New manpage for the Wavelan (wl) driver.
Submitted by: Jim Binkley <jrb@cs.pdx.edu>
This commit is contained in:
Mike Smith
parent
bfbee26298
commit
d28e8bd5da
|
|
@ -1,5 +1,5 @@
|
|||
.\"
|
||||
.\" Copyright (c) 1996 Michael Smith
|
||||
.\" Copyright (c) 1997, Jim Binkley
|
||||
.\" All rights reserved.
|
||||
.\"
|
||||
.\" Redistribution and use in source and binary forms, with or without
|
||||
|
|
@ -10,6 +10,11 @@
|
|||
.\" 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.
|
||||
.\" 3. All advertising materials mentioning features or use of this software
|
||||
.\" must display the following acknowledgement:
|
||||
.\" This product includes software developed by Jim Binkley
|
||||
.\" 4. The name of the author may not be used to endorse or promote products
|
||||
.\" derived from this software without specific prior written permission.
|
||||
.\"
|
||||
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
||||
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
|
|
@ -23,122 +28,120 @@
|
|||
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
||||
.\" SUCH DAMAGE.
|
||||
.\"
|
||||
.\" $Id$
|
||||
.\"
|
||||
.Dd March 4, 1997
|
||||
.Dt ED 4 i386
|
||||
.\" $Id$
|
||||
.Dd July 7, 1997
|
||||
.Dt WL 4 i386
|
||||
.Os FreeBSD
|
||||
.Sh NAME
|
||||
.Nm wl
|
||||
.Nd
|
||||
NCR Wavelan device driver
|
||||
.Nd T1 speed ISA/radio lan card
|
||||
.Sh SYNOPSIS
|
||||
.Cd "device wl0 at isa? port 0x300 net irq ? vector wlintr"
|
||||
.Cd "device wl0 at isa? port 0x300 net irq 5 vector wlintr"
|
||||
.Sh DESCRIPTION
|
||||
The
|
||||
.Nm
|
||||
driver provides support for the ISA bus AT&T/NCR/Lucent Wavelan
|
||||
ethernet-like wireless network cards.
|
||||
.Nm wl
|
||||
driver controls a radio lan card system made originally by
|
||||
NCR, then ATT, now Lucent. The system is spread-spectrum radio
|
||||
at around 915 mhz (or 2.4 ghz). With the supplied omni-directional antennae,
|
||||
about 400 feet (indoors, more outdoors) can be covered in circumference.
|
||||
This card can talk to the companion (wlp0) pccard. Speeds vary
|
||||
from 1 megabit to theoretically 2 megabits (roughly T1 in speed).
|
||||
.Pp
|
||||
All card configuration values except for the base address are read from the PSA
|
||||
(Parameter Storage Area) on the card, which can be configured with the
|
||||
.Xr wlconfig 8
|
||||
program. The base address for the card is set by dipswitches; see the manual
|
||||
supplied with the card for details.
|
||||
.Pp
|
||||
The
|
||||
.Nm
|
||||
driver does not accept any flags, however its behaviour can be controlled with
|
||||
a number of sysctl values:
|
||||
.Bl -hang -offset indent
|
||||
.It Em machdep.wl_xmit_delay
|
||||
On some systems, the driver is too fast for the Wavelan card. This value can be
|
||||
tweaked to slow the output rate of the driver at the expense of CPU performance.
|
||||
A value of 200 is believed to be acceptable for 486 systems, while 1000 is
|
||||
considered adequate for a 166MHz Pentium.
|
||||
.It Em machdep.wl_xmit_watch
|
||||
Causes extra debugging messages concerning the state of transmitted packets to
|
||||
be printed.
|
||||
.It Em machdep.wl_gather_snr
|
||||
SNR statistics are gathered. This is not completely implemented yet.
|
||||
.El
|
||||
.Sh DIAGNOSTICS
|
||||
The card has 3 fundmental hardware
|
||||
units, a so-called PSA or programmable storage area, a radio modem,
|
||||
and a ethernet lan controller. The latter component is the
|
||||
ancient (and not very honorable) Intel 82586 ethernet chip.
|
||||
Fundamentally it appears to the operating system as an ethernet system,
|
||||
and speaks IEEE MAC addresses. The radio modem simply translates
|
||||
ethernet packets to/from radio packets, that are either at 2.4 Ghz
|
||||
or 915 mhz depending on the radio modem. It supports a collision
|
||||
avoidance scheme. The lan controller
|
||||
supports promiscuous mode, broadcast, and multicasting
|
||||
(although there is a glitch
|
||||
in the latter). "It thinks it is ethernet".
