uint64_t *in_t, int index);
+.globl ecp_nistz256_gather_w5
+.type ecp_nistz256_gather_w5,\@abi-omnipotent
+.align 32
+ecp_nistz256_gather_w5:
___
$code.=<<___ if ($avx>1);
mov OPENSSL_ia32cap_P+8(%rip), %eax
test \$`1<<5`, %eax
- jnz .Lavx2_select_w5
+ jnz .Lavx2_gather_w5
___
$code.=<<___ if ($win64);
lea -0x88(%rsp), %rax
-.LSEH_begin_ecp_nistz256_select_w5:
+.LSEH_begin_ecp_nistz256_gather_w5:
.byte 0x48,0x8d,0x60,0xe0 #lea -0x20(%rax), %rsp
.byte 0x0f,0x29,0x70,0xe0 #movaps %xmm6, -0x20(%rax)
.byte 0x0f,0x29,0x78,0xf0 #movaps %xmm7, -0x10(%rax)
movaps 0x80(%rsp), %xmm14
movaps 0x90(%rsp), %xmm15
lea 0xa8(%rsp), %rsp
-.LSEH_end_ecp_nistz256_select_w5:
+.LSEH_end_ecp_nistz256_gather_w5:
___
$code.=<<___;
ret
-.size ecp_nistz256_select_w5,.-ecp_nistz256_select_w5
+.size ecp_nistz256_gather_w5,.-ecp_nistz256_gather_w5
################################################################################
-# void ecp_nistz256_select_w7(uint64_t *val, uint64_t *in_t, int index);
-.globl ecp_nistz256_select_w7
-.type ecp_nistz256_select_w7,\@abi-omnipotent
+# void ecp_nistz256_scatter_w7(uint64_t *val, uint64_t *in_t, int index);
+.globl ecp_nistz256_scatter_w7
+.type ecp_nistz256_scatter_w7,\@abi-omnipotent
.align 32
-ecp_nistz256_select_w7:
+ecp_nistz256_scatter_w7:
+ movdqu 0x00($in_t), %xmm0
+ shl \$6, $index
+ movdqu 0x10($in_t), %xmm1
+ movdqu 0x20($in_t), %xmm2
+ movdqu 0x30($in_t), %xmm3
+ movdqa %xmm0, 0x00($val,$index)
+ movdqa %xmm1, 0x10($val,$index)
+ movdqa %xmm2, 0x20($val,$index)
+ movdqa %xmm3, 0x30($val,$index)
+
+ ret
+.size ecp_nistz256_scatter_w7,.-ecp_nistz256_scatter_w7
+
+################################################################################
+# void ecp_nistz256_gather_w7(uint64_t *val, uint64_t *in_t, int index);
+.globl ecp_nistz256_gather_w7
+.type ecp_nistz256_gather_w7,\@abi-omnipotent
+.align 32
+ecp_nistz256_gather_w7:
___
$code.=<<___ if ($avx>1);
mov OPENSSL_ia32cap_P+8(%rip), %eax
test \$`1<<5`, %eax
- jnz .Lavx2_select_w7
+ jnz .Lavx2_gather_w7
___
$code.=<<___ if ($win64);
lea -0x88(%rsp), %rax
-.LSEH_begin_ecp_nistz256_select_w7:
+.LSEH_begin_ecp_nistz256_gather_w7:
.byte 0x48,0x8d,0x60,0xe0 #lea -0x20(%rax), %rsp
.byte 0x0f,0x29,0x70,0xe0 #movaps %xmm6, -0x20(%rax)
.byte 0x0f,0x29,0x78,0xf0 #movaps %xmm7, -0x10(%rax)
movaps 0x80(%rsp), %xmm14
movaps 0x90(%rsp), %xmm15
lea 0xa8(%rsp), %rsp
-.LSEH_end_ecp_nistz256_select_w7:
+.LSEH_end_ecp_nistz256_gather_w7:
___
$code.=<<___;
ret
-.size ecp_nistz256_select_w7,.-ecp_nistz256_select_w7
+.size ecp_nistz256_gather_w7,.-ecp_nistz256_gather_w7
___
}
if ($avx>1) {
$code.=<<___;
################################################################################
-# void ecp_nistz256_avx2_select_w5(uint64_t *val, uint64_t *in_t, int index);
