* Dave Hansen: > My assumption has always been that these large, potentially sparse > hardware tables *must* be mmap()'d with MAP_NORESERVE specified. That > should keep them from being problematic with respect to overcommit.
MAP_NORESERVE pages still count towards the commit limit. The flag only disables checks at allocation time, for this particular allocation. (At least this was the behavior the last time I looked into this, I believe.) Not sure if this makes a difference here. Thanks, Florian
