Re: c++/8047: [i386] Stack corruption with exceptions

Re: c++/8047: [i386] Stack corruption with exceptions

[Danny Smith]

The following reply was made to PR c++/8047; it has been noted by GNATS.

From: Danny Smith <dannysmith@clear.net.nz>
To: gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: c++/8047: [i386] Stack corruption with exceptions
Date: Mon, 18 Nov 2002 23:51:15 +0000

 [The following observations are for mingw32 and cygwin builds with
 Dwarf2 EH enabled.  The patches to enable Dwarf2 are in the
 cygwin-mingw-gcc-3_2-branch.]
 
 I have just tried this with i386-mingw32 (GCC3.3) with same result: Okay
 with
 march=i686, stack exhaustion with  i[345]86.  However, if add
 -maccumulate-ougoing-args, the stack is okay with default  -march=i386
 
 Ditto with cygwin
 
 
 Coincidentally, in mingw and cygwin, functions that call a __stdcall
 function
 (callee pops stack) show stack corruption with Dwarf2 exceptions when
 compiled with
 -fomit-frame-pointer, but are okay with -fomit-frame-pointer
 -maccumulate-outgoing-args.
 
 Also, perhaps related is PR6764, which still fails on mingw
 with --march=i386
 but not with -march=i686 or -maccumulate-outgoing-args
 
 Danny
 
 

Re: c++/8047: [i386] Stack corruption with exceptions

[Colin Burgess]

The following reply was made to PR c++/8047; it has been noted by GNATS.

From: Colin Burgess <cburgess@qnx.com>
To: wagner@rtna.daimlerchrysler.com, king@rtna.daimlerchrysler.com,
        nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org,
        schroedl@rtna.daimlerchrysler.com, gcc-prs@gcc.gnu.org,
        kloeckner@rtna.daimlerchrysler.com
Cc:  
Subject: Re: c++/8047: [i386] Stack corruption with exceptions
Date: Tue, 04 Feb 2003 15:58:45 -0500

 Hey guys, I think I saw this already.  See 
 http://gcc.gnu.org/ml/gcc-bugs/2001-12/msg00522.html
 
 You can check it by seeing if -fno-defer-pop helps the problem.  If so, 
 then here is a patch for 2.95.3
 
 Index: except.c
 ===================================================================
 RCS file: /product/tools/gcc/gcc/cp/except.c,v
 retrieving revision 1.1
 retrieving revision 1.2
 diff -c -r1.1 -r1.2
 *** except.c    28 Feb 2001 18:54:56 -0000      1.1
 --- except.c    12 Dec 2001 18:03:07 -0000      1.2
 ***************
 *** 1135,1140 ****
 --- 1135,1143 ----
         expand_expr (exp, const0_rtx, VOIDmode, EXPAND_NORMAL);
       }
 
 +   /* make sure we have cleaned up the stack */
 +   do_pending_stack_adjust();
 +
     expand_internal_throw ();
   }
 
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=8047 
 
 

Re: c++/8047: [i386] Stack corruption with exceptions

[Wolfgang Bangerth]

The following reply was made to PR c++/8047; it has been noted by GNATS.

From: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
To: gcc-gnats@gcc.gnu.org, <gcc-bugs@gcc.gnu.org>
Cc:  
Subject: Re: c++/8047: [i386] Stack corruption with exceptions
Date: Fri, 15 Nov 2002 18:31:44 -0600 (CST)

 I just looked at this report. A reduced testcase is this:
 -----------------------------------
 struct auto_ptr {
     auto_ptr(int) { }
 };
 
 int foo() { throw 5; }
 
 int main()
 {
   label:
   try {
     auto_ptr x(foo());
   }
   catch (...)
     {
     }
 
   goto label;
 }
 ----------------------------------
 
 On closer examination, I believe this to be actually a stack corruption 
 bug, because if I put a breakpoint into the loop, I can see how the stack 
 pointer is decreased by 8 bytes in each iteration, which eventually leads 
 to the program running out of space. Note that this behavior requires that 
 you use -march=386, using i686 does not trigger the problem.
 
 Regards
  Wolfgang
 
 -------------------------------------------------------------------------
 Wolfgang Bangerth              email:           bangerth@ticam.utexas.edu
                                www: http://www.ticam.utexas.edu/~bangerth