View 88E6060-XX-RCJ1I000_7083468.PDF datasheet online --- IC-ON-LINE

Datasheet File OCR Text:

marvell. moving forward faster doc. no. mv-s100952-u0, rev. -- january 3, 2008 document classification: proprietary information link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch
document conventions note: provides related information or information of special importance. caution: indicates potential damage to hardware or software, or loss of data. warning: indicates a risk of personal injury. document status doc status: preliminary te chnical publications: 1.10 for more information, visit our website at: www.marvell.com disclaimer no part of this document may be reproduced or transmitted in any form or by any means, electronic or mechanical, including phot ocopying and recording, for any purpose, without the express written permission of marvell. marvell retains the right to make changes to this document at any time, with out notice. marvell makes no warranty of any kind, expressed or implied, with regard to any information contained in this document, including, but not limited to, the impli ed warranties of merchantability or fitness for any particular purpose. further, marvell does not warrant the accuracy or completeness of the information, text, graphics, or other items contained within this document. marvell products are not designed for use in life-support equipment or applications that would cause a life-threatening situati on if any such products failed. do not use marvell products in these types of equipment or applications. with respect to the products described herein, the user or recipient, in the absence of appropriate u.s. government authorizati on, agrees: 1) not to re-export or release any such information consisting of technology, software or source code controlled for national s ecurity reasons by the u.s. export control regulations ("ear"), to a national of ear country groups d:1 or e:2; 2) not to export the direct product of such technology or such software, to ear country groups d:1 or e:2, if such technology o r software and direct products thereof are controlled for national security reasons by the ear; and, 3) in the case of technology controlled for national security reasons under the ear where the direct product of the technology is a complete plant or component of a plant, not to export to ear country groups d:1 or e:2 the direct product of the plant or major component thereof, if such direct produ ct is controlled for national security reasons by the ear, or is subject to controls under the u.s. munitions list ("usml"). at all times hereunder, the recipient of any such information agrees that they shall be deemed to have manually signed this doc ument in connection with their receipt of any such information. copyright ? 2008. marvell international ltd. all rights reserved . marvell, the marvell logo, moving forward faster, alaska, fas twriter, datacom systems on silicon, libertas, link street, netgx, phyadvantage, prestera, raising the technology bar, the technology within, virtual cable tester, and yukon are registered trademarks of marvell. ants, anyvoltage, discovery, dsp switcher, feroceon, galnet, galt is, horizon, marvell makes it all possible, radlan, unimac, an d vct are trademarks of marvell. all other trademarks are the property of their respective owners. link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0 rev. -- copyright ? 2008 marvell page 2 document classification: proprietary information january 3, 2008, preliminary
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch copyright ? 2008 marvell doc. no. mv-s100952-u0 rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 3 overview the marvell ? 88e6060 is a single chip integration of a complete 6-port fast ethernet switch with support for a cpu connection. it contains five 10base-t/100base- tx transceivers (phys), two that can be used to sup- port 100base-fx; six independent fast ethernet media access controllers (macs), a high-speed non- blocking switch fabric, a high-performance address lookup engine, and a 1/2 megabit frame buffer mem- ory. it is designed for cost-sensitive low port count switch systems and firewall routers. the phy transceivers are designed with marvell vir- tual cable tester ? (vct?) technology for advanced cable diagnostics. vct enables it managers to easily pinpoint the location of cabling issues down to a meter or less, reducing network installation and support costs. the marvell 88e6060 is designed to work in all envi- ronments. true plug-n-play is supported with auto crossover, auto polarity and auto negoti ation in the phys, along with bridge loop prevention (using port states implementing spanning tree support). the shared memory-based swit ch fabric uses the lat- est marvell switch architecture that provides non- blocking switching performance in all traffic environ- ments. back-pressure and pause-frame-based flow control schemes are included to support zero packet loss under temporary traffic congestion. the lookup engine allows for up to 1,024 active nodes to be con- nected with the switch. the sixth port, ?always-on? rmii/mii/sni interface sup- ports a direct connection to management or router cpus with integrated macs. it can be configured in either rmii mode, mii-phy or mii-mac mode or sni mode. these interfaces along with bpdu handling, programmable per port vlan configurations, and port states, support spanning tr ee and truly isolated wan vs. lan firewall applications. the phy units in the marvell 88e6060 are designed with marvell cutting-edge mixed-signal processing technology for digital implementation of adaptive equalization and clock data recovery. special power management techniques are used to facilitate low power dissipation and high port count integration. both the phy and mac units in the marvell 88e6060 com- ply fully with the applicable sections of ieee 802.3, ieee 802.3u, and ieee 802.3x standards. the many operating modes of the marvell 88e6060 can be configured using smi (serial management interface - mdc/mdio) and/ or a low cost serial eeprom (93c46, c56 or c66). ? single chip integration of a 6-port fast ethernet switch in a 14x20 mm 128-pin pqfp package ? integrates six independent media access control- lers fully compliant with the applicable sections of ieee802.3 ? integrates five indepen dent fast ethernet phy transceivers fully compliant with the applicable sections of ieee802.3 ? supports 802.1x implementation with port-based access control ? port based vlans supported in any combination ? supports a cpu header mode for accelerated router performance and wirespeed vlan control ? port states & bpdu handling supports spanning tree ? automatic mdi/mdix cr ossover for 100base-tx and 10base-t ports ? port 0 and port 1 can be configured as copper (100base-tx or 10base-t) or fiber (100base- fx) ? port 5 has dedicated, a lways on, mac mode (for- ward) or phy mode (rev erse) rmii/mii/sni inter- face for management and firewall applications ? port 4 can be either a copper (100base-tx or 10base-t) or an mii interface with the same interface options as port 5 ? each port works at 10 mbps or 100 mbps, full-duplex or half-duplex mode (forced or auto-negotiated) ? back-pressure flow control on half-duplex ports & pause-frame flow control on full-duplex ports ? shared on-chip memory-based switch fabric with true non-blocking switching performance ? high performance lookup engine with support for up to 1,024 mac address entries with automatic learning and aging ? flexible led support for link, speed, duplex mode, collision, and tx/rx activities ? supports a low cost 25 mhz xtal clock source or a 25 mhz or 50 mhz osc. ? pin compatible with the marvell ? 88e6063, 88e6061, 88e6065, 88e6031, and 88e6035 ? low power dissipation p ave = less than 0.6w ? available with commercial grade (88e6060 part) or industrial grade (88e6060-i part) temperature specifications ? integrates virtual circuit tester ? ( vct ?) with each marvell phy features
doc. no. mv-s100952-u0 rev. -- copyright ? 2008 marvell page 4 document classification: proprieta ry information january 3, 2008, preliminary link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch ? firewall router switch w/four 10/100base-t lan ports & one 10/100base-t wan port ? five port switch with spanning tree support ? firewall router switch supporting a fiber wan port ? fiber to copper industrial temperature media converter ? four-port lan switch with two mii connections for wan phy and wireless interface 10base-t 100base-tx/fx transceiver w/auto crossover 10base-t 100base-tx/fx transceiver w/auto crossover 10base-t 100base-tx transceiver w/auto crossover 10base-t 100base-tx transceiver w/auto crossover 10base-t 100base-tx transceiver w/auto crossover rx0 tx0 rx1 tx1 rx2 tx2 rx3 tx3 rx4 tx4 port 0's mac port 1's mac port 2's mac port 3's mac port 4's mac 1/2 mbit embedded memory switch fabric & address database time slot memory controller - data path queue controller cpu/reg interface register loader look-up engine time slot memory controller - address path resetn xtal_in mdc ee_dout xtal_out mdio ee_cs ee_din ee_clk intn clk_sel led controller led0[4:0] led1[4:0] led2[4:0] port 1 fiber enable and controls mac or phy mode mii/rmii/sni interface disable_mii4 port 5's mac mac or phy mode mii/rmii/sni interface enable_mii5 port 0 fiber enable and controls sw_mode[1:0] ledena ledclk ledser 88e6060 top level block diagram applications
table of contents copyright ? 2008 marvell doc. no. mv-s100952-u0 rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 5 table of contents s ection 1. s ignal d escription ................... ................. .............. .............. ........... 14 1.1 88e6060 128-pin pqfp package .............................................................................................14 1.2 pin description............................................................................................................. .............15 s ection 2. a pplication e xamples ............... ................. .............. .............. ........... 34 2.1 examples with the 88e6060 .... ............................................................................................... ..34 2.2 routing with the marvell? header...........................................................................................3 8 s ection 3. f unctional d escription ................. .............. .............. .............. ......... 39 3.1 switch data flow ............................................................................................................ ..........39 3.2 mii/sni/rmii ................................................................................................................ ...............41 3.2.1 mii phy mode .............................................................................................................. .............41 3.2.2 mii mac mode .............................................................................................................. ............42 3.2.3 sni phy mode .............................................................................................................. ............43 3.2.4 rmii phy mode ............................................................................................................. ...........44 3.2.5 rmii/mii/sni configuration ................................................................................................ .......45 3.2.6 enabling the rmii/mii/sni interfaces ...................................................................................... ..45 3.2.7 port status registers ..................................................................................................... ...........45 3.3 media access controllers (mac) ..................... .......................................................................46 3.3.1 backoff ................................................................................................................... ...................46 3.3.2 half-duplex flow control ........................... ....................................................................... ........46 3.3.3 full-duplex flow control .................................................................................................. .........47 3.3.4 forcing flow control .................................. .................................................................... ...........48 3.3.5 statistics counters .................................. ..................................................................... .............48 3.4 address management.......................................................................................................... .....49 3.4.1 address translation unit .................................................................................................. .........49 3.4.2 address searching or translation .................... ...................................................................... ...50 3.4.3 address learning .......................................................................................................... ............51 3.4.4 address aging ............................................................................................................. ..............51 3.4.5 address translation unit operations ....................................................................................... .52 3.5 ingress policy.............................................................................................................. ..............56 3.5.1 port-based vlans.......................................................................................................... ...........57 3.5.2 switching frames back to their source port.............................................................................59 3.5.3 port states............................................................................................................... ..................59 3.5.4 switch?s ingress header (port 4 and port 5 only) .....................................................................59 3.5.5 switch?s ingress trailer (port 4 and port 5 only) .......................................................................60 3.6 queue controller............................................................................................................ ...........62 3.6.1 no head-of-line blocking .... .............................................................................................. .......62
doc. no. mv-s100952-u0 rev. -- copyright ? 2008 marvell page 6 document classification: proprietary information january 3, 2008, preliminary link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch 3.6.2 the queues................................................................................................................ .............. 62 3.6.3 queue manager.......... .............. .............. .............. .............. ........... ........... ........... .......... ........... 62 3.6.4 output queues ........ .............. .............. .............. .............. ........... ........... ............ ........... ............ 63 3.6.5 multicast handler......................................................................................................... ............. 63 3.7 egress policy (port 4 and port 5 only) ................................................................................... 64 3.7.1 switch?s egress header .................................................................................................... ....... 64 3.7.2 switch?s egress trailer .............................. ..................................................................... .......... 65 3.8 spanning tree support....................................................................................................... ..... 66 3.9 embedded memory ............................................................................................................. ..... 66 3.10 interrupt controller....................................................................................................... ............ 66 3.11 port monitoring support ............................. ....................................................................... ...... 67 3.12 port trunking support .......... ................ ................ ................ ................ ................ ............ ....... 67 s ection 4. p hysical i nterface (phy) f unctional d escription ............ ........... 68 4.1 transmit pcs and pma........................................................................................................ .... 71 4.1.1 100base-tx transmitter .................................................................................................... ..... 71 4.1.2 4b/5b encoding............................................................................................................ ............ 71 4.1.3 scrambler ................................................................................................................. ................ 71 4.1.4 nrz to nrzi conversion.................................................................................................... ...... 71 4.1.5 pre-driver and transmit clock ....................... ...................................................................... .... 71 4.1.6 multimode transmit dac .................................................................................................... ..... 71 4.2 receive pcs and pma ......................................................................................................... .... 72 4.2.1 10-base-t/100base-tx receiver..... .............. .............. .............. .............. ........... ............ ...... 72 4.2.2 agc and baseline wander ................................................................................................... ... 72 4.2.3 adc and digital adaptive equaliz er ............ .............. .............. .............. ............ ........... ........... .72 4.2.4 digital phased locked loop (dpll) ........................................................................................ 7 2 4.2.5 nrzi to nrz conversion.................................................................................................... ...... 72 4.2.6 descrambler ............................................................................................................... .............. 72 4.2.7 serial-to-parallel conversion and 5b/4b code-g roup alignment ............... ........... ............ ...... 73 4.2.8 5b/4b decoder ............................................................................................................. ............ 73 4.2.9 setting cable characteristics ...................... ....................................................................... ...... 75 4.2.10 scrambler/descrambler.................................................................................................... ........ 75 4.2.11 digital clock recovery/generator ..................... .................................................................... ... 75 4.2.12 link monitor ............................................................................................................. ................. 75 4.2.13 auto-negotiation......................................................................................................... .............. 76 4.2.14 register update.......................................................................................................... .............. 76 4.2.15 next page support ........................................................................................................ ........... 76 4.2.16 status registers ......................................................................................................... .............. 76 4.3 power management............................................................................................................ ...... 77 4.3.1 low power modes ........................................................................................................... ......... 77 4.3.2 mac interface and phy configuration for low power modes ................................................. 77 4.3.3 ieee power down mode . .............. .............. .............. .............. ............ ........... ........... ........... .... 77 4.4 far end fault indication (f efi) ............................................................................................. .. 78
table of contents copyright ? 2008 marvell doc. no. mv-s100952-u0 rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 7 4.5 virtual cable tester?....................................................................................................... ........78 4.6 auto mdi/mdix crossover ..................................................................................................... ..79 4.7 led interface ............................................................................................................... ..............80 4.7.1 parallel led interface .................................................................................................... ...........80 4.7.2 serial led interface ...................................................................................................... ............82 s ection 5. s erial m anagement i nterface (smi) ............ ............... ........... ......... 87 5.1 mdc/mdio read and write oper ations ..................................................................................87 s ection 6. s witch r egister d escription ............... ................. ................ ........... 89 6.1 register types .............................................................................................................. ............90 6.2 switch core registers ....... ................................................................................................ .......90 6.2.1 switch core register map.... .............................................................................................. .......91 6.2.2 switch port registers ..................................................................................................... ...........92 6.2.3 switch global registers ......... .............. .............. .............. .............. ............ ........... ........... .........99 s ection 7. phy r egisters ................. .............. .............. .............. .............. ....... 106 s ection 8. eeprom p rogramming f ormat ................ .............. .............. ......... 133 8.1 eeprom programming details .. ................ ................ ................ ................ ................ ...........133 s ection 9. e lectrical s pecifications ............. .............. .............. .............. ....... 135 9.1 absolute maximum ratings ...................................................................................................1 35 9.2 recommended operating conditions ..................................................................................136 9.3 package thermal information................................................................................................1 37 9.3.1 thermal conditions for 128-pin pqfp package ... ..................................................................137 9.4 dc electrical characteristic s ............................................................................................... ..138 9.4.1 digital operating conditions.......................... .................................................................... ......139 9.4.2 ieee dc transceiver parameters ...........................................................................................1 41 9.5 ac electrical specifications ................................................................................................ ..142 9.5.1 reset and configuration timing............................................................................................ ..142 9.5.2 clock timing when using a 25 mhz oscillator ........................................................................144 9.5.3 mii receive timing?phy mode.. ...........................................................................................14 5 9.5.4 mii transmit timing?phy mode............................................................................................14 6 9.5.5 mac mode clock timing..................................................................................................... ....147 9.5.6 mii receive timing?mac mode ....................... .....................................................................148 9.5.7 mii transmit timing?mac mode ...........................................................................................149 9.5.8 sni falling edge receive timi ng........................................................................................... .150 9.5.9 sni falling edge transmit timing.......................................................................................... .151 9.5.10 sni rising edge receive timi ng ........................................................................................... .152
doc. no. mv-s100952-u0 rev. -- copyright ? 2008 marvell page 8 document classification: proprietary information january 3, 2008, preliminary link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch 9.5.11 sni rising edge transmit timing .......................................................................................... 153 9.5.12 rmii receive timing using i nclk ......................................................................................... 1 54 9.5.13 rmii transmit timing using inclk ........................................................................................ 1 55 9.5.14 serial led timing........................................................................................................ ........... 156 9.5.15 serial management interface clock timing.......... .............. .............. ........... ........... ............ .... 157 9.5.16 serial management interface ti ming........... .............. .............. .............. .............. ............ ....... 1 58 9.5.17 eeprom timing ............................................................................................................ ........ 159 9.5.18 ieee ac parameters.. .............. .............. .............. .............. ........... ........... ........... ............ ....... 160 s ection 10. m echanical d rawings ................ ................ .............. .............. ......... 161 10.1 128-pin pqfp package ............................... ........................................................................ ... 161 s ection 11. o rdering i nformation ................ ................ .............. .............. ......... 162 11.1 ordering part numbers............................ .......................................................................... .... 162 11.2 package markings ........................................................................................................... ....... 163
list of tables copyright ? 2008 marvell doc. no. mv-s100952-u0 rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 9 list of tables table 1: pin type definitions ..... .............................................................................................. ...........................15 table 2: network interface ...................................................................................................... ............................16 table 3: phy configuration...................................................................................................... ...........................17 table 4: regulator and reference ... ............................................................................................. ......................19 table 5: system ................................................................................................................. .................................19 table 6: register access interface ........................ ...................................................................... .......................20 table 7: serial eeprom interface... .............. .............. .............. .............. .............. .............. ......... ......................21 table 8: port 5?s enable ........................................................................................................ ..............................23 table 9: port 5?s input mii?if enable_mii5 = high ..... .......................................................................... ...........23 table 10: port 5?s output mii?if enable_mii5 = high ..... ........................................................................ ..........24 table 11: port 4?s enable ....................................................................................................... ...............................26 table 12: port 4?s input mii?if disable_mii4 = low.............................................................................. ............26 table 13: port 4?s output mii?if disabl e_mii4 = low ........ .............. .............. .............. .............. ............. ..........27 table 14: switch configuration interface ................. ....................................................................... ......................29 table 15: port status leds ...................................................................................................... .............................30 table 16: power and ground ...................................................................................................... ..........................31 table 18: 88e6060 port configuratio n ............... .............. .............. .............. .............. ............ ......... ......................34 table 19: rmii/mii/sni configuration options ............... ..................................................................... ..................45 table 20: pause frame format .................................................................................................... ........................47 table 21: atu operations registers .............................................................................................. ......................52 table 22: atu data fields ....................................................................................................... .............................53 table 23: atu get next operation register usage................................................................................. .............54 table 24: atu load/purge operation register usage ............................................................................... ..........55 table 25: vlantable settings for figure 14............... ........................................................................ ..................58 table 26: port state options .................................................................................................... .............................59 table 27: ingress trailer fields .......................... ...................................................................... .............................61 table 28: egress trailer fields ................................................................................................. ............................65 table 29: 5b/4b code mapping ............................... ..................................................................... ........................74 table 30: scrambler settings .................................................................................................... ............................75 table 31: operating mode power cons umption ...................................................................................... .............77 table 32: fefi select........................................................................................................... .................................78 table 33: mdi/mdix pin functions ................................................................................................ .......................79 table 34: parallel led hardware defaults............... ......................................................................... ....................80 table 35: parallel led display interpretation........... ........................................................................ .....................81 table 36: serial led display options (a = active) ..... .......................................................................... ................83 table 37: single led display mode............................................................................................... .......................84 table 38: dual led display mode ................................................................................................. .......................85 table 39: serial management interface protocol exampl e .......................................................................... .........88 table 40: register types ........................................................................................................ ..............................90 table 41: switch core register map. ............................................................................................. .......................91
doc. no. mv-s100952-u0 rev. -- copyright ? 2008 marvell page 10 document classification: proprietary information january 3, 2008, preliminary link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch table 42: port status register .................................................................................................. ........................... 92 table 43: switch identifier register ...................... ...................................................................... .......................... 93 table 44: port control register ................................................................................................. ........................... 94 table 45: port based vlan map .............................. ..................................................................... ...................... 96 table 46: port association vector ............................................................................................... ......................... 97 table 47: rx counter ............................................................................................................ ............................... 98 table 48: tx counter............................................................................................................ ................................ 98 table 49: switch global status register ................. ........................................................................ ..................... 99 table 50: switch mac address register bytes 0 & 1 ..... .......................................................................... ......... 100 table 51: switch mac address register bytes 2 & 3 ..... .......................................................................... ......... 100 table 52: switch mac address register bytes 4 & 5 ..... .......................................................................... ......... 100 table 53: switch global control register................ ........................................................................ ................... 101 table 54: atu control register .................................................................................................. ....................... 102 table 55: atu operation register ................................................................................................ ..................... 103 table 56: atu data register .................................. ................................................................... ........................ 104 table 57: atu switch mac address register bytes 0 & 1 ........................................................................... ..... 104 table 58: atu switch mac address register bytes 2 & 3 ........................................................................... ..... 104 table 59: atu switch mac address register bytes 4 & 5 ........................................................................... ..... 105 table 60: phy register map ...................................................................................................... ........................ 106 table 61: phy control register .................................................................................................. ....................... 107 table 62: phy status register................................................................................................... ........................ 109 table 63: phy identifier ........................................................................................................ ............................. 110 table 64: phy identifier ........................................................................................................ ............................. 110 table 65: auto-negotiation advertisement register ....... ........................................................................ ........... 111 table 66: link partner ability register (base page) .... ......................................................................... ............. 113 table 67: link partner ability register (next page) .. ........................................................................... .............. 113 table 68: auto-negotiation expansion register.............. ..................................................................... .............. 114 table 69: next page transmit regist er ........................................................................................... .................. 115 table 70: link partner next page register ............... ........................................................................ ................. 115 table 71: phy specific control register................... ...................................................................... ................... 116 table 72: phy specific status register .......................................................................................... ................... 119 table 73: phy interrupt enable.................................................................................................. ........................ 121 table 74: phy interrupt status.................................................................................................. ......................... 122 table 75: phy interrupt port summary (global) .......... ......................................................................... ............. 123 table 76: receive error counter ...... ........................................................................................... ....................... 124 table 77: led parallel select register (global) ...... .......................................................................... ............... 125 table 78: led stream select for seri al leds ..................................................................................... ............... 126 table 79: phy led control register (global) ..................................................................................... .............. 128 table 80: phy manual led override........................... .................................................................... .................. 129 table 81: vct? register for txp/n pins................. ......................................................................... ................ 130 table 82: vct? register for rxp/n pins................. ......................................................................... ................ 131 table 83: phy specific control register ii.............. ........................................................................ ................... 132 table 84: absolute maximum ratings . ............................................................................................. .................. 135 table 85: recommended operating conditions ................ ...................................................................... .......... 136
list of tables copyright ? 2008 marvell doc. no. mv-s100952-u0 rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 11 table 86: thermal conditions for 128-pin pqfp package .... .............. .............. ........... ........... ............ ......... ......137 table 87: dc electrical characteristics ................. ........................................................................ ......................138 table 88: digital operating conditions.................. ........................................................................ ......................139 table 89: internal resistor description ......................................................................................... ......................140 table 90: ieee dc transceiver paramete rs............ .............. .............. .............. ........... ........... ............ ...............141 table 91: reset and configuration timing ........................................................................................ ..................142 table 92: clock timing with a 25 mhz oscillator ................................................................................. ...............144 table 93: mii receive timing?phy mode.................... ....................................................................... ..............145 table 94: mii transmit timing?phy mode.......................................................................................... ..............146 table 95: mac mode clock timing............................. .................................................................... ....................147 table 96: mii receive timing?mac mode ................... ........................................................................ .............148 table 97: mii transmit timing?mac mode .......................................................................................... .............149 table 98: sni falling edge receive timing................ ....................................................................... .................150 table 99: sni falling edge transmit timing........... .............. .............. .............. ........... ........... .......... .................151 table 100:sni rising edge receive ti ming ....................................................................................... .................152 table 101:sni rising edge transmit timing ............... ....................................................................... .................153 table 102:rmii receive timing usi ng inclk ...................................................................................... ................154 table 103:rmii transmit timing using inclk ..................................................................................... ................155 table 104:serial led timing .............................. ...................................................................... ...........................156 table 105:serial management interface clock timing.... ......................................................................... ............157 table 106:serial management interface timing........... ........................................................................ ................158 table 107:eeprom timing .... .............. .............. .............. .............. ............ ........... ........... .......... .........................159 table 108:ieee ac paramete rs............ .............. .............. .............. ............ ........... ........... ........... ........................160 table 109:part order option - commercial....................................................................................... ...................162 table 110:part order option - industrial ....................................................................................... .......................162
doc. no. mv-s100952-u0 rev. -- copyright ? 2008 marvell page 12 document classification: proprietary information january 3, 2008, preliminary link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch list of figures figure 1: 88e6060 128-pin pqfp package (top view) ..... .......................................................................... ..... 14 figure 2: 88e6060 firewall router ex ample ....................................................................................... .............. 35 figure 3: firewall router switch supporting a fiber wan port.................................................................... ...... 36 figure 4: firewall router switch with lan ports and wa n phy..................................................................... .. 37 figure 5: 88e6060 device switch data flow................ ....................................................................... .............. 39 figure 6: switch operation........ .............................................................................................. ........................... 40 figure 7: mii phy interface pins........................... ..................................................................... ........................ 41 figure 8: mii mac interface pins ................................................................................................ ....................... 42 figure 9: sni phy interface pins.......................... ...................................................................... ....................... 43 figure 10: rmii phy interface pins using inclk........ .......................................................................... .............. 44 figure 11: atu size tradeoffs............................. ...................................................................... .......................... 50 figure 12: format of an atu entry ......................... ...................................................................... ....................... 53 figure 13: switch operation with vlans disabled ......... ........................................................................ ............. 57 figure 14: switch operation with a typical router vlan configuration ............................................................ .58 figure 15: ingress trailer format.. ............................................................................................. .......................... 60 figure 16: switch queues........................................................................................................ ............................ 62 figure 17: egress trailer fo rmat ................................................................................................ ......................... 65 figure 18: 88e6060 device transmit block diagram ........... ..................................................................... .......... 69 figure 19: 88e6060 device receive block diagram ............ ..................................................................... .......... 70 figure 20: serial ledena high clocking with colx in dual mode, error off, and duplex in single mode...................................... .................................................................... ........... 82 figure 21: serial led conversion........................... ..................................................................... ........................ 83 figure 22: serial led display order?(if all are select ed)....................................................................... ............ 83 figure 23: typical mdc/mdio read operation................. ..................................................................... ............. 88 figure 24: typical mdc/mdio write operation..................................................................................... .............. 88 figure 25: 88e6060 register map ............................ ..................................................................... ...................... 89 figure 26: cable fault distance trend line ............. ......................................................................... ................ 131 figure 27: eeprom data format ............................... .................................................................... .................. 134 figure 28: reset and configuration ti ming ............ .............. .............. .............. ........... ........... ........... ................ 143 figure 29: oscillator clock timing .......................... .................................................................... ....................... 144 figure 30: phy mode mii receive timi ng.......................................................................................... ............... 145 figure 31: phy mode mii transmit timi ng......................................................................................... ............... 146 figure 32: mac clock timing ................................. .................................................................... ....................... 147 figure 33: mac mode mii receive timing .......................................................................................... .............. 148 figure 34: mac mode mii transmit timing .................... ..................................................................... .............. 149 figure 35: sni falling edge receive timing ............... ....................................................................... ............... 150 figure 36: sni falling edge transmit timing ..................................................................................... ............... 151 figure 37: sni rising edge receive timing....................................................................................... ............... 152 figure 38: sni rising edge transmit timing............... ....................................................................... ............... 153 figure 39: phy mode rmii receive timing using inclk .... ......................................................................... .... 154
list of figures copyright ? 2008 marvell doc. no. mv-s100952-u0 rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 13 figure 40: phy mode rmii transmit timing using inclk............................................................................ .... 155 figure 41: serial led timing.................................................................................................... ......................... 156 figure 42: serial management interface clock timing.... ......................................................................... ......... 157 figure 43: serial management interface timing............ ....................................................................... ............. 158 figure 44: eeprom timing ..... .............. .............. .............. .............. ........... ............ ........... ......... ...................... 159 figure 45: sample part number ................................................................................................... ..................... 162 figure 46: 88e6060 128-pin pqfp commercial rohs 6/6 pa ckage marking and pin 1 loc ation ........ ........... 163 figure 47: 88e6060 128-pin pqfp indus trial rohs 6/6 package marking and pin 1 location....... ........... ...... 163
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 14 document classification: proprietary information january 3, 2008, preliminary section 1. signal description 1.1 88e6060 128-pin pqfp package figure 1: 88e6060 128-pin pqfp package (top view) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 102 101 100 99 98 97 96 95 94 93 92 91 90 89 88 87 86 85 84 83 82 81 80 79 78 77 76 75 74 73 72 71 70 69 68 67 66 65 88e6060 top view 128 127 126 125 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 105 104 103 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 vss xtal_in rsetn config_a vddo p1_config vss resetn p4_ind3 vdd p4_ind2 vss p4_ind1 p4_ind0 nc nc p0_sdet vddo control_15 rsetp p1_sdet vss fd_flow_dis intn nc vss ee_dout vdd p0_led0 p0_led1 p0_led2 p1_led0 vss p1_led1 p1_led2 p2_led0 p2_led1 vddo p2_led2 p3_led0 p3_led1 p3_led2 vss p4_led0 p4_led1 p4_led2 p4_inclk vdd p4_outdv p4_outclk p4_outd0/p4_mode0 p4_outd1/p4_mode1 p4_outd2/p4_mode2 p4_outd3/p4_mode3 p4_col config_b vss p4_crs vdd disable_mii4 nc p5_indv p5_ind0 p5_ind1 p5_ind2 p5_ind3 p5_inclk p5_outdv vss p5_outclk vddo p5_outd0/p5_mode0 p5_outd1/p5_mode1 p5_outd2/p5_mode2 p5_outd3/p5_mode3 p5_col p5_crs mdio mdc ee_din/hd_flow_dis vss vdd vss control_25 vddah p0_rxp p0_rxn vddal p0_txp p0_txn p1_txn p1_txp vddal p1_rxn p1_rxp vddah p2_rxp p2_rxn vddal p2_txp p2_txn vss vss ledser ledclk vdd sw_mode1 sw_mode0 p3_txn p3_txp vddal p3_rxn p3_rxp vddah p4_rxp p4_rxn vddal p4_txp p4_txn xtal_out nc p4_indv p0_config clk_sel enable_mii5 ee_clk/addr4 ee_cs/ee_1k ledena
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 15 1.2 pin description table 1: pin type definitions pin type definition i/o input and output input input only output output only open drain open drain output analog analog typically input may be input or output depending upon operating state; usually input typically output may be output or input depending upon operating state; usually output power device voltage and current supply ground device current return nc no connection, do not connect anything to these pins
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 16 document classification: proprietary information january 3, 2008, preliminary table 2: network interface 128-pqfp package pin # pin name pin type description 27 25 15 13 4 p[4:0]_rxp typically input output if in crossover mode receiver input ? positive. p[4:0]_rxp connects directly to the receiver magnetics. if the port is configured for 100base-fx mode (ports 0 and 1 only) rxp connects directly to the fiber- optic receiver?s positive output. these pins can become out- puts if auto mdi/mdix crossover is enabled (see section 4.6 ). if a phy port is not used, the rxp pins should be tied to vss. 28 24 16 12 5 p[4:0]_rxn typically input output if in crossover mode receiver input ? negative. p[4:0]_rxn connects directly to the receiver magnetics. if the port is configured for 100base- fx mode (ports 0 and 1 only) rxn connects directly to the fiber-optic receiver?s negative output. these pins can become outputs if auto mdi/mdix crossover is enabled (see section 4.6 ). if a phy port is not used, the rxn pins should be tied to vss. 30 22 18 10 7 p[4:0]_txp typically output input if in crossover mode transmitter output ? positive. p[4:0]_txp connects directly to the transmitter magnetics. if the port is configured for 100base-fx mode (ports 0 and 1 only) txp connects directly to the fiber-optic transmitter?s positive input. these pins can become inputs if auto mdi/mdix crossover is enabled (see section 4.6 ). if a phy port is not used, the txp pins should be tied to vss. 31 21 19 9 8 p[4:0]_txn typically output input if in crossover mode transmitter output ? negative. p[4:0]_txn connects directly to the transmitter magnetics. if the port is configured for 100base-fx mode (ports 0 and 1 only) txn connects directly to the fiber-optic transmitter?s negative input. these pins can become inputs if auto mdi/mdix crossover is enabled (see section 4.6 ). if a phy port is not used, the txn pins should be tied to vss. 127 122 p[1:0]_sdet input signal detect input. if port 0 and/or 1 is configured for 100base-fx mode sdet indicates whether a signal is detected by the fiber-optic transceiver. a positive level indi- cates that a signal is detected. if port 0 and/or 1 is configured for 10/100base-t mode sdet is not used, but cannot be left floating since these pins do not contain internal resistors. sdet must be tied to vss or vddo either directly or through a 4.7 k resistor.
