[Bug c++/25608] New: g++ miscompiles gcjx

[Bug c++/25608] New: g++ miscompiles gcjx

[gdr at gcc dot gnu dot org]

g++ from mainline (and 4.1.x) miscompilers gcjx from gcjx-branch.

How to repeat:

(1) check out gcjx-branch
(2) configure and build gcjx (make all-gcjx, make install-gcjx)
(3) Check out classpath
(4) configure classpath for build with --with-gcjx pointing to gcjx (2)
(5) make

The classpath build stalls at

 /home/gdr/bin/gcjx -bootclasspath '' -sourcepath '' -classpath ../lib:. -d
classes
/home/gdr/redhat/classpath/classpath/examples/gnu/classpath/examples/*/*.java
/home/gdr/redhat/classpath/classpath/examples/gnu/classpath/examples/*/*/*.java
/home/gdr/redhat/classpath/classpath/examples/gnu/classpath/examples/*/*/*/*.java 


Notice that using g++ from GCC-3.3.x and GCC-3.4.x just works fine.
GCC-4.0.x from FSF source has similar problem.  Tom reports that GCC-4.0.x
from Fedora Core works fine.

-- Gaby


-- 
           Summary: g++ miscompiles gcjx
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: gdr at gcc dot gnu dot org
 GCC build triplet: native
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: native


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug c++/25608] g++ miscompiles gcjx

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2005-12-30 22:18 -------
This needs more information than compiling and runing some piece of code.
I almost want to say gcjx uses some undefined code in it.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug c++/25608] g++ miscompiles gcjx

[gdr at integrable-solutions dot net]

------- Comment #2 from gdr at integrable-solutions dot net  2005-12-31 00:12 -------
Subject: Re:  g++ miscompiles gcjx

"pinskia at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| This needs more information than compiling and runing some piece of code.

The PR provides way to reproduce the bug.  And it is not random piece
of code.  You can help reduce the testcase or find better way to be
useful, at least be less helpful.

| I almost want to say gcjx uses some undefined code in it.

Maybe.  I looked at the code, could not find any.  I highky suspect
miscompilation, given recent breakage.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug c++/25608] g++ miscompiles gcjx

[pinskia at gcc dot gnu dot org]

------- Comment #3 from pinskia at gcc dot gnu dot org  2005-12-31 00:14 -------
Does anyone read:
http://gcc.gnu.org/bugs.html

What we do not want

A source file that #includes header files that are left out of the bug report
(see above)
That source file and a collection of header files.
An attached archive (tar, zip, shar, whatever) containing all (or some :-) of
the above.
A code snippet that won't cause the compiler to produce the exact output
mentioned in the bug report (e.g., a snippet with just a few lines around the
one that apparently triggers the bug, with some pieces replaced with ellipses
or comments for extra obfuscation :-)
The location (URL) of the package that failed to build (we won't download it,
anyway, since you've already given us what we need to duplicate the bug,
haven't you? :-)
An error that occurs only some of the times a certain file is compiled, such
that retrying a sufficient number of times results in a successful compilation;
this is a symptom of a hardware problem, not of a compiler bug (sorry)
Assembly files (*.s) produced by the compiler, or any binary files, such as
object files, executables, core files, or precompiled header files
Duplicate bug reports, or reports of bugs already fixed in the development
tree, especially those that have already been reported as fixed last week :-)
Bugs in the assembler, the linker or the C library. These are separate
projects, with separate mailing lists and different bug reporting procedures
Bugs in releases or snapshots of GCC not issued by the GNU Project. Report them
to whoever provided you with the release
Questions about the correctness or the expected behavior of certain constructs
that are not GCC extensions. Ask them in forums dedicated to the discussion of
the programming language


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug c++/25608] g++ miscompiles gcjx

[gdr at gcc dot gnu dot org]

