Re: Update: status of high-priority GNATS bugs

Re: Update: status of high-priority GNATS bugs

[Reichelt]

Hi,

I'm afraid I've got another bunch of bugs that qualify for "high
priority" :-(. Could somebody please change their priority in GNATS?

PR 7944: ice-on-legal-code, regression from 2.95.x (branch and trunk),
probably target-specific (ix86)

PR 8067: ice-on-legal-code, regression from 3.0.x (branch and trunk),
there were some attempts to tackle the problem, but no patch got applied

PR 5665: ice-on-illegal-code, regression from 3.0.x (trunk only),
the problem already appeared in 3.1-20020204, but was fixed in 3.1.

Greetings,
Volker Reichelt

Re: Update: status of high-priority GNATS bugs

[Paolo Carlini]

Reichelt wrote:

>Hi,
>
>I'm afraid I've got another bunch of bugs that qualify for "high
>priority" :-(. Could somebody please change their priority in GNATS?
>
Ok, thanks,
Paolo.

Re: Update: status of high-priority GNATS bugs

[Paul Jarc]

PR 8146 seems to me to be a regression from 2.95.3.  It might be just
a problem in my (rather unusual) environment, though - has anyone
tried to reproduce it?  (FWIW, I can reproduce it with stock gcc 3.2,
but not with the patched version Red Hat distributes with 8.0.)


paul

Re: Update: status of high-priority GNATS bugs

[Joe Buck]

> PR 8146 seems to me to be a regression from 2.95.3.  It might be just
> a problem in my (rather unusual) environment, though - has anyone
> tried to reproduce it?  (FWIW, I can reproduce it with stock gcc 3.2,
> but not with the patched version Red Hat distributes with 8.0.)

I suspect that almost all Red Hat's patches are in 3.2.1-pre as well.
Could you test the 3.2 CVS branch or a recent 3.2-branch snapshot?

Re: Update: status of high-priority GNATS bugs

[Janis Johnson]

On Wed, Oct 23, 2002 at 02:18:11PM -0700, Joe Buck wrote:
> 
> > PR 8146 seems to me to be a regression from 2.95.3.  It might be just
> > a problem in my (rather unusual) environment, though - has anyone
> > tried to reproduce it?  (FWIW, I can reproduce it with stock gcc 3.2,
> > but not with the patched version Red Hat distributes with 8.0.)
> 
> I suspect that almost all Red Hat's patches are in 3.2.1-pre as well.
> Could you test the 3.2 CVS branch or a recent 3.2-branch snapshot?

I tried this (bootstrapping GCC 2.95.3) with 3.2 that I built from a
release tarball and with the 3.2 branch updated an hour or so ago, and
with both of them I get the same bootstrap comparison failures that
were reported in the PR:

  Bootstrap comparison failure!
  cccp.o differs
  f/data.o differs
  f/target.o differs

This was on i686-pc-linux-gnu, with an old Caldera distribution plus
binutils 2.12.1.

Janis

Re: Update: status of high-priority GNATS bugs

[Paul Jarc]

Joe Buck <jbuck@synopsys.com> wrote:
>> PR 8146 seems to me to be a regression from 2.95.3.  It might be just
>> a problem in my (rather unusual) environment, though - has anyone
>> tried to reproduce it?  (FWIW, I can reproduce it with stock gcc 3.2,
>> but not with the patched version Red Hat distributes with 8.0.)
>
> I suspect that almost all Red Hat's patches are in 3.2.1-pre as well.
> Could you test the 3.2 CVS branch or a recent 3.2-branch snapshot?

I checked out the 3.2 branch yesterday and got the same error.


paul

Re: Update: status of high-priority GNATS bugs

[Jakub Jelinek]

On Thu, Oct 24, 2002 at 11:32:10AM -0400, Paul Jarc wrote:
> Joe Buck <jbuck@synopsys.com> wrote:
> >> PR 8146 seems to me to be a regression from 2.95.3.  It might be just
> >> a problem in my (rather unusual) environment, though - has anyone
> >> tried to reproduce it?  (FWIW, I can reproduce it with stock gcc 3.2,
> >> but not with the patched version Red Hat distributes with 8.0.)
> >
> > I suspect that almost all Red Hat's patches are in 3.2.1-pre as well.
> > Could you test the 3.2 CVS branch or a recent 3.2-branch snapshot?
> 
> I checked out the 3.2 branch yesterday and got the same error.

