Last active
December 27, 2020 03:22
-
-
Save seisvelas/ad57cacb26e30af74408f2a97587d598 to your computer and use it in GitHub Desktop.
String with backspaces function in ASM - broken because I can't correctly move one byte between two pointers
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
global strclr | |
extern printf | |
extern malloc | |
section .text | |
strclr: | |
sub rsp, 8 | |
; callee reserved registers | |
push rbx | |
push rcx | |
push rdi | |
xor rax, rax | |
call printf | |
pop rdi | |
mov r8, rdi | |
mov r12, 1 | |
push rax | |
; r12: char count | |
xor rax, rax | |
final_char_count: | |
cmp BYTE [r8], 0 | |
je after_count | |
cmp BYTE [r8], '#' | |
je found_backspace | |
inc r12 | |
jmp count_epilogue | |
found_backspace: | |
cmp r12, 1 | |
je count_epilogue | |
dec r12 | |
count_epilogue: | |
inc r8 | |
jmp final_char_count | |
after_count: | |
pop rax | |
; copy first arg into r8 so we can save the address | |
; of where the string began (for our loop condition) | |
mov r8, rdi | |
; Get last character in string | |
mov rdx, 0 | |
get_string_end: | |
cmp BYTE [r8], 0 | |
je end_attained | |
inc r8 | |
inc rdx | |
jmp get_string_end | |
end_attained: | |
push rdi | |
push r8 | |
push rdx | |
xor rax, rax | |
push r12 | |
call malloc | |
mov rbx, rax | |
add rbx, r12 | |
add rsp, 8 | |
pop rdx | |
pop r8 | |
pop rdi | |
xor rax, rax ; RAX <- the result | |
mov BYTE [rbx], 0 ; null terminator for output buffer | |
mov r9, 0 | |
algorithm: | |
cmp rdx, 0 | |
jl return | |
cmp BYTE [r8], '#' | |
jne non_backspace | |
inc r9 | |
jmp algo_postscript | |
non_backspace: | |
cmp r9, 0 | |
je no_pending_backspaces | |
dec r9 | |
jmp algo_postscript | |
no_pending_backspaces: | |
dec rbx | |
mov al, BYTE [r8] | |
mov [rbx], al | |
algo_postscript: | |
dec r8 | |
dec rdx | |
jmp algorithm | |
return: | |
lea rax, [rbx] | |
; restore callee reserved registers | |
pop rcx | |
pop rbx | |
add rsp, 8 | |
ret |
Perhaps a simpler hack (that still isn't sufficient educationally, but still) would be to use AL
, which is the first byte of RAX!
Okay, fixed. The function works now, but crashes in the codewar's test because they try to free()
the pointer to the string this function returns. That's no good because this function is writing to a buffer in .bss, so free'ing the pointer is invalid. To fix this, I'll have to get memory dynamically (ie, mmap syscall or C malloc)
Grr, everything works fine on machine and in sample tests. Then real tests all crash. So, clearly in the real tests something else is being done with the pointers I return, or am given.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The crux of the brokenness is this section:
movsb probably doesn't work how I'm expecting it to. I need to learn how to do this correctly. Or, I could hack around it by using external C functions (which I have access to in this context, since it's a codewars problem) to reverse the string and copy stuff. But that's just punting the real problem down the line, which is my ignorance of how to transfer one byte between two pointers using raw assembly.