This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug c/76783] [7 Regression] wrong code with conditional vector assignment @ -Og
- From: "rguenth at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Mon, 15 Aug 2016 14:00:21 +0000
- Subject: [Bug c/76783] [7 Regression] wrong code with conditional vector assignment @ -Og
- Authentication-results: sourceware.org; auth=none
- Auto-submitted: auto-generated
- References: <bug-76783-4@http.gcc.gnu.org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=76783
--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ok, the BB iteration order of the SSA propagator now visits the merger in a
half-diamond before the then block, breaking the setting of
prop_set_simulate_again in simulate_stmt for two constant uses.
Easy out is to consider ! executable edges as has_simulate_again_uses.
I guess we need that to be conservative anyway.
But then iteration order is not optimal here. Search for a better one than
PRE order (use reverse RPO order?)