Skip to content

Commit ef24243

Browse files
barracuda156barracuda156
committed
sp_int.c: fix ppc asm for macOS
1 parent b6bfae9 commit ef24243

1 file changed

Lines changed: 302 additions & 0 deletions

File tree

‎wolfcrypt/src/sp_int.c‎

Lines changed: 302 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -3477,6 +3477,156 @@ static WC_INLINE sp_int_digit sp_div_word(sp_int_digit hi, sp_int_digit lo,
34773477
* CPU: PPC64
34783478
*/
34793479

3480+
#ifdef __APPLE__
3481+
3482+
/* Multiply va by vb and store double size result in: vh | vl */
3483+
#define SP_ASM_MUL(vl, vh, va, vb) \
3484+
__asm__ __volatile__ ( \
3485+
"mulld %[l], %[a], %[b] \n\t" \
3486+
"mulhdu %[h], %[a], %[b] \n\t" \
3487+
: [h] "+r" (vh), [l] "+r" (vl) \
3488+
: [a] "r" (va), [b] "r" (vb) \
3489+
: "memory" \
3490+
)
3491+
/* Multiply va by vb and store double size result in: vo | vh | vl */
3492+
#define SP_ASM_MUL_SET(vl, vh, vo, va, vb) \
3493+
__asm__ __volatile__ ( \
3494+
"mulhdu %[h], %[a], %[b] \n\t" \
3495+
"mulld %[l], %[a], %[b] \n\t" \
3496+
"li %[o], 0 \n\t" \
3497+
: [l] "+r" (vl), [h] "+r" (vh), [o] "=r" (vo) \
3498+
: [a] "r" (va), [b] "r" (vb) \
3499+
: \
3500+
)
3501+
/* Multiply va by vb and add double size result into: vo | vh | vl */
3502+
#define SP_ASM_MUL_ADD(vl, vh, vo, va, vb) \
3503+
__asm__ __volatile__ ( \
3504+
"mulld r16, %[a], %[b] \n\t" \
3505+
"mulhdu r17, %[a], %[b] \n\t" \
3506+
"addc %[l], %[l], r16 \n\t" \
3507+
"adde %[h], %[h], r17 \n\t" \
3508+
"addze %[o], %[o] \n\t" \
3509+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3510+
: [a] "r" (va), [b] "r" (vb) \
3511+
: "r16", "r17", "cc" \
3512+
)
3513+
/* Multiply va by vb and add double size result into: vh | vl */
3514+
#define SP_ASM_MUL_ADD_NO(vl, vh, va, vb) \
3515+
__asm__ __volatile__ ( \
3516+
"mulld r16, %[a], %[b] \n\t" \
3517+
"mulhdu r17, %[a], %[b] \n\t" \
3518+
"addc %[l], %[l], r16 \n\t" \
3519+
"adde %[h], %[h], r17 \n\t" \
3520+
: [l] "+r" (vl), [h] "+r" (vh) \
3521+
: [a] "r" (va), [b] "r" (vb) \
3522+
: "r16", "r17", "cc" \
3523+
)
3524+
/* Multiply va by vb and add double size result twice into: vo | vh | vl */
3525+
#define SP_ASM_MUL_ADD2(vl, vh, vo, va, vb) \
3526+
__asm__ __volatile__ ( \
3527+
"mulld r16, %[a], %[b] \n\t" \
3528+
"mulhdu r17, %[a], %[b] \n\t" \
3529+
"addc %[l], %[l], r16 \n\t" \
3530+
"adde %[h], %[h], r17 \n\t" \
3531+
"addze %[o], %[o] \n\t" \
3532+
"addc %[l], %[l], r16 \n\t" \
3533+
"adde %[h], %[h], r17 \n\t" \
3534+
"addze %[o], %[o] \n\t" \
3535+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3536+
: [a] "r" (va), [b] "r" (vb) \
3537+
: "r16", "r17", "cc" \
3538+
)
3539+
/* Multiply va by vb and add double size result twice into: vo | vh | vl
3540+
* Assumes first add will not overflow vh | vl
3541+
*/
3542+
#define SP_ASM_MUL_ADD2_NO(vl, vh, vo, va, vb) \
3543+
__asm__ __volatile__ ( \
3544+
"mulld r16, %[a], %[b] \n\t" \
3545+
"mulhdu r17, %[a], %[b] \n\t" \
3546+
"addc %[l], %[l], r16 \n\t" \
3547+
"adde %[h], %[h], r17 \n\t" \
3548+
