second thoughts on using dl_iterate_phdr() for cache-validation

[Roland McGrath]

To avoid that locking overhead you need to do your short-circuit case
without taking any locks.  I presume you don't care about races producing
false short circuits, since addresses you care about shouldn't be coming or
going while you're unwinding.  The only thing I see is to give you a way to
see _dl_load_adds without calling __dl_iterate_phdr and taking locks, which
is what you asked for in the first place.  We really can't give you an
exported variable symbol, so it would have to be a new function interface.