Describe the bug
Sometimes the subroutine s_mpi_allreduce_integer_sum can hjave integer overflows on large enough cases, causing issues such as array instantiation in m_ibm for the ghost_points array. We need to do the all-reduce sum in 8-byte integers instead of 4-byte.
To Reproduce
I found this while running with about 4 Million spherical IBs, which overflows the number of ghost points
Expected behavior
The integer overflows to a negative number, creating a negative array size and therefor a crash.
Screenshots
Not relevant
Environment
Describe the bug
Sometimes the subroutine s_mpi_allreduce_integer_sum can hjave integer overflows on large enough cases, causing issues such as array instantiation in m_ibm for the ghost_points array. We need to do the all-reduce sum in 8-byte integers instead of 4-byte.
To Reproduce
I found this while running with about 4 Million spherical IBs, which overflows the number of ghost points
Expected behavior
The integer overflows to a negative number, creating a negative array size and therefor a crash.
Screenshots
Not relevant
Environment