Re: [PATCH v2] ARM: __attribute__ ((reg_return))

[Zack Weinberg <zack at codesourcery dot com>]

Paul Brook wrote:
> > I really don't think it is possible to make it a function attribute, for
> > one language reason and one QoI reason.  The language reason:  if we did
> > it your way, this fragment would be acceptable code:
> >
> >     struct S { int a, b; };
> >     struct S foo(void) __attribute__ ((reg_return));
> >     struct S (*ptr)(void) = foo;
> >
> > But any caller of 'ptr' will be looking for the return value in the
> > normal place, not where 'foo' will put it.  Therefore, that last
> > assignment needs to be a type error.  My way, we get that for free.  I
> > don't believe we have any way of requiring that function attributes
> > match in function pointer assignment.
>
> This can't be a new problem. x86 has at least 2 function attributes that 
> effect the ABI (stdcall and regparm).

And on x86, assigning a regparm function to a non-regparm function 
pointer does get me an 'incompatible types in assignment' warning.  I'll 
 have a look how x86 does that and maybe I can figure something out.

> > The QoI reason: any function that returns a structure must (yes, I know
> > there are exceptions) declare a local variable with the same type as the
> > structure.  On ARM, local variables with a structure type larger than
> > two words are normally put on the stack.  This produces suboptimal code
> > when the structure is going to be returned in registers; GCC is (even in
> > 4.3) not able to eliminate all the memory accesses or the stack slot.
> > This is why I invented the override_record_mode target hook, to force
> > reg_return structures to be kept in registers.  Obviously, if we make
> > the attribute be on the function, there is no way to do that.
>
> Yuck. This suggests that the attribute is somewhat mis-named. It effects code 
> generation as well as the ABI. On a related note, we need documentation for 
> the new attribute.

Well, I'd love to avoid the target hook and just make the optimizer 
smarter, do you have any thoughts on that?

zw