Structure problem

[trancexx]

Unexpected behaviour with function DllStructureCreate()when created with pointer and "dword":

ConsoleWrite("Binary data is 0x3378FF453D1133" & @CRLF & @CRLF)

$structure_Binary_1 = DllStructCreate("byte[7]")
DllStructSetData($structure_Binary_1, 1, "0x3378FF453D1133"); that is 0x3378 and 0xFF453D11, last byte 0x33 is just for testing 

; Testing on "short;dword"
$struct_1 =  DllStructCreate("short;dword", DllStructGetPtr($structure_Binary_1))
ConsoleWrite("Before align:" & @CRLF)
ConsoleWrite(Binary(DllStructGetData($struct_1, 1)) & " <-- Binary(short) - ok" & @CRLF)
ConsoleWrite(Binary(DllStructGetData($struct_1, 2)) & " <-- Binary(dword) - 0xFF is missing and 0x33 is there" & @CRLF)

ConsoleWrite(@CRLF)


; New structure is "short;dword" with keyword align
$structure_Binary_2 = DllStructCreate("byte[7]")
DllStructSetData($structure_Binary_2, 1, "0x3378FF453D1133")

$struct_2 =  DllStructCreate("align 1;short;dword", DllStructGetPtr($structure_Binary_2)) ; or "align 2"
ConsoleWrite("After:" & @CRLF)
ConsoleWrite(Binary(DllStructGetData($struct_2, 1)) & " <-- Binary(short)" & @CRLF)
ConsoleWrite(Binary(DllStructGetData($struct_2, 2)) & " <-- Binary(dword), this appears ok" & @CRLF)


ConsoleWrite(@CRLF)


; Testing on "dword;short"
$structure_Binary_3 = DllStructCreate("byte[7]")
DllStructSetData($structure_Binary_3, 1, "0x3378FF453D1133")

$struct_3 =  DllStructCreate("dword;short", DllStructGetPtr($structure_Binary_3)) 
ConsoleWrite("Reoganized (no align):" & @CRLF)
ConsoleWrite(Binary(DllStructGetData($struct_3, 1)) & " <-- Binary(dword) - ok" & @CRLF)
ConsoleWrite(Binary(DllStructGetData($struct_3, 2)) & " <-- Binary(short) - ok" & @CRLF)

Thread on forum:
​http://www.autoitscript.com/forum/index.php?showtopic=85213

[Valik]

The code is correct. The helpfile almost explains why with this line:

DllStructCreate("short;int") ; structure is 8 bytes, the "int" is at offset 4

The key there is, the int (dword in your example) starts at byte 4 (0-index) and not at byte 2 like you are thinking.

So, we have the data 0x3378FF453D1133. The 0x3378FF45 portion is mapped to the short but only the first 2 bytes are accessible (0x3378). The 0xFF45 portion is mapped to the 2 bytes of padding that appears between the short and the dword. That means that the dword is 0x3D113300. Visually it looks like:
3378FF453D113300 (Extra 00's added to pad length for clarity).
SSSSPPPPDDDDDDDD (Where S = short, P = padding and D = dword).

When you start changing the alignment, you remove the padding and the structures align correctly.

This is not a bug.

[Valik]

Ooops, here's the memory layout expressed in a fixed-width font so you may see the mapping.

3378FF453D113300 (Extra 00's added to pad length for clarity).
SSSSPPPPDDDDDDDD (Where S = short, P = padding and D = dword).

[anonymous]

Briliant and simple explanation.

...and if I set data manually (like copying from one structure to another) problem cannot occure because it's filled 8 and 8 and 8...

Thanks!