https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119786
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jakub at gcc dot gnu.org
--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
So, either we just count the make_return_insns into one of the valid passes, so
either
--- gcc/reorg.cc.jj 2025-04-08 14:08:59.000000000 +0200
+++ gcc/reorg.cc 2025-04-14 11:44:12.543606248 +0200
@@ -3697,6 +3697,7 @@ dbr_schedule (rtx_insn *first)
fill_eager_delay_slots ();
relax_delay_slots (first);
}
+ reorg_pass_number = 0;
/* If we made an end of function label, indicate that it is now
safe to delete it by undoing our prior adjustment to LABEL_NUSES.
or the same with = 1; or count them independently by using MAX_REORG_PASSES+1
rather than MAX_REORG_PASSES in those 2 statistics vars and tweaking the
dumping so that it reports even the return stuff.
