Well, yes. And inliner, in fact, turned out to be much less scary than clang codegen. See the patch in llvm-commits instead.
On Tue, Jun 25, 2013 at 2:32 PM, Chandler Carruth <[email protected]> wrote: > > On Tue, Jun 25, 2013 at 1:36 AM, Kostya Serebryany <[email protected]> wrote: >> >> I don't like that one attribute implies another attribute silently, but >> I see no other good way to fix the issue. >> At the very least we need to update the docs in this change. >> Also let's wait for a day if anyone has a better idea. > > > I really share your feelings here. > > Fundamentally, I don't think 'noinline' is what you want here. It has all > kinds of implications you don't really intend. > > For example, if you have one blacklisted function calling another, you can > inline freely. > > How about teaching the inliner specifically to not inline functions with the > attributes you care about into functions which don't have them (or vice > versa)? > > _______________________________________________ > cfe-commits mailing list > [email protected] > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits > _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
