xref: /dragonfly/share/man/man4/dc.4 (revision 8e1c6f81)
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.\" $DragonFly: src/share/man/man4/dc.4,v 1.12 2008/07/11 10:38:50 thomas Exp $
33.\"
34.Dd November 20, 1999
35.Dt DC 4
36.Os
37.Sh NAME
38.Nm dc
39.Nd "DEC/Intel 21143 and clone 10/100 ethernet driver"
40.Sh SYNOPSIS
41.Cd "device miibus"
42.Cd "device dc"
43.Sh DESCRIPTION
44The
45.Nm
46driver provides support for several PCI fast ethernet adapters and
47embedded controllers based on the following chipsets:
48.Pp
49.Bl -bullet -compact -offset indent
50.It
51DEC/Intel 21143
52.It
53Macronix 98713, 98713A, 98715, 98715A and 98725
54.It
55Davicom DM9100, DM9102 and DM9102A
56.It
57ASIX Electronics AX88140A and AX88141
58.It
59ADMtek AL981 Comet and AN985 Centaur
60.It
61Lite-On 82c168 and 82c169 PNIC
62.It
63Lite-On/Macronix 82c115 PNIC II
64.It
65Conexant LANfinity RS7112 (miniPCI)
66.El
67.Pp
68All of these chips have the same general register layout, DMA
69descriptor format and method of operation.
70All of the clone chips
71are based on the 21143 design with various modifications.
72The
7321143 itself has support for 10baseT, BNC, AUI, MII and symbol
74media attachments, 10 and 100Mbps speeds in full or half duplex,
75built in NWAY autonegotiation and wake on LAN.
76The 21143 also
77offers several receive filter programming options including
78perfect filtering, inverse perfect filtering and hash table
79filtering.
80.Pp
81Some clone chips duplicate the 21143 fairly closely while others
82only maintain superficial similarities.
83Some support only MII
84media attachments.
85Others use different receiver filter programming
86mechanisms.
87At least one supports only chained DMA descriptors
88(most support both chained descriptors and contiguously allocated
89fixed size rings). Some chips (especially the PNIC) also have
90peculiar bugs.
91The
92.Nm
93driver does its best to provide generalized support for all
94of these chipsets in order to keep special case code to a minimum.
95.Pp
96These chips are used by many vendors which makes it
97difficult to provide a complete list of all supported cards.
98The
99following NICs are known to work with the
100.Nm
101driver at this time:
102.Pp
103.Bl -bullet -compact -offset indent
104.It
105Digital DE500-BA 10/100 (21143, non-MII)
106.It
107Built in 10Mbps only ethernet on Compaq Presario 7900 series
108desktops (21143, non-MII)
109.It
110Built in ethernet on LinkSys EtherFast 10/100 Instant GigaDrive (DM9102, MII)
111.It
112Kingston KNE100TX (21143, MII)
113.It
114D-Link DFE-570TX (21143, MII, quad port)
115.It
116NDC SOHOware SFA110 (98713A)
117.It
118SVEC PN102-TX (98713)
119.It
120CNet Pro120A (98715A or 98713A) and CNet Pro120B (98715)
121.It
122Compex RL100-TX (98713 or 98713A)
123.It
124LinkSys LNE100TX (PNIC 82c168, 82c169)
125.It
126NetGear FA310-TX Rev. D1, D2 or D3 (PNIC 82c169)
127.It
128Matrox FastNIC 10/100 (PNIC 82c168, 82c169)
129.It
130Kingston KNE110TX (PNIC 82c169)
131.It
132LinkSys LNE100TX v2.0 (PNIC II 82c115)
133.It
134Jaton XpressNet (Davicom DM9102)
135.It
136Alfa Inc GFC2204 (ASIX AX88140A)
137.It
138CNet Pro110B (ASIX AX88140A)
139.El
140.Pp
141The
142.Nm
143driver supports the following media types:
144.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
145.It autoselect
146Enable autoselection of the media type and options.
147The user can manually override
148the autoselected mode by adding media options to the
149.Pa /etc/rc.conf
150file.
151.Pp
152Note: the built-in NWAY autonegotiation on the original PNIC 82c168
153chip is horribly broken and is not supported by the
154.Nm
155driver at this time (see the
156.Sx BUGS
157section for details).
158The original 82c168 appears
159on very early revisions of the LinkSys LNE100TX and Matrox FastNIC.
160.It 10baseT/UTP
161Set 10Mbps operation.
162The
163.Ar mediaopt
164option can also be used to enable
165.Ar full-duplex
166operation.
167Not specifying
168.Ar full duplex
169implies
170.Ar half-duplex
171mode.
172.It 100baseTX
173Set 100Mbps (fast ethernet) operation.
174The
175.Ar mediaopt
176option can also be used to enable
177.Ar full-duplex
178operation.
179Not specifying
180.Ar full duplex
181implies
182.Ar half-duplex
183mode.
184.El
185.Pp
186The
187.Nm
188driver supports the following media options:
189.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
190.It full-duplex
191Force full duplex operation.
192The interface will operate in
193half duplex mode if this media option is not specified.
194.El
195.Pp
196Note that the 100baseTX media type may not be available on certain
197Intel 21143 adapters which support 10mbps media attachments only.
198.Pp
199For more information on configuring this device, see
200.Xr ifconfig 8 .
201The
202.Nm
203driver supports
204.Xr polling 4 .
205.Sh DIAGNOSTICS
206.Bl -diag
207.It "dc%d: couldn't map ports/memory"
208A fatal initialization error has occurred.
209.It "dc%d: couldn't map interrupt"
210A fatal initialization error has occurred.
211.It "dc%d: watchdog timeout"
212A packet was queued for transmission and a transmit command was
213issued, but the device failed to acknowledge the transmission
214before a timeout expired.
