Re: c++/2858: internal compiler error

Re: c++/2858: internal compiler error

[rodrigc]

Synopsis: internal compiler error

State-Changed-From-To: open->feedback
State-Changed-By: rodrigc
State-Changed-When: Sun Aug 12 12:10:58 2001
State-Changed-Why:
    I could not reproduce your problem with the gcc 3.0.1 20010811
    for IA32 Linux.
    
    Can you report this bug to the Turbolinux IA64 mailing list
    at:
    http://www.turbolinux.com/mailman/listinfo/ia64-dist
    
    You may want to try one of the IA64 releases of
    gcc available from Redhat:
    ftp://ftp.ibiblio.org/pub/linux/distributions/redhat/rawhide/SRPMS/SRPMS/gcc-2.96-95.src.rpm
    
    rpm --rebuild gcc-2.96-95.src.rpm
    
    

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=2858&database=gcc

Re: c++/2858: internal compiler error

[wilson]

Synopsis: internal compiler error

Responsible-Changed-From-To: unassigned->wilson
Responsible-Changed-By: wilson
Responsible-Changed-When: Sun Sep 16 19:31:27 2001
Responsible-Changed-Why:
    IA-64 maintainer
State-Changed-From-To: feedback->closed
State-Changed-By: wilson
State-Changed-When: Sun Sep 16 19:31:27 2001
State-Changed-Why:
    I reproduced the bug with the 000717 compiler.  I debugged
    it enough to determine that it was a C++ front end bug.
    Bugs in the 000717 compiler are not interesting at this
    point, as we stopped maintaining it 6 months ago.  It
    was maintained only to support GNU/Linux system development
    as part of the Trillian (ia64-linux) project, and it was
    never intended that this be a fully production quality
    compiler release.
    
    Of more interest is gcc 3.0.x and the gcc mainline sources.
    I tried both.  Neither one has the bug.  Therefore I assume
    that it is already fixed.  GCC 3.x is the preferred compiler
    for IA-64 now, however, it is not ABI compatible with the
    000717 release, so it will not be easily usable by end
    users until the next major round of GNU/linux distros.
    RHL 8.x will use the gcc 3.0.2 release for instance.
    
    I also tried the RHL 7.2 beta (roswell2) compiler.  This
    is the infamous "2.96" compiler.  This compiler also does
    not have the bug.  This compiler is not maintained by the
    FSF, but it is maintained by Red Hat, and the intent is
    that it be a fully production quality compiler.
    
    I don't know if TurboLinux is trying to maintain the
    compiler used by thier OS releases.  You would have to
    ask them.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=2858&database=gcc

Re: c++/2858: internal compiler error

[Gabriel Dos Reis]

The following reply was made to PR c++/2858; it has been noted by GNATS.

From: Gabriel Dos Reis <gdr@codesourcery.com>
To: Craig Rodrigues <rodrigc@mediaone.net>
Cc: Gabriel Dos Reis <gdr@codesourcery.com>,
   Russell Mok <rmok@algorithmics.com>, rodrigc@gcc.gnu.org,
   gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-gnats@gcc.gnu.org,
   nobody@gcc.gnu.org
Subject: Re: c++/2858: internal compiler error
Date: 13 Aug 2001 23:10:38 +0200

 Craig Rodrigues <rodrigc@mediaone.net> writes:
 
 | On Mon, Aug 13, 2001 at 10:42:26PM +0200, Gabriel Dos Reis wrote:
 | > Craig Rodrigues <rodrigc@mediaone.net> writes:
 | > 
 | > | Tell me, who are the IA64 engineers working on gcc?
 | > 
 | >    merlin% grep 64 MAINTAINERS
 | >    ia64 port               Jim Wilson              wilson@redhat.com
 | 
 | OK, so Jim Wilson works for Red Hat (ie. the division of
 | Red Hat formerly known as Cygnus).  So I am not out of line
 | telling Mr. Mok to investigate the fixes in the Red Hat version
 | of IA64 gcc, since that is where a lot of effort to fixing
 | IA64 problems has gone.
 
 I'm not question that.  
 
 You wanted me to tell you "who are the IA64 engineers working on gcc",
 after I said to let the issue open until a IA64 specialist looks at it.
 That is done.  I'm not interested in peripheral details.
 We've got a lot of PRs to examinate, haven't we? :-)
 
 [...]
 
 | It's not clear to me whether these fixes have been incorporated
 | in the FSF version of gcc, but Jim Wilson would probably be
 | the best person to answer these questions.
 
 Certainly.
 
 -- Gaby
 CodeSourcery, LLC                       http://www.codesourcery.com

Re: c++/2858: internal compiler error

[Craig Rodrigues]

The following reply was made to PR c++/2858; it has been noted by GNATS.

