On Tue, Aug 27, 2013 at 5:06 PM, Mike Stump <mikest...@comcast.net> wrote:
> On Aug 27, 2013, at 4:08 AM, Richard Biener <richard.guent...@gmail.com>
> wrote:
>>> and converts:
>>> struct GTY(()) cgraph_node
>>> to:
>>> struct GTY((user)) cgraph_node : public symtab_node_base
>
> GTY didn't like single inheritance for me in in wide-int.h. I extended GTY
> to support it better. See the wide-int branch, if you need more beef here.
> It isn't flawless, but, it did left me to things to make it work nice enough
> for my purposes. In my case, all my data is in the base class, and the base
> class and the derived have the same address (single inheritance, no virtual
> bases), and I have no embedded pointers in my data. I don't use user, seemed
> less ideal to me.
>
> I also had to put:
>
> void gt_ggc_mx(max_wide_int*) { }
> void gt_pch_nx(max_wide_int*,void (*)(void*, void*), void*) { }
> void gt_pch_nx(max_wide_int*) { }
>
> in wide-int.cc and:
>
> extern void gt_ggc_mx(max_wide_int*);
> extern void gt_pch_nx(max_wide_int*,void (*)(void*, void*), void*);
> extern void gt_pch_nx(max_wide_int*);
>
> into wide-int.h to get it to go. These I think are merely because I'm using
> a typedef for a template with template args. In the wide-int branch, it'd be
> nice if a GTY god can figure out how to improves things so I don't need to do
> the above, and to review/approve/fix the GTY code to support single
> inheritance even better.
Huh? Why should wide-int need to be marked GTY at all?!
Richard.
>>> and:
>>> struct GTY(()) varpool_node
>>> to:
>>> class GTY((user)) varpool_node : public symtab_node_base
>>>
>>> dropping the symtab_node_def union.
>>>
>>> Since gengtype is unable to cope with inheritance, we have to mark the
>>> types with GTY((user)), and handcode the gty field-visiting functions.
>>> Given the simple hierarchy, we don't need virtual functions for this.
