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