WolfSSL
/
wolfssl
mirror of https://github.com/wolfSSL/wolfssl.git
1
0
Fork 0
Code Issues Releases Wiki Activity

ARMv8 32 Inline ASM: fixes 

Force parameters to use specific registers.
Shift/rotate amount must have '#' prepended.
pull/5152/head
Sean Parkinson 2022-09-21 10:45:50 +10:00
parent ca392fb200
commit 2578f2c8f2
3 changed files with 1044 additions and 897 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

181
wolfcrypt/src/port/arm/armv8-32-curve25519_c.c
View File

@ -39,6 +39,7 @@
void fe_init()
{
__asm__ __volatile__ (
"\n\t"
:
@ -47,8 +48,11 @@ void fe_init()
);
}
void fe_frombytes(fe out, const unsigned char* in)
void fe_frombytes(fe out_p, const unsigned char* in_p)
{
register fe out asm ("r0") = out_p;
register const unsigned char* in asm ("r1") = in_p;
__asm__ __volatile__ (
"ldrd r2, r3, [%[in]]\n\t"
"ldrd r12, lr, [%[in], #8]\n\t"
@ -65,8 +69,11 @@ void fe_frombytes(fe out, const unsigned char* in)
);
}
void fe_tobytes(unsigned char* out, const fe n)
void fe_tobytes(unsigned char* out_p, const fe n_p)
{
register unsigned char* out asm ("r0") = out_p;
register const fe n asm ("r1") = n_p;
__asm__ __volatile__ (
"ldrd r2, r3, [%[n]]\n\t"
"ldrd r12, lr, [%[n], #8]\n\t"
@ -101,8 +108,10 @@ void fe_tobytes(unsigned char* out, const fe n)
);
}
void fe_1(fe n)
void fe_1(fe n_p)
{
register fe n asm ("r0") = n_p;
__asm__ __volatile__ (
/* Set one */
"mov r2, #1\n\t"
@ -117,8 +126,10 @@ void fe_1(fe n)
);
}
void fe_0(fe n)
void fe_0(fe n_p)
{
register fe n asm ("r0") = n_p;
__asm__ __volatile__ (
/* Set zero */
"mov r1, #0\n\t"
@ -132,8 +143,11 @@ void fe_0(fe n)
);
}
void fe_copy(fe r, const fe a)
void fe_copy(fe r_p, const fe a_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
__asm__ __volatile__ (
/* Copy */
"ldrd r2, r3, [%[a]]\n\t"
@ -150,8 +164,12 @@ void fe_copy(fe r, const fe a)
);
}
void fe_sub(fe r, const fe a, const fe b)
void fe_sub(fe r_p, const fe a_p, const fe b_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
register const fe b asm ("r2") = b_p;
__asm__ __volatile__ (
/* Sub */
"ldrd r12, lr, [%[a]]\n\t"
@ -198,8 +216,12 @@ void fe_sub(fe r, const fe a, const fe b)
);
}
void fe_add(fe r, const fe a, const fe b)
void fe_add(fe r_p, const fe a_p, const fe b_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
register const fe b asm ("r2") = b_p;
__asm__ __volatile__ (
/* Add */
"ldrd r12, lr, [%[a]]\n\t"
@ -246,8 +268,11 @@ void fe_add(fe r, const fe a, const fe b)
);
}
void fe_neg(fe r, const fe a)
void fe_neg(fe r_p, const fe a_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
__asm__ __volatile__ (
"mov r5, #-1\n\t"
"mov r4, #-19\n\t"
@ -274,8 +299,10 @@ void fe_neg(fe r, const fe a)
);
}
int fe_isnonzero(const fe a)
int fe_isnonzero(const fe a_p)
{
register const fe a asm ("r0") = a_p;
__asm__ __volatile__ (
"ldrd r2, r3, [%[a]]\n\t"
"ldrd r12, lr, [%[a], #8]\n\t"
@ -314,8 +341,10 @@ int fe_isnonzero(const fe a)
return (uint32_t)(size_t)a;
}
int fe_isnegative(const fe a)
int fe_isnegative(const fe a_p)
{
register const fe a asm ("r0") = a_p;
__asm__ __volatile__ (
"ldrd r2, r3, [%[a]]\n\t"
"ldrd r12, lr, [%[a], #8]\n\t"
@ -340,8 +369,12 @@ int fe_isnegative(const fe a)
return (uint32_t)(size_t)a;
}
void fe_cmov_table(fe* r, fe* base, signed char b)
void fe_cmov_table(fe* r_p, fe* base_p, signed char b_p)
{
register fe* r asm ("r0") = r_p;
register fe* base asm ("r1") = base_p;
register signed char b asm ("r2") = b_p;
__asm__ __volatile__ (
"sxtb %[b], %[b]\n\t"
"sbfx r7, %[b], #7, #1\n\t"
@ -1312,8 +1345,12 @@ void fe_cmov_table(fe* r, fe* base, signed char b)
);
}
void fe_mul(fe r, const fe a, const fe b)
void fe_mul(fe r_p, const fe a_p, const fe b_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
register const fe b asm ("r2") = b_p;
__asm__ __volatile__ (
"sub sp, sp, #0x40\n\t"
/* Multiply */
