| Message ID | 20220510033219.2639364-1-eric.dumazet@gmail.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | tcp: BIG TCP implementation | expand |
On Mon, 2022-05-09 at 20:32 -0700, Eric Dumazet wrote: > From: Eric Dumazet <edumazet@google.com> > > This series implements BIG TCP as presented in netdev 0x15: > > https://netdevconf.info/0x15/session.html?BIG-TCP > > Jonathan Corbet made a nice summary: https://lwn.net/Articles/884104/ > > Standard TSO/GRO packet limit is 64KB > > With BIG TCP, we allow bigger TSO/GRO packet sizes for IPv6 traffic. > > Note that this feature is by default not enabled, because it might > break some eBPF programs assuming TCP header immediately follows IPv6 header. > > While tcpdump recognizes the HBH/Jumbo header, standard pcap filters > are unable to skip over IPv6 extension headers. > > Reducing number of packets traversing networking stack usually improves > performance, as shown on this experiment using a 100Gbit NIC, and 4K MTU. > > 'Standard' performance with current (74KB) limits. > for i in {1..10}; do ./netperf -t TCP_RR -H iroa23 -- -r80000,80000 -O MIN_LATENCY,P90_LATENCY,P99_LATENCY,THROUGHPUT|tail -1; done > 77 138 183 8542.19 > 79 143 178 8215.28 > 70 117 164 9543.39 > 80 144 176 8183.71 > 78 126 155 9108.47 > 80 146 184 8115.19 > 71 113 165 9510.96 > 74 113 164 9518.74 > 79 137 178 8575.04 > 73 111 171 9561.73 > > Now enable BIG TCP on both hosts. > > ip link set dev eth0 gro_max_size 185000 gso_max_size 185000 > for i in {1..10}; do ./netperf -t TCP_RR -H iroa23 -- -r80000,80000 -O MIN_LATENCY,P90_LATENCY,P99_LATENCY,THROUGHPUT|tail -1; done > 57 83 117 13871.38 > 64 118 155 11432.94 > 65 116 148 11507.62 > 60 105 136 12645.15 > 60 103 135 12760.34 > 60 102 134 12832.64 > 62 109 132 10877.68 > 58 82 115 14052.93 > 57 83 124 14212.58 > 57 82 119 14196.01 > > We see an increase of transactions per second, and lower latencies as well. > > v6: fix a compilation error for CONFIG_IPV6=n in > "net: allow gso_max_size to exceed 65536", reported by kernel bots. > > v5: Replaced two patches (that were adding new attributes) with patches > from Alexander Duyck. Idea is to reuse existing gso_max_size/gro_max_size > > v4: Rebased on top of Jakub series (Merge branch 'tso-gso-limit-split') > max_tso_size is now family independent. > > v3: Fixed a typo in RFC number (Alexander) > Added Reviewed-by: tags from Tariq on mlx4/mlx5 parts. > > v2: Removed the MAX_SKB_FRAGS change, this belongs to a different series. > Addressed feedback, for Alexander and nvidia folks. > > > Alexander Duyck (2): > net: allow gso_max_size to exceed 65536 > net: allow gro_max_size to exceed 65536 > > Coco Li (2): > ipv6: Add hop-by-hop header to jumbograms in ip6_output > mlx5: support BIG TCP packets > > Eric Dumazet (9): > net: add IFLA_TSO_{MAX_SIZE|SEGS} attributes > net: limit GSO_MAX_SIZE to 524280 bytes > tcp_cubic: make hystart_ack_delay() aware of BIG TCP > ipv6: add struct hop_jumbo_hdr definition > ipv6/gso: remove temporary HBH/jumbo header > ipv6/gro: insert temporary HBH/jumbo header > net: loopback: enable BIG TCP packets > veth: enable BIG TCP packets > mlx4: support BIG TCP packets Looked over the changes to my patches and they all look good (sorry for not catching that myself). This approach addresses all the concerns I had. For the series: Acked-by: Alexander Duyck <alexanderduyck@fb.com>
