From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <rguenth@sourceware.org>
Received: by sourceware.org (Postfix, from userid 1666)
 id CE72A3858401; Mon,  8 Nov 2021 12:35:27 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org CE72A3858401
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset="utf-8"
From: Richard Biener <rguenth@gcc.gnu.org>
To: gcc-cvs@gcc.gnu.org
Subject: [gcc r11-9222] middle-end/101480 - overloaded global new/delete
X-Act-Checkin: gcc
X-Git-Author: Richard Biener <rguenther@suse.de>
X-Git-Refname: refs/heads/releases/gcc-11
X-Git-Oldrev: 580d66f5a0bdcd88dcdab2d4ace3bbbaae121af3
X-Git-Newrev: 19dcea67ac40cfdeb396fa264ebbe04fbe61fdc0
Message-Id: <20211108123527.CE72A3858401@sourceware.org>
Date: Mon,  8 Nov 2021 12:35:27 +0000 (GMT)
X-BeenThere: gcc-cvs@gcc.gnu.org
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Gcc-cvs mailing list <gcc-cvs.gcc.gnu.org>
List-Unsubscribe: <https://gcc.gnu.org/mailman/options/gcc-cvs>,
 <mailto:gcc-cvs-request@gcc.gnu.org?subject=unsubscribe>
List-Archive: <https://gcc.gnu.org/pipermail/gcc-cvs/>
List-Help: <mailto:gcc-cvs-request@gcc.gnu.org?subject=help>
List-Subscribe: <https://gcc.gnu.org/mailman/listinfo/gcc-cvs>,
 <mailto:gcc-cvs-request@gcc.gnu.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Nov 2021 12:35:27 -0000

https://gcc.gnu.org/g:19dcea67ac40cfdeb396fa264ebbe04fbe61fdc0

commit r11-9222-g19dcea67ac40cfdeb396fa264ebbe04fbe61fdc0
Author: Richard Biener <rguenther@suse.de>
Date:   Mon Oct 11 16:06:03 2021 +0200

    middle-end/101480 - overloaded global new/delete
    
    The following fixes the issue of ignoring side-effects on memory
    from overloaded global new/delete operators by not marking them
    as effectively 'const' apart from other explicitely specified
    side-effects.
    
    This will cause
    
    FAIL: g++.dg/warn/Warray-bounds-16.C  -std=gnu++1? (test for excess errors)
    
    because we now no longer statically see the initialization loop
    never executes because the call to operator new can now clobber 'a.m'.
    This seems to be an issue with the warning code and/or ranger so
    I'm leaving this FAIL to be addressed as followup.
    
    2021-10-11  Richard Biener  <rguenther@suse.de>
    
            PR middle-end/101480
            * gimple.c (gimple_call_fnspec): Do not mark operator new/delete
            as const.
    
            * g++.dg/torture/pr10148.C: New testcase.
    
    (cherry picked from commit 09a0affdb0598a54835ac4bb0dd6b54122c12916)

Diff:
---
 gcc/gimple.c                           |  4 +--
 gcc/testsuite/g++.dg/torture/pr10148.C | 52 ++++++++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+), 2 deletions(-)

diff --git a/gcc/gimple.c b/gcc/gimple.c
index c6f7c2d4900..61126cf94b3 100644
--- a/gcc/gimple.c
+++ b/gcc/gimple.c
@@ -1516,12 +1516,12 @@ gimple_call_fnspec (const gcall *stmt)
       && DECL_IS_OPERATOR_DELETE_P (fndecl)
       && DECL_IS_REPLACEABLE_OPERATOR (fndecl)
       && gimple_call_from_new_or_delete (stmt))
-    return ".co ";
+    return ". o ";
   /* Similarly operator new can be treated as malloc.  */
   if (fndecl
       && DECL_IS_REPLACEABLE_OPERATOR_NEW_P (fndecl)
       && gimple_call_from_new_or_delete (stmt))
-    return "mC";
+    return "m ";
   return "";
 }
 
diff --git a/gcc/testsuite/g++.dg/torture/pr10148.C b/gcc/testsuite/g++.dg/torture/pr10148.C
new file mode 100644
index 00000000000..ed278f9f8d8
--- /dev/null
+++ b/gcc/testsuite/g++.dg/torture/pr10148.C
@@ -0,0 +1,52 @@
+/* { dg-do run } */
+
+#include <stdlib.h>
+#include <assert.h>
+
+static bool flag = false;
+
+class C
+{
+  bool prev;
+
+public:
+  C() : prev(flag)
+  {
+    flag = true;
+  }
+
+  ~C() {
+    flag = prev;
+  }
+};
+
+void* operator new(size_t size)
+{
+  assert(flag);
+  return malloc(size);
+}
+
+void operator delete(void *p)
+{
+  free(p);
+}
+
+void g(int* p)
+{
+  delete p;
+}
+
+void f()
+{
+  int* p;
+  {
+    C c;
+    p = new int;
+  }
+  g(p);
+}
+
+int main(int, char**)
+{
+  f();
+}