[QE-developers] pp.x segfaults when writing ELF, depending on cutoffs

[Paolo Giannozzi]

Apparently an extension to ELF included in v.6.8 broke case nspin=1:
   https://gitlab.com/QEF/q-e/-/merge_requests/1672
Note that ELF may or may not work, beyond the specific case, because there
are no tests for it.

Paolo

On Thu, Dec 9, 2021 at 10:43 AM MARTINEZ-CANALES Miguel <
Miguel.Martinez at ed.ac.uk> wrote:

> Dear QE developers,
>
> I’ve found a reproducible segmentation fault when running pp.x to compute
> ELF. When running the attached input with QE 6.8 and 7.0 rc1, pp.x crashes
> after printing
> "elf + US not fully implemented”
> And right before "Writing data to file  sielf” with the following message:
>
> […]
>      elf + US not fully implemented
>
> Program received signal SIGSEGV: Segmentation fault - invalid memory
> reference.
>
> Backtrace for this error:
> #0  0x7f80a434bd21 in ???
> #1  0x7f80a434aef5 in ???
> #2  0x7f80a3fc420f in ???
> #3  0x7f80a3ecf62a in ???
> #4  0x7f80a463ee62 in ???
> #5  0x559bc087b719 in __scatter_mod_MOD_gather_real_grid
>         at /scratch/qe-6.8/FFTXlib/scatter_mod.f90:86
> #6  0x559bc05163eb in punch_plot_
>         at /scratch/qe-6.8/PP/src/punch_plot.f90:305
> #7  0x559bc0505b60 in __pp_module_MOD_extract
>         at /scratch/qe-6.8/PP/src/postproc.f90:227
> #8  0x559bc0504f93 in pp
>         at /scratch/qe-6.8/PP/src/postproc.f90:280
> #9  0x559bc0504f93 in main
>         at /scratch/qe-6.8/PP/src/postproc.f90:250
> Segmentation fault (core dumped)
>
>
> This happens, on either serial or parallel, on the following systems:
> macOS 11.6.1 on Intel, macports GCC 10.3.0, OpenMPI 4.1.1, OpenBLAS 0.3.18
> & FFTW 3.3.10
> Ubuntu 20.04.3 on Intel, Ubuntu’s GCC 9.3.0, OpenMPI 4.0.3, OpenBLAS 0.3.8
> & FFTW 3.3.8
>
> The input works fine with QE 6.6 & 6.7MaX. For some reason, some values of
> ecutwfc/ecutrho work, and the ELF file is written (correctly?).
>
> Works on the Mac:
> ecutwfc=18
> ecutwfc=30, ecutrho=240
>
> Works on Ubuntu:
> ecutwfc=30, ecutrho=300
>
> I ignore what’s going on.
>
> ########  si.scf.in  ########
>  &control
>     calculation='scf'
>     restart_mode='from_scratch',
>     prefix='si'
>     pseudo_dir = './'
>     outdir='./tmp/'
>  /
>  &system
>     ibrav= 2, celldm(1)= 10.2, nat= 2, ntyp= 1,
>     ecutwfc = 30.0, ecutrho = 240.0
>  /
>  &electrons
>     conv_thr =  1.0d-8
>     mixing_beta = 0.7
>  /
> ATOMIC_SPECIES
>  Si  28.086   Si.pz-vbc.UPF
> ATOMIC_POSITIONS alat
>  Si 0.00 0.00 0.00
>  Si 0.25 0.25 0.25
> K_POINTS automatic
>  8 8 8   1 1 1
>
>
> ########  si.pp_elf.in  ########
>  &inputpp
>     prefix  = 'si'
>     outdir = './tmp/'
>     filplot = 'sielf'
>     plot_num= 8
>  /
>  &plot
>     nfile = 1
>     filepp(1) = 'sielf'
>     weight(1) = 1.0
>     iflag = 3
>     output_format = 5
>     fileout = 'si.elf.xsf'
>  /
>
> Regards,
>
> Miguel
>
> ----------------------------------------
> Dr. Miguel Martinez Canales
>   School of Physics & Astronomy
>   The University of Edinburgh
>   James Clerk Maxwell Building
>   Peter Guthrie Tait Road
>   Edinburgh EH9 3FD (UK)
> Tlf:  +44 131 650 5680
> Pronouns: He/Him
> ----------------------------------------
>
> The University of Edinburgh is a charitable body, registered in Scotland,
> with registration number SC005336. Is e buidheann carthannais a th’ ann an
> Oilthigh Dhùn Èideann, clàraichte an Alba, àireamh clàraidh SC005336.
> _______________________________________________
> developers mailing list
> developers at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/developers
>


-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 206, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20211209/501262ce/attachment.html>