https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125584
--- Comment #5 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> --- > --- Comment #4 from Andre Vehreschild <vehre at gcc dot gnu.org> --- > Have a look at the tests. Most of them are linear. I suspect all will fit into > a 2 MB segment. The tests are separated by the PID of the supervisor process > in > the shm-name, so no interference there, either. This is only apparent to someone knowing FORTRAN ;-) > The really interesting Question is: Why does the machine have problems with > 16GB shmem allocation? The segment size is virtual and only really allocated > when it is accessed. Or is Solaris going a different way and reQuires the > backing memory to be present? I mean on Linux the shmem-segment can be > exabytes > in size with the machine only having a few GB and the program runs fine as > long > as it does not reQuire more memory than is available in the machine. Solaris doesn't do Linux-style lazy allocation, but requires backing store. This way, processes aren't subject to the OOM killer... I suspect other OSes behave the same way.
