RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Kaveh R. Ghazi]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu>
To: David.Billinghurst@riotinto.com, gcc-bugs@gcc.gnu.org,
        gcc-gnats@gcc.gnu.org, kenner@vlsi1.ultra.nyu.edu, rth@gcc.gnu.org
Cc:  
Subject: RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Wed, 17 Apr 2002 20:48:29 -0400 (EDT)

  > From: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
  > 
  > test results with --disable-shared
  > 
  > http://gcc.gnu.org/ml/gcc-testresults/2002-04/msg00634.html
 
 I get similar results, i.e. still over 100 g++ -mabi=64 failures.
 
 --
 Kaveh R. Ghazi			Director of Systems Architecture
 ghazi@caip.rutgers.edu		Qwest Global Services

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[rth]

Synopsis: g++ testsuite EH regressions for irix6 -mabi=64

Responsible-Changed-From-To: kenner->rth
Responsible-Changed-By: rth
Responsible-Changed-When: Mon May  6 22:46:46 2002
Responsible-Changed-Why:
    My patch.
State-Changed-From-To: analyzed->closed
State-Changed-By: rth
State-Changed-When: Mon May  6 22:46:46 2002
State-Changed-Why:
    http://gcc.gnu.org/ml/gcc/2002-05/msg00436.html

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6212

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Richard Henderson]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: Richard Henderson <rth@redhat.com>
To: "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu>
Cc: David.Billinghurst@riotinto.com, gcc-bugs@gcc.gnu.org,
   gcc-gnats@gcc.gnu.org, kenner@vlsi1.ultra.nyu.edu
Subject: Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Tue, 30 Apr 2002 23:17:25 -0700

 On Wed, May 01, 2002 at 12:50:47AM -0400, Kaveh R. Ghazi wrote:
 > Does it make sense to put the testcase you wrote in the URL below into
 > the testsuite?
 
 No, not as-is.  You'd have to search for the unaligned store instructions.
 
 
 r~

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Kaveh R. Ghazi]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu>
To: David.Billinghurst@riotinto.com, rth@redhat.com
Cc: gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, kenner@vlsi1.ultra.nyu.edu
Subject: Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Wed, 1 May 2002 00:50:47 -0400 (EDT)

  > From: Richard Henderson <rth@redhat.com>
  > 
  > On Wed, May 01, 2002 at 01:31:07PM +1000, Billinghurst, David (CRTS) wrote:
  > > The tests now pass on 3.1 and 3.2.  I am unsure if the PR should be closed, 
  > > or if the problem that rth mentions is still present. 
  > 
  > It is.
  > r~
 
 Does it make sense to put the testcase you wrote in the URL below into
 the testsuite?  That way we won't lose the fact that it still needs to
 be fixed.
 
 http://gcc.gnu.org/ml/gcc/2002-04/msg01220.html
 
 		--Kaveh
 --
 Kaveh R. Ghazi			Director of Systems Architecture
 ghazi@caip.rutgers.edu		Qwest Global Services

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Richard Henderson]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: Richard Henderson <rth@redhat.com>
To: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
Cc: gcc-gnats@gcc.gnu.org, ghazi@caip.rutgers.edu, kenner@vlsi1.ultra.nyu.edu,
   gcc-bugs@gcc.gnu.org
Subject: Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Tue, 30 Apr 2002 21:27:00 -0700

 On Wed, May 01, 2002 at 01:31:07PM +1000, Billinghurst, David (CRTS) wrote:
 > The tests now pass on 3.1 and 3.2.  I am unsure if the PR should be closed, 
 > or if the problem that rth mentions is still present. 
 
