Reverse Engineering for Beginners

CHAPTER 52. NEGATIVE ARRAY INDICES CHAPTER 52. NEGATIVE ARRAY INDICES

23 $LN2@main:
24 mov ecx, DWORD PTR _i$[ebp]
25 add ecx, 1
26 mov DWORD PTR _i$[ebp], ecx
27 $LN3@main:
28 cmp DWORD PTR _i$[ebp], 10
29 jge SHORT $LN1@main
30 mov edx, DWORD PTR _i$[ebp]
31 mov al, BYTE PTR _i$[ebp]
32 mov BYTE PTR _array$[ebp+edx], al
33 jmp SHORT $LN2@main
34 $LN1@main:
35 mov ecx, DWORD PTR _fakearray$[ebp]
36 ; ecx=address of fakearray[0], ecx+1 is fakearray[1] or array[0]
37 movzx edx, BYTE PTR [ecx+1]
38 push edx
39 push OFFSET $SG2751 ; 'first element %d'
40 call _printf
41 add esp, 8
42 mov eax, DWORD PTR _fakearray$[ebp]
43 ; eax=address of fakearray[0], eax+2 is fakearray[2] or array[1]
44 movzx ecx, BYTE PTR [eax+2]
45 push ecx
46 push OFFSET $SG2752 ; 'second element %d'
47 call _printf
48 add esp, 8
49 mov edx, DWORD PTR _fakearray$[ebp]
50 ; edx=address of fakearray[0], edx+10 is fakearray[10] or array[9]
51 movzx eax, BYTE PTR [edx+10]
52 push eax
53 push OFFSET $SG2753 ; 'last element %d'
54 call _printf
55 add esp, 8
56 ; subtract 4, 3, 2 and 1 from pointer to array[0] in order to find values
before array[]
57 lea ecx, DWORD PTR _array$[ebp]
58 movzx edx, BYTE PTR [ecx-4]
59 push edx
60 lea eax, DWORD PTR _array$[ebp]
61 movzx ecx, BYTE PTR [eax-3]
62 push ecx
63 lea edx, DWORD PTR _array$[ebp]
64 movzx eax, BYTE PTR [edx-2]
65 push eax
66 lea ecx, DWORD PTR _array$[ebp]
67 movzx edx, BYTE PTR [ecx-1]
68 push edx
69 push OFFSET $SG2754 ; 'array[-1]=%02X, array[-2]=%02X, array[-3]=%02X, array[-4]=%02⤦
ÇX'
70 call _printf
71 add esp, 20
72 xor eax, eax
73 mov esp, ebp
74 pop ebp
75 ret 0
76 _main ENDP

So we havearray[]of ten elements, filled with 0 : : : 9 bytes. Then we have thefakearray[]pointer, which points

one byte beforearray[]. fakearray[1]points exactly toarray[0]. But we are still curious, what is there before

array[]? We have addedrandom_valuebeforearray[]and set it to0x11223344. The non-optimizing compiler

allocated the variables in the order they were declared, so yes, the 32-bitrandom_valueis right before the array.

We ran it, and:

first element 0

second element 1

last element 9

array[-1]=11, array[-2]=22, array[-3]=33, array[-4]=44

Here is the stack fragment we will copypaste from OllyDbg’s stack window (with comments added by the author):