------- Comment #4 from gdr at gcc dot gnu dot org  2005-12-31 01:13 -------
(In reply to comment #0)
> g++ from mainline (and 4.1.x) miscompilers gcjx from gcjx-branch.
> 
> How to repeat:
> 
> (1) check out gcjx-branch
> (2) configure and build gcjx (make all-gcjx, make install-gcjx)
> (3) Check out classpath
> (4) configure classpath for build with --with-gcjx pointing to gcjx (2)
> (5) make
> 
> The classpath build stalls at
> 
>  /home/gdr/bin/gcjx -bootclasspath '' -sourcepath '' -classpath ../lib:. -d
> classes
> /home/gdr/redhat/classpath/classpath/examples/gnu/classpath/examples/*/*.java
> /home/gdr/redhat/classpath/classpath/examples/gnu/classpath/examples/*/*/*.java
> /home/gdr/redhat/classpath/classpath/examples/gnu/classpath/examples/*/*/*/*.java 
> 

GDB seession seems to indicate a deadlock:
[Thread debugging using libthread_db enabled]
[New Thread 1078843072 (LWP 12228)]

Program received signal SIGINT, Interrupt.
[Switching to Thread 1078843072 (LWP 12228)]
0xffffe410 in ?? ()
(gdb) bt
#0  0xffffe410 in ?? ()
#1  0xbfffd2b8 in ?? ()
#2  0x00000002 in ?? ()
#3  0x00000000 in ?? ()
#4  0x404c305e in __lll_mutex_lock_wait () from /lib/tls/libpthread.so.0
#5  0x404bfe2d in _L_mutex_lock_33 () from /lib/tls/libpthread.so.0
#6  0x00000000 in ?? ()
#7  0x00000000 in ?? ()
#8  0x404b4ff4 in ?? () from /lib/tls/libc.so.6
#9  0xbfffd428 in ?? ()
#10 0xbfffd318 in ?? ()
#11 0xbfffd2c8 in ?? ()
#12 0x404656ba in pthread_mutex_lock () from /lib/tls/libc.so.6
#13 0x404656ba in pthread_mutex_lock () from /lib/tls/libc.so.6
#14 0x402bd3de in locale (this=0x2fc4) at gthr-default.h:521
#15 0x402ba010 in std::ios_base::_M_init (this=0xbfffd3bc)
    at /home/gdr/redhat/trunk.gcc/libstdc++-v3/src/ios_locale.cc:48
#16 0x402ccd58 in std::basic_ios<char, std::char_traits<char> >::init (
    this=0xbfffd3bc, __sb=0xbfffd394) at basic_ios.tcc:143
#17 0x401aa47d in format_repr::get_message (this=0x8513400) at sstream:441
#18 0x401aacdf in format_repr::dump (this=0x8513400, os=@0x8054308)
    at /home/gdr/redhat/gcjx.gcc/gcjx/format/format.cc:69
#19 0x401aad9b in operator<< (out=@0x8054308, objw=@0x85115a4)
    at /home/gdr/redhat/gcjx.gcc/gcjx/format/format.cc:59
#20 0x400afc18 in compiler::do_analyze_unit (this=0x8055248, unit=0x8089dc0)
    at exception.hh:63
#21 0x400afebf in compiler::semantic_analysis (this=0x8055248)
    at /home/gdr/redhat/gcjx.gcc/gcjx/compiler.cc:307
#22 0x0804a4d5 in gcjx_main (argv=0xbfffd7b4)
    at /home/gdr/redhat/gcjx.gcc/gcjx/main.cc:644
#23 0x0804abb6 in main (argc=Cannot access memory at address 0x0
) at /home/gdr/redhat/gcjx.gcc/gcjx/main.cc:661


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug c++/25608] g++ miscompiles gcjx

[gdr at gcc dot gnu dot org]

------- Comment #5 from gdr at gcc dot gnu dot org  2005-12-31 04:12 -------
Created an attachment (id=10572)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10572&action=view)
translation unit of format.cc

GDB session indicate the stall happens in the call to
format_repr::get_message().
More precisely, the use of ostringstream seems to led a deadlock.  This
works fine with GCC-3.4.x.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[pinskia at gcc dot gnu dot org]

------- Comment #6 from pinskia at gcc dot gnu dot org  2005-12-31 04:17 -------
This sounds more like a libstdc++ issue rather than a front-end or optimizer
issue.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c++                         |libstdc++


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug c++/25608] g++ miscompiles gcjx

[gdr at gcc dot gnu dot org]

------- Comment #7 from gdr at gcc dot gnu dot org  2005-12-31 04:20 -------
(In reply to comment #6)
> This sounds more like a libstdc++ issue rather than a front-end or optimizer
> issue.