Can you check gcc-3_2-rhl8-branch too? If you don't get an error there,
the diff is not that big (if omitting TLS and visibility changes).

	Jakub

Re: Update: status of high-priority GNATS bugs

[Paul Jarc]

Jakub Jelinek <jakub@redhat.com> wrote:
> Can you check gcc-3_2-rhl8-branch too?

I can't build that branch; apparently it doesn't like glibc 2.2.5.
./xgcc -B./ -B/package/misc/spf/gcc-3.2-rhl8-2002.10.24/i686-pc-linux-gnu/bin/ -isystem /package/misc/spf/gcc-3.2-rhl8-2002.10.24/i686-pc-linux-gnu/include -isystem /package/misc/spf/gcc-3.2-rhl8-2002.10.24/i686-pc-linux-gnu/sys-include -O2  -DIN_GCC    -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -isystem ./include  -fPIC -g -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED  -I. -I. -I../../gcc-3.2-rhl8-2002.10.24/gcc -I../../gcc-3.2-rhl8-2002.10.24/gcc/. -I../../gcc-3.2-rhl8-2002.10.24/gcc/config -I../../gcc-3.2-rhl8-2002.10.24/gcc/../include  -DL_muldi3 -c ../../gcc-3.2-rhl8-2002.10.24/gcc/libgcc2.c -o libgcc/./_muldi3.o
In file included from /package/misc/spf/gcc-3.2-rhl8-2002.10.24/spf/glibc/include/signal.h:358,
                 from ../../gcc-3.2-rhl8-2002.10.24/gcc/config/i386/linux.h:221,
                 from tconfig.h:21,
                 from ../../gcc-3.2-rhl8-2002.10.24/gcc/libgcc2.c:36:
/package/misc/spf/gcc-3.2-rhl8-2002.10.24/spf/glibc/include/bits/sigthread.h:36: storage class specified for parameter `type name'
make[3]: *** [libgcc/./_muldi3.o] Error 1
make[3]: Leaving directory `/fs/home/mount/home/prj/src/gcc-3_2-rhl8-branch/gcc-3.2-rhl8-2002.10.24-build/gcc'


paul

Re: Update: status of high-priority GNATS bugs

[Janis Johnson]

On Thu, Oct 24, 2002 at 12:17:56PM -0400, Paul Jarc wrote:
> Jakub Jelinek <jakub@redhat.com> wrote:
> > Can you check gcc-3_2-rhl8-branch too?
> 
> I can't build that branch; apparently it doesn't like glibc 2.2.5.

It built for me, with glibc 2.1.2, and it can be used to successfuly
bootstrap GCC 2.95.3.  So, there's a bug in gcc-3_2-branch that does
not exist in gcc-3_2-rhl8-branch.

Janis

Re: Update: status of high-priority GNATS bugs

[Paul Jarc]

Jakub Jelinek <jakub@redhat.com> wrote:
> Can you check gcc-3_2-rhl8-branch too?

Once I got that built, I got the same error again when using it to
build 2.95.3, unlike Janis - maybe my checkout is slightly different
from hers.

Should this PR be made high priority?  It seems to be a regression to
me - 2.95.3 can bootstrap 2.95.3, but 3.* can't.


paul

Re: Update: status of high-priority GNATS bugs

[Janis Johnson]

On Thu, Oct 24, 2002 at 04:04:28PM -0400, Paul Jarc wrote:
> Jakub Jelinek <jakub@redhat.com> wrote:
> > Can you check gcc-3_2-rhl8-branch too?
> 
> Once I got that built, I got the same error again when using it to
> build 2.95.3, unlike Janis - maybe my checkout is slightly different
> from hers.

I went back and tried it again with a gcc built from the
gcc-3_2-rhl8-branch and this time the 2.95.3 bootstrap failed; I
obviously did something wrong last time I tried it, like not really
using the compiler I thought I was using.

