Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!usc!jarthur!nntp-server.caltech.edu!tybalt.caltech.edu!chi
From: chi@tybalt.caltech.edu (Curt Hagenlocher)
Newsgroups: comp.sys.mac.programmer
Subject: Re: passing correct var size in MPW C/C++
Summary: MPW C always passes 32-bit stuff
Message-ID: <1990Jul24.235007.20545@laguna.ccsf.caltech.edu>
Date: 24 Jul 90 23:50:07 GMT
References: <26ACA0CB.542F@wilbur.coyote.trw.com> <43315@apple.Apple.COM>
Sender: chi@tybalt.caltech.edu
Distribution: usa
Organization: California Institute of Technology, Pasadena
Lines: 27

In article <43315@apple.Apple.COM> marc@Apple.COM (Mark Dawson) writes:
>I wrote a routine in assembly that I wanted to call from my C++ code, it
>didn't end up working the way I expected--the C++ call kept passing a long,
>when my assembly was expecting a short (2 bytes).  I've of course modifed
>my assembly to expect a long, but I'd like to know where my thinking went
>wrong.
>
>(code deleted)
>
>When I did a debugger trap in myProc(), the stack always had an $FFFF FFFE,
>not the $FFFE that I expected--it seemed to be passing 4 bytes instead of two.
>Why?
>

MPW C always converts everything to 32 bits that's less than
32 bits in length.  This includes characters and shorts.
The information is put into the low (byte/word) and sign extended
to make a long integer.

I imagine this is done so that everything stays longword aligned.

This information can be found in the chapter of the MPW C manual
entitled ``Calling Conventions.''  I believe it is Appendix B.

---
chi@tybalt.caltech.edu
Curt Hagenlocher