I'm seeing a 65541-Byte memory leak every time a call to "Magick::Image::profile(std::string) const" doesn't find the profile specified in the image.
(Version 6.6.9.7, ubuntu 12.04 package)
Example code to repoduce:
Code: Select all
#include <Magick++.h>
#include <iostream>
int main(int argc, const char * argv[]) {
Magick::Image img(argv[1]);
std::cout<<"Image loaded ("<<img.columns()<<"x"<<img.rows()<<")"<<std::endl;
Magick::Blob blob;
try {
blob = img.profile("8BIM");
} catch(const Magick::Exception & e) {
std::cerr << "ME: " << e.what() << std::endl;
} catch (...) {
std::cerr << "unknown exception" << std::endl;
}
std::cout << "data: "<< blob.length()<<std::endl;
return 0;
}
Image loaded (115x158)
data: 738
and no leaked memory.
Called with a plain jpeg without 8Bim, the output is:
Image loaded (115x158)
ME: Magick: no 8BIM data is available `' @ error/meta.c/WriteMETAImage/2269
data: 0
and the application leaks ~64KiB.
valgrind says:
==31033== 65,541 bytes in 1 blocks are possibly lost in loss record 34,205 of 34,392
==31033== at 0x402BE68: malloc (in /usr/lib/valgrind/vgpreload_memcheck-x86-linux.so)
==31033== by 0x4E1E505: AcquireMagickMemory (in /usr/lib/libMagickCore.so.4.0.1)
==31033== by 0x4E1E54C: AcquireQuantumMemory (in /usr/lib/libMagickCore.so.4.0.1)
==31033== by 0x4D3AFED: ImageToBlob (in /usr/lib/libMagickCore.so.4.0.1)
==31033== by 0x49834B7: Magick::Image::write(Magick::Blob*, std::string const&) (in /usr/lib/libMagick++.so.4.0.1)
==31033== by 0x49835C3: Magick::Image::profile(std::string) const (in /usr/lib/libMagick++.so.4.0.1)
My first guess would be a problem with exception handling inside Image::write, called on the temporary image copy created by Image::profile.
Any ideas for a quick workaround? Is there a way to check in advance if a 8BIM-profile exists, so no exception is thrown?
Thanks,
Ernst