From: Craig Rodrigues <rodrigc@mediaone.net>
To: Gabriel Dos Reis <gdr@codesourcery.com>
Cc: Russell Mok <rmok@algorithmics.com>, rodrigc@gcc.gnu.org,
   gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-gnats@gcc.gnu.org,
   nobody@gcc.gnu.org
Subject: Re: c++/2858: internal compiler error
Date: Mon, 13 Aug 2001 16:50:43 -0400

 On Mon, Aug 13, 2001 at 10:42:26PM +0200, Gabriel Dos Reis wrote:
 > Craig Rodrigues <rodrigc@mediaone.net> writes:
 > 
 > | Tell me, who are the IA64 engineers working on gcc?
 > 
 >    merlin% grep 64 MAINTAINERS
 >    ia64 port               Jim Wilson              wilson@redhat.com
 
 OK, so Jim Wilson works for Red Hat (ie. the division of
 Red Hat formerly known as Cygnus).  So I am not out of line
 telling Mr. Mok to investigate the fixes in the Red Hat version
 of IA64 gcc, since that is where a lot of effort to fixing
 IA64 problems has gone.
 
 For example, if I look in the ChangeLog file for Red Hat's
 gcc 2.96-94, I see entries like:
 
 =====================================================================
 * Wed Jun 06 2001 Jakub Jelinek <jakub@redhat.com> 2.96-86
  
 - 13 new patches
   - make jump threading work on ia-32, sparc (#39810)
   - optimize static constructors/destructors function
   - fix ICE on very questionable C++ code from JDK (#39858)
   - fix #ident handling if ident type was seen (#39929)
   - avoid regmove for unchanging pseudo (#40069)
   - fix diagnostic of invalid characters in identifiers (#41554)
   - fix another template mangling bug in the old ABI (#43071)
   - avoid moving CCmode regs around in loop and gcse on IA-64 (#42334)
   - fix ICE with complex float arguments to functions without
     prototypes on IA-64 (#42339)
   - fix Fortran ICHAR() (Toon Moene, #42281)
   - fix IA-64 stop bit handling in presence of asm with multiple
     outputs (Bernd Schmidt, #42560)
   - fix instantiation ICE (Nathan Sidwell, #43150)
   - fix -fsyntax-only (Nix, #43604)
   - fix default argument handling (Jason Merrill, #39466, #42123)
 =====================================================================
 
 It's not clear to me whether these fixes have been incorporated
 in the FSF version of gcc, but Jim Wilson would probably be
 the best person to answer these questions.
 
 -- 
 Craig Rodrigues        
 http://www.gis.net/~craigr    
 rodrigc@mediaone.net          

Re: c++/2858: internal compiler error

[Gabriel Dos Reis]

The following reply was made to PR c++/2858; it has been noted by GNATS.

From: Gabriel Dos Reis <gdr@codesourcery.com>
To: Craig Rodrigues <rodrigc@mediaone.net>
Cc: Gabriel Dos Reis <gdr@codesourcery.com>,
   Russell Mok <rmok@algorithmics.com>, rodrigc@gcc.gnu.org,
   gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-gnats@gcc.gnu.org,
   nobody@gcc.gnu.org
Subject: Re: c++/2858: internal compiler error
Date: 13 Aug 2001 22:42:26 +0200

 Craig Rodrigues <rodrigc@mediaone.net> writes:
 
 | Tell me, who are the IA64 engineers working on gcc?
 
    merlin% grep 64 MAINTAINERS
    ia64 port               Jim Wilson              wilson@redhat.com
 
 -- Gaby
 CodeSourcery, LLC                       http://www.codesourcery.com

Re: c++/2858: internal compiler error

[Gabriel Dos Reis]

The following reply was made to PR c++/2858; it has been noted by GNATS.

From: Gabriel Dos Reis <gdr@codesourcery.com>
To: Craig Rodrigues <rodrigc@mediaone.net>
Cc: Russell Mok <rmok@algorithmics.com>, rodrigc@gcc.gnu.org,
   gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-gnats@gcc.gnu.org,
   nobody@gcc.gnu.org
Subject: Re: c++/2858: internal compiler error
Date: 13 Aug 2001 22:31:23 +0200

 Craig Rodrigues <rodrigc@mediaone.net> writes:
 
 [...]
 
 | IA64 is not one of the primary testing platforms for the
 | FSF release of GCC.
 
 That is not a reason to close bug reported as occuring on those
 non-primary plateforms.  Please, keep this issue open until someone of
 the IA64 engineers looks at it.
 
 Thanks,
 
 -- Gaby

Re: c++/2858: internal compiler error

[Craig Rodrigues]

The following reply was made to PR c++/2858; it has been noted by GNATS.

