xref: /freebsd/share/man/man4/dc.4 (revision bec35b9a5f0cdb233b0601c507114d11a1093126)
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 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.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 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.Pp
145.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
146.It autoselect
147Enable autoselection of the media type and options.
148The user can manually override
149the autoselected mode by adding media options to the
150.Pa /etc/rc.conf
151file.
152.Pp
153Note: the built-in NWAY autonegotiation on the original PNIC 82c168
154chip is horribly broken and is not supported by the
155.Nm
156driver at this time: the chip will operate in any speed or duplex
157mode, however these must be set manually.
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.Pp
190.Bl -tag -width xxxxxxxxxxxxxxxxxxxx
191.It full-duplex
192Force full duplex operation.
193The interface will operate in
194half duplex mode if this media option is not specified.
195.El
196.Pp
197Note that the 100baseTX media type may not be available on certain
198Intel 21143 adapters which support 10mbps media attachments only.
199For more information on configuring this device, see
200.Xr ifconfig 8 .
201.Sh DIAGNOSTICS
202.Bl -diag
203.It "dc%d: couldn't map ports/memory"
204A fatal initialization error has occurred.
205.It "dc%d: couldn't map interrupt"
206A fatal initialization error has occurred.
207.It "dc%d: watchdog timeout"
208A packet was queued for transmission and a transmit command was
209issued, however the device failed to acknowledge the transmission
210before a timeout expired.
211This can happen if the device is unable
212to deliver interrupts for some reason, of if there is a problem with
213the network connection (cable).
214.It "dc%d: no memory for rx list"
215The driver failed to allocate an mbuf for the receiver ring.
216.It "dc%d: TX underrun -- increasing TX threshold"
217The device generated a transmit underrun error while attempting to
218DMA and transmit a packet.
219This happens if the host is not able to
220DMA the packet data into the NIC's FIFO fast enough.
221The driver
222will dynamically increase the transmit start threshold so that
223more data must be DMAed into the FIFO before the NIC will start
224transmitting it onto the wire.
225.It "dc%d: TX underrun -- using store and forward mode"
226The device continued to generate transmit underruns even after all
227possible transmit start threshold settings had been tried, so the
228driver programmed the chip for store and forward mode.
229In this mode,
230the NIC will not begin transmission until the entire packet has been
231transfered into its FIFO memory.
232.It "dc%d: chip is in D3 power state -- setting to D0"
233This message applies only to adapters which support power
234management.
235Some operating systems place the controller in low power
236mode when shutting down, and some PCI BIOSes fail to bring the chip
237out of this state before configuring it.
238The controller loses all of
239its PCI configuration in the D3 state, so if the BIOS does not set
240it back to full power mode in time, it won't be able to configure it
241correctly.
242The driver tries to detect this condition and bring
243the adapter back to the D0 (full power) state, but this may not be
244enough to return the driver to a fully operational condition.
245If
246you see this message at boot time and the driver fails to attach
247the device as a network interface, you will have to perform second
248warm boot to have the device properly configured.
249.Pp
250Note that this condition only occurs when warm booting from another
251operating system.
252If you power down your system prior to booting
253.Fx ,
254the card should be configured correctly.
255.El
256.Sh SEE ALSO
257.Xr arp 4 ,
258.Xr netintro 4 ,
259.Xr ng_ether 4 ,
260.Xr ifconfig 8
261.Rs
262.%T ADMtek AL981, AL983 and AL985 data sheets
263.%O http://www.admtek.com.tw
264.Re
265.Rs
266.%T ASIX Electronics AX88140A and AX88141 data sheets
267.%O http://www.asix.com.tw
268.Re
269.Rs
270.%T Davicom DM9102 data sheet
271.%O http://www.davicom8.com
272.Re
273.Rs
274.%T Intel 21143 Hardware Reference Manual
275.%O http://developer.intel.com
276.Re
277.Rs
278.%T Macronix 98713/A, 98715/A and 98725 data sheets
279.%O http://www.macronix.com
280.Re
281.Rs
282.%T Macronix 98713/A and 98715/A app notes
283.%O http://www.macronix.com
284.Re
285.Sh HISTORY
286The
287.Nm
288device driver first appeared in
289.Fx 4.0 .
290.Sh AUTHORS
291The
292.Nm
293driver was written by
294.An Bill Paul Aq wpaul@ee.columbia.edu .
295.Sh BUGS
296The Macronix application notes claim that in order to put the
297chips in normal operation, the driver must write a certian magic
298number into the CSR16 register.
299The numbers are documented in
300the app notes, but the exact meaning of the bits is not.
301.Pp
302The 98713A seems to have a problem with 10Mbps full duplex mode.
303The transmitter works but the receiver tends to produce many
304unexplained errors leading to very poor overall performance.
305The
30698715A does not exhibit this problem.
307All other modes on the
30898713A seem to work correctly.
309.Pp
310The original 82c168 PNIC chip has built in NWAY support which is
311used on certain early LinkSys LNE100TX and Matrox FastNIC cards,
312however it is horribly broken and difficult to use reliably.
313Consequently, autonegotiation is not currently supported for this
314chipset: the driver defaults the NIC to 10baseT half duplex, and it's
315up to the operator to manually select a different mode if necessary.
316(Later cards use an external MII transceiver to implement NWAY
317autonegotiation and work correctly.)
318.Pp
319The
320.Nm
321driver programs 82c168 and 82c169 PNIC chips to use the store and
322forward setting for the transmit start threshold by default.
323This
324is to work around problems with some NIC/PCI bus combinations where
325the PNIC can transmit corrupt frames when operating at 100Mbps,
326probably due to PCI DMA burst transfer errors.
327.Pp
328The 82c168 and 82c169 PNIC chips also have a receiver bug that
329sometimes manifests during periods of heavy receive and transmit
330activity, where the chip will improperly DMA received frames to
331the host.
332The chips appear to upload several kilobytes of garbage
333data along with the received frame data, dirtying several RX buffers
334instead of just the expected one.
335The
336.Nm
337driver detects this condition and will salvage the frame, however
338it incurs a serious performance penalty in the process.
339.Pp
340The PNIC chips also sometimes generate a transmit underrun error when
341the driver attempts to download the receiver filter setup frame, which
342can result in the receive filter being incorrectly programmed.
343The
344.Nm
345driver will watch for this condition and requeue the setup frame until
346it is transfered successfully.
347.Pp
348The ADMtek AL981 chip (and possibly the AN985 as well) has been observed
349to sometimes wedge on transmit: this appears to happen when the driver
350queues a sequence of frames which cause it to wrap from the end of the
351the transmit descriptor ring back to the beginning.
352The
353.Nm
354driver attempts to avoid this condition by not queing any frames past
355the end of the transmit ring during a single invocation of the
356.Fn dc_start
357routine.
358This workaround has a negligible impact on transmit performance.
359
360
361