Re: RFA: tolerate unavailable struct return values

[Jim Blandy <jimb@zwingli.cygnus.com>]

Elena Zannoni <ezannoni@cygnus.com> writes:
> Michael Snyder writes:
>  > Daniel Jacobowitz wrote:
>  > > On Fri, Nov 30, 2001 at 03:49:52PM -0500, Jim Blandy wrote:
>  > > > Daniel Jacobowitz <drow@mvista.com> writes:
>  > > > > On Thu, Nov 29, 2001 at 05:09:13PM -0500, Jim Blandy wrote:
>  > > > > >
>  > > > > > On some architectures, it's impossible for GDB to find structs
>  > > > > > returned by value.  These shouldn't be failures.  Should they be
>  > > > > > passes?
>  > > > >
>  > > > > Out of curiousity, which architectures?  And to be pedantic, I suspect
>  > > > > that it might be "not always possible" rather than actually
>  > > > > impossible.
>  > > >
>  > > > The one I have in mind is the S/390, although I'm pretty sure there
>  > > > are others.  I've included the bug report I sent to the S/390 GCC
>  > > > maintainers below.
>  > > >
>  > > > One approach would be to hope that the return buffer's address was
>  > > > still there in the register it was passed in.  But there's no way to
>  > > > tell when you're wrong.  GDB will just print garbage, and the user
>  > > > will think their program is wrong.  Better to simply say, "I can't
>  > > > find this information reliably", and let the user, who knows their
>  > > > program, find another way to get the info --- setting a breakpoint on
>  > > > the return statement, or looking at where the caller put the
>  > > > structure.
>  > > 
>  > > Hmmmm.  I wonder if MIPS could ever be affected by this?  I don't think
>  > > the MIPS ABI specifies that $a0 remains live.  It looks as if the value
>  > > of $a0 is always returned in $v0 in such functions, though.
>  > 
>  > It's not an uncommon problem, and I imagine we get it wrong a lot of the time.
> 
> Have you looked at the macro VALUE_RETURNED_FROM_STACK ? I defined that
> long time ago for hppa. It looks like the rs6000-tdep.c tries to deal
> with the same problem as well.
> 
> Maybe we should clean up that code, which came in as part of the HP
> merge :-(.

Looking at the following code in infcmd.c:

      /* We cannot determine the contents of the structure because
	 it is on the stack, and we don't know where, since we did not
	 initiate the call, as opposed to the call_function_by_hand case */
#ifdef VALUE_RETURNED_FROM_STACK
      value = 0;
#ifdef UI_OUT
      ui_out_text (uiout, "Value returned has type: ");
      ui_out_field_string (uiout, "return-type", TYPE_NAME (value_type));
      ui_out_text (uiout, ".");
      ui_out_text (uiout, " Cannot determine contents\n");
#else /* UI_OUT */
      printf_filtered ("Value returned has type: %s.", TYPE_NAME (value_type));
      printf_filtered (" Cannot determine contents\n");
#endif /* UI_OUT */
#else
      value = value_being_returned (value_type, stop_registers, structure_return);

and then at the following code in valops.c:

#ifdef VALUE_RETURNED_FROM_STACK
    if (struct_return)
      return (value_ptr) VALUE_RETURNED_FROM_STACK (value_type, struct_addr);
#endif

the stuff in infcmd.c looks backwards to me.  Wasn't the intention to
fall back to printing the type when VALUE_RETURNED_FROM_STACK is *not*
defined?

TYPE_NAME is just the structure tag; don't we want to print the
`struct' or `union' there?  The `type_print' function takes care of
that.

Frankly, I don't even think it's that useful to print the type.  The
user knows what function was called, and can use `ptype' or `whatis'
if they want to know the type.