public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "dorit at il dot ibm dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/19951] ICE in tree_split_edge, at tree-cfg.c:3199 with -ftree-vectorize
Date: Tue, 15 Feb 2005 20:56:00 -0000	[thread overview]
Message-ID: <20050215132356.25359.qmail@sourceware.org> (raw)
In-Reply-To: <20050214134328.19951.micis@gmx.de>


------- Additional Comments From dorit at il dot ibm dot com  2005-02-15 13:23 -------
I'm testing this patch (it solves this PR):

Index: tree-vectorizer.c
===============================================================
====
RCS file: /cvs/gcc/gcc/gcc/tree-vectorizer.c,v
retrieving revision 2.73
diff -c -3 -p -r2.73 tree-vectorizer.c
*** tree-vectorizer.c   12 Feb 2005 12:57:01 -0000      2.73
--- tree-vectorizer.c   15 Feb 2005 13:18:34 -0000
*************** vect_analyze_loop_form (struct loop *loo
*** 5581,5597 ****
      }

    /* Make sure we have a preheader basic block.  */
!   if (!loop->pre_header)
      {
!       rescan = true;
!       loop_split_edge_with (loop_preheader_edge (loop), NULL);
      }

    /* Make sure there exists a single-predecessor exit bb:  */
!   if (EDGE_COUNT (loop->exit_edges[0]->dest->preds) != 1)
      {
!       rescan = true;
!       loop_split_edge_with (loop->exit_edges[0], NULL);
      }

    if (rescan)
--- 5581,5617 ----
      }

    /* Make sure we have a preheader basic block.  */
!   if (!loop->pre_header || EDGE_COUNT (loop->pre_header->succs) != 1)
      {
!       edge e = loop_preheader_edge (loop);
!       if (!e->flags & EDGE_ABNORMAL)
!       {
!         loop_split_edge_with (e, NULL);
!         rescan = true;
!       }
!       else
!       {
!         if (vect_print_dump_info (REPORT_BAD_FORM_LOOPS, loop_loc))
!           fprintf (vect_dump, "not vectorized: no preheader bb.");
!         return NULL;
!       }
      }

    /* Make sure there exists a single-predecessor exit bb:  */
!   if (EDGE_COUNT (loop->single_exit->dest->preds) != 1)
      {
!       edge e = loop->single_exit;
!       if (!e->flags & EDGE_ABNORMAL)
!       {
!           loop_split_edge_with (e, NULL);
!           rescan = true;
!       }
!       else
!       {
!         if (vect_print_dump_info (REPORT_BAD_FORM_LOOPS, loop_loc))
!           fprintf (vect_dump, "not vectorized: exit-bb has multiple preds.");
!         return NULL;
!       }
      }

    if (rescan)

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19951


  parent reply	other threads:[~2005-02-15 13:24 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-02-14 18:33 [Bug tree-optimization/19951] New: " micis at gmx dot de
2005-02-14 18:34 ` [Bug tree-optimization/19951] " micis at gmx dot de
2005-02-14 20:05 ` pinskia at gcc dot gnu dot org
2005-02-15 20:56 ` dorit at il dot ibm dot com [this message]
2005-02-17 16:45 ` giovannibajo at libero dot it
2005-02-18 18:17 ` reichelt at gcc dot gnu dot org
2005-02-18 18:18 ` reichelt at gcc dot gnu dot org
2005-02-20 19:23 ` cvs-commit at gcc dot gnu dot org
2005-02-21 10:44 ` cvs-commit at gcc dot gnu dot org
2005-02-21 19:33 ` reichelt at gcc dot gnu dot org

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20050215132356.25359.qmail@sourceware.org \
    --to=gcc-bugzilla@gcc.gnu.org \
    --cc=gcc-bugs@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).