# Segfault when plotting FE function

**URL:** <https://community.freefem.org/t/segfault-when-plotting-fe-function/2222>\
**Category:** General Discussion\
**Created:** [January 16, 2023, 12:37pm UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222 "2023-01-16T12:37:03Z")\
**Posts on this page:** 8\
**Page:** 1

<div class="post-metadata">

**Author:** ![Eloi](https://avatars.discourse-cdn.com/v4/letter/e/3ec8ea/32.png) [@Eloi](https://community.freefem.org/u/Eloi)\
**Post date:** [January 16, 2023, 12:37pm UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222/1 "2023-01-16T12:37:03Z")

</div>

Hello,  
I used FreeFem for a while and today when I tried to run a simple script which involved plotting a function, it segfaulted. Here is the code :

```auto
int npoints = 100; // Number of points on the circonference of the disk
real radius = sqrt(1/pi); // The radius of the disk of area 1

border circle(t=0,2*pi){
  x = radius*cos(t);
  y = radius*sin(t);
}

mesh Th = buildmesh(circle(npoints));

fespace Vh(Th,P1);
Vh f = x;

plot(f, wait=true);

```

and the error i get :

```auto
 -- FreeFem++ v4.6 (Thu Apr 2 15:47:38 CEST 2020 - git v4.6)
 Load: lg_fem lg_mesh lg_mesh3 eigenvalue 
    1 : int eigCount = 10; // the number of eigenvalues to compute
    2 : int npoints = 100; // Number of points on the circonference of the disk
    3 : real radius = sqrt(1/pi); // The radius of the disk of area 1
    4 : 
    5 : border circle(t=0,2*pi){
    6 : x = radius*cos(t);
    7 : y = radius*sin(t);
    8 : }
    9 : 
   10 : mesh Th = buildmesh(circle(npoints));
   11 : 
   12 : fespace Vh(Th,P1);
   13 : Vh uh,vh;
   14 : Vh f = x;
   15 : 
   16 : plot(f, wait=true);
   17 : sizestack + 1024 =1504 ( 480 )

  -- mesh: Nb of Triangles = 1766, Nb of Vertices 934
times: compile 0.00399s, execution 0.009392s, mpirank:0
Segmentation fault (core dumped)
 CodeAlloc : nb ptr 3473, size :469344 mpirank: 0
Ok: Normal End

```

I don’t know what would be the cause of it, I hope you will be able to help me. I already tried to uninstall and reinstall freefem with no success.

Best,  
Eloi.

---

<div class="post-metadata">

**Author:** ![cmd](https://yyz2.discourse-cdn.com/flex030/user_avatar/community.freefem.org/cmd/32/67_2.png) [@cmd](https://community.freefem.org/u/cmd)\
**Post date:** [January 16, 2023, 4:00pm UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222/2 "2023-01-16T16:00:33Z")

</div>

I see no errors in the code. Must be a problem with your setup. Did you check that all of the necessary packages are available on your system before installing?

---

<div class="post-metadata">

**Author:** ![frederichecht](https://yyz2.discourse-cdn.com/flex030/user_avatar/community.freefem.org/frederichecht/32/15_2.png) [@frederichecht](https://community.freefem.org/u/frederichecht)\
**Post date:** [January 16, 2023, 4:43pm UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222/3 "2023-01-16T16:43:23Z")

</div>

You use a very very old version of FreeFem++

first see if due to the graphic part (ffglut)

launch without graphique (with flag -nw)

1. reinstall freefem++ (lot of bug are corrected).

---

<div class="post-metadata">

**Author:** ![Eloi](https://avatars.discourse-cdn.com/v4/letter/e/3ec8ea/32.png) [@Eloi](https://community.freefem.org/u/Eloi)\
**Post date:** [January 17, 2023, 12:37pm UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222/4 "2023-01-17T12:37:48Z")

</div>

Hello,  
I have réinstalled FreeFem with version 4.12 and now the previous code works. However, the thing that i want to do still segfault and I don’t know with (it doesn’t happens in nw mode) :

```auto
int eigCount = 10; // the number of eigenvalues to compute
int npoints = 100; // Number of points on the circonference of the disk
real radius = sqrt(1/pi); // The radius of the disk of area 1

border circle(t=0,2*pi){
  x = radius*cos(t);
  y = radius*sin(t);
}

mesh Th = buildmesh(circle(npoints));

fespace Vh(Th,P1);
Vh uh,vh;

varf va(uh, vh) = int2d(Th)( dx(uh)*dx(vh)+dy(uh)*dy(vh));
varf vb(uh, vh) = int2d(Th)(uh*vh);

matrix A = va(Vh, Vh ,solver = sparsesolver); // Matrix A on left side
matrix B = vb(Vh, Vh); // Matrix B on right side

// Get first Eigenvalues
real[int] ev(eigCount); // Holds Eigenvalues
Vh[int] eV(eigCount); // Holds Eigenfunctions

int numEigs = EigenValue(A,B,sym=true,sigma=0,value=ev,vector=eV);

plot(eV[1], wait=true);

