/*
 * Copyright (c) 2020 Raspberry Pi (Trading) Ltd.
 *
 * SPDX-License-Identifier: BSD-3-Clause
 */

.syntax unified
.cpu cortex-m0plus
.thumb

#include "pico/asm_helper.S"
__pre_init __aeabi_bits_init, 00010

.macro bits_section name
#if PICO_BITS_IN_RAM
.section RAM_SECTION_NAME(\name), "ax"
#else
.section SECTION_NAME(\name), "ax"
#endif
.endm

.section .data.aeabi_bits_funcs
.global aeabi_bits_funcs, aeabi_bits_funcs_end
.equ BITS_FUNC_COUNT, 4
.align 4
aeabi_bits_funcs:
    .word rom_table_code('P','3') // popcount32
    .word rom_table_code('L','3') // clz32
    .word rom_table_code('T','3') // ctz32
    .word rom_table_code('R','3') // reverse32
aeabi_bits_funcs_end:

.section .text
.thumb_func
__aeabi_bits_init:
    ldr r0, =aeabi_bits_funcs
    movs r1, #BITS_FUNC_COUNT
    ldr r3, =rom_funcs_lookup
    bx r3

.equ POPCOUNT32, 0
.equ CLZ32, 4
.equ CTZ32, 8
.equ REVERSE32, 12

bits_section clzsi
wrapper_func __clz
wrapper_func __clzl
wrapper_func __clzsi2
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #CLZ32]
    bx r3

bits_section ctzsi
wrapper_func __ctzsi2
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #CTZ32]
    bx r3

bits_section popcountsi
wrapper_func __popcountsi2
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #POPCOUNT32]
    bx r3

bits_section clzdi
wrapper_func __clzll
wrapper_func __clzdi2
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #CLZ32]
    cmp r1, #0
    bne 1f
    push {lr}
    blx r3
    adds r0, #32
    pop {pc}
1:
    mov r0, r1
    bx r3

bits_section ctzdi
wrapper_func __ctzdi2
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #CTZ32]
    cmp r0, #0
    bne 1f
    bx r3
1:
    push {lr}
    mov r0, r1
    blx r3
    adds r0, #32
    pop {pc}

bits_section popcountdi
wrapper_func __popcountdi2
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #POPCOUNT32]
    push {r1, r3, lr}
    blx r3
    mov ip, r0
    pop {r0, r3}
    blx r3
    mov r1, ip
    add r0, r1
    pop {pc}

bits_section reverse32
regular_func reverse32
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #REVERSE32]
    bx r3

bits_section __rev
regular_func __rev
regular_func __revl
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #REVERSE32]
    bx r3

bits_section __revll
regular_func __revll
    push {lr}
    ldr r3, =aeabi_bits_funcs
    ldr r3, [r3, #REVERSE32]
    push {r1, r3}
    blx r3
    mov ip, r0 // reverse32 preserves ip
    pop {r0, r3}
    blx r3
    mov r1, ip
    pop {pc}