https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92508
--- Comment #14 from Jan Hubicka <hubicka at ucw dot cz> ---
> Did you use bootstrap-lto-lean (and not bootstrap-lto)?
OK, I finally managed to reproduce and debug this.
It is triggered by fact that bootstrap-lean is not training objc and
other frontends which is probably not a good idea (at least from the
objc developer POV).
What happens is that we link trained libbacked with the untrained objc
FE code and ICF then produces alias from objc FE function to trained
funtion and the sanity check mistakely check count of alias instead of
its ultimate target.
I am testing:
Index: ipa-inline-analysis.c
===================================================================
--- ipa-inline-analysis.c (revision 278390)
+++ ipa-inline-analysis.c (working copy)
@@ -211,7 +211,7 @@ do_estimate_edge_time (struct cgraph_edg
nonspec_time = e->entry.nonspec_time;
hints = e->entry.hints;
if (flag_checking
- && !edge->callee->count.ipa_p ())
+ && !callee->count.ipa_p ())
{
sreal chk_time, chk_nonspec_time;
int chk_size, chk_min_size;
