xref: /freebsd/share/man/man4/dc.4 (revision dce6e6518b85561495cff38a3074a69d29d58a55)
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.\" $FreeBSD$
32.\"
33.Dd November 20, 1999
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, 98725, 98727 and 98732
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
64Xircom X3201 (cardbus only)
65.It
66Conexant LANfinity RS7112 (miniPCI)
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 DE500-BA on DEC Alpha workstations (21143, non-MII)
109.It
110Built in 10Mbps only ethernet on Compaq Presario 7900 series
111desktops (21143, non-MII)
112.It
113Built in ethernet on LinkSys EtherFast 10/100 Instant GigaDrive (DM9102, MII)
114.It
115Kingston KNE100TX (21143, MII)
116.It
117D-Link DFE-570TX (21143, MII, quad port)
118.It
119NDC SOHOware SFA110A (98713A)
120.It
121NDC SOHOware SFA110A Rev B4 (98715AEC-C)
122.It
123SVEC PN102-TX (98713)
124.It
125CNet Pro120A (98715A or 98713A) and CNet Pro120B (98715)
126.It
127Compex RL100-TX (98713 or 98713A)
128.It
129LinkSys LNE100TX (PNIC 82c168, 82c169)
130.It
131NetGear FA310-TX Rev. D1, D2 or D3 (PNIC 82c169)
132.It
133Matrox FastNIC 10/100 (PNIC 82c168, 82c169)
134.It
135Kingston KNE110TX (PNIC 82c169)
136.It
137LinkSys LNE100TX v2.0 (PNIC II 82c115)
138.It
139Jaton XpressNet (Davicom DM9102)
140.It
141Alfa Inc GFC2204 (ASIX AX88140A)
142.It
143CNet Pro110B (ASIX AX88140A)
144.It
145LinkSys LNE100TX v4.0/4.1 (ADMtek AN985 Centaur-P)
146.It
1473Com OfficeConnect 10/100B (ADMtek AN985 Centaur-P)
148.El
149.Pp
150The
151.Nm
152driver supports the following media types:
153.Pp
154.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
155.It autoselect
156Enable autoselection of the media type and options.
157The user can manually override
158the autoselected mode by adding media options to the
159.Pa /etc/rc.conf
160file.
161.Pp
162Note: the built-in NWAY autonegotiation on the original PNIC 82c168
163chip is horribly broken and is not supported by the
164.Nm
165driver at this time (see the
166.Sx BUGS
167section for details).
168The original 82c168 appears
169on very early revisions of the LinkSys LNE100TX and Matrox FastNIC.
170.It 10baseT/UTP
171Set 10Mbps operation.
172The
173.Ar mediaopt
174option can also be used to enable
175.Ar full-duplex
176operation.
177Not specifying
178.Ar full duplex
179implies
180.Ar half-duplex
181mode.
182.It 100baseTX
183Set 100Mbps (fast ethernet) operation.
184The
185.Ar mediaopt
186option can also be used to enable
187.Ar full-duplex
188operation.
189Not specifying
190.Ar full duplex
191implies
192.Ar half-duplex
193mode.
194.El
195.Pp
196The
197.Nm
198driver supports the following media options:
199.Pp
200.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
201.It full-duplex
202Force full duplex operation.
203The interface will operate in
204half duplex mode if this media option is not specified.
205.El
206.Pp
207Note that the 100baseTX media type may not be available on certain
208Intel 21143 adapters which support 10mbps media attachments only.
209For more information on configuring this device, see
210.Xr ifconfig 8 .
211.Sh DIAGNOSTICS
212.Bl -diag
213.It "dc%d: couldn't map ports/memory"
214A fatal initialization error has occurred.
215.It "dc%d: couldn't map interrupt"
216A fatal initialization error has occurred.
217.It "dc%d: watchdog timeout"
218A packet was queued for transmission and a transmit command was
219issued, but the device failed to acknowledge the transmission
220before a timeout expired.
221This can happen if the device is unable
222to deliver interrupts for some reason, of if there is a problem with
223the network connection (cable or network equipment) that results in a loss
224of link.
225.It "dc%d: no memory for rx list"
226The driver failed to allocate an mbuf for the receiver ring.
227.It "dc%d: TX underrun -- increasing TX threshold"
228The device generated a transmit underrun error while attempting to
229DMA and transmit a packet.
230This happens if the host is not able to
231DMA the packet data into the NIC's FIFO fast enough.
232The driver
233will dynamically increase the transmit start threshold so that
234more data must be DMAed into the FIFO before the NIC will start
235transmitting it onto the wire.
236.It "dc%d: TX underrun -- using store and forward mode"
237The device continued to generate transmit underruns even after all
238possible transmit start threshold settings had been tried, so the
239driver programmed the chip for store and forward mode.