@ -1842,8 +1879,11 @@ void fe_mul(fe r, const fe a, const fe b)
);
}
void fe_sq(fe r, const fe a)
void fe_sq(fe r_p, const fe a_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
__asm__ __volatile__ (
"sub sp, sp, #0x40\n\t"
/* Square */
@ -2264,8 +2304,11 @@ void fe_sq(fe r, const fe a)
);
}
void fe_mul121666(fe r, fe a)
void fe_mul121666(fe r_p, fe a_p)
{
register fe r asm ("r0") = r_p;
register fe a asm ("r1") = a_p;
__asm__ __volatile__ (
/* Multiply by 121666 */
"ldrd r2, r3, [%[a]]\n\t"
@ -2319,8 +2362,11 @@ void fe_mul121666(fe r, fe a)
);
}
void fe_sq2(fe r, const fe a)
void fe_sq2(fe r_p, const fe a_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
__asm__ __volatile__ (
"sub sp, sp, #0x40\n\t"
/* Square * 2 */
@ -2756,8 +2802,11 @@ void fe_sq2(fe r, const fe a)
);
}
void fe_invert(fe r, const fe a)
void fe_invert(fe r_p, const fe a_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
__asm__ __volatile__ (
"sub sp, sp, #0x88\n\t"
/* Invert */
@ -2915,8 +2964,12 @@ void fe_invert(fe r, const fe a)
);
}
int curve25519(byte* r, const byte* n, const byte* a)
int curve25519(byte* r_p, const byte* n_p, const byte* a_p)
{
register byte* r asm ("r0") = r_p;
register const byte* n asm ("r1") = n_p;
register const byte* a asm ("r2") = a_p;
__asm__ __volatile__ (
"sub sp, sp, #0xbc\n\t"
"str %[r], [sp, #160]\n\t"
@ -3694,8 +3747,11 @@ int curve25519(byte* r, const byte* n, const byte* a)
return (uint32_t)(size_t)r;
}
void fe_pow22523(fe r, const fe a)
void fe_pow22523(fe r_p, const fe a_p)
{
register fe r asm ("r0") = r_p;
register const fe a asm ("r1") = a_p;
__asm__ __volatile__ (
"sub sp, sp, #0x68\n\t"
/* pow22523 */
@ -3853,8 +3909,16 @@ void fe_pow22523(fe r, const fe a)
);
}
void fe_ge_to_p2(fe rx, fe ry, fe rz, const fe px, const fe py, const fe pz, const fe pt)
void fe_ge_to_p2(fe rx_p, fe ry_p, fe rz_p, const fe px_p, const fe py_p, const fe pz_p, const fe pt_p)
{
register fe rx asm ("r0") = rx_p;
register fe ry asm ("r1") = ry_p;
register fe rz asm ("r2") = rz_p;
register const fe px asm ("r3") = px_p;
register const fe py asm ("r4") = py_p;
register const fe pz asm ("r5") = pz_p;
register const fe pt asm ("r6") = pt_p;
__asm__ __volatile__ (
"sub sp, sp, #16\n\t"
"str %[rx], [sp]\n\t"
@ -3883,8 +3947,17 @@ void fe_ge_to_p2(fe rx, fe ry, fe rz, const fe px, const fe py, const fe pz, con
(void)pt;
}
void fe_ge_to_p3(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe pz, const fe pt)
void fe_ge_to_p3(fe rx_p, fe ry_p, fe rz_p, fe rt_p, const fe px_p, const fe py_p, const fe pz_p, const fe pt_p)
{
register fe rx asm ("r0") = rx_p;
register fe ry asm ("r1") = ry_p;
register fe rz asm ("r2") = rz_p;
register fe rt asm ("r3") = rt_p;
register const fe px asm ("r4") = px_p;
register const fe py asm ("r5") = py_p;
register const fe pz asm ("r6") = pz_p;
register const fe pt asm ("r7") = pt_p;
__asm__ __volatile__ (
"sub sp, sp, #16\n\t"
"str %[rx], [sp]\n\t"
@ -3918,8 +3991,16 @@ void fe_ge_to_p3(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe
(void)pt;
}
void fe_ge_dbl(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe pz)
void fe_ge_dbl(fe rx_p, fe ry_p, fe rz_p, fe rt_p, const fe px_p, const fe py_p, const fe pz_p)
{
register fe rx asm ("r0") = rx_p;
register fe ry asm ("r1") = ry_p;
register fe rz asm ("r2") = rz_p;
register fe rt asm ("r3") = rt_p;
register const fe px asm ("r4") = px_p;
register const fe py asm ("r5") = py_p;
register const fe pz asm ("r6") = pz_p;
__asm__ __volatile__ (
"sub sp, sp, #16\n\t"
"str %[rx], [sp]\n\t"
@ -4175,8 +4256,20 @@ void fe_ge_dbl(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe pz
(void)pz;
}
void fe_ge_madd(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe pz, const fe pt, const fe qxy2d, const fe qyplusx, const fe qyminusx)
