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