From: Craig Rodrigues <rodrigc@mediaone.net>
To: Gabriel Dos Reis <gdr@codesourcery.com>
Cc: Russell Mok <rmok@algorithmics.com>, rodrigc@gcc.gnu.org,
   gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-gnats@gcc.gnu.org,
   nobody@gcc.gnu.org
Subject: Re: c++/2858: internal compiler error
Date: Mon, 13 Aug 2001 16:34:50 -0400

 On Mon, Aug 13, 2001 at 10:31:23PM +0200, Gabriel Dos Reis wrote:
 > Craig Rodrigues <rodrigc@mediaone.net> writes:
 > 
 > [...]
 > 
 > | IA64 is not one of the primary testing platforms for the
 > | FSF release of GCC.
 > 
 > That is not a reason to close bug reported as occuring on those
 > non-primary plateforms.  Please, keep this issue open until someone of
 > the IA64 engineers looks at it.
 
 Read my e-mail, I said I wasn't going to close it!
 Tell me, who are the IA64 engineers working on gcc?
 -- 
 Craig Rodrigues        
 http://www.gis.net/~craigr    
 rodrigc@mediaone.net          

Re: c++/2858: internal compiler error

[Craig Rodrigues]

The following reply was made to PR c++/2858; it has been noted by GNATS.

From: Craig Rodrigues <rodrigc@mediaone.net>
To: Russell Mok <rmok@algorithmics.com>
Cc: rodrigc@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
   gcc-gnats@gcc.gnu.org, nobody@gcc.gnu.org
Subject: Re: c++/2858: internal compiler error
Date: Mon, 13 Aug 2001 16:04:40 -0400

 On Mon, Aug 13, 2001 at 03:51:43PM -0400, Russell Mok wrote:
 > I think this is more likely a problem in the compiler instead of the
 > OS, especially when the error disappears by removing a 'const'.  I
 > understand that you couldn't reproduce it using gcc 3.0.1 on IA32
 > linux, but I did file against a different platform (probably with a
 > different code generator and optimizer) and a different release.  I
 > certainly can report this problem to Turbolinux IA64 mailing list, but
 > I am afraid that I will be told this is a compiler problem.  As for
 > the RH alternative, it is beyond what I could try because I cannot
 > arbitrary install software on a machine.  Maybe you can run the test
 > for me if you do have access to a RH IA64 machine.
 > 
 > Please advice on how to keep this issue open if it's been closed.
 
 I am not closing this report.
 
 I do not have access to an IA64 machine so cannot test it.
 The reason why I suggested using the Redhat build of IA64
 is because a lot of IA64 work for gcc has been done
 by Cygnus engineers (which is now part of Redhat).
 
 Redhat is actively updating their compiler and adding IA64
 fixes, so your best bet is to download the Source RPM from
 them and test it on your setup.
 
 Contacting Turbolinux is also a good thing to do, since they
 can tell you how to obtain the latest patched version of their
 compiler (which is derived for the work done at Redhat).
 
 IA64 is not one of the primary testing platforms for the
 FSF release of GCC.
 -- 
 Craig Rodrigues        
 http://www.gis.net/~craigr    
 rodrigc@mediaone.net          

c++/2858: internal compiler error

[rmok]

>Number:         2858
>Category:       c++
>Synopsis:       internal compiler error
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu May 17 13:36:01 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     rmok@algorithmics.com
>Release:        unknown-1.0
>Organization:
>Environment:
Turbolinux for IA64 beta 3 release
gcc version: gcc version 2.96-ia64-000717 snap 001117
>Description:
Error message on screen:

xxx.C: In function `void 
__static_initialization_and_destruction_0 (int, int)':
/home/rmok/xxx.C:14: Unrecognizable insn:
(insn 39 37 40 (set (reg:SI 351)
        (mem/s/u (reg:DI 350) 0)) -1 (nil)
    (expr_list:REG_DEAD (reg:DI 350)
        (nil)))
/home/rmok/xxx.C:14: Internal compiler error in `extract_insn', at 
recog.c:2047
Please submit a full bug report.
See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.

Note: changing the declaration of ATime::Null to a non-const apparently
avoids the error.
>How-To-Repeat:
gcc -c xxx.C
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="xxx.C"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="xxx.C"

Y2xhc3MgQVRpbWUKewpwdWJsaWM6CiAgc3RhdGljIGNvbnN0IEFUaW1lIE51bGw7CiAgaW50IHQ7
Cn07CgpjbGFzcyBTdHJpbmdDb252ZXJ0ZXIKewpwdWJsaWM6CiAgc3RhdGljIEFUaW1lIGlfY3Vy
cmVudFRpbWU7Cn07CgpBVGltZSBTdHJpbmdDb252ZXJ0ZXI6OmlfY3VycmVudFRpbWUgPSBBVGlt
ZTo6TnVsbDsKCg==