[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: make check segmentation fault
From: |
John W. Eaton |
Subject: |
Re: make check segmentation fault |
Date: |
Fri, 28 Dec 2007 12:16:16 -0500 |
On 27-Dec-2007, Yan Wu wrote:
|
| Thanks David, here is the result:
|
| [New Thread 2305843009463104096 (LWP 9099)]
| Program received signal SIGSEGV, Segmentation fault.
| [Switching to Thread 2305843009269374240 (LWP 9096)]
| base_lu<Matrix, double, Matrix, double>::P (this=Variable "this" is not
| available.) at base-lu.cc:87
| 87 base-lu.cc: No such file or directory.
| in base-lu.cc
| Current language: auto; currently c++
So the crash is happening here (at the line marked with >>>):
template <class lu_type, class lu_elt_type, class p_type, class p_elt_type>
p_type
base_lu <lu_type, lu_elt_type, p_type, p_elt_type> :: P (void) const
{
octave_idx_type a_nr = a_fact.rows ();
Array<octave_idx_type> pvt (a_nr);
for (octave_idx_type i = 0; i < a_nr; i++)
pvt.xelem (i) = i;
for (octave_idx_type i = 0; i < ipvt.length(); i++)
{
octave_idx_type k = ipvt.xelem (i);
if (k != i)
{
>>> octave_idx_type tmp = pvt.xelem (k);
pvt.xelem (k) = pvt.xelem (i);
pvt.xelem (i) = tmp;
}
}
In your earlier message, I think you showed that you were compiling
with --enable-64. I suspect that your blas/lapack library is not
compiled such that integers are 8 bytes wide. Do you really need
--enable-64? Do you understand that it is an experimental feature and
not fully implemented?
jwe