Mercurial > octave
view libinterp/dldfcn/amd.cc @ 27940:ec0690a6955c stable rc-5-1-90
Lock amd and symbfact to avoid segmentation fault with SuiteSparse (bug #57435).
* dldfcn/amd.cc (Famd), dldfcn/symbfact.cc (Fsymbfact): Use DEFMETHOD_DLD. Call
"mlock" to avoid having invalid function pointers in global SuiteSparse_config
structure when the functions are unloaded (e.g. with "clear functions").
author | Markus Mützel <markus.muetzel@gmx.de> |
---|---|
date | Sun, 22 Dec 2019 01:01:27 +0100 |
parents | c0d8ce61c1c9 |
children | b442ec6dda5c |
line wrap: on
line source
/* Copyright (C) 2008-2019 David Bateman This file is part of Octave. Octave is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. Octave is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with Octave; see the file COPYING. If not, see <https://www.gnu.org/licenses/>. */ // This is the octave interface to amd, which bore the copyright given // in the help of the functions. #if defined (HAVE_CONFIG_H) # include "config.h" #endif #include <cstdlib> #include "CSparse.h" #include "Sparse.h" #include "dMatrix.h" #include "oct-locbuf.h" #include "oct-sparse.h" #include "defun-dld.h" #include "error.h" #include "errwarn.h" #include "oct-map.h" #include "ov.h" #include "ovl.h" #include "parse.h" DEFMETHOD_DLD (amd, interp, args, nargout, doc: /* -*- texinfo -*- @deftypefn {} {@var{p} =} amd (@var{S}) @deftypefnx {} {@var{p} =} amd (@var{S}, @var{opts}) Return the approximate minimum degree permutation of a matrix. This is a permutation such that the Cholesky@tie{}factorization of @code{@var{S} (@var{p}, @var{p})} tends to be sparser than the Cholesky@tie{}factorization of @var{S} itself. @code{amd} is typically faster than @code{symamd} but serves a similar purpose. The optional parameter @var{opts} is a structure that controls the behavior of @code{amd}. The fields of the structure are @table @asis @item @var{opts}.dense Determines what @code{amd} considers to be a dense row or column of the input matrix. Rows or columns with more than @code{max (16, (dense * sqrt (@var{n})))} entries, where @var{n} is the order of the matrix @var{S}, are ignored by @code{amd} during the calculation of the permutation. The value of dense must be a positive scalar and the default value is 10.0 @item @var{opts}.aggressive If this value is a nonzero scalar, then @code{amd} performs aggressive absorption. The default is not to perform aggressive absorption. @end table The author of the code itself is Timothy A. Davis (see @url{http://faculty.cse.tamu.edu/davis/suitesparse.html}). @seealso{symamd, colamd} @end deftypefn */) { #if defined (HAVE_AMD) int nargin = args.length (); if (nargin < 1 || nargin > 2) print_usage (); octave_idx_type n_row, n_col; const octave::suitesparse_integer *ridx, *cidx; SparseMatrix sm; SparseComplexMatrix scm; if (args(0).issparse ()) { if (args(0).iscomplex ()) { scm = args(0).sparse_complex_matrix_value (); n_row = scm.rows (); n_col = scm.cols (); ridx = octave::to_suitesparse_intptr (scm.xridx ()); cidx = octave::to_suitesparse_intptr (scm.xcidx ()); } else { sm = args(0).sparse_matrix_value (); n_row = sm.rows (); n_col = sm.cols (); ridx = octave::to_suitesparse_intptr (sm.xridx ()); cidx = octave::to_suitesparse_intptr (sm.xcidx ()); } } else { if (args(0).iscomplex ()) sm = SparseMatrix (real (args(0).complex_matrix_value ())); else sm = SparseMatrix (args(0).matrix_value ()); n_row = sm.rows (); n_col = sm.cols (); ridx = octave::to_suitesparse_intptr (sm.xridx ()); cidx = octave::to_suitesparse_intptr (sm.xcidx ()); } if (n_row != n_col) err_square_matrix_required ("amd", "S"); OCTAVE_LOCAL_BUFFER (double, Control, AMD_CONTROL); AMD_NAME (_defaults) (Control); if (nargin > 1) { octave_scalar_map arg1 = args(1).xscalar_map_value ("amd: OPTS argument must be a scalar structure"); octave_value tmp; tmp = arg1.getfield ("dense"); if (tmp.is_defined ()) Control[AMD_DENSE] = tmp.double_value (); tmp = arg1.getfield ("aggressive"); if (tmp.is_defined ()) Control[AMD_AGGRESSIVE] = tmp.double_value (); } OCTAVE_LOCAL_BUFFER (octave::suitesparse_integer, P, n_col); Matrix xinfo (AMD_INFO, 1); double *Info = xinfo.fortran_vec (); // Lock the function to not loose the SuiteSparse_config structure interp.mlock (); // FIXME: how can we manage the memory allocation of amd // in a cleaner manner? SUITESPARSE_ASSIGN_FPTR (malloc_func, amd_malloc, malloc); SUITESPARSE_ASSIGN_FPTR (free_func, amd_free, free); SUITESPARSE_ASSIGN_FPTR (calloc_func, amd_calloc, calloc); SUITESPARSE_ASSIGN_FPTR (realloc_func, amd_realloc, realloc); SUITESPARSE_ASSIGN_FPTR (printf_func, amd_printf, printf); octave_idx_type result = AMD_NAME (_order) (n_col, cidx, ridx, P, Control, Info); if (result == AMD_OUT_OF_MEMORY) error ("amd: out of memory"); else if (result == AMD_INVALID) error ("amd: matrix S is corrupted"); Matrix Pout (1, n_col); for (octave_idx_type i = 0; i < n_col; i++) Pout.xelem (i) = P[i] + 1; if (nargout > 1) return ovl (Pout, xinfo); else return ovl (Pout); #else octave_unused_parameter (args); octave_unused_parameter (nargout); err_disabled_feature ("amd", "AMD"); #endif } /* %!shared A, A2, opts %! A = ones (20, 30); %! A2 = ones (30, 30); %!testif HAVE_AMD %! assert(amd (A2), [1:30]); %! opts.dense = 25; %! assert(amd (A2, opts), [1:30]); %! opts.aggressive = 1; %! assert(amd (A2, opts), [1:30]); %!testif HAVE_AMD %! assert (amd ([]), zeros (1,0)) %!error <S must be a square matrix|was unavailable or disabled> amd (A) %!error amd (A2, 2) */