alistair23-linux/net/ipv4
Eric Dumazet 472c2e07ee tcp: add one skb cache for tx
On hosts with a lot of cores, RPC workloads suffer from heavy contention on slab spinlocks.

    20.69%  [kernel]       [k] queued_spin_lock_slowpath
     5.64%  [kernel]       [k] _raw_spin_lock
     3.83%  [kernel]       [k] syscall_return_via_sysret
     3.48%  [kernel]       [k] __entry_text_start
     1.76%  [kernel]       [k] __netif_receive_skb_core
     1.64%  [kernel]       [k] __fget

For each sendmsg(), we allocate one skb, and free it at the time ACK packet comes.

In many cases, ACK packets are handled by another cpus, and this unfortunately
incurs heavy costs for slab layer.

This patch uses an extra pointer in socket structure, so that we try to reuse
the same skb and avoid these expensive costs.

We cache at most one skb per socket so this should be safe as far as
memory pressure is concerned.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-03-23 21:57:38 -04:00
..
bpfilter net: bpfilter: disallow to remove bpfilter module while being used 2019-01-11 18:05:41 -08:00
netfilter netfilter: nf_tables: merge ipv4 and ipv6 nat chain types 2019-03-01 14:36:59 +01:00
af_inet.c gso: validate gso_type on ipip style tunnels 2019-02-20 11:24:27 -08:00
ah4.c net-ipv4: remove 2 always zero parameters from ipv4_redirect() 2018-09-26 20:30:55 -07:00
arp.c net: Evict neighbor entries on carrier down 2018-10-12 09:47:39 -07:00
cipso_ipv4.c netlabel: fix out-of-bounds memory accesses 2019-02-27 21:45:24 -08:00
datagram.c ipv4: Allow sending multicast packets on specific i/f using VRF socket 2018-10-02 22:28:17 -07:00
devinet.c net: ignore sysctl_devconf_inherit_init_net without SYSCTL 2019-03-04 13:14:34 -08:00
esp4.c esp: Skip TX bytes accounting when sending from a request socket 2019-01-28 11:20:58 +01:00
esp4_offload.c net: use skb_sec_path helper in more places 2018-12-19 11:21:37 -08:00
fib_frontend.c ipv4: Return error for RTA_VIA attribute 2019-02-26 13:23:17 -08:00
fib_lookup.h
fib_notifier.c
fib_rules.c ipv4: fib_rules: Fix possible infinite loop in fib_empty_table 2018-12-30 12:57:04 -08:00
fib_semantics.c ipv4: fib: use struct_size() in kzalloc() 2019-02-01 15:12:29 -08:00
fib_trie.c ipv4: Allow amount of dirty memory from fib resizing to be controllable 2019-03-21 13:29:53 -07:00
fou.c genetlink: make policy common to family 2019-03-22 10:38:23 -04:00
gre_demux.c net: ip_gre: use erspan key field for tunnel lookup 2019-01-22 11:52:17 -08:00
gre_offload.c
icmp.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2019-03-02 12:54:35 -08:00
igmp.c net: remove unneeded switch fall-through 2019-02-21 13:48:00 -08:00
inet_connection_sock.c inet: minor optimization for backlog setting in listen(2) 2018-11-07 22:31:07 -08:00
inet_diag.c inet_diag: fix reporting cgroup classid and fallback to priority 2019-02-12 13:35:57 -05:00
inet_fragment.c net: remove unused struct inet_frag_queue.fragments field 2019-02-26 08:27:05 -08:00
inet_hashtables.c net: dccp: fix kernel crash on module load 2018-12-24 15:27:56 -08:00
inet_timewait_sock.c
inetpeer.c net: ipv4: use a dedicated counter for icmp_v4 redirect packets 2019-02-08 21:50:15 -08:00
ip_forward.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2018-12-20 11:53:36 -08:00
ip_fragment.c net: remove unused struct inet_frag_queue.fragments field 2019-02-26 08:27:05 -08:00
ip_gre.c route: Add multipath_hash in flowi_common to make user-define hash 2019-02-27 12:50:17 -08:00
ip_input.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2019-03-02 12:54:35 -08:00
ip_options.c net: avoid use IPCB in cipso_v4_error 2019-02-25 14:32:35 -08:00
ip_output.c sk_buff: add skb extension infrastructure 2018-12-19 11:21:37 -08:00
ip_sockglue.c ip: on queued skb use skb_header_pointer instead of pskb_may_pull 2019-01-10 09:27:20 -05:00
ip_tunnel.c iptunnel: NULL pointer deref for ip_md_tunnel_xmit 2019-03-06 10:43:06 -08:00
