freebsd-skq/share/man/man9/netisr.9
bz 83f1495433 Update epair(4) to the new netisr implementation and polish
things a bit:
- use dpcpu data to track the ifps with packets queued up,
- per-cpu locking and driver flags
- along with .nh_drainedcpu and NETISR_POLICY_CPU.
- Put the mbufs in flight reference count, preventing interfaces
  from going away, under INVARIANTS as this is a general problem
  of the stack and should be solved in if.c/netisr but still good
  to verify the internal queuing logic.
- Permit changing the MTU to virtually everythinkg like we do for loopback.

Hook epair(4) up to the build.

Approved by:	re (kib)
2009-07-26 12:20:07 +00:00

219 lines
6.8 KiB
Groff

.\"
.\" Copyright (c) 2009 Robert N. M. Watson
.\" All rights reserved.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\" notice(s), this list of conditions and the following disclaimer as
.\" the first lines of this file unmodified other than the possible
.\" addition of one or more copyright notices.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\" notice(s), this list of conditions and the following disclaimer in the
.\" documentation and/or other materials provided with the distribution.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
.\" EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
.\" WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
.\" DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY
.\" DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
.\" (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
.\" SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
.\" CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
.\" DAMAGE.
.\"
.\" $FreeBSD$
.\"
.Dd July 26, 2009
.Dt NETISR 9
.Os
.Sh NAME
.Nm netisr
.Nd Kernel network dispatch service
.Sh SYNOPSIS
.In net/netisr.h
.Ft void
.Fn netisr_register "const struct netisr_handler *nhp"
.Ft void
.Fn netisr_unregister "const struct netisr_handler *nhp"
.Ft int
.Fn netisr_dispatch "u_int proto" "struct mbuf *m"
.Ft int
.Fn netisr_dispatch_src "u_int proto" "uintptr_t source" "struct mbuf *m"
.Ft int
.Fn netisr_queue "u_int proto" "struct mbuf *m"
.Ft int
.Fn netisr_queue_src "u_int proto" "uintptr_t source" "struct mbuf *m"
.Ft void
.Fn netisr_clearqdrops "const struct netisr_handler *nhp"
.Ft void
.Fn netisr_getqdrops "const struct netisr_handler *nhp" "u_int64_t *qdropsp"
.Ft void
.Fn netisr_getqlimit "const struct netisr_handler *nhp" "u_int *qlimitp"
.Ft int
.Fn netisr_setqlimit "const struct netisr_handler *nhp" "u_int qlimit"
.Ft u_int
.Fn netisr_default_flow2cpu "u_int flowid"
.Ft u_int
.Fn netisr_get_cpucount "void"
.Ft u_int
.Fn netisr_get_cpuid "u_int cpunumber"
.Sh DESCRIPTION
The
.Nm
kernel interface suite allows device drivers (and other packet sources) to
direct packets to protocols for directly dispatched or deferred processing.
.Ss Protocol registration
Protocols register and unregister handlers using
.Fn netisr_register
and
.Fn netisr_unregister ,
and may also manage queue limits and statistics using the
.Fn netisr_clearqdrops ,
.Fn netisr_getqdrops ,
.Fn netisr_getqlimit ,
and
.Fn netisr_setqlimit.
.Pp
.Nm
supports multi-processor execution of handlers, and relies on a combination
of source ordering and protocol-specific ordering and work-placement
policies to decide how do distribute work across one or more worker
threads.
Registering protocols will declare one of three policies:
.Bl -tag -width NETISR_POLICY_SOURCE
.It Dv NETISR_POLICY_SOURCE
.Nm
should maintain source ordering without advice from the protocol.
.Nm
will ignore any flow IDs present on
.Vt mbuf
headers for the purposes of work placement.
.It Dv NETISR_POLICY_FLOW
.Nm
should maintain flow ordering as defined by the
.Vt mbuf
header flow ID field.
If the protocol implements
.Va nh_m2flow ,
then
.Nm
will query the protocol in the evet that the
.Vt mbuf
doesn't have a flow ID, falling back on source ordering.
.It NETISR_POLICY_CPU
.Nm
will entirely delegate all work placement decisions to the protocol,
querying
.Va nh_m2cpuid
for each packet.
.El
.Pp
Registration is declared using
.Vt "struct netisr_handler" ,
whose fields are defined as follows:
.Bl -tag -width "netisr_handler_t nh_handler"
.It Vt "const char *" Va nh_name
Unique character string name of the protocol, which may be included in
.Xr 2 sysctl
MIB names, so should not contain whitespace.
.It Vt netisr_handler_t Va nh_handler
Protocol handler function that will be invoked on each packet received for
the protocol.
.It Vt netisr_m2flow_t Va nh_m2flow
Optional protocol function to generate a flow ID and set
.Dv M_FLOWID
for packets that do not enter
.Nm
with
.Dv M_FLOWID
defined.
Will be used only with
.Dv NETISR_POLICY_FLOW .
.It Vt netisr_m2cpuid_t Va nh_m2cpuid
Protocol function to determine what CPU a packet should be processed on.
Will be used only with
.Dv NETISR_POLICY_CPU .
.It Vt netisr_drainedcpu_t Va nh_drainedcpu
Optional callback function that will be invoked when a per-CPU queue
was drained.
It will never fire for directly dispatched packets.
Unless fully understood, this special-purpose function should not be used.
.\" In case you intend to use this please send 50 chocolate bars to each
.\" of rwatson and bz and wait for an answer.
.It Vt u_int Va nh_proto
Protocol number used by both protocols to identify themselves to
.Nm ,
and by packet sources to select what handler will be used to process
packets.
A table of supported protocol numbers appears below.
For implementation reasons, protocol numbers great than 15 are currently
unsupported.
.It Vt u_int Va nh_qlimit
The maximum per-CPU queue depth for the protocol; due to internal
implementation details, the effective queue depth may be as much as twice
this number.
.It Vt u_int Va nh_policy
The ordering and work placement policy for the protocol, as described
earlier.
.El
.Ss Packet source interface
Packet sources, such as network interfaces, may request protocol processing
using the
.Fn netisr_dispatch
and
.Fn netisr_queue
interfaces.
Both accept a protocol number and
.Vt mbuf
argument, but while
.Fn netisr_queue
will always execute the protocol handler asynchonously in a deferred
context,
.Fn netisr_dispatch
will optionally direct dispatch if permitted by global and per-protocol
policy.
.Pp
In order to provide additional load balancing and flow information,
packet sources may also specify an opaque source identifier, which in
practice might be a network interface number or socket pointer, using
the
.Fn netisr_dispatch_src
and
.Fn netisr_queue_src
variants.
.Ss Protocol number constants
The follow protocol numbers are currently defined:
.Bl -tag -width NETISR_ATALK1
.It Dv NETISR_IP
IPv4
.It Dv NETISR_IGMP
IGMPv3 loopback
.It Dv NETISR_ROUTE
Routing socket loopback
.It Dv NETISR_AARP
Appletalk AARP
.It Dv NETISR_ATALK1
Appletalk phase 1
.It Dv NETISR_ATALK2
Appletalk phase 2
.It Dv NETISR_ARP
ARP
.It Dv NETISR_IPX
IPX/SPX
.It Dv NETISR_IPV6
IPv6
.It Dv NETISR_NATM
ATM
.It Dv NETISR_EPAIR
.Xr epair 4
.El
.Sh AUTHORS
This manual page and the
.Nm
implementation were written by
.An Robert N. M. Watson .