"addc %[l], %[l], r16 \n\t" \
3549+
"adde %[h], %[h], r17 \n\t" \
3550+
"addze %[o], %[o] \n\t" \
3551+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3552+
: [a] "r" (va), [b] "r" (vb) \
3553+
: "r16", "r17", "cc" \
3554+
)
3555+
/* Square va and store double size result in: vh | vl */
3556+
#define SP_ASM_SQR(vl, vh, va) \
3557+
__asm__ __volatile__ ( \
3558+
"mulld %[l], %[a], %[a] \n\t" \
3559+
"mulhdu %[h], %[a], %[a] \n\t" \
3560+
: [h] "+r" (vh), [l] "+r" (vl) \
3561+
: [a] "r" (va) \
3562+
: "memory" \
3563+
)
3564+
/* Square va and add double size result into: vo | vh | vl */
3565+
#define SP_ASM_SQR_ADD(vl, vh, vo, va) \
3566+
__asm__ __volatile__ ( \
3567+
"mulld r16, %[a], %[a] \n\t" \
3568+
"mulhdu r17, %[a], %[a] \n\t" \
3569+
"addc %[l], %[l], r16 \n\t" \
3570+
"adde %[h], %[h], r17 \n\t" \
3571+
"addze %[o], %[o] \n\t" \
3572+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3573+
: [a] "r" (va) \
3574+
: "r16", "r17", "cc" \
3575+
)
3576+
/* Square va and add double size result into: vh | vl */
3577+
#define SP_ASM_SQR_ADD_NO(vl, vh, va) \
3578+
__asm__ __volatile__ ( \
3579+
"mulld r16, %[a], %[a] \n\t" \
3580+
"mulhdu r17, %[a], %[a] \n\t" \
3581+
"addc %[l], %[l], r16 \n\t" \
3582+
"adde %[h], %[h], r17 \n\t" \
3583+
: [l] "+r" (vl), [h] "+r" (vh) \
3584+
: [a] "r" (va) \
3585+
: "r16", "r17", "cc" \
3586+
)
3587+
/* Add va into: vh | vl */
3588+
#define SP_ASM_ADDC(vl, vh, va) \
3589+
__asm__ __volatile__ ( \
3590+
"addc %[l], %[l], %[a] \n\t" \
3591+
"addze %[h], %[h] \n\t" \
3592+
: [l] "+r" (vl), [h] "+r" (vh) \
3593+
: [a] "r" (va) \
3594+
: "cc" \
3595+
)
3596+
/* Sub va from: vh | vl */
3597+
#define SP_ASM_SUBB(vl, vh, va) \
3598+
__asm__ __volatile__ ( \
3599+
"subfc %[l], %[a], %[l] \n\t" \
3600+
"li r16, 0 \n\t" \
3601+
"subfe %[h], r16, %[h] \n\t" \
3602+
: [l] "+r" (vl), [h] "+r" (vh) \
3603+
: [a] "r" (va) \
3604+
: "r16", "cc" \
3605+
)
3606+
/* Add two times vc | vb | va into vo | vh | vl */
3607+
#define SP_ASM_ADD_DBL_3(vl, vh, vo, va, vb, vc) \
3608+
__asm__ __volatile__ ( \
3609+
"addc %[l], %[l], %[a] \n\t" \
3610+
"adde %[h], %[h], %[b] \n\t" \
3611+
"adde %[o], %[o], %[c] \n\t" \
3612+
"addc %[l], %[l], %[a] \n\t" \
3613+
"adde %[h], %[h], %[b] \n\t" \
3614+
"adde %[o], %[o], %[c] \n\t" \
3615+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3616+
: [a] "r" (va), [b] "r" (vb), [c] "r" (vc) \
3617+
: "cc" \
3618+
)
3619+
/* Count leading zeros. */
3620+
#define SP_ASM_LZCNT(va, vn) \
3621+
__asm__ __volatile__ ( \
3622+
"cntlzd %[n], %[a] \n\t" \
3623+
: [n] "=r" (vn) \
3624+
: [a] "r" (va) \
3625+
: \
3626+
)
3627+
3628+
#else /* !defined(__APPLE__) */
3629+
34803630
/* Multiply va by vb and store double size result in: vh | vl */
34813631
#define SP_ASM_MUL(vl, vh, va, vb) \
34823632
__asm__ __volatile__ ( \
@@ -3623,6 +3773,8 @@ static WC_INLINE sp_int_digit sp_div_word(sp_int_digit hi, sp_int_digit lo,
36233773
: \
36243774
)
36253775

3776+
#endif /* !defined(__APPLE__) */
3777+
36263778
#define SP_INT_ASM_AVAILABLE
36273779