 It is.
 
 
 r~

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Billinghurst, David (CRTS)]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
To: <gcc-gnats@gcc.gnu.org>,
	<ghazi@caip.rutgers.edu>,
	<kenner@vlsi1.ultra.nyu.edu>,
	<gcc-bugs@gcc.gnu.org>
Cc: <rth@redhat.com>
Subject: Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Wed, 1 May 2002 13:31:07 +1000

 The tests now pass on 3.1 and 3.2.  I am unsure if the PR should be =
 closed,=20
 or if the problem that rth mentions is still present.=20
 
 Running target unix/-mabi=3D64
 
 		=3D=3D=3D g++ Summary for unix/-mabi=3D64 =3D=3D=3D
 
 # of expected passes		7009
 # of expected failures		96
 # of untested testcases		23
 # of unsupported tests		4
 
  It was fixed:
  - after Wed Apr 24 02:19:44 2002 UTC, and
  - before Thu Apr 25 01:00:00 2002 UTC

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[rth]

Synopsis: g++ testsuite EH regressions for irix6 -mabi=64

Responsible-Changed-From-To: rth->kenner
Responsible-Changed-By: rth
Responsible-Changed-When: Wed Apr 24 11:14:13 2002
Responsible-Changed-Why:
    Your regression.
State-Changed-From-To: open->analyzed
State-Changed-By: rth
State-Changed-When: Wed Apr 24 11:14:13 2002
State-Changed-Why:
    http://gcc.gnu.org/ml/gcc/2002-04/msg01220.html

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6212

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[rth]

Synopsis: g++ testsuite EH regressions for irix6 -mabi=64

State-Changed-From-To: feedback->open
State-Changed-By: rth
State-Changed-When: Tue Apr 23 00:38:52 2002
State-Changed-Why:
    Lots o feedback.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6212

RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Billinghurst, David (CRTS)]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
To: "Richard Kenner" <kenner@vlsi1.ultra.nyu.edu>
Cc: <gcc-bug@gcc.gnu.org>,
	<gcc-gnats@gcc.gnu.org>,
	<rth@redhat.com>
Subject: RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Tue, 23 Apr 2002 11:56:16 +1000

 I am not sure I can adequately describe the problem.  Here goes anyway.
 
 Firstly, I have a simpler test case -  g++.old-deja/g++.mike/eh12.C
 
 // Special g++ Options: -fexceptions
 // excess errors test - XFAIL a29k-*-* sparc64-*-elf arm-*-pe
 
 class MyError { };
 
 int main (int argc, char **argv) {
   try {
     throw MyError();
   }
   catch (MyError x) {
   }
 
   return 0;
 }
 
 
 When compiled and executed by the testsuite (below) I see
 Abort (core dumped)
 
 If I compile it with -g -O0 I get the same results.
 
 When I try it under gdb-5.1.1 I see
 
 GNU gdb 5.1.1
 Copyright 2002 Free Software Foundation, Inc.
 GDB is free software, covered by the GNU General Public License, and you =
 are
 welcome to change it and/or distribute copies of it under certain =
 conditions.
 Type "show copying" to see the conditions.
 There is absolutely no warranty for GDB.  Type "show warranty" for =
 details.
 This GDB was configured as "mips-sgi-irix6.5"...Dwarf Error: wrong =
 version in compilation unit header.
 
 (gdb) run
 Starting program: =
 /exd6/billingd/obj/gcc-3.1/gcc/testsuite/./g++-mike-eh12-C.exe=20
 warning: shared library handler failed to enable breakpoint
 
 Program received signal SIGABRT, Aborted.
 warning: Warning: GDB can't find the start of the function at 0xd9e0904.
 
     GDB is unable to find the start of the function at 0xd9e0904
 and thus can't determine the size of that function's stack frame.
 This means that GDB may be unable to access that stack frame, or
 the frames below it.
     This problem is most likely caused by an invalid program counter or
 stack pointer.
     However, if you think GDB should simply search farther back
 from 0xd9e0904 for code which looks like the beginning of a
 function, you can increase the range of the search using the `set
 heuristic-fence-post' command.
 0x0d9e0904 in ?? ()
 (gdb)=20
 
 
 
 The exact compiler flags used are:
 
