using MASM 8086
program determines the largest value in the given arrays
Unhandled exception gets thrown when esi is increased. This does not appear to happen when largestVal is defined as a DWORD. However, this prevents .IF largestVal < ebx from working.
Include Irvine32.inc
FindLargest PROTO, pArr: PTR DWORD, Count: DWORD
.data
str1 BYTE "arr1: ", 0
str2 BYTE "arr2: ", 0
str3 BYTE "arr3: ", 0
arr1 SDWORD +1122h, +2233h, -3344h
arr2 SDWORD +2233h, +3344h, +4455h, +6677h
.code
main PROC
call Clrscr
INVOKE FindLargest, ADDR arr1, LENGTHOF arr1
mov edx, OFFSET str1
call WriteString
call WriteInt
INVOKE FindLargest, ADDR arr2, LENGTHOF arr2
mov edx, OFFSET str2
call WriteString
call WriteInt
exit
main ENDP
FindLargest PROC USES ebx ecx esi, pArr: PTR DWORD, Count: DWORD
LOCAL largestVal: SDWORD
mov esi, pArr
mov ecx, Count
mov largestVal, -2147483648
L1:
mov ebx, [esi]
.IF largestVal < ebx
mov largestVal, ebx
.ENDIF
add esi, 4 ; Where exception gets thrown
loop L1
mov eax, largestVal
ret
FindLargest ENDP
END main
It's a feature of VS2015. The code works fine with VS2013
Visual Studio 2015 stumbles over this line:
.IF largestVal < ebx
and produces total crap, amongst others the address ebp-7D for largestVal that is beyound the stack frame. Correct is ebp-4.
I've figured out two workarounds.
1) Decouple largestVal from the .IF directive:
...
mov eax, largestVal
.IF eax < ebx
mov largestVal, ebx
.ENDIF
...
2) Don't use the MASM preprocessor, use pure assembly instructions instead:
...
cmp largestVal, ebx
jge @F ; Jump forward to the next @@
mov largestVal, ebx
@@:
...