[m-dev.] Re: C Standard or GCC bug?

[Fergus Henderson]

On 15-Jan-2001, Levi Paul Cameron <l.cameron2 at ugrad.unimelb.edu.au> wrote:
> 
>  Either I have reached the limit of the C standard or there is a bug in
> GCC:
> 
>  If you create the following program:
> 
> 
> #include <mercury_imp.h>
> 
> typedef struct {
> 	int a0;
> 	int b1;
> 	int c2;
> 	int d3;
> } DATA;
> 
> DATA func(void) {
> 	DATA d = {0, 1, 2, 3};
> 	return d;
> }
> 
> it compiles fine.
> However, as soon as you increase the data structre size
> (add another int to the data structure) GCC aborts with an
> error about si doing something it is not supposed to. Is this
> because it is being used already (si, di & bx all appear to
> be used as mercury registers) and this is a expected or is the
> C standard only supposed to allow returning small structures on
> the stack?

This is a bug/feature of GCC.  It occurs only if you use GCC's global
register variables extension.

> (I can work around it by mallocing memory and returning a
> pointer, but for the part I am using it, it is going to be more
> efficent if this doesn't have to happen)

A more efficient work-around is to use pass-by-reference rather than
return by value: pass a pointer to the struct.  Then you can use
MR_memcpy() to assign the result.  This avoids the need to use
malloc().

-- 
Fergus Henderson <fjh at cs.mu.oz.au>  |  "I have always known that the pursuit
                                    |  of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh>  |     -- the last words of T. S. Garp.
--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------