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 July 16, 2005 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 40To compile this driver into the kernel, 41place the following lines in your 42kernel configuration file: 43.Bd -ragged -offset indent 44.Cd "device miibus" 45.Cd "device dc" 46.Ed 47.Pp 48Alternatively, to load the driver as a 49module at boot time, place the following line in 50.Xr loader.conf 5 : 51.Bd -literal -offset indent 52if_dc_load="YES" 53.Ed 54.Sh DESCRIPTION 55The 56.Nm 57driver provides support for several PCI Fast Ethernet adapters and 58embedded controllers based on the the DEC/Intel 21143 chipset and clones. 59.Pp 60All of supported chipsets have the same general register layout, DMA 61descriptor format and method of operation. 62All of the clone chips 63are based on the 21143 design with various modifications. 64The 6521143 itself has support for 10baseT, BNC, AUI, MII and symbol 66media attachments, 10 and 100Mbps speeds in full or half duplex, 67built in NWAY autonegotiation and wake on LAN. 68The 21143 also 69offers several receive filter programming options including 70perfect filtering, inverse perfect filtering and hash table 71filtering. 72.Pp 73Some clone chips duplicate the 21143 fairly closely while others 74only maintain superficial similarities. 75Some support only MII 76media attachments. 77Others use different receiver filter programming 78mechanisms. 79At least one supports only chained DMA descriptors 80(most support both chained descriptors and contiguously allocated 81fixed size rings). 82Some chips (especially the PNIC) also have 83peculiar bugs. 84The 85.Nm 86driver does its best to provide generalized support for all 87of these chipsets in order to keep special case code to a minimum. 88.Pp 89These chips are used by many vendors which makes it 90difficult to provide a complete list of all supported cards. 91.Pp 92The 93.Nm 94driver supports the following media types: 95.Pp 96.Bl -tag -width ".Cm 10baseT/UTP" 97.It Cm autoselect 98Enable autoselection of the media type and options. 99The user can manually override 100the autoselected mode by adding media options to the 101.Pa /etc/rc.conf 102file. 103.Pp 104Note: the built-in NWAY autonegotiation on the original PNIC 82c168 105chip is horribly broken and is not supported by the 106.Nm 107driver at this time (see the 108.Sx BUGS 109section for details). 110The original 82c168 appears 111on very early revisions of the LinkSys LNE100TX and Matrox FastNIC. 112.It Cm 10baseT/UTP 113Set 10Mbps operation. 114The 115.Cm mediaopt 116option can also be used to enable 117.Cm full-duplex 118operation. 119Not specifying 120.Cm full-duplex 121implies 122.Cm half-duplex 123mode. 124.It Cm 100baseTX 125Set 100Mbps (Fast Ethernet) operation. 126The 127.Cm mediaopt 128option can also be used to enable 129.Cm full-duplex 130operation. 131Not specifying 132.Cm full-duplex 133implies 134.Cm half-duplex 135mode. 136.El 137.Pp 138The 139.Nm 140driver supports the following media options: 141.Pp 142.Bl -tag -width ".Cm full-duplex" 143.It Cm full-duplex 144Force full duplex operation. 145The interface will operate in 146half duplex mode if this media option is not specified. 147.El 148.Pp 149Note that the 100baseTX media type may not be available on certain 150Intel 21143 adapters which support 10Mbps media attachments only. 151For more information on configuring this device, see 152.Xr ifconfig 8 . 153.Sh HARDWARE 154The 155.Nm 156driver provides support for the following chipsets: 157.Pp 158.Bl -bullet -compact 159.It 160DEC/Intel 21143 161.It 162ADMtek AL981 Comet, AN985 Centaur, ADM9511 Centaur II and ADM9513 163Centaur II 164.It 165ASIX Electronics AX88140A and AX88141 166.It 167Conexant LANfinity RS7112 (miniPCI) 168.It 169Davicom DM9009, DM9100, DM9102 and DM9102A 170.It 171Lite-On 82c168 and 82c169 PNIC 172.It 173Lite-On/Macronix 82c115 PNIC II 174.It 175Macronix 98713, 98713A, 98715, 98715A, 98715AEC-C, 98725, 98727 and 98732 176.It 177Xircom X3201 (cardbus only) 178.El 179.Pp 180The 181following NICs are known to work with the 182.Nm 183driver at this time: 184.Pp 185.Bl -bullet -compact 186.It 1873Com OfficeConnect 10/100B (ADMtek AN985 Centaur-P) 188.It 189Abocom FE2500 190.It 191Accton EN1217 (98715A) 192.It 193Accton EN2242 MiniPCI 194.It 195Adico AE310TX (98715A) 196.It 