xref: /freebsd/share/man/man4/dc.4 (revision f9218d3d4fd34f082473b3a021c6d4d109fb47cf)
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.El
147.Pp
148The
149.Nm
150driver supports the following media types:
151.Pp
152.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
153.It autoselect
154Enable autoselection of the media type and options.
155The user can manually override
156the autoselected mode by adding media options to the
157.Pa /etc/rc.conf
158file.
159.Pp
160Note: the built-in NWAY autonegotiation on the original PNIC 82c168
161chip is horribly broken and is not supported by the
162.Nm
163driver at this time (see the
164.Sx BUGS
165section for details).
166The original 82c168 appears
167on very early revisions of the LinkSys LNE100TX and Matrox FastNIC.
168.It 10baseT/UTP
169Set 10Mbps operation.
170The
171.Ar mediaopt
172option can also be used to enable
173.Ar full-duplex
174operation.
175Not specifying
176.Ar full duplex
177implies
178.Ar half-duplex
179mode.
180.It 100baseTX
181Set 100Mbps (fast ethernet) operation.
182The
183.Ar mediaopt
184option can also be used to enable
185.Ar full-duplex
186operation.
187Not specifying
188.Ar full duplex
189implies
190.Ar half-duplex
191mode.
192.El
193.Pp
194The
195.Nm
196driver supports the following media options:
197.Pp
198.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
199.It full-duplex
200Force full duplex operation.
201The interface will operate in
202half duplex mode if this media option is not specified.
203.El
204.Pp
205Note that the 100baseTX media type may not be available on certain
206Intel 21143 adapters which support 10mbps media attachments only.
207For more information on configuring this device, see
208.Xr ifconfig 8 .
209.Sh DIAGNOSTICS
210.Bl -diag
211.It "dc%d: couldn't map ports/memory"
212A fatal initialization error has occurred.
213.It "dc%d: couldn't map interrupt"
214A fatal initialization error has occurred.
215.It "dc%d: watchdog timeout"
216A packet was queued for transmission and a transmit command was
217issued, but the device failed to acknowledge the transmission
218before a timeout expired.
219This can happen if the device is unable
220to deliver interrupts for some reason, of if there is a problem with
221the network connection (cable or network equipment) that results in a loss
222of link.
223.It "dc%d: no memory for rx list"
224The driver failed to allocate an mbuf for the receiver ring.
225.It "dc%d: TX underrun -- increasing TX threshold"
226The device generated a transmit underrun error while attempting to
227DMA and transmit a packet.
228This happens if the host is not able to
229DMA the packet data into the NIC's FIFO fast enough.
230The driver
231will dynamically increase the transmit start threshold so that
232more data must be DMAed into the FIFO before the NIC will start
233transmitting it onto the wire.
234.It "dc%d: TX underrun -- using store and forward mode"
235The device continued to generate transmit underruns even after all
236possible transmit start threshold settings had been tried, so the
237driver programmed the chip for store and forward mode.
238In this mode,
239the NIC will not begin transmission until the entire packet has been
240transfered into its FIFO memory.
241.It "dc%d: chip is in D3 power state -- setting to D0"
242This message applies only to adapters which support power
243management.
244Some operating systems place the controller in low power
245mode when shutting down, and some PCI BIOSes fail to bring the chip
246out of this state before configuring it.
247The controller loses all of
248its PCI configuration in the D3 state, so if the BIOS does not set
249it back to full power mode in time, it won't be able to configure it
250correctly.
251The driver tries to detect this condition and bring
252the adapter back to the D0 (full power) state, but this may not be
253enough to return the driver to a fully operational condition.
254If
255you see this message at boot time and the driver fails to attach
256the device as a network interface, you will have to perform a second
257warm boot to have the device properly configured.
258.Pp
259Note that this condition only occurs when warm booting from another
260operating system.
261If you power down your system prior to booting
262.Fx ,
263the card should be configured correctly.
264.El
265.Sh SEE ALSO
266.Xr arp 4 ,
267.Xr miibus 4 ,
268.Xr netintro 4 ,
269.Xr ng_ether 4 ,
270.Xr ifconfig 8
271.Rs
272.%T ADMtek AL981, AL983 and AL985 data sheets
273.%O http://www.admtek.com.tw
274.Re
275.Rs
276.%T ASIX Electronics AX88140A and AX88141 data sheets
277.%O http://www.asix.com.tw
278.Re
279.Rs
280.%T Davicom DM9102 data sheet
281.%O http://www.davicom8.com
282.Re
283.Rs
284.%T Intel 21143 Hardware Reference Manual
285.%O http://developer.intel.com
286.Re
287.Rs
288.%T Macronix 98713/A, 98715/A and 98725 data sheets
289.%O http://www.macronix.com
290.Re
291.Rs
292.%T Macronix 98713/A and 98715/A app notes
293.%O http://www.macronix.com
294.Re
295.Sh HISTORY
296The
297.Nm
298device driver first appeared in
299.Fx 4.0 .
300.Sh AUTHORS
301The
302.Nm
303driver was written by
304.An Bill Paul Aq wpaul@ee.columbia.edu .
305.Sh BUGS
306The Macronix application notes claim that in order to put the
307chips in normal operation, the driver must write a certain magic
308number into the CSR16 register.
309The numbers are documented in
310the app notes, but the exact meaning of the bits is not.
311.Pp
312The 98713A seems to have a problem with 10Mbps full duplex mode.
313The transmitter works but the receiver tends to produce many
314unexplained errors leading to very poor overall performance.
315The
31698715A does not exhibit this problem.
317All other modes on the
31898713A seem to work correctly.
319.Pp
320The original 82c168 PNIC chip has built in NWAY support which is
321used on certain early LinkSys LNE100TX and Matrox FastNIC cards,
322however it is horribly broken and difficult to use reliably.
323Consequently, autonegotiation is not currently supported for this
324chipset: the driver defaults the NIC to 10baseT half duplex, and it's
325up to the operator to manually select a different mode if necessary.
326(Later cards use an external MII transceiver to implement NWAY
327autonegotiation and work correctly.)
328.Pp
329The
330.Nm
331driver programs 82c168 and 82c169 PNIC chips to use the store and
332forward setting for the transmit start threshold by default.
333This
334is to work around problems with some NIC/PCI bus combinations where
335the PNIC can transmit corrupt frames when operating at 100Mbps,
336probably due to PCI DMA burst transfer errors.
337.Pp
338The 82c168 and 82c169 PNIC chips also have a receiver bug that
339sometimes manifests during periods of heavy receive and transmit
340activity, where the chip will improperly DMA received frames to
341the host.
342The chips appear to upload several kilobytes of garbage
343data along with the received frame data, dirtying several RX buffers
344instead of just the expected one.
345The
346.Nm
347driver detects this condition and will salvage the frame; however,
348it incurs a serious performance penalty in the process.
349.Pp
350The PNIC chips also sometimes generate a transmit underrun error when
351the driver attempts to download the receiver filter setup frame, which
352can result in the receive filter being incorrectly programmed.
353The
354.Nm
355driver will watch for this condition and requeue the setup frame until
356it is transfered successfully.
357.Pp
358The ADMtek AL981 chip (and possibly the AN985 as well) has been observed
359to sometimes wedge on transmit: this appears to happen when the driver
360queues a sequence of frames which cause it to wrap from the end of the
361transmit descriptor ring back to the beginning.
362The
363.Nm
364driver attempts to avoid this condition by not queuing any frames past
365the end of the transmit ring during a single invocation of the
366.Fn dc_start
367routine.
368This workaround has a negligible impact on transmit performance.
369