 /exd6/billingd/obj/gcc-3.1/gcc/testsuite/../g++ =
 -B/exd6/billingd/obj/gcc-3.1/gcc/testsuite/../ =
 /exd2/billingd/src/gcc-3.1/gcc/testsuite/g++.old-deja/g++.mike/eh12.C =
 -nostdinc++  -fmessage-length=3D0 -fexceptions -lstdc++ =
 -L/exd6/billingd/obj/gcc-3.1/mips-sgi-irix6.5/libstdc++-v3/src/.libs =
 -L/exd6/billingd/obj/gcc-3.1/mips-sgi-irix6.5/libiberty -lm -o =
 /exd6/billingd/obj/gcc-3.1/gcc/testsuite/j32.exe -O0 -g

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Richard Kenner]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: kenner@vlsi1.ultra.nyu.edu (Richard Kenner)
To: David.Billinghurst@riotinto.com
Cc: gcc-bug@gcc.gnu.org, gcc-gnats@gcc.gnu.org, rth@redhat.com
Subject: Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Mon, 22 Apr 02 06:40:42 EDT

 And what is being miscompiled here?

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Billinghurst, David (CRTS)]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
To: <rth@gcc.gnu.org>,
	<gcc-gnats@gcc.gnu.org>,
	<ghazi@caip.rutgers.edu>,
	<kenner@vlsi1.ultra.nyu.edu>,
	<gcc-bugs@gcc.gnu.org>
Cc:  
Subject: Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Mon, 22 Apr 2002 14:01:00 +1000

 The smallest test case I have found is g++.old-deja/g++.other/eh3.C.  =
 The=20
 preprocessed file is attached, but the original file doesn't have any =
 preprocessor=20
 directives.
 
 # 2 =
 "/exd6/billingd/src/gcc-j/gcc/testsuite/g++.old-deja/g++.other/eh3.C"
 typedef struct { } e;
 
 char *p;
 
 void _Jv_throw ();
 
 int barf (int len)
 {
   char a[len];
 
   p =3D a;
   _Jv_throw ();
   return 0;
 }
 
 void _Jv_throw ()
 {
   e ex;
   throw ex;
 }
 
 int main ()
 {
   try {
     barf (2);
   }
   catch (...) {
   }
 
   return 0;
 }

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Billinghurst, David (CRTS)]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
To: <rth@gcc.gnu.org>,
	<gcc-gnats@gcc.gnu.org>,
	<ghazi@caip.rutgers.edu>,
	<kenner@vlsi1.ultra.nyu.edu>,
	<gcc-bugs@gcc.gnu.org>
Cc:  
Subject: Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Mon, 22 Apr 2002 13:38:15 +1000

 I have just confirmed the original analysis.
 
 CVS code from 2001-06-12 12:00 UTC has these failures.  With the =
 following change, reverting
 part of the suspect patch, the failures are gone
 
 Index: stor-layout.c
 =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
 =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
 =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
 RCS file: /cvs/gcc/gcc/gcc/stor-layout.c,v
 retrieving revision 1.98
 diff -u -r1.98 stor-layout.c
 --- stor-layout.c       2 Jun 2001 11:14:06 -0000       1.98
 +++ stor-layout.c       22 Apr 2002 03:33:27 -0000
 @@ -381,7 +381,7 @@
        DECL_BIT_FIELD_TYPE (decl) =3D DECL_BIT_FIELD (decl) ? type : 0;
        if (maximum_field_alignment !=3D 0)
         DECL_ALIGN (decl) =3D MIN (DECL_ALIGN (decl), =
 maximum_field_alignment);
 -      else if (DECL_PACKED (decl) && known_align % DECL_ALIGN (decl) =
 !=3D 0)
 +      else if (DECL_PACKED (decl))
         {
           DECL_ALIGN (decl) =3D MIN (DECL_ALIGN (decl), BITS_PER_UNIT);
           DECL_USER_ALIGN (decl) =3D 0;
 
 
 