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 17 table 3: phy configuration 128-pqfp package pin # pin name pin type description 108 111 p[1:0]_config input port 0 and 1 configuration. the config pin is used to set the default configuration for port 0 and 1 by connecting these pins to other device pins as follows: vss = auto-negotiation enabled - default p0_led1 = forced 10base-t half-duplex p0_led2 = forced 10base-t full-duplex p1_led0 = forced 100base-tx half-duplex p1_led1 = forced 100base-tx full-duplex p1_led2 = forced 100base-fx half-duplex vddo = forced 100base-fx full-duplex ports 2, 3 and 4?s default configuration is auto-negotiation enabled. any port?s default configuration can be modified by accessing the phy registers by a cpu or a serial eeprom. fiber vs. copper mode cannot be configured in this way. how- ever, fiber vs. copper must be selected at reset by using these pins. the config pins are configured after reset and contain inter- nal pull-down resistors so they can be left floating to select auto-negotiation. 106 config_a input global configuration a. this global configuration pin is used to set the default led mode and far end fault indication (fefi) mode for 100base-fx by connecting these pins to other device pins as follows: vss = led mode 0, fefi disabled p0_led0 = led mode 0, fefi enabled p0_led1 = led mode 1, fefi disabled p0_led2 = led mode 1, fefi enabled p1_led0 = led mode 2, fefi disabled p1_led1 = led mode 2, fefi enabled p1_led2 = led mode 3, fefi disabled vddo = led mode 3, fefi enabled - default the led modes are covered in section 4.7.1 and fefi is cov- ered in section 4.2.8.3 . the config_a pin is configured after reset and contains an internal pull-up resistor.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 18 document classification: proprietary information january 3, 2008, preliminary 93 config_b input global configuration b. this global configuration pin is used to set the default mode for auto crossover, the phy driver type, and energy detect by connecting these pins to other device pins as follows: vss = no crossover, class a 1 drivers, energy detect dis- abled p0_led0 = no crossover, class a drivers, energy detect enabled p0_led1 = no crossover, class b 2 drivers, energydetect disabled p0_led2 = no crossover, class b drivers, energy detect enabled p1_led0 = auto crossover, cla ss a drivers, energy detect disabled p1_led1 = auto crossover, cla ss a drivers, energy detect enabled p1_led2 = auto crossover, cla ss b drivers, energy detect disabled vddo = auto crossover, cla ss b drivers, energy detect enabled?default auto crossover is covered in section 4.6 , class b vs. class a drivers are covered in table 83 on page 132 and energy detect is covered in section table 71 through table 74. the config_b pin is configured after reset and contains an internal pull-up resistor. 1. a class a driver is available for 100base-tx mode only and ty pically used in backplane or direct connect applications. 2. a class b driver is typically used in cat 5 applications. table 3: phy configuration (continued) 128-pqfp package pin # pin name pin type description
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 19 table 4: regulator and reference 128-pqfp package pin # pin name pin type description 126 rsetp analog resistor reference. a 2 k 1% resistor is placed between the rsetp and rsetn. this resistor is used to set an internal bias reference current. 125 rsetn analog resistor reference. a 2 k 1% resistor is placed between the rsetn and rsetp. this resistor is used to set an internal bias reference current. 124 control_15 analog voltage control to external 1.5v regulator. this signal con- trols an external pnp transistor to generate the 1.5v power supply for the vdd and vddal pins. 2 control_25 analog voltage control to external 2.5v regulator. this signal con- trols an external pnp transistor to generate the 2.5v power supply for the vddah pins. table 5: system 128-pqfp package pin # pin name pin type description 104 xtal_in input 25 mhz or 50 mhz syst em reference clock input. the fre- quency of this clock input is selected by the clk_sel pin. the clock source can come from a crystal (25 mhz only) or an oscillator (25 or 50 mhz). this is the only clock required as it is used for both the switch and the phys. 105 xtal_out output system reference clock output. this output can only be used to drive an external crystal (25 mhz only). it cannot be used to drive external logic. if an oscillator is connected to xtal_in this pin should be left unconnected. 68 clk_sel input clock frequency select. connect this pin to vss if xtal_in is 25 mhz. connect this pin to vddo or leave it unconnected if xtal_in is 50 mhz. this pin must be stable before and after reset. clk_sel is internally pulled high via a resistor. 110 resetn input hardware reset. active low. the 88e6060 is configured dur- ing reset. when resetn is low all configuration pins become inputs and the value seen on these pins is latched on the ris- ing edge of resetn or some time after.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 20 document classification: proprietary information january 3, 2008, preliminary table 6: register access interface 128-pqfp package pin # pin name pin type description 70 mdc input mdc is the management data clock reference for the serial management interface (smi). a continuous clock stream is not expected. the maximum frequency supported is 8.3 mhz. the smi is used to access the registers in the phy and in the switch if the serial eeprom is not accessing the registers. it is available in all combinations of sw_mode[1:0]. mdc is internally pulled high via a resistor. 71 mdio i/o mdio is the management data input/output for the smi. mdio is used to transfer management data in and out of the device synchronously with mdc. this pin requires an external pull-up resistor in the range of 4.7 k to 10 k . the 88e6060 device uses 16 of the 32 possible smi port addresses. the 16 that are used are selectable using the ee_clk/addr4 pin. mdio is internally pulled high via a resistor. 63 intn open drain output intn is an active low, open drain pin that is asserted to indi- cate an unmasked interrupt event occurred. a single external pull-up resistor is required to achieve a logic high when this signal is inactive. the intn pin is asserted active low if sw_mode[1:0] (see table 14) bits are not 0b10 (standalone mode) and the eeprom data has been completely read into the device. this eeprom done interrupt indicates to any attached cpu that it may use the mdc/mdio lines to access the internal registers because the eeprom has finished using the registers. this pin also goes low when any other unmasked interrupt becomes active inside the device.
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 21 table 7: serial eeprom interface 128-pqfp package pin # pin name pin type description 61 ee_cs /ee_1k i/o input during reset serial eeprom chip select. ee_cs is the serial eeprom chip select referenced to ee_clk. it is used to enable the external eeprom (if present), and to delineate each data transfer. ee_cs is a multi-function pin used to configure the 88e6060 during a hardware reset. when reset is asserted, ee_cs becomes an input and the desired eeprom type configura- tion is latched at the rising edge of resetn as follows: low = use 8-bit addresses (for 2k bit 93c56 & 4k bit 93c66) high = use 6-bit addresses (for 1k bit 93c46) the external eeprom must be configured in the x16 organi- zation. ee_cs is internally pulled high via a resistor. use a 4.7 k resistor to vss for a configuration low. 60 ee_clk /addr4 i/o output during eeprom loading serial eeprom clock. ee_clk is the serial eeprom clock reference output by the 88e6060. it is used to shift the exter- nal serial eeprom (if installed) to the next data bit so the default values of the internal registers can be overridden. ee_clk is a multi-function pin used to configure the 88e6060 during a hardware reset. when reset is asserted, ee_clk becomes an input and the desired smi addr4 address space configuration (table 5) is latched at the rising edge of resetn as follows: low = use smi device addresses 0x00 to 0x0f high = use smi device addresses 0x10 to 0x1f ee_clk is internally pulled high via a resistor. use a 4.7 k resistor to vss for a configuration low.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 22 document classification: proprietary information january 3, 2008, preliminary 69 ee_din/ hd_flow_dis typically input output during eeprom loading serial eeprom data into the eeprom device. ee_din is serial eeprom data referenced to ee_clk used to transmit the eeprom command and address to the external serial eeprom (if present). ee_din is a multifunction pin used to configure the 88e6060 during a hardware reset. when reset is asserted, ee_din becomes an input and the configured half-duplex flow con- trol disable value is latched at the rising edge of reset as fol- lows: low = enable ?forced collision? flow control on all half- duplex ports high = disable flow control on all half-duplex ports ee_din is internally pulled high via a resistor. use a 4.7 k resistor to vss for a configuration low. fd_flow_dis is used to select flow control on full-duplex ports (see table 13). 58 ee_dout input serial eeprom data out from the eeprom device. ee_dout is serial eeprom data referenced to ee_clk used to receive the eeprom configuration data from the external serial eeprom (if present). ee_dout is internally pulled high via a resistor table 7: serial eeprom interface (continued) 128-pqfp package pin # pin name pin type description
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 23 table 8: port 5?s enable 128-pqfp package pin # pin name pin type description 66 enable_mii5 input enable mii port 5. this pin is used to enable port 5. a high enables the port. a low disables port 5 (i.e., the drivers are tri-stated). enable_p5 is internally pulled high via a resistor. table 9: port 5?s input mii?if enable_mii5 = high 128-pqfp package pin # pin name pin type description 82 p5_inclk i/o input clock. p5_inclk is a reference for p5_indv and p5_ind[3:0]. the direction and speed of p5_inclk is deter- mined by p5_mode[3:0] at the end of reset. if the port is in phy mode, p5_inclk is an output. in this mode the frequency of the clock is 25 mhz if the port is in 100base-x mode, 2.5 mhz if the port is in 10base-t mode and 50 mhz for rmii mode. if the port is in mac mode, p5_inclk is an input. in this mode the frequency of the clock can be anywhere from dc to 25 mhz although it should be 25 mhz for 100base-x mode and 2.5 mhz for 10base-t mode. p5_inclk is tri-stated during reset and it is internally pulled high. 83 84 85 86 p5_ind[3:0] input input data. p5_ind[3:0] receives the data nibble to be trans- mitted into the switch in 100base-x and 10base-t modes. p5_ind[3:0] is synchronous to p5_inclk. these pins are inputs regardless of the port?s mode (i.e., phy mode or mac mode). only p5_ind0 is used when sni mode is selected. p5_ind[1:0] are used when rmii mode is selected 1 . p5_ind[3:0] are internally pulled high via resistor. 1. when rmii mode is selected p5_ind[1:0] are synchronous to p5_outclk or p5_inclk which are phase shifted to each other. 87 p5_indv input input data valid. when p5_indv is asserted high, data on p5_ind[3:0] is accepted into the switch. p5_indv must be synchronous to p5_inclk for sni and mii operation. it must be synchronous to p5_outclk or p5_inclk in rmii opera- tion. p5_indv is internally pulled low via resistor.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 24 document classification: proprietary information january 3, 2008, preliminary table 10: port 5?s output mii?if enable_mii5 = high 128-pqfp package pin # pin type pin name description 79 p5_outclk i/o output clock. p5_outclk is a reference for p5_outdv and p5_outd[3:0]. the direction and speed of p5_outclk is determined by p5_mode[3:0] at the end of reset. if the port is in phy mode, p5_outclk is an output. in this mode the frequency of the clock will be 25 mhz if the port is in 100base-x mode, and 2.5 mhz if the port is in 10base-t mode and 50 mhz for rmii mode. if the port is in mac mode, p5_outclk is an input. in this mode the frequency of the clock can be anywhere from dc to 25mhz although it should be 25 mhz for 100base-x mode and 2.5 mhz for 10base-t mode. p5_outclk is tri-stated during reset and it is internally pulled high. 74 75 76 77 p5_outd[3:0] /p5_mode[3:0] normally output input only when resetn is low output data. data transmitted from the switch is decoded and presented on p5_outd[3:0] pins synchronous to p5_out_clk. these pins are outputs regardless of the port?s mode (i.e., phy or mac mode). only p5_outd0 con- tains meaningful data when sni mode is selected. p5_outd[1:0] are used when rmii mode 1 is selected. during reset, these internally pulled high pins are tri-stated and used to latch in the required operating mode for the port (see section. 3.2.5 ). 1. p5_outd[1:0] can be synchronous to either p5_outclk or p5_inclk, which are phase shifted to each other. 81 p5_outdv output output data valid. when p5_outdv is asserted high, data transmitted from the switch on p5_outd[3:0] is valid. p5_outdv is synchronous with p5_outclk in mii mode. when rmii mode is selected, p5_outdv can be synchro- nous to either p5_inclk or p5_outclk. p5_outdv is tri-stated during reset and it is internally pulled high. 72 p5_crs i/o carrier sense. after reset, p5_crs becomes an output if phy mode is selected for this port. it remains an input if mac mode is selected. p5_crs asserts (or is expected to be asserted) when the receive data path is non-idle. in half- duplex mode p5_crs is also asserted (or is expected to be asserted) during transmission. p5_crs is asynchronous to p5_outclk and p5_inclk. p5_crs is tri-stated during reset and it is internally pulled low so the pin can be left unconnected if not used.
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 25 table 10: port 5?s output mii?if enable_mii5 = high (continued) 128-pqfp package pin # pin type pin name description 73 p5_col i/o collision. after reset, p5_col becomes an output if phy mode is selected for this port. it remains an input if mac mode is selected. in phy mode, p5_col asserts when both the transmit and receive paths are non-idle in both half and full-duplex modes. in half-duplex, mac mode p5_col is expected to be asserted when both the transmit and receive paths are non-idle. in full-duplex mac mode, p5_col is ignored. p5_col is asynchronous with p5_outclk and p5_inclk. p5_col is tri-stated during reset and it is internally pulled low.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 26 document classification: proprietary information january 3, 2008, preliminary table 11: port 4?s enable 128-pqfp package pin # pin name pin type description 89 disable_mii4 input disable mii port 4. this pin is used to disable port 4?s mii interface drivers. a high disables port 4?s mii interface driv- ers?s mii interface (i.e., the drivers are tri-stated) and enables port 4?s phy interface (its mdi pins). a low enables port 4?s mii interface and its drivers and disables port 4?s phy inter- face. disable_mii4 is internally pulled high via a resistor. table 12: port 4?s input mii?if disable_mii4 = low 128-pqfp package pin # pin name pin type description 102 p4_inclk i/o input clock. p4_inclk is a reference for p4_indv and p4_ind[3:0]. the direction and speed of p4_inclk is deter- mined by p4_mode[3:0] at the end of reset. if the port is in phy mode, p4_inclk is an output. in this mode the frequency of the clock will be 25 mhz if the port is in 100base-x mode, and 2.5 mhz if the port is in 10base-t mode and 50 mhz for rmii mode. if the port is in mac mode, p4_inclk is an input. in this mode the frequency of the clock can be anywhere from dc to 25 mhz although it should be 25 mhz for 100base-x mode and 2.5 mhz for 10base-t mode. p4_inclk is tri-stated during reset and it is internally pulled high. 112 114 116 117 p4_ind[3:0] input input data. p4_ind[3:0] receives a data nibble to be transmitted into the switch in 100base-x and 10base-t modes. p4_ind[3:0] is synchronous with p4_inclk. these pins are inputs regardless of the port?s mode (i.e., phy or mac mode). only p4_ind0 is used when sni mode is selected. p4_ind[1:0] are used when rmii mode is selected 1 . p4_ind[3:0] are internally pulled high via resistor. 1. when rmii mode is selected, p4_ind[1:0] are synchronous to p4_outclk or p4_inclk, which are phase shifted to each other. 118 p4_indv input input data valid. when p4_indv is asserted high, data on p4_ind[3:0] is accepted into the switch. p4_indv must be syn- chronous to p4_inclk for sni and mii operation. it must be synchronous to p4_outclk or p4_inclk in rmii operation. p4_indv is internally pulled low via resistor.
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 27 table 13: port 4?s output mii?if disable_mii4 = low 128-pqfp package pin # pin name pin type description 99 p4_outclk i/o output clock. p4_outclk is a reference for p4_outdv and p4_outd[3:0]. the direction and speed of p4_outclk is determined by p4_mode[3:0] at the end of reset. if the port is in phy mode, p4_outclk is an output. in this mode the frequency of the clock will be 25 mhz if the port is in 100base-x mode, and 2.5 mhz if the port is in 10base-t mode and 50 mhz for rmii mode. if the port is in mac mode, p4_outclk is an input. in this mode the frequency of the clock can be anywhere from dc to 25mhz although it should be 25 mhz for 100base-x mode and 2.5 mhz for 10base-t mode. p4_outclk is tri-stated during reset and it is internally pulled high. 95 96 97 98 p4_outd[3:0] /p4_mode[3:0] normally output input only when resetn is low output data. data transmitted from the switch is decoded and presented on p4_outd[3:0] pins synchronous to p4_out_clk. these pins are outputs regardless of the port?s mode (i.e., phy or mac mode). only p4_outd0 con- tains data when sni mode is selected. p4_outd[1:0] are used when rmii mode 1 is selected. during reset these internally pulled high pins are tri-stated and used to latch in the desired operating mode for the port (see section 3.2.5 ). 1. p4_outd[1:0] can be synchronous to either p4_outclk or p4_inclk, which are phase shifted to each other. 100 p4_outdv output output data valid. when p4_outdv is asserted high, data transmitted from the switch on p4_outd[3:0] is valid. p4_outdv is synchronous with p4_outclk in mii mode. when rmii mode is selected, p4_outdv can be synchro- nous to either p5_inclk or p4_outclk. p4_outdv is tri-stated during reset and it is internally pulled high. 91 p4_crs i/o carrier sense. after reset, p4_crs becomes an output if phy mode is selected for this port. it remains an input if mac mode is selected. p4_crs asserts (or is expected to be asserted) when the receive data path is non-idle. in half- duplex mode p4_crs is also asserted (or is expected to be asserted) during transmission. p4_crs is asynchronous to p4_outclk and p4_inclk. p4_crs is tri-stated during reset and it is internally pulled low.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 28 document classification: proprietary information january 3, 2008, preliminary table 13: port 4?s output mii?if disable_mii4 = low (continued) 128-pqfp package pin # pin name pin type description 94 p4_col i/o collision. after reset, p4_col becomes an output if phy mode is selected for this port. it remains an input if mac mode is selected. in phy mode, p4_col asserts when both the transmit and receive paths are non-idle in both half and full-duplex modes. in half-duplex, mac mode p4_col is expected to be asserted when both the transmit and receive paths are non-idle. in full-duplex mac mode, p4_col is ignored. p4_col is asynchronous to p4_outclk and p4_inclk. p4_col is tri-stated during reset and it is internally pulled low.
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 29 table 14: switch configuration interface 128-pqfp package pin # pin name pin type description 34 33 sw_mode[1:0] input switch mode. these pins are used to configure the 88e6060 after reset. switch mode pins work as follows: 1 0 description 0 0 cpu attached mode ? ports come up disabled 1 0 1 reserved 1 0 stand alone mode - ignore eeprom 1 1 eeprom attached mode 2 the eeprom attached mode (when both sw_mode pins = high) can be used with a cpu. in all but the standalone modes, the intn pin asserts active low after the eeprom completes initializing the internal registers (i.e. a halt com- mand has been executed) 3 . sw_mode[1:0] are not latched on the rising edge of reset and they must remain static for proper device operation. they are internally pulled high via resistors. 1. the ports come up in the disabled port state in the cpu mode so the cpu can perform bridge loop detection on link up and/or per- form switch initialization/configuration prior to letting packets flow. 2. in eeprom attached mode the ports come up in the forwarding port state unless the port control register is overwritten by th e eeprom data. 3. the intn pin is activated so the cpu can access the register s using the mdc/mdio pins. the cpu cannot access these registers as long as the eeprom is still being executed. 64 fd_flow_dis input full-duplex flow control disable. high = disable flow control on all full-duplex ports low = enable ieee 802.3x pause based flow control on all supported full-duplex ports full-duplex flow control requires support from the end station. full-duplex flow control is supported on any full-duplex port that has auto-negotiation enabled, advertises that it supports pause (i.e., fd_flow_dis = low at reset), and sees that the end station supports pause as well (from data returned during auto-negotiation). if any of these requirements are not met the port does not generate pause frames and it does not pause when the port receives pause frames. fd_flow_dis is latched on the rising edge of reset into all the port?s phy auto-negotiation advertisement registers. it is internally pulled high via a resistor. the ee_din/hd_flow_dis multi-function pin is used to select flow control on half-duplex ports (see table 7)
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 30 document classification: proprietary information january 3, 2008, preliminary table 15: port status leds 128-pqfp package pin # pin name pin type description 39 43 46 50 54 p[4:0]_led2 output parallel led outputs ? one for each port. this active low led pin directly drives an led in parallel led mode. it can be con- figured to display many options. p[4:0]_led2 are driven active low whenever resetn is asserted. 40 44 48 51 55 p[4:0]_led1 output parallel led outputs ? one for each port. this active low led pin directly drives an led in parallel led mode. it can be con- figured to display many options. p[4:0]_led1 are driven active low whenever resetn is asserted. 41 45 49 53 56 p[4:0]_led0 output parallel led outputs ? one for each port. this active low led pin directly drives an led in parallel led mode. it can be con- figured to display many options. p[4:0]_led0 are driven active low whenever resetn is asserted. 36 ledser output ledser outputs serial status bits that can be shifted into a shift register to be displayed via leds. ledser is output synchro- nously to ledclk. 38 ledena output ledena asserts high whenever ledser has valid status that is to be stored into the shift register. ledena is output synchro- nously to ledclk. 37 ledclk output ledclk is the reference clock for the serial led signals.
signal description pin description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 31 table 16: power and ground 128-pqfp package pin # pin name pin type description 47 78 107 123 vddo power power to p4 and p5 miis, led, eeprom, and i/o drivers. vddo must be connected to 3.3v. 3 14 26 vddah power 2.5 volt analog power to phys 6 11 17 23 29 vddal power 1.5 volt power to analog core 35 57 65 90 101 113 vdd power 1.5 volt power to digital core 1 20 32 42 52 59 67 80 92 103 109 115 128 vss ground ground to device 62 88 119 120 121 nc - no connect. these pins must be left unconnected.
doc. no. mv-s100952-u0 rev. -- copyright ? 2008 marvell page 32 document classification: proprieta ry information january 3, 2008, preliminary link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch 1.3 128-pin lqfp pin assignment list?by signal name table 17: package pin list?alp habetical by signal name pin number pin name 68 clk_sel 106 config_a 93 config_b 124 control_15 2 control_25 89 disable_mii4 60 ee_clk/addr4 61 ee_cs/ee_1k 69 ee_din/hd_flow_dis 58 ee_dout 66 enable_mii5 64 fd_flow_dis 63 intn 37 ledclk 38 ledena 36 ledser 70 mdc 71 mdio 62 nc 88 nc 119 nc 120 nc 121 nc 111 p0_config 56 p0_led0 55 p0_led1 54 p0_ led2 5p0_rxn 4p0_rxp 122 p0_sdet 8p0_txn 7p0_txp 108 p1_config 53 p1_led0 51 p1_led1 50 p1_led2 12 p1_rxn 13 p1_rxp 127 p1_sdet 9p1_txn 10 p1_txp 49 p2_led0 48 p2_led1 46 p2_led2 16 p2_rxn 15 p2_rxp 19 p2_txn 18 p2_txp 45 p3_led0 44 p3_led1 43 p3_led2 24 p3_rxn 25 p3_rxp 21 p3_txn 22 p3_txp 41 p4_led0 40 p4_led1 39 p4_led2 28 p4_rxn 27 p4_rxp 31 p4_txn 30 p4_txp 73 p5_col 72 p5_crs pin number pin name
copyright ? 2008 marvell doc. no. mv-s100952-u0 rev. -- january 3, 2008, preliminary document classification: proprietary information page 33 signal description 128-pin lqfp pin assignment list?by signal name 82 p5_inclk 86 p5_ind0 85 p5_ind1 84 p5_ind2 83 p5_ind3 87 p5_indv 79 p5_outclk 77 p5_outd0/p5_mode0 76 p5_outd1/p5_mode1 75 p5_outd2/p5_mode2 74 p5_outd3/p5_mode3 81 p5_outdv 94 p4_col 91 p4_crs 102 p4_inclk 117 p4_ind0 116 p4_ind1 114 p4_ind2 112 p4_ind3 118 p4_indv 99 p4_outclk 98 p4_outd0/p4_mode0 97 p4_outd1/p4_mode1 96 p4_outd2/p4_mode2 95 p4_outd3/p4_mode3 100 p4_outdv 110 resetn 125 rsetn 126 rsetp 33 sw_mode0 34 sw_mode1 35 vdd 57 vdd 65 vdd 90 vdd pin number pin name 101 vdd 113 vdd 3 vddah 14 vddah 26 vddah 6 vddal 11 vddal 17 vddal 23 vddal 29 vddal 47 vddo 78 vddo 107 vddo 123 vddo 1vss 20 vss 32 vss 42 vss 52 vss 59 vss 67 vss 80 vss 92 vss 103 vss 115 vss 109 vss 128 vss 104 xtal_in 105 xtal_out pin number pin name
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 34 document classification: proprietary information january 3, 2008, preliminary section 2. application examples the marvell ? 88e6060 device supports many different applicatio ns with few additional active components in a small footprint. the 88e6060 contains six ports with five ph ys. this architecture allows all five phys to be active and available while at the same time supporting one independent mii interface. the higher integration, low power, and an extra port saves bom costs on designs requiring six ports or two mii interfaces. the flexibility of the 88e6060 device comes from its configuration options on port 0 and port 1, and its configuration options on its two mii ports (port 5?s mii inte rface is always on, while port 4 can be an mii or a phy). the device also supports an optional eeprom (93c46 type) to override any of the default settings. the only active components required to implement a complete 88e6060 10/100 ethernet switch are: ? 25 mhz crystal clock source or a 25 or 50 mhz oscillator ? low-cost pnp transistor used to regulate 2.5v down from 3.3v ? low-cost pnp transistor used to regulate 1.5v down from 3.3v the 88e6060 is a full system on a chip containing the phys, led drivers, voltage regulator logic, switch logic, and memory. in addition, it is pin compatible with the marvell ? 88e6063 qos switch. 2.1 examples with the 88e6060 the 88e6060 has multiple configuration options selectab le through four pins: p[1:0]_config, enable_mii5, and disable_mii4. table 1 shows port 0 and port 1 copper or fiber configuration settings, and port 5 and port 4 configuration settings. ports 2, 3, and 4?s def ault configuration is auto-negotiation enabled. table 18: 88e6060 port configuration p0_config p1_config enable_mii5 disable_mii4 port mode low x x x port 0 = copper auto-negotiation high x x x port 0 = fiber mode x low x x port 1 = copper auto-negotiation x high x x port 1 = fiber mode x x low x port 5 = disabled x x high x port 5 = rmii/mii/sni mode x x x low port 4 = rmii/mii/sni mode x x x high port 4 = copper auto-negotiation
application examples examples with the 88e6060 copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classifi cation: proprietary information page 35 each of the following modes are described in detail below: ? firewall router switch w/four 10/100b ase-t lan ports & one 10/100base-t wan port ? firewall router switch supporting a fiber wan port (using port 0 or port 1) ? firewall router switch with four 10/100base-t lan po rts & an mii port for a wan phy (using port 4) for the firewall router examples, port 5 is configured in phy mode, and ports 0 to 4 support 100base-tx and/or 10base-t operation. port 0 could be 10 0base-fx if desired as a fiber inte rface from the wan (see figure 3). the cpu?s smi is used to access all the phy and switch registers inside the 88e6060. in figure 2, port 0 is used as a wan interface in the lan. to make this work, the wan port needs to be isolated from the lan ports, and vice versa. lan frames cannot go directly to the wan, and wan frames cannot go directly to the lan. the cpu?s port needs to be the bridge by transmitting, receiving, and translating frames from both the wan and to and from the la n. the 88e6060 supports port based vlans in any programmable configu- ration that support this requirement and a marvell header mode supports dynamic vlans on the cpu?s port. bringing the 802.3 wan phy into the 88e6060?s port 0 not only reduces bom costs, but it also adds the feature of a 10/100 mbps interface with auto crossover c able support (supported on all copper ports). figure 2: 88e6060 firewall router example port 5 5x tx or 4x tx + 1 fx port 1port 2port 3port 4 88e6060 parallel led lines nc port 0 wan lan enable_mii5 tx_er tx_en txd[3:0] tx_clk rx_dv rx_clk rxd[3:0] col crs router chip with mac mdio mdc smi 25 mhz xtal ? ? 2.5v control_15 3.3v 3.3v power control_25 1.5v nc disable_mii4 link/activity duplex/collision speed 1234 0
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 36 document classification: proprietary information january 3, 2008, preliminary figure 3: firewall router switch supporting a fiber wan port port 5 port 1 port 2 port 3 port 4 88e6060 link/activity duplex/collision speed 1234 parallel led lines 2.5v control_15 3.3v 3.3v power nc port 0 enable_mii5 tx_er tx_en txd[3:0] tx_clk rx_dv rx_clk rxd[3:0] col crs router chip with mac mdio mdc smi 25 mhz xtal ? ? 100 mbit laser 1x fx + 4x tx 1x laser + 4x magnetics control_25 1.5v nc disable_mii4 0
application examples examples with the 88e6060 copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classifi cation: proprietary information page 37 figure 4: firewall router switch with lan ports and wan phy port 5 4x tx port 0 port 1 port 2 port 3 88e6060 link/activity duplex/collision speed 0123 parallel led lines disable_mii4 nc lan enable_mii5 port 4 - mii tx_er tx_en txd[3:0] tx_clk rx_dv rx_clk rxd[3:0] col crs marvell ? wireless router chip mdio mdc smi 25 mhz xtal ? ? 2.5v control_15 3.3v 3.3v power control_25 1.5v dsl, cable modem, or 802.3ah phy wan phy port 4
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 38 document classification: proprietary information january 3, 2008, preliminary 2.2 routing with the marvell ? header the detailed information regarding this feature requires an nda with marvell semiconductor. please contact your local marvell sales representative for more information.
functional description switch data flow copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 39 section 3. functional description this section describes the wire-speed, non-blocking 6-port 10/100-mbps fast ethernet sw itch core that is inte- grated into the 88e6060 device. the six ports include ph y, rmii/mii/sni, and internal mii ports, as described below. 3.1 switch data flow the 88e6060 device accepts ieee 802.3 fram es and either discards them or trans mits them out of one or more of the switchports. the decision on what to do with each frame is one of the many jobs handled inside the switch. figure 5 shows the data path inside the switch along with the major functional blocks that process the frame as it travels through the 88e6060 device. figure 5: 88e6060 device switch data flow the phy, or physical layer interface, is used to receive an d transmit frames over cat 5 twisted pair cable or fiber- optic cables (only port 0 and port 1 support a fiber option). the 88e6060 device contains five phys connected to port 0 to port 4. the phy block is covered in detail in section 4. port 5 does not contain phys. instead, a short distance industry standard digital interface is supported and gener- ically called the port?s media independent interface (mii see ?mii/sni/rmii? on page 41 ). many interface modes and timings are supported so that a large nu mber of external device types can be used. the two miis (port 4 can be mii or phy) on the 88e60 60 support four major modes of operation and each media independent interface (mii) can be configured independently. the 88e6060 contains six independent macs ( section 3.3 ) that perform all of the functions in the 802.3 protocol: ? frame formatting ? crc checking ? csma/cd enforcement ? collision handling phy rx 802.3 mac ingress policy port 0 (optional fiber interface) port 4 port 5 queue controller phy or rmii/ mii/sni pins rx 802.3 mac ingress policy rmii/mii/sni pins rx 802.3 mac ingress policy phy tx 802.3 mac port 0 (optional fiber interface) egress policy phy or rmii/ mii/sni pins tx 802.3 mac port 4 egress policy rmii/mii/sni pins tx 802.3 mac port 5 output queue output queue output queue ports 1-3 ports 1-3
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 40 document classification: proprietary information january 3, 2008, preliminary the switch portion of the 88e6060 receives good packets from the macs, processes th em, and forwards them to the appropriate macs for transmission. processing frames is the key activity, and it involves the ingress policy, the queue controller, the output queues, an d the egress policy blocks shown in figure 6 . the egress policy block is only applicable to the mii ports (port 4 and port 5). these blocks modify the normal or default packet flow through the switch and are discuss ed in section 3.5 to section 3.7. figure 6: switch operation ingress policy queue controller egress policy (port 4 and port 5 only) output queue
functional description mii/sni/rmii copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 41 3.2 mii/sni/rmii the media independent interface (mii) su pports four major modes of operation: ? mii phy mode (the 88e6060 device drives the interface clocks - see section 3.2.1 for details.) ? mii mac mode (the external device drives the interface clocks - see section 3.2.2 for details.) ? sni phy mode ? rmii phy mode the two mii ports can be configur ed differently from each other. 3.2.1 mii phy mode the mii phy mode (reverse mii) configures the selected mac inside the 88e6060 device to emulate a phy. this enables the 88e6060 device to be directly connected to an external mac (for example, one inside a router cpu). in this mode, the 88e6060 device drives the interface clocks (px_ inclk and px_outclk) ; therefore, the required frequency needs to be selected. both full-dupl ex and half-duplex modes are supported and need to be selected to match the mode of the lin k partner?s mac. the mii phy interfac e is compliant with the ieee 802.3u clause 22. figure 7: mii phy interface pins port 4 mii or port 5 mii acting like a phy cpu module with mii mac 88e6060 tx_clk txd[3:0] tx_en in_clk in_d[3:0] in_dv out_clk rx_clk rxd[3:0] rx_dv crs col out_d[3:0] out_dv crs col
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 42 document classification: proprietary information january 3, 2008, preliminary 3.2.2 mii mac mode the mii mac mode (forward mii) config ures the desired mac inside the 88e6060 device to act as a mac so that it can be directly connected to an external phy. in this mode, the 88e6060 device receives the interface clocks (px_inclk and px_outclk) and works at any frequency from dc to 25 mhz. the two clocks can be asynchro- nous with respect to each other. both fu ll-duplex and half-duplex modes are supported and need to be selected to match the mode of the link partner? s mac. the mii mac interface is co mpliant to the i eee 802.3u clause 22. figure 8: mii mac interface pins port 4 mii or port 5 mii acting like a mac phy module with mii interface 88e6060 rx_clk rxd[3:0] rx_dv in_clk in_d[3:0] in_dv out_clk tx_clk txd[3:0] tx_en crs col out_d[3:0] out_dv crs col
functional description mii/sni/rmii copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 43 3.2.3 sni phy mode the sni phy mode (7-wire interface) configures the selected mac inside the 88e6060 device to act as a 10 mbps phy, enabling it to be directly connected to an external 10 mbps-only mac (suc h as one inside a router cpu). in this mode, only one data bit is used on each of input and output (px_ind0 and px_outd0). the inter- face clocks, px_inclk and px_outcl k, are driven by the 88e6060 device. since sni was never standardized, the 88e6060 device supports various sni modes. the active edge of the clock (either rising or falling) and the active level on the collision signal (either active high or low) can be selected. in sni mode, the output data fr om the switch is indicated to be valid by the crs signal. crs is not synchronous with out_clk, however. so the receiving mac needs to de tect the beginning of the frame by looking for the pre- amble and then the start of frame de limiter (sfd). crs will be deasserted asynchronously at the end of the frame so the receiving mac needs to tolerate the presence of these extra 'dribble' bits figure 9: sni phy interface pins port 4 mii or port 5 mii acting like a phy cpu module with sni mac 88e6060 tx_clk txd tx_en in_clk in_d0 in_dv out_clk rx_clk rxd crs col out_d0 out_dv crs col
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 44 document classification: proprietary information january 3, 2008, preliminary 3.2.4 rmii phy mode rmii phy mode (reduced mii) conf igures the selected mac inside the 88e6060 device to act as a 100 mbps phy with a reduced media independent interface (rmii) enab ling it to be directly connected to an external 100 mbps rmii mac (for example, one inside an asic or fpga ). in this mode, only two data bits are used on each of input and output (px_ind[1:0] and px _outd[1:0]). the interface clock (p x_inclk) is driven by the 88e6060 device at the rmii consta nt frequency of 50 mhz. figure 10: rmii phy inte rface pins using inclk port 4 mii or port 5 mii acting like a phy cpu module with rmii mac 88e6060 txd[1:0] tx_en in_clk in_d[1:0] in_dv out_clk refclk rxd[1:0] crs_dv out_d[1:0] out_dv crs col
functional description mii/sni/rmii copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 45 3.2.5 rmii/mii/sni configuration mii/sni/rmii interfaces in the 88e6060 device are config ured at the rising edges of resetn. during reset the px_outd[3:0]/px_mode[3 :0] pins become tri-stated, and the values found on these pins become the interface?s mode as defined in table 19. 3.2.6 enabling the rmii/mii/sni interfaces port 5 (the 6th port) is enabled when enable_mii5 is tied high to vddo. port 4?s mii is enabled when disable_mii4 is tied to vss. 3.2.7 port status registers each switch port of the 88e6060 device has a status regi ster that reports informatio n about that port?s phy or rmii/mii/sni interface. see section 6.2.1 "switch core register map" for more information. table 19: rmii/mii/sni configuration options px_ mode[3:0] (at reset) phy/ mac mode duplex mii/sni mode description 0 0 0 0 phy half-duplex sni 10 mbps rising edge clock with collision active low 0 0 0 1 phy half-duplex sni 10 mbps rising edge clock with collision active high 0 0 1 0 phy full-duplex sni 10 mbps rising edge clock (collision is don?t care) 0 0 1 1 mac full-duplex reserved reserved for future use 0 1 0 0 phy half-duplex sni 10 mbps falling edge clock with collision active low 0 1 0 1 phy half-duplex sni 10 mbps falling edge clock with collision active high 0 1 1 0 phy full-duplex sni 10 mbps falling edge clock (collision is don?t care) 0 1 1 1 phy full-duplex reserved reserved for future use. 1 0 0 0 mac half-duplex mii 0 -100 mbps dc to 25 mhz mii input clock mode 1 0 0 1 phy half-duplex rmii 100 mbps 50 mhz reduced mii output clock mode 1 0 1 0 mac full-duplex mii 0 -100 mbps dc to 25 mhz mii input clock mode 1 0 1 1 phy full-duplex rmii 100 mbps 50 mhz reduced mii output clock mode 1 1 0 0 phy half-duplex mii 10 mbps 2.5 mhz mii output clock mode 1 1 0 1 phy half-duplex mii 100 mbps 25 mhz mii output clock mode 1 1 1 0 phy full-duplex mii 10 mbps 2.5 mhz mii output clock mode 1 1 1 1 phy full-duplex mii 100 mbps 25 mhz mii output clock mode
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 46 document classification: proprietary information january 3, 2008, preliminary 3.3 media access controllers (mac) the 88e6060 device contains six independent macs that perf orm all of the functions in the 802.3 pr otocol that include, among others, frame formatt ing, frame stripping, fcs checking , csma/cd enforcement, and collision handling. each mac receive block checks incoming packets and disca rds packets with crc errors, alignment errors, short packets (less than 64 bytes), or long packets (more than 1522 bytes) 1 . each mac constantly monitors its receive lines waiting for preamble bytes followed by the start of frame delimiter (sfd). the first six bytes after the sfd are used as the packet?s destination address (da), and the next six bytes are used as the packet?s source address (sa). these two addresses are fundamental to the oper ation of the switch (see se ction 3.4). the last four bytes of the packet contain the packet?s frame check s equence (fcs). the packet is discarded if the fcs does not meet the ieee 802. 3 crc-32 requirements. before a packet can be sent out, the transmit block must check whether the line is available for transmission. the transmit line is available all the time when the port is in full-duplex mode, but the line could be busy receiving a packet when the port is in half-duplex mode. in this ca se, the transmitter defers its transmission until the line becomes available, when it ens ures that a minimum interpacket gap of at least 96 bits occurs before transmitting a 56-bit preamble and an 8-bit start of frame delimiter (sfd) ahead of the bas ic frame. transmission of the frame begins immediately after the sfd. for the half-duplex mode, the 88e6060 device also monito rs the collision signal while it is transmitting. when a collision is detected (i.e., both transmitter and receiver of a half-duplex mac are active at the same time), the mac transmits a jam pattern and then delays the retransmi ssion for a random time period determined by the ieee 802.3 backoff algorithm. in full-duplex mode, the co llision signal and backoff algorithm are ignored. 3.3.1 backoff in half-duplex mode, the 88e6060 devic e?s macs implement the truncated binary exponential backoff algorithm defined in the ieee 802.3 standard. this algorithm starts with a randomly-selected small backoff time and follows by generating progressively longer random backoff times. the random times prevent two or more macs from always attempting re-transmission at the same time. the progressively longer backoff times give a wider random range at the expense of a longer delay, giving congest ed links a better chance of finding a winning transmitter. each mac in the 88e6060 device resets the progressi vely longer backoff time circuit after 16 consecutive retransmit trials. each mac then restarts the backof f algorithm with the shortest random backoff time and continues to retry and retransmit the frame. a packet that successively collides with a retransmit signal is re- transmitted until transmission is successful. this algori thm prevents packet loss in highly-congested environments. the macs in the switch can be configured to meet the ieee 802.3 specification and discard a frame after 16 consecutive collisions instead of restarting the backoff algorithm. to discard a frame after 16 consecutive collisions, set the discardexcessive bit to a one in the global c ontrol register?see table 53. 3.3.2 half-duplex flow control half-duplex flow control is used to throttle the end stat ion to avoid dropping packets during network congestion. it is enabled on all half-du plex ports when the ee_din/hd_flow_dis pin is low at the rising edge of resetn. the 88e6060 device uses a collision-based scheme to perform ha lf-duplex flow control. when the free buffer space is almost exhausted, the mac forces a collision in the input port when the 88e6060 device senses an incoming packet. only those ports that are involved in the conges tion are flow controlled. when the half-duplex flow control mode is not set and no packet buffer space is available, the incoming packet is discarded. 1. a maximum frame size of 1536 bytes is supported by setting the maxframesize bit in the global control register (section 6.2.3).
functional description media access controllers (mac) copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 47 3.3.3 full-duplex flow control the purpose of full-duplex flow control is the same as th at of flow control in the ha lf-duplex case?to avoid drop- ping packets during congestion. full-duplex flow control is enabled on all full-duplex ports when: ? fd_flow_dis pin is low at the rising edge of resetn, and ? auto-negotiation is enabled on the port (see section 4.2.13), and ? the link partner ?advertises? that it supports pause during auto-negotiation 1 full-duplex flow control is not automatically su pported on ports 0?1 when configured for 100base-fx operation, nor for port 5 and port 4 when it is in mii mode. this is because auto-n egotiation is not de fined for 100base-fx or mii. therefore, the link partners cannot advertise that they support pause. it can be forced, however?see sec- tion 3.3.4. when the full-duplex flow control mode is not set and no packet buf fer space is available, the incoming packet is discarded. in full-duplex mode, the 88e6060 macs support the st andard flow control defined in the ieee 802.3x specifica- tion. this flow control enables stopping and restoring the transmission from the remote node. the basic mecha- nism for performing full-duplex flow control is by means of a pause frame. the format of the pause frame is shown in table 20. full-duplex flow control functions as follows. when the free buffer space is almost exhausted, the 88e6060 device sends out a pause frame with the maximum pause time (a value of all ones?0xffff) to stop the remote node from sending more frames into the switch. only the node that is involved in the congestion is paused. when the event that invoked flow control disappears, the 88e606 0 device sends out a pause frame with the pause time equal to zero, indicating that the remote node can resu me transmission. the 88e6060 device also responds to the pause command in the mac receiving block. when the pause command is detected, the mac responds within one slot time (512 bit times) to stop transmission of the new data frame for the amount of time defined in the pause time field of the pause command packet. the source address of received pause frames is not learned (see address learning in section 3.4.3 ) since it may not represent the source address of th e transmitting port. this is generally the case if the link partner is an unmanaged switch. the 88e6060 device can be configur ed to transmit a unique source address on pause frames (the default is all zeros). global swit ch mac address registers 1, 2, and 3 (see ta b l e 5 0 , table 51 , and table 52 ) can be set to the source address to use. a single fixed source address can be used for all ports, or a unique source address per port can be selected by changing the value of the diffaddr bit in ta b l e 5 0 . the macs always discard all ieee 802.3x pause frames that they receive, even when fu ll-duplex flow control is disabled or even when the port is in half-duplex mode. 1. full-duplex flow control is not automatically supported on ports configured for 100base-fx operation since auto-negotiation is not defined for 100base-fx, and hence, the link partners cannot adve rtise that they support pause. it can be forced, however ? see section 3.3.4. table 20: pause frame format destination address (6 bytes) source address (6 bytes) type (2 bytes) op code (2 bytes) pause time (2 bytes) padding (42 bytes) fcs (4 bytes) 01-80-c2-00-00-01 see text 88-08 00-01 see text all zeros computed
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 48 document classification: proprietary information january 3, 2008, preliminary 3.3.4 forcing flow control when flow control is enabled using the 88e6060 device?s pins , it is enabled for all ports of the same type (i.e., on all half-duplex ports or on all full-duplex ports that have a flow-controllable link partner). it may be required to have flow control enabled on only one or two ports and have a ll the other ports disabled. in this case, flow control should be disabled using the appropriate device pins. refer to the fd_flow_dis and hd_flow_dis pin descriptions for disabling flow control then use the port?s forceflowcontrol bit in the port control register for enabling flow control on individual ports. (see section 6.2.2). 3.3.5 statistics counters sometimes it is necessary to debug network occurrences. for example, a technician may want to view the net- work remotely to solve a customer problem or a software programmer may want to trace transmitted frames. in these situations, two basic types of data are important: ? number of good frames entering and leaving each port of the switch ? quality of network segment performance frame size and the distribution of frames between multicast and unicast types are less important in this kind of debugging for an edge switch. the 88e6060 statistics counters support basic debugging needs. each port can captur e two kinds of statistics: ? a count of the number of good frames received with the numb er of frames transmitted, or ? a count of the number of bad frames receiv ed with the number of collisions encountered the first statistic answers the question ?where did all t he frames go??, while the second statistic answers the question ?does the network segment have any performa nce problems??. these counters are described in table 47 , and in table 48 . the counters can be cleared, and their mo de chosen by the ctrm ode bit in the switch global control regist er (section 6.2.3).
functional description address management copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 49 3.4 address management the primary function of the switch por tion of the 88e6060 device is to receive good packets from the macs, pro- cesses them, and forward them to the appropriate macs fo r transmission. this frame processing involves the ingress policy, queue controller, output queues, and egress policy blocks shown in figure 6 . these blocks mod- ify the normal or default packet flow through the swit ch and are discussed following section 3.5. the normal packet flow and processing is discussed first. the normal packet flow involves learning how to switch packets only to the correct macs. the switch learns which port an end station is connected to by remembering each pa cket?s source address along with the port number on which the packet arrived. when a packet is directed to a new, currently unlearned ma c address, the packet is transmitted out of all of the ports 1 except for the one on which it arrived 2 . once a mac address/port number mapping is learned, all future packets directed to that end station?s mac address (as defin ed in a frame?s destination address field) are directed to the learned port number only. this ensures that the packet is received by the correct end station (if it exists), and when the end station responds, its address is learned by the switch for the next series of packets. owing to the limitation of physical me mory, switches learn only the current ly ?active? mac addresses?a small subset of the 2 48 possible mac addresses. when an end station is moved from one port to another, a new mac address/port number association must be learned, and t he old one replaced. these issues are handled by the ?aging? and ?station move handling? functions. a mac address/port number association is allowed to be ?active? for only a limited time. this time limi t is typically set to five minutes. 3.4.1 address translation unit the 88e6060 device?s lookup engine or address translation unit (atu) gets the da and sa from each frame received from each port. it performs all address searchi ng, address learning, and address aging functions for all ports at wire speed rates. for example, a da and an sa lookup/learn function can be performed for all ports in less time than it takes to receive a 64-byte frame on all ports. the address database uses a hashing technique for quick storage and retrieval. hashing a 48-bit address into fewer bits results in some mac addresses having the same hash address. this situation is called a hash collision and is solved in the 88e6060 device by using a four entry bin per hash location that can store up to four different mac addresses at each hash location. the four-entry bin is twice as deep as that of many competing switching devices and allows for a reduced size of the address dat abase while still holding the same number of active ran- dom value mac addresses. the address database is stored in the embedded sram a nd has a default size of 1024 entries with a default aging time of about 300 seconds or 5 mi nutes. the size of the address database can be modified to a maximum of 256, 512, or 1024 entries. decreasing the size of the address database incr eases the number of buffers available for frames (see figure 11). the age time can be modi fied in 16 second increments from 0 seconds (aging dis- abled) to 4080 seconds (or 68 minutes). these op tions are set in the atu control register ( table 54 ). note changing the atu size will result in atu reset an d swreset. swreset will reset the mac, and the port state bits will be set back to their configuration reset value. the atu reset will reset the entire atu database. see atusize and swre set register descriptions in table 54 . 1. port-based vlans modify this operation (section 3.5.1). 2. the 88e6060 device can be configured to transmit frames out the same port that they came in on?see table 45 .
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 50 document classification: proprietary information january 3, 2008, preliminary figure 11: atu size tradeoffs 3.4.2 address searching or translation the address search engine searches the address database to get the output port number(s), called the destina- tion port vector (dpv), for each frame?s destination addr ess (da). when an address is found, it can switch the frame instead of flooding it. if the destination address is not found, then the packet is flooded. flooding refers to the action of sending frames out to a ll the ports of the switch that have link up with portstate not "disabled", except for the port the frame came in on. the switch arbi trates destination address lo okup requests from the ports and grants one lookup at a time. the mac address is has hed, and then data is read from the sram table and compared to the mac address for a match. four different addresses can be stored at each hash location. when a match is found, the address translation unit (atu) return s the dpv to the ingress policy block where it may get modified before the packet is queued to the output port(s). the dpv returned from the atu may get modified by the vlantable data. when no mac address match is found, the ingress policy block uses a unique default dpv for each ingress port, which typically floods the frame. t he default dpv for each port is the port?s vlantable data ? see table 45 . when the destination address in the frame is a multicast address or broadcast address, the address is searched in the same way as a unicast addr ess, and the frame is proce ssed identically. multicast addresses cannot be learned, so they app ear in the address database only when they are loaded into it by a cpu or the eeprom ? see section 3.4.5.3. this feature is used for multicast filtering and bridge protocol data unit (bpdu) handling. bpdu frames are special frames used for spanning tree or bridge loop detection. multiple separate address databases are supported in the 88e6060 device. th e database that is searched is con- trolled by the port?s default database number (dbnum - table 45 ). mac addresses that are not members of the port?s dbnum cannot be found. 0x0000 atu data frame buffers 62k bytes 0x1f00 256 entries 2k bytes 0 63 0x1fff 0x0000 atu data frame buffers 60k bytes 0x1e00 512 entries 4k bytes 0 63 0x1fff 0x0000 atu data frame buffers 56k bytes default 0x1c00 1024 entries 8k bytes 0 63 0x1fff
functional description address management copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 51 3.4.3 address learning the address learning engine learns source addresses of incoming frames. up to 1024 mac address/port number mappings can be stored in the address database. see se ction 3.4.1. when the source address from an input frame cannot be found in the address database, the atu enters the self-learning mode, places the new mac address/port number mapping into the database, and refreshes its age time. the age time on a mac address entry is refreshed by setting its entry_state field to 0xe. when the mac address is already in the database, the port number 1 and age associated with the entry is updated and/or refreshed. the port number is updated in case the end station has moved, and the port number needs to be corrected. the entry?s age is refreshed since the mac address is still ?active?. this refreshing prevents the mac address/ port number mapping from being prema- turely removed as being ?inactive?. when an address is added into the database it is hashed a nd then stored in the 1st empty bin found at the hashed location. when all four address bins are full, a ?least recently used? algorithm is used to scan each entry?s age time in the entry_state field. the entry_state field is de scribed in section 3.4.5.1. w hen all four address bins have the same age time, then the 1st unlocked bin is used (s ee section 3.4.5.1 for more information about locked or static addresses). if all four bins ar e locked then the address is not learned and an atufull interrupt is generated. see the switch global status register ? ta b l e 4 9 . multiple separate address databases are supported in the 88e6060 device. the port?s database number (dbnum - ta b l e 4 5 ) determines the mac address database into whic h the learned address is stored. the same mac address can be learned multiple times with different po rt mappings if different dbnum values are used. learning can be disabled for all ports by setting the lear ndis bit to a one in the atu control register (table 54). learning is disabled on any port that has a portstate of disabled or blocking/listening (see the port control regis- ter?table 44) or whose port association vector (pav bit ? table 46) is cleared to a zero. 3.4.4 address aging address aging makes room for new active addresses by ensuring that when a node is disconnected from the net- work segment or when it becomes inactive, its entry is removed from the address database. an address is removed from the database after a predetermined interval from the time it last appeared as an ingress frame?s source address (sa). this interval is programmable in the 88e6060 device. the default aging interval is about 5 minutes (282 to 304 seconds), but it can be set from 0 se conds (i.e., aging is disabled) to 4,080 seconds in 16 second increments. see agetime in atu control register?see table 54 . the 88e6060 device runs the address aging process contin uously unless disabled by clearing the agetime field in the atu control register to zero. aging is accomplis hed by a periodic sweeping of the address database. the speed of these sweeps is determined by the agetime fi eld. on each aging sweep of the database, the atu reads each valid entry and updates its age time by decrementing its entry_state field as long as the entry is not locked. the entry_state field is described in section 3.4.5.1. when the entry_state fi eld reaches zero, the entry is consid- ered invalid and purged from the database. the time taken to age out any entry in the mac address database is a function of the agetime value in the atu control register and the value in the entry_state fiel d. a new or just-refreshed unicast mac address has an entry_state value of 0xe. see section 3.4.2. a purged or invalid entry has an entry_state value of 0x0. the values from 0xd to 0x1 indicate the age time on a valid unic ast mac address with 0x1 being the oldest. this scheme pro- vides 14 possible age values in the entry_state field whic h increases precision in the age of entries in the mac address database. this precision is relayed to the ?least recently used? algorithm that is employed by the address 1. the port?s port association vector (pav - table 46) is used as the port?s port vector when an address is auto learned or upda ted.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 52 document classification: proprietary information january 3, 2008, preliminary learning process. an address is purged from the database within 1/14th of the programmed agetime value in the atu control register. 3.4.5 address translation unit operations the atu in the 88e6060 device supports user commands to access and modify the contents of the mac address database. all atu operations have the same user interface and protoc ol. five global registers are used and are shown in table 21 on page 52. the protocol for an atu operation is as follows: 1. ensuring the atu is available by checking the atubusy bit in the atu operation re gister. the atu can only perform one user command at a time. 2. loading the atu data and atu mac registers, if required by the operation. 3. starting the atu operation by defining the dbnum, atuop, and setting the atubusy bit to a one in the atu operation register. the dbnum, at uop and the atubusy bits setting can be done at the same time. 4. waiting for the atu operation to complete. completion can be verified by polling the atubusy bit in the atu operation register or by receiving an atudone interrupt. see switch global control register ta b l e 5 3 and switch global status register, ta b l e 4 9 . 5. reading the results, if appropriate. 3.4.5.1 format of the atu database each mac address entry in the atu database is 64 bits in size. the lower 48 bits contain the 48-bit mac address, and the upper 16 bits contain information about the entry as shown in figure 12. the database is accessed 16 bits at a time via the switch global registers shown in the fi gure. for more information a bout these registers, see atu mac switch address registers, table 57 through ta b l e 5 9 . table 21: atu operations registers register offset section before the operation starts after the operation completes atu operation 0x0b or decimal 11 table 55 used to define the required opera- tion (including which database to search, i.e., the dbnum field) and start it. used to indicate the atu?s busy status. atu data 0x0c or decimal 12 table 56 used further to define the required operation and used as the required atu data that is to be associated with the mac address below. returns the atu data that is associated with the resulting mac address below. atu mac (3 registers) 0x0d to 0x0f or decimal 13 to 15 table 57 used to define the required mac address upon which to operate returns the resulting mac address from the required operation.
