xref: /freebsd/share/man/man4/dc.4 (revision 5521ff5a4d1929056e7ffc982fac3341ca54df7c)
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 FreeBSD
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.El
64.Pp
65All of these chips have the same general register layout, DMA
66descriptor format and method of operation.
67All of the clone chips
68are based on the 21143 design with various modifications.
69The
7021143 itself has support for 10baseT, BNC, AUI, MII and symbol
71media attachments, 10 and 100Mbps speeds in full or half duplex,
72built in NWAY autonegotiation and wake on LAN.
73The 21143 also
74offers several receive filter programming options including
75perfect filtering, inverse perfect filtering and hash table
76filtering.
77.Pp
78Some clone chips duplicate the 21143 fairly closely while others
79only maintain superficial similarities.
80Some support only MII
81media attachments.
82Others use different receiver filter programming
83mechanisms.
84At least one supports only chained DMA descriptors
85(most support both chained descriptors and contiguously allocated
86fixed size rings). Some chips (especially the PNIC) also have
87peculiar bugs.
88The
89.Nm
90driver does its best to provide generalized support for all
91of these chipsets in order to keep special case code to a minimun.
92.Pp
93These chips are used by many vendors which makes it
94difficult to provide a complete list of all supported cards.
95The
96following NICs are known to work with the
97.Nm
98driver at this time:
99.Pp
100.Bl -bullet -compact -offset indent
101.It
102Digital DE500-BA 10/100 (21143, non-MII)
103.It
104Built in DE500-BA on DEC Alpha workstations (21143, non-MII)
105.It
106Built in 10Mbps only ethernet on Compaq Presario 7900 series
107desktops (21143, non-MII)
108.It
109Built in ethernet on LinkSys EtherFast 10/100 Instant GigaDrive (DM9102, MII)
110.It
111Kingston KNE100TX (21143, MII)
112.It
113D-Link DFE-570TX (21143, MII, quad port)
114.It
115NDC SOHOware SFA110A (98713A)
116.It
117NDC SOHOware SFA110A Rev B4 (98715AEC-C)
118.It
119SVEC PN102-TX (98713)
120.It
121CNet Pro120A (98715A or 98713A) and CNet Pro120B (98715)
122.It
123Compex RL100-TX (98713 or 98713A)
124.It
125LinkSys LNE100TX (PNIC 82c168, 82c169)
126.It
127NetGear FA310-TX Rev. D1, D2 or D3 (PNIC 82c169)
128.It
129Matrox FastNIC 10/100 (PNIC 82c168, 82c169)
130.It
131Kingston KNE110TX (PNIC 82c169)
132.It
133LinkSys LNE100TX v2.0 (PNIC II 82c115)
134.It
135Jaton XpressNet (Davicom DM9102)
136.It
137Alfa Inc GFC2204 (ASIX AX88140A)
138.It
139CNet Pro110B (ASIX AX88140A)
140.It
141LinkSys LNE100TX v4.0/4.1 (ADMtek AN985 Centaur-P)
142.El
143.Pp
144The
145.Nm
146driver supports the following media types:
147.Pp
148.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
149.It autoselect
150Enable autoselection of the media type and options.
151The user can manually override
152the autoselected mode by adding media options to the
153.Pa /etc/rc.conf
154file.
155.Pp
156Note: the built-in NWAY autonegotiation on the original PNIC 82c168
157chip is horribly broken and is not supported by the
158.Nm
159driver at this time (see the
160.Sx BUGS
161section for details).
162The original 82c168 appears
163on very early revisions of the LinkSys LNE100TX and Matrox FastNIC.
164.It 10baseT/UTP
165Set 10Mbps operation.
166The
167.Ar mediaopt
168option can also be used to enable
169.Ar full-duplex
170operation.
171Not specifying
172.Ar full duplex
173implies
174.Ar half-duplex
175mode.
176.It 100baseTX
177Set 100Mbps (fast ethernet) operation.
178The
179.Ar mediaopt
180option can also be used to enable
181.Ar full-duplex
182operation.
183Not specifying
184.Ar full duplex
185implies
186.Ar half-duplex
187mode.
188.El
189.Pp
190The
191.Nm
192driver supports the following media options:
193.Pp
194.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
195.It full-duplex
196Force full duplex operation.
197The interface will operate in
198half duplex mode if this media option is not specified.
199.El
200.Pp
201Note that the 100baseTX media type may not be available on certain
202Intel 21143 adapters which support 10mbps media attachments only.
203For more information on configuring this device, see
204.Xr ifconfig 8 .
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
236transfered 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.Fx ,
259the card should be configured correctly.
260.El
261.Sh SEE ALSO
262.Xr arp 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