Then I tried with some other versions.  GCC 3.1 fails in the same way,
and GCC 3.0.4 succeeds.

Paul, you said that you don't get the failure with the GCC 3.2 that
is distributed with Red Hat 8.0, right?
 
> Should this PR be made high priority?  It seems to be a regression to
> me - 2.95.3 can bootstrap 2.95.3, but 3.* can't.
> paul

It would seem that way to me.

Janis

Re: Update: status of high-priority GNATS bugs

[Paul Jarc]

Janis Johnson <janis187@us.ibm.com> wrote:
> Then I tried with some other versions.  GCC 3.1 fails in the same way,

It's the same for me.

> and GCC 3.0.4 succeeds.

I had thought 3.0 also failed (for just cccp.o), but it's been to long
since I tried it to remember for sure.

> Paul, you said that you don't get the failure with the GCC 3.2 that
> is distributed with Red Hat 8.0, right?

Right.  I think.  I've since replaced that system, so I can't check it
again at the moment.


paul

Update: status of high-priority GNATS bugs

[Joe Buck]

Thanks to Kaveh Ghazi, Matthias Klose, Joseph Myers, David Miller, Janis
Johnson, and Brad Lucier for the feedback on the last version of this
message.  Here's a revision.

6994 and 8017 are now marked "high".

There are now 29 high-priority GNATS bugs; the main reason for reduction
was that the in-feedback-for-ages bugs were closed.

Verified as present in 3.2.1-pre as of 22-oct-02 (on i686-pc-linux-gnu):

11 bugs: 5509, 6579, 6718, 6746, 6981, 6994, 7228, 7266, 7363, 7385, 7679
(Mark says "working on a fix" for 7385)

6579 and 7679 appear to be the same bug.
Possible patch for 6994 from Zack pending?

Reported against 3.2 (or earlier), not tested by me yet:

7227 (sparc/solaris and mips/irix)
7856 (arm)

Can the following two bugs be closed?  The bugs don't show up on the branch,
however I haven't tested the mainline.

6745 (can't duplicate, evidently fixed)
7426 (can't duplicate, evidently fixed)

We need more info on the following:

6545 in feedback state since Aug.
7090 it appears that a patch was applied for this one, but it is still
     in "feedback" state.
7623 (SCO, patch candidate in review as of Aug 17)

Reported against trunk/3.3 branch, confirmed not present in 3.2 branch:

5856, 6558, 6919, 7639, 7794, 7796, 7928, 8080, 8017, 8085, 8186

Re: Update: status of high-priority GNATS bugs

[David Edelsohn]

	Mark just fixed PR c++/6579, but the patch did not fix
PR c++/7679.

David

Re: Update: status of high-priority GNATS bugs

[Joe Buck]

> 	Mark just fixed PR c++/6579, but the patch did not fix
> PR c++/7679.

Sigh, so much for my theory ... in my brief gdb debug session the
infinite loop looked the same.

Re: Update: status of high-priority GNATS bugs

[David Edelsohn]

>>>>> Joe Buck writes:

Joe> Sigh, so much for my theory ... in my brief gdb debug session the
Joe> infinite loop looked the same.

	I think they *are* similar.  I am trying to poke at the problem,
but parsers and G++'s parser are not my expertise.  It may simply be that
Mark was too conservative in his patch for the other PR, like limiting it
to blev==0 (brace level).

	Many of the open problems are C++-specific, but we unfortunately
have a limited group of C++ developers who can efficiently debug and fix
the problems.  G++ is developed and maintained with care and competence,
but C++ is complex and G++ is delicate (especially the current parser
which Mark is rewriting).

	We need to get the bugs fixed, but I also think that we should be
proud of the small number of bugs overall and the small number of bugs in
a C++ implementation which has become extremely more conformant to the
specs.

