Messages in this thread | | | Date | Fri, 22 Aug 2008 11:29:15 -0700 | From | "Paul E. McKenney" <> | Subject | Re: [PATCH 2/2] smp_call_function: use rwlocks on queues rather than rcu |
| |
On Fri, Aug 22, 2008 at 12:14:37PM -0500, Christoph Lameter wrote: > Paul E. McKenney wrote: > > >> RCU is problematic because it lets cachelines get cold. A hot cacheline that > >> is used frequently read and written to by the same cpu is very good thing for > >> performace. > > > > So on your these large boxes, read-only cachelines are preferentially > > ejected from the cache, so that one should write to per-CPU data > > occasionally to keep it resident? Or is the issue the long RCU grace > > periods which allow the structure being freed to age out of all relevant > > caches? (My guess would be the second.) > > The issue are the RCU grace period that are generally long enough to make the > cacheline fall out of all caches.
Would it make sense to push the freed-by-RCU memory further up the hierarchy, so that such memory is not mistaken for recently freed hot-in-cache memory?
Thanx, Paul
| |