QCxMS.v5.0.2.tar.xz release SIGSEGV, segmentation fault occurred

[tobigithub]

Hi, I am running one of the first public releases (QCxMS.v5.0.2.tar.xz).

(1) I tested QCxMS.v5.0.2 on a newly setup VM Ubuntu 18, just fresh install. (2) I additionally tested QCxMS.v5.0.2 on a production machine with XTB, QCEIMS, orca, TurboMole, MNDO99 etc, basically a machine that has had no prior issues, intel compilers and intel MKL installed.

The two initial qcxms runs run fine, I get the trjM and xtb.last files as well as the individual TMP directories. But on both machines i get the following error when executing pqcxms when running the THF CID example. While the qcxms runs show CPU use, the pqcxms just shows 2% CPU use across all cores and doesn't even finish. Thanks. Tobias

~/QCxMS.v5.0/EXAMPLE/CID/Tetrahydrofuran/TMPQCXMS/TMP.1$ cat qcxms.out
Sun Jun  6 19:40:19 PDT 2021

                      *********************************************
                      *                                           *
                      *            Q   C   x   M   S              *
                      *                                           *
                      *                  V5.0.2                   *
                      *                                           *
                      *                S. Grimme                  *
                      * Mulliken Center for Theoretical Chemistry *
                      *             Universitaet Bonn             *
                      *                  2008-21                  *
                      *        Jun 1 15:00:00 CEST 2021           *
                      *                                           *
                      *********************************************

Please cite as:
S.Grimme, Angew.Chem.Int.Ed. 52 (2013) 6306-6312.

Non-commercial. Free for academic use only.

Former Dev. : V.Asgeirsson, C.Bauer
Current Dev.: J. Koopman
 xtbhome directory:/home/user/QCxMS.v5.0/.XTBPARAM/
 changed by input:
>xtb2
>cid
>elab 40
>maxcoll 6
>noesi

 QC method: XTB2
 "Fock"-exchange ax = 0.000

              ************************************************************
                       Mode: Collision Induced Dissociation (CID)
                                   Positive Ion mode
              ************************************************************

 Internal program parameters (<0 means automatic):
 method             PosCID
 prog               XTB2
 Dispersion          D4
 prog for IPs       XTB2
 MO spectrum        XTB
 ntraj              350
 tstep (fs)        0.50
 tmax  (ps)        5.00
 eimpact (eV)    40.00
 CollAuto       F
 eTemp (K)        -1.00
 base eTemp (K) 5000.00
 Tinit (K)       500.00
 nfragexit            3
 # ion tracks         7
 iseed               42
----------------------------------------
 qc path       /usr/local/bin/
----------------------------------------

 *Program selected for CID is OK*

--------------------------------------------------------------------------------
 !!! M+ collisions are user set !!!
================================================================================
      trajectory     1      collision   1 / 6
================================================================================

 initial Cartesian coordinates:
  1  13.03706  13.12782   7.35086    O     15.999
  2  12.67154  10.52935   7.59640    C     12.011
  3  15.07256   9.30773   8.56976    C     12.011
  4  16.90149  11.49722   9.21147    C     12.011
  5  15.75398  13.54767   7.27951    C     12.011
  6  11.30941  10.22276   9.18509    H      1.008
  7  11.85540   9.75627   5.77384    H      1.008
  8  15.72251   8.29798   6.89231    H      1.008
  9  14.84536   7.92434   9.99760    H      1.008
 10  18.72586  10.54356   9.00469    H      1.008
 11  16.80263  12.26704  11.13826    H      1.008
 12  16.29815  13.31201   5.26867    H      1.008
 13  16.13578  15.44135   8.06894    H      1.008
 14  12.86643  14.14376   9.09913    H      1.008

 statistical charge  =     1.0000

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source
qcxms              00000000014D9FE4  Unknown               Unknown  Unknown
qcxms              00000000016A41B0  Unknown               Unknown  Unknown
qcxms              00000000004EAFE0  areavolume_               306  arvo.f
qcxms              00000000004EA6E3  arvo_                      95  arvo.f
qcxms              00000000004915BE  mvoldrv1_                  46  mvol.f90
qcxms              0000000000426BA7  cid_                      218  cid.f90
qcxms              00000000004083AE  MAIN__                   1793  main.f90
qcxms              0000000000400DEE  Unknown               Unknown  Unknown
qcxms              00000000016A6049  Unknown               Unknown  Unknown
qcxms              0000000000400CCA  Unknown               Unknown  Unknown

[zyt0y]

I had this problem during my first try. ulimit -s unlimited solves my problem.

[tobigithub]

@zyt0y Thanks, that worked indeed. I wonder if we also have to set the OMP_STACKSIZE correctly.

[JayTheDog]

This is a good point and might be the issue because of the xTB version implemented. The implementation of xTB will be redesigned in the near future, until then the solution provided by @zyt0y is working fine.

However, I have redesigned the way the program calculates the molecular radii because the arvo.f file was a quick fix for the issue. In the 5.0.3 release (https://github.com/qcxms/QCxMS/releases/tag/v.5.0.3), the radii are calculated by an other approach and the file is obsolete. For this reason, this issue can be regarded as fixed.