oops -- since actual body of this very email wasn't already duplicated, here it is:
Dear ITK maintainer, see attached patch which fixes memory leak for
subsequent loads of NIFTI files. I've tested the patch with ants --
this huge leak seems to be gone... there is something left (see below)
but that might still be ANTs issue (thus feel free to ignore, for now
;)):
==12852== 1,648 (784 direct, 864 indirect) bytes in 2 blocks are definitely
lost in loss record 7 of 7
==12852== at 0x4C249C7: operator new(unsigned long) (vg_replace_malloc.c:220)
==12852== by 0x483391: itk::MatrixOffsetTransformBase<double, 3u, 3u>::New()
(itkMatrixOffsetTransformBase.h:95)
==12852== by 0x4BC962:
itk::CreateObjectFunction<itk::MatrixOffsetTransformBase<double, 3u, 3u>
>::CreateObject() (itkCreateObjectFunction.h:69)
==12852== by 0x4EFA132: itk::ObjectFactoryBase::CreateObject(char const*)
(in /usr/lib/libITKCommon.so.3.18.0)
==12852== by 0x4EFBA74: itk::ObjectFactoryBase::CreateInstance(char const*)
(in /usr/lib/libITKCommon.so.3.18.0)
==12852== by 0x5306FDB: itk::TransformFileReader::Update() (in
/usr/lib/libITKIO.so.3.18.0)
==12852== by 0x47A97E: void WarpImageMultiTransformFourD<4>(char*, char*,
std::vector<TRAN_OPT, std::allocator<TRAN_OPT> >&, MISC_OPT&)
(WarpTimeSeriesImageMultiTransform.cxx:569)
==12852== by 0x473235: main (WarpTimeSeriesImageMultiTransform.cxx:1004)
--
.-.
=------------------------------ /v\ ----------------------------=
Keep in touch // \\ (yoh@|www.)onerussian.com
Yaroslav Halchenko /( )\ ICQ#: 60653192
Linux User ^^-^^ [175555]
--- insighttoolkit-3.18.0.orig/Code/IO/itkNiftiImageIO.cxx
+++ insighttoolkit-3.18.0/Code/IO/itkNiftiImageIO.cxx
@@ -457,6 +457,10 @@
_size[4] = numComponents;
}
+ // Free memory if any was occupied already
+ if (this->m_NiftiImage != NULL)
+ nifti_image_free(this->m_NiftiImage);
+
//
// allocate nifti image...
this->m_NiftiImage = nifti_image_read(this->GetFileName(),false);
signature.asc
Description: Digital signature