```

and the trace in the command-line :

```auto
-- FreeFem++ v4.12 (Tue Dec 6 19:14:11 CET 2022 - git v4.12)
   file : neumann_disk.edp
 Load: lg_fem lg_mesh lg_mesh3 eigenvalue 
    1 : int eigCount = 10; // the number of eigenvalues to compute
    2 : int npoints = 100; // Number of points on the circonference of the disk
    3 : real radius = sqrt(1/pi); // The radius of the disk of area 1
    4 : 
    5 : border circle(t=0,2*pi){
    6 : x = radius*cos(t);
    7 : y = radius*sin(t);
    8 : }
    9 : 
   10 : mesh Th = buildmesh(circle(npoints));
   11 : 
   12 : fespace Vh(Th,P1);
   13 : Vh uh,vh;
   14 : 
   15 : varf va(uh, vh) = int2d(Th)( dx(uh)*dx(vh)+dy(uh)*dy(vh));
   16 : varf vb(uh, vh) = int2d(Th)(uh*vh);
   17 : 
   18 : matrix A = va(Vh, Vh ,solver = sparsesolver); // Matrix A on left side
   19 : matrix B = vb(Vh, Vh); // Matrix B on right side
   20 : 
   21 : // Get first Eigenvalues
   22 : real[int] ev(eigCount); // Holds Eigenvalues
   23 : Vh[int] eV(eigCount); // Holds Eigenfunctions
   24 : 
   25 : int numEigs = EigenValue(A,B,sym=true,sigma=0,value=ev,vector=eV);
   26 : 
   27 : plot(eV[1], wait=true);
   28 : sizestack + 1024 =1688 ( 664 )

  -- mesh: Nb of Triangles = 1766, Nb of Vertices 934
Real symmetric eigenvalue problem: A*x - B*x*lambda
times: compile 0.016681s, execution 0.038446s, mpirank:0
Segmentation fault (core dumped)
 CodeAlloc : nb ptr 3791, size :506664 mpirank: 0
Ok: Normal End

```

I really don’t understand what is goind on.

Thank you for your help !  
Best,  
Eloi.

P.S.: in the installation guide [Installation guide](https://doc.freefem.org/introduction/installation.html), the page given for the binary packages is [Download](https://doc.freefem.org/introduction/download.html#download). However the latest versions are said to be 4.6. Personally I used the github repo FreeFem/FreeFem-sources/releases but I don’t know if it is the right thing to do.

---

<div class="post-metadata">

**Author:** ![Eloi](https://avatars.discourse-cdn.com/v4/letter/e/3ec8ea/32.png) [@Eloi](https://community.freefem.org/u/Eloi)\
**Post date:** [January 20, 2023, 1:19pm UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222/5 "2023-01-20T13:19:34Z")

</div>

Hello,  
I allow myself to up this thread.  
I guess the issue comes from ffglut, is it possible to reinstall this only ?  
Have a good day.  
Best,  
Eloi.

---

<div class="post-metadata">

**Author:** ![marchywka](https://avatars.discourse-cdn.com/v4/letter/m/ee59a6/32.png) [@marchywka](https://community.freefem.org/u/marchywka)\
**Post date:** [January 20, 2023, 4:06pm UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222/6 "2023-01-20T16:06:15Z")

</div>

I ran your second set of code and it ran ok AFAICT  
You can try running it under gdb. Just type "gdb --args FreeFem++ file.edp  
type run and when it segfailts type “bt” and copy that to this forum.

int eigCount = 10; // the number of eigenvalues to compute  
int npoints = 100; // Number of points on the circonference of the disk  
real radius = sqrt(1/pi); // The radius of the disk of area 1

border circle(t=0,2_pi){  
x = radius_cos(t);  
y = radius\*sin(t);  
}

mesh Th = buildmesh(circle(npoints));

fespace Vh(Th,P1);  
Vh uh,vh;

varf va(uh, vh) = int2d(Th)( dx(uh)\*dx(vh)+dy(uh)_dy(vh));  
varf vb(uh, vh) = int2d(Th)(uh_vh);

matrix A = va(Vh, Vh ,solver = sparsesolver); // Matrix A on left side  
matrix B = vb(Vh, Vh); // Matrix B on right side

// Get first Eigenvalues  
real[int] ev(eigCount); // Holds Eigenvalues  
Vh[int] eV(eigCount); // Holds Eigenfunctions

int numEigs = EigenValue(A,B,sym=true,sigma=0,value=ev,vector=eV);

plot(eV[1], wait=true);

---

<div class="post-metadata">

**Author:** ![Eloi](https://avatars.discourse-cdn.com/v4/letter/e/3ec8ea/32.png) [@Eloi](https://community.freefem.org/u/Eloi)\
**Post date:** [January 27, 2023, 10:58am UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222/7 "2023-01-27T10:58:27Z")

</div>

Hello,  
Sorry for the late reply. Here is the trace :

```auto
(gdb) run
Starting program: /usr/local/bin/FreeFem++ neumann_disk.edp
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
[Detaching after vfork from child process 125864]
-- FreeFem++ v4.12 (Tue Dec 6 19:14:11 CET 2022 - git v4.12)
   file : neumann_disk.edp
 Load: lg_fem lg_mesh lg_mesh3 eigenvalue 
    1 : load "medit"
    2 : int eigCount = 10; // the number of eigenvalues to compute
    3 : int npoints = 400; // Number of points on the circonference of the disk
    4 : real radius = sqrt(1/pi); // The radius of the disk of area 1
    5 : 
    6 : border circle(t=0,2*pi){
    7 : x = radius*cos(t);
    8 : y = radius*sin(t);
    9 : }
   10 : 
   11 : /* Save scalar function u as a .sol file */
   12 : func int printsol(string sout, real[int] & u) {
   13 : 
   14 : int n = u.n;
   15 : ofstream fout(sout);
   16 : 
   17 : /* Header */
   18 : fout << "MeshVersionFormatted 1\n\nDimension 2\n\nSolAtVer
  ... : tices\n"<<n<<"\n1 1\n"<<endl;
   19 : 
   20 : /* Print values */
   21 : for (int k=0; k<n; k++)
   22 : fout<<u[k]<<endl;
   23 : 
   24 : /* End keyword */
   25 : fout<<"End"<<endl;
   26 : 
   27 : return(1);
   28 : }
   29 : 
   30 : mesh Th = buildmesh(circle(npoints));
   31 : 
   32 : fespace Vh(Th,P1);
   33 : Vh uh,vh;
   34 : 
   35 : varf va(uh, vh) = int2d(Th)( dx(uh)*dx(vh)+dy(uh)*dy(vh));
   36 : varf vb(uh, vh) = int2d(Th)(uh*vh);
   37 : 
   38 : matrix A = va(Vh, Vh ,solver = sparsesolver); // Matrix A on left side
   39 : matrix B = vb(Vh, Vh); // Matrix B on right side
   40 : 
   41 : // Get first Eigenvalues
   42 : real[int] ev(eigCount); // Holds Eigenvalues
   43 : Vh[int] eV(eigCount); // Holds Eigenfunctions
   44 : 
   45 : int numEigs = EigenValue(A,B,sym=true,sigma=0,value=ev,vector=eV);
   46 : 
   47 : plot(eV[1], wait=1);
   48 : sizestack + 1024 =1984 ( 960 )

  -- mesh: Nb of Triangles = 27650, Nb of Vertices 14026
Real symmetric eigenvalue problem: A*x - B*x*lambda
times: compile 0.003337s, execution 0.511229s, mpirank:0
Segmentation fault (core dumped)
 CodeAlloc : nb ptr 3853, size :509688 mpirank: 0
Ok: Normal End
[Inferior 1 (process 125860) exited normally]
(gdb) bt
No stack.

```

---

<div class="post-metadata">

**Author:** ![marchywka](https://avatars.discourse-cdn.com/v4/letter/m/ee59a6/32.png) [@marchywka](https://community.freefem.org/u/marchywka)\
**Post date:** [January 27, 2023, 11:49am UTC](https://community.freefem.org/t/segfault-when-plotting-fe-function/2222/8 "2023-01-27T11:49:15Z")

</div>

Usually gdb catches the segfault before the app but I guess if its launching a new  
process for ffglut that is not attached to gdb.

There are a couple of ideas here, I would imagine FF invokes the fflglut in  
the FF bin dir but maybe you could replace that with a script to help attach  
or see if the follow process thing works,

[https://stackoverflow.com/questions/15126925/debugging-child-process-after-fork-follow-fork-mode-child-configured](https://stackoverflow.com/questions/15126925/debugging-child-process-after-fork-follow-fork-mode-child-configured)