 Testsuite results for 2001-06-12 12:00 UTC with above change
 
 Native configuration is mips-sgi-irix6.5
 
 		=3D=3D=3D g++ tests =3D=3D=3D
 
 
 Running target unix/-mabi=3D64
 XPASS: g++.dg/vtgc1.C (test for excess errors)
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 Multivv3, 0
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*VTT for =
 Multivv3, 0
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 Multiss2, vtable for Base2
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 Multivs1, vtable for Base2
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*VTT for =
 Multivs1, vtable for Base2
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 Multisv0, vtable for Side0
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*VTT for =
 Multisv0, vtable for Side0
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 Side0, 0
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 VbasedA, 0
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*VTT for =
 VbasedA, 0
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 Base2, vtable for Base1
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 Base1, vtable for Base0
 XPASS: g++.dg/vtgc1.C scan-assembler .vtable_inherit[ 	]*vtable for =
 Base0, 0
 FAIL: g++.ext/instantiate1.C not instantiated (test for errors, line 20)
 FAIL: g++.ext/instantiate1.C not instantiated (test for errors, line 22)
 XPASS: g++.jason/thunk3.C (test for excess errors)
 XPASS: g++.jason/thunk3.C  Execution test
 FAIL: g++.other/crash18.C (test for excess errors)
 XPASS: g++.other/crash27.C (test for excess errors)
 XPASS: g++.robertl/eb17.C  Execution test
 
 		=3D=3D=3D g++ Summary =3D=3D=3D
 
 # of expected passes		6480
 # of unexpected failures	3
 # of unexpected successes	18
 # of expected failures		93
 # of untested testcases		20
 /exd6/billingd/obj/gcc-j/gcc/testsuite/../g++ version 3.1 20010611 =
 (experimental)
 
 

RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Billinghurst, David (CRTS)]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
To: <rth@gcc.gnu.org>,
	<gcc-bugs@gcc.gnu.org>,
	<ghazi@caip.rutgers.edu>,
	<kenner@vlsi1.ultra.nyu.edu>,
	<gcc-gnats@gcc.gnu.org>
Cc:  
Subject: RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Thu, 18 Apr 2002 10:07:20 +1000

 test results with --disable-shared
 
 http://gcc.gnu.org/ml/gcc-testresults/2002-04/msg00634.html
 

RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[Billinghurst, David (CRTS)]

The following reply was made to PR c++/6212; it has been noted by GNATS.

From: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
To: <gcc-bugs@gcc.gnu.org>,
	<gcc-gnats@gcc.gnu.org>
Cc: <rth@gcc.gnu.org>
Subject: RE: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64
Date: Wed, 17 Apr 2002 12:22:29 +1000

 It isn't a harness issue (fixed that).  I see a link warning:
 
 ld64: WARNING 127: Two shared objects with the same soname, =
 /usr/lib64/mips3/libm.so and /usr/lib64/libm.so, have been been linked. =
 This is probably due to a missing -L specification. Ignoring the latter.
 
 I will rebuild with --disable-shared (overnight).
 
 
 -----Original Message-----
 From: rth@gcc.gnu.org [mailto:rth@gcc.gnu.org]
 Sent: Wednesday, 17 April 2002 12:06=20
 To: Billinghurst, David (CRTS); gcc-bugs@gcc.gnu.org;
 gcc-prs@gcc.gnu.org; ghazi@caip.rutgers.edu; kenner@vlsi1.ultra.nyu.edu;
 rth@gcc.gnu.org
 Subject: Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=3D64
 
 
 Synopsis: g++ testsuite EH regressions for irix6 -mabi=3D64
 
 State-Changed-From-To: open->feedback
 State-Changed-By: rth
 State-Changed-When: Tue Apr 16 19:06:28 2002
 State-Changed-Why:
     Are you not seeing a testsuite harness problem?  E.g.
    =20
     884955:./array1.exe: rld: Fatal Error: Cannot Successfully map =
 soname 'libstdc++.so.5' under any of the filenames =
 ./libstdc++.so.5:/holodeck/work/law/rth/build/gcc/libstdc++.so.5:/usr/lib=
 64/libstdc++.so.5:/usr/lib64/internal/libstdc++.so.5:/lib64/libstdc++.so.=
 5:/opt/lib64/libstdc++.so.5:./libstdc++.so.5.5:/holodeck/work/law/rth/bui=
 ld/gcc/libstdc++.so.5.5:/usr/lib64/libstdc++.so.5.5:/usr/lib64/internal/l=
 ibstdc++.so.5.5:/lib64/libstdc++.so.5.5:/opt/lib64/libstdc++.so.5.5:=20
    =20
     Do you see the same failures if you configure with --disable-shared?
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=3Dview%20audit-trail&database=3D=
 gcc&pr=3D6212