197Alfa Inc GFC2204 (ASIX AX88140A) 198.It 199Built in 10Mbps only Ethernet on Compaq Presario 7900 series 200desktops (21143, non-MII) 201.It 202Built in Sun DMFE 10/100 Mbps Ethernet on Sun Netra X1 and Sun Fire V100 203(DM9102A, MII) 204.It 205Built in Ethernet on LinkSys EtherFast 10/100 Instant GigaDrive (DM9102, MII) 206.It 207CNet Pro110B (ASIX AX88140A) 208.It 209CNet Pro120A (98715A or 98713A) and CNet Pro120B (98715) 210.It 211Compex RL100-TX (98713 or 98713A) 212.It 213D-Link DFE-570TX (21143, MII, quad port) 214.It 215Digital DE500-BA 10/100 (21143, non-MII) 216.It 217ELECOM Laneed LD-CBL/TXA (ADMtek AN985) 218.It 219Hawking CB102 CardBus 220.It 221IBM EtherJet Cardbus Adapter 222.It 223Intel PRO/100 Mobile Cardbus (versions that use the X3201 chipset) 224.It 225Jaton XpressNet (Davicom DM9102) 226.It 227Kingston KNE100TX (21143, MII) 228.It 229Kingston KNE110TX (PNIC 82c169) 230.It 231LinkSys LNE100TX (PNIC 82c168, 82c169) 232.It 233LinkSys LNE100TX v2.0 (PNIC II 82c115) 234.It 235LinkSys LNE100TX v4.0/4.1 (ADMtek AN985 Centaur-P) 236.It 237Matrox FastNIC 10/100 (PNIC 82c168, 82c169) 238.It 239Melco LGY-PCI-TXL 240.It 241Microsoft MN-120 10/100 CardBus (ADMTek Centaur-C) 242.It 243Microsoft MN-130 10/100 PCI (ADMTek Centaur-P) 244.It 245NDC SOHOware SFA110A (98713A) 246.It 247NDC SOHOware SFA110A Rev B4 (98715AEC-C) 248.It 249NetGear FA310-TX Rev.\& D1, D2 or D3 (PNIC 82c169) 250.It 251Netgear FA511 252.It 253PlaneX FNW-3602-T (ADMtek AN985) 254.It 255SMC EZ Card 10/100 1233A-TX (ADMtek AN985) 256.It 257SVEC PN102-TX (98713) 258.It 259Xircom Cardbus Realport 260.It 261Xircom Cardbus Ethernet 10/100 262.It 263Xircom Cardbus Ethernet II 10/100 264.El 265.Sh NOTES 266On sparc64 the 267.Nm 268driver respects the 269.Va local-mac-address? 270system configuration variable for the built in Sun DMFE 10/100 Mbps Ethernet 271interfaces on Sun Netra X1 and Sun Fire V100. 272This system configuration variable can be set in the Open Firmware boot 273monitor using the 274.Ic setenv 275command or by 276.Xr eeprom 8 . 277If set to 278.Dq Li false 279(the default), the 280.Nm 281driver will use the system's default MAC address for both of the built in 282devices. 283If set to 284.Dq Li true , 285the unique MAC address of each interface is used rather than the system's 286default MAC address. 287.Sh DIAGNOSTICS 288.Bl -diag 289.It "dc%d: couldn't map ports/memory" 290A fatal initialization error has occurred. 291.It "dc%d: couldn't map interrupt" 292A fatal initialization error has occurred. 293.It "dc%d: watchdog timeout" 294A packet was queued for transmission and a transmit command was 295issued, but the device failed to acknowledge the transmission 296before a timeout expired. 297This can happen if the device is unable 298to deliver interrupts for some reason, of if there is a problem with 299the network connection (cable or network equipment) that results in a loss 300of link. 301.It "dc%d: no memory for rx list" 302The driver failed to allocate an mbuf for the receiver ring. 303.It "dc%d: TX underrun -- increasing TX threshold" 304The device generated a transmit underrun error while attempting to 305DMA and transmit a packet. 306This happens if the host is not able to 307DMA the packet data into the NIC's FIFO fast enough. 308The driver 309will dynamically increase the transmit start threshold so that 310more data must be DMAed into the FIFO before the NIC will start 311transmitting it onto the wire. 312.It "dc%d: TX underrun -- using store and forward mode" 313The device continued to generate transmit underruns even after all 314possible transmit start threshold settings had been tried, so the 315driver programmed the chip for store and forward mode. 316In this mode, 317the NIC will not begin transmission until the entire packet has been 318transfered into its FIFO memory. 319.It "dc%d: chip is in D3 power state -- setting to D0" 320This message applies only to adapters which support power 321management. 322Some operating systems place the controller in low power 323mode when shutting down, and some PCI BIOSes fail to bring the chip 324out of this state before configuring it. 325The controller loses all of 326its PCI configuration in the D3 state, so if the BIOS does not set 327it back to full power mode in time, it will not be able to configure it 328correctly. 