michael@0: /* This Source Code Form is subject to the terms of the Mozilla Public
michael@0:  * License, v. 2.0. If a copy of the MPL was not distributed with this
michael@0:  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
michael@0: 
michael@0: /* This inlined version is for 32-bit ARM platform only */
michael@0: 
michael@0: #if !defined(__arm__)
michael@0: #error "This is for ARM only"
michael@0: #endif
michael@0: 
michael@0: /* 16-bit thumb doesn't work inlined assember version */
michael@0: #if (!defined(__thumb__) || defined(__thumb2__)) && !defined(__ARM_ARCH_3__)
michael@0: 
michael@0: #include "mpi-priv.h"
michael@0: 
michael@0: #ifdef MP_ASSEMBLY_MULTIPLY
michael@0: void s_mpv_mul_d(const mp_digit *a, mp_size a_len, mp_digit b, mp_digit *c)
michael@0: {
michael@0:   __asm__ __volatile__(
michael@0:     "mov     r5, #0\n"
michael@0: #ifdef __thumb2__
michael@0:     "cbz     %1, 2f\n"
michael@0: #else
michael@0:     "cmp     %1, r5\n" /* r5 is 0 now */
michael@0:     "beq     2f\n"
michael@0: #endif
michael@0: 
michael@0:     "1:\n"
michael@0:     "mov     r4, #0\n"
michael@0:     "ldr     r6, [%0], #4\n"
michael@0:     "umlal   r5, r4, r6, %2\n"
michael@0:     "str     r5, [%3], #4\n"
michael@0:     "mov     r5, r4\n"
michael@0: 
michael@0:     "subs    %1, #1\n"
michael@0:     "bne     1b\n"
michael@0: 
michael@0:     "2:\n"
michael@0:     "str     r5, [%3]\n"
michael@0:   :
michael@0:   : "r"(a), "r"(a_len), "r"(b), "r"(c)
michael@0:   : "memory", "cc", "%r4", "%r5", "%r6");
michael@0: }
michael@0: 
michael@0: void s_mpv_mul_d_add(const mp_digit *a, mp_size a_len, mp_digit b, mp_digit *c)
michael@0: {
michael@0:   __asm__ __volatile__(
michael@0:     "mov     r5, #0\n"
michael@0: #ifdef __thumb2__
michael@0:     "cbz     %1, 2f\n"
michael@0: #else
michael@0:     "cmp     %1, r5\n" /* r5 is 0 now */
michael@0:     "beq     2f\n"
michael@0: #endif
michael@0: 
michael@0:     "1:\n"
michael@0:     "mov     r4, #0\n"
michael@0:     "ldr     r6, [%3]\n"
michael@0:     "adds    r5, r6\n"
michael@0:     "adc     r4, r4, #0\n"
michael@0: 
michael@0:     "ldr     r6, [%0], #4\n"
michael@0:     "umlal   r5, r4, r6, %2\n"
michael@0:     "str     r5, [%3], #4\n"
michael@0:     "mov     r5, r4\n"
michael@0: 
michael@0:     "subs    %1, #1\n"
michael@0:     "bne     1b\n"
michael@0: 
michael@0:     "2:\n"
michael@0:     "str     r5, [%3]\n"
michael@0:     :
michael@0:     : "r"(a), "r"(a_len), "r"(b), "r"(c)
michael@0:     : "memory", "cc", "%r4", "%r5", "%r6");
michael@0: }
michael@0: 
michael@0: void s_mpv_mul_d_add_prop(const mp_digit *a, mp_size a_len, mp_digit b, mp_digit *c)
michael@0: {
michael@0:   if (!a_len)
michael@0:     return;
michael@0: 
michael@0:   __asm__ __volatile__(
michael@0:     "mov     r5, #0\n"
michael@0: 
michael@0:     "1:\n"
michael@0:     "mov     r4, #0\n"
michael@0:     "ldr     r6, [%3]\n"
michael@0:     "adds    r5, r6\n"
michael@0:     "adc     r4, r4, #0\n"
michael@0:     "ldr     r6, [%0], #4\n"
michael@0:     "umlal   r5, r4, r6, %2\n"
michael@0:     "str     r5, [%3], #4\n"
michael@0:     "mov     r5, r4\n"
michael@0: 
michael@0:     "subs    %1, #1\n"
michael@0:     "bne     1b\n"
michael@0: 
michael@0: #ifdef __thumb2__
michael@0:     "cbz     r4, 3f\n"
michael@0: #else
michael@0:     "cmp     r4, #0\n"
michael@0:     "beq     3f\n"
michael@0: #endif
michael@0: 
michael@0:     "2:\n"
michael@0:     "mov     r4, #0\n"
michael@0:     "ldr     r6, [%3]\n"
michael@0:     "adds    r5, r6\n"
michael@0:     "adc     r4, r4, #0\n"
michael@0:     "str     r5, [%3], #4\n"
michael@0:     "movs    r5, r4\n"
michael@0:     "bne     2b\n"
michael@0: 
michael@0:     "3:\n"
michael@0:     :
michael@0:     : "r"(a), "r"(a_len), "r"(b), "r"(c)
michael@0:     : "memory", "cc", "%r4", "%r5", "%r6");
michael@0: }
michael@0: #endif
michael@0: 
michael@0: #ifdef MP_ASSEMBLY_SQUARE
michael@0: void s_mpv_sqr_add_prop(const mp_digit *pa, mp_size a_len, mp_digit *ps)
michael@0: {
michael@0:   if (!a_len)
michael@0:     return;
michael@0: 
michael@0:   __asm__ __volatile__(
michael@0:     "mov     r3, #0\n"
michael@0: 
michael@0:     "1:\n"
michael@0:     "mov     r4, #0\n"
michael@0:     "ldr     r6, [%0], #4\n"
michael@0:     "ldr     r5, [%2]\n"
michael@0:     "adds    r3, r5\n"
michael@0:     "adc     r4, r4, #0\n"
michael@0:     "umlal   r3, r4, r6, r6\n" /* w = r3:r4 */
michael@0:     "str     r3, [%2], #4\n"
michael@0: 
michael@0:     "ldr     r5, [%2]\n"
michael@0:     "adds    r3, r4, r5\n"
michael@0:     "mov     r4, #0\n"
michael@0:     "adc     r4, r4, #0\n"
michael@0:     "str     r3, [%2], #4\n"
michael@0:     "mov     r3, r4\n"
michael@0: 
michael@0:     "subs    %1, #1\n"
michael@0:     "bne     1b\n"
michael@0: 
michael@0: #ifdef __thumb2__
michael@0:     "cbz     r3, 3f\n"
michael@0: #else
michael@0:     "cmp     r3, #0\n"
michael@0:     "beq     3f\n"
michael@0: #endif
michael@0: 
michael@0:     "2:\n"
michael@0:     "mov     r4, #0\n"
michael@0:     "ldr     r5, [%2]\n"
michael@0:     "adds    r3, r5\n"
michael@0:     "adc     r4, r4, #0\n"
michael@0:     "str     r3, [%2], #4\n"
michael@0:     "movs    r3, r4\n"
michael@0:     "bne     2b\n"
michael@0: 
michael@0:     "3:"
michael@0:     :
michael@0:     : "r"(pa), "r"(a_len), "r"(ps)
michael@0:     : "memory", "cc", "%r3", "%r4", "%r5", "%r6");
michael@0: }
michael@0: #endif
michael@0: #endif