solve

    .intel_syntax noprefix
    .section    .rodata
.LC0:
    .string "%lld"
.LC1:
    .string "r"
.LC2:
    .string "w"
    .text
    .globl  solve
    .type   solve, @function
solve:
    push    rbp
    mov rbp, rsp
    sub rsp, 48
        push     rbx
        push     r10
        push     r12
        push     r13
        push     r14
        push     r15

    #mov    QWORD PTR -16[rbp], rdi # pass pointer to s into rbp-16
    mov r10, rdi # pass pointer to s into r10
    #mov    QWORD PTR -40[rbp], rsi # pass n into rbp-40
    mov r12, rsi # pass n into r12
    mov QWORD PTR -16[rbp], rdx # pass output into rcx

    #mov    QWORD PTR -32[rbp], 0 # p1
    mov r13, 0 # p1
        cmp     r12, 1
        jne     .L17
        mov     r13, rdi
.L17:
    add rdi, r12 # rdi = s + n #-40
    #mov    QWORD PTR -24[rbp], rdi # rbp-24 = p2 = s + n
    mov r14, rdi # r14 = p2 = s + n


    #mov    DWORD PTR -48[rbp], eax # rbp-48 = i = n-1
        mov rbx, r12
    dec rbx# rbx = i = n-1
    jmp .L4
.L9:
    lea rdx, -1[rbx]
    mov rax, r10 # rax = s
    add rax, rdx
    movzx   edx, BYTE PTR [rax] # edx = s[i-1]
    mov rax, rbx
    add rax, r10
    movzx   eax, BYTE PTR [rax] # rax = s[i]
    cmp dl, al # if (s[i - 1] < s[i])
    jge .L5
    lea r13, -1[rbx]
    add r13, r10 # p1 = s + i - 1
    jmp .L6
.L5: # else
    cmp r13, 0 # if (p1 != NULL)
    jne .L8 # break
    mov r14, rbx
    add r14, r10 # p2 = s + i
.L6:
    dec rbx # --i
.L4:
    cmp rbx, 0
    jg  .L9
.L8:
        cmp      r13, 0
        je      .L15
    #mov    QWORD PTR -8[rbp], 0 # i = 0
    mov r15, 0 # i = 0
    jmp .L10
.L11:
    lea rax, [r15+r13]
    movzx   eax, BYTE PTR [rax]
    movsx   eax, al
    mov edi, eax # arg1=p1[i]
    mov rsi, QWORD PTR -16[rbp] # arg2=output
    call    fputc@PLT # fputc=fprintf("%c",...)
    inc r15
.L10:
    lea rax, [r13+r15]
    cmp r14, rax
    jne .L11
.L15:
    pop      r15
    pop      r14
    pop      r13
    pop      r12
    pop      r10
    pop      rbx
    leave
    ret
    .size   solve, .-solve