Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!uwm.edu!zaphod.mps.ohio-state.edu!usc!ucsd!nosc!crash!ncr-sd!sagpd1!jharkins
From: jharkins@sagpd1.UUCP (Jim Harkins)
Newsgroups: comp.lang.c
Subject: Re: Return Graph varies from Call Graph?
Message-ID: <807@sagpd1.UUCP>
Date: 6 Jun 90 20:03:58 GMT
References: <1990Jun5.142836.9821@msuinfo.cl.msu.edu>
Reply-To: jharkins@sagpd1.UUCP (Jim Harkins)
Organization: Scientific Atlanta, Government Products Div, San Diego, CA
Lines: 26

In article <1990Jun5.142836.9821@msuinfo.cl.msu.edu> reid@cpswh.cps.msu.edu (Dr Richard J. Reid) writes:
>Do you have any suggestions about how to vary the
>return graph of function invocation from the call graph?

Lots of them, but this is a family newgroup :-)

>  Given the call sequence:     A -> B -> C -> D
>
>  Can the return sequence be:  A <- C <- B <- D  ?

I don't really see why you would want to do this.  If you did do it I think
that, once maintanance is factored in, you're much better off coming up with a
better algorithm for what you're trying to do.

A slightly more palatable thing to do would be to change the return sequence
to A <- B <- D (using your call sequence).  Then all you do is figure out how
many doo hickeys get put onto the stack when your C compiler does a function
call (can you say 'implementation dependent'.  I knew you could), then write
some in-line assembly to adjust the stack pointer.  Me?  If I ended up
maintaining code that did this sort of thing I'd look for a bucket of tar and
some feathers.

-- 
jim		jharkins@sagpd1

I hate to see you go, but I love to see you walk away.