ip_tunnel_core.c ip_tunnel: Add dst_cache support in lwtunnel_state of ip tunnel 2019-02-24 22:13:49 -08:00
ip_vti.c vti4: Fix a ipip packet processing bug in 'IPCOMP' virtual tunnel 2019-01-09 14:00:37 +01:00
ipcomp.c net-ipv4: remove 2 always zero parameters from ipv4_redirect() 2018-09-26 20:30:55 -07:00
ipconfig.c ipconfig: add carrier_timeout kernel parameter 2019-02-01 15:24:13 -08:00
ipip.c ip_tunnel: Add tnl_update_pmtu in ip_md_tunnel_xmit 2019-01-26 09:43:03 -08:00
ipmr.c ipmr: ip6mr: Create new sockopt to clear mfc cache or vifs 2019-02-21 13:05:05 -08:00
ipmr_base.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2018-10-19 11:03:06 -07:00
Kconfig
Makefile bpf, sockmap: convert to generic sk_msg interface 2018-10-15 12:23:19 -07:00
metrics.c net: Add extack argument to ip_fib_metrics_init 2018-11-06 15:00:45 -08:00
netfilter.c netfilter: ipv4: remove useless export_symbol 2019-01-28 11:32:58 +01:00
netlink.c ipv4: Add ICMPv6 support when parse route ipproto 2019-03-01 16:41:27 -08:00
ping.c ipv4: Allow sending multicast packets on specific i/f using VRF socket 2018-10-02 22:28:17 -07:00
proc.c tcp: implement coalescing on backlog queue 2018-11-30 13:26:54 -08:00
protocol.c fou, fou6: ICMP error handlers for FoU and GUE 2018-11-08 17:13:08 -08:00
raw.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2018-12-20 11:53:36 -08:00
raw_diag.c
route.c net: dst: remove gc leftovers 2019-03-21 13:39:25 -07:00
syncookies.c tcp: free request sock directly upon TFO or syncookies error 2019-03-19 14:13:01 -07:00
sysctl_net_ipv4.c ipv4: Allow amount of dirty memory from fib resizing to be controllable 2019-03-21 13:29:53 -07:00
tcp.c tcp: add one skb cache for tx 2019-03-23 21:57:38 -04:00
tcp_bbr.c tcp_bbr: adapt cwnd based on ack aggregation estimation 2019-01-24 22:27:27 -08:00
tcp_bic.c
tcp_bpf.c bpf: sk_msg, sock{map|hash} redirect through ULP 2018-12-20 23:47:09 +01:00
tcp_cdg.c tcp: cdg: use tcp high resolution clock cache 2018-10-15 22:56:42 -07:00
tcp_cong.c
tcp_cubic.c
tcp_dctcp.c tcp: refactor DCTCP ECN ACK handling 2018-10-10 22:26:00 -07:00
tcp_dctcp.h tcp: refactor DCTCP ECN ACK handling 2018-10-10 22:26:00 -07:00
tcp_diag.c
tcp_fastopen.c
tcp_highspeed.c
tcp_htcp.c
tcp_hybla.c
tcp_illinois.c
tcp_input.c tcp: free request sock directly upon TFO or syncookies error 2019-03-19 14:13:01 -07:00
tcp_ipv4.c tcp: Don't access TCP_SKB_CB before initializing it 2019-03-11 15:36:49 -07:00
tcp_lp.c
tcp_metrics.c genetlink: make policy common to family 2019-03-22 10:38:23 -04:00
tcp_minisocks.c tcp: use tcp_md5_needed for timewait sockets 2019-02-26 13:16:03 -08:00
tcp_nv.c
tcp_offload.c net: use indirect call wrappers at GRO transport layer 2018-12-15 13:23:02 -08:00
tcp_output.c tcp: remove conditional branches from tcp_mstamp_refresh() 2019-03-23 21:43:21 -04:00
tcp_rate.c
tcp_recovery.c
tcp_scalable.c
tcp_timer.c tcp: Refactor pingpong code 2019-01-27 13:29:43 -08:00
tcp_ulp.c tcp, ulp: remove socket lock assertion on ULP cleanup 2018-10-16 12:38:41 -07:00
tcp_vegas.c
tcp_vegas.h
tcp_veno.c
tcp_westwood.c
tcp_yeah.c
tunnel4.c net: Convert protocol error handlers from void to int 2018-11-08 17:13:08 -08:00
udp.c udp: fix possible user after free in error handler 2019-02-22 16:05:11 -08:00
udp_diag.c net: diag: document swapped src/dst in udp_dump_one. 2018-10-28 19:27:21 -07:00
udp_impl.h udp: add missing rehash callback to udplite 2019-01-17 15:01:08 -08:00
udp_offload.c udp: use indirect call wrappers for GRO socket lookup 2018-12-15 13:23:02 -08:00
udp_tunnel.c net/ipv4/udp_tunnel: prefer SO_BINDTOIFINDEX over SO_BINDTODEVICE 2019-01-17 14:55:52 -08:00
udplite.c udp: add missing rehash callback to udplite 2019-01-17 15:01:08 -08:00
xfrm4_input.c
xfrm4_mode_beet.c
xfrm4_mode_transport.c
xfrm4_mode_tunnel.c
xfrm4_output.c
xfrm4_policy.c
xfrm4_protocol.c net: Convert protocol error handlers from void to int 2018-11-08 17:13:08 -08:00
xfrm4_state.c
xfrm4_tunnel.c