[Bug c++/29529] New: purify with iostream reports 128 uninitialized memory reads

[Bug c++/29529] New: purify with iostream reports 128 uninitialized memory reads

[mstaley at lanl dot gov]

When I compile this code:

   #include <iostream>
   int main(void)
   {
   }

with g++, instrumented with the "purify" utility, purify reports
the following:

   UMR: Uninitialized memory read (128 times)
   This is occurring while in:
         __gconv_get_alias_db [libc.so.6]
         wctob          [libc.so.6]
         std::ctype< wchar_t>::_M_initialize_ctype( void) [libstdc++.so.6]
         std::ctype< wchar_t>::ctype( unsigned) [libstdc++.so.6]
         std::locale::_Impl::_Impl( unsigned) [libstdc++.so.6]
         std::locale::_Impl::_Impl( unsigned) [libstdc++.so.6]
   Reading 4 bytes from 0xbfa77a24 on the stack.
   Address 0xbfa77a24 is       84 bytes below frame pointer in function wctob

Dunno if this is a g++ problem, purify problem, OS problem, system
library problem, not really an error, or whatever, but it would be
nice to make it go away.

System/compiler info:

   uname -a
      Linux banat 2.6.15-26-686 #1 SMP PREEMPT Mon Jul 17 20:14:14 UTC 2006
      i686 GNU/Linux

   g++ --version
      g++ (GCC) 4.0.3 (Ubuntu 4.0.3-1ubuntu5)

Without <iostream>, no problem.


-- 
           Summary: purify with iostream reports 128 uninitialized memory
                    reads
           Product: gcc
           Version: 4.0.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: mstaley at lanl dot gov


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29529

[Bug libstdc++/29529] purify with iostream reports 128 uninitialized memory reads

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2006-10-20 21:34 -------
This is most likely a purify problem.  Have you tried using valgrind instead?


-- 

pinskia at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29529

[Bug libstdc++/29529] purify with iostream reports 128 uninitialized memory reads

[mstaley at lanl dot gov]

------- Comment #2 from mstaley at lanl dot gov  2006-10-25 22:38 -------
(In reply to comment #1)
> This is most likely a purify problem.  Have you tried using valgrind instead?
> 

I just tried valgrind. I'm not really familiar with valgrind, but as far
as I can tell it reports no problems. So, either purify is wrong, valgrind
is wrong, or they're both wrong in different ways.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29529

[Bug libstdc++/29529] purify with iostream reports 128 uninitialized memory reads

[pcarlini at suse dot de]

------- Comment #3 from pcarlini at suse dot de  2006-10-26 04:22 -------
(In reply to comment #2)
> I just tried valgrind. I'm not really familiar with valgrind, but as far
> as I can tell it reports no problems. So, either purify is wrong, valgrind
> is wrong, or they're both wrong in different ways.

Maybe "wrong" is too strong a statement, in this case. In practice, what often
makes a difference is that valgrind knows about many details of the GNU
environment (for instance about the underlying glibc) and is thus able to
automatically filter out many potential false positives. It does that via a
constantly updated so-called suppression file (see the manual, about that).
Anyway, I will double check myself in a few days...


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29529

[Bug libstdc++/29529] purify with iostream reports 128 uninitialized memory reads

[pcarlini at suse dot de]

------- Comment #4 from pcarlini at suse dot de  2006-10-28 22:34 -------
Ok, I have double checked that both on 4.0.3 and the active branches everything
is fine per Valgrind (3.2.1). You may want to pass -v to Valgrind and look at
the "supp:" lines in the output, showing the used suppressions: interestingly,
one pretty commonly used is called "Ubuntu-stripped-ld.so", that is, it has
been added to cope with Ubuntu-specific false positives...


-- 

pcarlini at suse dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29529