-.type ecp_nistz256_avx2_select_w5,\@abi-omnipotent
+# void ecp_nistz256_avx2_gather_w5(uint64_t *val, uint64_t *in_t, int index);
+.type ecp_nistz256_avx2_gather_w5,\@abi-omnipotent
.align 32
-ecp_nistz256_avx2_select_w5:
-.Lavx2_select_w5:
+ecp_nistz256_avx2_gather_w5:
+.Lavx2_gather_w5:
vzeroupper
___
$code.=<<___ if ($win64);
lea -0x88(%rsp), %rax
-.LSEH_begin_ecp_nistz256_avx2_select_w5:
+.LSEH_begin_ecp_nistz256_avx2_gather_w5:
.byte 0x48,0x8d,0x60,0xe0 #lea -0x20(%rax), %rsp
.byte 0xc5,0xf8,0x29,0x70,0xe0 #vmovaps %xmm6, -0x20(%rax)
.byte 0xc5,0xf8,0x29,0x78,0xf0 #vmovaps %xmm7, -0x10(%rax)
movaps 0x80(%rsp), %xmm14
movaps 0x90(%rsp), %xmm15
lea 0xa8(%rsp), %rsp
-.LSEH_end_ecp_nistz256_avx2_select_w5:
+.LSEH_end_ecp_nistz256_avx2_gather_w5:
___
$code.=<<___;
ret
-.size ecp_nistz256_avx2_select_w5,.-ecp_nistz256_avx2_select_w5
+.size ecp_nistz256_avx2_gather_w5,.-ecp_nistz256_avx2_gather_w5
___
}
if ($avx>1) {
$code.=<<___;
################################################################################
-# void ecp_nistz256_avx2_select_w7(uint64_t *val, uint64_t *in_t, int index);
-.globl ecp_nistz256_avx2_select_w7
-.type ecp_nistz256_avx2_select_w7,\@abi-omnipotent
+# void ecp_nistz256_avx2_gather_w7(uint64_t *val, uint64_t *in_t, int index);
+.globl ecp_nistz256_avx2_gather_w7
+.type ecp_nistz256_avx2_gather_w7,\@abi-omnipotent
.align 32
-ecp_nistz256_avx2_select_w7:
-.Lavx2_select_w7:
+ecp_nistz256_avx2_gather_w7:
+.Lavx2_gather_w7:
vzeroupper
___
$code.=<<___ if ($win64);
lea -0x88(%rsp), %rax
-.LSEH_begin_ecp_nistz256_avx2_select_w7:
+.LSEH_begin_ecp_nistz256_avx2_gather_w7:
.byte 0x48,0x8d,0x60,0xe0 #lea -0x20(%rax), %rsp
.byte 0xc5,0xf8,0x29,0x70,0xe0 #vmovaps %xmm6, -0x20(%rax)
.byte 0xc5,0xf8,0x29,0x78,0xf0 #vmovaps %xmm7, -0x10(%rax)
movaps 0x80(%rsp), %xmm14
movaps 0x90(%rsp), %xmm15
lea 0xa8(%rsp), %rsp
-.LSEH_end_ecp_nistz256_avx2_select_w7:
+.LSEH_end_ecp_nistz256_avx2_gather_w7:
___
$code.=<<___;
ret
-.size ecp_nistz256_avx2_select_w7,.-ecp_nistz256_avx2_select_w7
+.size ecp_nistz256_avx2_gather_w7,.-ecp_nistz256_avx2_gather_w7
___
} else {
$code.=<<___;
-.globl ecp_nistz256_avx2_select_w7
-.type ecp_nistz256_avx2_select_w7,\@function,3
+.globl ecp_nistz256_avx2_gather_w7
+.type ecp_nistz256_avx2_gather_w7,\@function,3
.align 32
-ecp_nistz256_avx2_select_w7:
+ecp_nistz256_avx2_gather_w7:
.byte 0x0f,0x0b # ud2
ret
-.size ecp_nistz256_avx2_select_w7,.-ecp_nistz256_avx2_select_w7
+.size ecp_nistz256_avx2_gather_w7,.-ecp_nistz256_avx2_gather_w7
___
}
{{{
}
}}}
+########################################################################
+# Convert ecp_nistz256_table.c to layout expected by ecp_nistz_gather_w7
+#
+open TABLE,"<ecp_nistz256_table.c" or
