octave-maintainers
[Top][All Lists]
Advanced

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

Re: compiling development sources


From: Michael D Godfrey
Subject: Re: compiling development sources
Date: Wed, 10 Feb 2010 12:35:17 -0800
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.7) Gecko/20100120 Fedora/3.0.1-1.fc12 Thunderbird/3.0.1

On 02/10/2010 11:33 AM, Carlo de Falco wrote:

On 10 Feb 2010, at 18:46, John W. Eaton wrote:

OK, then try running Octave with

 ./run-octave -g

then at the (gdb) prompt, type "run", then at the Octave prompt, cd to
the src/DLD-FUNCTIONS directory and type "test lu".  That should
result in the same segfault and you should be able to get information
about where the crash happens and examine the variables in scope at
the time of the crash.

jwe

running the tests manually I found where the crash is:

>> [l, u, p] = lu ([1, 2; 3, 4]);
Reading symbols for shared libraries . done
>> p
p =

Permutation Matrix

   0   1
   1   0

>> assert(p(:,:), [0, 1; 1, 0], sqrt (eps));

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0x9b12d1e0
get_dispatch_type (address@hidden, address@hidden) at symtab.cc:528
528                btyp = sup_table[btyp][args(i).builtin_type ()];
(gdb)


It seems that the problem is in comparing the permutation matrix p with the full matrix [0, 1; 1, 0]

Indeed, changing the test to:

>> [l, u, p] = lu ([1, 2; 3, 4]);
>> assert(full(p), [0, 1; 1, 0], sqrt (eps));
>>


I don't get the crash anymore...
c.


My experience with compiling Octave on Mac 10.6.2 is that optimization -O3
cannot be used, at least with any of the compilers that I could find.  I set -O0
and the result runs and passes the tests.

Michael


reply via email to

[Prev in Thread] Current Thread [Next in Thread]