from:"t.h.shorrock at gmail dot com"

[Bug c++/48751] New: Incorrect delete[] in c++/4.4/parallel/par_loop.h line 127

2011-04-24 Thread t.h.shorrock at gmail dot com

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48751

   Summary: Incorrect delete[] in c++/4.4/parallel/par_loop.h line
127
   Product: gcc
   Version: 4.4.5
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: c++
AssignedTo: unassig...@gcc.gnu.org
ReportedBy: t.h.shorr...@gmail.com


__gnu_parallel::accumulate gives a segfault when accumulating over 1000
elements (but not 999).

This was originally asked as a question on  StackOverflow -
http://stackoverflow.com/questions/5769908/unexpected-segfault-with-gnu-parallelaccumulate:

Sehe diagnosed the problem as a bug in c++/4.4/parallel/par_loop.h:87

"
#   pragma omp single
  {
num_threads = omp_get_num_threads();
thread_results = static_cast(
::operator new(num_threads * sizeof(Result)));
constructed = new bool[num_threads];
  }
But line 127 deletes it with

delete[] thread_results;

_Apparently, the construction of thread_results was optimized at one stage, but
the deletion statement was never updated to reflect this. The optimization
makes sense over just newing up the array (new Result[num_threads]) because it
avoids constructing the elements._

Fixing that to

delete thread_results;

removes the bug. You will want to report this to the gnu devs.
"

SYSTEM
Ubuntu 10-10 on AMD64 dual core.

TO RECREATE:



#include  
#include  
#include  


class NaturalParameters
{
public:

  NaturalParameters()
: m_data(2) //defualt constructor
  {  }

  std::vector::const_iterator
  begin() const
  {
return m_data.begin();
  }

  std::vector::const_iterator
  end() const
  {
return m_data.begin();
  }

  NaturalParameters& 
  operator+=(const NaturalParameters& other)
  {
//do something - well actually ... nothing for the purposes of this bug
report.
return *this;
  }

private:
  std::vector m_data;
};

inline
NaturalParameters
operator+(const NaturalParameters& a, const NaturalParameters& b)
{
  NaturalParameters tmp = a;
  return tmp+=b;
}  

int
main  (int ac, char **av)
{
  //This one succeeds
  for(size_t i=0;i<1000;++i){
std::vector ChildrenNP(999);
NaturalParameters init;
NaturalParameters NP = __gnu_parallel::accumulate(ChildrenNP.begin(),
ChildrenNP.end(), init ); 
  }
  //This one fails with a segfault
  for(size_t i=0;i<1000;++i){
std::vector ChildrenNP(1000);
NaturalParameters init;
NaturalParameters NP = __gnu_parallel::accumulate(ChildrenNP.begin(),
ChildrenNP.end(), init ); 
  }
}


Preprocessed file attached

COMMAND:

 g++ -v  gnu_parallel_bug.cpp -g -fopenmp -Wall -D_GLIBCXX_PARALLEL -save-temps
Using built-in specs.
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Ubuntu/Linaro
4.4.4-14ubuntu5' --with-bugurl=file:///usr/share/doc/gcc-4.4/README.Bugs
--enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr
--program-suffix=-4.4 --enable-shared --enable-multiarch
--enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib
--without-included-gettext --enable-threads=posix
--with-gxx-include-dir=/usr/include/c++/4.4 --libdir=/usr/lib --enable-nls
--with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug --enable-objc-gc
--disable-werror --with-arch-32=i686 --with-tune=generic
--enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu
--target=x86_64-linux-gnu
Thread model: posix
gcc version 4.4.5 (Ubuntu/Linaro 4.4.4-14ubuntu5) 
COLLECT_GCC_OPTIONS='-v' '-g' '-fopenmp' '-Wall' '-D_GLIBCXX_PARALLEL'
'-save-temps' '-shared-libgcc' '-mtune=generic' '-pthread'
 /usr/lib/gcc/x86_64-linux-gnu/4.4.5/cc1plus -E -quiet -v -D_GNU_SOURCE
-D_REENTRANT -D_GLIBCXX_PARALLEL gnu_parallel_bug.cpp -D_FORTIFY_SOURCE=2
-mtune=generic -Wall -fopenmp -g -fworking-directory -fpch-preprocess
-fstack-protector -o gnu_parallel_bug.ii
ignoring nonexistent directory "/usr/local/include/x86_64-linux-gnu"
ignoring nonexistent directory
"/usr/lib/gcc/x86_64-linux-gnu/4.4.5/../../../../x86_64-linux-gnu/include"
ignoring nonexistent directory "/usr/include/x86_64-linux-gnu"
#include "..." search starts here:
#include <...> search starts here:
 /usr/include/c++/4.4
 /usr/include/c++/4.4/x86_64-linux-gnu
 /usr/include/c++/4.4/backward
 /usr/local/include
 /usr/lib/gcc/x86_64-linux-gnu/4.4.5/include
 /usr/lib/gcc/x86_64-linux-gnu/4.4.5/include-fixed
 /usr/include
End of search list.
COLLECT_GCC_OPTIONS='-v' '-g' '-fopenmp' '-Wall' '-D_GLIBCXX_PARALLEL'
'-save-temps' '-shared-libgcc' '-mtune=generic' '-pthread'
 /usr/lib/gcc/x86_64-linux-gnu/4.4.5/cc1plus -fpreprocessed gnu_parallel_bug.ii
-quiet -dumpbase gnu_parallel_bug.cpp -mtune=generic -auxbase gnu_parallel_bug
-g -Wall -version -fopenmp -fstack-protector -o gnu_parallel_bug.s
GNU C++ (Ubuntu/Linaro 4.4.4-14ubuntu5) version 4.4.5 (x86_64-linux-gnu)
compiled by GNU C version 4.4.5, GMP version 4.3.2, MPFR version 3.0.0-p3.
GGC heuristics: --param ggc-min-expand=