+open TABLE,"<${dir}/../ecp_nistz256_table.c" or
+die "failed to open ecp_nistz256_table.c:",$!;
+
+use integer;
+
+foreach(<TABLE>) {
+ s/TOBN\(\s*(0x[0-9a-f]+),\s*(0x[0-9a-f]+)\s*\)/push @arr,hex($2),hex($1)/geo;
+}
+close TABLE;
+
+die "insane number of elements" if ($#arr != 64*16*37-1);
+
+print <<___;
+.text
+.globl ecp_nistz256_precomputed
+.type ecp_nistz256_precomputed,\@object
+.align 4096
+ecp_nistz256_precomputed:
+___
+while (@line=splice(@arr,0,16)) {
+ print ".long\t",join(',',map { sprintf "0x%08x",$_} @line),"\n";
+}
+print <<___;
+.size ecp_nistz256_precomputed,.-ecp_nistz256_precomputed
+___
+
$code =~ s/\`([^\`]*)\`/eval $1/gem;
print $code;
close STDOUT;
void ecp_nistz256_to_mont(BN_ULONG res[P256_LIMBS],
const BN_ULONG in[P256_LIMBS]);
/* Functions that perform constant time access to the precomputed tables */
-void ecp_nistz256_select_w5(P256_POINT * val,
+void ecp_nistz256_scatter_w5(P256_POINT * val,
const P256_POINT * in_t, int index);
-void ecp_nistz256_select_w7(P256_POINT_AFFINE * val,
+void ecp_nistz256_gather_w5(P256_POINT * val,
+ const P256_POINT * in_t, int index);
+void ecp_nistz256_scatter_w7(P256_POINT_AFFINE * val,
+ const P256_POINT_AFFINE * in_t, int index);
+void ecp_nistz256_gather_w7(P256_POINT_AFFINE * val,
const P256_POINT_AFFINE * in_t, int index);
/* One converted into the Montgomery domain */
static EC_PRE_COMP *ecp_nistz256_pre_comp_new(const EC_GROUP * group);
/* Precomputed tables for the default generator */
-#include "ecp_nistz256_table.c"
+extern const PRECOMP256_ROW ecp_nistz256_precomputed[37];
/* Recode window to a signed digit, see ecp_nistputil.c for details */
static unsigned int _booth_recode_w5(unsigned int in)
const unsigned int window_size = 5;
const unsigned int mask = (1 << (window_size + 1)) - 1;
unsigned int wvalue;
- BN_ULONG tmp[P256_LIMBS];
- ALIGN32 P256_POINT h;
+ P256_POINT *temp; /* place for 5 temporary points */
const BIGNUM **scalars = NULL;
P256_POINT(*table)[16] = NULL;
void *table_storage = NULL;
if ((table_storage =
- OPENSSL_malloc(num * 16 * sizeof(P256_POINT) + 64)) == NULL
+ OPENSSL_malloc((num * 16 + 5) * sizeof(P256_POINT) + 64)) == NULL
|| (p_str =
OPENSSL_malloc(num * 33 * sizeof(unsigned char))) == NULL
|| (scalars = OPENSSL_malloc(num * sizeof(BIGNUM *))) == NULL) {
ECerr(EC_F_ECP_NISTZ256_WINDOWED_MUL, ERR_R_MALLOC_FAILURE);
goto err;
- } else {
- table = (void *)ALIGNPTR(table_storage, 64);
}
+ table = (void *)ALIGNPTR(table_storage, 64);
+ temp = (P256_POINT *)(table + num);
+
for (i = 0; i < num; i++) {
P256_POINT *row = table[i];
for (; j < 33; j++)
p_str[i][j] = 0;
- /* table[0] is implicitly (0,0,0) (the point at infinity),
- * therefore it is not stored. All other values are actually
- * stored with an offset of -1 in table.