215This can happen if the device is unable
216to deliver interrupts for some reason, of if there is a problem with
217the network connection (cable or network equipment) that results in a loss
218of link.
219.It "dc%d: no memory for rx list"
220The driver failed to allocate an mbuf for the receiver ring.
221.It "dc%d: TX underrun -- increasing TX threshold"
222The device generated a transmit underrun error while attempting to
223DMA and transmit a packet.
224This happens if the host is not able to
225DMA the packet data into the NIC's FIFO fast enough.
226The driver
227will dynamically increase the transmit start threshold so that
228more data must be DMAed into the FIFO before the NIC will start
229transmitting it onto the wire.
230.It "dc%d: TX underrun -- using store and forward mode"
231The device continued to generate transmit underruns even after all
232possible transmit start threshold settings had been tried, so the
233driver programmed the chip for store and forward mode.
234In this mode,
235the NIC will not begin transmission until the entire packet has been
236transferred into its FIFO memory.
237.It "dc%d: chip is in D3 power state -- setting to D0"
238This message applies only to adapters which support power
239management.
240Some operating systems place the controller in low power
241mode when shutting down, and some PCI BIOSes fail to bring the chip
242out of this state before configuring it.
243The controller loses all of
244its PCI configuration in the D3 state, so if the BIOS does not set
245it back to full power mode in time, it won't be able to configure it
246correctly.
247The driver tries to detect this condition and bring
248the adapter back to the D0 (full power) state, but this may not be
249enough to return the driver to a fully operational condition.
250If
251you see this message at boot time and the driver fails to attach
252the device as a network interface, you will have to perform a second
253warm boot to have the device properly configured.
254.Pp
255Note that this condition only occurs when warm booting from another
256operating system.
257If you power down your system prior to booting
258.Dx ,
259the card should be configured correctly.
260.El
261.Sh SEE ALSO
262.Xr arp 4 ,
263.Xr ifmedia 4 ,
264.Xr miibus 4 ,
265.Xr netintro 4 ,
266.Xr ng_ether 4 ,
267.Xr polling 4 ,
268.Xr ifconfig 8
269.Rs
270.%T ADMtek AL981, AL983 and AL985 data sheets
271.%O http://www.admtek.com.tw
272.Re
273.Rs
274.%T ASIX Electronics AX88140A and AX88141 data sheets
275.%O http://www.asix.com.tw
276.Re
277.Rs
278.%T Davicom DM9102 data sheet
279.%O http://www.davicom.com.tw
280.Re
281.Rs
282.%T Intel 21143 Hardware Reference Manual
283.%O http://developer.intel.com
284.Re
285.Rs
286.%T Macronix 98713/A, 98715/A and 98725 data sheets
287.%O http://www.macronix.com
288.Re
289.Rs
290.%T Macronix 98713/A and 98715/A app notes
291.%O http://www.macronix.com
292.Re
293.Sh HISTORY
294The
295.Nm
296device driver first appeared in
297.Fx 4.0 .
298.Sh AUTHORS
299The
300.Nm
301driver was written by
302.An Bill Paul Aq wpaul@ee.columbia.edu .
303.Sh BUGS
304The Macronix application notes claim that in order to put the
305chips in normal operation, the driver must write a certain magic
306number into the CSR16 register.
307The numbers are documented in
308the app notes, but the exact meaning of the bits is not.
309.Pp
310The 98713A seems to have a problem with 10Mbps full duplex mode.
311The transmitter works but the receiver tends to produce many
312unexplained errors leading to very poor overall performance.
313The
31498715A does not exhibit this problem.
315All other modes on the
31698713A seem to work correctly.
317.Pp
318The original 82c168 PNIC chip has built in NWAY support which is
319used on certain early LinkSys LNE100TX and Matrox FastNIC cards,
320however it is horribly broken and difficult to use reliably.
321Consequently, autonegotiation is not currently supported for this
322chipset: the driver defaults the NIC to 10baseT half duplex, and it's
323up to the operator to manually select a different mode if necessary.
324(Later cards use an external MII transceiver to implement NWAY
325autonegotiation and work correctly.)
326.Pp
327The
328.Nm
329driver programs 82c168 and 82c169 PNIC chips to use the store and
330forward setting for the transmit start threshold by default.
331This
332is to work around problems with some NIC/PCI bus combinations where
333the PNIC can transmit corrupt frames when operating at 100Mbps,
334probably due to PCI DMA burst transfer errors.
335.Pp
336The 82c168 and 82c169 PNIC chips also have a receiver bug that
337sometimes manifests during periods of heavy receive and transmit
338activity, where the chip will improperly DMA received frames to
339the host.
340The chips appear to upload several kilobytes of garbage
341data along with the received frame data, dirtying several RX buffers
342instead of just the expected one.
343The
344.Nm
345driver detects this condition and will salvage the frame; however,
346it incurs a serious performance penalty in the process.
347.Pp
348The PNIC chips also sometimes generate a transmit underrun error when
349the driver attempts to download the receiver filter setup frame, which
350can result in the receive filter being incorrectly programmed.
351The
352.Nm
353driver will watch for this condition and requeue the setup frame until
354it is transferred successfully.
355.Pp
356The ADMtek AL981 chip (and possibly the AN985 as well) has been observed
357to sometimes wedge on transmit: this appears to happen when the driver
358queues a sequence of frames which cause it to wrap from the end of the
359transmit descriptor ring back to the beginning.
360The
361.Nm
362driver attempts to avoid this condition by not queueing any frames past
363the end of the transmit ring during a single invocation of the
364.Fn dc_start
365routine.
366This workaround has a negligible impact on transmit performance.
367