Sending valgrind logs to...
Cyril Giraud
cgiraud at free.fr
Mon Jul 13 19:15:41 BST 2026
Command launched:
$ valgrind --tool=memcheck --leak-check=full --show-leak-kinds=all
--error-limit=no digikam 2>&1 | tee ./digiKam-9.0.0.valgrind2.log
Then, digikams opens my database, I preview a photo and zoom inside,
then exit.
This is the default Ubuntu Studio 26.04 LTS configuration.
I don't have any problem with digikam (unless I am a newbee user ;-)),
just trying to help.
Regards,
Le 13/07/2026 à 15:23, Gilles Caulier a écrit :
> Look at the end of the log : the main memory leak is due to OPenCV DNN
> model internal management :
>
> ==1415992== 2,964,624 bytes in 2 blocks are possibly lost in loss
> record 4,664 of 4,669
> ==1415992== at 0x485880F: malloc (vg_replace_malloc.c:447)
> ==1415992== by 0x9F88CC0: cv::fastMalloc(unsigned long) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA08D13F: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA088315: cv::Mat::create(int, int const*, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA0D9E29: cv::_OutputArray::create(int, int const*,
> int, int, bool, cv::_OutputArray::DepthMask) const (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0x9FEC872: cv::Mat::copyTo(cv::_OutputArray const&)
> const (in /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0x96E8EDE: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96F5A17: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x972A5F7: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9722DE4:
> cv::dnn::dnn4_v20240521::readNetFromONNX(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9520AD1:
> cv::dnn::dnn4_v20240521::readNet(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&,
> std::__cxx11::basic_string<char, std::char_traits<char>,
> std::allocator<char> > const&, std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x62935D0: Digikam::DNNModelNet::callLoader(QString
> const&) (in /usr/lib/digikam/libdigikamcore.so.9.0.0)
> ==1415992==
> ==1415992== 4,172,486 (40 direct, 4,172,446 indirect) bytes in 1
> blocks are definitely lost in loss record 4,666 of 4,669
> ==1415992== at 0x4858F95: operator new(unsigned long)
> (vg_replace_malloc.c:488)
> ==1415992== by 0x969125C:
> cv::dnn::dnn4_v20240521::Net::Impl::connect(int, int, int, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96913D8:
> cv::dnn::dnn4_v20240521::Net::connect(int, int, int, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96ED023: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96FCEFD: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96F3C25: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96F7297: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x972A5F7: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9722DE4:
> cv::dnn::dnn4_v20240521::readNetFromONNX(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9520AD1:
> cv::dnn::dnn4_v20240521::readNet(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&,
> std::__cxx11::basic_string<char, std::char_traits<char>,
> std::allocator<char> > const&, std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x10E0A1EE:
> cv::FaceRecognizerSF::create(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&,
> std::__cxx11::basic_string<char, std::char_traits<char>,
> std::allocator<char> > const&, int, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_objdetect.so.4.10.0)
> ==1415992== by 0x62946FC: Digikam::DNNModelSFace::callLoader() (in
> /usr/lib/digikam/libdigikamcore.so.9.0.0)
> ==1415992==
> ==1415992== 9,175,472 bytes in 6 blocks are possibly lost in loss
> record 4,667 of 4,669
> ==1415992== at 0x485880F: malloc (vg_replace_malloc.c:447)
> ==1415992== by 0x9F88CC0: cv::fastMalloc(unsigned long) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA08D13F: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA088315: cv::Mat::create(int, int const*, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA0D9E29: cv::_OutputArray::create(int, int const*,
> int, int, bool, cv::_OutputArray::DepthMask) const (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0x9FEC872: cv::Mat::copyTo(cv::_OutputArray const&)
> const (in /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0x96E8AA8: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96F5A17: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x972A5F7: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9722DE4:
> cv::dnn::dnn4_v20240521::readNetFromONNX(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9520AD1:
> cv::dnn::dnn4_v20240521::readNet(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&,
> std::__cxx11::basic_string<char, std::char_traits<char>,
> std::allocator<char> > const&, std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x10E0A1EE:
> cv::FaceRecognizerSF::create(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&,
> std::__cxx11::basic_string<char, std::char_traits<char>,
> std::allocator<char> > const&, int, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_objdetect.so.4.10.0)
> ==1415992==
> ==1415992== 25,690,184 bytes in 1 blocks are possibly lost in loss
> record 4,668 of 4,669
> ==1415992== at 0x485880F: malloc (vg_replace_malloc.c:447)
> ==1415992== by 0x9F88CC0: cv::fastMalloc(unsigned long) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA08D13F: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA088315: cv::Mat::create(int, int const*, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA08869F: cv::Mat::create(int, int, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0x9FEC71B: cv::Mat::copyTo(cv::_OutputArray const&)
> const (in /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0x96E8AA8: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96F5A17: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x972A5F7: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9722DE4:
> cv::dnn::dnn4_v20240521::readNetFromONNX(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9520AD1:
> cv::dnn::dnn4_v20240521::readNet(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&,
> std::__cxx11::basic_string<char, std::char_traits<char>,
> std::allocator<char> > const&, std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x10E0A1EE:
> cv::FaceRecognizerSF::create(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&,
> std::__cxx11::basic_string<char, std::char_traits<char>,
> std::allocator<char> > const&, int, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_objdetect.so.4.10.0)
> ==1415992==
> ==1415992== 342,113,140 bytes in 83 blocks are possibly lost in loss
> record 4,669 of 4,669
> ==1415992== at 0x485880F: malloc (vg_replace_malloc.c:447)
> ==1415992== by 0x9F88CC0: cv::fastMalloc(unsigned long) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA08D13F: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA088315: cv::Mat::create(int, int const*, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0xA08869F: cv::Mat::create(int, int, int) (in
> /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0x9FEC71B: cv::Mat::copyTo(cv::_OutputArray const&)
> const (in /usr/lib/x86_64-linux-gnu/libopencv_core.so.4.10.0)
> ==1415992== by 0x96E8EDE: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x96F5A17: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x972A5F7: ??? (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9722DE4:
> cv::dnn::dnn4_v20240521::readNetFromONNX(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x9520AD1:
> cv::dnn::dnn4_v20240521::readNet(std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&,
> std::__cxx11::basic_string<char, std::char_traits<char>,
> std::allocator<char> > const&, std::__cxx11::basic_string<char,
> std::char_traits<char>, std::allocator<char> > const&) (in
> /usr/lib/x86_64-linux-gnu/libopencv_dnn.so.4.10.0)
> ==1415992== by 0x62935D0: Digikam::DNNModelNet::callLoader(QString
> const&) (in /usr/lib/digikam/libdigikamcore.so.9.0.0)
> ==1415992==
>
> It's not at all a dysfunction of digiKAm as well. OPenCV4.10 is used
> on your system, OPenCV 4.13 is out with probably the fixes for this
> kind of memory leak.
