freebsd-nq/share/man/man4/polling.4

.\" Copyright (c) 2002 Luigi Rizzo
.\" 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, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\"    notice, 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 AUTHOR AND CONTRIBUTORS ``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 AUTHOR OR CONTRIBUTORS 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 December 14, 2004
.Dt POLLING 4
.Os
.Sh NAME
.Nm polling
.Nd device polling support
.Sh SYNOPSIS
.Cd "options DEVICE_POLLING"
.Cd "options HZ=1000"
.Sh DESCRIPTION
Device polling
.Nm (
for brevity) refers to a technique that
lets the operating system periodically poll devices, instead of
relying on the devices to generate interrupts when they need attention.
This might seem inefficient and counterintuitive, but when done
properly,
.Nm
gives more control to the operating system on
when and how to handle devices, with a number of advantages in terms
of system responsiveness and performance.
.Pp
In particular,
.Nm
reduces the overhead for context
switches which is incurred when servicing interrupts, and
gives more control on the scheduling of the CPU between various
tasks (user processes, software interrupts, device handling)
which ultimately reduces the chances of livelock in the system.
.Ss Principles of Operation
In the normal, interrupt-based mode, devices generate an interrupt
whenever they need attention.
This in turn causes a
context switch and the execution of an interrupt handler
which performs whatever processing is needed by the device.
The duration of the interrupt handler is potentially unbounded
unless the device driver has been programmed with real-time
concerns in mind (which is generally not the case for
.Fx
drivers).
Furthermore, under heavy traffic load, the system might be
persistently processing interrupts without being able to
complete other work, either in the kernel or in userland.
.Pp
Device polling disables interrupts by polling devices at appropriate
times, i.e., on clock interrupts, system calls and within the idle loop.
This way, the context switch overhead is removed.
Furthermore,
the operating system can control accurately how much work to spend
in handling device events, and thus prevent livelock by reserving
some amount of CPU to other tasks.
.Pp
Enabling
.Nm
also changes the way software network interrupts
are scheduled, so there is never the risk of livelock because
packets are not processed to completion.
.Ss MIB Variables
The operation of
.Nm
is controlled by the following
.Xr sysctl 8
MIB variables:
.Pp
.Bl -tag -width indent -compact
.It Va kern.polling.enable
If set to non-zero,
.Nm
is enabled.
Default is disabled.
.Pp
.It Va kern.polling.user_frac
When
.Nm
is enabled, and provided that there is some work to do,
up to this percent of the CPU cycles is reserved to userland tasks,
the remaining fraction being available for
.Nm
processing.
Default is 50.
.Pp
.It Va kern.polling.burst
Maximum number of packets grabbed from each network interface in
each timer tick.
This number is dynamically adjusted by the kernel,
according to the programmed
.Va user_frac , burst_max ,
CPU speed, and system load.
.Pp
.It Va kern.polling.each_burst
The burst above is split into smaller chunks of this number of
packets, going round-robin among all interfaces registered for
.Nm .
This prevents the case that a large burst from a single interface
can saturate the IP interrupt queue
.Pq Va net.inet.ip.intr_queue_maxlen .
Default is 5.
.Pp
.It Va kern.polling.burst_max
Upper bound for
.Va kern.polling.burst .
Note that when
.Nm
is enabled, each interface can receive at most
.Pq Va HZ No * Va burst_max
packets per second unless there are spare CPU cycles available for
.Nm
in the idle loop.
This number should be tuned to match the expected load
(which can be quite high with GigE cards).
Default is 150 which is adequate for 100Mbit network and HZ=1000.
.Pp
.It Va kern.polling.idle_poll
Controls if
.Nm
is enabled in the idle loop.
There are no reasons (other than power saving or bugs in the scheduler's
handling of idle priority kernel threads) to disable this.
Note that -CURRENT apparently has some problems in this respect now,
so default is disabled.
.Pp
.It Va kern.polling.poll_in_trap
Controls if
.Nm
is enabled during hardware traps.
Enabling this can be useful to improve the network responsiveness
of boxes with 100% CPU usage.
Default is disabled.
.Pp
.It Va kern.polling.reg_frac
Controls how often (every
.Va reg_frac No / Va HZ
seconds) the status registers of the device are checked for error
conditions and the like.
Increasing this value reduces the load on the bus, but also delays
the error detection.
Default is 20.
.Pp
.It Va kern.polling.handlers
How many active devices have registered for
.Nm .
.Pp
.It Va kern.polling.short_ticks
.It Va kern.polling.lost_polls
.It Va kern.polling.pending_polls
.It Va kern.polling.residual_burst
.It Va kern.polling.phase
.It Va kern.polling.suspect
.It Va kern.polling.stalled
Debugging variables.
.El
.Sh SUPPORTED DEVICES
Device polling requires explicit modifications to the device drivers.
As of this writing, the
.Xr dc 4 ,
.Xr em 4 ,
.Xr fwe 4 ,
.Xr fwip 4 ,
.Xr fxp 4 ,
.Xr ixgb 4 ,
.Xr nge 4 ,
.Xr re 4 ,
.Xr rl 4 ,
.Xr sf 4 ,
.Xr sis 4 ,
.Xr ste 4 ,
.Xr vge 4 ,
and
.Xr vr 4
devices are supported, with others in the works.
The modifications are rather straightforward, consisting in
the extraction of the inner part of the interrupt service routine
and writing a callback function,
.Fn *_poll ,
which is invoked
to probe the device for events and process them.
(See the
conditionally compiled sections of the devices mentioned above
for more details.)
.Pp
As in the worst case the devices are only polled on
clock interrupts, in order to reduce the latency in processing
packets, it is advisable to increase the frequency of the clock
to at least 1000 HZ.
.Sh HISTORY
Device polling first appeared in
.Fx 4.6
and
.Fx 5.0 .
.Sh AUTHORS
Device polling was written by
.An Luigi Rizzo Aq luigi@iet.unipi.it .