InjectAhkDll - Fix malformed VirtualFreeEx calls

[lvalkov]

MEM_RELEASE in DllCall("VirtualFreeEx","PTR",hProc,"PTR",pBufferRemote,"PTR",nDirLength,MEM_RELEASE) is missing its DWORD type. This causes the value of MEM_RELEASE to be incorrectly interpreted as being the calling convention/return type for the DllCall, causing InjectAhkDll to throw whenever VirtualFreeEx is called.

The Struct definition _MODULEENTRY32 was changed to use Format for populating the array sizes, instead of relying on the continuation section. I imagine you wanted to keep the V1 and V2 versions as similar as possible, but that's not how continuation sections work in V2. Instead of passing in the correct array size, a string is passed. The struct size reported later on by sizeof is, therefore, wrong, which is likely what has been causing the crashes on 64-bit.

The example has also been updated. It injects into notepad x86 on x86 systems, and notepad x86/64 on x64 systems, depending on the bitness of the interpreter.

I need to know what your intentions were regarding constructs such as:

return (DllCall("MessageBox", "Ptr", 0, "Str", "Could not reseve memory for process.", "Str", "Error", "UInt", 0)
      , DllCall("CloseHandle", "Ptr", hProc))

In V1 the result of MessageBox will be returned, whereas in V2, the result of evaluating the last expression will be returned. What did you intend to return here? I'd guess the MessageBox result. Is there an explicit reason for CloseHandle to be called after MessageBox? If not, I'd recommend restructuring these bits.

Lastly, I understand you want to keep both versions similar, but aren't checks like A_IsUnicode ? 2 : 1 pointless, since the V2 interpreters are Unicode only?

[HotKeyIt]

Many thanks ;)
The return MessageBox was just to show the message, return value is not important.