From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jarek Poplawski Subject: Re: rib_trie / Fix inflate_threshold_root. Now=15 size=11 bits Date: Fri, 26 Jun 2009 18:15:00 +0200 Message-ID: <20090626161500.GB6526@ami.dom.local> References: <20090626080302.GB5204@ff.dom.local> <19012.37515.146191.198843@robur.slu.se> <20090626093727.GA6832@ff.dom.local> <19012.49700.908412.410984@robur.slu.se> <20090626125449.GA8897@ff.dom.local> <20090626132820.GB8897@ff.dom.local> <19012.53943.734747.493480@robur.slu.se> <20090626151051.GA2714@ami.dom.local> <20090626153010.GC6771@linux.vnet.ibm.com> <20090626155410.GA6526@ami.dom.local> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Robert Olsson , Robert Olsson , Eric Dumazet , =?us-ascii?B?PT9JU08tODg1OS0yP1E/UGF3ZT1CM19TdGFzemV3c2tpPz0=?= , Robert Olsson , Linux Network Development list To: "Paul E. McKenney" Return-path: Received: from mail-fx0-f213.google.com ([209.85.220.213]:38869 "EHLO mail-fx0-f213.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750784AbZFZQPd (ORCPT ); Fri, 26 Jun 2009 12:15:33 -0400 Received: by fxm9 with SMTP id 9so2201171fxm.37 for ; Fri, 26 Jun 2009 09:15:35 -0700 (PDT) Content-Disposition: inline In-Reply-To: <20090626155410.GA6526@ami.dom.local> Sender: netdev-owner@vger.kernel.org List-ID: On Fri, Jun 26, 2009 at 05:54:10PM +0200, Jarek Poplawski wrote: > On Fri, Jun 26, 2009 at 08:30:10AM -0700, Paul E. McKenney wrote: > > On Fri, Jun 26, 2009 at 05:10:52PM +0200, Jarek Poplawski wrote: > > > On Fri, Jun 26, 2009 at 03:52:55PM +0200, Robert Olsson wrote: > > > > > > > > Jarek Poplawski writes: > > > > > > > > Thanks, > > > > > > > > Should be worth testing so we synchronize_rcu instead of doing call_rcu's > > > > > > > > > > Alas take 2 (nor 1) doesn't compile, so here it is again. > > > > So the idea is to balance memory and latency, so that large changes > > (those affecting the root node) get at least one synchronize_rcu(), > > while smaller changes just use call_rcu(), correct? This means that > > the amount of memory awaiting an RCU grace period is limited, but > > the algorithm avoids per-node synchronize_rcu() overhead. > > > > If I understand the goal correctly, looks good! (Give or take my > > limited understanding of fib_trie and is usage, of course.) > > The goal is practically to replace all call_rcu() during > trie_rebalance() with synchronize_rcu() (except some freeing after > ENOMEM). I guess currently (<= 2.6.30) call_rcu() can free this > memory after trie_rebalance() has finished, that's why there were > problems with enabled preemption. So this patch tries to do/force > this a bit earlier - at least before the top/largest node is > rebalanced. On the other hand, we could probably stay with call_rcu() plus only one synchronize_rcu() before the top node's resize() if you think it's enough here? Thanks, Jarek P. > > > > > Thanx, Paul > > > > > Thanks, > > > Jarek P. > > > --- (take 3 - for testing) > > > > > > net/ipv4/fib_trie.c | 30 ++++++++++++++++++++++++------ > > > 1 files changed, 24 insertions(+), 6 deletions(-) > > > > > > diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c > > > index 012cf5a..1a4c4b7 100644 > > > --- a/net/ipv4/fib_trie.c > > > +++ b/net/ipv4/fib_trie.c > > > @@ -366,6 +366,17 @@ static void __tnode_vfree(struct work_struct *arg) > > > vfree(tn); > > > } > > > > > > +static void __tnode_free(struct tnode *tn) > > > +{ > > > + size_t size = sizeof(struct tnode) + > > > + (sizeof(struct node *) << tn->bits); > > > + > > > + if (size <= PAGE_SIZE) > > > + kfree(tn); > > > + else > > > + vfree(tn); > > > +} > > > + > > > static void __tnode_free_rcu(struct rcu_head *head) > > > { > > > struct tnode *tn = container_of(head, struct tnode, rcu); > > > @@ -402,7 +413,7 @@ static void tnode_free_flush(void) > > > while ((tn = tnode_free_head)) { > > > tnode_free_head = tn->tnode_free; > > > tn->tnode_free = NULL; > > > - tnode_free(tn); > > > + __tnode_free(tn); > > > } > > > } > > > > > > @@ -1021,18 +1032,25 @@ static void trie_rebalance(struct trie *t, struct tnode *tn) > > > (struct node *)tn, wasfull); > > > > > > tp = node_parent((struct node *) tn); > > > - tnode_free_flush(); > > > if (!tp) > > > break; > > > tn = tp; > > > } > > > > > > + if (tnode_free_head) { > > > + synchronize_rcu(); > > > + tnode_free_flush(); > > > + } > > > + > > > /* Handle last (top) tnode */ > > > - if (IS_TNODE(tn)) > > > + if (IS_TNODE(tn)) { > > > tn = (struct tnode *)resize(t, (struct tnode *)tn); > > > - > > > - rcu_assign_pointer(t->trie, (struct node *)tn); > > > - tnode_free_flush(); > > > + rcu_assign_pointer(t->trie, (struct node *)tn); > > > + synchronize_rcu(); > > > + tnode_free_flush(); > > > + } else { > > > + rcu_assign_pointer(t->trie, (struct node *)tn); > > > + } > > > > > > return; > > > } > > > -- > > > To unsubscribe from this list: send the line "unsubscribe netdev" in > > > the body of a message to majordomo@vger.kernel.org > > > More majordomo info at http://vger.kernel.org/majordomo-info.html