|
||||
.Pp
|
||||
How it is used
|
||||
depends on the kind of antennae deployed with it. Point to point
|
||||
applications are possible as are ethernet-like lan use. The vendor
|
||||
ships an omni-directional antennae that works in the
|
||||
vicinity of 400 feet (indoors).
|
||||
Point to point antennae can be purchased that will go miles.
|
||||
.Sh SETUP
|
||||
The card can either be initialized with the vendor supplied DOS setup software.
|
||||
Typically minimally an IRQ, port, and Network ID must be supplied.
|
||||
Michael Smith's
|
||||
.Nm wlconfig(8)
|
||||
utility can now be used to do this work from
|
||||
the UNIX side. The card is "not" plug and play.
|
||||
The network id controls whether one set of cards can hear another.
|
||||
If different, cards will read physical packets, but they will be discarded
|
||||
by the radio modem.
|
||||
.Sh CONTROL
|
||||
In addition to the config utility, there are several sysctl
|
||||
switches that can be used to modify runtime parameters.
|
||||
The
|
||||
.Nm sysctl(8)
|
||||
variables are as follows:
|
||||
.Bl -diag
|
||||
.It "wl%d: PSA corrupt (invalid IRQ value)"
|
||||
The IRQ value in the PSA is illegal. The card will be configured not to generate
|
||||
any interrupts, ie. it will not work as a network device, however it will be possible
|
||||
to use
|
||||
.Xr wlconfig 8
|
||||
to reconfigure the IRQ to a sensible value.
|
||||
.It "wl%d init(): trouble resetting board."
|
||||
.It "wl%d bldcu(): iscp_busy timeout."
|
||||
.It "wl%d bldcu(): not ready after reset."
|
||||
The board failed to respond correctly during reset and configuration.
|
||||
.It "wl%d: busy?!"
|
||||
While attempting to queue a packet for transmission, the card claims to
|
||||
be busy although we were sure it wasn't.
|
||||
.It "wl%d read(): board is not running."
|
||||
An attempt was made to read a packet from the card, but we don't think that
|
||||
it's running. Should never happen.
|
||||
.It "wl%d read(): Invalid buffer"
|
||||
The i82586 on the card has corrupted its buffers.
|
||||
.It "wl%d read(): hwrst trouble."
|
||||
Having failed to meet expectations, the i82586 on the card is refusing to reset.
|
||||
.It "wl%d read(): No mbuf nth"
|
||||
An error occurred allocating mbufs to copy a received packet to.
|
||||
.It "wl%d ioctl(): board is not running"
|
||||
The interface was marked as DOWN but still RUNNING. Harmless.
|
||||
.It "wlack() int bits disappeared : %04x != int_type %04x"
|
||||
While acknowledging an interrupt, some of the interrupt sources on the i83586
|
||||
went away.
|
||||
.It "wl%d: xmt intr but not busy, CU %04x"
|
||||
A transmit interrupt occurred but we are not expecting one.
|
||||
.It "wl%d: xmt intr but ac_status == 0"
|
||||
A transmit interrupt occurred but there is no status to indicate that one should
|
||||
have been generated.
|
||||
.It "wl%d: xmt aborted"
|
||||
The transmission of a packet was aborted.
|
||||
.It "wl%d: no CTS"
|
||||
A packet was not transmitted because the CTS input to the i82586 went away.
|
||||
.It "wl%d: DMA underrun"
|
||||
The DMA hardware on the Wavelan card failed, causing an outgoing packet to be
|
||||
lost.
|
||||
.It "wl%d: xmt deferred"
|
||||
Transmission of a packet was deferred to avoid other traffic.
|
||||
.It "wl%d: heart beat"
|
||||
The i82586 thought it heard a heartbeat signal. This usually means that there's
|
||||
someone else transmitting at the same time.
|
||||
.It "wl%d: too many collisions"
|
||||
A packet was aborted because too many collisions occurred trying to send it.
|
||||
.It "wl%d RCV: !OK %x"
|
||||
A receive operation did not complete OK.
|
||||
.It "wl%d: slow accepting xmit"
|
||||
The i82586 on the card is taking a long time to accept a transmit command.
|
||||
.It "wl%d: i82586 Self Test failed!"
|
||||
Diagnostics on the i82586 failed.
|
||||
.It "wl%d: %s failed; status = %d, inw = %x, outw = %x"
|
||||
A command to the i82586 failed.
|
||||
.It "wl%d wlack(): board not accepting command."
|
||||
The card is not responding to an interrupt acknowledge.
