From e1b21f3c8ca989dc94b25526fda1bb107691f1af Mon Sep 17 00:00:00 2001 From: Kevin Krakauer Date: Wed, 6 Nov 2019 14:24:38 -0800 Subject: Use PacketBuffers, rather than VectorisedViews, in netstack. PacketBuffers are analogous to Linux's sk_buff. They hold all information about a packet, headers, and payload. This is important for: * iptables to access various headers of packets * Preventing the clutter of passing different net and link headers along with VectorisedViews to packet handling functions. This change only affects the incoming packet path, and a future change will change the outgoing path. Benchmark Regular PacketBufferPtr PacketBufferConcrete -------------------------------------------------------------------------------- BM_Recvmsg 400.715MB/s 373.676MB/s 396.276MB/s BM_Sendmsg 361.832MB/s 333.003MB/s 335.571MB/s BM_Recvfrom 453.336MB/s 393.321MB/s 381.650MB/s BM_Sendto 378.052MB/s 372.134MB/s 341.342MB/s BM_SendmsgTCP/0/1k 353.711MB/s 316.216MB/s 322.747MB/s BM_SendmsgTCP/0/2k 600.681MB/s 588.776MB/s 565.050MB/s BM_SendmsgTCP/0/4k 995.301MB/s 888.808MB/s 941.888MB/s BM_SendmsgTCP/0/8k 1.517GB/s 1.274GB/s 1.345GB/s BM_SendmsgTCP/0/16k 1.872GB/s 1.586GB/s 1.698GB/s BM_SendmsgTCP/0/32k 1.017GB/s 1.020GB/s 1.133GB/s BM_SendmsgTCP/0/64k 475.626MB/s 584.587MB/s 627.027MB/s BM_SendmsgTCP/0/128k 416.371MB/s 503.434MB/s 409.850MB/s BM_SendmsgTCP/0/256k 323.449MB/s 449.599MB/s 388.852MB/s BM_SendmsgTCP/0/512k 243.992MB/s 267.676MB/s 314.474MB/s BM_SendmsgTCP/0/1M 95.138MB/s 95.874MB/s 95.417MB/s BM_SendmsgTCP/0/2M 96.261MB/s 94.977MB/s 96.005MB/s BM_SendmsgTCP/0/4M 96.512MB/s 95.978MB/s 95.370MB/s BM_SendmsgTCP/0/8M 95.603MB/s 95.541MB/s 94.935MB/s BM_SendmsgTCP/0/16M 94.598MB/s 94.696MB/s 94.521MB/s BM_SendmsgTCP/0/32M 94.006MB/s 94.671MB/s 94.768MB/s BM_SendmsgTCP/0/64M 94.133MB/s 94.333MB/s 94.746MB/s BM_SendmsgTCP/0/128M 93.615MB/s 93.497MB/s 93.573MB/s BM_SendmsgTCP/0/256M 93.241MB/s 95.100MB/s 93.272MB/s BM_SendmsgTCP/1/1k 303.644MB/s 316.074MB/s 308.430MB/s BM_SendmsgTCP/1/2k 537.093MB/s 584.962MB/s 529.020MB/s BM_SendmsgTCP/1/4k 882.362MB/s 939.087MB/s 892.285MB/s BM_SendmsgTCP/1/8k 1.272GB/s 1.394GB/s 1.296GB/s BM_SendmsgTCP/1/16k 1.802GB/s 2.019GB/s 1.830GB/s BM_SendmsgTCP/1/32k 2.084GB/s 2.173GB/s 2.156GB/s BM_SendmsgTCP/1/64k 2.515GB/s 2.463GB/s 2.473GB/s BM_SendmsgTCP/1/128k 2.811GB/s 3.004GB/s 2.946GB/s BM_SendmsgTCP/1/256k 3.008GB/s 3.159GB/s 3.171GB/s BM_SendmsgTCP/1/512k 2.980GB/s 3.150GB/s 3.126GB/s BM_SendmsgTCP/1/1M 2.165GB/s 2.233GB/s 2.163GB/s BM_SendmsgTCP/1/2M 2.370GB/s 2.219GB/s 2.453GB/s BM_SendmsgTCP/1/4M 2.005GB/s 2.091GB/s 2.214GB/s BM_SendmsgTCP/1/8M 2.111GB/s 2.013GB/s 2.109GB/s BM_SendmsgTCP/1/16M 1.902GB/s 1.868GB/s 1.897GB/s BM_SendmsgTCP/1/32M 1.655GB/s 1.665GB/s 1.635GB/s BM_SendmsgTCP/1/64M 1.575GB/s 1.547GB/s 1.575GB/s BM_SendmsgTCP/1/128M 1.524GB/s 1.584GB/s 1.580GB/s BM_SendmsgTCP/1/256M 1.579GB/s 1.607GB/s 1.593GB/s PiperOrigin-RevId: 278940079 --- pkg/tcpip/stack/registration.go | 64 ++++++++++++++++++++++++----------------- 1 file changed, 37 insertions(+), 27 deletions(-) (limited to 'pkg/tcpip/stack/registration.go') diff --git a/pkg/tcpip/stack/registration.go b/pkg/tcpip/stack/registration.go index d7c124e81..5806d294c 100644 --- a/pkg/tcpip/stack/registration.go +++ b/pkg/tcpip/stack/registration.go @@ -64,16 +64,15 @@ type TransportEndpoint interface { UniqueID() uint64 // HandlePacket is called by the stack when new packets arrive to - // this transport endpoint. + // this transport endpoint. It sets pkt.TransportHeader. // - // HandlePacket takes ownership of vv. - HandlePacket(r *Route, id TransportEndpointID, vv buffer.VectorisedView) + // HandlePacket takes ownership of pkt. + HandlePacket(r *Route, id TransportEndpointID, pkt tcpip.PacketBuffer) - // HandleControlPacket is called by the stack when new control (e.g., + // HandleControlPacket is called by the stack when new control (e.g. // ICMP) packets arrive to this transport endpoint. - // - // HandleControlPacket takes ownership of vv. - HandleControlPacket(id TransportEndpointID, typ ControlType, extra uint32, vv buffer.VectorisedView) + // HandleControlPacket takes ownership of pkt. + HandleControlPacket(id TransportEndpointID, typ ControlType, extra uint32, pkt tcpip.PacketBuffer) // Close puts the endpoint in a closed state and frees all resources // associated with it. This cleanup may happen asynchronously. Wait can @@ -99,8 +98,8 @@ type RawTransportEndpoint interface { // this transport endpoint. The packet contains all data from the link // layer up. // - // HandlePacket takes ownership of packet and netHeader. - HandlePacket(r *Route, netHeader buffer.View, packet buffer.VectorisedView) + // HandlePacket takes ownership of pkt. + HandlePacket(r *Route, pkt tcpip.PacketBuffer) } // PacketEndpoint is the interface that needs to be implemented by packet @@ -117,8 +116,8 @@ type PacketEndpoint interface { // linkHeader may have a length of 0, in which case the PacketEndpoint // should construct its own ethernet header for applications. // - // HandlePacket takes ownership of packet and linkHeader. - HandlePacket(nicid tcpip.NICID, addr tcpip.LinkAddress, netProto tcpip.NetworkProtocolNumber, packet buffer.VectorisedView, linkHeader buffer.View) + // HandlePacket takes ownership of pkt. + HandlePacket(nicid tcpip.NICID, addr tcpip.LinkAddress, netProto tcpip.NetworkProtocolNumber, pkt tcpip.PacketBuffer) } // TransportProtocol is the interface that needs to be implemented by transport @@ -148,7 +147,9 @@ type TransportProtocol interface { // // The return value indicates whether the packet was well-formed (for // stats purposes only). - HandleUnknownDestinationPacket(r *Route, id TransportEndpointID, netHeader buffer.View, vv buffer.VectorisedView) bool + // + // HandleUnknownDestinationPacket takes ownership of pkt. + HandleUnknownDestinationPacket(r *Route, id TransportEndpointID, pkt tcpip.PacketBuffer) bool // SetOption allows enabling/disabling protocol specific features. // SetOption returns an error if the option is not supported or the @@ -166,17 +167,21 @@ type TransportProtocol interface { // the network layer. type TransportDispatcher interface { // DeliverTransportPacket delivers packets to the appropriate - // transport protocol endpoint. It also returns the network layer - // header for the enpoint to inspect or pass up the stack. + // transport protocol endpoint. + // + // pkt.NetworkHeader must be set before calling DeliverTransportPacket. // - // DeliverTransportPacket takes ownership of vv and netHeader. - DeliverTransportPacket(r *Route, protocol tcpip.TransportProtocolNumber, netHeader buffer.View, vv buffer.VectorisedView) + // DeliverTransportPacket takes ownership of