Index
> Main > Weird code problem |
| Author |
|
|
LocoDelAssembly 15 Aug 2007, 01:39
With "AND_______________________" commented compiles for me.
Which version are you using? The mine is 1.67.22 |
|||
15 Aug 2007, 01:39 |
|
|
shism2 15 Aug 2007, 02:08
No no no.... The and I put in their intentionally.... I wanted to know the difference between the two...
mov dword [ebx-1],-1 cmp eax,320 ret db 0dbh AND mov dword [dword ebx-1],-1 ; Why does this compile ? cmp eax,320 ret db 0dbh |
|||
|
15 Aug 2007, 02:08 |
|
|
r22 15 Aug 2007, 03:35
mov dword [dword ebx-1],-1
It compiles because it's technically correct. The [dword denotes that the address inside is a DWORD address which is true (being 32bit and all). Replace mov dword [dword ebx-1],-1 With mov word [word ebx-1],-1 And you'll get an "Address Size Error" |
|||
|
15 Aug 2007, 03:35 |
|
|
asmfan 15 Aug 2007, 06:58
Somethind like this. Makes compiler to produce the full form of instruction, not the shortest possible.
Code: .00401000: C743FFFFFFFFFF mov d,[ebx][-01],-1 .00401007: 3D40010000 cmp eax,000000140 ;' @' .0040100C: C3 retn .0040100D: C783FFFFFFFFFFFFFFFF mov d,[ebx][-00000001],-1 .00401017: 3D40010000 cmp eax,000000140 ;' @' .0040101C: C3 retn _________________ Any offers? |
|||
|
15 Aug 2007, 06:58 |
|
|
shism2 15 Aug 2007, 11:44
Asmfasm :
That's what you get when you look at the compiled code in a disasmbeller? |
|||
|
15 Aug 2007, 11:44 |
|
|
LocoDelAssembly 15 Aug 2007, 13:22
shism2, I tested your code as you posted but I commented that line for obvious reasons.
My output is the same as asmfan, here my output with OllyDbg: Code: 00401000 >/$ C743 FF FFFFFFFF MOV DWORD PTR DS:[EBX-1],-1 00401007 |. 3D 40010000 CMP EAX,140 0040100C \. C3 RETN 0040100D DB DB DB 0040100E . C783 FFFFFFFF FFFFFFFF MOV DWORD PTR DS:[EBX-1],-1 00401018 . 3D 40010000 CMP EAX,140 0040101D . C3 RETN 0040101E DB DB DB And to clarify even more I tested exactly this: Code: format PE gui mov dword [ebx-1],-1 cmp eax,320 ret db 0dbh ;AND_______________________ mov dword [dword ebx-1],-1 ; Why does this compile ? cmp eax,320 ret db 0dbh Note that removing format also compiles fine and using "use32" also works, etc. Seems to work always. I insist, which version are you using? Is exactly that code that gives you troubles? [edit]Sorry, I'm clearly seriously stupid here... The difference is the size of the offset. With "dword" inside the square bracket you prevent fasm from using the short signed 8-bit immediate form. You also prevent the suppression of the offset when it is zero ("mov [dword ebx+0], eax" is not converted into "mov [ebx], eax" while "mov [ebx+0], eax" yes). Sorry for misreading you[/edit] |
|||
|
15 Aug 2007, 13:22 |
|
|
DOS386 15 Aug 2007, 22:01
Code:
use32
mov dword [ebx-1],-1
cmp eax,320
ret
db 0dbh
nop
nop
nop
nop
nop
mov dword [dword ebx-1],-1 ; Why does this compile ?
cmp eax,320
ret
db 0dbh
nop
nop
nop
nop
nop
00000000 C743FFFFFFFFFF mov dword [ebx-1],$ffffffff
00000007 3D40010000 cmp eax,$140
0000000C C3 ret
0000000D DB9090909090 fist dword [eax+$90909090]
00000013 C783FFFFFFFFFFFF mov dword [ebx+$ffffffff],$ffffffff
-FFFF
0000001D 3D40010000 cmp eax,$140
00000022 C3 ret
00000023 DB9090909090 fist dword [eax+$90909090]
Works perfectly as supposed. The only "weird" thing is the syntax of some disassemblers. See correct FASM syntax coming from NDISASM.  _________________ Bug Nr.: 12345 Title: Hello World program compiles to 100 KB !!! Status: Closed: NOT a Bug |
|||
|
15 Aug 2007, 22:01 |
|
|
shism2 16 Aug 2007, 00:34
I didn't know about this addressing. Thank you everyone.
|
|||
|
16 Aug 2007, 00:34 |
|
< Last Thread | Next Thread > |
Forum Rules:
|