- */
-
- if (!ecp_nistz256_bignum_to_field_elem(row[1 - 1].X, &point[i]->X)
- || !ecp_nistz256_bignum_to_field_elem(row[1 - 1].Y, &point[i]->Y)
- || !ecp_nistz256_bignum_to_field_elem(row[1 - 1].Z, &point[i]->Z)) {
+ if (!ecp_nistz256_bignum_to_field_elem(temp[0].X, &point[i]->X)
+ || !ecp_nistz256_bignum_to_field_elem(temp[0].Y, &point[i]->Y)
+ || !ecp_nistz256_bignum_to_field_elem(temp[0].Z, &point[i]->Z)) {
ECerr(EC_F_ECP_NISTZ256_WINDOWED_MUL, EC_R_COORDINATES_OUT_OF_RANGE);
goto err;
}
- ecp_nistz256_point_double(&row[ 2 - 1], &row[ 1 - 1]);
- ecp_nistz256_point_add (&row[ 3 - 1], &row[ 2 - 1], &row[1 - 1]);
- ecp_nistz256_point_double(&row[ 4 - 1], &row[ 2 - 1]);
- ecp_nistz256_point_double(&row[ 6 - 1], &row[ 3 - 1]);
- ecp_nistz256_point_double(&row[ 8 - 1], &row[ 4 - 1]);
- ecp_nistz256_point_double(&row[12 - 1], &row[ 6 - 1]);
- ecp_nistz256_point_add (&row[ 5 - 1], &row[ 4 - 1], &row[1 - 1]);
- ecp_nistz256_point_add (&row[ 7 - 1], &row[ 6 - 1], &row[1 - 1]);
- ecp_nistz256_point_add (&row[ 9 - 1], &row[ 8 - 1], &row[1 - 1]);
- ecp_nistz256_point_add (&row[13 - 1], &row[12 - 1], &row[1 - 1]);
- ecp_nistz256_point_double(&row[14 - 1], &row[ 7 - 1]);
- ecp_nistz256_point_double(&row[10 - 1], &row[ 5 - 1]);
- ecp_nistz256_point_add (&row[15 - 1], &row[14 - 1], &row[1 - 1]);
- ecp_nistz256_point_add (&row[11 - 1], &row[10 - 1], &row[1 - 1]);
- ecp_nistz256_point_add (&row[16 - 1], &row[15 - 1], &row[1 - 1]);
+ /* row[0] is implicitly (0,0,0) (the point at infinity),
+ * therefore it is not stored. All other values are actually
+ * stored with an offset of -1 in table.
+ */
+
+ ecp_nistz256_scatter_w5 (row, &temp[0], 1);
+ ecp_nistz256_point_double(&temp[1], &temp[0]); /*1+1=2 */
+ ecp_nistz256_scatter_w5 (row, &temp[1], 2);
+ ecp_nistz256_point_add (&temp[2], &temp[1], &temp[0]); /*2+1=3 */
+ ecp_nistz256_scatter_w5 (row, &temp[2], 3);
+ ecp_nistz256_point_double(&temp[1], &temp[1]); /*2*2=4 */
+ ecp_nistz256_scatter_w5 (row, &temp[1], 4);
+ ecp_nistz256_point_double(&temp[2], &temp[2]); /*2*3=6 */
+ ecp_nistz256_scatter_w5 (row, &temp[2], 6);
+ ecp_nistz256_point_add (&temp[3], &temp[1], &temp[0]); /*4+1=5 */
+ ecp_nistz256_scatter_w5 (row, &temp[3], 5);
+ ecp_nistz256_point_add (&temp[4], &temp[2], &temp[0]); /*6+1=7 */
+ ecp_nistz256_scatter_w5 (row, &temp[4], 7);
+ ecp_nistz256_point_double(&temp[1], &temp[1]); /*2*4=8 */
+ ecp_nistz256_scatter_w5 (row, &temp[1], 8);
+ ecp_nistz256_point_double(&temp[2], &temp[2]); /*2*6=12 */
+ ecp_nistz256_scatter_w5 (row, &temp[2], 12);
+ ecp_nistz256_point_double(&temp[3], &temp[3]); /*2*5=10 */
+ ecp_nistz256_scatter_w5 (row, &temp[3], 10);
+ ecp_nistz256_point_double(&temp[4], &temp[4]); /*2*7=14 */
+ ecp_nistz256_scatter_w5 (row, &temp[4], 14);
+ ecp_nistz256_point_add (&temp[2], &temp[2], &temp[0]); /*12+1=13*/