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[rth]

Synopsis: g++ testsuite EH regressions for irix6 -mabi=64

State-Changed-From-To: open->feedback
State-Changed-By: rth
State-Changed-When: Tue Apr 16 19:06:28 2002
State-Changed-Why:
    Are you not seeing a testsuite harness problem?  E.g.
    
    884955:./array1.exe: rld: Fatal Error: Cannot Successfully map soname 'libstdc++.so.5' under any of the filenames ./libstdc++.so.5:/holodeck/work/law/rth/build/gcc/libstdc++.so.5:/usr/lib64/libstdc++.so.5:/usr/lib64/internal/libstdc++.so.5:/lib64/libstdc++.so.5:/opt/lib64/libstdc++.so.5:./libstdc++.so.5.5:/holodeck/work/law/rth/build/gcc/libstdc++.so.5.5:/usr/lib64/libstdc++.so.5.5:/usr/lib64/internal/libstdc++.so.5.5:/lib64/libstdc++.so.5.5:/opt/lib64/libstdc++.so.5.5: 
    
    Do you see the same failures if you configure with --disable-shared?

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6212

Re: c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[rth]

Synopsis: g++ testsuite EH regressions for irix6 -mabi=64

Responsible-Changed-From-To: unassigned->rth
Responsible-Changed-By: rth
Responsible-Changed-When: Tue Apr 16 18:07:10 2002
Responsible-Changed-Why:
    .

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6212

c++/6212: g++ testsuite EH regressions for irix6 -mabi=64

[David.Billinghurst]

>Number:         6212
>Category:       c++
>Synopsis:       g++ testsuite EH regressions for irix6 -mabi=64
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Apr 06 22:56:01 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     David Billinghurst
>Release:        3.1 and 3.2
>Organization:
>Environment:
mips-sgi-irix6.5
mips-sgi-irix6.2
>Description:
There are around 100 failures in g++ testsuite with 3.1 on 
irix 6.5.  They are runtime failures consistent with broken 
exception handling.

Looking back through gcc-testresults
 - 20010529 is OK (http://gcc.gnu.org/ml/gcc-testresults/2001-05/msg00714.html)
 - 20010611 fails (http://gcc.gnu.org/ml/gcc-testresults/2001-06/msg00276.html)


There was some discussion at the time:
 - http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00894.html
 - http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg01005.html
 - http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg01489.html

which pointed to patch:

Sat Jun  2 06:53:50 2001  Richard Kenner  <kenner@vlsi1.ultra.nyu.edu>

	* tree.h (struct record_layout_info_s): New field unpadded_align.
	(set_lang_adjust_rli): New declaration.
	* stor-layout.c (layout_decl): If DECL is packed, but at alignment
	it would have if not packed, do not downgrade DECL_ALIGN.
	(lang_adjust_rli, set_lang_adjust_rli): New.
	(start_record_layout): Initialize new field unpadded_align.
	(debug_rli): Display it.
	(place_union_field, place_field): Set it.
	(layout_type, case RECORD_TYPE): Call via lang_adjust_rli if set.

	* print-tree.c (print_node): Don't print "regdecl" when bit doesn't
	mean that; use proper names instead.
	Print DECL_NO_STATIC_CHAIN.

	* except.c (expand_builtin_frob_return_addr): Add missing call
	to convert_memory_address #ifdef POINTERS_EXTEND_UNSIGNED.
	(expand_builtin_eh_return): Likewise.
	(expand_eh_return): Clean up ptr_mode != Pmode cases.

>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: