External Email - Use Caution        

Hi again,

I have solved the issue; the option -openmp was creating the segmentation fault. Of note, the option -openmp was not an issue when running all steps at once (recon-all -all). I tried with option -threads instead, but there was again a segmentation fault. Without these options, I was able to finish the pipeline.

Best,

Emmanuelle Renauld
De : Emmanuelle Renauld <Emmanuelle.Renauld@USherbrooke.ca>
Envoyé : 12 août 2026 12:43
À : freesurfer@nmr.mgh.harvard.edu <freesurfer@nmr.mgh.harvard.edu>
Objet : [Freesurfer] segmentation fault
 

        External Email - Use Caution        

Hi freesurfer,

I have a segmentation fault error and I would like your help, either in fixing the issue or in running recon-all in another way. 

Here is what I tried:  I decided to skullstrip my images first, and 1) use autorecon-1 with -noskullstrip, 2) place the brainmask myself. 3) run autorecon-2 and autorecon3. I am trying to do that on AllianceCanada's clusters. They have already installed Freesurfer for us, and I can do "module load freesurfer". I have access to version 8.2. I am launching tasks through sbatch, using environments with 32G memory.

This fails nearly at the end of the process, with: 
     mri_brainvol_stats --subject my_subj
     Segmentation fault (core dumped)
     Linux nc10227 5.14.0-611.49.1.el9_7.x86_64 #1 SMP PREEMPT_DYNAMIC Tue Apr 21 16:39:08 EDT 2026 x86_64 AMD EPYC 7532 32-Core Processor AuthenticAMD GNU/Linux

Note that I also get many warnings along the way, with: "libtiff.so.5: no version information available, if this could be a clue."

When testing manually this line only:
    >> mri_brainvol_stats --subject $SUBJECT_ID
It works (most of the time. I have run into the same issue once, I don't know what was different). But if I run it through recon-all, it fails again.

I thought I could use option -make all and continue from there, considering that I did run it manually, but make is not longer supported. Then I thought I could start at the next step. I saw from the log that mri_brainvol_stats happens during the step APas-to-ASeg, after it says "mri_surf2volseg done". I tried to run recon-all with option -noapas2aseg, but it still fails with the same error.

Do you know what causes this? Can you help?

Thank you,

Emmanuelle Renauld