Re: [GOLD] Multiple stub sections and long branch trampolines for powerpc.

[Alan Modra <amodra at gmail dot com>]

On Sun, Dec 02, 2012 at 08:06:59PM -0800, Ian Lance Taylor wrote:
> On Fri, Nov 30, 2012 at 8:09 PM, Alan Modra <amodra@gmail.com> wrote:
> >
> > @@ -1560,7 +1562,8 @@ class Output_data_reloc<elfcpp::SHT_REL,
> >
> >    void
> >    add_global(Symbol* gsym, unsigned int type, Output_data* od, Address address)
> > -  { this->add(od, Output_reloc_type(gsym, type, od, address, false, false, false)); }
> > +  { this->add(od, Output_reloc_type(gsym, type, od, address,
> > +                                   false, false, false)); }
> 
> I don't see an actual change here.

No, this was just wrapping an overly long line.

> If this were to change, a formatting nit: I only put the braces on the
> same line as the code when the inline function fits on a single line.
> otherwise the braces should be on lines by themselves.

I'll leave these formatting changes to a followup patch.  There
are quite a number of lines similar to my change above in output.h.

> 
> > -  { this->section_offsets_need_adjustment_ = true; }
> > +  set_section_offsets_need_adjustment(bool val = true)
> > +  { this->section_offsets_need_adjustment_ = val; }
> 
> Pure C++ style issue: I don't like this kind of default argument.
> This should either be two functions
> (set_section_offsets_need_adjustment and
> clear_section_offsets_need_adjustment) or it should take a parameter
> without all default and the two existing calls should be changed to
> pass true.
> 
> This is OK with those changes.

OK, I went for clear_section_offsets_need_adjustment.

Applied, thanks for the review!

-- 
Alan Modra
Australia Development Lab, IBM