public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Alexander Monakov <amonakov@ispras.ru>
To: gcc-patches@gcc.gnu.org
Cc: Alexander Monakov <amonakov@ispras.ru>
Subject: [committed] tree-ssa-sink: do not touch calls that return twice
Date: Mon,  7 Nov 2022 16:03:02 +0300	[thread overview]
Message-ID: <20221107130302.22073-1-amonakov@ispras.ru> (raw)

Avoid moving pure/const calls that may return twice in tree-ssa-sink:
properly redirecting the associated abnormal edge is non-trivial.

gcc/ChangeLog:

	PR tree-optimization/107505
	* tree-ssa-sink.cc (statement_sink_location): Additionally
	reject ECF_RETURNS_TWICE calls.

gcc/testsuite/ChangeLog:

	PR tree-optimization/107505
	* gcc.dg/pr107505.c: New test.
---

Preapproved by Richi on the Bugzilla, pushed after testing.

 gcc/testsuite/gcc.dg/pr107505.c | 26 ++++++++++++++++++++++++++
 gcc/tree-ssa-sink.cc            |  5 +++--
 2 files changed, 29 insertions(+), 2 deletions(-)
 create mode 100644 gcc/testsuite/gcc.dg/pr107505.c

diff --git a/gcc/testsuite/gcc.dg/pr107505.c b/gcc/testsuite/gcc.dg/pr107505.c
new file mode 100644
index 000000000..01270eac9
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/pr107505.c
@@ -0,0 +1,26 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -fno-guess-branch-probability" } */
+
+int n;
+
+void
+bar (void);
+
+__attribute__ ((noinline, returns_twice)) int
+zero (void)
+{
+  return 0;
+}
+
+void
+foo (void)
+{
+  int a = zero ();
+
+  for (n = 0; n < 2; ++n)
+    {
+    }
+
+  if (a)
+    bar ();
+}
diff --git a/gcc/tree-ssa-sink.cc b/gcc/tree-ssa-sink.cc
index 921305201..eb7c2e681 100644
--- a/gcc/tree-ssa-sink.cc
+++ b/gcc/tree-ssa-sink.cc
@@ -263,12 +263,13 @@ statement_sink_location (gimple *stmt, basic_block frombb,
 
   *zero_uses_p = false;
 
-  /* We only can sink assignments and non-looping const/pure calls.  */
+  /* We only can sink assignments and const/pure calls that are guaranteed
+     to return exactly once.  */
   int cf;
   if (!is_gimple_assign (stmt)
       && (!is_gimple_call (stmt)
 	  || !((cf = gimple_call_flags (stmt)) & (ECF_CONST|ECF_PURE))
-	  || (cf & ECF_LOOPING_CONST_OR_PURE)))
+	  || (cf & (ECF_LOOPING_CONST_OR_PURE|ECF_RETURNS_TWICE))))
     return false;
 
   /* We only can sink stmts with a single definition.  */
-- 
2.37.2


                 reply	other threads:[~2022-11-07 13:03 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

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=20221107130302.22073-1-amonakov@ispras.ru \
    --to=amonakov@ispras.ru \
    --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).