xref: /dragonfly/share/man/man4/dc.4 (revision bbb35c81)
1.\" Copyright (c) 1997, 1998, 1999
2.\"	Bill Paul <wpaul@ee.columbia.edu>. All rights reserved.
3.\"
4.\" Redistribution and use in source and binary forms, with or without
5.\" modification, are permitted provided that the following conditions
6.\" are met:
7.\" 1. Redistributions of source code must retain the above copyright
8.\"    notice, this list of conditions and the following disclaimer.
9.\" 2. Redistributions in binary form must reproduce the above copyright
10.\"    notice, this list of conditions and the following disclaimer in the
11.\"    documentation and/or other materials provided with the distribution.
12.\" 3. All advertising materials mentioning features or use of this software
13.\"    must display the following acknowledgement:
14.\"	This product includes software developed by Bill Paul.
15.\" 4. Neither the name of the author nor the names of any co-contributors
16.\"    may be used to endorse or promote products derived from this software
17.\"   without specific prior written permission.
18.\"
19.\" THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND
20.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22.\" ARE DISCLAIMED.  IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD
23.\" BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24.\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25.\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26.\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27.\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28.\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
29.\" THE POSSIBILITY OF SUCH DAMAGE.
30.\"
31.\" $FreeBSD: src/share/man/man4/dc.4,v 1.6.2.9 2003/02/17 21:20:39 trhodes Exp $
32.\"
33.Dd November 14, 2010
34.Dt DC 4
35.Os
36.Sh NAME
37.Nm dc
38.Nd "DEC/Intel 21143 and clone 10/100 ethernet driver"
39.Sh SYNOPSIS
40.Cd "device miibus"
41.Cd "device dc"
42.Sh DESCRIPTION
43The
44.Nm
45driver provides support for several PCI fast ethernet adapters and
46embedded controllers based on the following chipsets:
47.Pp
48.Bl -bullet -compact -offset indent
49.It
50DEC/Intel 21143
51.It
52Macronix 98713, 98713A, 98715, 98715A and 98725
53.It
54Davicom DM9100, DM9102 and DM9102A
55.It
56ASIX Electronics AX88140A and AX88141
57.It
58ADMtek AL981 Comet and AN985 Centaur
59.It
60Lite-On 82c168 and 82c169 PNIC
61.It
62Lite-On/Macronix 82c115 PNIC II
63.It
64Conexant LANfinity RS7112 (miniPCI)
65.It
66Abocom FE2500
67.El
68.Pp
69All of these chips have the same general register layout, DMA
70descriptor format and method of operation.
71All of the clone chips
72are based on the 21143 design with various modifications.
73The
7421143 itself has support for 10baseT, BNC, AUI, MII and symbol
75media attachments, 10 and 100Mbps speeds in full or half duplex,
76built in NWAY autonegotiation and wake on LAN.
77The 21143 also
78offers several receive filter programming options including
79perfect filtering, inverse perfect filtering and hash table
80filtering.
81.Pp
82Some clone chips duplicate the 21143 fairly closely while others
83only maintain superficial similarities.
84Some support only MII
85media attachments.
86Others use different receiver filter programming
87mechanisms.
88At least one supports only chained DMA descriptors
89(most support both chained descriptors and contiguously allocated
90fixed size rings). Some chips (especially the PNIC) also have
91peculiar bugs.
92The
93.Nm
94driver does its best to provide generalized support for all
95of these chipsets in order to keep special case code to a minimum.
96.Pp
97These chips are used by many vendors which makes it
98difficult to provide a complete list of all supported cards.