David

Re: Update: status of high-priority GNATS bugs

[Mark Mitchell]

--On Wednesday, October 23, 2002 01:05:44 PM -0400 David Edelsohn 
<dje@watson.ibm.com> wrote:

>>>>>> Joe Buck writes:
>
> Joe> Sigh, so much for my theory ... in my brief gdb debug session the
> Joe> infinite loop looked the same.
>
> 	I think they *are* similar.  I am trying to poke at the problem,
> but parsers and G++'s parser are not my expertise.  It may simply be that
> Mark was too conservative in his patch for the other PR, like limiting it
> to blev==0 (brace level).

They were related.  My second patch would have caused the first test
case not to loop forever -- but G++ would still have issued a spurious
error.

The first patch made sure that we identified the start and end of
the inline function correctly; the second patch made sure that an
error in an inline function doesn't cause us to loop forever.

As you say, G++ has made great strides towards conformance, and a lot
of the regressions we are fixing now are truly corner-cases, albeit
corner-cases people have found in real code.

The most important thing about the new parser is, in a way, not that
it will be more conformant (it will be), but that it will allow us to
considerably tighten the interfaces throughout the front end.  We have
a ton of code to compensate for what our parser cannot do; once we have
a parser that gets it right we can get rid of all that, and make things
a lot more transparent.

-- 
Mark Mitchell                mark@codesourcery.com
CodeSourcery, LLC            http://www.codesourcery.com

Re: Update: status of high-priority GNATS bugs

[Zack Weinberg]

On Tue, Oct 22, 2002 at 06:52:41PM -0700, Joe Buck wrote:
>
> Possible patch for 6994 from Zack pending?

The first half would be checked in now but there was a power outage on
my block so the bootstrap isn't done yet.

The second half is unfortunately still in development.

zw

Re: Update: status of high-priority GNATS bugs

[Kaveh R. Ghazi]

 > Reported against 3.2 (or earlier), not tested by me yet:
 > 7227 (sparc/solaris and mips/irix)

Sorry, the version number here is unfortunately misleading.

At the time when I filed the report in July, the trunk was called
3.2.x.  But as you are well aware, the trunk was renamed to 3.3.x
later that month when we changed the C++ ABI on the release branch.

So this PR can be moved into the "Reported against trunk" category.
I'll update the PR to reflect the current version number on the trunk.

		--Kaveh
--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu

Re: Update: status of high-priority GNATS bugs

[Robert Lipe]

> 7623 (SCO, patch candidate in review as of Aug 17)

I like this one better.  Bootstrap is still in progress, but it does now
sail past the compilation error.

May I commit this?

RJL

Index: config.gcc
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config.gcc,v
retrieving revision 1.167.2.23.2.1
diff -u -p -r1.167.2.23.2.1 config.gcc
--- config.gcc  6 Aug 2002 15:20:55 -0000       1.167.2.23.2.1
+++ config.gcc  23 Oct 2002 13:38:13 -0000
@@ -1405,7 +1405,7 @@ i[34567]86-*-rtems*|i[34567]86-*-rtemsel
 i[34567]86-*-sco3.2v5*)        # 80386 running SCO Open Server 5
        xm_defines=POSIX
        install_headers_dir=install-headers-cpio
-       tm_file=i386/sco5.h
+       tm_file="${tm_file} i386/att.h i386/sco5.h"
        if test x$gas = xyes
        then
                tm_file="usegas.h ${tm_file}"

Re: Update: status of high-priority GNATS bugs

[Richard Henderson]

On Wed, Oct 23, 2002 at 08:41:33AM -0500, Robert Lipe wrote:
> > 7623 (SCO, patch candidate in review as of Aug 17)
> 
> I like this one better.  Bootstrap is still in progress, but it does now
> sail past the compilation error.
> 
> May I commit this?

Yes, this looks fine.


r~

Re: Update: status of high-priority GNATS bugs

[Robert Lipe]

Richard Henderson wrote:
> On Wed, Oct 23, 2002 at 08:41:33AM -0500, Robert Lipe wrote:
> > > 7623 (SCO, patch candidate in review as of Aug 17)
> > 
> > I like this one better.  Bootstrap is still in progress, but it does now
> > sail past the compilation error.
> > 
> > May I commit this?
> 
> Yes, this looks fine.

Done.  By inspection, this problem doesn't appear on the trunk. 

Can someone please dispose of this PR?

Thanx,
RJL