[Bug c++/48751] Incorrect delete[] in c++/4.4/parallel/par_loop.h line 127

2011-04-24 Thread t.h.shorrock at gmail dot com

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48751

--- Comment #1 from Tom  2011-04-24 12:05:10 UTC 
---
Sehe - who found the bug in response to my question submitted the bug report
too, its here: 

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48750

This is a repetition of this report.

[Bug libstdc++/48750] for_each_template_random_access_ed has unbalanced new/delete[]

2011-04-26 Thread t.h.shorrock at gmail dot com

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48750

--- Comment #10 from Tom  2011-04-26 18:09:05 
UTC ---
Hi Paulo,

Yes, this certainly fixes my segfault.

Valgrind complains a little - although I am not sure if this is important (I
am very new to the helgrind tool).  In case it is useful, I attach the
output I get,

So the line numbers are meaningful, I include my modified par_loop.h and my
test file.


Here is my g++ -v

Using built-in specs.
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Ubuntu/Linaro
4.4.4-14ubuntu5' --with-bugurl=file:///usr/share/doc/gcc-4.4/README.Bugs
--enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr
--program-suffix=-4.4 --enable-shared --enable-multiarch
--enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib
--without-included-gettext --enable-threads=posix
--with-gxx-include-dir=/usr/include/c++/4.4 --libdir=/usr/lib --enable-nls
--with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug
--enable-objc-gc --disable-werror --with-arch-32=i686 --with-tune=generic
--enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu
--target=x86_64-linux-gnu
Thread model: posix
gcc version 4.4.5 (Ubuntu/Linaro 4.4.4-14ubuntu5)

running on Ubuntu 10-10


Many thanks for your assistance with this. My gratitude to the FSF grows
daily.

Tom



On 26 April 2011 16:20, paolo.carlini at oracle dot com <
gcc-bugzi...@gcc.gnu.org> wrote:

> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48750
>
> --- Comment #9 from Paolo Carlini 
> 2011-04-26 15:19:12 UTC ---
> Seth, Tom, if you get a chance to test the changes I propose in Comment 7
> (suitably, trivially tweaked for 4.4.x), please let me know as soon as
> possible.
>
> If Johannes cannot provide feedback on this issue over the next few days I
> mean
> to go ahead with my patch, likely for 4.6.x too, and work on a other
> similar
> tweaks in the same spirit elsewhere in the parallel code (luckily we have
> only
> an handful of operator new / operator delete pairs needing attention).
>
> --
> Configure bugmail: http://gcc.gnu.org/bugzilla/userprefs.cgi?tab=email
> --- You are receiving this mail because: ---
> You are on the CC list for the bug.
>

[Bug c++/48751] New: Incorrect delete[] in c++/4.4/parallel/par_loop.h line 127

[Bug c++/48751] Incorrect delete[] in c++/4.4/parallel/par_loop.h line 127

[Bug libstdc++/48750] for_each_template_random_access_ed has unbalanced new/delete[]

3 matches

Site Navigation

Mail list logo

Footer information