xref: /freebsd/share/man/man9/netisr.9 (revision 830940567b49bb0c08dfaed40418999e76616909)
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 July 26, 2009
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" "u_int64_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.
69.Ss Protocol registration
70Protocols register and unregister handlers using
71.Fn netisr_register
72and
73.Fn netisr_unregister ,
74and may also manage queue limits and statistics using the
75.Fn netisr_clearqdrops ,
76.Fn netisr_getqdrops ,
77.Fn netisr_getqlimit ,
78and
79.Fn netisr_setqlimit.
80.Pp
81.Nm
82supports multi-processor execution of handlers, and relies on a combination
83of source ordering and protocol-specific ordering and work-placement
84policies to decide how do distribute work across one or more worker
85threads.
86Registering protocols will declare one of three policies:
87.Bl -tag -width NETISR_POLICY_SOURCE
88.It Dv NETISR_POLICY_SOURCE
89.Nm
90should maintain source ordering without advice from the protocol.
91.Nm
92will ignore any flow IDs present on
93.Vt mbuf
94headers for the purposes of work placement.
95.It Dv NETISR_POLICY_FLOW
96.Nm
97should maintain flow ordering as defined by the
98.Vt mbuf
99header flow ID field.
100If the protocol implements
101.Va nh_m2flow ,
102then
103.Nm
104will query the protocol in the evet that the
105.Vt mbuf
106doesn't have a flow ID, falling back on source ordering.
107.It NETISR_POLICY_CPU
108.Nm
109will entirely delegate all work placement decisions to the protocol,
110querying
111.Va nh_m2cpuid
112for each packet.
113.El
114.Pp
115Registration is declared using
116.Vt "struct netisr_handler" ,
117whose fields are defined as follows:
118.Bl -tag -width "netisr_handler_t nh_handler"
119.It Vt "const char *" Va nh_name
120Unique character string name of the protocol, which may be included in
121.Xr 2 sysctl
122MIB names, so should not contain whitespace.
123.It Vt netisr_handler_t Va nh_handler
124Protocol handler function that will be invoked on each packet received for
125the protocol.
126.It Vt netisr_m2flow_t Va nh_m2flow
127Optional protocol function to generate a flow ID and set
128.Dv M_FLOWID
129for packets that do not enter
130.Nm
131with
132.Dv M_FLOWID
133defined.
134Will be used only with
135.Dv NETISR_POLICY_FLOW .
136.It Vt netisr_m2cpuid_t Va nh_m2cpuid
137Protocol function to determine what CPU a packet should be processed on.
138Will be used only with
139.Dv NETISR_POLICY_CPU .
140.It Vt netisr_drainedcpu_t Va nh_drainedcpu
141Optional callback function that will be invoked when a per-CPU queue
142was drained.
143It will never fire for directly dispatched packets.
144Unless fully understood, this special-purpose function should not be used.
145.\" In case you intend to use this please send 50 chocolate bars to each
146.\" of rwatson and bz and wait for an answer.
147.It Vt u_int Va nh_proto
148Protocol number used by both protocols to identify themselves to
149.Nm ,
150and by packet sources to select what handler will be used to process
151packets.
152A table of supported protocol numbers appears below.
153For implementation reasons, protocol numbers great than 15 are currently
154unsupported.
155.It Vt u_int Va nh_qlimit
156The maximum per-CPU queue depth for the protocol; due to internal
157implementation details, the effective queue depth may be as much as twice
158this number.
159.It Vt u_int Va nh_policy
160The ordering and work placement policy for the protocol, as described
161earlier.
162.El
163.Ss Packet source interface
164Packet sources, such as network interfaces, may request protocol processing
165using the
166.Fn netisr_dispatch
167and
168.Fn netisr_queue
169interfaces.
170Both accept a protocol number and
171.Vt mbuf
172argument, but while
173.Fn netisr_queue
174will always execute the protocol handler asynchonously in a deferred
175context,
176.Fn netisr_dispatch
177will optionally direct dispatch if permitted by global and per-protocol
178policy.
179.Pp
180In order to provide additional load balancing and flow information,
181packet sources may also specify an opaque source identifier, which in
182practice might be a network interface number or socket pointer, using
183the
184.Fn netisr_dispatch_src
185and
186.Fn netisr_queue_src
187variants.
188.Ss Protocol number constants
189The follow protocol numbers are currently defined:
190.Bl -tag -width NETISR_ATALK1
191.It Dv NETISR_IP
192IPv4
193.It Dv NETISR_IGMP
194IGMPv3 loopback
195.It Dv NETISR_ROUTE
196Routing socket loopback
197.It Dv NETISR_AARP
198Appletalk AARP
199.It Dv NETISR_ATALK1
200Appletalk phase 1
201.It Dv NETISR_ATALK2
202Appletalk phase 2
203.It Dv NETISR_ARP
204ARP
205.It Dv NETISR_IPX
206IPX/SPX
207.It Dv NETISR_IPV6
208IPv6
209.It Dv NETISR_NATM
210ATM
211.It Dv NETISR_EPAIR
212.Xr epair 4
213.El
214.Sh AUTHORS
215This manual page and the
216.Nm
217implementation were written by
218.An Robert N. M. Watson .
219