This is the mail archive of the
mailing list for the GCC project.
Re: [C++/cgraph] Middle-end thunk support
> Jan Hubicka wrote:
> > I will update my local version with the static variable and re-test then.
> Can you confirm that we will still follow the invariant that if we put
> out the main function we will always emit all thunks, even if
> unreferenced locally, now that we're depending more heavily on cgraph?
Yes, the code works by adding thunks into list of "aliases" of the
function. This list hangs on cgraph nodes and is removed only when the
function node is removed (or at conflict in lto-symtab symbol table when
i.e. two units define thunk of same name).
> If so, the C++ part of the patch is OK.
Thanks! I am leaving for 2 days off mail, so I will proceed with the
patch after returning.
> Mark Mitchell
> (650) 331-3385 x713