Re: compiling development sources

[Jaroslav Hajek]

On Thu, Feb 11, 2010 at 12:58 PM, Carlo de Falco
<address@hidden> wrote:
>
> On 11 Feb 2010, at 12:50, Jaroslav Hajek wrote:
>
>> It's surely worth trying, although as I said, not a single line of
>> code in the library routines has changed since 1.1.0, it's just the
>> test suite, so in theory the binary should be identical to 1.1.0.
>> Compared to 1.0, only 8 new routines were added. If you start seeing
>> the test failures in Octave again, try rebuilding and re-running the
>> qrupdate test suite. For me, the manifestation of the bug seemed to be
>> dependent on previous calls to BLAS made, and it apparently made a
>> difference whether the BLAS remained in memory or not.
>> In any case, it will help if you discover which test was failing for you:
>> ./run-octave
>> octave:1> test qr verbose
>>
>> this should dump the tests as they're tried so you'll see which one
>> crashed.
>>
>
> As I mentioned in a previous post, I could reproduce the segfault in Octave
> with the following test from the qrupdate testsuite:
>
>> Ac = [0.620405 + 0.956953i  0.480013 + 0.048806i  0.402627 + 0.338171i;
>>     0.589077 + 0.658457i  0.013205 + 0.279323i  0.229284 + 0.721929i;
>>     0.092758 + 0.345687i  0.928679 + 0.241052i  0.764536 + 0.832406i;
>>     0.912098 + 0.721024i  0.049018 + 0.269452i  0.730029 + 0.796517i;
>>     0.112849 + 0.603871i  0.486352 + 0.142337i  0.355646 + 0.151496i ];
>>
>>  uc = [0.20351 + 0.05401i;
>>     0.13141 + 0.43708i;
>>     0.29808 + 0.08789i;
>>     0.69821 + 0.38844i;
>>     0.74871 + 0.25821i ];
>>
>>  vc = [0.85839 + 0.29468i;
>>     0.20820 + 0.93090i;
>>     0.86184 + 0.34689i ];
>>
>>
>>  [Q,R] = qr(Ac);
>>  [Q,R] = qrupdate(Q,R,uc,vc);
>
> and the result was
>
>> Program received signal EXC_BAD_ACCESS, Could not access memory.
>> Reason: 13 at address: 0x00000000
>> 0x8fe18c02 in __dyld_misaligned_stack_error ()
>> (gdb)
>
> a similar test with real data gives no problem.
>
> the test above was carried out with an Octave build linked to qrupdate 1.1.0
> I nothing has changed in the library code in 1.1.1 I don't expect things to
> be different with th new version.
>
> For the time being I will therefore build with --without-qrupdate to
> proceed,
> but if there is any more useful tests I can run to try debug this, please
> let me know.
>
> Carlo
>

Do you have valgrind available? If yes, please try:

1. compile qrupdate with debug info (add -g to FFLAGS in Makeconf)
2. link Octave to it
3. run the binary in-place with valgrind
./run-octave -valgrind
4. run the test (preferably saved in an extra script)
5. report results


-- 
RNDr. Jaroslav Hajek, PhD
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz