On Mon, Dec 19, 2016 at 12:50:00PM -0700, Jeff Law wrote:
> > Unrelated to where the warning is issued, it might be a good idea to use
> > %K to emit it with inlining stack, otherwise figuring out why it warns
> > will be harder than needed.
> I would think that would apply to any warning triggered once we've started
> optimizing the code. Don't you?
Probably just any warning post-einline, yes.
Note %K takes a tree, which we have in the expansion (the CALL_EXPR), but
for gimple either we need to build some random tree with the location_t
of gimple_location (stmt), or add another modifier that takes a location_t
and otherwise does the similar thing as %K.
Jakub