functional description address management copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 53 figure 12: format of an atu entry the upper 16 bits of the atu entry are called the atu data which are separated into three fields?see atu data fields,table 22 and atu data register, table 56 . table 22: atu data fields field bits description entry_state 51:48 the entry_state field, together with the entry?s multicast bit (bit 40) is used to deter- mine the entry?s age or its type as follows: for unicast mac addresses (bit 40 = 0): 0x0 = invalid, empty, or purged entry. 0x1 to 0xe = valid entry where the entry_state = the entry?s age, and the dpv indicates the port or ports mapped to this mac address. 0xf = valid entry that is locked and does not age. the dpv indicates the port or ports mapped to this mac address. for multicast mac addresses (bit 40 = 1): 0x0 = invalid, empty, or purged entry. 0x7 = valid entry that is locked does not age. the dpv indicates the port or ports mapped to this mac address. used for multicast filtering. 0xe = valid entry that is locked and does not age. the dpv indicates the port or ports mapped to this mac address. frames with a mac address that return this entry_state are considered mgmt (managem ent) frames and are allowed to tun- nel through blocked ports (see section 3.5). used for bpdu handling. dpv 57:52 the destination port vector. these bits indicate which port or ports are associated with this mac address (i.e., where frames should be switch to) when they are set to a one. a dpv of all zeros indicates frames with this da should be discarded. bit 52 is assigned to physical port 0, 53 to port 1, 54 to port 2, 55 to port 3, and so on. reserved 63:58 reserved for future use. dpv 63 es 52 51 48 addr byte 0 addr byte 1 addr byte 2 addr byte 3 addr byte 4 addr byte 5 0 7 8 15 16 23 24 31 32 39 40 47 register 0x0f or decimal 15 register 0x0e or decimal 14 register 0x0d or decimal 13 register 0x0c or decimal 12 48-bit mac address res 57 58 bit 40 is the multicast bit atu data
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 54 document classification: proprietary information january 3, 2008, preliminary 3.4.5.2 reading the address databa se - the get next operation the contents of the address database can be dumped or searched. the get next operation returns the active contents of the address database in asc ending network byte order. a search operation can also be done using the get next operation. if multiple address databases are be ing used, the get next function returns all unique mac addresses from the selected database. the get next operation starts with the mac address c ontained in the atu mac registers and returns the next higher active mac address currently active in the addr ess database. use an atu mac address of all ones to get the first or lowest active mac address. the returned mac address and its data is accessible in the atu mac and the atu data registers. to get the next higher active mac address, the get next operation can be started again without setting the atu mac registers si nce they already contain the ?last? address. a returned atu mac address of all ones indicates that the end of the database has been reached 1 . a summary of how th e get next operation uses the atu?s registers is shown in table 23. to search for a particular mac addr ess, start the get next operation wit h a mac address numerically one less than the particular mac address using the dbnum of the selected database to search. when the searched mac address is found, it is returned in the atu mac registers along with its associated data in the atu data register. when the searched mac address is not found active, then the atu mac registers will not equal the required searched address. 3.4.5.3 loading and purging an entry in the address database any mac address (unicast or multicast) can be loaded into , or removed from, the address database by using the load operation. an address is loaded into the database if the entry_state in the atu data register ( table 56 ) is non-zero. a value of zero indicates the required atu operation is a purge. 1. when the returned mac address is all ones, it always indicates that the end of the table has been reached. if the returned entry_state bits are non-zero, it also indicates that the mac address of all ones is active in the database. table 23: atu get next operation register usage register offset section before the operation starts after the operation completes atu operation 0x0b or decimal 11 ta b l e 5 5 used to define the required opera- tion (including which database to search, i.e., the dbnum field) and start it. used to indicate the atu?s busy status. atu data 0x0c or decimal 12 ta b l e 5 6 ignored. returns the atu data that is associated with the resulting mac address below. if entry_state = 0x0 the returned data is not a valid entry. atu mac (3 registers) 0x0d to 0x0f or decimal 13 to 15 ta b l e 5 7 through ta b l e 5 9 used to define the starting mac address to search. use an address of all zeros to find the first or lowest mac address. use the last address to find the next address. there is no need to write to this register in this case. returns the next higher active mac address if found, or all ones are returned indi- cating the end of the table has been reached.
functional description address management copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 55 the load operation searches the address database indica ted by the database number, dbnum (in the atu oper- ation register), for the mac address contained in the atu mac registers. when the address is found, it is updated by the information found in the atu data register. note a load operation becomes a purge operation when the atu data?s entry_state equals zero. also, locked addresses can be modified without their needing to be purged first. if the address is not found an d if the atu data?s entry_state does not equ al zero, the address is loaded into the address database using the same protocol as is used by automatic address learning. see section 3.4.3. the 16 bits of the atu data register are written into bits 63:48 of the atu entry . see section 3.4.5.1. a summary of how the load operation uses the atu?s registers is shown in table 24. table 24: atu load/purge operation register usage 3.4.5.4 flushing entries all mac addresses, or just the unlo cked mac addresses, can be purged fr om the entire set of address databases or from just a particular address database usi ng single atu operations. these atu operations are: ? flush all entries ? flush all unlocked entries ? flush all entries in a particular dbnum database ? flush all unlocked entries in a particular dbnum database the atu data and atu mac address registers are not used for these operations and they are left unmodified. the dbnum field of the atu operation register is used fo r the flush operations that require a database number to be defined. register offset section before the operation starts after the operation completes atu operation 0x0b or decimal 11 ta b l e 5 5 used to define the required opera- tion (including which database to load or purge, i.e., the dbnum field) and start it. used to indicate the atu?s busy status. atu data 0x0c or decimal 12 ta b l e 5 6 used to define the associated data that is loaded with the mac address below. when entry_state = 0, the load becomes a purge. no change. atu mac (3 registers) 0x0d to 0x0f or decimal 13 to 15 ta b l e 5 7 through ta b l e 5 9 used to define the mac address to load or purge. no change.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 56 document classification: proprietary information january 3, 2008, preliminary 3.5 ingress policy the ingress policy block modifies the normal packet flow through the switch. frames are prevented from going out of certain ports by the use of port-based vlans, and frames are prevented from ent ering the switch by the use of switch management port states.
functional description ingress policy copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 57 3.5.1 port-based vlans the 88e6060 device supports a very flexible port-based vlan feature. each ingress port is associated with the vlantable field of the port-based vl an map register that restricts which egress ports its frames may use. ( table 45 on page 96 ). when bit 0 of a port?s vlantable field is set to a one, that port is allowed to send frames to port 0. when bit 1 of this field is set to a one, that po rt is allowed to send frames to port 1. if bit 2 equals 1, port 2 may receive frames, if bit 3 equals 1, port 3 may receive frames, and so on. at reset the vlantable value for each port is set to all ones, except for each port?s own bit, which is cleared to a zero. this prevents frames from going back out of the port at which they arrived. this default vlan configuration allows all of the ports to send frames to al l of the other ports as shown in figure 13. figure 13: switch operat ion with vlans disabled one reason for vlan support in the 88e6060 device is to is olate a port for firewall router applications. figure 14 shows a typical vlan configuration for a firewall router. po rt 0 is the wan port. the frames arriving at this port must not go out to any of the lan ports, but they must be able to go to the router cpu. all the lan ports are able to send frames directly to each other without the need fo r cpu intervention but they cann ot send frames directly to the wan port. to accomplish routing, the cpu is able to send frames to all of the ports. the use of the marvell header (section 3.5.4 ) enables a cpu to define dynamically to which port or ports a frame is allowed to go. this feature is useful for purposes of wan and lan isolat ion on multicast or flooded traffic generated by the cpu. 0 2 1 5 3 4
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 58 document classification: proprietary information january 3, 2008, preliminary figure 14: switch operation with a typical router vlan configuration this specific vlan configuration is accomplished by setting the port?s vlantable registers as follows: 3.5.1.1 tunneling frames through port-based vlans normally frames cannot pass through the port based vlan barr iers. however, some frames can be made to pass through the vlan barriers on the 88e6060 device. before a frame can tunnel through a port based vlan barrier, its destination address (da) must be locked into the address database (sec tion 3.4.5.1), and the vlantunnel bit on the frame?s ingress port must be set to a one ( table 44 on page 94 ). when both of these conditions are met, the frame is sent out of the port or ports indicated in the locked address?s dpv field for the da entry in the address database. the vlantable data is ignored in this case. this feature is enabled only on those ports that have their vlantunnel bit set to a one. table 25: vlantable settings for figure 14 port # port type vlantable setting 0wan0x20 1 lan 0x3c 2 lan 0x3a 3lan0x36 4 lan 0x2e 5cpu0x1f 0 2 1 5 3 4 wan lan lan lan lan cpu/ router
functional description ingress policy copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 59 3.5.2 switching frames back to their source port the 88e6060 device supports the ability to return frames to the port at which they arriv ed. while this is not a stan- dard way to handle ethernet frames, some applications may require this ability on some ports. this feature can be enabled on a port-by-port basis by setting the port?s own bi t in its vlantable register to a one. see section 3.5.1 and table 45 . 3.5.3 port states the 88e6060 device supports four port states per port as shown in table 26. the port states are used by the queue controller (section 3.6) in the 88e6060 device to ad just buffer allocation. they are used by the ingress pol- icy blocks to control which frame types are allowed to ent er and leave the switch, so that spanning tree or other bridge loop detection software can be supported. the portst ate bits in the port control register (table 44) deter- mine each port?s port state, and they can be modified at any time. table 26 below lists the port states and their function. two of the port states requi re the detection of mgmt frames. a mgmt frame in the 88e6060 device is any multicast frame whose da address is locked into the address database with an entry_state value of 0xe (secti on 3.4.5.1). mgmt frames can tunnel through blocked ports. mgmt frames always ignore vl ans (i.e., they always go to the port indicated by the da?s dpv). these mgmt frames are typically used for 802.1d spanning tree br idge protocol data units (bpdus), but any multicast address can be used supporting new and/or proprietary protocols. the default port state for all the ports in the switch can be either disabled or forwarding depending upon the value of the sw_mode pins (see pin description in part 1 of this datasheet). the ports come up in the forward- ing port state unless the sw_mode is fo r cpu-attached mode. this enables t he cpu to bring up the ports slowly after port-based vlans are configured (f or router wan to lan isolation) and so a spanning tree protocol can be run (if required). 3.5.4 switch?s ingress header (port 4 and port 5 only) the detailed information regarding this feature requires an nda with marvell ? semiconductor. please contact your local marvell sales repres entative for more information. table 26: port state options port state description disabled frames are not allowed to enter (ingress) or leave (egress) a disabled port. learning does not take place on disabled ports. blocking/ listening only mgmt frames are allowed to enter or leav e a blocked port. all other frame types are dis- carded. learning is disabled on blocked ports. learning only mgmt frames are allowed to enter or leave a learning port. all other frame types are dis- carded, but learning takes place on all good frames even if they are not mgmt frames. forwarding normal operation. all frames are allowed to enter and leave a forwarding port. learning takes place on all good frames.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 60 document classification: proprietary information january 3, 2008, preliminary 3.5.5 switch?s ingress traile r (port 4 and port 5 only) when a cpu needs to perform spanning tree or bridge loop de tection or when it needs to be able to send frames out any port or ports, then the cpu?s port needs to be c onfigured for ingress trailer mode. any mii port (port 4 and port 5) can be configured this way by setting the in gressmode bits in the port?s port control register ( table 44 on page 94 ), but only the cpu?s port should be configured this way. when the ingress trailer mode is enabled on a port, the last four bytes of the data portion of the frame are used to configure the switch?see figure 15. the ingress policy bl ock removes the trailer from the frame and overwrites it with a new crc, causing the frame to be four bytes smaller in size. this adjustment makes the frame ?normal? for the rest of the network since the trailer?s data is in tended for the switch only. fram e size checking is performed on the adjusted frame size. this means the cpu must always add four bytes of data to the end of every frame it sends into the switch when the trailer mode is enabled. the ingress trailer gives the cpu the ability to override normal switch operation on the fr ame that it just transmit- ted but this override may not always be necessary. when the cpu requires the switch to process the frame with the switch?s current ingress policy, the cpu sets the traile r data in the frame to all zeros by inserting a four-byte pad. this zero padding clears the override bit in the traile r causing the switch to ignore the trailer?s data and pro- cess the frame normally after removing the trailer from the frame. when the cpu needs to override all of the switch?s ingress policy including port based vlans, it sets the fields in the trailer as shown in figure 15 and defined in table 27. the ingress trailer is used to force a frame to egress out certain ports and it is the only way for the cpu to tunnel a frame through blocked ports.(by using the mgmt bit). wh en the cpu uses the trailer it must "know" where it wants the frame to go. figure 15: ingress trailer format preamble 7 octets destination address 1 octet sfd source address 6 octets 6 octets 2 octets mac client data length/type pad fcs marvell trailer 4 octets octets within frame transmitted top to bottom into the switch ieee 802.3 frame with marvell trailer in front of fcs res = 0 dpv [5:0] res = 0 b 7 b 0 1st octet 2nd octet 3rd octet 4th octet ignorefcs res = 0 mgmt res = 0 4 octets marvell ingress trailer format reserved = 0 override reserved = 0 learndisable
functional description ingress policy copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 61 table 27: ingress trailer fields port state description override when this bit is set to a one, the trailer data is used to override the switch?s operation. when this bit is cleared to a zero, the trailer?s dpv and mgmt bits are ignored. learndisable when this bit is set to a one (with or withou t the override bit being set) the source address con- tained in this frame will not be learned in the address database. ignorefcs when this bit is set to a one (with or withou t the override bit being set) the frame?s fcs is ignored (i.e., the fcs is not checked to see if the frame should be discarded due to a crc error). the fcs is still a required part of the frame, however, its value is not verified if this bit is set in the trailer. a new, correct crc is written to the last four bytes of the frame and the frame is pro- cessed by the switch with the good crc. dpv[5:0] the destination port vector. these bits indicate which port or ports the frame is to be sent to if the override bit is set to a one. a dpv of all zeros discards the frame. bit 0 is assigned to physi- cal port 0 and must be set to a one for this frame to egress from that port, bit 1 for port 1, bit 2 for port 2, bit 3 for port 3, and so on. mgmt the frame?s management bit. when this bit is set to a one (along with the override bit) indicates the frame is a mgmt frame and is allowed to ingress and egress through blocked ports (section 3.5.3). must be set on bpdu frames. res = 0 these fields are reserved for future use and must be set to zeros.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 62 document classification: proprietary information january 3, 2008, preliminary 3.6 queue controller the 88e6060 queue controller uses an advanced non-blocki ng, output port queue architecture with resource reservation. as a result, the 88e6060 supports definable frame latencies without head- of-line blocking problems or non-blocked flow disturbances in any congested environment. 3.6.1 no head-of-line blocking an output port that is slow or congested never affects t he transmission of frames to ports that are not congested. the 88e6060 device is designed to ensure that all flows that are not congested traver se the switch without degra- dation regardless of the congest ion elsewhere in the switch. 3.6.2 the queues the queues in the 88e6060 device are shown in figure 16. figure 16: switch queues 3.6.3 queue manager at reset 1 , the queue manager initializes the fr ee queue by setting all of the buffer pointers to point into it and ensuring that all of t he other queues are empty. the queue manager t hen takes the first available free buffer pointers from the free queue and assigns them to any ingress port that is not disabled 2 and whose link 3 is up. when these conditions are met, the switch is ready to a ccept and switch packets. whenever any port?s link goes down or the port is set to the disabled port state, the po rt?s ingress buffers and output queue buffers are immedi- ately returned to the free queue. this feature prevents ?stale buffers? or ?lost buffers? conditions and maximizes the size of the free queue so that of momentary congestion can be handled. when an enabled port?s link comes back up, the port gets its ingress buffers ba ck and it can start receiving frames again. 1. the queue manager is reset either by toggling the hardware reset n pin, a software reset by the swreset bit, or by an atusize change (both in the atu control register? table 54 ). 2. when a port is in the disabled port state (section 3.5.3), its ingress buffers are left in the free queue for other ports to use. 3. phy based ports need to get a link up signal from the phy. mii based ports have a link up state if they are enabled. ingress buffers port 0 port 1 port n queue manager ingress buffers ingress buffers port 0 port 1 port n free queue multicast handler output queue output queue output queue
functional description queue controller copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 63 when a mac receives a packet, it places it into the embed ded memory at the address indicated by the input point- ers that the mac received from the queue manager. when packet reception is complete, the mac transfers the pointers to the queue manager and requests new buffers fr om the free queue. when the free queue is empty, the mac is not allocated any pointers until they become available. if the ma c starts to receive a packet when it has no pointers allocated, the packet is dropped. if flow control is enabled, it prevents this condition from occur- ring. the queue manager uses the data retu rned from the lookup engine (section 3.4.1) and the ingress policy (sec- tion 3.5) to determine which output queue or queues t he packet?s pointer(s) should go to. at this point, the queue manager modifies the desired mapping of the fr ame, depending upon the mode of the switch and its level of congestion. two modes are supported, with and without flow control. both modes are handled at the same time and can be different per port. one port can have flow co ntrol enabled, while another has it disabled. when flow control is enabled on an ingress port, the fr ame is switched to the requ ired output queue without mod- ification. this operation is done so that frames are not dr opped. the queue manager monitors which output queues are congested and enables or disables flow contro l on the ingress ports that are causing the congestion. this approach allows flows that are not congest ed to progress through the switch without degradation. when flow control is disabled on an ingress port, the frame can be discarded instead of being switched to the required output queue. if a frame is destined for more than one output queue, it can be switched to some queues and not to others. the decisions are quite complex bec ause the queue manager takes many pieces of information into account before the decision is made. the queue manager monitors the current level of congest ion in the output queues to which the frame is being switched, and the current number of free buffers in the free queue. as a result, fl ows that are not congested traverse the switch unimpeded. 3.6.4 output queues the output queues receive and transmit packets in the order received. this is very important for some forms of ethernet traffic. the output queues are emptied as fast as possible, but they can empty at different rates, possibly owing to a port's being configured for a slower or faster speed, or because of network congestion (collisions or flow control). after a packet has been completely transmitted to the ma c, the output queue passes the transmitted packet?s pointers to the multicast handler for processing, afte r which the mac begins transmitting the next packet. 3.6.5 multicast handler the multicast handler receives the pointers from all of the packets that are transmitted. it looks up each pointer to determine whether each packet were directed to more than one output queue. if not, the po inter is returned to the free queue where it can be used again. when the frame is switched to mult iple output queues, the multicast han- dler ensures that the frame has exited all of the ports to which it was switc hed before returning the pointers to the free queue.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 64 document classification: proprietary information january 3, 2008, preliminary 3.7 egress policy (port 4 and port 5 only) the egress policy block is used to modify frames, if directed to, as they exit the switch . specific switch information can be added to the frame for the switch?s cpu. 3.7.1 switch?s egress header the detailed information regarding this feature requires an nda with marvell semiconductor. please contact your local marvell sales representative for more information.
functional description egress policy (port 4 and port 5 only) copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 65 3.7.2 switch?s egress trailer if a cpu needs to perform spanning tree or bridge loop de tection, the cpu must have information about the orig- inating physical source ports of its received frames, since those frames' sa information cannot be relied upon. to get this physical source port data, the cpu?s port needs to be configured for egress trailer mode (or egress header mode?section 3.7.1 ). any mii port (port 4 and port 5) can be configured this way by setting the trailer- mode bit in the port?s port control register ( table 44 on page 94 ), but only the cpu?s port should be configured this way. when the egress trailer mode is enabled on a port, four extra bytes are added to the end of the fr ame before the frame?s crc or fcs, and a new crc is appended to the end of the frame. when the frame is received by the cpu, the trailer occupies the last four bytes of the frame. the cpu?s softwa re can examine portions of the frame to determine if it needs the frame?s source port informati on. generally it is needed only on mgmt/spanning tree frames. if the information is not needed, the cpu reduces th e frame size variable by four and passes the frame to the appropriate routines for processing. removing the tr ailer from a frame essentially a subtraction operation without the need to move any data. consequently cpu overhead is kept to a minimum. the format of the egress trailer is shown in figure 17 and its fields are defined in table 28. figure 17: egress trailer format table 28: egress trailer fields field description valid when this bit is set to a one, it indicates the trailer data is present. this bit is always a one. spid[2:0] the source port id. these bits indicate which physical port the frame entered the switch from. an spid of all zeros indicates port 0. an spid of 0x1 indicates port 1, 0x2 indicates port 2, 0x3 indicates port 3, and so on. mgmt the frame management bit. when this bit is set to a one, it indicates the frame is a mgmt frame and is allowed to ingress and egress through blocked ports (section 3.5.3). should be set on bpdu frames. preamble 7 octets destination address 1 octet sfd source address 6 octets 6 octets 2 octets mac client data length/type pad fcs marvell trailer 4 octets octets within frame transmitted top to bottom by the switch ieee 802.3 frame with marvell trailer in front of fcs res = 0 res = 0 spid [2:0] res = 0 b 7 b 0 1st octet 2nd octet 3rd octet 4th octet valid - always a 1 mgmt res = 0 4 octets marvell egress trailer format res = 0
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 66 document classification: proprietary information january 3, 2008, preliminary 3.8 spanning tree support ieee 802.1d spanning tree is supported in the 88e6060 de vice with the help of an ex ternal cpu that runs the spanning tree algorithm. the 88e6060 device supports spanning tree by: ? detection of bridge protocol data unit (bpdu) frames. these fr ames are called mgmt (management) frames in the 88e6060 device. they are detec ted by loading the bpdu?s multicast address (01:80:c2:00:00:00) into the address database with a mgmt entry_state indicator (see section 3.4.5.1). ? tunnelling of bpdu frames through blocked ports. block ed ports are controlled by the port?s portstate bits (section 3.5.3). when a port is in the blocked state, a ll frames are discarded except for multicast frames with a da that is contained in the address dat abase with a mgmt indicator (see above). ? redirection of bpdu frames. bpdu frames need to go to the cpu only, even though they are multicast frames. this is handled in the detection of bpdu fr ames above by mapping the bpdu?s multicast address to the cpu port. (the value of the dpv bits when the address is loaded). ? source port information. the cpu needs information of the physical source port of origin of the bpdu frame. the source port is supplied in the frame?s egress trailer that is sent to the cpu (section 3.7.2). ? cpu transmission of bpdu frames. the cpu needs to be able to transmit bpdu frames out of any physical port of the switch. this is supported only in the ingr ess trailer data that is supplied by the cpu (section 3.5.5). the ingress header cannot be used for this since it cannot force mgmt frames to go out a particular port. the 88e6060 device can support 802.1d spanning tree, or it can be used to perform simpler bridge loop detec- tion on new link up. these different opt ions are accommodated by running appr opriate software on the attached cpu. any vendor?s proprietary protocol units can be handled with the same mechanism. 3.9 embedded memory the 88e6060 device contains an embedded 512kb (8kx64) synchronous sram (ssram). the ssram is run- ning at 50 mhz, and the data bus is 64 bits wide. the memory interface provides up to 3.2 gbps bandwidth for packet reception/transmission and address mapping data accesses. this memory bandwidth is enough for all the ports running at full wire speed in full-duplex mode with minimum size frames of 64 bytes. 3.10 interrupt controller the 88e6060 device contains a switch interrupt controller that is used to merge various interrupts on to the cpu?s interrupt signal. each switch interrupt can be individua lly masked by an enable bit contained in the switch global control register ( table 53 ). when an unmasked interrupt occurs and the interrupt pin goes active low, the cpu needs to read the switch global status register ( ta b l e 4 9 ) to determine the source of the interrupt. when the inter- rupt comes from the switch core (from atufull, atudone or eeint), the switch?s interrupt pin goes inactive after the switch global status register is r ead. the interrupt status bits are clear ed on read. if the interrupt comes from the phy (phyint), then the switch?s interrupt pin goes inac tive only after the phy?s interrupt is cleared by reading the appropriate registers in the phy. see table 60 for details.
functional description port monitoring support copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 67 3.11 port monitoring support port monitoring is supported by the 88e6060 device with egress only monitoring or egress and ingress monitor- ing. egress monitoring duplicates egress fr ames from a particular port to a se lected monitor port. ingress monitor- ing duplicates any good ingress frames fo r a particular port to a selected m onitor port (such frames are processed normally through the switch). port monitoring is enabled by modifying the port association vector ( ta b l e 4 6 ) for the particular port that is to be monitored. 3.12 port trunking support port trunking is supported by the 88e6060 device with any combinations of ports. the ports that are to be associ- ated with the trunk need to have all the port member?s bits set in each of their port association vectors ( table 46 ). port based vlans are then used to prevent loops and to load balance the trunk.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 68 document classification: proprieta ry information january 3, 2008, preliminary section 4. physical interface (phy) functional description the 88e6060 device contains five ieee 802.3 100ba se-tx and 10base-t compliant media-dependent inter- faces for support of ethernet over unshielded twisted pair (utp) copper cable. dsp-based advanced mixed signal processing technology supports attachment of up 150 mete rs of cat 5 cable to each of these interfaces. an optional, per port, automatic mdi/mdix crossover detect ion function gives true ?plug and play? capability without the need for confusing crossover cables or crossover ports. the port 0 and port 1 interface can be configured to support ieee 802.3 100base-fx by utilizing a pseudo-ecl (pecl) interface for fiber-optics.
physical interface (phy) functional description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 69 figure 18: 88e6060 device transmit block diagram mac pre-driver/ pecl driver 4b/5b encoder multimode dac physical coding sublayer pcs physical media attachment pma function physical media dependent pmd sublayer parallel to serial conversion internal interfaces twisted pair (on all ports) or fiber cable (only port 0 and port 1) driver p[n]_txp p[n]_txn media dependent interface mdi nrz to nrzi encoder scrambler parallel to serial conversion digital fir nrz to manchester encoder 100base-t 10base-t p[n]_tx_en p[1:0]config p[n]_txd
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 70 document classification: proprieta ry information january 3, 2008, preliminary figure 19: 88e6060 device receive block diagram adc digital adaptive equalizer nrzi to nrz decoder descrambler 4b/5b decoder serial to parallel conversion code group alignment agc sample/hold mlt-3 to binary decoder clock recovery 125mhz synchronizing fifo 10mbps receiver manchester to nrz decoder serial to parallel conversion sfd alignment baseline wander compensation physical coding sublayer pcs physical media attachment pma function physical media dependent pmd sublayer internal interfaces twisted pair (on all ports) or fiber cable (only port 0 and port 1) p[n]_crs_d v p[n]_rxd media dependent interface mdi p[x]_rxp p[x]_rxn pecl receiver p[1:0]_sdet 100base-t 10base-t p[1:0]config
physical interface (phy) functional description transmit pcs and pma copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 71 4.1 transmit pcs and pma 4.1.1 100base-tx transmitter the 100base-tx transmitter consists of several functional blocks t hat convert synchronous 4- bit nibble data to a scrambled mlt-3 125 mbps serial data stream. 4.1.2 4b/5b encoding for 100base-tx mode, the 4-bit nibble is converted to a 5-bit symbol with /j/k/ start- of-stream delim iters and /t/ r/ end-of-stream delimiters inse rted as needed. the 5-bit symbol is then serialized and scrambled. 4.1.3 scrambler in 100base-tx mode, the transmit data st ream is scrambled in order to redu ce radiated emissions on the twisted pair cable. the data is scrambled by exclusive oring t he nrz signal with the output of an 11-bit wide linear feed- back shift register (lfsr), which produces a 2047-bi t repeating pseudo-random sequence. the scrambler reduces peak emissions by randomly spreading the signa l energy over the transmit frequency range and eliminat- ing peaks at certain frequencies. note the enabling and disabling of the scrambler and the fa r end fault generator are controlled in the same way as for the descrambler detection and far end fault detection on the receive side. 4.1.4 nrz to nrzi conversion the data stream is converted from nrz to nrzi. 4.1.5 pre-driver and transmit clock the 88e6060 device uses an all-digital clock generator ci rcuit to create the various receive and transmit clocks necessary for 100base-tx, 100base-fx, and 10base-t modes of operation. for 100base-tx mode, the transmit data is converted to mlt-3-coded symbols. the digital time base generator (tbg) produces the locked 125 mhz transmit clock. for 100base-fx mode, nrzi data is present ed directly to the multimode dac. for 10base-t mode, the transmit data is converted to manchester encoding. the digital time base generator (tbg) produces the 10 mhz transmit reference clock as well as the ov er-sampling clock for 10base-t waveshap- ing. 4.1.6 multimode transmit dac the multimode transmit digital to analog converter (dac ) transmits mlt3-coded symbols in 100base-tx mode, nrzi symbols in 100base-fx mode, a nd manchester-coded symbols in 10ba se-t mode. the transmit dac uti- lizes a direct-drive current driver which is well ba lanced to produce very low common mode transmit noise. in 100base-tx mode, the multimode transmit dac performs slew control to minimize high frequency emi.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 72 document classification: proprieta ry information january 3, 2008, preliminary in 100base-fx mode, the pseudo ecl level is gen erated through external resistive terminations. in 10base-t mode, the multimode transmit dac gener ates the needed pre-equalization waveform. this pre- equalization is achieved by using a digital fir filter. 4.2 receive pcs and pma 4.2.1 10-base-t/100base-tx receiver the differential rxp and rxn pins are shared by the 100base-tx, 100base-fx (support ed on port 0 and port 1) and 10base-t receivers. the 100base-tx receiver consists of several functional blocks t hat convert the scrambled mlt-3 125 mbps serial data stream to the synchronous 4-bit nibble data presented to the mac interfaces. 4.2.2 agc and baseline wander in 100base-tx mode, after input to the agc block, the signal is compensated for baseline wander by means of a digitally controlled digital to analog converter (dac). it automatically removes the dc offset from the received sig- nal before it reaches the input to the sample and hold stage of the adc. 4.2.3 adc and digital adaptive equalizer in 100base-t mode, an analog to digital converter (adc ) samples and quantizes the input analog signal and sends the result into the digital adaptive equalizer. this equalizer removes inter-symbol interference at the receiver. the digital adaptive equalizer takes unequalized signals from the adc output and uses a combination of feed-forward equalizer (ffe) and decision feedback equaliz er (dfe) for the best optimized signal-to-noise (snr) ratio. 4.2.4 digital phased locked loop (dpll) in 100base-tx mode, the receive clock is locked to the incoming data stream and extracts a 125 mhz reference clock. the input data stream is quantiz ed by the recovered clock and sent thro ugh to the digital adaptive equalizer from each port. digital interpolator clock recovery circuits are optim ized for mlt-3, nrzi, and manchester modes. a digital approach makes the 88e6060 receiver path robust in the pr esence of variations in process, temperature, on-chip noise, and supply voltage. 4.2.5 nrzi to nrz conversion in 100base-tx mode, the re covered 100base-tx nrzi signa l from the receiver is converted to nrz data, descrambled, aligned, parallelized, and 5b/4b decoded. 4.2.6 descrambler the descrambler is initially enabled upon hardware re set if 100base-tx is sele cted. the scrambler can be enabled or disabled via software by setting the descrambler bit ( ta b l e 7 1 ).
