[Bug bootstrap/69513] LTO bootstrap fails with bootstrap-profiled during linking gnat1 in stagefeedback

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69513

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ah, I guess a simpler patch could be to do

  while ((node = limbo_die_list))
   {
     limbo_die_list = node->next;

... old code ...
   }

I'd even approve that based on the fact we use get_context_die here.  We could
also avoid generating new DIEs here by falling back to comp_unit_die () if
the context DIE doesn't exist yet (which it should with early debug, not with
LTO in some circumstances of course as we see here).