- The segfault occurs at a possition unrelated to the error.
- It occurs both with and without mpi, just at different positions.
- @alek's commit e2a9c5ec seems not to be at fault.
- With the current compiler the error also occurs for the commits back to 1st of February.
- I also get it on Jureca (at a different position), but both my tower and jureca use ifort version 184.108.40.206
My best guess is an error in ifort. It seems that they ruined ifort in version 19.
(generated from) inpSi.txt
P.S: since this file runs fine inp.xml sym.out inpSi.txt it could also be an error in libxml2. There the calculations are identical aside from the xml-files and libxml is written in C, which is prone to memory corruption. @wortmann @micha did you change anything about libxml?