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.\" 33.Dd November 14, 2010 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 and 98725 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 64Conexant LANfinity RS7112 (miniPCI) 65.It 66Abocom FE2500 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 10Mbps only ethernet on Compaq Presario 7900 series 109desktops (21143, non-MII) 110.It 111Built in ethernet on LinkSys EtherFast 10/100 Instant GigaDrive (DM9102, MII) 112.It 113Kingston KNE100TX (21143, MII) 114.It 115D-Link DFE-570TX (21143, MII, quad port) 116.It 117NDC SOHOware SFA110 (98713A) 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.El 141.Pp 142The 143.Nm 144driver supports the following media types: 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 (see the 157.Sx BUGS 158section for details). 159The original 82c168 appears 160on very early revisions of the LinkSys LNE100TX and Matrox FastNIC. 161.It 10baseT/UTP 162Set 10Mbps operation. 163The 164.Ar mediaopt 165option can also be used to enable 166.Ar full-duplex 167operation. 168Not specifying 169.Ar full duplex 170implies 171.Ar half-duplex 172mode. 173.It 100baseTX 174Set 100Mbps (fast ethernet) operation. 175The 176.Ar mediaopt 177option can also be used to enable 178.Ar full-duplex 179operation. 180Not specifying 181.Ar full duplex 182implies 183.Ar half-duplex 184mode. 185.El 186.Pp 187The 188.Nm 189driver supports the following media options: 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. 199.Pp 200For more information on configuring this device, see 201.Xr ifconfig 8 . 202The 203.Nm 204driver supports 205.Xr polling 4 . 206.Sh DIAGNOSTICS 207.Bl -diag 208.It "dc%d: couldn't map ports/memory" 209A fatal initialization error has occurred. 210.It "dc%d: couldn't map interrupt" 211A fatal initialization error has occurred. 212.It "dc%d: watchdog timeout" 213A packet was queued for transmission and a transmit command was 214issued, but the device failed to acknowledge the transmission 215before a timeout expired. 216This can happen if the device is unable 217to deliver interrupts for some reason, of if there is a problem with 218the network connection (cable or network equipment) that results in a loss 219of link. 220.It "dc%d: no memory for rx list" 221The driver failed to allocate an mbuf for the receiver ring. 222.It "dc%d: TX underrun -- increasing TX threshold" 223The device generated a transmit underrun error while attempting to 224DMA and transmit a packet. 225This happens if the host is not able to 226DMA the packet data into the NIC's FIFO fast enough. 227The driver 228will dynamically increase the transmit start threshold so that 229more data must be DMAed into the FIFO before the NIC will start 230transmitting it onto the wire. 231.It "dc%d: TX underrun -- using store and forward mode" 232The device continued to generate transmit underruns even after all 233possible transmit start threshold settings had been tried, so the 234driver programmed the chip for store and forward mode. 235In this mode, 236the NIC will not begin transmission until the entire packet has been 237transferred into its FIFO memory. 238.It "dc%d: chip is in D3 power state -- setting to D0" 239This message applies only to adapters which support power 240management. 241Some operating systems place the controller in low power 242mode when shutting down, and some PCI BIOSes fail to bring the chip 243out of this state before configuring it. 244The controller loses all of 245its PCI configuration in the D3 state, so if the BIOS does not set 246it back to full power mode in time, it won't be able to configure it 247correctly. 248The driver tries to detect this condition and bring 249the adapter back to the D0 (full power) state, but this may not be 250enough to return the driver to a fully operational condition. 251If 252you see this message at boot time and the driver fails to attach 253the device as a network interface, you will have to perform a second 254warm boot to have the device properly configured. 255.Pp 256Note that this condition only occurs when warm booting from another 257operating system. 258If you power down your system prior to booting 259.Dx , 260the card should be configured correctly. 261.El 262.Sh SEE ALSO 263.Xr arp 4 , 264.Xr ifmedia 4 , 265.Xr miibus 4 , 266.Xr netintro 4 , 267.Xr ng_ether 4 , 268.Xr polling 4 , 269.Xr ifconfig 8 270.Rs 271.%T ADMtek AL981, AL983 and AL985 data sheets 272.%O http://www.admtek.com.tw 273.Re 274.Rs 275.%T ASIX Electronics AX88140A and AX88141 data sheets 276.%O http://www.asix.com.tw 277.Re 278.Rs 279.%T Davicom DM9102 data sheet 280.%O http://www.davicom.com.tw 281.Re 282.Rs 283.%T Intel 21143 Hardware Reference Manual 284.%O http://developer.intel.com 285.Re 286.Rs 287.%T Macronix 98713/A, 98715/A and 98725 data sheets 288.%O http://www.macronix.com 289.Re 290.Rs 291.%T Macronix 98713/A and 98715/A app notes 292.%O http://www.macronix.com 293.Re 294.Sh HISTORY 295The 296.Nm 297device driver first appeared in 298.Fx 4.0 . 299.Sh AUTHORS 300The 301.Nm 302driver was written by 303.An Bill Paul Aq Mt wpaul@ee.columbia.edu . 304.Sh BUGS 305The Macronix application notes claim that in order to put the 306chips in normal operation, the driver must write a certain magic 307number into the CSR16 register. 308The numbers are documented in 309the app notes, but the exact meaning of the bits is not. 310.Pp 311The 98713A seems to have a problem with 10Mbps full duplex mode. 312The transmitter works but the receiver tends to produce many 313unexplained errors leading to very poor overall performance. 314The 31598715A does not exhibit this problem. 316All other modes on the 31798713A seem to work correctly. 318.Pp 319The original 82c168 PNIC chip has built in NWAY support which is 320used on certain early LinkSys LNE100TX and Matrox FastNIC cards, 321however it is horribly broken and difficult to use reliably. 322Consequently, autonegotiation is not currently supported for this 323chipset: the driver defaults the NIC to 10baseT half duplex, and it's 324up to the operator to manually select a different mode if necessary. 325(Later cards use an external MII transceiver to implement NWAY 326autonegotiation and work correctly.) 327.Pp 328The 329.Nm 330driver programs 82c168 and 82c169 PNIC chips to use the store and 331forward setting for the transmit start threshold by default. 332This 333is to work around problems with some NIC/PCI bus combinations where 334the PNIC can transmit corrupt frames when operating at 100Mbps, 335probably due to PCI DMA burst transfer errors. 336.Pp 337The 82c168 and 82c169 PNIC chips also have a receiver bug that 338sometimes manifests during periods of heavy receive and transmit 339activity, where the chip will improperly DMA received frames to 340the host. 341The chips appear to upload several kilobytes of garbage 342data along with the received frame data, dirtying several RX buffers 343instead of just the expected one. 344The 345.Nm 346driver detects this condition and will salvage the frame; however, 347it incurs a serious performance penalty in the process. 348.Pp 349The PNIC chips also sometimes generate a transmit underrun error when 350the driver attempts to download the receiver filter setup frame, which 351can result in the receive filter being incorrectly programmed. 352The 353.Nm 354driver will watch for this condition and requeue the setup frame until 355it is transferred successfully. 356.Pp 357The ADMtek AL981 chip (and possibly the AN985 as well) has been observed 358to sometimes wedge on transmit: this appears to happen when the driver 359queues a sequence of frames which cause it to wrap from the end of the 360transmit descriptor ring back to the beginning. 361The 362.Nm 363driver attempts to avoid this condition by not queueing any frames past 364the end of the transmit ring during a single invocation of the 365.Fn dc_start 366routine. 367This workaround has a negligible impact on transmit performance. 368