Re: [PATCH v2] guile: fix make-value with pointer type

[George Barrett <bob@bob131.so>]

On Wed, Jul 28, 2021 at 10:07:26PM -0400, Simon Marchi wrote:
> On 2021-06-06 3:56 p.m., George Barrett via Gdb-patches wrote:
> > @@ -1931,6 +1931,22 @@ get_signed_type_minmax (struct type *type, LONGEST *min, LONGEST *max)
> >    *max = ((ULONGEST) 1 << (n - 1)) - 1;
> >  }
> >  
> > +/* Store in *MAX the largest value representable by pointer type
> > +   TYPE. */
> > +
> > +void
> > +get_pointer_type_max (struct type *type, CORE_ADDR *max)
> > +{
> 
> Let's use the function's return value to return the max:
> 
>   CORE_ADDR
>   get_pointer_type_max (struct type *type)
> 
> I'd suggest changing get_unsigned_type_max the same way, as a separate
> patch.

Ack.

> > @@ -558,6 +557,21 @@ vlscm_convert_typed_number (const char *func_name, int obj_arg_pos, SCM obj,
> >  	  return value_from_longest (type, gdbscm_scm_to_longest (obj));
> >  	}
> >      }
> > +  else if (type->code () == TYPE_CODE_PTR)
> > +    {
> > +      CORE_ADDR max;
> > +
> > +      get_pointer_type_max (type, &max);
> > +      if (!scm_is_unsigned_integer (obj, 0, max))
> > +	{
> > +	  *except_scmp
> > +	    = gdbscm_make_out_of_range_error (func_name,
> > +					      obj_arg_pos, obj,
> > +					_("value out of range for type"));
> 
> The indentaion of that last line is not correct, but I see that you just
> copied the code above.  Can you make a follow-patch to fix them all?  In
> this case, I would typically do:
> 
> 	  *except_scmp
> 	    = gdbscm_make_out_of_range_error
> 		(func_name, obj_arg_pos, obj, _("value out of range for type"));

So leave the indentation in the patch as-is, but have a follow-up patch fixing
both? Alright.

Thanks.