1.\" 2.\" Copyright (c) 2009 Robert N. M. Watson 3.\" All rights reserved. 4.\" 5.\" Redistribution and use in source and binary forms, with or without 6.\" modification, are permitted provided that the following conditions 7.\" are met: 8.\" 1. Redistributions of source code must retain the above copyright 9.\" notice(s), this list of conditions and the following disclaimer as 10.\" the first lines of this file unmodified other than the possible 11.\" addition of one or more copyright notices. 12.\" 2. Redistributions in binary form must reproduce the above copyright 13.\" notice(s), this list of conditions and the following disclaimer in the 14.\" documentation and/or other materials provided with the distribution. 15.\" 16.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY 17.\" EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 18.\" WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 19.\" DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY 20.\" DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 21.\" (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 22.\" SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER 23.\" CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 24.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH 26.\" DAMAGE. 27.\" 28.\" $FreeBSD$ 29.\" 30.Dd February 22, 2010 31.Dt NETISR 9 32.Os 33.Sh NAME 34.Nm netisr 35.Nd Kernel network dispatch service 36.Sh SYNOPSIS 37.In net/netisr.h 38.Ft void 39.Fn netisr_register "const struct netisr_handler *nhp" 40.Ft void 41.Fn netisr_unregister "const struct netisr_handler *nhp" 42.Ft int 43.Fn netisr_dispatch "u_int proto" "struct mbuf *m" 44.Ft int 45.Fn netisr_dispatch_src "u_int proto" "uintptr_t source" "struct mbuf *m" 46.Ft int 47.Fn netisr_queue "u_int proto" "struct mbuf *m" 48.Ft int 49.Fn netisr_queue_src "u_int proto" "uintptr_t source" "struct mbuf *m" 50.Ft void 51.Fn netisr_clearqdrops "const struct netisr_handler *nhp" 52.Ft void 53.Fn netisr_getqdrops "const struct netisr_handler *nhp" "uint64_t *qdropsp" 54.Ft void 55.Fn netisr_getqlimit "const struct netisr_handler *nhp" "u_int *qlimitp" 56.Ft int 57.Fn netisr_setqlimit "const struct netisr_handler *nhp" "u_int qlimit" 58.Ft u_int 59.Fn netisr_default_flow2cpu "u_int flowid" 60.Ft u_int 61.Fn netisr_get_cpucount "void" 62.Ft u_int 63.Fn netisr_get_cpuid "u_int cpunumber" 64.Sh DESCRIPTION 65The 66.Nm 67kernel interface suite allows device drivers (and other packet sources) to 68direct packets to protocols for directly dispatched or deferred processing. 69Protocol registration and work stream statistics may be monitored using 70.Xr netstat 1 . 71.Ss Protocol registration 72Protocols register and unregister handlers using 73.Fn netisr_register 74and 75.Fn netisr_unregister , 76and may also manage queue limits and statistics using the 77.Fn netisr_clearqdrops , 78.Fn netisr_getqdrops , 79.Fn netisr_getqlimit , 80and 81.Fn netisr_setqlimit . 82.Pp 83.Nm 84supports multi-processor execution of handlers, and relies on a combination 85of source ordering and protocol-specific ordering and work-placement 86policies to decide how to distribute work across one or more worker 87threads. 88Registering protocols will declare one of three policies: 89.Bl -tag -width NETISR_POLICY_SOURCE 90.It Dv NETISR_POLICY_SOURCE 91.Nm 92should maintain source ordering without advice from the protocol. 93.Nm 94will ignore any flow IDs present on 95.Vt mbuf 96headers for the purposes of work placement. 97.It Dv NETISR_POLICY_FLOW 98.Nm 99should maintain flow ordering as defined by the 100.Vt mbuf 101header flow ID field. 102If the protocol implements 103.Va nh_m2flow , 104then 105.Nm 106will query the protocol in the event that the 107.Vt mbuf 108doesn't have a flow ID, falling back on source ordering. 109.It NETISR_POLICY_CPU 110.Nm 111will entirely delegate all work placement decisions to the protocol, 112querying 113.Va nh_m2cpuid 114for each packet. 115.El 116.Pp 117Registration is declared using 118.Vt "struct netisr_handler" , 119whose fields are defined as follows: 120.Bl -tag -width "netisr_handler_t nh_handler" 121.It Vt "const char *" Va nh_name 122Unique character string name of the protocol, which may be included in 123.Xr sysctl 2 124MIB names, so should not contain whitespace. 125.It Vt netisr_handler_t Va nh_handler 126Protocol handler function that will be invoked on each packet received for 127the protocol. 128.It Vt netisr_m2flow_t Va nh_m2flow 129Optional protocol function to generate a flow ID and set 130.Dv M_FLOWID 131for packets that do not enter 132.Nm 133with 134.Dv M_FLOWID 135defined. 136Will be used only with 137.Dv NETISR_POLICY_FLOW . 138.It Vt netisr_m2cpuid_t Va nh_m2cpuid 139Protocol function to determine what CPU a packet should be processed on. 140Will be used only with 141.Dv NETISR_POLICY_CPU . 142.It Vt netisr_drainedcpu_t Va nh_drainedcpu 143Optional callback function that will be invoked when a per-CPU queue 144was drained. 145It will never fire for directly dispatched packets. 146Unless fully understood, this special-purpose function should not be used. 147.\" In case you intend to use this please send 50 chocolate bars to each 148.\" of rwatson and bz and wait for an answer. 149.It Vt u_int Va nh_proto 150Protocol number used by both protocols to identify themselves to 151.Nm , 152and by packet sources to select what handler will be used to process 153packets. 154A table of supported protocol numbers appears below. 155For implementation reasons, protocol numbers great than 15 are currently 156unsupported. 157.It Vt u_int Va nh_qlimit 158The maximum per-CPU queue depth for the protocol; due to internal 159implementation details, the effective queue depth may be as much as twice 160this number. 161.It Vt u_int Va nh_policy 162The ordering and work placement policy for the protocol, as described 163earlier. 164.El 165.Ss Packet source interface 166Packet sources, such as network interfaces, may request protocol processing 167using the 168.Fn netisr_dispatch 169and 170.Fn netisr_queue 171interfaces. 172Both accept a protocol number and 173.Vt mbuf 174argument, but while 175.Fn netisr_queue 176will always execute the protocol handler asynchronously in a deferred 177context, 178.Fn netisr_dispatch 179will optionally direct dispatch if permitted by global and per-protocol 180policy. 181.Pp 182In order to provide additional load balancing and flow information, 183packet sources may also specify an opaque source identifier, which in 184practice might be a network interface number or socket pointer, using 185the 186.Fn netisr_dispatch_src 187and 188.Fn netisr_queue_src 189variants. 190.Ss Protocol number constants 191The follow protocol numbers are currently defined: 192.Bl -tag -width NETISR_ROUTE 193.It Dv NETISR_IP 194IPv4 195.It Dv NETISR_IGMP 196IGMPv3 loopback 197.It Dv NETISR_ROUTE 198Routing socket loopback 199.It Dv NETISR_ARP 200ARP 201.It Dv NETISR_IPV6 202IPv6 203.It Dv NETISR_NATM 204ATM 205.It Dv NETISR_EPAIR 206.Xr netstat 1 , 207.Xr epair 4 208.El 209.Sh AUTHORS 210This manual page and the 211.Nm 212implementation were written by 213.An Robert N. M. Watson . 214