On Tue, Mar 12, 2019 at 03:34:02PM -0700, Roman Gushchin wrote: > Flush percpu stats and events data to corresponding before releasing > percpu memory. > > Although per-cpu stats are never exactly precise, dropping them on > floor regularly may lead to an accumulation of an error. So, it's > safer to flush them before releasing. > > To minimize the number of atomic updates, let's sum all stats/events > on all cpus locally, and then make a single update per entry. > > Signed-off-by: Roman Gushchin <g...@fb.com>
Acked-by: Johannes Weiner <han...@cmpxchg.org> Do you mind merging 6/6 into this one? That would make it easier to verify that the code added in this patch and the code removed in 6/6 are indeed functionally equivalent.
