[Neurodebian-users] fslview segmentation fault
romain valabregue
romain.valabregue at upmc.fr
Mon Jan 21 14:44:29 UTC 2013
Thanks for the information
strangly gdb does not complain and fslview works (when running whithin gdb)
with valgrind fslview to I get the window and I can open volume (whereas
I do not get any window with fslview)
~ > fslview
Erreur de segmentation
~ > valgrind fslview
==6231== Memcheck, a memory error detector
==6231== Copyright (C) 2002-2010, and GNU GPL'd, by Julian Seward et al.
==6231== Using Valgrind-3.6.0.SVN-Debian and LibVEX; rerun with -h for
copyright info
==6231== Command: fslview
==6231==
==6231== Conditional jump or move depends on uninitialised value(s)
==6231== at 0xAA20290: inflateReset2 (in /usr/lib/libz.so.1.2.3.4)
==6231== by 0xAA2037F: inflateInit2_ (in /usr/lib/libz.so.1.2.3.4)
==6231== by 0xA802FC3: png_create_read_struct_2 (in
/lib/libpng12.so.0.44.0)
==6231== by 0xA803176: png_create_read_struct (in
/lib/libpng12.so.0.44.0)
==6231== by 0x53FF31E: ??? (in /usr/lib/libqt-mt.so.3.3.8)
==6231== by 0x517B749: QImageIO::read() (in /usr/lib/libqt-mt.so.3.3.8)
==6231== by 0x517B924: QImage::loadFromData(unsigned char const*,
unsigned int, char const*) (in /usr/lib/libqt-mt.so.3.3.8)
==6231== by 0x510C81:
ApplicationWindowBase::ApplicationWindowBase(QWidget*, char const*,
unsigned int) (in /usr/bin/fslview)
==6231== by 0x488257:
ApplicationWindow::ApplicationWindow(ApplicationOptions&) (in
/usr/bin/fslview)
==6231== by 0x48D62C: SplashScreen::runApplication() (in
/usr/bin/fslview)
==6231== by 0x545F3C: SplashScreen::qt_invoke(int, QUObject*) (in
/usr/bin/fslview)
==6231== by 0x519BE33: QObject::activate_signal(QConnectionList*,
QUObject*) (in /usr/lib/libqt-mt.so.3.3.8)
==6231==
==6231==
==6231== HEAP SUMMARY:
==6231== in use at exit: 266,259 bytes in 1,804 blocks
==6231== total heap usage: 1,290,156 allocs, 1,288,352 frees,
26,927,539 bytes allocated
==6231==
==6231== LEAK SUMMARY:
==6231== definitely lost: 1,940 bytes in 5 blocks
==6231== indirectly lost: 114,464 bytes in 204 blocks
==6231== possibly lost: 5,612 bytes in 171 blocks
==6231== still reachable: 144,243 bytes in 1,424 blocks
==6231== suppressed: 0 bytes in 0 blocks
==6231== Rerun with --leak-check=full to see details of leaked memory
==6231==
==6231== For counts of detected and suppressed errors, rerun with: -v
==6231== Use --track-origins=yes to see where uninitialised values come from
==6231== ERROR SUMMARY: 2 errors from 1 contexts (suppressed: 8 from 6)
Le 21/01/2013 15:18, Yaroslav Halchenko a écrit :
> fslview has quite a bit of not fixed quirks (e.g. inability to deal with
> int8 dtype niftis etc). Since there is no proper bugtracking (besides
> Debian's [1]) for it AFAIK -- we have no formal record of all of them.
>
> you might get a better idea if you run it via gdb and checkout the crash
> stack:
>
> gdb fslview
> type r<enter>
>
> and when crashes, it no trackback, type 'bt', and share
>
> alternative to hint the problem might be to run via valgrind,
>
> sudo apt-get install valgrind
> valgrind fslview
>
> it will be slow but can pin point to specific problems with memory
> access etc
>
> [1] http://bugs.debian.org/cgi-bin/pkgreport.cgi?ordering=normal;archive=0;src=fslview;dist=unstable;repeatmerged=0
>
> On Mon, 21 Jan 2013, romain valabregue wrote:
>
>> Hello
>
>> I am working with Debian GNU/Linux 6.0.6 (squeeze)
>
>> and I get a segmentation fault when I try to launch fslview.
>> This is on computeurs I recently installed whereas it works fine on
>> other computer (runing the same debian).
>> Does anyone know how can I have a clue of what is wrong ?
>> the only information I have is from dmesg
>> [ 2562.225990] fslview[18031]: segfault at 0 ip 00007f75b64241ba sp
>> 00007fffee65da08 error 4 in libc-2.11.3.so[7f75b631d000+159000]
>
>> My guess would be it comes from the nvidia installation although I
>> install properly the nvidia driver ...
>> The strange thing is that other viewer that use a graphic
>> acceleration works fine (mrview (from mrtrix) freeview (freesurfer)
>> anatomist (brainvisa) )
>
>> Many thanks
>
>> Romain
More information about the Neurodebian-users
mailing list