x86-64汇编输入验证问题：末尾校验致首字符值异常

Hey there, let's break down exactly why your code is mangling the first character when checking the end of the input, and get it working right.

The Root Cause #

Your logic for grabbing the last character of the input is completely off—this is why the first character gets corrupted:

mov al, [character]   ; You load the first character into al
add rax, r8           ; Then you add the input length (r8) to rax!

This isn't calculating the memory address of the last character—it's adding the length value directly to the ASCII value of the first character. That's why your first character gets messed up, and you're not actually checking the real last character at all.

How to Correctly Grab the Last Character #

To validate the final character properly without touching the first one, you need to:

1. Calculate the memory address of the last input byte: character + number_of_bytes_read - 1
2. Load the value from that address into a separate register
3. Run your validation checks on that new value

Fixed Full Code #

section .data
text1 db "ENTER TEXT: "
len_text1 equ $ - text1
text2 db "THIS IS WHAT YOU ENTERED: "
len_text2 equ $ - text2
text3 db "invalid message, keeping current "
len_text3 equ $ - text3

section .bss
character resb 255

section .text
global main
main:
 ; Print input prompt
 mov rax, 1
 mov rdi, 1
 mov rsi, text1
 mov rdx, len_text1
 syscall
 call validate

validate:
 ; Read user input from stdin
 mov rax, 0
 mov rdi, 0
 mov rsi, character
 mov rdx, 10
 syscall
 mov r8, rax  ; Store number of bytes read in r8

 ; Validate first character is uppercase A-Z
 mov al, [character]
 cmp al, 'A'
 jl invalid
 cmp al, 'Z'
 jg invalid

 ; Calculate address of the last input character
 mov rbx, character
 add rbx, r8
 sub rbx, 1  ; rbx now points to the final byte of input
 mov cl, [rbx]  ; Load last character into cl (separate from al)

 ; Check if last character is ! . or ?
 cmp cl, '!'
 je valid
 cmp cl, '.'
 je valid
 cmp cl, '?'
 je valid

 ; If we reach here, last character fails validation
 jmp invalid

valid:
 ; Print confirmation prompt
 mov rax, 1
 mov rdi, 1
 mov rsi, text2
 mov rdx, len_text2
 syscall
 ; Print the valid user input
 mov rax, 1
 mov rdi, 1
 mov rsi, character
 mov rdx, r8
 syscall
 jmp exit

invalid:
 ; Print error message
 mov rax, 1
 mov rdi, 1
 mov rsi, text3
 mov rdx, len_text3
 syscall
 ; Print original input to "keep current message"
 mov rax, 1
 mov rdi, 1
 mov rsi, character
 mov rdx, r8
 syscall
 jmp exit

exit:
 mov rax, 60
 xor rdi, rdi
 syscall

Key Fixes Explained #

1. Proper Last Character Addressing: We use rbx as a pointer to calculate the exact memory location of the last input byte, so we never modify the first character's original value.
2. Dynamic String Lengths: Instead of hardcoding lengths like 12, we use equ $ - label to calculate the exact length of each string. This fixes truncated error messages and makes the code more maintainable.
3. Cleaner Branch Logic: Split first-character and last-character validation into distinct steps, making the code easier to follow and debug.
4. Full Requirement Compliance: The invalid branch now prints both the error message and the original input, which matches your requirement to "keep current message" when input is invalid.

内容的提问来源于stack exchange，提问作者Kadi