Hi, I posted my question at http://stackoverflow.com/questions/29620730/strange-behaviour-of-fftw-mpi-plan-dft-r2c-3d <http://stackoverflow.com/questions/29620730/strange-behaviour-of-fftw-mpi-plan-dft-r2c-3d>. And was told to report my question here.
Besides information provided in that link, here is some extra hopefully useful information: 1. L, M and N can be exchanged, the program give the same error/result. 2. The last flag does not affect the problem. 3. Whether _rin_ and _cout_ point to the same area of memory does not affect the problem, too. 4. The signal is usually received by high rank process, like rank 23 when running 24 processes or rank 46 running 48 processes.