329The driver tries to detect this condition and bring 330the adapter back to the D0 (full power) state, but this may not be 331enough to return the driver to a fully operational condition. 332If 333you see this message at boot time and the driver fails to attach 334the device as a network interface, you will have to perform a second 335warm boot to have the device properly configured. 336.Pp 337Note that this condition only occurs when warm booting from another 338operating system. 339If you power down your system prior to booting 340.Fx , 341the card should be configured correctly. 342.El 343.Sh SEE ALSO 344.Xr altq 4 , 345.Xr arp 4 , 346.Xr miibus 4 , 347.Xr netintro 4 , 348.Xr ng_ether 4 , 349.Xr polling 4 , 350.Xr eeprom 8 , 351.Xr ifconfig 8 352.Rs 353.%T ADMtek AL981, AL983 and AL985 data sheets 354.%O http://www.admtek.com.tw 355.Re 356.Rs 357.%T ASIX Electronics AX88140A and AX88141 data sheets 358.%O http://www.asix.com.tw 359.Re 360.Rs 361.%T Davicom DM9102 data sheet 362.%O http://www.davicom8.com 363.Re 364.Rs 365.%T Intel 21143 Hardware Reference Manual 366.%O http://developer.intel.com 367.Re 368.Rs 369.%T Macronix 98713/A, 98715/A and 98725 data sheets 370.%O http://www.macronix.com 371.Re 372.Rs 373.%T Macronix 98713/A and 98715/A app notes 374.%O http://www.macronix.com 375.Re 376.Sh HISTORY 377The 378.Nm 379device driver first appeared in 380.Fx 4.0 . 381.Sh AUTHORS 382The 383.Nm 384driver was written by 385.An Bill Paul Aq wpaul@ee.columbia.edu . 386.Sh BUGS 387The Macronix application notes claim that in order to put the 388chips in normal operation, the driver must write a certain magic 389number into the CSR16 register. 390The numbers are documented in 391the app notes, but the exact meaning of the bits is not. 392.Pp 393The 98713A seems to have a problem with 10Mbps full duplex mode. 394The transmitter works but the receiver tends to produce many 395unexplained errors leading to very poor overall performance. 396The 39798715A does not exhibit this problem. 398All other modes on the 39998713A seem to work correctly. 400.Pp 401The original 82c168 PNIC chip has built in NWAY support which is 402used on certain early LinkSys LNE100TX and Matrox FastNIC cards, 403however it is horribly broken and difficult to use reliably. 404Consequently, autonegotiation is not currently supported for this 405chipset: the driver defaults the NIC to 10baseT half duplex, and it is 406up to the operator to manually select a different mode if necessary. 407(Later cards use an external MII transceiver to implement NWAY 408autonegotiation and work correctly.) 409.Pp 410The 411.Nm 412driver programs 82c168 and 82c169 PNIC chips to use the store and 413forward setting for the transmit start threshold by default. 414This 415is to work around problems with some NIC/PCI bus combinations where 416the PNIC can transmit corrupt frames when operating at 100Mbps, 417probably due to PCI DMA burst transfer errors. 418.Pp 419The 82c168 and 82c169 PNIC chips also have a receiver bug that 420sometimes manifests during periods of heavy receive and transmit 421activity, where the chip will improperly DMA received frames to 422the host. 423The chips appear to upload several kilobytes of garbage 424data along with the received frame data, dirtying several RX buffers 425instead of just the expected one. 426The 427.Nm 428driver detects this condition and will salvage the frame; however, 429it incurs a serious performance penalty in the process. 430.Pp 431The PNIC chips also sometimes generate a transmit underrun error when 432the driver attempts to download the receiver filter setup frame, which 433can result in the receive filter being incorrectly programmed. 434The 435.Nm 436driver will watch for this condition and requeue the setup frame until 437it is transfered successfully. 438.Pp 439The ADMtek AL981 chip (and possibly the AN985 as well) has been observed 440to sometimes wedge on transmit: this appears to happen when the driver 441queues a sequence of frames which cause it to wrap from the end of the 442transmit descriptor ring back to the beginning. 443The 444.Nm 445driver attempts to avoid this condition by not queuing any frames past 446the end of the transmit ring during a single invocation of the 447.Fn dc_start 448routine. 449This workaround has a negligible impact on transmit performance. 450