240In this mode,
241the NIC will not begin transmission until the entire packet has been
242transfered into its FIFO memory.
243.It "dc%d: chip is in D3 power state -- setting to D0"
244This message applies only to adapters which support power
245management.
246Some operating systems place the controller in low power
247mode when shutting down, and some PCI BIOSes fail to bring the chip
248out of this state before configuring it.
249The controller loses all of
250its PCI configuration in the D3 state, so if the BIOS does not set
251it back to full power mode in time, it won't be able to configure it
252correctly.
253The driver tries to detect this condition and bring
254the adapter back to the D0 (full power) state, but this may not be
255enough to return the driver to a fully operational condition.
256If
257you see this message at boot time and the driver fails to attach
258the device as a network interface, you will have to perform a second
259warm boot to have the device properly configured.
260.Pp
261Note that this condition only occurs when warm booting from another
262operating system.
263If you power down your system prior to booting
264.Fx ,
265the card should be configured correctly.
266.El
267.Sh SEE ALSO
268.Xr arp 4 ,
269.Xr miibus 4 ,
270.Xr netintro 4 ,
271.Xr ng_ether 4 ,
272.Xr ifconfig 8
273.Rs
274.%T ADMtek AL981, AL983 and AL985 data sheets
275.%O http://www.admtek.com.tw
276.Re
277.Rs
278.%T ASIX Electronics AX88140A and AX88141 data sheets
279.%O http://www.asix.com.tw
280.Re
281.Rs
282.%T Davicom DM9102 data sheet
283.%O http://www.davicom8.com
284.Re
285.Rs
286.%T Intel 21143 Hardware Reference Manual
287.%O http://developer.intel.com
288.Re
289.Rs
290.%T Macronix 98713/A, 98715/A and 98725 data sheets
291.%O http://www.macronix.com
292.Re
293.Rs
294.%T Macronix 98713/A and 98715/A app notes
295.%O http://www.macronix.com
296.Re
297.Sh HISTORY
298The
299.Nm
300device driver first appeared in
301.Fx 4.0 .
302.Sh AUTHORS
303The
304.Nm
305driver was written by
306.An Bill Paul Aq wpaul@ee.columbia.edu .
307.Sh BUGS
308The Macronix application notes claim that in order to put the
309chips in normal operation, the driver must write a certain magic
310number into the CSR16 register.
311The numbers are documented in
312the app notes, but the exact meaning of the bits is not.
313.Pp
314The 98713A seems to have a problem with 10Mbps full duplex mode.
315The transmitter works but the receiver tends to produce many
316unexplained errors leading to very poor overall performance.
317The
31898715A does not exhibit this problem.
319All other modes on the
32098713A seem to work correctly.
321.Pp
322The original 82c168 PNIC chip has built in NWAY support which is
323used on certain early LinkSys LNE100TX and Matrox FastNIC cards,
324however it is horribly broken and difficult to use reliably.
325Consequently, autonegotiation is not currently supported for this
326chipset: the driver defaults the NIC to 10baseT half duplex, and it's
327up to the operator to manually select a different mode if necessary.
328(Later cards use an external MII transceiver to implement NWAY
329autonegotiation and work correctly.)
330.Pp
331The
332.Nm
333driver programs 82c168 and 82c169 PNIC chips to use the store and
334forward setting for the transmit start threshold by default.
335This
336is to work around problems with some NIC/PCI bus combinations where
337the PNIC can transmit corrupt frames when operating at 100Mbps,
338probably due to PCI DMA burst transfer errors.
339.Pp
340The 82c168 and 82c169 PNIC chips also have a receiver bug that
341sometimes manifests during periods of heavy receive and transmit
342activity, where the chip will improperly DMA received frames to
343the host.
344The chips appear to upload several kilobytes of garbage
345data along with the received frame data, dirtying several RX buffers
346instead of just the expected one.
347The
348.Nm
349driver detects this condition and will salvage the frame; however,
350it incurs a serious performance penalty in the process.
351.Pp
352The PNIC chips also sometimes generate a transmit underrun error when
353the driver attempts to download the receiver filter setup frame, which
354can result in the receive filter being incorrectly programmed.
355The
356.Nm
357driver will watch for this condition and requeue the setup frame until
358it is transfered successfully.
359.Pp
360The ADMtek AL981 chip (and possibly the AN985 as well) has been observed
361to sometimes wedge on transmit: this appears to happen when the driver
362queues a sequence of frames which cause it to wrap from the end of the
363transmit descriptor ring back to the beginning.
364The
365.Nm
366driver attempts to avoid this condition by not queuing any frames past
367the end of the transmit ring during a single invocation of the
368.Fn dc_start
369routine.
370This workaround has a negligible impact on transmit performance.
371