bugGNU Octave - Bugs: bug #55956, build overlinks with SUNDIALS...

 
 

bug #55956: build overlinks with SUNDIALS libraries even when HAVE_SUNDIALS is unset

Submitter:  Mike Miller <mtmiller>
Submitted:  Tue 19 Mar 2019 06:34:49 PM UTC
   
 
Category:  Configuration and Build System Severity:  3 - Normal
Priority:  5 - Normal Item Group:  Other
Status:  Fixed Assigned to:  mtmiller
Originator Name:  Open/Closed:  * Closed
Release:  * 5.1.0 Operating System:  * GNU/Linux
Fixed Release:  None Planned Release:  None
* Mandatory Fields

Add a New Comment Rich Markup
   

Jump to the original submission

Thu 28 Mar 2019 06:29:55 PM UTC, comment #12: 

Michael, that is the intended behavior until Fedora updates to sundials version 4.

I personally choose to build without that option and just let the optional sparse Jacobian feature be disabled because it's easier.

 Mike Miller <mtmiller>
Group Member
Thu 28 Mar 2019 12:19:48 PM UTC, comment #11: 

As of latest patch I still need:

 CPPFLAGS=-I/usr/include/suitesparse

Fedora 29:
Linux pbdsl4 5.0.3-200.fc29.x86_64 #1 SMP Tue Mar 19 15:07:58 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux

 Michael Godfrey <godfrey>
Group Member
Wed 27 Mar 2019 07:04:49 PM UTC, comment #10: 

Seems good to me, thanks.

 Mike Miller <mtmiller>
Group Member
Mon 25 Mar 2019 08:43:01 PM UTC, comment #9: 

Okay to close now?

 Rik <rik5>
Group administrator
Wed 20 Mar 2019 11:45:44 AM UTC, comment #8: 

The CPPFLAGS is what fixed. Thanks!

On 3/20/19 10:27 AM, Carlo de Falco wrote:

> @godfrey, did you specify something like
>
>
> CPPFLAGS=-I/usr/include/suitesparse
>
>
> when running Octave's configure?
> This is needed when using sundials version 3

 Michael Godfrey <godfrey>
Group Member
Wed 20 Mar 2019 10:35:51 AM UTC, comment #7: 

@godfrey and of course you need to have suitesparse-devel installed

 Carlo de Falco <cdf>
Group Member
Wed 20 Mar 2019 10:27:37 AM UTC, comment #6: 

@godfrey, did you specify something like


CPPFLAGS=-I/usr/include/suitesparse


when running Octave's configure?
This is needed when using sundials version 3

 Carlo de Falco <cdf>
Group Member
Wed 20 Mar 2019 10:02:10 AM UTC, comment #5: 

And, my sundials packages are:
[godfrey@pbdsl4 octave]$ rpm -qa|grep -i sund
656:sundials-3.1.1-3.fc29.x86_64
1207:sundials-mpich-3.1.1-3.fc29.x86_64
2601:sundials-mpich-devel-3.1.1-3.fc29.x86_64
2795:sundials-doc-3.1.1-3.fc29.noarch
2827:sundials-openmpi-3.1.1-3.fc29.x86_64
2849:sundials-devel-3.1.1-3.fc29.x86_64
3148:sundials-openmpi-devel-3.1.1-3.fc29.x86_64
=======================
Do I need something else?

 Michael Godfrey <godfrey>
Group Member
Wed 20 Mar 2019 09:34:56 AM UTC, comment #4: 

With current Fedora 29 I get:

checking whether SUNDIALS IDA is configured with double precision realtype... yes
checking sunlinsol/sunlinsol_dense.h usability... yes
checking sunlinsol/sunlinsol_dense.h presence... yes
checking for sunlinsol/sunlinsol_dense.h... yes
checking whether SUNDIALS IDA includes the SUNLINSOL_DENSE linear solver... yes
checking klu.h usability... no
checking klu.h presence... no
checking for klu.h... no
checking klu/klu.h usability... no
checking klu/klu.h presence... no
checking for klu/klu.h... no
checking for suitesparse/klu.h... (cached) yes
checking ufsparse/klu.h usability... no
checking ufsparse/klu.h presence... no
checking for ufsparse/klu.h... no
checking for sunlinsol/sunlinsol_klu.h... no
checking for SUNKLU in -lsundials_sunlinsolklu... yes
checking whether compiling a program that calls SUNKLU works... yes
configure: WARNING: SUNDIALS IDA library not configured with SUNLINSOL_KLU or sunlinksol_klu.h is not usable; ode15i and ode15s will not support the sparse Jacobian feature
checking for IDASetJacFn... no
checking for IDASetLinearSolver... no
checking for SUNLinSol_Dense... no
checking for SUNLinSol_KLU... no
checking ida/ida_direct.h usability... yes
checking ida/ida_direct.h presence... yes
checking for ida/ida_direct.h... yes
checking ida_direct.h usability... no
checking ida_direct.h presence... no
checking for ida_direct.h... no
checking for IDADlsSetJacFn... yes
checking for IDADlsSetLinearSolver... yes
checking for SUNDenseLinearSolver... yes
checking for SUNKLU... no
checking for dseupd in -larpack ... yes
checking whether the arpack library works... yes

and:
configure: WARNING: SUNDIALS IDA library not configured with SUNLINSOL_KLU or sunlinksol_klu.h is not usable;

 Michael Godfrey <godfrey>
Group Member
Wed 20 Mar 2019 04:29:12 AM UTC, comment #3: 

Ok, that looks right, I hadn't gotten around to testing with -I/usr/include/suitesparse yet.