void fe_ge_madd(fe rx_p, fe ry_p, fe rz_p, fe rt_p, const fe px_p, const fe py_p, const fe pz_p, const fe pt_p, const fe qxy2d_p, const fe qyplusx_p, const fe qyminusx_p)
{
register fe rx asm ("r0") = rx_p;
register fe ry asm ("r1") = ry_p;
register fe rz asm ("r2") = rz_p;
register fe rt asm ("r3") = rt_p;
register const fe px asm ("r4") = px_p;
register const fe py asm ("r5") = py_p;
register const fe pz asm ("r6") = pz_p;
register const fe pt asm ("r7") = pt_p;
register const fe qxy2d asm ("r8") = qxy2d_p;
register const fe qyplusx asm ("r9") = qyplusx_p;
register const fe qyminusx asm ("r10") = qyminusx_p;
__asm__ __volatile__ (
"sub sp, sp, #32\n\t"
"str %[rx], [sp]\n\t"
@ -4529,8 +4622,20 @@ void fe_ge_madd(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe p
(void)qyminusx;
}
void fe_ge_msub(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe pz, const fe pt, const fe qxy2d, const fe qyplusx, const fe qyminusx)
void fe_ge_msub(fe rx_p, fe ry_p, fe rz_p, fe rt_p, const fe px_p, const fe py_p, const fe pz_p, const fe pt_p, const fe qxy2d_p, const fe qyplusx_p, const fe qyminusx_p)
{
register fe rx asm ("r0") = rx_p;
register fe ry asm ("r1") = ry_p;
register fe rz asm ("r2") = rz_p;
register fe rt asm ("r3") = rt_p;
register const fe px asm ("r4") = px_p;
register const fe py asm ("r5") = py_p;
register const fe pz asm ("r6") = pz_p;
register const fe pt asm ("r7") = pt_p;
register const fe qxy2d asm ("r8") = qxy2d_p;
register const fe qyplusx asm ("r9") = qyplusx_p;
register const fe qyminusx asm ("r10") = qyminusx_p;
__asm__ __volatile__ (
"sub sp, sp, #32\n\t"
"str %[rx], [sp]\n\t"
@ -4883,8 +4988,21 @@ void fe_ge_msub(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe p
(void)qyminusx;
}
void fe_ge_add(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe pz, const fe pt, const fe qz, const fe qt2d, const fe qyplusx, const fe qyminusx)
void fe_ge_add(fe rx_p, fe ry_p, fe rz_p, fe rt_p, const fe px_p, const fe py_p, const fe pz_p, const fe pt_p, const fe qz_p, const fe qt2d_p, const fe qyplusx_p, const fe qyminusx_p)
{
register fe rx asm ("r0") = rx_p;
register fe ry asm ("r1") = ry_p;
register fe rz asm ("r2") = rz_p;
register fe rt asm ("r3") = rt_p;
register const fe px asm ("r4") = px_p;
register const fe py asm ("r5") = py_p;
register const fe pz asm ("r6") = pz_p;
register const fe pt asm ("r7") = pt_p;
register const fe qz asm ("r8") = qz_p;
register const fe qt2d asm ("r9") = qt2d_p;
register const fe qyplusx asm ("r10") = qyplusx_p;
register const fe qyminusx asm ("r11") = qyminusx_p;
__asm__ __volatile__ (
"sub sp, sp, #0x60\n\t"
"str %[rx], [sp]\n\t"
@ -5243,8 +5361,21 @@ void fe_ge_add(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe pz
(void)qyminusx;
}
void fe_ge_sub(fe rx, fe ry, fe rz, fe rt, const fe px, const fe py, const fe pz, const fe pt, const fe qz, const fe qt2d, const fe qyplusx, const fe qyminusx)
void fe_ge_sub(fe rx_p, fe ry_p, fe rz_p, fe rt_p, const fe px_p, const fe py_p, const fe pz_p, const fe pt_p, const fe qz_p, const fe qt2d_p, const fe qyplusx_p, const fe qyminusx_p)
{
register fe rx asm ("r0") = rx_p;
register fe ry asm ("r1") = ry_p;
register fe rz asm ("r2") = rz_p;
register fe rt asm ("r3") = rt_p;
register const fe px asm ("r4") = px_p;
register const fe py asm ("r5") = py_p;
register const fe pz asm ("r6") = pz_p;
register const fe pt asm ("r7") = pt_p;
register const fe qz asm ("r8") = qz_p;
register const fe qt2d asm ("r9") = qt2d_p;
register const fe qyplusx asm ("r10") = qyplusx_p;
register const fe qyminusx asm ("r11") = qyminusx_p;
__asm__ __volatile__ (
"sub sp, sp, #0x60\n\t"
"str %[rx], [sp]\n\t"

656
wolfcrypt/src/port/arm/armv8-32-sha256-asm_c.c
View File

File diff suppressed because it is too large Load Diff

1104
wolfcrypt/src/port/arm/armv8-32-sha512-asm_c.c
View File

File diff suppressed because it is too large Load Diff