From: Eric Dumazet <edumazet@google.com> This series implements BIG TCP as presented in netdev 0x15: https://netdevconf.info/0x15/session.html?BIG-TCP Jonathan Corbet made a nice summary: https://lwn.net/Articles/884104/ Standard TSO/GRO packet limit is 64KB With BIG TCP, we allow bigger TSO/GRO packet sizes for IPv6 traffic. Note that this feature is by default not enabled, because it might break some eBPF programs assuming TCP header immediately follows IPv6 header. While tcpdump recognizes the HBH/Jumbo header, standard pcap filters are unable to skip over IPv6 extension headers. Reducing number of packets traversing networking stack usually improves performance, as shown on this experiment using a 100Gbit NIC, and 4K MTU. 'Standard' performance with current (74KB) limits. for i in {1..10}; do ./netperf -t TCP_RR -H iroa23 -- -r80000,80000 -O MIN_LATENCY,P90_LATENCY,P99_LATENCY,THROUGHPUT|tail -1; done 77 138 183 8542.19 79 143 178 8215.28 70 117 164 9543.39 80 144 176 8183.71 78 126 155 9108.47 80 146 184 8115.19 71 113 165 9510.96 74 113 164 9518.74 79 137 178 8575.04 73 111 171 9561.73 Now enable BIG TCP on both hosts. ip link set dev eth0 gro_max_size 185000 gso_max_size 185000 for i in {1..10}; do ./netperf -t TCP_RR -H iroa23 -- -r80000,80000 -O MIN_LATENCY,P90_LATENCY,P99_LATENCY,THROUGHPUT|tail -1; done 57 83 117 13871.38 64 118 155 11432.94 65 116 148 11507.62 60 105 136 12645.15 60 103 135 12760.34 60 102 134 12832.64 62 109 132 10877.68 58 82 115 14052.93 57 83 124 14212.58 57 82 119 14196.01 We see an increase of transactions per second, and lower latencies as well. v6: fix a compilation error for CONFIG_IPV6=n in "net: allow gso_max_size to exceed 65536", reported by kernel bots. v5: Replaced two patches (that were adding new attributes) with patches from Alexander Duyck. Idea is to reuse existing gso_max_size/gro_max_size v4: Rebased on top of Jakub series (Merge branch 'tso-gso-limit-split') max_tso_size is now family independent. v3: Fixed a typo in RFC number (Alexander) Added Reviewed-by: tags from Tariq on mlx4/mlx5 parts. v2: Removed the MAX_SKB_FRAGS change, this belongs to a different series. Addressed feedback, for Alexander and nvidia folks. Alexander Duyck (2): net: allow gso_max_size to exceed 65536 net: allow gro_max_size to exceed 65536 Coco Li (2): ipv6: Add hop-by-hop header to jumbograms in ip6_output mlx5: support BIG TCP packets Eric Dumazet (9): net: add IFLA_TSO_{MAX_SIZE|SEGS} attributes net: limit GSO_MAX_SIZE to 524280 bytes tcp_cubic: make hystart_ack_delay() aware of BIG TCP ipv6: add struct hop_jumbo_hdr definition ipv6/gso: remove temporary HBH/jumbo header ipv6/gro: insert temporary HBH/jumbo header net: loopback: enable BIG TCP packets veth: enable BIG TCP packets mlx4: support BIG TCP packets drivers/net/ethernet/amd/xgbe/xgbe.h | 3 +- .../net/ethernet/mellanox/mlx4/en_netdev.c | 3 + drivers/net/ethernet/mellanox/mlx4/en_tx.c | 47 +++++++++-- .../net/ethernet/mellanox/mlx5/core/en_main.c | 1 + .../net/ethernet/mellanox/mlx5/core/en_rx.c | 2 +- .../net/ethernet/mellanox/mlx5/core/en_tx.c | 84 +++++++++++++++---- drivers/net/ethernet/sfc/ef100_nic.c | 3 +- drivers/net/ethernet/sfc/falcon/tx.c | 3 +- drivers/net/ethernet/sfc/tx_common.c | 3 +- drivers/net/ethernet/synopsys/dwc-xlgmac.h | 3 +- drivers/net/hyperv/rndis_filter.c | 2 +- drivers/net/loopback.c | 2 + drivers/net/veth.c | 1 + drivers/scsi/fcoe/fcoe.c | 2 +- include/linux/ipv6.h | 1 + include/linux/netdevice.h | 15 +++- include/net/ipv6.h | 44 ++++++++++ include/uapi/linux/if_link.h | 2 + net/bpf/test_run.c | 2 +- net/core/dev.c | 9 +- net/core/gro.c | 8 ++ net/core/rtnetlink.c | 16 ++-- net/core/sock.c | 14 ++++ net/ipv4/tcp_bbr.c | 2 +- net/ipv4/tcp_cubic.c | 4 +- net/ipv4/tcp_output.c | 2 +- net/ipv6/ip6_offload.c | 56 ++++++++++++- net/ipv6/ip6_output.c | 22 ++++- net/sctp/output.c | 3 +- tools/include/uapi/linux/if_link.h | 2 + 30 files changed, 301 insertions(+), 60 deletions(-)