This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH]: Interprocedural detection of readonly andnon-addressable static variables
On Wed, 2005-06-08 at 20:48 +0200, Jan Hubicka wrote:
> > > I made this decision before danny started breaking the decls into
> > > separate structures. The right thing to do most likely to wait until
> > > danny gets the function decl broken out and shove it in there.
> > >
> >
> > Maybe.
> >
> > But anything that goes only on global vars definitely belongs in the
> > cgraph stuff.
>
> Yeah. We need to avoid allocating datastructures we won't use. If it
> is in the decl nodes, we get it for all vars and all functions. We
> probably need it only for available functions analyzed by IPA and this
> is what global substructure if cgraph node is meant for. We will need
> to re-thing this scheme anyway. As discussed while back with Richard,
> currently we have struct function and cgraph node with similar meaning
> and both not fitting very well (lifetime wise) for most of datas we use
> to store in them and both quite a memory hogs....
>
And just a heads up:
FUNCTION_DECL will be the largest DECL, and probably pretty close to the
current size (108 bytes).
I'm guessing (92+ bytes) at this point.
Most of the extra cost is flags, arguments tree, and result tree.
--Dan