>
> Nota : OpenCV 5.0 available since few weeks ago is not yet supported
> in digiKam, as the DNN engine have been completely rewritten and lack
> features provided by 4.x version (eg. the GPU support for ex)...
> So OpenCV 4 must be used with digiKAm for the moment, not the OpenCV 5
> (and it will still like this until OPenCV taem fix all the
> regressions.
>
> Best regards
>
> Gilles Caulier
>
> Le lun. 13 juil. 2026 à 15:16, Gilles Caulier
> <caulier.gilles at gmail.com> a écrit :
>> Did you use the native digiKam package from your distribution ?
>>
>> Which operations did you run in digiKam exactly ?
>>
>> Best regards
>>
>> Gilles Caulier
>>
>> Le lun. 13 juil. 2026 à 13:44, Cyril Giraud <cgiraud at free.fr> a écrit :
>>> Hi Gilles,
>>>
>>> Probably yes for digikam 9.0.0 (native for UbuntuStudio 26.04):
>>> ==1415992== LEAK SUMMARY:
>>> ==1415992== definitely lost: 9,208 bytes in 93 blocks
>>> ==1415992== indirectly lost: 17,819,646 bytes in 71,055 blocks
>>> ==1415992== possibly lost: 387,519,486 bytes in 44,750 blocks
>>> ==1415992== still reachable: 1,377,846 bytes in 4,720 blocks
>>> ==1415992== of which reachable via heuristic:
>>> ==1415992== newarray : 381,304 bytes in 12 blocks
>>> ==1415992== suppressed: 0 bytes in 0 blocks
>>> ==1415992== Reachable blocks (those to which a pointer was found) are not shown.
>>> ==1415992== To see them, rerun with: --leak-check=full --show-leak-kinds=all
>>>
>>> Rerun today with --show-leak-kinds=all (probably not exactly the same GUI actions):
>>>
>>> ==7409== LEAK SUMMARY:
>>> ==7409== definitely lost: 10,144 bytes in 96 blocks
>>> ==7409== indirectly lost: 17,853,762 bytes in 76,878 blocks
>>> ==7409== possibly lost: 387,503,190 bytes in 39,004 blocks
>>> ==7409== still reachable: 1,366,375 bytes in 4,637 blocks
>>> ==7409== of which reachable via heuristic:
>>> ==7409== newarray : 376,536 bytes in 10 blocks
>>> ==7409== suppressed: 0 bytes in 0 blocks
>>>
>>>
>>> Full logs here:
>>> J’ai lié 2 fichiers à ce message :
>>>
>>> digiKam-9.0.0.valgrind.log
>>> Taille : 797 Ko
>>>
>>> Date d’expiration : 14/08/2026
>>> NextCloud
>>> Lien : https://nextcloud.cgiraud.fr/index.php/s/CPQKTzGGPwkHpmy/download
>>>
>>> digiKam-9.0.0.valgrind2.log
>>> Taille : 5.6 Mo
>>>
>>> Date d’expiration : 14/08/2026
>>> Service Filelink : NextCloud
>>> Lien : https://nextcloud.cgiraud.fr/index.php/s/ap2AqoKHatCKGpG/download
>>>
>>> I tried to run digikam 9.1.0 but running appimage with valgrind seems not working.
>>>
>>> Best regards,
>>> Cyril
>>>
>>>
>>>
>>>
>>> Le 12/07/2026 à 22:57, Gilles Caulier a écrit :
>>>
>>> Hi
>>> did you observe a potential memory leak? In which context?
>>>
>>> You can share the trace in this room using a cloud storage link.
>>>
>>> Best regards
>>>
>>> Gilles Caulier
>>>
>>> Le dim. 12 juil. 2026 à 22:51, Cyril Giraud <cgiraud at free.fr> a écrit :
>>>
>>> Hi,
>>>
>>> https://www.digikam.org/contribute / Checking for Memory Leaks asks to
>>> send valgrind output to developers.
>>>
>>> Could someone give me an email address or a way to send these outputs?
>>>
>>> Thanks !
>>>
>>> Cyril
>>>
>>>
More information about the Digikam-users
mailing list