36283780
#endif /* WOLFSSL_SP_PPC64 && SP_WORD_SIZE == 64 */
@@ -3632,6 +3784,154 @@ static WC_INLINE sp_int_digit sp_div_word(sp_int_digit hi, sp_int_digit lo,
36323784
* CPU: PPC 32-bit
36333785
*/
36343786

3787+
#ifdef __APPLE__
3788+
3789+
/* Multiply va by vb and store double size result in: vh | vl */
3790+
#define SP_ASM_MUL(vl, vh, va, vb) \
3791+
__asm__ __volatile__ ( \
3792+
"mullw %[l], %[a], %[b] \n\t" \
3793+
"mulhwu %[h], %[a], %[b] \n\t" \
3794+
: [h] "+r" (vh), [l] "+r" (vl) \
3795+
: [a] "r" (va), [b] "r" (vb) \
3796+
: "memory" \
3797+
)
3798+
/* Multiply va by vb and store double size result in: vo | vh | vl */
3799+
#define SP_ASM_MUL_SET(vl, vh, vo, va, vb) \
3800+
__asm__ __volatile__ ( \
3801+
"mulhwu %[h], %[a], %[b] \n\t" \
3802+
"mullw %[l], %[a], %[b] \n\t" \
3803+
"li %[o], 0 \n\t" \
3804+
: [l] "+r" (vl), [h] "+r" (vh), [o] "=r" (vo) \
3805+
: [a] "r" (va), [b] "r" (vb) \
3806+
)
3807+
/* Multiply va by vb and add double size result into: vo | vh | vl */
3808+
#define SP_ASM_MUL_ADD(vl, vh, vo, va, vb) \
3809+
__asm__ __volatile__ ( \
3810+
"mullw r16, %[a], %[b] \n\t" \
3811+
"mulhwu r17, %[a], %[b] \n\t" \
3812+
"addc %[l], %[l], r16 \n\t" \
3813+
"adde %[h], %[h], r17 \n\t" \
3814+
"addze %[o], %[o] \n\t" \
3815+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3816+
: [a] "r" (va), [b] "r" (vb) \
3817+
: "r16", "r17", "cc" \
3818+
)
3819+
/* Multiply va by vb and add double size result into: vh | vl */
3820+
#define SP_ASM_MUL_ADD_NO(vl, vh, va, vb) \
3821+
__asm__ __volatile__ ( \
3822+
"mullw r16, %[a], %[b] \n\t" \
3823+
"mulhwu r17, %[a], %[b] \n\t" \
3824+
"addc %[l], %[l], r16 \n\t" \
3825+
"adde %[h], %[h], r17 \n\t" \
3826+
: [l] "+r" (vl), [h] "+r" (vh) \
3827+
: [a] "r" (va), [b] "r" (vb) \
3828+
: "r16", "r17", "cc" \
3829+
)
3830+
/* Multiply va by vb and add double size result twice into: vo | vh | vl */
3831+
#define SP_ASM_MUL_ADD2(vl, vh, vo, va, vb) \
3832+
__asm__ __volatile__ ( \
3833+
"mullw r16, %[a], %[b] \n\t" \
3834+
"mulhwu r17, %[a], %[b] \n\t" \
3835+
"addc %[l], %[l], r16 \n\t" \
3836+
"adde %[h], %[h], r17 \n\t" \
3837+
"addze %[o], %[o] \n\t" \
3838+
"addc %[l], %[l], r16 \n\t" \
3839+
"adde %[h], %[h], r17 \n\t" \
3840+
"addze %[o], %[o] \n\t" \
3841+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3842+
: [a] "r" (va), [b] "r" (vb) \
3843+
: "r16", "r17", "cc" \
3844+
)
3845+
/* Multiply va by vb and add double size result twice into: vo | vh | vl
3846+
* Assumes first add will not overflow vh | vl
3847+
*/
3848+
#define SP_ASM_MUL_ADD2_NO(vl, vh, vo, va, vb) \
3849+
__asm__ __volatile__ ( \
3850+
"mullw r16, %[a], %[b] \n\t" \
3851+
"mulhwu r17, %[a], %[b] \n\t" \
3852+
"addc %[l], %[l], r16 \n\t" \
3853+
"adde %[h], %[h], r17 \n\t" \
3854+
"addc %[l], %[l], r16 \n\t" \
3855+
"adde %[h], %[h], r17 \n\t" \
3856+
"addze %[o], %[o] \n\t" \
3857+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3858+
: [a] "r" (va), [b] "r" (vb) \
