From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-prs-return-17273-listarch-gcc-prs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 5116 invoked by alias); 14 Apr 2002 12:16:04 -0000
Mailing-List: contact gcc-prs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc-prs/>
List-Post: <mailto:gcc-prs@gcc.gnu.org>
List-Help: <mailto:gcc-prs-help@gcc.gnu.org>
Sender: gcc-prs-owner@gcc.gnu.org
Received: (qmail 5079 invoked by uid 71); 14 Apr 2002 12:16:02 -0000
Date: Sun, 14 Apr 2002 05:16:00 -0000
Message-ID: <20020414121602.5076.qmail@sources.redhat.com>
To: nobody@gcc.gnu.org
Cc: gcc-prs@gcc.gnu.org,
From: "Billinghurst, David \(CRTS\)" <David.Billinghurst@riotinto.com>
Subject: Re: fortran/5651: Optimization (-funroll-loops) error with LAPACK sstebz.f
Reply-To: "Billinghurst, David \(CRTS\)" <David.Billinghurst@riotinto.com>
X-SW-Source: 2002-04/txt/msg00736.txt.bz2
List-Id: <gcc-prs.sourceware.org>

The following reply was made to PR fortran/5651; it has been noted by GNATS.

From: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
To: "gcc-bugs@gcc.gnu.org" <gcc-gnats@gcc.gnu.org,>,
	<gcc-gnats@gcc.gnu.org>
Cc: <rth@redhat.com>
Subject: Re: fortran/5651: Optimization (-funroll-loops) error with LAPACK sstebz.f
Date: Sun, 14 Apr 2002 22:07:23 +1000

 I have done a search through CVS.  The test case:
  - passes on 2002-01-27 04:30 UTC
  - fails at 05:15 UTC
 
 The only change is:
 
  2002-01-26  Richard Henderson  <rth@redhat.com>
 
         * sched-deps.c (reg_pending_uses_head): New.
         (reg_pending_barrier): Rename from reg_pending_sets_all.
         (find_insn_list): Don't mark inline.
         (find_insn_mem_list): Remove.
         (add_dependence_list, add_dependence_list_and_free): New.
         (flush_pending_lists): Replace only_write param with separate
         for_read and for_write parameters.  Update all callers.  Use
         add_dependence_list_and_free.
         (sched_analyze_1): Do not add reg dependencies here; just set
         the pending bits.  Use add_dependence_list.
         (sched_analyze_2): Likewise.
         (sched_analyze_insn): Replace schedule_barrier_found with
         reg_pending_barrier.  Add all dependencies for pending reg
         uses, sets, and clobbers.
         (sched_analyze): Don't add reg dependencies for calls, just
         set pending bits.  Use regs_invalidated_by_call.  Treat
         sched_before_next_call as a normal list, not a fake insn.
         (init_deps): No funny init for sched_before_next_call.
         (free_deps): Free pending mems lists.  Don't zero reg_last.
         (init_deps_global): Init reg_pending_uses.
         (finish_deps_global): Free it.
         * sched-int.h (deps): Make in_post_call_group_p boolean.  Update =
 docs.
         (find_insn_mem_list): Remove.
         * sched-rgn.c (concat_INSN_LIST, concat_insn_mem_list): New.
         (propagate_deps): Use them.  Zero temp mem lists.