Before reclassifying, it would useful to provide data.  If you cannot,
please move on something else.  You're not being helpful here.


-- 

gdr at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
          Component|libstdc++                   |c++
     Ever Confirmed|0                           |1
   Last reconfirmed|2005-12-31 00:13:09         |2005-12-31 04:20:59
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[pinskia at gcc dot gnu dot org]

------- Comment #8 from pinskia at gcc dot gnu dot org  2005-12-31 04:33 -------
(In reply to comment #7)
> Before reclassifying, it would useful to provide data.  If you cannot,
> please move on something else.  You're not being helpful here.

Since this is reproducible from 4.0 upwards and it is way down in the libstdc++
and it is a dead lock,
it is more likely to be a bug in libstdc++ than anywhere else.


Are you sure that you cannot reproduce this with 3.4.x where x>=2?


This seems very related to PR 12658.

It seems like this could be a glibc bug reading that other bug, specifically
"testsuite/22_locale/locale/cons/12658_thread.cc" test.

What glibc version do you have?


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c++                         |libstdc++


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[gdr at integrable-solutions dot net]

------- Comment #9 from gdr at integrable-solutions dot net  2005-12-31 04:44 -------
Subject: Re:  g++ miscompiles gcjx

"pinskia at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| (In reply to comment #7)
| > Before reclassifying, it would useful to provide data.  If you cannot,
| > please move on something else.  You're not being helpful here.
| 
| Since this is reproducible from 4.0 upwards and it is way down in the
libstdc++
| and it is a dead lock,
| it is more likely to be a bug in libstdc++ than anywhere else.

OK, let see.

| Are you sure that you cannot reproduce this with 3.4.x where x>=2?

I used the latest GCC-3.4.5 to compile gcjx, classpath, jacks and
produce some profile data.  I did not run into the dead lock.  Several
time.  However, with mainline, I get a stall exactly at the same place.

| This seems very related to PR 12658.
| 
| It seems like this could be a glibc bug reading that other bug, specifically
| "testsuite/22_locale/locale/cons/12658_thread.cc" test.
| 
| What glibc version do you have?

glibc-2.3.4-23.4 (SuSE 9.3)

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[gdr at integrable-solutions dot net]

------- Comment #10 from gdr at integrable-solutions dot net  2005-12-31 05:07 -------
Subject: Re:  g++ miscompiles gcjx

"pinskia at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| This seems very related to PR 12658.

it was fixed in all branches.


-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[pinskia at gcc dot gnu dot org]

------- Comment #11 from pinskia at gcc dot gnu dot org  2005-12-31 05:46 -------
Hmm, can you try the following C++ program and see if it works for you:
#include <sstream>

int main(void)
{
  for(int j=0;j<1000;j++){
  std::ostringstream a;
  char *a1 = "aaaaaaaaaaaaaaaaaaaaaaaaaa";
  for(int i =0;i<strlen(a1);i++)
  a << *a1;
  a.str();
}
}

This is emulating what format does in gcjx except multiple times over.  I tried
it on my machine which has a pre SVN 4.1 and it worked there.

The only really big thing which I keep on reading in the libstdc++ source which
is a performance issue rather than anything else is that the locale on the
ios_base can be set a couple of times.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[gdr at integrable-solutions dot net]

------- Comment #12 from gdr at integrable-solutions dot net  2005-12-31 05:56 -------
Subject: Re:  g++ miscompiles gcjx

"pinskia at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| Hmm, can you try the following C++ program and see if it works for you:

It works for me at -O0, -O1, -O2, -O3 and combinations with -g.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[tromey at gcc dot gnu dot org]

------- Comment #13 from tromey at gcc dot gnu dot org  2006-01-19 18:23 -------
FWIW I suspect there is undefined code in gcjx.
For instance I think the constant evaluation code assumes
-fwrapv behavior.  There could well be other undefined code,
but I don't know of any.

That said, it seems unlikely that this particular problem
would cause this bug.  I suppose compiling with -fwrapv
might shed some light on this.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[bkoz at gcc dot gnu dot org]

------- Comment #14 from bkoz at gcc dot gnu dot org  2006-01-26 16:32 -------

Gaby, here's the one thing that I can think of that changed between 3.4.x and
mainline/4.1 in this code path:

2005-10-10  Ian Lance Taylor  <ian@airs.com>

        PR libstdc++/13583
        * include/bits/locale_classes.h (locale::_Impl::_M_install_cache):
        Move out of line.
        * src/locale.cc: Define here, add mutex.

