MIRRORS/linux-yocto
1
0
Fork 0
mirror of git://git.yoctoproject.org/linux-yocto.git synced 2025-10-22 23:13:01 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

net: rfs: add sock_rps_delete_flow() helper

Browse Source 
RFS can exhibit lower performance for workloads using short-lived
flows and a small set of 4-tuple.

This is often the case for load-testers, using a pair of hosts,
if the server has a single listener port.

Typical use case :

Server : tcp_crr -T128 -F1000 -6 -U -l30 -R 14250
Client : tcp_crr -T128 -F1000 -6 -U -l30 -c -H server | grep local_throughput

This is because RFS global hash table contains stale information,
when the same RSS key is recycled for another socket and another cpu.

Make sure to undo the changes and go back to initial state when
a flow is disconnected.

Performance of the above test is increased by 22 %,
going from 372604 transactions per second to 457773.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Octavian Purdila <tavip@google.com>
Reviewed-by: Neal Cardwell <ncardwell@google.com>
Link: https://patch.msgid.link/20250515100354.3339920-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This commit is contained in:
Eric Dumazet 2025-05-15 10:03:54 +00:00 committed by Jakub Kicinski
parent f24f7b2f3a
commit 9cd5ef0b8c
4 changed files with 31 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
include/net/rps.h
View File

@ -123,6 +123,30 @@ static inline void sock_rps_record_flow(const struct sock *sk)
#endif
}
static inline void sock_rps_delete_flow(const struct sock *sk)
{
#ifdef CONFIG_RPS
struct rps_sock_flow_table *table;
u32 hash, index;
if (!static_branch_unlikely(&rfs_needed))
return;
hash = READ_ONCE(sk->sk_rxhash);
if (!hash)
return;
rcu_read_lock();
table = rcu_dereference(net_hotdata.rps_sock_flow_table);
if (table) {
index = hash & table->mask;
if (READ_ONCE(table->ents[index]) != RPS_NO_CPU)
WRITE_ONCE(table->ents[index], RPS_NO_CPU);
}
rcu_read_unlock();
#endif
}
static inline u32 rps_input_queue_tail_incr(struct softnet_data *sd)
{
#ifdef CONFIG_RPS

6
net/ipv4/inet_hashtables.c
View File

@ -23,11 +23,12 @@
#if IS_ENABLED(CONFIG_IPV6)
#include <net/inet6_hashtables.h>
#endif
#include <net/secure_seq.h>
#include <net/hotdata.h>
#include <net/ip.h>
#include <net/tcp.h>
#include <net/rps.h>
#include <net/secure_seq.h>
#include <net/sock_reuseport.h>
#include <net/tcp.h>
u32 inet_ehashfn(const struct net *net, const __be32 laddr,
const __u16 lport, const __be32 faddr,
@ -790,6 +791,7 @@ void inet_unhash(struct sock *sk)
if (sk_unhashed(sk))
return;
sock_rps_delete_flow(sk);
if (sk->sk_state == TCP_LISTEN) {
struct inet_listen_hashbucket *ilb2;

2
net/ipv4/udp.c
View File

@ -120,6 +120,7 @@
#if IS_ENABLED(CONFIG_IPV6)
#include <net/ipv6_stubs.h>
#endif
#include <net/rps.h>
struct udp_table udp_table __read_mostly;
@ -2200,6 +2201,7 @@ void udp_lib_unhash(struct sock *sk)
struct udp_table *udptable = udp_get_table_prot(sk);
struct udp_hslot *hslot, *hslot2;
sock_rps_delete_flow(sk);
hslot = udp_hashslot(udptable, sock_net(sk),
udp_sk(sk)->udp_port_hash);
hslot2 = udp_hashslot2(udptable, udp_sk(sk)->udp_portaddr_hash);

2
net/sctp/socket.c
View File

@ -8321,7 +8321,7 @@ static int sctp_hash(struct sock *sk)
static void sctp_unhash(struct sock *sk)
{
/* STUB */
sock_rps_delete_flow(sk);
}
/* Check if port is acceptable. Possibly find first available port.