xref: /freebsd/share/man/man4/dc.4 (revision b601c69bdbe8755d26570261d7fd4c02ee4eff74)
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 i386
35.Os FreeBSD
36.Sh NAME
37.Nm dc
38.Nd
39DEC/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, 98725, 98727 and 98732
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.El
65.Pp
66All of these chips have the same general register layout, DMA
67descriptor format and method of operation.
68All of the clone chips
69are based on the 21143 design with various modifications.
70The
7121143 itself has support for 10baseT, BNC, AUI, MII and symbol
72media attachments, 10 and 100Mbps speeds in full or half duplex,
73built in NWAY autonegotiation and wake on LAN.
74The 21143 also
75offers several receive filter programming options including
76perfect filtering, inverse perfect filtering and hash table
77filtering.
78.Pp
79Some clone chips duplicate the 21143 fairly closely while others
80only maintain superficial simularities.
81Some support only MII
82media attachments.
83Others use different receiver filter programming
84mechanisms.
85At least one supports only chained DMA descriptors
86(most support both chained descriptors and contiguously allocated
87fixed size rings). Some chips (especially the PNIC) also have
88peculiar bugs.
89The
90.Nm
91driver does its best to provide generalized support for all
92of these chipsets in order to keep special case code to a minimun.
93.Pp
94These chips are used by many vendors which makes it
95difficult provide a complete list of all supported cards.
96The
97following NICs are known to work with the
98.Nm
99driver at this time:
100.Pp
101.Bl -bullet -compact -offset indent
102.It
103Digital DE500-BA 10/100 (21143, non-MII)
104.It
105Built in DE500-BA on DEC Alpha workstations (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 SFA110A (98713A)
117.It
118NDC SOHOware SFA110A Rev B4 (98715AEC-C)
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: the chip will operate in any speed or duplex
159mode, however these must be set manually.
160The original 82c168 appears
161on very early revisions of the LinkSys LNE100TX and Matrox FastNIC.
162.It 10baseT/UTP
163Set 10Mbps operation.
164The
165.Ar mediaopt
166option can also be used to enable
167.Ar full-duplex
168operation.
169Not specifying
170.Ar full duplex
171implies
172.Ar half-duplex
173mode.
174.It 100baseTX
175Set 100Mbps (fast ethernet) operation.
176The
177.Ar mediaopt
178option can also be used to enable
179.Ar full-duplex
180operation.
181Not specifying
182.Ar full duplex
183implies
184.Ar half-duplex
185mode.
186.El
187.Pp
188The
189.Nm
190driver supports the following media options:
191.Pp
192.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
193.It full-duplex
194Force full duplex operation.
195The interface will operate in
196half duplex mode if this media option is not specified.
197.El
198.Pp
199Note that the 100baseTX media type may not be available on certain
200Intel 21143 adapters which support 10mbps media attachments only.
201For more information on configuring this device, see
202.Xr ifconfig 8 .
203.Sh DIAGNOSTICS
204.Bl -diag
205.It "dc%d: couldn't map ports/memory"
206A fatal initialization error has occurred.
207.It "dc%d: couldn't map interrupt"
208A fatal initialization error has occurred.
209.It "dc%d: watchdog timeout"
210A packet was queued for transmission and a transmit command was
211issued, however the device failed to acknowledge the transmission
212before a timeout expired.
213This can happen if the device is unable
214to deliver interrupts for some reason, of if there is a problem with
215the network connection (cable).
216.It "dc%d: no memory for rx list"
217The driver failed to allocate an mbuf for the receiver ring.
218.It "dc%d: TX underrun -- increasing TX threshold"
219The device generated a transmit underrun error while attempting to
220DMA and transmit a packet.
221This happens if the host is not able to
222DMA the packet data into the NIC's FIFO fast enough.
223The driver
224will dynamically increase the transmit start threshold so that
225more data must be DMAed into the FIFO before the NIC will start
226transmitting it onto the wire.
227.It "dc%d: TX underrun -- using store and forward mode"
228The device continued to generate transmit underruns even after all
229possible transmit start threshold settings had been tried, so the
230driver programmed the chip for store and forward mode.
231In this mode,
232the NIC will not begin transmission until the entire packet has been
233transfered into its FIFO memory.
234.It "dc%d: chip is in D3 power state -- setting to D0"
235This message applies only to adapters which support power
236management.