You might want to try reverting this and seeing if it has any impact.

I kind of doubt it, but it's the only thing that really changed WRT ios/locale
init.

I'm surprised to see this issue. I would imagine that if this is a libstdc++
issue everybody and their dog would have found it by now.

best,
benjamin


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[bkoz at gcc dot gnu dot org]

------- Comment #15 from bkoz at gcc dot gnu dot org  2006-01-26 16:35 -------
ps tom this kind of organization for compiling looks dreamy:

#20 0x400afc18 in compiler::do_analyze_unit (this=0x8055248, unit=0x8089dc0)
    at exception.hh:63
#21 0x400afebf in compiler::semantic_analysis (this=0x8055248)


NICE!!!!


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[gdr at cs dot tamu dot edu]

------- Comment #16 from gdr at cs dot tamu dot edu  2006-01-26 16:44 -------
Subject: Re:  g++ miscompiles gcjx

"bkoz at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| Gaby, here's the one thing that I can think of that changed between 3.4.x and
| mainline/4.1 in this code path:

Thank you very much for looking into this.  
I'll do that, once I get to office.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[gdr at cs dot tamu dot edu]

------- Comment #17 from gdr at cs dot tamu dot edu  2006-01-26 16:45 -------
Subject: Re:  g++ miscompiles gcjx

"bkoz at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| ps tom this kind of organization for compiling looks dreamy:

Yup!  I dream  of when other GCC parts would be similarly organized.
(the source code is well structured, with separate concepts developed
separately.)

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[bkoz at gcc dot gnu dot org]

------- Comment #18 from bkoz at gcc dot gnu dot org  2006-03-17 18:33 -------

is this still an active issue, or was it indicative of some temporary or
transient thing in the gcc sources?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[gdr at integrable-solutions dot net]

------- Comment #19 from gdr at integrable-solutions dot net  2006-03-17 23:37 -------
Subject: Re:  g++ miscompiles gcjx

"bkoz at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| is this still an active issue, or was it indicative of some temporary or
| transient thing in the gcc sources?

Now, I get a "segmentation fault" -- instead of dead lock.
gcjx is compiled with mainline GCC/g++.
I was trying to compile classptah.

Adding generated files in builddir '..'.
/home/gdr/bin/gcjx -encoding UTF-8 -classpath .: -d . @classes
make[1]: *** [compile-classes] Segmentation fault
make[1]: Leaving directory `/home/gdr/build/classpath/lib'


-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

Re: [Bug libstdc++/25608] g++ miscompiles gcjx

[Gabriel Dos Reis]

"bkoz at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| is this still an active issue, or was it indicative of some temporary or
| transient thing in the gcc sources?

Now, I get a "segmentation fault" -- instead of dead lock.
gcjx is compiled with mainline GCC/g++.
I was trying to compile classptah.

Adding generated files in builddir '..'.
/home/gdr/bin/gcjx -encoding UTF-8 -classpath .: -d . @classes
make[1]: *** [compile-classes] Segmentation fault
make[1]: Leaving directory `/home/gdr/build/classpath/lib'


-- Gaby

[Bug libstdc++/25608] g++ miscompiles gcjx

[pcarlini at suse dot de]

------- Comment #20 from pcarlini at suse dot de  2006-10-17 09:28 -------
Gaby, any news about this issue?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[gdr at integrable-solutions dot net]

------- Comment #21 from gdr at integrable-solutions dot net  2006-10-24 06:18 -------
Subject: Re:  g++ miscompiles gcjx

"pcarlini at suse dot de" <gcc-bugzilla@gcc.gnu.org> writes:

| Gaby, any news about this issue?

gcjx has been declared dead.  I never got to have the issue fixed.
So, it is up to you.  It was a blocking issue for work on gcjx, and
since the latter no longer exists it is less of an issue.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608

[Bug libstdc++/25608] g++ miscompiles gcjx

[bkoz at gcc dot gnu dot org]

------- Comment #22 from bkoz at gcc dot gnu dot org  2006-12-12 11:37 -------

This is closed as gcjx is dead.


-- 

bkoz at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25608