3859+
: "r16", "r17", "cc" \
3860+
)
3861+
/* Square va and store double size result in: vh | vl */
3862+
#define SP_ASM_SQR(vl, vh, va) \
3863+
__asm__ __volatile__ ( \
3864+
"mullw %[l], %[a], %[a] \n\t" \
3865+
"mulhwu %[h], %[a], %[a] \n\t" \
3866+
: [h] "+r" (vh), [l] "+r" (vl) \
3867+
: [a] "r" (va) \
3868+
: "memory" \
3869+
)
3870+
/* Square va and add double size result into: vo | vh | vl */
3871+
#define SP_ASM_SQR_ADD(vl, vh, vo, va) \
3872+
__asm__ __volatile__ ( \
3873+
"mullw r16, %[a], %[a] \n\t" \
3874+
"mulhwu r17, %[a], %[a] \n\t" \
3875+
"addc %[l], %[l], r16 \n\t" \
3876+
"adde %[h], %[h], r17 \n\t" \
3877+
"addze %[o], %[o] \n\t" \
3878+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3879+
: [a] "r" (va) \
3880+
: "r16", "r17", "cc" \
3881+
)
3882+
/* Square va and add double size result into: vh | vl */
3883+
#define SP_ASM_SQR_ADD_NO(vl, vh, va) \
3884+
__asm__ __volatile__ ( \
3885+
"mullw r16, %[a], %[a] \n\t" \
3886+
"mulhwu r17, %[a], %[a] \n\t" \
3887+
"addc %[l], %[l], r16 \n\t" \
3888+
"adde %[h], %[h], r17 \n\t" \
3889+
: [l] "+r" (vl), [h] "+r" (vh) \
3890+
: [a] "r" (va) \
3891+
: "r16", "r17", "cc" \
3892+
)
3893+
/* Add va into: vh | vl */
3894+
#define SP_ASM_ADDC(vl, vh, va) \
3895+
__asm__ __volatile__ ( \
3896+
"addc %[l], %[l], %[a] \n\t" \
3897+
"addze %[h], %[h] \n\t" \
3898+
: [l] "+r" (vl), [h] "+r" (vh) \
3899+
: [a] "r" (va) \
3900+
: "cc" \
3901+
)
3902+
/* Sub va from: vh | vl */
3903+
#define SP_ASM_SUBB(vl, vh, va) \
3904+
__asm__ __volatile__ ( \
3905+
"subfc %[l], %[a], %[l] \n\t" \
3906+
"li r16, 0 \n\t" \
3907+
"subfe %[h], r16, %[h] \n\t" \
3908+
: [l] "+r" (vl), [h] "+r" (vh) \
3909+
: [a] "r" (va) \
3910+
: "r16", "cc" \
3911+
)
3912+
/* Add two times vc | vb | va into vo | vh | vl */
3913+
#define SP_ASM_ADD_DBL_3(vl, vh, vo, va, vb, vc) \
3914+
__asm__ __volatile__ ( \
3915+
"addc %[l], %[l], %[a] \n\t" \
3916+
"adde %[h], %[h], %[b] \n\t" \
3917+
"adde %[o], %[o], %[c] \n\t" \
3918+
"addc %[l], %[l], %[a] \n\t" \
3919+
"adde %[h], %[h], %[b] \n\t" \
3920+
"adde %[o], %[o], %[c] \n\t" \
3921+
: [l] "+r" (vl), [h] "+r" (vh), [o] "+r" (vo) \
3922+
: [a] "r" (va), [b] "r" (vb), [c] "r" (vc) \
3923+
: "cc" \
3924+
)
3925+
/* Count leading zeros. */
3926+
#define SP_ASM_LZCNT(va, vn) \
3927+
__asm__ __volatile__ ( \
3928+
"cntlzw %[n], %[a] \n\t" \
3929+
: [n] "=r" (vn) \
3930+
: [a] "r" (va) \
3931+
)
3932+
3933+
#else /* !defined(__APPLE__) */
3934+
36353935
/* Multiply va by vb and store double size result in: vh | vl */
36363936
#define SP_ASM_MUL(vl, vh, va, vb) \
36373937
__asm__ __volatile__ ( \
@@ -3776,6 +4076,8 @@ static WC_INLINE sp_int_digit sp_div_word(sp_int_digit hi, sp_int_digit lo,
37764076
: [a] "r" (va) \
37774077
)
37784078

4079+
#endif /* !defined(__APPLE__) */
4080+
37794081
#define SP_INT_ASM_AVAILABLE
37804082

37814083
#endif /* WOLFSSL_SP_PPC && SP_WORD_SIZE == 64 */

0 commit comments

Comments
 (0)