 Mike Miller <mtmiller>
Group Member
Wed 20 Mar 2019 03:23:34 AM UTC, comment #2: 

After your most recent changes I'm seeing the following results for configure when using CPPFLAGS=/usr/include/suitesparse and building with Sundiasl 3.x on a Debian system:


checking klu.h usability... yes
checking klu.h presence... yes
checking for klu.h... yes
checking klu/klu.h usability... no
checking klu/klu.h presence... no
checking for klu/klu.h... no
checking for suitesparse/klu.h... (cached) yes
checking ufsparse/klu.h usability... no
checking ufsparse/klu.h presence... no
checking for ufsparse/klu.h... no
checking for sunlinsol/sunlinsol_klu.h... yes
checking for SUNKLU in -lsundials_sunlinsolklu... yes
checking whether compiling a program that calls SUNKLU works... yes
checking for IDASetJacFn... no
checking for IDASetLinearSolver... no
checking for SUNLinSol_Dense... no
checking for SUNLinSol_KLU... no
checking ida/ida_direct.h usability... yes
checking ida/ida_direct.h presence... yes
checking for ida/ida_direct.h... yes
checking ida_direct.h usability... no
checking ida_direct.h presence... no
checking for ida_direct.h... no
checking for IDADlsSetJacFn... yes
checking for IDADlsSetLinearSolver... yes
checking for SUNDenseLinearSolver... yes
checking for SUNKLU... no


and then the following output when running make:


/home/jwe/src/octave/libinterp/dldfcn/__ode15__.cc: In member function 'void octave::IDA::set_up(const ColumnVector&)':
/home/jwe/src/octave/libinterp/dldfcn/__ode15__.cc:401:27: error: 'SUNLinSol_KLU' was not declared in this scope
         sunLinearSolver = SUNLinSol_KLU (yy, sunJacMatrix);
                           ^~~~~~~~~~~~~
/home/jwe/src/octave/libinterp/dldfcn/__ode15__.cc:401:27: note: suggested alternative: 'SUNLinSolFree_KLU'
         sunLinearSolver = SUNLinSol_KLU (yy, sunJacMatrix);
                           ^~~~~~~~~~~~~
                           SUNLinSolFree_KLU
make[2]: *** [Makefile:19254: libinterp/dldfcn/__ode15___la-__ode15__.lo] Error 1
make[2]: *** Waiting for unfinished jobs....
make[2]: Leaving directory '/net/devnull/scratch/jwe/build/octave'
make[1]: *** [Makefile:27143: all-recursive] Error 1
make[1]: Leaving directory '/net/devnull/scratch/jwe/build/octave'
make: *** [Makefile:10919: all] Error 2


Note that in the configure output, SUNKLU is found in the custom test but not later.

It is fixed for me by the following change:

http://hg.savannah.gnu.org/hgweb/octave/rev/8f819b00b28e

 John W. Eaton <jwe>
Group administrator
Tue 19 Mar 2019 06:37:58 PM UTC, comment #1: 

I pushed this fix on the stable branch

https://hg.savannah.gnu.org/hgweb/octave/rev/b4cb230ced82

 Mike Miller <mtmiller>
Group Member
Tue 19 Mar 2019 06:34:49 PM UTC, original submission:  

When building Octave 5.1.0 on a system with SUNDIALS version 3 or 4 installed, configure correctly detects that the library is not compatible and correctly does not set HAVE_SUNDIALS, so all features in '__ode15__.oct' are disabled. But the oct file is still linked to libsundials_ida and libsundials_nvecserial, even though it uses no functions from those libraries (overlinking).

I have a patch to configure that will clear all of the SUNDIALS associated build variables if and only if HAVE_SUNDIALS is not set, suitable for stable, and already tested on one system with SUNDIALS 3.1.2 and one with SUNDIALS 2.7.0. There are no regressions on the latter that I can see, and the oct file is now not linked with the libraries on the former, which is the desired outcome.

Will push to stable once I have a bug number.

 Mike Miller <mtmiller>
Group Member

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach Files:
   
   
Comment:
   

No files currently attached

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by rik5 (Posted a comment)
  • -email is unavailable- added by cdf (Posted a comment)
  • -email is unavailable- added by godfrey (Posted a comment)
  • -email is unavailable- added by jwe (Posted a comment)

    •  

    There are 0 votes so far. Votes easily highlight which items people would like to see resolved in priority, independently of the priority of the item set by tracker managers.

    Only group members can vote.

     

    Follow 4 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2019-03-27 rik5 StatusReady For Test Fixed
        Open/ClosedOpen Closed
    2019-03-19 mtmiller Carbon-CopyRemoved 80942 -
    2019-03-19 mtmiller StatusIn Progress Ready For Test

    Back to the top

    Copyright © 2024 Free Software Foundation, Inc.
    Verbatim copying and distribution of this entire article is permitted in any medium, provided this notice is preserved.
    The Levitating, Meditating, Flute-playing Gnu logo is a GNU GPL'ed image provided by the Nevrax Design Team.
    Page source code

    Powered by Savane 3.14-f13d.
    Corresponding source code