AT&T汇编字符数组遍历问题：简化版sprintf的%格式符定位

Got it, let's work through converting that Intel syntax to AT&T for your simplified sprintf task—this is a common pain point, but once you nail the key syntax differences, it clicks fast.

First, let's recap the core syntax gaps between Intel and AT&T that matter for your code:

• Registers: AT&T requires a % prefix (e.g., %edx instead of edx)
• Immediates: AT&T requires a $ prefix (e.g., $0 instead of 0)
• Instruction suffixes: AT&T uses suffixes to specify operand size (b for byte, l for 32-bit word) — your Intel code already uses cmpb, so we'll keep that suffix
• Memory addressing: Intel’s (base, index, scale) translates directly to AT&T’s (base, index, scale) (no brackets, just parentheses)
• Jump mnemonics: Most jump instructions (like je) are identical between the two syntaxes

Converting Your Snippet #

Here’s your original Intel code snippet, converted to AT&T, with explanations:

Intel Original

find_index: cmpb $0, (%edx, %ecx, 1) #check if end of string
je reslength #go to next part of the code if true
cmpb $'%', (%edx, %ecx, 1) #check if byte = %
inc ecx # increment index counter
...

AT&T Conversion

find_index:
    cmpb    $0, (%edx, %ecx, 1)  # Check if we've hit the null terminator (end of string)
    je      reslength             # Jump to result handling if we're done
    cmpb    $'%', (%edx, %ecx, 1) # Check if current byte is the format specifier '%'
    incl    %ecx                  # Increment our index counter to move to the next character
    # Add your loop continuation or format handling logic here

Full Loop Example for Your sprintf Task #

To make this useful for your string traversal, here’s a complete loop that either finds % or reaches the end of the string:

find_index:
    cmpb    $0, (%edx, %ecx, 1)  # Check for null terminator
    je      reslength             # Exit loop if we've reached the end of the string
    cmpb    $'%', (%edx, %ecx, 1) # Check if current byte is the format trigger '%'
    je      handle_format         # Jump to your format specifier handling code if found
    incl    %ecx                  # Move to next character if not '%'
    jmp     find_index            # Repeat the loop

handle_format:
    # Insert your code to process the format specifier here
    # Don't forget to increment %ecx past the '%' before continuing the loop!
    incl    %ecx
    jmp     find_index

reslength:
    # Insert your code to handle the end of the string (e.g., return the length stored in %ecx)

Quick Notes for Your Context #

• %edx should hold the starting pointer to your input string (the char* you’re traversing)
• %ecx acts as your index counter, starting at 0
• When you hit %, jump to handle_format to process whatever comes next (like d, s, etc.) before resuming the loop

内容的提问来源于stack exchange，提问作者Telanore