[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: For those who would like to debug some old C++ code over the weekend



On 02/14/2014 02:51 PM, Andreas Tille wrote:
> Hi,
>
> I'm currently working on preconditions for ngopt[1] which is used in our
> institute.  It includes some third party pieces of code and one of this
> is FISH[2].  I have created some packaging for this at
>
>   svn://anonscm.debian.org/debian-med/trunk/packages/unc-fish/trunk/
>
> but when I try to process the provided test data it segfaults:
>
> $ /usr/lib/unc-fish/fish -b blocks.txt -B simpleblocks.txt -g grids.txt -C contigs.txt
>      Segmentation fault
It fails at:
Program terminated with signal 11, Segmentation fault.
#0  0xf74fcad6 in vfprintf () from /lib/i386-linux-gnu/libc.so.6
(gdb) where
#0  0xf74fcad6 in vfprintf () from /lib/i386-linux-gnu/libc.so.6
#1  0xf7506ebf in fprintf () from /lib/i386-linux-gnu/libc.so.6
#2  0x0805166d in grids () at grids.h:278
#3  0x080540aa in main (argc=11, argv=0xffbd9654) at fish.cpp:341

on fprintf(print_grid, "%s",get_time_stamp());

maybe an issue because get_time_stamp returns a char* or in time-stamp
settings.

Olivier


>
> There are some suspicious format warnings when building but the problem
> might be also somewhere else.  Since ngopt is my final target I dived
> into their repository and have takoen the content of
>
>    http://code.google.com/p/ngopt/source/browse/#svn%2Ftrunk%2FFISH%2Fsrc
>
> as a quilt patch - which at least solves the segfault issue.  So without
> this patch the original code segfaults with the sample data set but due
> to some other changes the result set of the test code can not be
> reproduced.  (BTW, I found a binary of fish on some non-Debian host used
> in my institute which runs as well and produces a third result which is
> different from the original test result and the patched version.)
>
> This issue makes me quite suspicious and I wonder whether we could at
> least reproduce the original test suite with the original code without
> the segfault problem.  Any volunteer to check the code for this issue?
>
> Kind regards
>
>        Andreas.
>
> [1] http://code.google.com/p/ngopt/
> [2] http://labs.bio.unc.edu/Vision/FISH/
>

-- 
Olivier Sallou
IRISA / University of Rennes 1
Campus de Beaulieu, 35000 RENNES - FRANCE
Tel: 02.99.84.71.95

gpg key id: 4096R/326D8438  (keyring.debian.org)
Key fingerprint = 5FB4 6F83 D3B9 5204 6335  D26D 78DC 68DB 326D 8438


Reply to: