|
»
»
»
Possible memory allocation bug
|
View:
New views
5 Messages
—
Rating Filter:
Alert me
|
 |
Possible memory allocation bug
by Jose Luis Blanco
::
Rate this Message:
Hello,
I've not been able to get a consistent test code, but I'm having program crashes (in Linux) and asserts (in Windows) when calling "cvGoodFeaturesToTrack". Switching from the current SVN code to the old 1.0.0 solves the issue. Under Windows, the assert I got is from cvMorph.cpp@260, so it seems it has to do with misaligned memory. In further debugging, I got no more clues... In Linux, I've tried to debug my program with Valgrind to detect memory corruption from either opencv or my code but it detects no errors and the program just dies with the message "Killed"... However, it can still be an issue of memory corruption within valid memory addresses so Valgrind doesn't comply. Perhaps it's a problem of my code, but since switching to opencv 1.0.0 solves it I wanted to let you know. Best, JL -- ___________________________________________________________ Jose-Luis Blanco-Claraco Phone: +34 952 132848 Dpto. Ingenieria de Sistemas y Automatica E.T.S.I. Telecomunicacion Fax: +34 952 133361 Universidad de Malaga Campus Universitario de Teatinos 29071 Malaga, Spain http://www.isa.uma.es/jlblanco ___________________________________________________________ ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Opencvlibrary-devel mailing list Opencvlibrary-devel@... https://lists.sourceforge.net/lists/listinfo/opencvlibrary-devel |
|
|
Re: Possible memory allocation bug
by Vadim Pisarevsky
::
Rate this Message:
Hello Jose,
I can not reproduce it with the current SVN on Mac. At least, lkdemo runs fine. There was such problem with misaligned buffers a couple of weeks ago, but it is solved in the trunk version now.
Regards, Vadim On Thu, Sep 3, 2009 at 2:10 PM, Jose Luis Blanco <joseluisblancoc@...> wrote: Hello, ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Opencvlibrary-devel mailing list Opencvlibrary-devel@... https://lists.sourceforge.net/lists/listinfo/opencvlibrary-devel |
|
|
Re: Possible memory allocation bug
by Guy K. Kloss-2
::
Rate this Message:
On Thu, 03 Sep 2009 22:10:11 Jose Luis Blanco wrote:
> I've not been able to get a consistent test code, but I'm having > program crashes (in Linux) and asserts (in Windows) when calling > "cvGoodFeaturesToTrack". It might help to to run the code under questioning through valgrind or compile it with libefence to see whether there's some boundary violation. Guy -- Guy K. Kloss Institute of Information and Mathematical Sciences Te Kura Pūtaiao o Mōhiohio me Pāngarau Massey University, Albany (North Shore City, Auckland) 473 State Highway 17, Gate 1, Mailroom, Quad B Building voice: +64 9 414-0800 ext. 9585 fax: +64 9 441-8181 G.Kloss@... http://www.massey.ac.nz/~gkloss ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Opencvlibrary-devel mailing list Opencvlibrary-devel@... https://lists.sourceforge.net/lists/listinfo/opencvlibrary-devel |
|
|
Re: Possible memory allocation bug
by Jose Luis Blanco
::
Rate this Message:
Running the test program (attached in my last mail) with Valgrind
reveals a few invalid memory accesses upon the first call to cvGoodFeaturesToTrack. It's possible that the error only happens for some precise image sizes, hence it passed unnoticed before... I attach the log file. Best, JL On Fri, Sep 4, 2009 at 2:52 AM, Guy K. Kloss<g.kloss@...> wrote: > On Thu, 03 Sep 2009 22:10:11 Jose Luis Blanco wrote: >> I've not been able to get a consistent test code, but I'm having >> program crashes (in Linux) and asserts (in Windows) when calling >> "cvGoodFeaturesToTrack". > > It might help to to run the code under questioning through valgrind or compile > it with libefence to see whether there's some boundary violation. > > Guy > > -- > Guy K. Kloss > Institute of Information and Mathematical Sciences > Te Kura Pūtaiao o Mōhiohio me Pāngarau > Massey University, Albany (North Shore City, Auckland) > 473 State Highway 17, Gate 1, Mailroom, Quad B Building > voice: +64 9 414-0800 ext. 9585 fax: +64 9 441-8181 > G.Kloss@... http://www.massey.ac.nz/~gkloss > > ------------------------------------------------------------------------------ > Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day > trial. Simplify your report design, integration and deployment - and focus on > what you do best, core application coding. Discover what's new with > Crystal Reports now. http://p.sf.net/sfu/bobj-july > _______________________________________________ > Opencvlibrary-devel mailing list > Opencvlibrary-devel@... > https://lists.sourceforge.net/lists/listinfo/opencvlibrary-devel > -- ___________________________________________________________ Jose-Luis Blanco-Claraco Phone: +34 952 132848 