+ ecp_nistz256_scatter_w5 (row, &temp[2], 13);
+ ecp_nistz256_point_add (&temp[3], &temp[3], &temp[0]); /*10+1=11*/
+ ecp_nistz256_scatter_w5 (row, &temp[3], 11);
+ ecp_nistz256_point_add (&temp[4], &temp[4], &temp[0]); /*14+1=15*/
+ ecp_nistz256_scatter_w5 (row, &temp[4], 15);
+ ecp_nistz256_point_add (&temp[2], &temp[1], &temp[0]); /*8+1=9 */
+ ecp_nistz256_scatter_w5 (row, &temp[2], 9);
+ ecp_nistz256_point_double(&temp[1], &temp[1]); /*2*8=16 */
+ ecp_nistz256_scatter_w5 (row, &temp[1], 16);
}
index = 255;
wvalue = p_str[0][(index - 1) / 8];
wvalue = (wvalue >> ((index - 1) % 8)) & mask;
- ecp_nistz256_select_w5(r, table[0], _booth_recode_w5(wvalue) >> 1);
+ /*
+ * We gather to temp[0], because we know it's position relative
+ * to table
+ */
+ ecp_nistz256_gather_w5(&temp[0], table[0], _booth_recode_w5(wvalue) >> 1);
+ memcpy(r, &temp[0], sizeof(temp[0]));
while (index >= 5) {
for (i = (index == 255 ? 1 : 0); i < num; i++) {
wvalue = _booth_recode_w5(wvalue);
- ecp_nistz256_select_w5(&h, table[i], wvalue >> 1);
+ ecp_nistz256_gather_w5(&temp[0], table[i], wvalue >> 1);
- ecp_nistz256_neg(tmp, h.Y);
- copy_conditional(h.Y, tmp, (wvalue & 1));
+ ecp_nistz256_neg(temp[1].Y, temp[0].Y);
+ copy_conditional(temp[0].Y, temp[1].Y, (wvalue & 1));
- ecp_nistz256_point_add(r, r, &h);
+ ecp_nistz256_point_add(r, r, &temp[0]);
}
index -= window_size;
wvalue = _booth_recode_w5(wvalue);
- ecp_nistz256_select_w5(&h, table[i], wvalue >> 1);
+ ecp_nistz256_gather_w5(&temp[0], table[i], wvalue >> 1);
- ecp_nistz256_neg(tmp, h.Y);
- copy_conditional(h.Y, tmp, wvalue & 1);
+ ecp_nistz256_neg(temp[1].Y, temp[0].Y);
+ copy_conditional(temp[0].Y, temp[1].Y, wvalue & 1);
- ecp_nistz256_point_add(r, r, &h);
+ ecp_nistz256_point_add(r, r, &temp[0]);
}
err:
OPENSSL_malloc(37 * 64 * sizeof(P256_POINT_AFFINE) + 64)) == NULL) {
ECerr(EC_F_ECP_NISTZ256_MULT_PRECOMPUTE, ERR_R_MALLOC_FAILURE);
goto err;
- } else {
- preComputedTable = (void *)ALIGNPTR(precomp_storage, 64);
}
+ preComputedTable = (void *)ALIGNPTR(precomp_storage, 64);
+
P = EC_POINT_new(group);
T = EC_POINT_new(group);
for (k = 0; k < 64; k++) {
EC_POINT_copy(P, T);
for (j = 0; j < 37; j++) {
+ P256_POINT_AFFINE temp;
/* It would be faster to use
* ec_GFp_simple_points_make_affine and make multiple
* points affine at the same time. */
ec_GFp_simple_make_affine(group, P, ctx);
- ecp_nistz256_bignum_to_field_elem(preComputedTable[j]
- [k].X, &P->X);
- ecp_nistz256_bignum_to_field_elem(preComputedTable[j]
- [k].Y, &P->Y);
+ ecp_nistz256_bignum_to_field_elem(temp.X, &P->X);
+ ecp_nistz256_bignum_to_field_elem(temp.Y, &P->Y);
+ ecp_nistz256_scatter_w7(preComputedTable[j], &temp, k);
for (i = 0; i < 7; i++)
ec_GFp_simple_dbl(group, P, P, ctx);
}
* you'd need to compile even asm/ecp_nistz256-avx.pl module.