physical interface (phy) functional description receive pcs and pma copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 73 the descrambler ?locks? to the descrambler state after detecting a sufficient number of consecutive idle code- groups. the receiver does not attempt to decode the dat a stream unless the descrambler is locked. once locked, the descrambler continuously monitors the data stream to make sure that it ha s not lost synchronization. the receiver descrambles the incoming data stream by exclusive oring it with the out put of an 11-bit wide linear feedback shift register (lfsr), which pr oduces a 2047-bit non-repeating sequence. the descrambler is always forced into the ?unlocked? state when a link failure condition is detected or when insuf- ficient idle symbols are detected. 4.2.7 serial-to-parallel conver sion and 5b/4b code-group alignment the serial-to-parallel /symbol alignment block performs se rial to parallel conversion and aligns 5b code-groups to a nibble boundary. 4.2.8 5b/4b decoder the 5b/4b decoder translates 5b code-g roups into 4b nibbles to be present ed to the mac interfaces. the 5b/4b code mapping is shown in table 29 . 4.2.8.1 fifo the 100base-x or 10base-t packet is placed into the fifo in order to correct for an y clock mismatch between the recovered clock and the reference clock refclk. 4.2.8.2 100base-fx receiver in 100base-fx mode, a pseudo-ecl (pecl) receiver is used to decode the in coming nrzi signal passed to the nrzi-nrz decoder. the nrzi signal from the receiver is converted to nrz data, aligned, parallelized, and 5b/4b decoded as in the 100base-tx mode. 4.2.8.3 far end faul t indication (fefi) when 100base-fx is selected and bit 0 of config_a is low at hardware reset, the far end fault detect (fefd) circuit is enabled. the fefd enable state can be overridden by programming the fefi bit ( table 71 ). note the fefi function is always dis abled if 100base-tx is selected. 4.2.8.4 10base-t receiver in 10base-t mode, the recovered 10base-t signal is dec oded from manchester to nrz and then aligned. the alignment is necessary to ensure that the start of fr ame delimiter (sfd) is aligned to the nibble boundary. in 10base-t mode, a receiver is used to decode the differential vo ltage offset of the ma nchester data. carrier sense is decoded by measuring t he magnitude of the voltage offset. in this mode, the recovered 10base-t si gnal is decoded from manchester to nrz data. the data stream is con- verted from serial to parallel format and aligned. the ali gnment is necessary to ensure that the start of frame delimiter (sfd) is aligned to a byte or nibble boundary. for cable lengths greater than 100 meters, the incoming signal has more attenuation. hence, the receive voltage threshold should be lowered via the extendeddistance bit in the phy specific control register ( ta b l e 7 1 ).
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 74 document classification: proprieta ry information january 3, 2008, preliminary table 29: 5b/4b code mapping pcs code-group [4:0] 4 3 2 1 0 name txd/rxd <3:0> 3 2 1 0 interpretation 1 1 1 1 0 0 0 0 0 0 data 0 0 1 0 0 1 1 0 0 0 1 data 1 1 0 1 0 0 2 0 0 1 0 data 2 1 0 1 0 1 3 0 0 1 1 data 3 0 1 0 1 0 4 0 1 0 0 data 4 0 1 1 1 0 6 0 1 1 0 data 6 0 1 1 1 1 7 0 1 1 1 data 7 1 0 0 1 0 8 1 0 0 0 data 8 1 0 0 1 1 9 1 0 0 1 data 9 1 0 1 1 0 a 1 0 1 0 data a 1 0 1 1 1 b 1 0 1 1 data b 1 1 0 1 0 c 1 1 0 0 data c 1 1 0 1 1 d 1 1 0 1 data d 1 1 1 0 0 e 1 1 1 0 data e 1 1 1 0 1 f 1 1 1 1 data f 1 1 1 1 1 i undefined idle; used as inter-stream fill code 1 1 0 0 0 j 0 1 0 1 start-of-stream delimiter, part 1 of 2; always used in pairs with k 1 0 0 0 1 k 0 1 0 1 start-of-stream delimiter, part 2 of 2; always used in pairs with j 0 1 1 0 1 t undefined end-of-stream delimiter, part 1 of 2; always used in pairs with r 0 0 1 1 1 r undefined end-of-stream delimiter, part 2 of 2; always used in pairs with t 0 0 1 0 0 h undefined transmit error; used to force signaling errors 0 0 0 0 0 v undefined invalid code 0 0 0 0 1 v undefined invalid code 0 0 0 1 0 v undefined invalid code 0 0 0 1 1 v undefined invalid code 0 0 1 0 1 v undefined invalid code 0 0 1 1 0 v undefined invalid code 0 1 0 0 0 v undefined invalid code 0 1 1 0 0 v undefined invalid code 1 0 0 0 0 v undefined invalid code 1 1 0 0 1 v undefined invalid code
physical interface (phy) functional description receive pcs and pma copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 75 4.2.9 setting cable characteristics since cable characteristics differ between unshielded twisted pair and shielded twisted pair cable, optimal receiver performance can be obtained in 100base -tx and 10base-t modes by setting the tpselect bit in the phy spe- cific control register ( ta b l e 7 1 ) for cable type. 4.2.10 scrambler/descrambler the scrambler block is initially enabl ed upon hardware reset if 100base-tx is se lected. if 100base-fx or 10base-t is selected, the scram bler is disabled by defau lt. the scrambler is contro lled by programming the diss- crambler bit in the phy sp ecific control register ( ta b l e 7 1 ). the scrambler setting is also controlled by hardwar e configuration at the end of hardware reset. table 30 shows the effect of various configuration settings on the scrambler. 4.2.11 digital clock recovery/generator the 88e6060 device uses an all-digital clock recovery and generator circuit to create all of the needed receive and transmit clocks. the digital time base generator (tbg ) takes the 25 mhz or 50 mhz reference input clock (xtal_in) and produces the locked 25 mhz transmit clock for the mac in 100base-tx mode. it produces a 2.5 mhz transmit clock for the mac in 10base-t mode as well as producing the over-sample clock for 10base-t waveshaping. 4.2.12 link monitor the link monitor is responsible for determining wh ether link is established with a link partner. in 10base-t mode, link monitor function is performed by detecting the presence of the valid link pulses on the rxp/n pins. in 100base-tx mode, the link is established by scrambled idles. in 100base-fx mode, the external fiber-optic rece iver performs the signal dete ction function and communicates this information with the 88e6060 device through sdet pin for port 0 and port 1. if force link good is asserted (forcelink bi t is set high - phy specific control register, table 71 ), the link is forced to be good, and the link monitor is bypassed. puls e checking is disabled if auto-negotiation is disabled, and disnlpcheck (phy specific control register, table 71 ) is set high. if auto-negotiation is disabled and disn- lpgen (phy specific control register, ta b l e 7 1 ) is set high, then the link pulse transmission is disabled. table 30: scrambler settings p[1:0]_config (if fx is selected) disscrambler bit () scrambler/ descrambler high hw reset to 1 disabled low hw reset to 0 enabled x user set to 1 disabled x user set to 0 enabled
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 76 document classification: proprieta ry information january 3, 2008, preliminary 4.2.13 auto-negotiation auto-negotiation is initiated upon any of the following conditions: ? power up reset ? hardware reset ? software reset (swreset bit?phy control register, ta b l e 6 1 ) ? restart auto-negotiation (restartaneg bit?phy control register, table 61 ) ? transition from power down to power up (pwrdwn bit?phy control register, ta b l e 6 1 ) ? change from the linkfail st ate to the link-up state if auto-negotiation is enabled, the 88e6060 device negotiat es with its link partner to determine the speed and duplex mode at which to operate. if the link partner is unable to auto-negotiate, the 88e6060 device goes into the parallel detect mode to determine the speed of the link partner. under para llel detect mode, the duplex mode is fixed at half-duplex. after hardware reset, auto-negotiation can be enabled an d disabled via the anegen bit (phy control register - table 61 ). when auto-negotiation is disabled, the speed and duplex can be changed via the speedlsb and duplex bits (phy control register - table 61 ), respectively. the abilities that are advertised can be changed via the auto-negotiation advertisement register ( table 65 ). 4.2.14 register update changes to the anegen, speedlsb, and duplex bits ( ta b l e 6 5 ) do not take effect unless one of the following takes place: ? software reset (swreset bit - table 61 ) ? restart auto-negotiation (restartaneg bit - ta b l e 6 1 ) ? transition from power down to power up (pwrdwn bit - ta b l e 6 1 ) ? loss of the link the auto-negotiation advertisement register ( table 65 ) is internally latched once every time auto-negotiation enters the ability detect state in the ar bitration state machine. hence, a writ e into the auto-negotiation advertise- ment register has no effect once the 88e6060 device begins to transmit fast link pulses (flps). this guaran- tees that a sequence of flps transmit ted is consistent with one another. the next page transmit register ( ta b l e 6 9 ) is internally latched once every time auto-negotiation enters the next page exchange state in the arbitration state machine. 4.2.15 next page support the 88e6060 device supports the use of next page during auto-negotiation. by default, the received base page and next page are stored in the link partner ability register - base page ( table 66 ). the 88e6060 device has an option to write the received next page into the link partner next page register - ta b l e 7 0 - (similar to the descrip- tion provided in the ieee 802.3ab standard) by programming the reg8nxtpg bit (phy specific control register - table 71 ). 4.2.16 status registers once the 88e6060 device completes auto-negotiation it updates the various status in the phy status ( ta b l e 7 2 ), link partner ability (next page) ( table 67 ), and auto-negotiation expansion ( table 68 ) registers. speed, duplex, page received, and auto-negotiation completed status are also available in the phy specific status ( ta b l e 7 2 ) and phy interrupt status registers ( ta b l e 7 4 ).
physical interface (phy) functional description power management copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 77 4.3 power management the 88e6060 supports advanced power management modes that conserve power. 4.3.1 low power modes two low power modes are supported in the 88e6060. ? ieee 802.3 clause 22.2.4.1.5 complia nt power down ? energy detect+ tm ieee 802.3 clause 22 .2.4.1.5 power-dow n compliance allows for the phy to be placed in a low-power consump- tion state by register control. energy detect+ tm allows the 88e6060 to wake up when energy is detected on the wire with the additional capabil- ity to wake up a link partner. the 10base-t link pulses are sent once ever y second while list ening for energy on the line. 4.3.2 mac interface and phy conf iguration for low power modes the 88e6060 has one config bit (in config_b - ta b l e 3 ) dedicated to support the low power modes. low power modes are also register programmable. the edet bit ( ta b l e 7 1 ) enables the user to turn on energy detect+ ? . when the low power mode is not selected, the pwrdwn bit ( ta b l e 6 1 ) can be used. if during the energy detect mode, the phy wakes up and starts operating in norm al mode, the edet bit settings are retained. when the link is lost and energy is no longer detected, the 88e6060 returns to the mode stored in the edet bit. table 31 shows how these modes are entered . 4.3.3 ieee power down mode the standard ieee power down mode is entered by setting the pwrdwn ( table 61 on page 107 ) bit equal to one. in this mode, the phy does not respond to any mac inte rface signals except the mdc/mdio. it also does not respond to any activity on the cat 5 cable. in this power down mode, the phy cannot wake up on its ow n by detecting activity on the cat 5 cable. it can only wake up by clearing the pwrdwn bit to 0. 4.3.3.1 energy detect + tm in this mode, the phy sends out a single 10 mbps nlp (normal link pulse) every second. if the 88e6060 is in energy detect+ tm mode, it can wake a connected device and it wakes upon the detection of a connected device. when ena_edet is 1, the mode of operation is energy detect+ tm . table 31: operating mode power consumption power mode est. power how to activate mode ieee power down see section 9. pwrdwn bit write ( table 61 ) energy detect+ tm same as ieee power down mode?see section 9. configuration option & register edet bit write ( ta b l e 7 1 )
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 78 document classification: proprieta ry information january 3, 2008, preliminary 4.4 far end fault indication (fefi) far-end fault indication provides a mechanism for transferri ng information from the local station to the link partner that a remote fault has occurred in 100base-fx mode (port 0 and port 1 only). a remote fault is an error in the link that one station can de tect while the other one cannot. an example of this is a disconnected fiber at a station?s transmitter. this station is receiving valid data and detects that the link is good via the link monitor, but is not able to detect that it s transmission is not propa gating to the other station. a 100base-fx station that detect s this remote fault modifies its transmitted idle stream pattern from all ones to a group of 84 ones followed by one zero. this is referred to as the fefi idle pattern. the fefi function is controlled by config_a connection and the disfefi bit ( ta b l e 7 1 ). table 32 shows the various configuration settings affecting the fefi function on hardware reset. 4.5 virtual cable tester? the 88e6060 phy virtual cable tester? feature uses time domain reflectometry (tdr) to determine the quality of the cables, connectors, and terminat ions. some of the possible problems that can be diagnosed include opens, shorts, cable impedance mismatches, bad connect ors, termination mismatches, and bad magnetics. the 88e6060 switch core conducts a cable diagnostic te st by transmitting a signal of known amplitude (+1v) sequentially along each of the tx and rx pairs of an at tached cable. the transmitted signal continues along the cable until it is reflected from a cabl e imperfection. the magnitude of this echo signal and its return time are shown in the vct registers ( table 81 and ta b l e 8 2 ) on the amprfln and distrfln bits respectively. using the information from these registers, the vct registers ( table 81 and table 82), the distance to the problem location and the type of problem can be determined. fo r example, the echo time can be converted to distance using ta b l e 8 1 and figure 26 . the polarity and magnitude of the reflection together with t he distance indicates the type of discontinuity. for example, a +1v reflection indicates an open close to the phy and a -1v reflection indi- cates a short close to the phy. when the cable diagnostic feature is activated by setting the envct bit to one ( table 81 ), a pre-determined amount of time elapses before a test puls e is transmitted. this is to ensure th at the link partner loses link, so that it stops sending100base-tx idles or 10 mbit data packets. th is is necessary to be able to perform the tdr test. the tdr test can be performed either when there is no lin k partner or when the link pa rtner is auto-negotiating or sending 10 mbit idle link pulses. if the 88e6060 device receiv es a continuous signal for 125 ms, it declares a test table 32: fefi select pn_config 1 1. "n" in pn_config may only take the values "0" or "1" en_fefi (config_a) fefi disfefi bit ( table 71 ) auto-negotiation x disabled hw set to 1 10base-t x disabled hw set to 1 100base-tx x disabled hw set to 1 100base-fx 0 disabled hw set to 1 100base-fx 1 enabled hw reset to 0
physical interface (phy) functional description auto mdi/mdix crossover copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 79 failure because it cannot start the tdr test. in the test failure case, the received data is not valid. the results of the test are also summarized in the vcttst bits ( table 81 and ta b l e 8 2 ). ? 11 = test fail (the tdr test could not be run for reasons explained above) ? 00 = valid test, normal cable (no short or open in cable) ? 10 = valid test, open in cable (impedance > 333 ohms) ? 01 = valid test, short in cable (impedance < 33 ohms) the definition for shorts and opens is arbitrary and can be user-defined using the information in the vct registers the impedance mismatch at the location of the discontinuity can also be calculated knowing the magnitude of the echo signal. refer to the application note ?virtual cable tester -- how to use tdr results? for details. 4.6 auto mdi/mdix crossover the 88e6060 device automatically determines whether or not it needs to interchan ge cable sense between pairs so that an external crossover cable is not required. if the 88e6060 device interoperates with a device that cannot automatically correct for crossover, the 88e6060 phy ma kes the necessary adjustm ent prior to commencing auto-negotiation. if the 88e6060 device interoperates with a device that implements mdi/mdix crossover, a ran- dom algorithm as described in ieee 802.3 section 40.4 .4 determines which device performs the crossover. when the 88e6060 device interoperates with legacy 10base-t device s that do not implemen t auto-negotiation, it follows the same algorithm as described above since li nk pulses are present. however, when interoperating with legacy 100base-tx devices that do not implement auto-n egotiation (i.e. link pulses are not present), the 88e6060 device uses signal detection to dete rmine whether to implement crossover. the auto mdi/mdix crossover function can be disabled via the automdi[x] bits ( table 71 ). the 88e6060 device is set to normal mode by default if auto mdi/mdix crossover is disabled at hardware reset. the pin mapping in normal and crossed modes is specified in table 33. table 33: mdi/mdix pin functions physical pin normal crossed 100base-tx 10base-t 100base-tx 10base-t txp/txn transmit transmit receive receive rxp/rxn receive receiv e transmit transmit
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 80 document classification: proprieta ry information january 3, 2008, preliminary 4.7 led interface the led interface pins can either be controlled by a port?s phy or controlled directly, independently of the state of the phy. four display options are available and both interf aces (phy or direct control) can be used together to provide even more led combinations. the leds can be controlled through a parallel interface or via a serial interface that supports up to seven leds per port. direct control of the parallel leds is achieved by writing to the phy manual led override register ( ta b l e 8 0 ). any of the parallel leds can be turned on, off, or made to b link at variable rates independent of the state of the phy. when the parallel leds are controlled by a port?s phy, th eir activity is determined by the phy?s state. each led can be programmed to indicate various phy states, with variable blink rate. some port phy events configured to drive led pins are too short to result in an observable change in led state. for these events, pulse stretching is provided to transl ate a short phy event into an observable led response. the duration of a pulse stretch can be programmed via the pulsestretch bits ( ta b l e 7 9 ). the default pulse stretch duration is set to 170 to 340 ms. the pulse stretch duration applies to all applicable led interface pins. some of the status indicators signal multiple events by t oggling led interface pins resulting in a blinking led. the blink period can be programmed via the blinkrate bits ( ta b l e 7 9 ). the default blink period is set to 84 ms. the blink rate information is true for all applicable leds. 4.7.1 parallel led interface these led interface pins can be used to display port stat us information. the led interface has three different sta- tus indicators for each port with four different default display options, using the px_led2, px_led1, and px_led0 pins.the led parallel select register ( table 77 ) specifies which single led mode status to display on the led interface pins. the default display for each mode is s hown in table 34 and the default mode that applies depends upon the hardware configuration established using the config_a pin?see ta b l e 3 . table 34: parallel led hardware defaults led mode ?set by config_a at reset p[4:0]_led2 p[4:0]_led1 p[4:0]_led0 0linkrxtx 1 link act speed 2 link/rx tx speed 3 link/act duplex/colx speed
physical interface (phy) functional description led interface copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 81 table 77 shows additional display modes that can be set up by software after startup. table 35 defines all the pos- sible parallel led display modes. table 35: parallel led display interpretation status description colx low = collision activity high = no collision activity this status is pulse stretched to 170 ms. error low = jabber, received error, false carrier, or fifo over/underflow occurred high = none of the above occurred this status is pulse stretched to 170 ms. duplex low = full-duplex high = half-duplex duplex/colx low = full-duplex high = half-duplex blink = collision activity (blink rate is 84ms active then 84 ms inactive) the collision activity is pulse stretched to 84 ms. speed low = speed is 100 mbps high = speed is 10 mbps link low = link up high = link down tx low = transmit activity high = no transmit activity this status is pulse stretched to 170 ms. rx low = receive activity high = no receive activity this status is pulse stretched to 170 ms. act low = transmit or received activity high = no transmit or receive activity this status is pulse stretched to 170 ms. link/rx low = link up high = link down blink = receive activity (blink rate is 84 ms active then 84ms inactive) the receive activity is pulse stretched to 84 ms. act (blink mode) high = no transmit or receive activity blink = transmit or receive activity (blink rate is 84ms active then 84 ms inactive) the transmit and receive activity is pulse stretched to 84 ms.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 82 document classification: proprieta ry information january 3, 2008, preliminary 4.7.2 serial led interface the ledser, ledena, and ledclk pins are used for the serial interface. the config_a pin (see ta b l e 3 ) is used to select one of four possible default led modes wh ich are numerically the same as those set up by the config_a pin in the parallel case. however, the default displays in this case are different. t he serial led interface can display a variety of different status indications in 100base-tx and 10base-t modes (see table 37 and table 38). status to display, pulse stretching, and blink speed can be programmed via the led stream select for serial leds register ( ta b l e 7 8 ) 4.7.2.1 single and dual led modes single led mode is initiated at power up according to the config_a pin settings. dual led mode can be entered (or single mode can be reentered) by setti ng the appropriate register fields according to table 78 . these registers enable led functions to be programmed individua lly in dual led or single led mode. the data states are sent out on the ledser data stream and may be extracted using the ledclk and ledena signals as shown in the example of figure 20. 4.7.2.2 single led modes in the single led display mode, the sa me status is driven on both status 100 and status 10 positions in the bit stream. however, the ledena signal asserts only over the status that is set and de-asserts over the other posi- tion that is turned off in the bit stream. for exam ple, duplex shows the same status for duplex100 and duplex10. however, the ledena signal is high over duplex100 position only for one clock period. refer to figure 20 for more details. 4.7.2.3 dual led display mode in the dual led display mode, two leds are used: one for 10 mbps, and one for 100 mbps activity. a different sta- tus is driven on status 100 and status 10 positions in the bit stream. in this case, the ledena signal asserts over both 100 and 10 positions in this stream. for example, the ledena signal asserts over colx100 and colx10 in figure 20. and remains high for two clock periods. if one of these status bits (shown in table 38) is turned off, ledena is not asserted in both positions. figure 20: serial ledena high clocking with colx in dual mode, error off, and duplex in single mode colx100 colx10 ledclk 80 ns error 80 ns 160 ns duplex 160 ns dual led mode ledena high off ledena not asserted single led mode ledena asserted only 1 clock period ledena ledser
physical interface (phy) functional description led interface copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 83 the bit stream on ledser can be clocked into a shift re gister with ledena as the shift enable signal as shown in figure 21. the rate of update of the serial led inte rface is controlled by progra mming register 24.8:6. the default value is set to 42 ms. figure 21: serial led conversion after the led data is shifted into the correct position, the shift sequence is suspended to allow the appropriate leds to light or extinguish depending on status. the led implementation used in the 88e6060 device is self-syn- chronizing. the default display options are given in ta b l e 3 6 . the led status bits are output in the order shown on th e ledser pin synchronously with ledclk. all status sig- nals for port 5 are sent out first followed by those for port s 4 through 0. each bit in the stream occupies a period of 80 ns. figure 22: serial led display order?(if all are selected) table 37 and table 38 show the status events that can be displayed by programming the 88e6060 device in single and dual led display modes. table 36: serial led disp lay options (a = active) led mode colx error duplex duplex /colx speed link tx rx act link /rx link /act 0- a - a a -----a 1a a a - a -a--a- 2a a a - a a--a-- 3a a a - a aaa--- shift register ledser ledena ledclk in ena
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 84 document classification: proprieta ry information january 3, 2008, preliminary 4.7.2.4 single led display mode table 37: single led display mode status description colx low = collision activity high = no collision activity this status has a default pulse stretch duration of 170 ms. error low = jabber, received error, false carrier, or fifo over/underflow occurred high = none of the above occurred this status has a default pulse stretch duration of 170 ms. duplex low = full-duplex high = half-duplex duplex/colx low = full-duplex high = half-duplex blink = collision activity (blink rate is 84 ms active then 84 ms inactive) the collision activity is pulse stretched to 84 ms. speed low = speed is 100 mbps high = speed is 10 mbps link low = link up high = link down tx low = transmit activity high = no transmit activity this status is pulse stretched to 170 ms. rx low = receive activity high = no receive activity this status is pulse stretched to 170 ms. act low = transmit or received activity high = no transmit or receive activity this status is pulse stretched to 170 ms. link/rx low = link up high = link down blink = receive activity (blink rate is 84 ms active then 84 ms inactive) the receive activity is pulse stretched to 84 ms. link/act low = link up high = link down blink = transmit or receive activity (blink rate is 84ms active then 84 ms inactive) the transmit and receive activity is pulse stretched to 170 ms.
physical interface (phy) functional description led interface copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 85 4.7.2.5 dual led display mode table 38: dual led display mode event description colx100 0 = 100 mbps collision activity 1 = no 100 mbps collision activity this status can be pulse stretched. colx10 0 = 10 mbps collision activity 1 = no 10 mbps collision activity this status can be pulse stretched. error100 0 = received error, false carrier, or 100 mbps fifo over/underflow occurred. 1 = none of the above occurred this status can be pulse stretched. error10 0 = jabber or 10 mbps fifo over/underflow occurred 1 = none of the above occurred this status can be pulse stretched. duplex100 0 = 100 mbps full-duplex 1 = 100 mbps half-duplex duplex10 0 = 10 mbps full-duplex 1 = 10 mbps half-duplex duplex/colx100 0 = 100 mbps full-duplex 1 = half-duplex blink = 100 mbps collision activity the collision activity can be pulse stretched. the blink rate can be programmed. duplex/colx10 0 = 10 mbps full-duplex 1 = 10 mbps half-duplex blink = 10 mbps collision activity the collision activity can be pulse stretched. the blink rate can be programmed. speed100 0 = speed is 100 mbps 1 = speed is 10 mbps speed10 0 = speed is 10 mbps 1 = speed is 100 mbps link100 0 = 100 mbps link up 1 = 100 mbps link down link10 0 = 10 mbps link up 1 = 10 mbps link down tx100 0 = 100 mbps transmit activity 1 = no 100 mbps transmit activity this status can be pulse stretched. tx10 0 = 10 mbps transmit activity 1 = no 10 mbps transmit activity this status can be pulse stretched. rx100 0 = 100 mbps receive activity 1 = no 100 mbps receive activity this status can be pulse stretched.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 86 document classification: proprieta ry information january 3, 2008, preliminary rx10 0 = 10 mbps receive activity 1 = no 10 mbps receive activity this status can be pulse stretched. act100 0 = 100 mbps transmit or 100 mbps receive activity 1 = no 100 mbps transmit or 100 mbps receive activity this status can be pulse stretched. act10 0 = 10 mbps transmit or 10 mbps receive activity 1 = no 10 mbps transmit or 10 mbps receive activity this status can be pulse stretched. link/rx100 0 = 100 mbps link up 1 = 100 mbps link down blink = 100 mbps receive activity the receive activity can be pulse stretched. the blink rate can be programmed. link/rx10 0 = 10 mbps link up 1 = 10 mbps link down blink = 10 mbps receive activity the receive activity is can be pulse stretched the blink rate can be programmed. link/act100 0 = 100 mbps link up 1 = 100 mbps link down blink = 100 mbps transmit or 100 mbps receive activity the transmit and receive activity can be pulse stretched. the blink rate can be programmed. link/act10 0 = 10 mbps link up 1 = 10 mbps link down blink = 10 mbps transmit or 10 mbps receive activity the transmit and receive activity can be pulse stretched. the blink rate can be programmed. table 38: dual led display mode (continued) event description
serial management interface (smi) mdc/mdio read and write operations copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 87 section 5. serial management interface (smi) the 88e6060 serial management interface provides access to the internal registers via the mdc and mdio sig- nals and is compliant with i eee 802.3u clause 22. mdc is the manage ment data clock input whose frequency can run from dc to a maximum rate of 8.3 mhz. mdio is the management data input/output which carries a bidirec- tional signal that runs synchronously with the mdc. the mdio pin requires a pull-up resistor to pull the mdio high during idle and turnaround times. 5.1 mdc/mdio read and write operations all of the relevant serial management registers, as we ll as several optional registers, are implemented in the 88e6060 switch core. a description of these registers can be found in section 6. "switch register description" . note access to the 88e6060 device?s switch and phy register s is not possible when the serial eeprom is still loading the registers. a cpu can monitor the 88e6060 device intn pin, which will go active (low) when the serial eeprom has been fully processed (i.e., a halt instruct ion has been reached - see section 8).
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 88 document classification: proprieta ry information january 3, 2008, preliminary figure 23: typical mdc/mdio read operation figure 24: typical mdc/mdio write operation table 39 shows an example of a read operation of phy address 04, register 0, with data of 04c0. table 39: serial management interface protocol example 32-bit preamble start of frame opcode read = 10 write = 01 5-bit phy device address 5-bit phy register address 2-bit turnaround read = z0 write = 10 16-bit data field idle 11111111 01 10 00100 00000 z0 000 0010011000000 111 11111 mdc mdio (sta) 0 1 1 0 0 1 1 0 0 0 0 0 0 0 z 0 0 0 0 1 0 0 1 1 0 0 0 0 0 0 0 0 z z mdio (phy) z z z z preamble start opcode (read) phy address register address ta register data idle mdc mdio (sta) 0 1 0 1 0 1 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 z z z preamble start opcode (write) phy address register address ta register data idle 0 z
switch register description copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 89 section 6. switch register description all of the 88e6060 registers are accessible using the ieee serial management interface (smi) used for phy devices (see mdc/mdio pin description in section 1 and section 5 of this datasheet). the 88e6060 device uses 16 of the 32 possible device addresses. the16 device addresses are configurable at reset by use of the ee_clk/addr4 pin (see signal description in part 1 of this datasheet). figure 25 shows the register map assum- ing the lower 16 smi device addresses are being used. figure 25: 88e6060 register map phy control 0123456789abcdef 0 1 2 3 4 5 6 7 8 9 a b c d e f 10 11 12 13 14 15 16 17 18 19 1a 1b 1c phy status phy identifier phy identifier auto-neg advertisement link partner ability auto-neg expansion next page transmit link partner next page reserved 1d 1e 1f phy specific control 1 phy specific status phy interrupt enable phy interrupt status interrupt port summary receive error counter led parallel select led stream select led control reserved reserved rx frame counter tx frame counter port status global status port control global control reserved reserved atu-mac atu operation port based vlan map reserved reserved switch identifier smi register address smi device address atu control atu data switch-mac reserved free plus port association vector reserved reserved reserved led override vct control vct status phy specific control 2
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 90 document classification: proprietary information january 3, 2008, preliminary 6.1 register types the registers in the 88e6060 device are made up of one or more fields. the way in which each of these fields operate is defined by the field?s type. the function of each type is described in table 40 . 6.2 switch core registers switch core registers are of two types: ? switch port registers?are separately configured for each port comprised in the switch. each port has a unique associated smi device address and this address is used to access that port ?s switch register set. ? switch global registers?are configured using a singl e smi device address. a global register?s setting affects all ports of the switch. the 88e6060 contains six ports (macs). the supported ports are accessible using smi device addresses 0x08 to 0x0d, or 0x18 to 0x1d depending upon the value of the ee_clk/addr4 pin at reset (section 1). the macs are fully ieee802.3 compliant. sinc e there is no ieee standard covering required mac regi sters, these registers are 88e6060 device specific. the switch contains many global registers that are used to control features and functions that are common to all ports in the switch. the global registers are accessibl e using smi device address 0x0f (or 0x1f) depending upon the value of the ee_clk/addr4 pin at reset. table 40: register types type description lh register field with latching high function. if status is high, then the register is set to a one and remains set until a read operation is performed through the management interface or a reset occurs. ll register field with latching low function. if status is low, then the register is cleared to a zero and remains cleared until a read operation is performed through the management interface or until a reset occurs. res reserved for future use. all reserved bits are read as zero unless otherwise noted. ro read only. r/w read and write with no definable initial value rwr read/write reset. all bits are readable and writable. after reset the register field is cleared to zero. rws read/write set. all bits are readable and writable. after reset the register field is set to a non-zero value specified in the text. sc self-clear. writing a one to this register causes the required function to be immediately executed, then the register field is cleared to zero when the function is complete. update value written to the register field does not take effect until a soft reset is executed. retain value written to the register field d0 wo write only. reads to this type of register field return undefined data.
switch register description switch core registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 91 6.2.1 switch core register map table 41: switch core register map description offset hex offset decimal page number switch port registers (smi device address) port status register 0x00 0 page 92 reserved registers 0x01 - 0x02 1 - 2 page 93 switch identifier register 0x03 3 page 93 port control register 0x04 4 page 94 reserved registers 0x05 5 page 95 port based vlan map 0x06 6 page 96 reserved registers 0x07 - 0x0a 7 - 10 page 96 port association vector 0x0b 11 page 97 reserved registers 0x0c - 0x0f 12 - 15 page 97 rx counter 0x10 16 page 98 tx counter 0x11 17 page 98 reserved registers 0x12 - 0x1f 18 - 31 page 98 switch global registers switch global status register 0x00 0 page 99 switch mac address register bytes 0 & 1 0x01 1 page 100 switch mac address register bytes 2 & 3 0x02 2 page 100 switch mac address register bytes 4 & 5 0x03 3 page 100 switch global control register 0x04 4 page 101 reserved registers 0x05 - 0x09 5 - 9 page 101 atu control register 0x0a 10 page 102 atu operation register 0x0b 11 page 103 atu data register 0x0c 12 page 104 atu mac address register bytes 0 & 1 0x0d 13 page 104 atu switch mac address regi ster bytes 2 & 3 0x0e 14 page 104 atu switch mac address regi ster bytes 4 & 5 0x0f 15 page 105 reserved registers 0x10 - 0x1f 16 - 31 page 105
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 92 document classification: proprietary information january 3, 2008, preliminary 6.2.2 switch port registers table 42: port status register offset: 0x00 (hex), or 0 (decimal) bits field type description 15 linkpause ro link partner?s pause bit, returned from the link partner through auto-negotia- tion. this bit is valid for ports with phys and when the resolved bit is set to a one. 0 = mac pause not implemented in the link partner 1 = mac pause is implemented in the link partner 14 mypause ro my pause bit, sent to the link partner during auto-negotiation. this bit is valid for ports with phys. it is set high if fd_flow_dis ( section 1 ) is low during reset. 0 = mac pause not implemented in the local mac 1 = mac pause is implemented in the local mac 13 resolved ro link mode is resolved. 0 = link is undergoing auto-negotiation or the port is disabled 1 = link has determined its speed, duplex and linkpause settings 12 link ro link status in real time (i.e., it is not latched). 0 = link is down 1 = link is up 11 portmode ro port mode. the value of this bit is always a one for ports 0 thru 4 and it comes from px_mode3 during configuration for port 5 and port 4. 0 = sni mode is enabled 1 = mii 10/100 or rmii 100 mbps mode is enabled 10 phymode ro phy mode. this bit is valid for all ports but it is meaningful for port 5 and port 4 only and only when the portmode (bit 11 above) indicates an mii mode of operation (i.e., portmode is a one). this bit may be either value for sni con- figured ports ? but all sni port configurations are phy mode. the value of this bit is always zero for ports 0 to 3 (and port 4 if port 4?s mii is disabled) and it comes from px_mode2 during configuration for port 5 and port 4 (if port 4?s mii is enabled). 0 = pins are in mii mac mode (i.e., inclk and outclk are inputs) or the pins are in rmii phy mode (all rmii modes are phy mode where inclk and outclk are outputs). 1 = pins are in mii phy mode (i.e., inclk and outclk are output) 9duplex ro r/w on port 4 & 5 only duplex mode. this bit is valid when the resolved bit is set to a one. 0 = half-duplex 1 = full-duplex this bit can be written to on port 5 and port 4 only so that the port?s duplex can be modified after reset occurs. care must be taken to ensure this port?s duplex matches the duplex of its link partner.
switch register description switch core registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 93 the portmode, phymode, duplex, and speed bits for port 5 and port 4 come from t he px_mode[3:0] pins at reset (if port 4 is in mii mode). wh en the portmode is a one, the phymode, duplex, and speed bits map the same as the other 10/100 ports that contain phys. when the portmode is a zero the mapping is easier to see by looking at " rmii/mii/sni configuration" table 19: . note registers 0x01 and 0x02 (hex), or 1 and 2 (decimal) are reserved. 8 speed ro speed mode. this bit is valid when the resolved bit is set to a one and the portmode (bit 11 above) is also a one. if the portmode bit is a zero then the port?s speed is 10 mbps regardless of the value of this bit. when the port is configured in mii mac mode, the speed bit is undefined. 0 = 10 mbps 1 = 100 mbps 7:0 reserved res reserved for future use. table 43: switch identifier register offset: 0x03 (hex), or 3 (decimal) bits field type description 15:4 deviceid ro device identifier. the 88e6060 is identified by the value 0x060. 3:0 revid ro revision identifier. the initial version of the 88e6060 has a revid of 0x0. this revid field may change at any time. contact a marvell fae for current information on the device revision identifier. table 42: port status register (continued) offset: 0x00 (hex), or 0 (decimal) bits field type description
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 94 document classification: proprietary information january 3, 2008, preliminary table 44: port control register offset: 0x04 (hex), or 4 (decimal) bits field type description 15 forceflow control rwr force flow control. this bit is used to enable full-duplex flow control or half- duplex back pressure on this port depending upon the port?s current duplex. this bit can only be changed when the port is in the disabled port state (bits 1:0 below). 0 = use configuration pin settings for flow control/back pressure 1 = force flow control/back pressure to be enabled on this port 14 trailer rwr egress trailer mode (port 4 and 5 only). 0 = normal mode ? frames are transmitted unmodified 1 = add a trailer to all frames transmitted out of the port. egress trailer mode is intended for management cpu ports for source port information on bpdu frame. 13:12 reserved res reserved for future use. 11 header rwr the detailed information regarding this feature requires an nda with marvell ? semiconductor. please contact your local marvell sales represen- tative for more information. 10:9 reserved res reserved for future use. 8ingress mode rwr ingress mode (port 4 and 5 only). 0 = normal mode ? frames are received unmodified 1 = cpu port with ingress trailer mode ? frames must be received with a trailer and the trailer is checked, potentially used and then removed from the frame. trailer mode is intended for ingress data control from a management cpu port so that bpdu frames can be directed. 7vlan tunnel rwr vlan tunnel. when this bit is cleared to a zero, the vlans defined in the vlantable (table 45, on page 96) are enforced for all frames. when this bit is set to a one, the vlantable masking is bypassed for any frame enter- ing this port with a da that is currently ?locked? in the atu. this includes uni- cast as well as multicast frames. 6:2 reserved res reserved for future use.
switch register description switch core registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 95 note register 0x05 (hex), or 5 (decimal) is reserved. 1:0 portstate rwr or rws to 0b11 port state. these bits are used to manage a port to determine what kind of frames, if any, are allowed to enter or leave a port for simple bridge loop detection or 802.1d spanning tree. the state of these bits can be changed at any time without disrupting frames currently in transit. the port states are: 00 = disabled. the switch port is completely disabled and it will not receive or transmit any frames. the qc will return any pre-allocated ingress queue buffers when the port is in this mode. 01 = blocking/listening. the switch will examine all frames without learning any sa addresses, and discard all but mgmt frames. mgmt (man- agement) frames are the only kind of frame that can be tunneled through blocked ports. a mgmt frame is any frame whose multicast da address appears in the atu database with the mgmt entry state. it will allow mgmt frames only to exit the port. this mode is used for bpdu handling for bridge loop detection and spanning tree support. 10 = learning. the switch will examine all frames, learning all sa addresses, and still discard all but mgmt frames. it will allow mgmt frames only to exit the port. 11 = forwarding. the switch will examine all frames, learning all sa addresses, and receive and transmit all frames like a normal switch. the portstate bits for all ports come up in the forwarding state unless the sw_mode[1:0] pins are set to 0b00, the cpu attached mode. software reset, (swreset - table 54 on page 102 ) causes the portstate bits to revert back to their reset state (i.e., either disabled or forwarding). table 44: port control register (continued) offset: 0x04 (hex), or 4 (decimal) bits field type description
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 96 document classification: proprietary information january 3, 2008, preliminary , note registers 0x07 - 0x0a (hex), or 7 - 10 (decimal) are reserved. table 45: port based vlan map offset: 0x06 (hex), or 6 (decimal) bits field type/ reset value description 15:12 dbnum rwr port?s default database vlan number. this field can be used with vlans to keep each vlan?s mac address mapping database separate from the other vlans. this allows the same mac address to appear multiple times in the address database (at most one time per vlan) with a different port mapping per entry. this field should be zero if not being used. it needs to be a unique number for each independent vlan if used. 11:6 reserved res reserved for future use. 5:0 vlantable rws to all ones except for this port?s bit port based vlan table. the bits in this table are used to restrict which output ports this input port can send frames to. these bits restrict where a port can send frames to (unless a vlantunnel frame is being received - see port control register, table 44 ). to send frames to port 0, bit 0 of this register must be a one. to send frames to port 1, bit 1 of this register must be a one, etc. after reset, all ports are accessible since all the other port number bits are set to a one. this port?s bit will always be zero at reset. this prevents frames exiting the port on which they arrived. this port?s bit can be set to a one in the 88e6060 switch core enabling frames to be returned on their ingress port; consequently care should be taken in writing code to control these bits. this register is reset to 0x3e for po rt 0 (smi device address 0x8), and it resets to 0x3d for port 1 (addr 0x9) and to 0x3b for port 2 (addr 0xa), etc.
switch register description switch core registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 97 note registers 0x0c - 0x0f (hex), or 12 - 15 (decimal) are reserved. table 46: port association vector offset: 0x0b (hex), or 11 (decimal) bits field type description 15 ingress monitor rwr ingress monitor enable. when this bit is set to a one ingress port monitoring is enabled on the same ports upon which egress port monitoring is enabled, as defined by the pav bits below. this is the recommended operation mode for port monitoring. 14:6 reserved res reserved for future use. 5:0 pav rws to all zeros except for this port?s bit port association vector for atu learning. the value of these bits is used as the port?s dpv on automatic atu learning or entry_state refresh whenever these bits contain a non-zero value. when these bits are all zero automatic learning and entry_state refresh is disabled on this port. for normal switch operation this port?s bit should be the only bit set in the vector. these bits must only be changed when frames are not entering the port (see portstate bits in port control ? table 44, on page 94). the pav bits can be used to set up an egress port monitor. to configure port 0 to receive a copy of the frames that exit port 1, set bit 0 in port 1?s pav reg- ister (along with port 1?s bit). to configure port 0 to get a copy of the frames that exit port 2, set bit 0 in port 2?s pav register, etc. the pav bits can be used to set up an ingress port monitor along with an egress port monitor. to configure port 0 to receive a copy of the frames that enter port 1 as well as exit port 1, set bit 0 in port 1?s pav register (along with port 1?s bit?bit 1) and set the ingressmonitor bit above. the pav bits can be used to set up port trunking (along with the vlantable bits ( table 45 on page 96 ). for the two ports that form a trunk, set both of the port?s bits in both port?s pav registers. then use the vlantable to isolate the two ports from each other and to steer the traffic from the other ports down the required trunk line of the pair.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 98 document classification: proprietary information january 3, 2008, preliminary note registers 0x12 - 0x1f (hex), or 18 - 31 (decimal) are reserved. table 47: rx counter offset: 0x10 (hex), or 16 (decimal) bits field type/ reset value description 15:0 rxctr ro received counter. when ctrmode is cleared to a zero (global control ? table 53, on page 101) this counter increments each time a good frame enters this port. it does not matter if the frame is switched or discarded by the switch. when ctrmode is set to a one this counter increments each time an error frame enters this port. an error frame is one that is 64 bytes or greater with a bad crc (including alignment errors but not dribbles). fragments and prop- erly formed frames are not counted. the counter wraps back to zero. the only time this counter does not incre- ment is when this port is disabled (see portstate, table 44, on page 94). this register can be cleared by changing the state of the ctrmode bit in the switch global control register ( table 53 on page 101 ). table 48: tx counter offset: 0x11 (hex), or 17 (decimal) bits field type/ reset value description 15:0 txctr ro transmit counter. when ctrmode is cleared to a zero (see global control ? table 53 on page 101 ) this counter increments each time a frame success- fully exits this port. when ctrmode is set to a one this counter increments each time a collision occurs during an attempted transmission. it no longer counts all transmitted frames ? but only those transmission attempts that resulted in a collision. the counter wraps back to zero. the only time this counter does not incre- ment is when this port is disabled (see portstate field of port control regis- ter (table 44, on page 94). this register can be cleared by changing the state of the ctrmode bit in global control ( table 53 on page 101 ).
switch register description switch core registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 99 6.2.3 switch global registers table 49: switch global status register offset: 0x00 (hex), or 0 (decimal) bits field type description 15:14 reserved res reserved for future use. 13:12 sw_mode ro switch mode. these bits return the value of the sw_mode[1:0] pins. 11 initready ro switch ready. this bit is set to a one when all the blocks inside the device have finished their initialization and are ready to accept frames. 10:4 reserved res reserved for future use. 3 atufull lh atu full interrupt. this bit is set to a one if the atu cannot load or learn a new mapping owing to all of the available locations for an address being locked. it is automatically cleared when read. this bit?s being high causes the 88e6060 intn pin to go low if the atufullinten bit in globalcontrol (table 53, on page 101) is set to a one. 2 atudone lh atu done interrupt. this bit is set to a one whenever the atubusy bit (table 55, on page 103) changes from a one to a zero. it is automatically cleared when read. this bit?s being high causes the 88e6060 intn pin to go low if the atudoneinten bit in global control (table 53, on page 101) is set to a one. 1 phyint ro phy interrupt. this bit is set to a one when the phys interrupt logic has at least one active interrupt. this bit?s being high causes the 88e6060 intn pin to go low if the phyinten bit in global control (table 53, on page 101) is set to a one. 0 eeint lh eeprom done interrupt. this bit is set to a one after the eeprom finishes loading registers, and it is automatically cleared when read. this bit?s being high causes the 88e6060 intn pin to go low when the eeinten bit in global control ( table 53 on page 101 ) is set to a one.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 100 document classification: proprietary information january 3, 2008, preliminary table 50: switch mac address register bytes 0 & 1 offset: 0x01 (hex), or 1 (decimal) bits field type description 15:9 macbyte0 rwr mac address byte 0 (bits 47:41) is used as the switch?s source address (sa) in transmitted full-duplex pause frames. since bit 0 of byte 0 (bit 40) is the multicast bit (it is the 1 st bit down the wire) it is always transmitted as a zero, and its value cannot be changed. 8 diffaddr rwr different mac addresses per port. this bit is used to have all ports transmit the same or different source addr esses in full-duplex pause frames. 0 = all ports transmit the same sa 1 = each port uses a different sa where bits 47:3 of the mac address are the same, but bits 2:0 are the port number (port 0 = 0, port 1 = 1, and so on.) 7:0 macbyte1 rwr mac address byte 1 (bits 39:32) is used as the switch?s source address (sa) in transmitted full-duplex pause frames. table 51: switch mac address register bytes 2 & 3 offset: 0x02 (hex), or 2 (decimal) bits field type description 15:8 macbyte2 rwr mac address byte 2 (bits 31:24) is used as the switch?s source address (sa) in transmitted full-duplex pause frames. 7:0 macbyte3 rwr mac address byte 3 (bits 23:16) is used as the switch?s source address (sa) in transmitted full-duplex pause frames. table 52: switch mac address register bytes 4 & 5 offset: 0x03 (hex), or 3 (decimal) bits field type description 15:8 macbyte4 rwr mac address byte 4 (bits 15:8) is used as the switch?s source address (sa) in transmitted full-duplex pause frames. 7:0 macbyte5 rwr mac address byte 5 (bits 7:0) is used as the switch?s source address (sa) in transmitted full-duplex pause frames. note: bits 2:0 of this register are ignored when diffaddr is set to a one.
switch register description switch core registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 101 note registers 0x05 - 0x09 (hex), or 5 - 9 (decimal) are reserved. table 53: switch global control register offset: 0x04 (hex), or 4 (decimal) bits field type description 15:14 reserved res reserved for future use. 13 discard excessive rwr discard frames with excessive collisions. when this bit is set to a one frames that encounter 16 consecutive collisions are discarded. when this bit is cleared to a zero egress frames are never discarded and the backoff range is reset after 16 consecutive collisions on a single frame. 12:11 reserved res reserved for future use. 10 maxframe size rwr maximum frame size allowed. the ingress block discards all frames that are less than 64 bytes in size. it also discards all frames that are greater than a certain size as follows: 0 = max size is 1522 if ieee 802.3ac tagged, or 1518 if not tagged 1 = max size is 1536 9 reload sc reload the registers using the eeprom. when this bit is set to a one the contents of the external eeprom are used to load the registers just as if a reset had occurred. when the reload operation is done, this bit is cleared to a zero automatically, and the eeint interrupt is set. 8 ctrmode rwr counter mode. when this bit is set to a one, the rx counters for all ports (table 47, on page 98) count rx errors and the tx counters for all ports (table 48, on page 98) count tx collisions. when this bit is cleared to a zero, the rx counters for all ports count rx frames and the tx counters for all ports count tx frames. the rx counters and the tx counters for all ports are cleared to a zero whenever this bit changes state (i.e., it transitions from a one to a zero or from a zero to a one). 7:4 reserved res reserved for future use. 3 atufull inten rwr atu full interrupt enable. this bit must be set to a one to allow the atu full interrupt to drive the 88e6060 intn pin low. 2atudone inten rwr atu done interrupt enable. this bit must be set to a one to allow the atu done interrupt to drive the 88e6060 intn pin low. 1 phyinten rwr phy interrupt enable. this bit must be set to a one to allow active interrupts enabled in phy registers 0x12 to drive the 88e6060 intn pin low. 0 eeinten rws eeprom done interrupt enable. this bit must be set to a one to allow the eeprom done interrupt to driv e the 88e6060 intn pin low.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 102 document classification: proprietary information january 3, 2008, preliminary table 54: atu control register offset: 0x0a (hex), or 10 (decimal) bits field type description 15 swreset sc switch software reset. writing a one to this bit causes the qc and the mac state machines in the switch to be reset. register values are not modified, except for the portstate bits ( table 44 on page 94 ) and the eeprom is not re-read. the phys and atu are not affected by this bit. when the reset oper- ation finishes, this bit is cleared to a zero automatically. the reset occurs immediately. to prevent transmission of crc frames, set all of the ports to the disabled state ( table 44 on page 94 ) and wait for 2 ms (i.e., the time for a maximum frame to be transmitted at 10 mbps) before setting the swreset bit to a one. 14 learndis rwr atu learn disable. 0 = normal operation, learning is determined by the portstate (table 44, on page 94) 1 = automatic learning is disabled on all ports ? cpu atu loads still work 13:12 atusize rws to 0x2 address translation unit table size. the initial size of the atu database is 1024 entries. the size of the atu database can be modified at any time, but an atu reset and a switch software reset (bit 15 above) occurs automati- cally if the new atusize is different from the old atusize. 00 = 256 entry address database 01 = 512 entry address database 10 = 1024 entry address database = default 11 = reserved 11:4 agetime rws to 0x13 atu age time. these bits determine the time that each atu entry remains valid in the database, since its last access as a source address, before being purged. the value in this register times 16 is the age time in seconds. for example: the default value of 0x13 is 19 decimal. 19 x 16 = 304 seconds or just over 5 minutes. the minimum age time is 0x1 or 16 seconds. the maximum age time is 0xff or 4080 seconds or 68 minutes. when the agetime is set to 0x0 the aging function is disabled, and all learned addresses will remain in the database forever. 3:0 reserved res reserved for future use.
switch register description switch core registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 103 table 55: atu operation register offset: 0x0b (hex), or 11 (decimal) bits field type/ reset value description 15 atubusy sc address translation unit busy. this bit must be set to a one to start an atu operation (see atuop below). since only one atu operation can be execut- ing at one time, this bit must be zero before setting it to a one. when the requested atu operation completes, this bit is automatically cleared to a zero. the transition of this bit from a one to a zero can be used to generate an interrupt (table 49, on page 99). 14:12 atuop rwr address translation unit table opcode. the 88e6060 device supports the following atu operations. all of these operations can be executed while frames are transiting through the switch: 000 = no operation 001 = flush all entries 1 010 = flush all unlocked entries 2 011 = load or purge an entry 3 in a particular dbnum database 100 = get next from a particular dbnum database 4 101 = flush all entries in a particular dbnum database 110 = flush all unlocked entries in a particular dbnum database 111 = reserved 1. all frames flood until new addresses are learned. 2. an unlocked entry is any unicast address with an entrystate less than 0xf. 3. an entry is loaded if the entrystate (table 56, on page 104) is non-zero. an entry is purged if it exists and if the entrystat e is zero. 4. a get next operation finds the next higher mac address current ly in a particular atu database (defined by the dbnum field). the atubyte[5:0] values (table 57, on page 104) are used as the address to start from. to find the lowest mac address set atubyte[5:0] to ones. when the operation is done atubyte[5:0] contains the next higher mac address found. to find the next address simply issue the get next opcode again. if atubyte[5:0] is returned set to all ones, no higher mac address was found (i f entrystate = 0) or the broadcast address was found (if entrystate 0) in either case, the end of the table was reached. to search for a particular address, perform a get next operation using a mac address with a value one less than the one being searched fo r. 11:4 reserved res reserved for future use. 3:0 dbnum rwr atu mac address database number . if multiple address databases are not being used, these bits must remain zero. if multiple address databases are being used these bits are used to set the required address database number that is to be used on the database supported commands (atuops 0x3, 0x4, 0x5, and 0x6 above).