Dpto. Ingenieria de Sistemas y Automatica E.T.S.I. Telecomunicacion Fax: +34 952 133361 Universidad de Malaga Campus Universitario de Teatinos 29071 Malaga, Spain http://www.isa.uma.es/jlblanco ___________________________________________________________ [valgrind.log] valgrind ./opencv_example ==9362== Memcheck, a memory error detector. ==9362== Copyright (C) 2002-2008, and GNU GPL'd, by Julian Seward et al. ==9362== Using LibVEX rev 1884, a library for dynamic binary translation. ==9362== Copyright (C) 2004-2008, and GNU GPL'd, by OpenWorks LLP. ==9362== Using valgrind-3.4.1-Debian, a dynamic binary instrumentation framework. ==9362== Copyright (C) 2000-2008, and GNU GPL'd, by Julian Seward et al. ==9362== For more details, rerun with: -v ==9362== 0... ==9362== Invalid read of size 1 ==9362== at 0x4C28500: memcpy (mc_replace_strmem.c:402) ==9362== by 0x539C9C3: cv::FilterEngine::start(cv::Size_<int>, cv::Rect_<int>, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x539CDD5: cv::FilterEngine::start(cv::Mat const&, cv::Rect_<int> const&, bool, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x53A600D: cv::FilterEngine::apply(cv::Mat const&, cv::Mat&, cv::Rect_<int> const&, cv::Point_<int>, bool) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C113: cv::morphOp(int, cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C7E9: cv::dilate(cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5382FD9: cv::goodFeaturesToTrack(cv::Mat const&, std::vector<cv::Point_<float>, std::allocator<cv::Point_<float> > >&, int, double, double, cv::Mat const&, int, bool, double) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5383B53: cvGoodFeaturesToTrack (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x401968: test_cv_bug() (test.cpp:26) ==9362== by 0x401A37: main (test.cpp:35) ==9362== Address 0x108417bf is not stack'd, malloc'd or (recently) free'd ==9362== ==9362== Invalid read of size 1 ==9362== at 0x4C28508: memcpy (mc_replace_strmem.c:402) ==9362== by 0x539C9C3: cv::FilterEngine::start(cv::Size_<int>, cv::Rect_<int>, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x539CDD5: cv::FilterEngine::start(cv::Mat const&, cv::Rect_<int> const&, bool, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x53A600D: cv::FilterEngine::apply(cv::Mat const&, cv::Mat&, cv::Rect_<int> const&, cv::Point_<int>, bool) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C113: cv::morphOp(int, cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C7E9: cv::dilate(cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5382FD9: cv::goodFeaturesToTrack(cv::Mat const&, std::vector<cv::Point_<float>, std::allocator<cv::Point_<float> > >&, int, double, double, cv::Mat const&, int, bool, double) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5383B53: cvGoodFeaturesToTrack (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x401968: test_cv_bug() (test.cpp:26) ==9362== by 0x401A37: main (test.cpp:35) ==9362== Address 0x108417be is not stack'd, malloc'd or (recently) free'd ==9362== ==9362== Invalid read of size 1 ==9362== at 0x4C28512: memcpy (mc_replace_strmem.c:402) ==9362== by 0x539C9C3: cv::FilterEngine::start(cv::Size_<int>, cv::Rect_<int>, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x539CDD5: cv::FilterEngine::start(cv::Mat const&, cv::Rect_<int> const&, bool, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x53A600D: cv::FilterEngine::apply(cv::Mat const&, cv::Mat&, cv::Rect_<int> const&, cv::Point_<int>, bool) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C113: cv::morphOp(int, cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C7E9: cv::dilate(cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5382FD9: cv::goodFeaturesToTrack(cv::Mat const&, std::vector<cv::Point_<float>, std::allocator<cv::Point_<float> > >&, int, double, double, cv::Mat const&, int, bool, double) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5383B53: cvGoodFeaturesToTrack (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x401968: test_cv_bug() (test.cpp:26) ==9362== by 0x401A37: main (test.cpp:35) ==9362== Address 0x108417bd is not stack'd, malloc'd or (recently) free'd ==9362== ==9362== Invalid read of size 1 ==9362== at 0x4C2851C: memcpy (mc_replace_strmem.c:402) ==9362== by 0x539C9C3: cv::FilterEngine::start(cv::Size_<int>, cv::Rect_<int>, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x539CDD5: cv::FilterEngine::start(cv::Mat const&, cv::Rect_<int> const&, bool, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x53A600D: cv::FilterEngine::apply(cv::Mat const&, cv::Mat&, cv::Rect_<int> const&, cv::Point_<int>, bool) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C113: cv::morphOp(int, cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C7E9: cv::dilate(cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5382FD9: cv::goodFeaturesToTrack(cv::Mat const&, std::vector<cv::Point_<float>, std::allocator<cv::Point_<float> > >&, int, double, double, cv::Mat const&, int, bool, double) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5383B53: cvGoodFeaturesToTrack (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x401968: test_cv_bug() (test.cpp:26) ==9362== by 0x401A37: main (test.cpp:35) ==9362== Address 0x108417bc is not stack'd, malloc'd or (recently) free'd ==9362== ==9362== Invalid read of size 8 ==9362== at 0x547D163: cv::MorphRowFilter<cv::MaxOp<float>, cv::MorphRowFVec<cv::VMax32f> >::operator()(unsigned char const*, unsigned char*, int, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x539CB09: cv::FilterEngine::start(cv::Size_<int>, cv::Rect_<int>, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x539CDD5: cv::FilterEngine::start(cv::Mat const&, cv::Rect_<int> const&, bool, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x53A600D: cv::FilterEngine::apply(cv::Mat const&, cv::Mat&, cv::Rect_<int> const&, cv::Point_<int>, bool) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C113: cv::morphOp(int, cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C7E9: cv::dilate(cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5382FD9: cv::goodFeaturesToTrack(cv::Mat const&, std::vector<cv::Point_<float>, std::allocator<cv::Point_<float> > >&, int, double, double, cv::Mat const&, int, bool, double) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5383B53: cvGoodFeaturesToTrack (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x401968: test_cv_bug() (test.cpp:26) ==9362== by 0x401A37: main (test.cpp:35) ==9362== Address 0x10844644 is 5,124 bytes inside a block of size 5,128 alloc'd ==9362== at 0x4C2726C: operator new(unsigned long) (vg_replace_malloc.c:230) ==9362== by 0x4F46B9F: std::vector<unsigned char, std::allocator<unsigned char> >::_M_fill_insert(__gnu_cxx::__normal_iterator<unsigned char*, std::vector<unsigned char, std::allocator<unsigned char> > >, unsigned long, unsigned char const&) (in /usr/local/lib/libcxcore.so.1.1.1) ==9362== by 0x539C8F2: cv::FilterEngine::start(cv::Size_<int>, cv::Rect_<int>, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x539CDD5: cv::FilterEngine::start(cv::Mat const&, cv::Rect_<int> const&, bool, int) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x53A600D: cv::FilterEngine::apply(cv::Mat const&, cv::Mat&, cv::Rect_<int> const&, cv::Point_<int>, bool) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C113: cv::morphOp(int, cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x548C7E9: cv::dilate(cv::Mat const&, cv::Mat&, cv::Mat const&, cv::Point_<int>, int, int, cv::Scalar_<double> const&) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5382FD9: cv::goodFeaturesToTrack(cv::Mat const&, std::vector<cv::Point_<float>, std::allocator<cv::Point_<float> > >&, int, double, double, cv::Mat const&, int, bool, double) (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x5383B53: cvGoodFeaturesToTrack (in /usr/local/lib/libcv.so.1.1.1) ==9362== by 0x401968: test_cv_bug() (test.cpp:26) ==9362== by 0x401A37: main (test.cpp:35) OpenCV Error: Assertion failed (((size_t)_src[i] & 15) == 0) in operator(), file /home/jlblanco/code/opencv-svn/src/cv/cvmorph.cpp, line 260 terminate called after throwing an instance of 'cv::Exception' ==9362== ==9362== ERROR SUMMARY: 3841 errors from 5 contexts (suppressed: 8 from 1) ==9362== malloc/free: in use at exit: 13,543,360 bytes in 145 blocks. ==9362== malloc/free: 234 allocs, 89 frees, 29,751,434 bytes allocated. ==9362== For counts of detected errors, rerun with: -v ==9362== searching for pointers to 145 not-freed blocks. ==9362== checked 11,246,624 bytes. ==9362== ==9362== LEAK SUMMARY: ==9362== definitely lost: 0 bytes in 0 blocks. ==9362== possibly lost: 718 bytes in 15 blocks. ==9362== still reachable: 13,542,642 bytes in 130 blocks. ==9362== suppressed: 0 bytes in 0 blocks. ==9362== Rerun with --leak-check=full to see details of leaked memory. ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Opencvlibrary-devel mailing list Opencvlibrary-devel@... https://lists.sourceforge.net/lists/listinfo/opencvlibrary-devel |
|
|
Re: Possible memory allocation bug
by Vadim Pisarevsky
::
Rate this Message:
Hello Jose Luis,
thank you for the sample code, I fixed the problem; please, take the updated cvfilter.cpp from SVN. Regards, Vadim On Fri, Sep 4, 2009 at 11:33 AM, Jose Luis Blanco <joseluisblancoc@...> wrote: Running the test program (attached in my last mail) with Valgrind ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Opencvlibrary-devel mailing list Opencvlibrary-devel@... https://lists.sourceforge.net/lists/listinfo/opencvlibrary-devel |
| Free embeddable forum powered by Nabble | Forum Help |
