| • heime | |
| Nov 14 2012, 3:52 PM |
| F926: valgrind.txt | |
| Nov 15 2012, 11:24 AM |
| F925: stacktrace.png | |
| Nov 14 2012, 3:53 PM |
If you load the test data spacing-ok-ct.dcm or any other .dcm file two times via the "open" filedialog of the mitkWorkbench, the application crashes. This also happens if you close the opened file / Project and load it again. See attached stacktrace.
Right now Thomas and I have no idea why this happens.
@Daniel: Do you think this could be related to your latest changes to dicom loading?
Well, at least the crash is happening only after my changes and NOT before. So at least I changed something that makes this crash possible. I'll look into it.
Eric or Thomas (or some other rendering guy), would you perhaps have some time to look into this one?
I tried to trace the error but ended up with a meaningless call stack. Running valgrind's memcheck pointed me to something that I don't understand and that should be related to the crash (see attachment).
As I understand the valgrind output, a part of ExtractSliceFilter tries to read a piece of (VTK) memory which has previously been destroyed.
Can you tell what is happening? Any idea how this is related to my changes? I don't use VTK code, at least not on purpose. The whole loading is done by ITK/GDCM.
Your valgrind output has the same classes involved than the stacktrace that gdb gives me in qtcreator. I also find it strange that there are VTK and other renderingclasses involved and the crash happens there. Since the Dicomseriesreader only uses ITK and GDCM.
What is the status of this bug? I could not reproduce a crash. Please provide a quick answer so that we can decide whether this bug is relevant for the upcoming release
(In reply to comment #9)
What is the status of this bug? I could not reproduce a crash. Please
provide a quick answer so that we can decide whether this bug is relevant
for the upcoming release
If by "you" you mean the assignee, there is nothing new since comment 3. Marco pointed out that there is similar behavior with PIC files.
Since I do not understand the problem properly and I cannot see that this is a regularly occurring problem, I am currently not actively working on this bug. Would be helpful to get feedback from the rendering team about what the rendering classes are missing here.. (since they crash)