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 104 document classification: proprietary information january 3, 2008, preliminary table 56: atu data register offset: 0x0c (hex), or 12 (decimal) bits field type description 15:10 reserved res reserv ed for future use. 9:4 portvec rwr port vector. these bits are used as the input port vector for atu load oper- ation. it is the resulting port vector from the atu get next operation. 3:0 entrystate rwr atu entry state. these bits are used as the input entry state for atu load or purge operations. it is the resulting entry state from the atu get next operation. (see ta b l e 2 2 for a description of this function.) table 57: atu switch mac address register bytes 0 & 1 offset: 0x0d (hex), or 13 (decimal) bits field type description 15:8 atubyte0 rwr atu mac address byte 0 (bits 47:40) is used as the input mac address for atu load, purge, or get next operations. it is the resulting mac address from atu get next operation. bit 0 of byte 0 (bit 40) is the multicast bit (it is the 1 st bit down the wire). any mac address with the multicast bit set to a one is considered locked by the atu. 7:0 atubyte1 rwr atu mac address byte 1 (bits 39:32) is used as the input mac address for atu load, purge, or get next operations. it is the resulting mac address from the atu get next operation. table 58: atu switch mac address register bytes 2 & 3 offset: 0x0e (hex), or 14 (decimal) bits field type description 15:8 atubyte2 rwr atu mac address byte 2 (bits 31:24) is used as the input mac address for atu load, purge or get next operations. it is the resulting mac address from atu get next operations. 7:0 atubyte3 rwr atu mac address byte 3 (bits 23:16) that are used as the input mac address for atu load, purge or get next operations. it is the resulting mac address from atu get next operations.
switch register description switch core registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 105 note registers 0x10 - 0x1f (hex), or 16 - 31 (decimal) are reserved. table 59: atu switch mac addr ess register bytes 4 & 5 offset: 0x0f (hex), or 15 (decimal) bits field type description 15:8 atubyte4 rwr atu mac address byte 4 (bits 15:8) is used as the input mac address for atu load, purge or get next operations. it is the resulting mac address from atu get next operations. 7:0 atubyte5 rwr atu mac address byte 5 (bits 7:0) is used as the input mac address for atu load, purge or get next operations. it is the resulting mac address from atu get next operations.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 106 document classification: proprietary information january 3, 2008, preliminary section 7. phy registers the 88e6060 device contains five physical layer devices (phys). these devices are accessible using smi device addresses 0x00 to 0x04 (or 0x10 to 0x14) depending upon the value of the ee_clk/addr4 pin at reset. the phys are fully ieee 802. 3 compliant including t heir register interface. the phys in the 88e6060 device are identical to those in the marvell ? 88e3082 octal transceiver except that there are five transceivers (transceivers 5 to 7 do not exist and are not accessible). table 60: phy register map description offset hex offset decimal page number phy control register 0x00 0 page 107 phy status register 0x01 1 page 109 phy identifier 0x02 2 page 110 phy identifier 0x03 3 page 110 auto-negotiation advertisement register 0x04 4 page 111 link partner ability register (base page) 0x05 5 page 113 link partner ability register (next page) 0x05 5 page 113 auto-negotiation expansion register 0x06 6 page 114 next page transmit register 0x07 7 page 115 link partner next page register 0x08 8 page 115 reserved registers 0x09-0x0f 9 - 15 page 116 phy specific control register i 0x10 16 page 116 phy specific status register 0x11 17 page 119 phy interrupt enable 0x12 18 page 121 phy interrupt status 0x13 19 page 122 phy interrupt port summary (common register to all ports) 0x14 20 page 123 phy receive error counter 0x15 21 page 124 led parallel select register 0x16 22 page 125 led stream select register 0x17 23 page 126 phy led control register 0x18 24 page 128 phy manual led override register 0x19 25 page 129 vct? control register 0x1a 26 page 130 vct? status register 0x1b 27 page 131 phy specific control register ii 0x1c 28 page 132 reserved registers 0x1d to 0x1f 29 - 31 page 132
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 107 table 61: phy control register offset: 0x00 (hex ), or 0 (decimal) bits field mode hw rst sw rst description 15 swreset sc 0x0 self clear phy software reset writing a 1 to this bit causes the phy state machines to be reset. when the reset operation is done, this bit is cleared to 0 automatically. the reset occurs immedi- ately. 1 = phy reset 0 = normal operation 14 loopback r/w retain 0 enable loopback mode when loopback mode is activated, the transmitter data presented on txd is looped back to rxd internally. the phy has to be in forced 10 or 100 mbps mode. auto- negotiation must be disabled. 1 = enable loopback 0 = disable loopback 13 speedlsb r/w aneg [2:0] update speed selection (lsb) when a speed change occurs, the phy drops link and tries to determine speed when auto-negotiation is on. speed, auto-negotiation enable, and duplex enable take on the values set by aneg[2:0] on hardware reset. a write to these registers has no effect unless any one of the following also occurs: software reset is asserted (bit 15) or power down (bit 11) transitions from power down to nor- mal operation. 1 = 100 mbps 0 = 10 mbps 12 anegen r/w aneg [2:0] update auto-negotiation enable speed, auto-negotiation enable, and duplex enable take on the values set by aneg[2:0] on hardware reset. a write to these registers has no effect unless any one of the following also occurs: software reset is asserted (bit 15, above), power down (bit 11, below), or transitions from power down to normal operation. if the anegen bit is set to 0, the speed and duplex bits of the phy control register (offset 0x00) take effect. if the anegen bit is set to 1, speed and duplex advertise- ment is found in the auto-negotiation advertisement register (offset 0x04). 1 = enable auto-negotiation process 0 = disable auto-negotiation process
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 108 document classification: proprietary information january 3, 2008, preliminary 11 pwrdwn r/w 0x0 0x0 power down mode when the port is switched from power down to normal operation, software reset and restart auto-negotiation are performed even when bits reset (bit 15, above) and restart auto-negotiation (bit 9, below) are not set by the user. 1 = power down 0 = normal operation 10 isolate ro always 0 always 0 isolate mode will always be 0. the isolate function is not available, since full mii is not implemented. 0 = normal operation 9 restartaneg sc 0x0 self clear restart auto-negotiation auto-negotiation automatically restarts after hardware or software reset regardless of whether or not the restart bit is set. 1 = restart auto-negotiation process 0 = normal operation 8 duplex r/w aneg [2:0] update duplex mode selection speed, auto-negotiation enable, and duplex enable take on the values set by aneg[2:0] on hardware reset. a write to these registers has no effect unless any one of the following also occurs: software reset is asserted (b it 15), power down (bit 11), or transitions from power down to normal operation. 1 = full-duplex 0 = half-duplex 7 coltest ro always 0 always 0 collision test mode will always be 0. the collision test is not available, since full mii is not implemented. 0 = disable col signal test 6 speedmsb ro always 0 always 0 speed selection mode (msb) will always be 0. 0 = 100 mbps or 10 mbps 5:0 reserved ro always 0 always 0 will always be 0. table 61: phy control register (continued) offset: 0x00 (hex), or 0 (decimal) bits field mode hw rst sw rst description
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 109 table 62: phy status register offset: 0x01 (hex ), or 1 (decimal) bits field mode hw rst sw rst description 15 100t4 ro always 0 always 0 100base-t4 this protocol is not available. 0 = phy not able to perform 100base-t4 14 100fdx ro always 1 always 1 100base-t and 100base-x full-duplex 1 = phy able to perform full-duplex 13 100hdx ro always 1 always 1 100base-t and 100bas e-x half-duplex 1 = phy able to perform half-duplex 12 10fdx ro always 1 always 1 10base-t full-duplex 1 = phy able to perform full-duplex 11 10hpx ro always 1 always 1 10base-t half-duplex 1 = phy able to perform half-duplex 10 100t2fdx ro always 0 always 0 100base-t2 full-duplex. this protocol is not available. 0 = phy not able to perform full-duplex 9 100t2hdx ro always 0 always 0 100base-t2 half-duplex this protocol is not available. 0 = phy not able to perform half-duplex 8extdstatus ro always 0 always 0 extended status 0 = no extended status information in register 15 7reserved ro always 0 always 0 must always be 0. 6mfpresup ro always 1 always 1 mf preamble suppression mode must be always 1. 1 = phy accepts management frames with preamble suppressed 5 anegdone ro 0x0 0 auto-negotiation complete 1 = auto-negotiation process completed 0 = auto-negotiation process not completed 4 remotefault ro, lh 0x0 0 remote fault mode 1 = remote fault condition detected 0 = remote fault condition not detected 3 anegable ro always 1 always 1 auto-negotiation ability mode 1 = phy able to perform auto-negotiation
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 110 document classification: proprieta ry information january 3, 2008, preliminary 2link ro, ll 0x0 0 link status mode this register indicates when link was lost since the last read. for the current link status, either read this register back-to-back or read rtlink in table 72 on page 119 . 1 = link is up 0 = link is down 1 jabberdet ro, lh 0x0 0 jabber detect 1 = jabber condition detected 0 = jabber condition not detected 0 extdreg ro always 1 always 1 extended capability mode. 1 = extended register capabilities table 63: phy identifier offset: 0x02 (hex), or 2 (decimal) bits field mode hw rst sw rst description 15:0 oui msb ro 0x0141 0x0141 organizationally unique identifier bits 3:18 0000000101000001 marvell ? oui is 0x005043 table 64: phy identifier offset: 0x03 (hex), or 3 (decimal) bits field mode hw rst sw rst description 15:10 oui lsb ro always 000011 always 000011 organizationally unique identifier bits19:24 9:4 modelnum ro varies varies model number = 001000 3:0 revnum ro varies varies revision number contact marvell ? faes for information on the device revision number. table 62: phy status register (continued) offset: 0x01 (hex), or 1 (decimal) bits field mode hw rst sw rst description
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classifi cation: proprietary information page 111 table 65: auto-negotiation advertisement register offset: 0x04 (hex), or 4 (decimal) bits field mode hw rst sw rst description 15 anegad nxtpage r/w 0x0 retain next page 1 = advertise 0 = not advertised values programmed into the auto-negotiation adver- tisement register have no effect unless auto-negotia- tion is restarted (restartaneg - table 61 on page 107 ) or link goes down. 14 ack ro always 0 always 0 must be 0. 13 anegad refault r/w 0x0 retain remote fault mode 1 = set remote fault bit 0 = do not set remote fault bit values programmed into the auto-negotiation adver- tisement register have no effect unless auto-negotia- tion is restarted (restartaneg - table 61 on page 107 ) or link goes down. 12:11 reserved r/w 0x0 retain must be 00. reserved bits are r/w to allow for forward compatibility with future ieee standards. values programmed into the auto-negotiation adver- tisement register have no effect unless auto-negotia- tion is restarted (restartaneg - table 61 on page 107 ) or link goes down. 10 anegad pause r/w 0x0 retain pause mode 1 = mac pause implemented 0 = mac pause not implemented values programmed into the auto-negotiation adver- tisement register have no effect unless auto-negotia- tion is restarted (restartaneg - table 61 on page 107 ) or link goes down. 9 anegad 100t4 ro always 0 always 0 100base-t4 mode 0 = not capable of 100base-t4 8 anegad 100fdx r/w 0x1 retain 100base-tx full-duplex mode 1 = advertise 0 = not advertised values programmed into the auto-negotiation adver- tisement register have no effect unless auto-negotia- tion is restarted (restartaneg - table 61 on page 107 ) or link goes down.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 112 document classification: proprieta ry information january 3, 2008, preliminary 7 anegad 100hdx r/w 0x1 retain 100base-tx half-duplex mode 1 = advertise 0 = not advertised values programmed into the auto-negotiation adver- tisement register have no effect unless auto-negotia- tion is restarted (restartaneg - table 61 on page 107 ) or link goes down. 6 anegad 10fdx r/w 0x1 retain 10base-tx full-duplex mode 1 = advertise 0 = not advertised values programmed into the auto-negotiation adver- tisement register have no effect unless auto-negotia- tion is restarted (restartaneg - table 61 on page 107 ) or link goes down. 5 anegad 10hdx r/w 0x1 retain 10base-tx half-duplex mode 1 = advertise 0 = not advertised values programmed into the auto-negotiation adver- tisement register have no effect unless auto-negotia- tion is restarted (restartaneg - table 61 on page 107 ) or link goes down. 4:0 anegad selector ro always 0x01 always 0x01 selector field mode 00001 = 802.3 table 65: auto-negotiation advertisement register (continued) offset: 0x04 (hex), or 4 (decimal) bits field mode hw rst sw rst description
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classifi cation: proprietary information page 113 table 66: link partner ability register (base page) offset: 0x05 (hex ), or 5 (decimal) bits field mode hw rst sw rst description 15 lpnxt page ro 0x0 0 next page mode base page will be overwritten if next page is received and if reg8nxtpg ( table 71 on page 116 ) is disabled. when reg8nxtpg ( table 71 on page 116 ) is enabled, then next page is stored in the link partner next page register ( table 70 on page 115 ) , and the link partner ability register ( table 66 on page 113 ) holds the base page. received code word bit 15 14 lpack ro 0x0 0 acknowledge received code word bit 14 13 lpremote fault ro 0x0 0 remote fault received code word bit 13 12:5 lptechable ro 0x00 0x00 technology ability field received code word bit 12:5 4:0 lpselector ro 00000 00000 selector field received code word bit 4:0 table 67: link partner ability register (next page) offset: 0x05 (hex ), or 5 (decimal) bits field mode hw rst sw rst description 15 lpnxtpage ro -- -- next page mode base page will be overwritten if next page is received and if reg8nxtpg ( table 71 on page 116 ) is disabled. when reg8nxtpg ( table 71 on page 116 ) is enabled, then next page is stored in the link partner next page register ( table 70 on page 115 ) , and link partner abil- ity register ( table 66 on page 113 ) holds the base page. received code word bit 15 14 lpack ro -- -- acknowledge received code word bit 14 13 lpmessage ro -- -- message page received code word bit 13 12 lpack2 ro -- -- acknowledge 2 received code word bit 12 11 lptoggle ro -- -- toggle received code word bit 11
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 114 document classification: proprieta ry information january 3, 2008, preliminary 10:0 lpdata ro -- -- message/unformatted field received code word bit 10:0 table 68: auto-negotiation expansion register offset: 0x06 (hex), or 6 (decimal) bits field mode hw rst sw rst description 15:5 reserved ro always 0x000 always 0x000 reserved. must be 00000000000. the auto-negotiation expansion register is not valid until the anegdone ( table 62 on page 109 ) indicates completed. 4 parfaultdet ro 0x0 0x0 parallel detection level 1 = a fault has been detected via the parallel detection function 0 = a fault has not been detected via the parallel detec- tion function 3 lpnxtpg able ro 0x0 0x0 link partner next page able 1 = link partner is next page able 0 = link partner is not next page able 2 localnxtpg able ro always 0x1 always 0x1 local next page able this bit is equivalent to anegable ( table 62 on page 109 ). 1 = local device is next page able 1rxnewpage ro/lh 0x0 0x0 page received 1 = a new page has been received 0 = a new page has not been received 0 lpanegable ro 0x0 0x0 link partner auto-negotiation able 1 = link partner is auto-negotiation able 0 = link partner is not auto-negotiation able table 67: link partner ability register (next page) (continued) offset: 0x05 (hex), or 5 (decimal) bits field mode hw rst sw rst description
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classifi cation: proprietary information page 115 table 69: next page transmit register offset: 0x07 (hex ), or 7 (decimal) bits field mode hw rst sw rst description 15 txnxtpage r/w 0x0 0x0 next page a write to the next page transmit register implicitly sets a variable in the auto-negotiation state machine indicat- ing that the next page has been loaded. transmit code word bit 15 14 reserved ro 0x0 0x0 reserved transmit code word bit 14 13 txmessage r/w 0x1 0x1 message page mode transmit code word bit 13 12 txack2 r/w 0x0 0x0 acknowledge2 transmit code word bit 12 11 txtoggle ro 0x0 0x0 toggle transmit code word bit 11 10:0 txdata r/w 0x001 0x001 message/unformatted field transmit code word bit 10:0 table 70: link partner next page register offset: 0x08 (hex ), or 8 (decimal) bits field mode hw rst sw rst description 15 rxnxtpage ro 0x0 0x0 next page if reg8nxtpg ( table 71 on page 116 ) is enabled, then next page is stored in the link partner next page reg- ister ( table 70 on page 115 ) ; otherwise, the link part- ner next page register ( table 70 on page 115 ) is cleared to all 0?s. received code word bit 15 14 rxack ro 0x0 0x0 acknowledge received code word bit 14 13 rxmessage ro 0x0 0x0 message page received code word bit 13 12 rxack2 ro 0x0 0x0 acknowledge 2 received code word bit 12 11 rxtoggle ro 0x0 0x0 toggle received code word bit 11 10:0 rxdata ro 0x001 0x000 message/unformatted field received code word bit 10:0
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 116 document classification: proprieta ry information january 3, 2008, preliminary note registers 0x09 through 0x0f (hexadecimal (9 through 15 decimal) are reserved. do not read or write to these registers. table 71: phy specific control register offset: 0x10 (hex), or 16 (decimal) bits field mode hw rst sw rst description 15 reserved res -- -- reserved. 14 edet r/w ena_ edet retain energy detect 1 = enable with sense and pulse 0 = disable enable with sense only is not supported enable energy detect takes on the appropriate value defined by the config9 pin at hardware reset. 13 disnlp check r/w 0x0 0x0 disable normal linkpulse check linkpulse check and generation disable have no effect, if auto-negotiation is enabled locally. 1 = disable linkpulse check 0 = enable linkpulse check 12 reg8nxtpg r/w 0x0 0x0 enable the link partner next page register ( table 67 on page 113 ) to store next page. if set to store next page in the link partner next page register ( table 67 on page 113 ), then 802.3u is violated to emulate 802.3ab. 1 = store next page in the link partner next page regis- ter ( table 67 on page 113 ) 0 = store next page in the link partner ability register (base page) register ( table 66 on page 113 ). 11 disnlpgen r/w 0x0 0x0 disable linkpulse generation. linkpulse check and generation disable have no effect, when auto-negotiation is enabled locally. 1 = disable linkpulse generation 0 = enable linkpulse generation 10 forcelink r/w 0x0 0x0 force link good when link is forced to be good, the link state machine is bypassed and the link is always up. 1 = force link good 0 = normal operation 9disscrambler r/w aneg [2:0] retain disable scrambler if fiber mode is selected then the scrambler is disabled at hardware reset. 1 = disable scrambler 0 = enable scrambler
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classifi cation: proprietary information page 117 8disfefi r/w dis_ fefi aneg [2:0] retain disable fefi in 100base-fx mode, disable fefi takes on the appro- priate value defined by the config8 pin at hardware reset. fefi is automatically disabled regardless of the state of this bit if copper mode is selected. 1 = disable fefi 0 = enable fefi for the 88e3083 device, this bit applies to port 7 only. 7 extddistance r/w 0x0 0x0 enable extended distance when using cable exceeding 100 meters, the 10base-t receive threshold must be lowered in order to detect incoming signals. 1 = lower 10base-t receive threshold 0 = normal 10base-t receive threshold 6tpselect r/w aneg [2:0] update (un)shielded twisted pair this setting can be changed by writing to this bit fol- lowed by software reset. 1 = shielded twisted pair 0 = unshielded twisted pair - default 5:4 automdi[x] r/w ena_ xc,1 update mdi/mdix crossover this setting can be changed by writing to these bits fol- lowed by software reset. if ena_xc is 1 at hardware reset then bits 5:4 = 11 00 = transmit on pins rxp/rxn, receive on pins txp/ txn 01 = transmit on pins txp/txn, receive on pins rxp/ rxn 1x = enable automatic crossover 3:2 rxfifo depth r/w 0x0 0x0 receive fifo depth 00 = 4 bytes 01 = 6 bytes 10 = 8 bytes 11 = reserved 1 autopol r/w 0x0 00 polarity reversal if polarity is disabled, then the polarity is forced to be normal in 10base-t mode. polarity reversal has no effect in 100base-tx mode. 1 = disable automatic polarity reversal 0 = enable automatic polarity reversal table 71: phy specific control register (continued) offset: 0x10 (hex), or 16 (decimal) bits field mode hw rst sw rst description
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 118 document classification: proprieta ry information january 3, 2008, preliminary 0 disjabber r/w 0x0 00 disable jabber jabber has no effect in full-duplex or in 100base-x mode. 1 = disable jabber function 0 = enable jabber function table 71: phy specific control register (continued) offset: 0x10 (hex), or 16 (decimal) bits field mode hw rst sw rst description
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classifi cation: proprietary information page 119 table 72: phy specific status register offset: 0x11 (hex), or 17 (decimal) bits field mode hw rst sw rst description 15 reserved res -- -- reserved 14 resspeed ro aneg [2:0] retain resolved speed speed and duplex takes on the values set by aneg[2:0] pins on hardware reset only. the values are updated after the completion of auto-negotiation. the registers retain their values during software reset. 1 = 100 mbps 0 = 10 mbps 13 resduplex ro aneg [2:0] retain resolved duplex mode speed and duplex takes on the values set by aneg[2:0] pins on hardware reset only. the values are updated after the completion of auto-negotiation. the registers retain their values during software reset. this bit is valid only after the resolved bit 11 is set. 1 = full-duplex 0 = half-duplex 12 rcvpage ro, lh 0x0 0x0 page receive mode 1 = page received 0 = page not received 11 resolved ro 0x0 0x0 speed and duplex resolved. speed and duplex bits (14 and 13) are valid only after the resolved bit is set. the resolved bit is set when auto-negotiation is completed or auto-negotiation is disabled. 1 = resolved 0 = not resolved 10 rtlink ro 0x0 0x0 link (real time) 1 = link up 0 = link down 9:7 reserved res 0x0 0x0 must be 000. 6 mdi/mdix ro 0x1 0x1 mdi/mdix crossover status 1 = transmit on pins rxp/rxn, receive on pins txp/ txn 0 = transmit on pins txp/txn, receive on pins rxp/ rxn 5reserved res 0x4 0x4 must be 0. 4 sleep ro 0 0x0 energy detect status 1 = chip is in sleep mode (no wire activity) 0 = chip is not in sleep mode (active) 3:2 reserved res 0x0 0x0 must be 00.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 120 document classification: proprietary information january 3, 2008, preliminary 1 rtpolarity ro 0x0 0x0 polarity (real time) 1 = reversed 0 = normal 0 rtjabber ro retain 0x0 jabber (real time) 1 = jabber 0 = no jabber table 72: phy specific stat us register (continued) offset: 0x11 (hex), or 17 (decimal) bits field mode hw rst sw rst description
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 121 table 73: phy interrupt enable offset: 0x12 (hex), or 18 (decimal) bits field mode hw rst sw rst description 15 reserved res -- -- reserved 14 speedinten r/w 0x0 retain speed changed interrupt enable 1 = interrupt enable 0 = interrupt disable 13 duplexinten r/w 0x0 retain duplex changed interrupt enable 1 = interrupt enable 0 = interrupt disable 12 rcvpageinten r/w 0x0 retain page received interrupt enable 1 = interrupt enable 0 = interrupt disable 11 anegdone inten r/w 0x0 retain auto-negotiation completed interrupt enable 1 = interrupt enable 0 = interrupt disable 10 linkinten r/w 0x0 retain link status changed interrupt enable 1 = interrupt enable 0 = interrupt disable 9symerrinten r/w 0x0 retain symbol error interrupt enable 1 = interrupt enable 0 = interrupt disable 8flscrsinten r/w 0x0 retain false carrier interrupt enable 1 = interrupt enable 0 = interrupt disable 7 fifoerrint r/w 0x0 retain fifo over/underflow interrupt enable 1 = interrupt enable 0 = interrupt disable 6 mdi[x]inten r/w 0x0 0x0 mdi/mdix crossover changed interrupt enable 1 = interrupt enable 0 = interrupt disable 5reserved res 0x0 retain must be 0. 4edetinten r/w 0x0 retain energy detect interrupt enable 1 = enable 0 = disable 3:2 reserved res 0x0 retain must be 00. 1 polarityinten r/w 0x0 retain polarity changed interrupt enable 1 = interrupt enable 0 = interrupt disable 0 jabberinten r/w 0x0 retain jabber interrupt enable 1 = interrupt enable 0 = interrupt disable
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 122 document classification: proprietary information january 3, 2008, preliminary table 74: phy interrupt status offset: 0x13 (hex), or 19 (decimal) bits field mode hw rst sw rst description 15 reserved res -- -- reserved 14 speedint ro, lh 0x0 0x0 speed changed 1 = speed changed 0 = speed not changed 13 duplexint ro, lh 0x0 0x0 duplex changed 1 = duplex changed 0 = duplex not changed 12 rxpageint ro, lh 0x0 0x0 1 = page received 0 = page not received 11 anegdoneint ro, lh 0x0 0x0 auto-negotiation completed 1 = auto-negotiation completed 0 = auto-negotiation not completed 10 linkint ro, lh 0x0 0x0 link status changed 1 = link status changed 0 = link status not changed 9 symerrint ro, lh 0x0 0x0 symbol error 1 = symbol error 0 = no symbol error 8flscrsint ro, lh 0x0 0x0 false carrier 1 = false carrier 0 = no false carrier 7 fifoerrint ro, lh 0x0 0x0 fifo over /underflow error 1 = over/underflow error 0 = no over/underflow error 6 mdimdixint ro, lh 0x0 0x0 mdi/mdix crossover changed 1 = mdi/mdix crossover changed 0 = mdi/mdix crossover not changed 5reserved ro always 0 always 0 must be 0 4edetchg ro, lh 0x0 0x0 energy detect changed 1 = changed 0 = no change 3:2 reserved ro 0x0 0x0 must be 00 1 polarityint ro 0x0 0x0 polarity changed 1 = polarity changed 0 = polarity not changed 0 jabberint ro, lh 0x0 0x0 jabber mode 1 = jabber 0 = no jabber
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 123 table 75: phy interrupt port summary (global 1 ) offset: 0x14 (hex), or 20 (decimal) 1. this global register is accessible for writing or reading from any phy port (i.e., only one ph ysical register exists for all the phy ports). bits field mode hw rst sw rst description 15:5 reserved ro 0x0 0x0 must be 00000000000. 4 port4int active ro 0x0 0x0 port 4 interrupt active bit is set high, if any enabled interrupt is active for the port. bit is cleared only when all bits in register 19 are cleared. 1 = port has active interrupt 0 = port does not have active interrupt 3 port3int active ro 0x0 0x0 port 3 interrupt active bit is set high, if any enabled interrupt is active for the port. bit is cleared only when all bits in register 19 are cleared. 1 = port has active interrupt 0 = port does not have active interrupt 2 port2int active ro 0x0 0x0 port 2 interrupt active bit is set high, if any enabled interrupt is active for the port. bit is cleared only when all bits in register 19 are cleared. 1 = port has active interrupt 0 = port does not have active interrupt 1 port1int active ro 0x0 0x0 port 1 interrupt active bit is set high, if any enabled interrupt is active for the port. bit is cleared only when all bits in register 19 are cleared. 1 = port has active interrupt 0 = port does not have active interrupt 0 port0int active ro 0x0 0x0 port 0 interrupt active bit is set high, if any enabled interrupt is active for the port. bit is cleared only when all bits in register 19 are cleared. 1 = port has active interrupt 0 = port does not have active interrupt
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 124 document classification: proprietary information january 3, 2008, preliminary table 76: receive error counter offset: 0x15 (hex ), or 21 (decimal) bits field mode hw rst sw rst description 15:0 rxerrcnt ro 0x0000 0x0000 receive error count this register counts receive errors on the media inter- face. when the maximum receive error count reaches 0xffff, the counter will not roll over.
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 125 table 77: led parallel select register (global) 1 2 offset: 0x16 (hex), or 22 (decimal) 1. see table 34 on page 80 and table 35 on page 81 for parallel led display modes that can be selected by hardware pin configuration at reset. 2. this global register is accessible for writing or reading from any phy port (i.e., only one physical register exists for all the phy ports). bits field mode hw rst sw rst description 15:12 reserved res 0000 retain must be 0000 11:8 led2 r/w led[1:0] 00 = link 01 = link 10 = link/rx 11= link act retain led2 control the parallel led settings take on the appropriate value defined by the config_a pin at hardware reset. 0000 = colx, 0001 = error, 0010 = duplex, 0011 = duplex/colx, 0100 = speed, 0101 = link, 0110 = tx, 0111 = rx, 1000 = act, 1001 = link/rx, 1011 = act (blink mode), 1100 to 1111 = force to 1 7:4 led1 r/w led[1:0] 00 = rx 01 = act 10 = tx 11= duplex /colx retain led1 control the parallel led settings take on the appropriate value defined by the config_a pin at hardware reset. 0000 = colx, 0001 = error, 0010 = duplex, 0011 = duplex/colx, 0100 = speed, 0101 = link, 0110 = tx, 0111 = rx, 1000 = act, 1001 = link/rx, 1011 = act (blink mode), 1100 to 1111 = force to 1 3:0 led0 r/w led[1:0] 00 = tx 01 = speed 10 = speed 11= speed retain led0 control the parallel led settings take on the appropriate value defined b y the config_a pin at hardware reset. 0000 = colx, 0001 = error, 0010 = duplex, 0011 = duplex/colx, 0100 = speed, 0101 = link, 0110 = tx, 0111 = rx, 1000 = act, 1001 = link/rx, 1011 = act (blink mode), 1100 to 1111 = force to 1
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 126 document classification: proprietary information january 3, 2008, preliminary table 78: led stream select for serial leds offset: 0x17 (hex), or 23 (decimal) bits function mode hw rst sw rst description 15:14 ledlnkacty r/w led[1:0] retain led link activity the serial led settings take on the appropriate value defined by the config_a pin at hardware reset. 00 = off 01 = reserved 10 = dual 11 = single 13:12 ledrcvlnk r/w led[1:0] retain led receive link the serial led settings take on the appropriate value defined b y the config_a pin at hardware reset. 00 = off 01 = reserved 10 = dual 11 = single 11:10 ledacty r/w led[1:0] retain led activity the serial led settings take on the appropriate value defined b y the config_a pin at hardware reset. 00 = off 01 = reserved 10 = dual 11 = single 9:8 ledrcv r/w led[1:0] retain led receive the serial led settings take on the appropriate value defined b y the config_a pin at hardware reset. 00 = off 01 = reserved 10 = dual 11 = single 7:6 ledtx r/w led[1:0] retain transmit the serial led settings take on the appropriate value defined by the config_a pin at hardware reset. 00 = off 01 = reserved 10 = dual 11 = single 5:4 ledlnk r/w led[1:0] retain link the serial led settings take on the appropriate value defined by the config_a pin at hardware reset. 00 = off 01 = reserved 10 = dual 11 = single
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 127 3:2 ledspd r/w led[1:0] retain speed the serial led settings take on the appropriate value defined by the config_a pin at hardware reset. 00 = off 01 = reserved 10 = dual 11 = single 1:0 leddx/ colx r/w led[1:0] retain led duplex/ colx the serial led settings take on the appropriate value defined by the config_a pin at hardware res et. 00 = off 01 = reserved 10 = dual 11 = single table 78: led stream select for serial leds (continued) offset: 0x17 (hex), or 23 (decimal) bits function mode hw rst sw rst description
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 128 document classification: proprietary information january 3, 2008, preliminary table 79: phy led control register (global 1 ) offset: 0x18 (hex), 0r 24 (decimal) 1. this global register is accessible for writing or reading fr om any phy port (i.e., only one phys ical register exists for all the phy ports). bits field mode hw rst sw rst description 15 reserved ro always 0 always 0 must be 0. 14:12 pulsestretch r/w 0x4 retain pulse stretch duration. default value = 100. 000 = no pulse stretching 001 = 21 ms to 42 ms 010 = 42 ms to 84 ms 011 = 84 ms to 170 ms 100 = 170 ms to 340 ms 101 = 340 ms to 670 ms 110 = 670 ms to 1.3s 111 = 1.3s to 2.7s 11:9 blinkrate r/w 0x1 retain blink rate. this is a global setting. default value = 001 000 = 42 ms 001 = 84 ms 010 = 170 ms 011 = 340 ms 100 = 670 ms 101 to 111 = reserved 8:6 srstrupdate r/w 0x2 retain serial stream update 000 = 10 ms 001 = 21 ms 010 = 42 ms 011 = 84 ms 100 = 170 ms 101 = 340 ms 110 to 111 = reserved 5:4 duplex r/w led[1:0] 00 = off 01 = single 10 = single 11 = single retain 00 = off 01 = reserved 10 = dual 11 = single 3:2 error r/w 11 retain 00 = off 01 = reserved 10 = dual 11 = single 1:0 colx r/w led[1:0] 00 = off 01 = single 10 = single 11 = single retain 00 = off 01 = reserved 10 = dual 11 = single
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 129 table 80: phy manual led override offset: 0x19 (hex), or 25 (decimal) bits field mode hw rst sw rst description 15:6 reserved r/w 0x0 retain 0000000000 5:4 led2 r/w 0x0 retain 00 = normal 01 = blink 10 = led off 11 = led on 3:2 led1 r/w 0x0 retain 00 = normal 01 = blink 10 = led off 11 = led on 1:0 led0 r/w 0x0 retain 00 = normal 01 = blink 10 = led off 11 = led on
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 130 document classification: proprietary information january 3, 2008, preliminary table 81: vct? register for txp/n pins offset: 0x1a 1 (hex), or 26 (decimal) 1. the results stored in this register apply to the tx pin pair. bits field mode hw rst sw rst description 15 envct r/w, sc 0x0 0x0 enable vct the 88e6060 device must be in forced 100 mbps mode before enabling this bit. 1 = run vct 0 = vct completed after running vct once, bit 15 = 0 indicates vct com- pleted. the cable status is reported in the vcttst bits in regis- ters 26 and 27. refer to the ?virtual cable tester?? feature in section 4.5. 14:13 vcttst ro 0x0 retain vct test status these vct test status bits are valid after completion of vct. 11 = test fail 00 = valid test, normal cable (no short or open in cable) 10 = valid test, open in cable (impedance > 333 ohm) 01 = valid test, short in cable (impedance < 33 ohm) 12:8 amprfln ro 0x0 retain amplitude of reflection the amplitude of reflection is stored in these register bits. these amplitude bits range from 0x07 to 0x1f. 0x1f = maximum positive amplitude 0x13 = zero amplitude 0x07 = maximum negative amplitude these bits are valid after completion of vct (bit 15) and if the vct test status bits (bits 14:13) have not indicated test failure. 7:0 distrfln ro 0x0 retain distance of reflection these bits refer to the approximate distance (1m) to the open/short location, measured at nominal conditions (room temperature and typical vdds). see figure 26 . these bits are valid after completion of vct (bit 15) and if the vct test status bits (bit 14:13) have not indicated test failure.