pkt. + DeliverTransportPacket(r *Route, protocol tcpip.TransportProtocolNumber, pkt tcpip.PacketBuffer) // DeliverTransportControlPacket delivers control packets to the // appropriate transport protocol endpoint. // - // DeliverTransportControlPacket takes ownership of vv. - DeliverTransportControlPacket(local, remote tcpip.Address, net tcpip.NetworkProtocolNumber, trans tcpip.TransportProtocolNumber, typ ControlType, extra uint32, vv buffer.VectorisedView) + // pkt.NetworkHeader must be set before calling + // DeliverTransportControlPacket. + // + // DeliverTransportControlPacket takes ownership of pkt. + DeliverTransportControlPacket(local, remote tcpip.Address, net tcpip.NetworkProtocolNumber, trans tcpip.TransportProtocolNumber, typ ControlType, extra uint32, pkt tcpip.PacketBuffer) } // PacketLooping specifies where an outbound packet should be sent. @@ -248,10 +253,10 @@ type NetworkEndpoint interface { NICID() tcpip.NICID // HandlePacket is called by the link layer when new packets arrive to - // this network endpoint. + // this network endpoint. It sets pkt.NetworkHeader. // - // HandlePacket takes ownership of vv. - HandlePacket(r *Route, vv buffer.VectorisedView) + // HandlePacket takes ownership of pkt. + HandlePacket(r *Route, pkt tcpip.PacketBuffer) // Close is called when the endpoint is reomved from a stack. Close() @@ -294,11 +299,14 @@ type NetworkProtocol interface { // the data link layer. type NetworkDispatcher interface { // DeliverNetworkPacket finds the appropriate network protocol endpoint - // and hands the packet over for further processing. linkHeader may have - // length 0 when the caller does not have ethernet data. + // and hands the packet over for further processing. + // + // pkt.LinkHeader may or may not be set before calling + // DeliverNetworkPacket. Some packets do not have link headers (e.g. + // packets sent via loopback), and won't have the field set. // - // DeliverNetworkPacket takes ownership of vv and linkHeader. - DeliverNetworkPacket(linkEP LinkEndpoint, remote, local tcpip.LinkAddress, protocol tcpip.NetworkProtocolNumber, vv buffer.VectorisedView, linkHeader buffer.View) + // DeliverNetworkPacket takes ownership of pkt. + DeliverNetworkPacket(linkEP LinkEndpoint, remote, local tcpip.LinkAddress, protocol tcpip.NetworkProtocolNumber, pkt tcpip.PacketBuffer) } // LinkEndpointCapabilities is the type associated with the capabilities @@ -329,7 +337,9 @@ const ( // LinkEndpoint is the interface implemented by data link layer protocols (e.g., // ethernet, loopback, raw) and used by network layer protocols to send packets -// out through the implementer's data link endpoint. +// out through the implementer's data link endpoint. When a link header exists, +// it sets each tcpip.PacketBuffer's LinkHeader field before passing it up the +// stack. type LinkEndpoint interface { // MTU is the maximum transmission unit for this endpoint. This is // usually dictated by the backing physical network; when such a @@ -395,7 +405,7 @@ type InjectableLinkEndpoint interface { LinkEndpoint // InjectInbound injects an inbound packet. - InjectInbound(protocol tcpip.NetworkProtocolNumber, vv buffer.VectorisedView) + InjectInbound(protocol tcpip.NetworkProtocolNumber, pkt tcpip.PacketBuffer) // InjectOutbound writes a fully formed outbound packet directly to the // link. -- cgit v1.2.3