From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugzilla@gcc.gnu.org>
Received: by sourceware.org (Postfix, from userid 48)
 id 64A9E3857831; Thu,  1 Apr 2021 13:23:10 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 64A9E3857831
From: "jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/99863] [10/11 Regression] wrong code with -O
 -fno-tree-forwprop -mno-sse2 since r10-7268-g529ea7d9596b26ba
Date: Thu, 01 Apr 2021 13:23:10 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: rtl-optimization
X-Bugzilla-Version: 11.0
X-Bugzilla-Keywords: wrong-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jakub at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 10.3
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-99863-4-2Ws9ZPqWLT@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-99863-4@http.gcc.gnu.org/bugzilla/>
References: <bug-99863-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-BeenThere: gcc-bugs@gcc.gnu.org
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Gcc-bugs mailing list <gcc-bugs.gcc.gnu.org>
List-Unsubscribe: <https://gcc.gnu.org/mailman/options/gcc-bugs>,
 <mailto:gcc-bugs-request@gcc.gnu.org?subject=unsubscribe>
List-Archive: <https://gcc.gnu.org/pipermail/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-request@gcc.gnu.org?subject=help>
List-Subscribe: <https://gcc.gnu.org/mailman/listinfo/gcc-bugs>,
 <mailto:gcc-bugs-request@gcc.gnu.org?subject=subscribe>
X-List-Received-Date: Thu, 01 Apr 2021 13:23:10 -0000

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D99863

--- Comment #13 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
So, I think we want to do:
--- gcc/dse.c.jj        2021-01-28 09:59:11.973750676 +0100
+++ gcc/dse.c   2021-04-01 15:16:22.003913061 +0200
@@ -1970,8 +1970,7 @@ get_stored_val (store_info *store_info,

 static bool
 replace_read (store_info *store_info, insn_info_t store_insn,
-             read_info_t read_info, insn_info_t read_insn, rtx *loc,
-             bitmap regs_live)
+             read_info_t read_info, insn_info_t read_insn, rtx *loc)
 {
   machine_mode store_mode =3D GET_MODE (store_info->mem);
   machine_mode read_mode =3D GET_MODE (read_info->mem);
@@ -2040,7 +2039,8 @@ replace_read (store_info *store_info, in
          note_stores (this_insn, look_for_hardregs, regs_set);
        }

-      bitmap_and_into (regs_set, regs_live);
+      if (store_insn->fixed_regs_live)
+       bitmap_and_into (regs_set, store_insn->fixed_regs_live);
       if (!bitmap_empty_p (regs_set))
        {
          if (dump_file && (dump_flags & TDF_DETAILS))
@@ -2286,7 +2286,7 @@ check_mem_read_rtx (rtx *loc, bb_info_t
                                                 offset - store_info->offse=
t,
                                                 width)
                      && replace_read (store_info, i_ptr, read_info,
-                                      insn_info, loc, bb_info->regs_live))
+                                      insn_info, loc))
                    return;

                  /* The bases are the same, just see if the offsets
@@ -2352,8 +2352,7 @@ check_mem_read_rtx (rtx *loc, bb_info_t
                                   store_info->width)
              && all_positions_needed_p (store_info,
                                         offset - store_info->offset, width)
-             && replace_read (store_info, i_ptr,  read_info, insn_info, lo=
c,
-                              bb_info->regs_live))
+             && replace_read (store_info, i_ptr,  read_info, insn_info, lo=
c))
            return;

          remove =3D canon_true_dependence (store_info->mem,
to fix this.  Perhaps together with filling fixed_regs_live on more stores,
currently we fill it in only for insn with exactly one store or on memset
calls, perhaps we want to fill it on insns with more than one store?=