99The
100following NICs are known to work with the
101.Nm
102driver at this time:
103.Pp
104.Bl -bullet -compact -offset indent
105.It
106Digital DE500-BA 10/100 (21143, non-MII)
107.It
108Built in 10Mbps only ethernet on Compaq Presario 7900 series
109desktops (21143, non-MII)
110.It
111Built in ethernet on LinkSys EtherFast 10/100 Instant GigaDrive (DM9102, MII)
112.It
113Kingston KNE100TX (21143, MII)
114.It
115D-Link DFE-570TX (21143, MII, quad port)
116.It
117NDC SOHOware SFA110 (98713A)
118.It
119SVEC PN102-TX (98713)
120.It
121CNet Pro120A (98715A or 98713A) and CNet Pro120B (98715)
122.It
123Compex RL100-TX (98713 or 98713A)
124.It
125LinkSys LNE100TX (PNIC 82c168, 82c169)
126.It
127NetGear FA310-TX Rev. D1, D2 or D3 (PNIC 82c169)
128.It
129Matrox FastNIC 10/100 (PNIC 82c168, 82c169)
130.It
131Kingston KNE110TX (PNIC 82c169)
132.It
133LinkSys LNE100TX v2.0 (PNIC II 82c115)
134.It
135Jaton XpressNet (Davicom DM9102)
136.It
137Alfa Inc GFC2204 (ASIX AX88140A)
138.It
139CNet Pro110B (ASIX AX88140A)
140.El
141.Pp
142The
143.Nm
144driver supports the following media types:
145.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
146.It autoselect
147Enable autoselection of the media type and options.
148The user can manually override
149the autoselected mode by adding media options to the
150.Pa /etc/rc.conf
151file.
152.Pp
153Note: the built-in NWAY autonegotiation on the original PNIC 82c168
154chip is horribly broken and is not supported by the
155.Nm
156driver at this time (see the
157.Sx BUGS
158section for details).
159The original 82c168 appears
160on very early revisions of the LinkSys LNE100TX and Matrox FastNIC.
161.It 10baseT/UTP
162Set 10Mbps operation.
163The
164.Ar mediaopt
165option can also be used to enable
166.Ar full-duplex
167operation.
168Not specifying
169.Ar full duplex
170implies
171.Ar half-duplex
172mode.
173.It 100baseTX
174Set 100Mbps (fast ethernet) operation.
175The
176.Ar mediaopt
177option can also be used to enable
178.Ar full-duplex
179operation.
180Not specifying
181.Ar full duplex
182implies
183.Ar half-duplex
184mode.
185.El
186.Pp
187The
188.Nm
189driver supports the following media options:
190.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
191.It full-duplex
192Force full duplex operation.
193The interface will operate in
194half duplex mode if this media option is not specified.
195.El
196.Pp
197Note that the 100baseTX media type may not be available on certain
198Intel 21143 adapters which support 10mbps media attachments only.
199.Pp
200For more information on configuring this device, see
201.Xr ifconfig 8 .
202The
203.Nm
204driver supports
205.Xr polling 4 .
206.Sh DIAGNOSTICS
207.Bl -diag
208.It "dc%d: couldn't map ports/memory"
209A fatal initialization error has occurred.
210.It "dc%d: couldn't map interrupt"
211A fatal initialization error has occurred.
212.It "dc%d: watchdog timeout"
213A packet was queued for transmission and a transmit command was
214issued, but the device failed to acknowledge the transmission
215before a timeout expired.
216This can happen if the device is unable
217to deliver interrupts for some reason, of if there is a problem with
218the network connection (cable or network equipment) that results in a loss
219of link.
220.It "dc%d: no memory for rx list"
221The driver failed to allocate an mbuf for the receiver ring.
222.It "dc%d: TX underrun -- increasing TX threshold"
223The device generated a transmit underrun error while attempting to
224DMA and transmit a packet.
225This happens if the host is not able to
226DMA the packet data into the NIC's FIFO fast enough.
227The driver
228will dynamically increase the transmit start threshold so that
229more data must be DMAed into the FIFO before the NIC will start
230transmitting it onto the wire.
231.It "dc%d: TX underrun -- using store and forward mode"
232The device continued to generate transmit underruns even after all
233possible transmit start threshold settings had been tried, so the
234driver programmed the chip for store and forward mode.
235In this mode,
236the NIC will not begin transmission until the entire packet has been
237transferred into its FIFO memory.
238.It "dc%d: chip is in D3 power state -- setting to D0"
239This message applies only to adapters which support power
240management.
241Some operating systems place the controller in low power
242mode when shutting down, and some PCI BIOSes fail to bring the chip
243out of this state before configuring it.
244The controller loses all of
245its PCI configuration in the D3 state, so if the BIOS does not set
246it back to full power mode in time, it won't be able to configure it
247correctly.
248The driver tries to detect this condition and bring
249the adapter back to the D0 (full power) state, but this may not be
250enough to return the driver to a fully operational condition.
