Re: segfault during test

[Max Brister]

On Thu, Aug 2, 2012 at 12:15 PM, JuanPi <address@hidden> wrote:
> On Thu, Aug 2, 2012 at 6:45 PM, John W. Eaton <address@hidden> wrote:
>> On  2-Aug-2012, Rik wrote:
>>
>> | It looks like all the easy channels for debug are closed.
>> |
>> | I would start by configuring a debug build of Octave with CFLAGS '-g -O0'.
>> | When you have that built, run it under gdb with './run-octave -g'.
>> | This will put you at the gdb prompt.  Type 'run' to start Octave.
>> | When Octave has started, type 'cd test' and then 'fntests'
>> | This will start the integrated test suite usually run by 'make check'.  If
>> | you have separate build and src directories you will need to cd to the
>> | actual location of fntests.m to run it.
>>
>> Maybe we should make something like the following change?  Then we
>> could simplify these instructions to
>>
>>   run "make check RUN_OCTAVE_FLAGS=-g", then at the (gdb) prompt type "r"
>>
>> jwe
>>
>>
>> diff --git a/test/Makefile.am b/test/Makefile.am
>> --- a/test/Makefile.am
>> +++ b/test/Makefile.am
>> @@ -58,8 +58,11 @@
>>  include fcn-handle-derived-resolution/module.mk
>>  include nest/module.mk
>>
>> +RUN_OCTAVE := $(top_builddir)/run-octave
>> +RUN_OCTAVE_FLAGS := --norc --silent --no-history
>> +
>>  check: test_sparse.m test_bc_overloads.m
>> -       $(top_builddir)/run-octave --norc --silent --no-history 
>> $(srcdir)/fntests.m $(srcdir)
>> +       $(RUN_OCTAVE) $(RUN_OCTAVE_FLAGS) $(srcdir)/fntests.m $(srcdir)
>>
>>  test_sparse.m: build_sparse_tests.sh
>>         $(srcdir)/build_sparse_tests.sh

That would be nice.

> Using ./run-octave -g after the suggested build produces no segfault.
> fntests finishes normally with the summary
>
>   PASS     10476
>   FAIL         2
>   XFAIL        3
>
> However if I do "make check" I still get the segmentation fault. It is
> something after fntests returns.

Can you try the same thing, but with
./run-octave -valgrind
instead? It will take forever to run, but valgrind is generally pretty
good about reporting memory related errors.

> --
> JuanPi Carbajal
> -----
> "The bad economist pursues a small present good, which will be
> followed by a great evil to come, while the true economist pursues a
> great good to come, at the risk of a small present evil." - Frédéric
> Bastiat
> -----
> http://ailab.ifi.uzh.ch/carbajal/

Max Brister