*/
#if defined(ECP_NISTZ256_AVX2)
-# if !(defined(__x86_64) || defined(__x86_64__)) || \
+# if !(defined(__x86_64) || defined(__x86_64__) || \
defined(_M_AMD64) || defined(_MX64)) || \
!(defined(__GNUC__) || defined(_MSC_VER)) /* this is for ALIGN32 */
# undef ECP_NISTZ256_AVX2
# else
/* Constant time access, loading four values, from four consecutive tables */
-void ecp_nistz256_avx2_select_w7(P256_POINT_AFFINE * val,
- const P256_POINT_AFFINE * in_t, int index);
-void ecp_nistz256_avx2_multi_select_w7(void *result, const void *in, int index0,
+void ecp_nistz256_avx2_multi_gather_w7(void *result, const void *in, int index0,
int index1, int index2, int index3);
void ecp_nistz256_avx2_transpose_convert(void *RESULTx4, const void *in);
void ecp_nistz256_avx2_convert_transpose_back(void *result, const void *Ax4);
ALIGN32 BN_ULONG aX4[4 * 9 * 3] = { 0 };
ALIGN32 BN_ULONG bX4[4 * 9 * 2] = { 0 };
- ALIGN32 P256_POINT_AFFINE point_arr[P256_LIMBS];
- ALIGN32 P256_POINT res_point_arr[P256_LIMBS];
+ ALIGN32 P256_POINT_AFFINE point_arr[4];
+ ALIGN32 P256_POINT res_point_arr[4];
/* Initial four windows */
wvalue = *((u16 *) & p_str[0]);
index += window_size;
booth_recode_w7(&sign3, &digit3, wvalue);
- ecp_nistz256_avx2_multi_select_w7(point_arr, preComputedTable[0],
+ ecp_nistz256_avx2_multi_gather_w7(point_arr, preComputedTable[0],
digit0, digit1, digit2, digit3);
ecp_nistz256_neg(tmp, point_arr[0].Y);
index += window_size;
booth_recode_w7(&sign3, &digit3, wvalue);
- ecp_nistz256_avx2_multi_select_w7(point_arr, preComputedTable[4 * 1],
+ ecp_nistz256_avx2_multi_gather_w7(point_arr, preComputedTable[4 * 1],
digit0, digit1, digit2, digit3);
ecp_nistz256_neg(tmp, point_arr[0].Y);
index += window_size;
booth_recode_w7(&sign3, &digit3, wvalue);
- ecp_nistz256_avx2_multi_select_w7(point_arr,
+ ecp_nistz256_avx2_multi_gather_w7(point_arr,
preComputedTable[4 * i],
digit0, digit1, digit2, digit3);
wvalue = *((u16 *) & p_str[(index - 1) / 8]);
wvalue = (wvalue >> ((index - 1) % 8)) & mask;
booth_recode_w7(&sign0, &digit0, wvalue);
- ecp_nistz256_avx2_select_w7((P256_POINT_AFFINE *) r,
+ ecp_nistz256_gather_w7((P256_POINT_AFFINE *) r,
preComputedTable[36], digit0);
ecp_nistz256_neg(tmp, r->Y);
copy_conditional(r->Y, tmp, sign0);
} t, p;
BIGNUM *tmp_scalar;
+ if ((num+1) == 0 || (num+1) > OPENSSL_MALLOC_MAX_NELEMS(void *)) {
+ ECerr(EC_F_ECP_NISTZ256_POINTS_MUL, ERR_R_MALLOC_FAILURE);
+ return 0;
+ }
+
if (group->meth != r->meth) {
ECerr(EC_F_ECP_NISTZ256_POINTS_MUL, EC_R_INCOMPATIBLE_OBJECTS);
return 0;
if (pre_comp_generator == NULL)
goto err;
- if (!ecp_nistz256_set_from_affine
- (pre_comp_generator, group, pre_comp->precomp[0], ctx))
+ if (!ecp_nistz256_set_from_affine(pre_comp_generator,
+ group, pre_comp->precomp[0],
+ ctx))
goto err;
if (0 == EC_POINT_cmp(group, generator, pre_comp_generator, ctx))
* is the default, a hardcoded table of precomputed
* data is used. This is because applications, such as
* Apache, do not use EC_KEY_precompute_mult. */
- preComputedTable = (const PRECOMP256_ROW *)ecp_nistz256_precomputed;
+ preComputedTable = ecp_nistz256_precomputed;
}
if (preComputedTable) {
wvalue = _booth_recode_w7(wvalue);
- ecp_nistz256_select_w7(&p.a, preComputedTable[0], wvalue >> 1);
+ ecp_nistz256_gather_w7(&p.a, preComputedTable[0], wvalue >> 1);
ecp_nistz256_neg(p.p.Z, p.p.Y);
copy_conditional(p.p.Y, p.p.Z, wvalue & 1);
wvalue = _booth_recode_w7(wvalue);
- ecp_nistz256_select_w7(&t.a,
+ ecp_nistz256_gather_w7(&t.a,
preComputedTable[i], wvalue >> 1);
ecp_nistz256_neg(t.p.Z, t.a.Y);
projects / openssl.git / commitdiff