This sounds great, but to my naive, unschooled mind doesn’t this scheme fragment the maximum memory you can request in a single vmalloc request? What am I missi

This sounds great, but to my naive, unschooled mind doesn’t this scheme fragment the maximum memory you can request in a single vmalloc request? What am I missing? Are the per-core pools dynamic in size?