251If
252you see this message at boot time and the driver fails to attach
253the device as a network interface, you will have to perform a second
254warm boot to have the device properly configured.
255.Pp
256Note that this condition only occurs when warm booting from another
257operating system.
258If you power down your system prior to booting
259.Dx ,
260the card should be configured correctly.
261.El
262.Sh SEE ALSO
263.Xr arp 4 ,
264.Xr ifmedia 4 ,
265.Xr miibus 4 ,
266.Xr netintro 4 ,
267.Xr ng_ether 4 ,
268.Xr polling 4 ,
269.Xr ifconfig 8
270.Rs
271.%T ADMtek AL981, AL983 and AL985 data sheets
272.%U http://www.admtek.com.tw
273.Re
274.Rs
275.%T ASIX Electronics AX88140A and AX88141 data sheets
276.%U http://www.asix.com.tw
277.Re
278.Rs
279.%T Davicom DM9102 data sheet
280.%U http://www.davicom.com.tw
281.Re
282.Rs
283.%T Intel 21143 Hardware Reference Manual
284.%U http://developer.intel.com
285.Re
286.Rs
287.%T Macronix 98713/A, 98715/A and 98725 data sheets
288.%U http://www.macronix.com
289.Re
290.Rs
291.%T Macronix 98713/A and 98715/A app notes
292.%U http://www.macronix.com
293.Re
294.Sh HISTORY
295The
296.Nm
297device driver first appeared in
298.Fx 4.0 .
299.Sh AUTHORS
300The
301.Nm
302driver was written by
303.An Bill Paul Aq Mt wpaul@ee.columbia.edu .
304.Sh BUGS
305The Macronix application notes claim that in order to put the
306chips in normal operation, the driver must write a certain magic
307number into the CSR16 register.
308The numbers are documented in
309the app notes, but the exact meaning of the bits is not.
310.Pp
311The 98713A seems to have a problem with 10Mbps full duplex mode.
312The transmitter works but the receiver tends to produce many
313unexplained errors leading to very poor overall performance.
314The
31598715A does not exhibit this problem.
316All other modes on the
31798713A seem to work correctly.
318.Pp
319The original 82c168 PNIC chip has built in NWAY support which is
320used on certain early LinkSys LNE100TX and Matrox FastNIC cards,
321however it is horribly broken and difficult to use reliably.
322Consequently, autonegotiation is not currently supported for this
323chipset: the driver defaults the NIC to 10baseT half duplex, and it's
324up to the operator to manually select a different mode if necessary.
325(Later cards use an external MII transceiver to implement NWAY
326autonegotiation and work correctly.)
327.Pp
328The
329.Nm
330driver programs 82c168 and 82c169 PNIC chips to use the store and
331forward setting for the transmit start threshold by default.
332This
333is to work around problems with some NIC/PCI bus combinations where
334the PNIC can transmit corrupt frames when operating at 100Mbps,
335probably due to PCI DMA burst transfer errors.
336.Pp
337The 82c168 and 82c169 PNIC chips also have a receiver bug that
338sometimes manifests during periods of heavy receive and transmit
339activity, where the chip will improperly DMA received frames to
340the host.
341The chips appear to upload several kilobytes of garbage
342data along with the received frame data, dirtying several RX buffers
343instead of just the expected one.
344The
345.Nm
346driver detects this condition and will salvage the frame; however,
347it incurs a serious performance penalty in the process.
348.Pp
349The PNIC chips also sometimes generate a transmit underrun error when
350the driver attempts to download the receiver filter setup frame, which
351can result in the receive filter being incorrectly programmed.
352The
353.Nm
354driver will watch for this condition and requeue the setup frame until
355it is transferred successfully.
356.Pp
357The ADMtek AL981 chip (and possibly the AN985 as well) has been observed
358to sometimes wedge on transmit: this appears to happen when the driver
359queues a sequence of frames which cause it to wrap from the end of the
360transmit descriptor ring back to the beginning.
361The
362.Nm
363driver attempts to avoid this condition by not queueing any frames past
364the end of the transmit ring during a single invocation of the
365.Fn dc_start
366routine.
367This workaround has a negligible impact on transmit performance.
368