Re: [RFC net-next] net ipv6: convert fib6_table rwlock to a percpu lock

2017-08-01 Thread Stephen Hemminger
On Mon, 31 Jul 2017 19:57:04 -0700 Shaohua Li wrote: > On Mon, Jul 31, 2017 at 04:10:07PM -0700, Stephen Hemminger wrote: > > On Mon, 31 Jul 2017 10:18:57 -0700 > > Shaohua Li wrote: > > > > > From: Shaohua Li > > > > > > In a syn flooding test, the fib6_table rwlock is a significant > > >

Re: [RFC net-next] net ipv6: convert fib6_table rwlock to a percpu lock

2017-08-01 Thread Eric Dumazet
On Mon, 2017-07-31 at 19:57 -0700, Shaohua Li wrote: > On Mon, Jul 31, 2017 at 04:10:07PM -0700, Stephen Hemminger wrote: > > On Mon, 31 Jul 2017 10:18:57 -0700 > > Shaohua Li wrote: > > > > > From: Shaohua Li > > > > > > In a syn flooding test, the fib6_table rwlock is a significant > > > bott

Re: [RFC net-next] net ipv6: convert fib6_table rwlock to a percpu lock

2017-07-31 Thread Shaohua Li
On Mon, Jul 31, 2017 at 04:10:07PM -0700, Stephen Hemminger wrote: > On Mon, 31 Jul 2017 10:18:57 -0700 > Shaohua Li wrote: > > > From: Shaohua Li > > > > In a syn flooding test, the fib6_table rwlock is a significant > > bottleneck. While converting the rwlock to rcu sounds straighforward, > >

Re: [RFC net-next] net ipv6: convert fib6_table rwlock to a percpu lock

2017-07-31 Thread Stephen Hemminger
On Mon, 31 Jul 2017 10:18:57 -0700 Shaohua Li wrote: > From: Shaohua Li > > In a syn flooding test, the fib6_table rwlock is a significant > bottleneck. While converting the rwlock to rcu sounds straighforward, > but is very challenging if it's possible. A percpu spinlock is quite > trival for

Re: [RFC net-next] net ipv6: convert fib6_table rwlock to a percpu lock

2017-07-31 Thread Shaohua Li
On Mon, Jul 31, 2017 at 11:30:53AM -0700, Eric Dumazet wrote: > On Mon, 2017-07-31 at 10:18 -0700, Shaohua Li wrote: > > From: Shaohua Li > > > > In a syn flooding test, the fib6_table rwlock is a significant > > bottleneck. While converting the rwlock to rcu sounds straighforward, > > but is ver

Re: [RFC net-next] net ipv6: convert fib6_table rwlock to a percpu lock

2017-07-31 Thread Eric Dumazet
On Mon, 2017-07-31 at 10:18 -0700, Shaohua Li wrote: > From: Shaohua Li > > In a syn flooding test, the fib6_table rwlock is a significant > bottleneck. While converting the rwlock to rcu sounds straighforward, > but is very challenging if it's possible. A percpu spinlock is quite > trival for th

[RFC net-next] net ipv6: convert fib6_table rwlock to a percpu lock

2017-07-31 Thread Shaohua Li
From: Shaohua Li In a syn flooding test, the fib6_table rwlock is a significant bottleneck. While converting the rwlock to rcu sounds straighforward, but is very challenging if it's possible. A percpu spinlock is quite trival for this problem since updating the routing table is a rare event. In m