From cf86a086a18095e33e0637cb78cda1fcf5280852 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 7 May 2020 18:58:10 -0700 Subject: net/dst: use a smaller percpu_counter batch for dst entries accounting percpu_counter_add() uses a default batch size which is quite big on platforms with 256 cpus. (2*256 -> 512) This means dst_entries_get_fast() can be off by +/- 2*(nr_cpus^2) (131072 on servers with 256 cpus) Reduce the batch size to something more reasonable, and add logic to ip6_dst_gc() to call dst_entries_get_slow() before calling the _very_ expensive fib6_run_gc() function. Signed-off-by: Eric Dumazet Signed-off-by: Jakub Kicinski --- net/ipv6/route.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'net/ipv6/route.c') diff --git a/net/ipv6/route.c b/net/ipv6/route.c index 1ff142393c76..a9072dba00f4 100644 --- a/net/ipv6/route.c +++ b/net/ipv6/route.c @@ -3195,6 +3195,9 @@ static int ip6_dst_gc(struct dst_ops *ops) int entries; entries = dst_entries_get_fast(ops); + if (entries > rt_max_size) + entries = dst_entries_get_slow(ops); + if (time_after(rt_last_gc + rt_min_interval, jiffies) && entries <= rt_max_size) goto out; -- cgit v1.2.3