xref: /dragonfly/share/man/man4/dc.4 (revision 9c600e7d)
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 1.2 2003/06/17 04:36:58 dillon 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 minimun.
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 DE500-BA on DEC Alpha workstations (21143, non-MII)
108.It
109Built in 10Mbps only ethernet on Compaq Presario 7900 series
110desktops (21143, non-MII)
111.It
112Built in ethernet on LinkSys EtherFast 10/100 Instant GigaDrive (DM9102, MII)
113.It
114Kingston KNE100TX (21143, MII)
115.It
116D-Link DFE-570TX (21143, MII, quad port)
117.It
118NDC SOHOware SFA110 (98713A)
119.It
120SVEC PN102-TX (98713)
121.It
122CNet Pro120A (98715A or 98713A) and CNet Pro120B (98715)
123.It
124Compex RL100-TX (98713 or 98713A)
125.It
126LinkSys LNE100TX (PNIC 82c168, 82c169)
127.It
128NetGear FA310-TX Rev. D1, D2 or D3 (PNIC 82c169)
129.It
130Matrox FastNIC 10/100 (PNIC 82c168, 82c169)
131.It
132Kingston KNE110TX (PNIC 82c169)
133.It
134LinkSys LNE100TX v2.0 (PNIC II 82c115)
135.It
136Jaton XpressNet (Davicom DM9102)
137.It
138Alfa Inc GFC2204 (ASIX AX88140A)
139.It
140CNet Pro110B (ASIX AX88140A)
141.El
142.Pp
143The
144.Nm
145driver supports the following media types:
146.Pp
147.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
148.It autoselect
149Enable autoselection of the media type and options.
150The user can manually override
151the autoselected mode by adding media options to the
152.Pa /etc/rc.conf
153file.
154.Pp
155Note: the built-in NWAY autonegotiation on the original PNIC 82c168
156chip is horribly broken and is not supported by the
157.Nm
158driver at this time (see the
159.Sx BUGS
160section for details).
161The original 82c168 appears
162on very early revisions of the LinkSys LNE100TX and Matrox FastNIC.
163.It 10baseT/UTP
164Set 10Mbps operation.
165The
166.Ar mediaopt
167option can also be used to enable
168.Ar full-duplex
169operation.
170Not specifying
171.Ar full duplex
172implies
173.Ar half-duplex
174mode.
175.It 100baseTX
176Set 100Mbps (fast ethernet) operation.
177The
178.Ar mediaopt
179option can also be used to enable
180.Ar full-duplex
181operation.
182Not specifying
183.Ar full duplex
184implies
185.Ar half-duplex
186mode.
187.El
188.Pp
189The
190.Nm
191driver supports the following media options:
192.Pp
193.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
194.It full-duplex
195Force full duplex operation.
196The interface will operate in
197half duplex mode if this media option is not specified.
198.El
199.Pp
200Note that the 100baseTX media type may not be available on certain
201Intel 21143 adapters which support 10mbps media attachments only.
202For more information on configuring this device, see
203.Xr ifconfig 8 .
204.Sh DIAGNOSTICS
205.Bl -diag
206.It "dc%d: couldn't map ports/memory"
207A fatal initialization error has occurred.
208.It "dc%d: couldn't map interrupt"
209A fatal initialization error has occurred.
210.It "dc%d: watchdog timeout"
211A packet was queued for transmission and a transmit command was
212issued, but the device failed to acknowledge the transmission
213before a timeout expired.
214This can happen if the device is unable
215to deliver interrupts for some reason, of if there is a problem with
216the network connection (cable or network equipment) that results in a loss
217of link.
218.It "dc%d: no memory for rx list"
219The driver failed to allocate an mbuf for the receiver ring.
220.It "dc%d: TX underrun -- increasing TX threshold"
221The device generated a transmit underrun error while attempting to
222DMA and transmit a packet.
223This happens if the host is not able to
224DMA the packet data into the NIC's FIFO fast enough.
225The driver
226will dynamically increase the transmit start threshold so that
227more data must be DMAed into the FIFO before the NIC will start
228transmitting it onto the wire.
229.It "dc%d: TX underrun -- using store and forward mode"
230The device continued to generate transmit underruns even after all
231possible transmit start threshold settings had been tried, so the
232driver programmed the chip for store and forward mode.
233In this mode,
234the NIC will not begin transmission until the entire packet has been
235transfered into its FIFO memory.
236.It "dc%d: chip is in D3 power state -- setting to D0"
237This message applies only to adapters which support power
238management.
239Some operating systems place the controller in low power
240mode when shutting down, and some PCI BIOSes fail to bring the chip
241out of this state before configuring it.
242The controller loses all of
243its PCI configuration in the D3 state, so if the BIOS does not set
244it back to full power mode in time, it won't be able to configure it
245correctly.
246The driver tries to detect this condition and bring
247the adapter back to the D0 (full power) state, but this may not be
248enough to return the driver to a fully operational condition.
249If
250you see this message at boot time and the driver fails to attach
251the device as a network interface, you will have to perform a second
252warm boot to have the device properly configured.
253.Pp
254Note that this condition only occurs when warm booting from another
255operating system.
256If you power down your system prior to booting
257.Fx ,
258the card should be configured correctly.
259.El
260.Sh SEE ALSO
261.Xr arp 4 ,
262.Xr miibus 4 ,
263.Xr netintro 4 ,
264.Xr ng_ether 4 ,
265.Xr ifconfig 8
266.Rs
267.%T ADMtek AL981, AL983 and AL985 data sheets
268.%O http://www.admtek.com.tw
269.Re
270.Rs
271.%T ASIX Electronics AX88140A and AX88141 data sheets
272.%O http://www.asix.com.tw
273.Re
274.Rs
275.%T Davicom DM9102 data sheet
276.%O http://www.davicom8.com
277.Re
278.Rs
279.%T Intel 21143 Hardware Reference Manual
280.%O http://developer.intel.com
281.Re
282.Rs
283.%T Macronix 98713/A, 98715/A and 98725 data sheets
284.%O http://www.macronix.com
285.Re
286.Rs
287.%T Macronix 98713/A and 98715/A app notes
288.%O http://www.macronix.com
289.Re
290.Sh HISTORY
291The
292.Nm
293device driver first appeared in
294.Fx 4.0 .
295.Sh AUTHORS
296The
297.Nm
298driver was written by
299.An Bill Paul Aq wpaul@ee.columbia.edu .
300.Sh BUGS
301The Macronix application notes claim that in order to put the
302chips in normal operation, the driver must write a certain magic
303number into the CSR16 register.
304The numbers are documented in
305the app notes, but the exact meaning of the bits is not.
306.Pp
307The 98713A seems to have a problem with 10Mbps full duplex mode.
308The transmitter works but the receiver tends to produce many
309unexplained errors leading to very poor overall performance.
310The
31198715A does not exhibit this problem.
312All other modes on the
31398713A seem to work correctly.
314.Pp
315The original 82c168 PNIC chip has built in NWAY support which is
316used on certain early LinkSys LNE100TX and Matrox FastNIC cards,
317however it is horribly broken and difficult to use reliably.
318Consequently, autonegotiation is not currently supported for this
319chipset: the driver defaults the NIC to 10baseT half duplex, and it's
320up to the operator to manually select a different mode if necessary.
321(Later cards use an external MII transceiver to implement NWAY
322autonegotiation and work correctly.)
323.Pp
324The
325.Nm
326driver programs 82c168 and 82c169 PNIC chips to use the store and
327forward setting for the transmit start threshold by default.
328This
329is to work around problems with some NIC/PCI bus combinations where
330the PNIC can transmit corrupt frames when operating at 100Mbps,
331probably due to PCI DMA burst transfer errors.
332.Pp
333The 82c168 and 82c169 PNIC chips also have a receiver bug that
334sometimes manifests during periods of heavy receive and transmit
335activity, where the chip will improperly DMA received frames to
336the host.
337The chips appear to upload several kilobytes of garbage
338data along with the received frame data, dirtying several RX buffers
339instead of just the expected one.
340The
341.Nm
342driver detects this condition and will salvage the frame; however,
343it incurs a serious performance penalty in the process.
344.Pp
345The PNIC chips also sometimes generate a transmit underrun error when
346the driver attempts to download the receiver filter setup frame, which
347can result in the receive filter being incorrectly programmed.
348The
349.Nm
350driver will watch for this condition and requeue the setup frame until
351it is transfered successfully.
352.Pp
353The ADMtek AL981 chip (and possibly the AN985 as well) has been observed
354to sometimes wedge on transmit: this appears to happen when the driver
355queues a sequence of frames which cause it to wrap from the end of the
356transmit descriptor ring back to the beginning.
357The
358.Nm
359driver attempts to avoid this condition by not queing any frames past
360the end of the transmit ring during a single invocation of the
361.Fn dc_start
362routine.
363This workaround has a negligible impact on transmit performance.
364