|
||||
.It "machdep.wl_xmit_delay <useconds>"
|
||||
This variable will cause the driver to insert a delay on transmit.
|
||||
250 is the default. The delay should probably be a bit longer
|
||||
on faster cpus and less on slower cpus. It exists because the 82586
|
||||
was not designed to work with pentium-speed cpu systems and if overdriven
|
||||
will have copious xmit side errors.
|
||||
.It machdep.wl_ignore_nwid <0 | 1>
|
||||
This switch defaults to 0; i.e., the nwid is not ignored. It can
|
||||
be set to 1 to cause the nwid to not be used. This may be useful
|
||||
when the device is in promiscuous mode as one can watch for all
|
||||
packets and ignore nwid differences.
|
||||
.It machdep.wl_xmit_watch <milliseconds>
|
||||
This switch is not currently useful.
|
||||
.It machdep.wl_gather_snr <milliseconds>
|
||||
This switch is not currently useful.
|
||||
.Pp
|
||||
There is also a signal strength cache in the driver. It may be interrogated
|
||||
with
|
||||
.Nm wlconfig(8) .
|
||||
Incoming packets
|
||||
are checked for certain hardware radio-modem values including signal
|
||||
strength, silence, and quality, which range fro 0..63, 0..63, and 0..15
|
||||
respectively. Thus one can read out signal strenth values to see
|
||||
how close/far peer nodes are. The signal strength cache is indexed by
|
||||
sender MAC address.
|
||||
There are two sysctls that change how it filters packets. Both are on
|
||||
by default.
|
||||
.It machdep.wl_wlcache_mcastonly <0 | 1>
|
||||
By default this switch is on. It forces the cache to filter out
|
||||
unicast packets. Only broadcast or multicast packets are accepted.
|
||||
.It machdep.wl_wlcache_iponly <0 | 1>
|
||||
By default this switch is on. It forces the driver to discard non-IP
|
||||
packets and also stores the IP src address. ARP packets are ignored,
|
||||
as are any other network protocol barring IPv4 packets.
|
||||
.El
|
||||
.Sh CAVEATS
|
||||
The cards are known to lose transmit interrupts under heavy load. Ajust the
|
||||
machdep.wl_xmit_delay parameter to avoid this.
|
||||
.Pp
|
||||
On one test link, packet loss was significant enough to causes major lagging.
|
||||
.Sh BUGS
|
||||
The
|
||||
.Nm
|
||||
driver is very old and has been hacked on by many people. The i82586 is an
|
||||
odd and funky device. The two are not necessarily a good mix.
|
||||
The 82586 has numerous defects. It may experience transmit-side
|
||||
errors when modern faster cpus send packets at it faster than it can handle.
|
||||
The driver (and probably the chip) does not support an all multicast mode.
|
||||
As a result, it can be used with applications like mrouted(8), but
|
||||
it must go into promiscuous mode for that to work. The driver
|
||||
is slow to change modes from "normal" to promiscuous mode, presumably
|
||||
due to delays in the configuration code.
|
||||
.Sh SEE ALSO
|
||||
.Xr wlconfig 8 ,
|
||||
.Xr netintro 4 ,
|
||||
.Xr ifconfig 8
|
||||
.Xr sysctl 8 ,
|
||||
.Xr http://www.wavelan.com
|
||||
.Sh HISTORY
|
||||
The
|
||||
.Nm
|
||||
device driver was written/ported for Mach by Anders Klemets, ported to BSD/OS
|
||||
by Richard Morris, to FreeBSD 2.1 by Jim Binkley and to FreeBSD 2.2 by Michael
|
||||
Smith. It has not appeared in any FreeBSD releases yet.
|
||||
|
||||
.Nm wl
|
||||
driver was written by Anders Klemets (thousands of years ago?) and
|
||||
appears to be based on an even older Intel 82586 driver. The 82586
|
||||
controller was one of the first (if not the first?) integrated lan
|
||||
controller on the block. That does not mean it was the best either.
|
||||
Anders ported and or created a driver for the ISA wavelan and PCCARD
|
||||
wavelan system too (wlp). Robert T. Morris, Jr. ported the Mach
|
||||
drivers to BSDI. Jim Binkley ported them to FreeBSD 2.1. Michael
|
||||
Smith ported the wl driver only to 2.2.2. Jim and Michael have been
|
||||
maintaining them. The current state of the driver is NOT ANYONE'S
|
||||
FAULT. Thanks to Bernie Doehner and Robert Buaas for contributions.
|
||||
.Sh AUTHOR
|
||||
Too numerous to mention. See above.
|
||||
|
|
|
|||
Loading…
Reference in New Issue