A student and I have independently been trying to reconstruct a dataset. We've tried it a couple different ways using either an additional FLAIR or a T2 image and always hit a point where the process halts because of a missing surface file:

This has happened when running just the autorecon1 step or both autorecon1 and autorecon2. I'm using freesurfer 7.2.0 on a linux computer, and my student is running 7.3.2 on a Mac. In my most recent test, I ran

recon-all -autorecon1 -autorecon2 -i <path to T1 file> -T2 <path to T2 file> -subjid <ID>

and I have several surface files (e.g., ?h.orig and even ?h.white.preaparc) but not the ?h.white surface. I'm unclear what's causing the recon-all script to look for the wm surface before it's been created.