Hi, Petr,

I see a raw pointer named pname in ibis::part::init that is allocated
with new [].  Just replace it with std::string.  Please take a look at
SVN 531 to see if it removes the memory leak problem.

Thanks.

John




On 7/30/12 8:33 AM, Petr Kramolis wrote:
> Hey there
> 
> i work with the fastbit library and everything works fine but i can't
> solve one memory leak. I store data in multiple directories (multiple
> tables) and sometimes i call a function to build indexes  (so i don't
> have to build all indexes at query time).
> Its really short function:
> 
> index_table = ibis::table::create(dir.c_str());
> index_table->buildIndexes();
> delete index_table;
> 
> and at this point valgrind reports me memory leak:
> 
> ==16727== 128 bytes in 2 blocks are definitely lost in loss record 7 of 14
> ==16727==    at 0x4C25D99: operator new[](unsigned long) (in
> /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
> ==16727==    by 0x67B62BC: ibis::part::init(char const*) (in
> /usr/lib64/libfastbit.so.0.0.9)
> ==16727==    by 0x67B8510: ibis::part::part(char const*, bool) (in
> /usr/lib64/libfastbit.so.0.0.9)
> ==16727==    by 0x67B8661:
> ibis::util::gatherParts(std::vector<ibis::part*,
> std::allocator<ibis::part*> >&, char const*, bool) (in
> /usr/lib64/libfastbit.so.0.0.9)
> ==16727==    by 0x76B35F1: ibis::mensa::mensa(char const*) (in
> /usr/lib64/libfastbit.so.0.0.9)
> ==16727==    by 0x76B36C8: ibis::table::create(char const*) (in
> /usr/lib64/libfastbit.so.0.0.9)
> ==16727==    by 0x63A012B: storage_close (in
> /usr/lib64/ipfixcol-fastbit-output.so)
> ==16727==    by 0x40395A: data_manager_free (in
> /home/kramolis/git/ipfixcol/base/src/ipfixcol)
> ==16727==    by 0x404A6A: data_manager_close (in
> /home/kramolis/git/ipfixcol/base/src/ipfixcol)
> ==16727==    by 0x4062D6: preprocessor_close (in
> /home/kramolis/git/ipfixcol/base/src/ipfixcol)
> ==16727==    by 0x406014: main (in
> /home/kramolis/git/ipfixcol/base/src/ipfixcol)
> 
> can it be memory leak cased by ibis:table and its destructor? (i tried
> to free created parts manually but it seams that they are freed in
> destructor)
> Thanks for any hits
> 
> Petr K.
> 
> 
> 
> _______________________________________________
> FastBit-users mailing list
> [email protected]
> https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users
> 
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users

Reply via email to