On Wednesday, July 07, 2010 08:26:10 pm Daniel Jacobowitz wrote:
> On Wed, Jul 07, 2010 at 07:50:14PM +0200, Ulrich Weigand wrote:
> > Ken Werner wrote:
> > > As can be seen the GDB behaves incorrect for vector types. A quick look
> > > to the valops.c:value_assign function shows that
> > > value_coerce_to_target creates a value with lval set to lval_memory
> > > for array types (including vectors). The code was introduced with the
> > > following patch: http://sourceware.org/ml/gdb-
> > > patches/2008-03/msg00079.html. I have to admit that I do not entirely
> > > understand why value_coerce_to_target is called here.
> > 
> > Dan, do you recall why you added a value_coerce_to_target for the
> > *destination* of an assignment?  It seems this shouldn't really be
> > necessary ...
> 
> I'm not sure now.  We could try dropping it.

Ok, the attached patch removes coerce_array call as well. Tested on powerpc64-
*-linux-gnu and i686-*-linux-gnu, no regressions.

> > > @@ -1424,6 +1424,9 @@ value_must_coerce_to_target (struct valu
> > > 
> > >    valtype = check_typedef (value_type (val));
> > > 
> > > +  if (TYPE_VECTOR (valtype))
> > > +    return 0;
> > > +
> > > 
> > >    switch (TYPE_CODE (valtype))
> > >    
> > >      {
> > 
> > >      case TYPE_CODE_ARRAY:
> > Ken, this doesn't look quite right: the TYPE_VECTOR flag is defined only
> > for TYPE_CODE_ARRAY types; you should never look at TYPE_VECTOR for any
> > other type.

Fixed, thanks.

> > Otherwise, this looks reasonable to me ...
> 
> It does seem reasonable that a vector does not have to live in target
> memory.

Regards
-ken