Yes! I updated the AFNI to Feb 26 2018 (Version AFNI_18.0.22) and the problem solved .... Sorry, I should check and update AFNI regularly.

But I do have another question. I used freesurfer to generate all the surface information and then used afni_proc.py as below to do preprocessing.
afni_proc.py -subj_id $subj.surf \
-blocks tshift align volreg surf blur scale \
-copy_anat anat+orig \
-dsets run_*+orig.HEAD \
-surf_anat $suma_dir/FS_templa_SurfVol.nii \
-surf_spec $suma_dir/FS_templa_?h.spec \
-tcat_remove_first_trs 0 \
-align_opts_aea -cost lpc+ZZ -partial_coverage \
-volreg_align_to MIN_OUTLIER \
-volreg_align_e2a \
-blur_size $blur_fwhm \
-execute

Then I used @RetinoProc to get the retinotopic map.
@RetinoProc \
-sid $suma_subj_name \
-surf_vol $suma_dir/FS_templa_SurfVol.nii \
-spec_left $suma_dir/FS_templa_lh.spec \
-spec_right $suma_dir/FS_templa_rh.spec \
"-$procType" \
....

So I suppose the retinotopic epi data will be aligned to FS_templa_SurfVol.nii, right?
But my question is how can I check the alignment. I tried to check the alignment with FS_templa_SurfVol.nii and sub01.surf_SurfVol_Alnd_Exp+orig which is generated by @RetinoProc. They don't align with each other. I'm quite confused about it. Did I use the wrong files or the alignment is indeed not very well.

Thanks~