public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Richard Biener <rguenther@suse.de>
To: gcc-patches@gcc.gnu.org
Subject: [PATCH] tree-optimization/104716 - check if we can copy loop in loop distribution
Date: Tue, 1 Mar 2022 09:52:50 +0100 (CET)	[thread overview]
Message-ID: <20220301085250.F301F13B32@imap2.suse-dmz.suse.de> (raw)

The following checks whether we can copy the loop before attempting
to do so in loop distribution.  In the testcase there's a computed
goto and thus abnormal edges which we cannot redirect.

Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.

2022-03-01  Richard Biener  <rguenther@suse.de>

	PR tree-optimization/104716
	* tree-loop-distribution.cc (find_seed_stmts_for_distribution):
	Check if we can copy the loop.

	* gfortran.dg/pr104716.f: New testcase.
---
 gcc/testsuite/gfortran.dg/pr104716.f | 31 ++++++++++++++++++++++++++++
 gcc/tree-loop-distribution.cc        | 14 +++++++++++++
 2 files changed, 45 insertions(+)
 create mode 100644 gcc/testsuite/gfortran.dg/pr104716.f

diff --git a/gcc/testsuite/gfortran.dg/pr104716.f b/gcc/testsuite/gfortran.dg/pr104716.f
new file mode 100644
index 00000000000..97f899a9201
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/pr104716.f
@@ -0,0 +1,31 @@
+! { dg-do compile }
+! { dg-options "-std=legacy -O2 -ftree-loop-distribution -fno-move-loop-stores -fno-tree-dominator-opts" }
+
+      SUBROUTINE FOO()
+      
+      COMMON /WORK/ C2(2, 2)
+      
+      DIMENSION D11(2)
+
+      EQUIVALENCE (D11(1), C2(1, 1))
+
+      DO 40 I = 1, 2
+         DO 30 J = 1, 2
+            ASSIGN 10 TO ILBL
+            IF (C2(J, I) .NE. 0.0) THEN
+               ASSIGN 20 TO ILBL
+            ENDIF
+            GO TO ILBL
+ 10         CONTINUE
+ 20         CONTINUE
+            C2(J, I) = C2(J, I) + 1
+ 30      CONTINUE
+ 40   CONTINUE
+
+      DO 50 I = 1, 2
+         PRINT 90, I
+ 50   CONTINUE
+      
+      RETURN
+ 90   FORMAT(I5)
+      END
diff --git a/gcc/tree-loop-distribution.cc b/gcc/tree-loop-distribution.cc
index 8ee40d88816..c2ca746e237 100644
--- a/gcc/tree-loop-distribution.cc
+++ b/gcc/tree-loop-distribution.cc
@@ -3237,6 +3237,20 @@ find_seed_stmts_for_distribution (class loop *loop, vec<gimple *> *work_list)
 	  work_list->truncate (0);
 	  break;
 	}
+      edge_iterator ei;
+      edge e;
+      /* We cannot redirect abnormal edges so loop copying will fail.  */
+      FOR_EACH_EDGE (e, ei, bbs[i]->succs)
+	if (e->flags & EDGE_ABNORMAL)
+	  {
+	    if (dump_file && (dump_flags & TDF_DETAILS))
+	      fprintf (dump_file, "loop %d is the source of abnormal edges.\n",
+		       loop->num);
+	    work_list->truncate (0);
+	    break;
+	  }
+      if (e)
+	break;
       for (gphi_iterator gsi = gsi_start_phis (bbs[i]);
 	   !gsi_end_p (gsi); gsi_next (&gsi))
 	{
-- 
2.34.1

             reply	other threads:[~2022-03-01  8:52 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-03-01  8:52 Richard Biener [this message]
2022-03-01 14:25 Richard Biener

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=20220301085250.F301F13B32@imap2.suse-dmz.suse.de \
    --to=rguenther@suse.de \
    --cc=gcc-patches@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).