237Some operating systems place the controller in low power
238mode when shutting down, and some PCI BIOSes fail to bring the chip
239out of this state before configuring it.
240The controller loses all of
241its PCI configuration in the D3 state, so if the BIOS does not set
242it back to full power mode in time, it won't be able to configure it
243correctly.
244The driver tries to detect this condition and bring
245the adapter back to the D0 (full power) state, but this may not be
246enough to return the driver to a fully operational condition.
247If
248you see this message at boot time and the driver fails to attach
249the device as a network interface, you will have to perform a second
250warm boot to have the device properly configured.
251.Pp
252Note that this condition only occurs when warm booting from another
253operating system.
254If you power down your system prior to booting
255.Fx ,
256the card should be configured correctly.
257.El
258.Sh SEE ALSO
259.Xr arp 4 ,
260.Xr netintro 4 ,
261.Xr ng_ether 4 ,
262.Xr ifconfig 8
263.Rs
264.%T ADMtek AL981, AL983 and AL985 data sheets
265.%O http://www.admtek.com.tw
266.Re
267.Rs
268.%T ASIX Electronics AX88140A and AX88141 data sheets
269.%O http://www.asix.com.tw
270.Re
271.Rs
272.%T Davicom DM9102 data sheet
273.%O http://www.davicom8.com
274.Re
275.Rs
276.%T Intel 21143 Hardware Reference Manual
277.%O http://developer.intel.com
278.Re
279.Rs
280.%T Macronix 98713/A, 98715/A and 98725 data sheets
281.%O http://www.macronix.com
282.Re
283.Rs
284.%T Macronix 98713/A and 98715/A app notes
285.%O http://www.macronix.com
286.Re
287.Sh HISTORY
288The
289.Nm
290device driver first appeared in
291.Fx 4.0 .
292.Sh AUTHORS
293The
294.Nm
295driver was written by
296.An Bill Paul Aq wpaul@ee.columbia.edu .
297.Sh BUGS
298The Macronix application notes claim that in order to put the
299chips in normal operation, the driver must write a certain magic
300number into the CSR16 register.
301The numbers are documented in
302the app notes, but the exact meaning of the bits is not.
303.Pp
304The 98713A seems to have a problem with 10Mbps full duplex mode.
305The transmitter works but the receiver tends to produce many
306unexplained errors leading to very poor overall performance.
307The
30898715A does not exhibit this problem.
309All other modes on the
31098713A seem to work correctly.
311.Pp
312The original 82c168 PNIC chip has built in NWAY support which is
313used on certain early LinkSys LNE100TX and Matrox FastNIC cards,
314however it is horribly broken and difficult to use reliably.
315Consequently, autonegotiation is not currently supported for this
316chipset: the driver defaults the NIC to 10baseT half duplex, and it's
317up to the operator to manually select a different mode if necessary.
318(Later cards use an external MII transceiver to implement NWAY
319autonegotiation and work correctly.)
320.Pp
321The
322.Nm
323driver programs 82c168 and 82c169 PNIC chips to use the store and
324forward setting for the transmit start threshold by default.
325This
326is to work around problems with some NIC/PCI bus combinations where
327the PNIC can transmit corrupt frames when operating at 100Mbps,
328probably due to PCI DMA burst transfer errors.
329.Pp
330The 82c168 and 82c169 PNIC chips also have a receiver bug that
331sometimes manifests during periods of heavy receive and transmit
332activity, where the chip will improperly DMA received frames to
333the host.
334The chips appear to upload several kilobytes of garbage
335data along with the received frame data, dirtying several RX buffers
336instead of just the expected one.
337The
338.Nm
339driver detects this condition and will salvage the frame, however
340it incurs a serious performance penalty in the process.
341.Pp
342The PNIC chips also sometimes generate a transmit underrun error when
343the driver attempts to download the receiver filter setup frame, which
344can result in the receive filter being incorrectly programmed.
345The
346.Nm
347driver will watch for this condition and requeue the setup frame until
348it is transfered successfully.
349.Pp
350The ADMtek AL981 chip (and possibly the AN985 as well) has been observed
351to sometimes wedge on transmit: this appears to happen when the driver
352queues a sequence of frames which cause it to wrap from the end of the
353the transmit descriptor ring back to the beginning.
354The
355.Nm
356driver attempts to avoid this condition by not queing any frames past
357the end of the transmit ring during a single invocation of the
358.Fn dc_start
359routine.
360This workaround has a negligible impact on transmit performance.
361
362
363