Intel® Fortran Compiler
Build applications that can scale for the future with optimized code designed for Intel® Xeon® and compatible processors.
28381 Discussions

catastrophic error with ifort 19.1.2.254

ccorrado
Beginner
1,978 Views

I got a catastrophic error compiling my program expo (http://www.ba.ic.cnr.it/softwareic/expo/) with ifort 19.1.2.254 20200623 on Ubuntu 18.04.5 LTS.
To quickly reproduce the problem I created a small program in the attached file catastrophic.zip.
To generate the error:
1) unzip the file catastrophic.zip
2) cd castrophic
3) make
and during the compilation of the file spginfo.f90 you will obtain the following error:
spginfo.f90(2292): catastrophic error: **Internal compiler error: internal abort** Please report this error along with the circumstances in which it occurred in a Software Problem Report. Note: File and line given may not be explicit cause of this error.
I don't have similar problem in the previous version of compilers. The problem seems platform independent.

0 Kudos
6 Replies
Ron_Green
Moderator
1,960 Views

I'll write up a bug report.  thank you for putting together the shorter reproducer.

 

0 Kudos
Ron_Green
Moderator
1,939 Views

The issue is in the "-check all" option.  As you may know, check has a number of arguments in "all".  The specific check causing the crash is:

  -check shape

Just that 1 argument.  So you could do

  -check arg_temp_created,assume,bounds,continguous,format,output_conversion,pointers,stack,udio_iostat,uninit

or any combo therein and this would not get the error.  Just avoid -check shape

I'll get the bug ID here shortly.

Thank you for such a complete reproducer.

0 Kudos
Ron_Green
Moderator
1,932 Views

bug ID is CMPLRIL0-33252


0 Kudos
Ron_Green
Moderator
1,931 Views

as FYI, the line triggering the crash is 2292.

vet = matmul((spaceg%symop(i)%rot(:,:) - identity_mat(:,:)),tryor) - spaceg%coper(:,nc)

0 Kudos
ccorrado
Beginner
1,906 Views

Thanks for the quick response and the suggestion on how to avoid -check shape.
As you probably know, the error disappears by rearranging the line 2292, for example, in this way:

real, dimension(3,3) :: mat

mat = spaceg%symop(i)%rot(:,:) - identity_mat(:,:)
vet = matmul(mat,tryor) - spaceg%coper(:,nc)

 

0 Kudos
Devorah_H_Intel
Moderator
1,586 Views

This was fixed in 2021.4 ifort.

0 Kudos
Reply