phy registers copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 131 figure 26: cable fault distance trend line table 82: vct? register for rxp/n pins offset: 0x1b 1 (hex), or 27 (decimal) 1. the results stored in this register apply to the rx pin pair. bits field mode hw rst sw rst description 15 reserved ro always 0 always 0 reserved 14:13 vcttst ro 0 retain vct test status the vct test status bits are valid after completion of vct. 11 = test fail 00 = valid test, normal cable (no short or open in cable) 10 = valid test, open in cable (impedance > 333 ohm) 01 = valid test, short in cable (impedance < 33 ohm) 12:8 amprfln ro 0 retain amplitude of reflection the amplitude of reflection is stored in these register bits. these amplitude bits range from 0x07 to 0x1f. 0x1f = maximum positive amplitude 0x13 = zero amplitude 0x07 = maximum negative amplitude these bits are valid after completion of vct (bit 15) and if vct test status bits (bit 14:13) have not indicated test failure. 7:0 distrfln ro 0 retain distance of reflection these bits refer to the approximate distance (1m) to the open/short location, measured at nominal conditions (room temperature and typical vdds). see figure 26 . these bits are valid after completion of vct (bit 15) and if vct test status bits (bits 14:13) have not indicated test failure. register 27[7:0] or 26[7:0] rx/tx 200 150 100 50 0 0 50 100 150 150 200 250 300 y = 0.7816x - 18.261
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 132 document classification: proprietary information january 3, 2008, preliminary note registers 0x1d through 0x1f (hexadecimal (29 through 31 decimal) are reserved. do not read or write to these registers. table 83: phy specific control register ii offset: 0x1c (hex), or 28 (decimal) bits field mode hw rst sw rst description 15:1 reserved r/w 0x0 0x0 must be 000000000000000 0selclsa r/w sel_ class a update 1 = select class a driver ; available fo r 100base-tx mode only ?typically used in backplane or direct connect applications 0 = select class b driver?typically used in cat 5 applications
eeprom programming format eeprom programming details copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classifi cation: proprietary information page 133 section 8. eeprom programming format 8.1 eeprom programming details the 88e6060 device supports an optional external serial eeprom device for programing its internal registers. the eeprom data is read in once after resetn is dea sserted if the stand alone switch mode is not selected (see table 14 on page 29). the 88e6060 supports 1k bit (93c46), 2k bit (93c56), or 4k bit (93c66) eeprom devices. the size of the device is selected by the ee_din/ee_1k pi n at reset?see table 7 on page 21. th e external eeprom device must be configured in x16 data organization mode. regardless of which particular dev ice is attached, the eeprom is re ad and processed in the same way: 1. start at eeprom address 0x00. 2. read in the 16 bits of data from the even address. this step is called the command. 3. if the just read in command is all ones, terminate the serial eeprom reading process and go to step 8. 4. increment the address by 1?to the next address. 5. read in the 16 bits of data from th at address?this is called the regdata. 6. write regdata to the locations defined by the previous command. 7. go to step 2. 8. set the eeint bit in global status to a one (table 49, on page 99) generating an interrupt (if enabled). 9. done. the 16-bit command determines which register or registers inside the 88e6060 are updated as follows. refer to figure 27 below. ? bit 15 determines which set of registers can be written. if bit 15 = 0 the phy registers can be written (smi device addresses 0x00 to 0x04) or the switch global reg- isters can be written (smi device address 0x7). if bit 15 = 1 the switch registers can be written (smi device addresses 0x08 to 0x0d). see section 6 and figure 25 for more information on the registers and their addresses. ? bits 10:5 (or 9:5), the device vector, determines which de vice addresses are written. each bit of the device vector that is set to a one causes one device address to be written. bit 5 controls writes for port 0 (either phy address 0x00 or switch address 0x08). ? bit 6 controls writes for port 1 (eithe r phy address 0x01 or switch address 0x09). ? bit 7 controls writes for port 2, etc. ? the switch global registers (switch address 0x0f) are wr itten when bit 15 is a zero and bits 14:5 are ones. care is needed to ensure that reserved registers are not written. ? bits 4:0, the register address, determi ne which smi register address is written.
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 134 document classification: proprietary information january 3, 2008, preliminary the format of the eeprom commands sup ports writing the same regdata to all the phys or all the switch?s macs with one command/regdata pair. the command/regdata list can be as short or as long as needed. this makes optimum use of the limited number of comma nd/regdata pairs that can fit in a given size eeprom 1 . the maximum number of command/regdata pairs is one fewe r than might be expected from a simple calculation because the last entry must be the end of list or halt indicator of all ones. figure 27: eeprom data format 1. 31 in the 93c46, 63 in the 93c56, and 127 in the 93c66 eeprom. register address 15 14 5 4 0 addr 0 any even addr any odd addr register data written to switch register(s) device vector 1 register address register data written to phy register(s) device vector 0 register address register data written to a switch global register 0 last used even addr end of list indicator = all 1's at 1st even addr addr 1 addr 2 addr 3 switch command phy command global command command regdata regdata regdata 7 8 1 1 1 1 1 1 1 1 1 1 6 15 14 5 4 0 7 86 0 0 0 0 0 0 0 0 9 10 11 0
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 135 section 9. electrical specifications 9.1 absolute maximum ratings stresses above those listed in absolute maximum ratings may cause permanent device failure. functionality at or above these limits is not implied. exposure to absolute maximum ratings for extended periods may affect device reliability. table 84: absolute maximum ratings symbol parameter min typ max units v dd(3.3) power supply voltage on v ddo with respect to v ss -0.5 3.3 +3.6 v v dd(2.5) power supply voltage on v ddah with respect to v ss -0.5 2.5 +3.6 or v dd(3.3) +0.5 1 whichever is less 1. vdd(2.5) must never be more than 0.5v greater than vdd(3.3) or damage will result. this implies that power must be applied t o vdd(3.3) before or at the same time as vdd(2.5). v v dd(1.5) power supply voltage on v dd , or v ddal with respect to v ss -0.5 1.5 +3.6 or v dd(2.5) +0.5 2 whichever is less 2. vdd(1.5) must never be more than 0.5v greater than vdd(2.5) or damage will result. this implies that power must be applied t o vdd(2.5) before or at the same time as vdd(1.5). v v pin voltage applied to any input pin with respect to v ss -0.5 +3.6 or v ddo +0.5 3 whichever is less 3. vpin must never be more than 0.5v greater than vddo or damage will result. v t storage storage temperature -55 +125 4 4. 125 c is the re-bake temperature. for extended storage time greater than 24 hours, +85 c should be the maximum. c
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 136 document classification: proprietary information january 3, 2008, preliminary 9.2 recommended operating conditions table 85: recommended operating conditions symbol parameter condition min typ max units v dd(3.3) 3.3v power supply for pins v ddo 3.135 3.3 3.465 v v dd(2.5) 2.5v power supply for pins v ddah 2.375 2.5 2.625 v v dd(1.5) 1.5v power supply for pins v dd , v ddal 1.425 1.5 1.575 v t a ambient operating temperature commercial parts 0 70 c industrial parts 1 1. industrial part numbers have an ?i? following the commercial part numbers. see ?ordering information? on page 162. -40 85 c t j maximum junction temperature 125 2 2. refer to white paper on tj thermal calculations for more information. c rset internal bias reference resistor value placed between rset- and rset+ pins 1980 2000 2020
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 137 9.3 package thermal information 9.3.1 thermal conditions for 128-pin pqfp package table 86: thermal conditions for 128-pin pqfp package symbol parameter condition min typ max units ja thermal resistance 1 - junc- tion to ambient of the 88e6060 device 128-pin pqfp package ja = (t j - t a )/ p p = total power dissipation 1. refer to white paper on tj thermal calculations for more information. jedec 4 in. x 4.5 in. 4- layer pcb with no air flow 33.1 c/w jedec 4 in. x 4.5 in. 4- layer pcb with 1 meter/sec air flow 30.3 c/w jedec 4 in. x 4.5 in. 4- layer pcb with 2 meter/sec air flow 29.1 c/w jedec 4 in. x 4.5 in. 4- layer pcb with 3 meter/sec air flow 28.3 c/w jt thermal characteristic parameter 1 - junction to top center of the 88e6060 device 128-pin pqfp package jt = (t j - t c )/p. p = total power dissipation jedec 4 in. x 4.5 in. 4- layer pcb with no air flow 2.3 c/w jc thermal resistance 1 - junc- tion to case of the 88e6060 device 128-pin pqfp pack- age jc = (t j - t c )/ p to p p to p = power dissipation from the top of the package thermal resistance 1 - junc- tion to board of the 88e6060 device 128-pin pqfp pack- age jedec with no air flow 16.7 c/w jb jb = (t j - t b )/ p bottom p bottom = power dissipation from the bottom of the pack- age to the pcb surface. jedec with no air flow 25.7 c/w
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 138 document classification: proprietary information january 3, 2008, preliminary 9.4 dc electrical characteristics (over full range of values listed in the recommended operating conditions unless otherwise specified) table 87: dc electrical characteristics symbol parameter pins condition min typ 1 1. the values listed are typical values with led mode 3, 3 leds per port, and auto-negotiation on. max units i dd(3.3) 3.3 volt power to outputs v ddo no link on any port 15 ma all ports 10 mbps 17 ma all ports 100 mbps 22 ma power down mode 18 ma sleep mode 18 ma i dd(2.5) 2.5 volt 2 power to analog core 2. the 2.5v power source must supply an additional 100 ma (typ ical) current from the center tap on the magnetics. (this number is not included in this table.) vddah no link on any port 40 ma all ports 10 mbps and active 126 ma all ports 100 mbps 111 ma power down mode 32 ma sleep mode 36 ma 2.5 volt 3 power to center tap 3. the 2.5v power source must supply an additional 100 ma (typ ical) current from the center tap on the magnetics. (this number is not included in this table.) external magnetics center tap pin no link on any port 10 ma all ports 10 mbps and active 301 ma all ports 100 mbps 103 ma power down mode 0 ma sleep mode 0 ma i dd(1.5) 1.5 volt power to analog core v ddal no link on any port 0 ma all ports 10 mbps and active 1ma all ports 100 mbps 29 ma power down mode 1 ma sleep mode 0 ma 1.5 volt power to digital core v dd no link on any port 40 ma all ports 10 mbps and active 54 ma all ports 100 mbps and active 119 ma power down mode 39 ma sleep mode 36 ma
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 139 9.4.1 digital operating conditions (over full range of values listed in the recommended operating conditions unless otherwise specified) table 88: digital operating conditions symbol parameter pins con dition min typ max units v ih high level input voltage all pins 2.0 v ddo +0.5 v xtal_in 1.4 v v il low level input voltage all pins -0.5 0.8 v xtal_in 0.6 v v oh high level output voltage led pins 1 1. the led pins are as follows: p4_led2[2:0], p3 _led1[2:0], p2_led2[2:0], p1_led1[2:0], p0_led1[2:0] i oh = -8 ma v ddo = min 2.4 v xtal_out i oh = -1 ma v ih(xtal_in) +0.2 v all others i oh = -4 ma v ddo = min 2.4 v v ol low level output voltage led pins 1 i ol = 8 ma 0.4 v xtal_out i ol = 1 ma v il(xtal_in) -0.2 v intn pin 2 i ol = 8 ma v all others i ol = 4 ma 0.4 v i ilk input leakage current with pull-up resistor (typ. 120 k ) 0 link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 140 document classification: proprietary information january 3, 2008, preliminary table 89: internal resistor description pin # pin name resistor pin # pin name resistor 34, 33 sw_mode[1:0] internal pull-up 82 p5_inclk internal pull-up 36 ledser internal pull-up 83, 84, 85, 86 p5_ind[3:0] internal pull-up 37 ledclk internal pull-up 87 p5_indv internal pull-down 38 ledena internal pull-up 89 disable_mii4 internal pull-up 58 ee_dout internal pull-up 91 p4_crs internal pull-down 60 ee_clk/ addr4 internal pull-up 93 config_b internal pull-up 61 ee_cs/ ee_1k internal pull-up 94 p4_col internal pull-down 64 fd_flow_dis internal pull-up 95, 96, 97, 98 p4_outd[3:0]/ p4_mode[3:0] internal pull-up 66 enable_mii5 internal pull-up 99 p4_outclk internal pull-up 68 clk_sel internal pull-up 102 p4_inclk internal pull-up 69 ee_din/ hd_flow_dis internal pull-up 106 config_a internal pull-up 70 mdc internal pull-up 108 p1_config internal pull-down 71 mdio internal pull-up 110 resetn none 72 p5_crs internal pull-down 111 p0_config internal pull-down 73 p5_col internal pull-down 112, 114, 116, 117 p4_ind[3:0] internal pull-up 74, 75, 76, 77 p5_outd[3:0]/ p5_mode[3:0] internal pull-up 118 p4_indv internal pull-down 79 p5_outclk internal pull-up 122 p0_sdet none 81 p5_outdv/ wd_dis internal pull-up 127 p1_sdet none
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 141 9.4.2 ieee dc transceiver parameters ieee tests are typically based on templates and cannot simply be specified by a number. for an exact description of the template and the test conditions, refer to the ieee specifications: -10base-t ieee 802.3 clause 14 -100base-tx ansi x3.263-1995 (over full range of values listed in the recommended operating conditions unless otherwise specified) table 90: ieee dc tran sceiver parameters symbol parameter pins condition min typ max units v odiff absolute peak differential output voltage txp/n[1:0] 10base-t no cable 2.2 2.5 2.8 v txp/n[1:0] 10base-t cable model 585 1 1. ieee 802.3 clause 14, figure 14.9 shows the template for the ?far end? wave form. this template allows as little as 495 mv peak dif- ferential voltage at the far end receiver. mv txp/n[0] 100base-fx mode 0.4 0.8 1.2 v txp/n[1:0] 100base-tx mode 0.950 1.0 1.05 v overshoot 2 2. ansi x3.263-1995 figure 9-1. txp/n[4:0] 100base-tx mode 0 5% v amplitude symmetry (positive/ negative) txp/n[1:0] 100base-tx mode 0.98x 1.02x v+/v- v idiff peak differential input voltage accept level rxp/n[1:0] 10base-t mode 585 3 3. the input test is actually a template test. ieee 802.3 cla use 14, figure 14.17 shows the template for the receive wave form. mv rxp/n[0] p[1:0]_sdet p/n 100base-fx mode 200 mv signal detect assertion rxp/n[1:0] 100base-tx mode 1000 460 4 4. the ansi tp-pmd specification requires that any received signal with peak-to-peak differential amplitude greater than 1000 m v should turn on signal detect (internal signal in 100base-tx m ode). the 88e6060 will accept signals typically with 460 mv peak-t o- peak differential amplitude. mv peak- peak signal detect de-assertion rxp/n[1:0] 100base-tx mode 200 360 5 5. the ansi tp-pmd specification requires that any received si gnal with peak-to-peak differential amplitude less than 200 mv sh ould be de-assert signal detect (internal signal in 100base-tx mode ). the 88e6060 will reject signal s typically with peak-to-peak di ffer- ential amplitude less than 360 mv. mv peak- peak
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 142 document classification: proprietary information january 3, 2008, preliminary 9.5 ac electrical specifications 9.5.1 reset and configuration timing (over full range of values listed in the recommended operating conditions unless otherwise specified.) table 91: reset and configuration timing symbol parameter conditi on min typ max units notes t pu_reset valid power to resetn de-asserted or resetn assertion time at power up or subse- quent resets after power up 10 ms t su_clk number of valid xtal_in cycles prior to resetn de-asserted 10 cycles t su configuration data valid prior to resetn de- asserted 200 ns 1 1. when resetn is low all configuration pins become inputs, and the value seen on these pins is latched on the rising edge of resetn. t hd configuration data valid after resetn de-asserted 0 ns t co configuration output driven after resetn de- asserted 40 ns 2 2. p[x]_outd[3:0]/p[x]_mode[3:0] are normally outputs that are al so used to configure the 88e6060 device during hardware reset. when reset is asserted, these pins become inputs and the required led configuration is latched at the rising edge of resetn.
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 143 figure 28: reset and configuration timing power clk resetn config data t pu_reset t su_clk t su t hd config pins output t co
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 144 document classification: proprietary information january 3, 2008, preliminary 9.5.2 clock timing when using a 25 mhz oscillator (over full range of values listed in the recommended operating conditions unless otherwise specified.) figure 29: oscillator clock timing table 92: clock timing with a 25 mhz oscillator symbol parameter condition min typ max units notes t p xtal_in period 40 -50 ppm 40 40 +50 ppm ns 25 mhz t h xtal_in high time 16 ns t l xtal_in low time 16 ns t r xtal_in rise 3 ns t f xtal_in fall 3 ns t p t h t l t r 2.0 v 0.8v t f xtal_in
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 145 9.5.3 mii receive timing?phy mode in phy mode, the p[x]_inclk pins are outputs. (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 30: phy mode mii receive timing table 93: mii receive timing?phy mode symbol parameter conditi on min typ max units notes t p_tx_clk p[x]_inclk period 10base mode 400 ns 1 1. 2.5 mhz for 10 mbps or 25 mhz for 100 mbps. 100base mode 40 ns 1 t h_tx_clk p[x]_inclk high 10base mode 160 200 240 ns 100base mode 16 20 24 ns t l_tx_clk p[x]_inclk low 10base mode 160 200 240 ns 100base mode 16 20 24 ns t su_tx mii inputs (p[x]_ind[3:0], p[x]_indv) valid prior to p[x]_inclk going high. 15 ns t hd_tx mii inputs (p[x]_ind[3:0], p[x]_indv) valid after p[x]_inclk going high. 0ns inclk inputs t hd_tx t p_tx_clk t su_tx t h_tx_clk t l_tx_clk inclk is the clock used to clock the input data. it is an output in this mode. note:
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 146 document classification: proprietary information january 3, 2008, preliminary 9.5.4 mii transmit timing?phy mode in phy mode, the p[x]_outclk pins are outputs. (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 31: phy mode mii transmit timing table 94: mii transmit timing?phy mode symbol parameter condition min typ max units notes t p_rx_clk p[x]_outclk period 10base mode 400 ns 1 1. 2.5 mhz for 10 mbps or 25 mhz for 100 mbps. 100base mode 40 ns 1 t h_rx_clk p[x]_outclk high 10base mode 160 200 240 ns 100base mode 16 20 24 ns t l_rx_clk p[x]_outclk low 10base mode 160 200 240 ns 100base mode 16 20 24 ns t cq_max p[x]_outclk to outputs (p[x]_outd[3:0], p[x]_outdv) valid 25 ns t cq_min p[x]_outclk to outputs p[x]_outd[3:0], p[x]_outdv) invalid 10 ns outclk outputs t p_rx_clk t cq_max t cq_min t h_rx_clk t l_rx_clk outclk is the clock used to clock the output data. it is an output in this mode. note:
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classi fication: proprietary information page 147 9.5.5 mac mode clock timing in mac mode, inclk and outclk are inputs. (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 32: mac clock timing table 95: mac mode clock timing symbol parameter condition min typ max units notes t p macclk_in period 04 or 40 40 +50 ppm ns dc to 25 mhz t h macclk_in high time 16 ns t l macclk_in low time 16 ns t r macclk_in rise 3 ns t f macclk_in fall 3 ns t p t h t l t r 2.0 v 0.8v t f macclk
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 148 document classification: proprietary information january 3, 2008, preliminary 9.5.6 mii receive timing?mac mode in mac mode, the p[x]_inclk pins are inputs. (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 33: mac mode mii receive timing table 96: mii receive timing?mac mode symbol parameter condition min typ max units notes t su_rx mii inputs (p[x]_ind[3:0], p[x]_indv) valid prior to p[x]_inclk going high with 10 pf load 10 ns t hd_rx mii inputs (p[x]_ind[3:0], p[x]_indv) valid after p[x]_inclk going high with 10 pf load 10 ns inclk inputs t hd_rx t su_rx inclk is the clock used to clock the input data. it is an input in this mode. note:
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classification: proprietary information page 149 9.5.7 mii transmit timing?mac mode in mac mode, the p[x]_outclk pins are inputs. (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 34: mac mode mii transmit timing table 97: mii transmit timing?mac mode symbol parameter conditi on min typ max units notes t cq_max p[x]_outclk to outputs (p[x]_outd[3:0], p[x]_outdv) valid with 10 pf load 25 ns t cq_min p[x]_outclk to outputs (p[x]_outd[3:0], p[x]_outdv) invalid with 10 pf load 0ns outclk outputs t cq_max t cq_min outclk is the clock used to clock the output data. it is an input in this mode. note:
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 150 document classification: proprietary information january 3, 2008, preliminary 9.5.8 sni falling edge receive timing (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 35: sni falling edge receive timing table 98: sni falling edge receive timing symbol parameter condition min typ max units notes t p_frx_clk sni falling edge inclk period 10base-t mode 100 ns t h_frx_clk sni falling edge inclk high 35 50 65 ns t l_frx_clk sni falling edge inclk low 35 50 65 ns t su_frx sni receive data valid prior to inclk going low 20 ns t hd_frx sni receive data valid after inclk going low 10 ns inclk inputs t hd_frx t p_frx_clk t su_frx t h_frx_clk t l_frx_clk inclk is the clock used to clock the input data. it is an output in this mode. note:
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classification: proprietary information page 151 9.5.9 sni falling edge transmit timing (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 36: sni falling edge transmit timing table 99: sni falling edge transmit timing symbol parameter condition min typ max units notes t p_ftx_clk sni falling edge outclk period 10base-t mode 100 ns t h_ftx_clk sni falling edge outclk high 35 50 65 ns t l_ftx_clk sni falling edge outclk low 35 50 65 ns t cq_mxftx sni falling edge outclk to output valid 65 ns t cq_mnftx sni falling edge outclk to output invalid 35 ns outclk outputs t p_ftx_clk t cq_mnftx t l_ftx_clk t h_ftx_clk t cq_mxftx outclk is the clock used to clock the output data. it is an output in this mode. note:
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 152 document classification: proprietary information january 3, 2008, preliminary 9.5.10 sni rising edge receive timing (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 37: sni rising edge receive timing table 100: sni rising edge receive timing symbol parameter condition min typ max units notes t p_rrx_clk sni rising edge inclk period 10base-t mode 100 ns t h_rrx_clk sni rising edge inclk high 35 50 65 ns t l_rrx_clk sni rising edge inclk low 35 50 65 ns t su_rrx sni receive data valid prior to inclk going high 20 ns t hd_rrx sni receive data valid after inclk going high 10 ns inclk inputs t hd_rrx t p_rrx_clk t su_rrx t h_rrx_clk t l_rrx_clk inclk is the clock used to clock the input data. it is an output in this mode. note:
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classification: proprietary information page 153 9.5.11 sni rising edge transmit timing (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 38: sni rising edge transmit timing table 101: sni rising edge transmit timing symbol parameter condition min typ max units notes t p_rtx_clk sni rising edge outclk period 10base-t mode 100 ns t h_rtx_clk sni rising edge outclk high 35 50 65 ns t l_rtx_clk sni rising edge outclk low 35 50 65 ns t cq_mxrtx outclk to output valid 65 ns t cq_mnrtx outclk to output invalid 35 ns outclk outputs t p_rtx_clk t cq_mnrtx t l_rtx_clk t h_rtx_clk t cq_mxrtx outclk is the clock used to clock the output data. it is an output in this mode. note:
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 154 document classification: proprietary information january 3, 2008, preliminary 9.5.12 rmii receive timing using inclk (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 39: phy mode rmii receive timing using inclk table 102: rmii receive timing using inclk symbol parameter condition min typ max units notes t p_tx_clk p[x]_inclk period 100base mode 20 ns 1 1. 50 mhz for 100 mbps. t h_tx_clk p[x]_inclk high 100base mode 81012ns t l_tx_clk p[x]_inclk low 100base mode 81012ns t su_tx mii inputs (p[x]_ind[1:0], p[x]_indv) valid prior to p[x]_inclk going high. 9.5 ns t hd_tx mii inputs (p[x]_ind[1:0], p[x]_indv) valid after p[x]_inclk going high. 0 ns inclk inputs t hd_tx t p_tx_clk t su_tx t h_tx_clk t l_tx_clk inclk is the clock used to clock the input data. it is an output in this mode. note:
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classification: proprietary information page 155 9.5.13 rmii transmit timing using inclk (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 40: phy mode rmii transmit timing using inclk table 103: rmii transmit timing using inclk symbol parameter condition min typ max units notes t p_rx_clk p[x]_inclk period 100base mode 20 ns 1 1. 50 mhz for 100 mbps. t h_rx_clk p[x]_inclk high 100base mode 8 10 12 ns t l_rx_clk p[x]_inclk low 100base mode 8 10 12 ns t cq_max p[x]_inclk to outputs (p[x]_outd[1:0], p[x]_outdv) valid 8.0 ns t cq_min p[x]_inclk to outputs p[x]_outd[1:0], p[x]_outdv) invalid 0ns inclk outputs t p_rx_clk t cq_max t cq_min t h_rx_clk t l_rx_clk inclk is the clock used to clock the output data. it is an output in this mode. note:
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 156 document classification: proprietary information january 3, 2008, preliminary 9.5.14 serial led timing (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 41: serial led timing table 104: serial led timing symbol parameter condition min typ max units notes t p_ledclk ledclk period 80 ns t h_ledclk ledclk high 32 40 48 ns t l_ledclk ledclk low 32 40 48 ns t cq_max ledclk falling edge to link outputs (ledser, ledena) valid with 10 pf load 20 ns t cq_min ledclk falling edge to link outputs (ledser, ledena) invalid with 10 pf load 0ns ledclk outputs t p_ledclk t cq_min t l_ledclk t h_ledclk t cq_max
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classification: proprietary information page 157 9.5.15 serial management interface clock timing (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 42: serial management interface clock timing table 105: serial management interface clock timing symbol parameter condition min typ max units notes t p mdcclk_in period 120 ns 8.33 mhz t h mdcclk_in high time 48 ns t l mdcclk_in low time 48 ns t r mdcclk_in rise 6 ns t f mdcclk_in fall 6 ns t p t h t l t r 2.0 v 0.8v t f mdc
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 158 document classification: proprietary information january 3, 2008, preliminary 9.5.16 serial management interface timing (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 43: serial management interface timing table 106: serial management interface timing symbol parameter conditi on min typ max units notes t cq_mdio mdc to mdio (output) data valid time 20 ns t dly_mdio mdc to mdio (output) delay time 0 ns t su mdio (input) to mdc setup time 10 ns t hd mdio (input) to mdc hold time 10 ns mdc mdio (input) mdc mdio (output) t dly_mdio t hd t su valid data t cq_mdio valid data
electrical specifications absolute maximum ratings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classification: proprietary information page 159 9.5.17 eeprom timing (over full range of values listed in the recommended operating conditions unless otherwise specified) figure 44: eeprom timing table 107: eeprom timing symbol parameter condition min typ max units notes t p ee_clk period 5120 ns t h ee_clk high time 2560 ns t l ee_clk low time 2560 ns t cqcsmx serial eeprom chip select valid referenced to ee_clk 5ns t cqcsmn serial eeprom chip select invalid 5ns t cqdmx serial eeprom data transmitted to eeprom valid 10 ns t cqdmn serial eeprom data transmitted to eeprom invalid 3ns t s setup time for data received from eeprom 10 ns t h hold time for data received from eeprom 10 ns ee_din (hd_flow_dis) ee_dout ee_cs ee_clk t cqcsmx t cqcsmn t cqdmx t cqdmn t s t h t p t l t h
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 160 document classification: proprietary information january 3, 2008, preliminary 9.5.18 ieee ac parameters ieee tests are typically based on templates and cannot simp ly be specified by number. for an exact description of the templates and the test conditions, refer to the ieee specifications: ? 10base-t ieee 802.3 clause 14-2000 ? 100base-tx ansi x3.263-1995 ? 1000base-t ieee 802.3ab clause 40 section 40.6.1.2 figure 40-26 shows the template waveforms for transmitter electrical specifications. (over full range of values listed in the recommended operating conditions unless otherwise specified) table 108: ieee ac parameters symbol parameter pins condition min typ max units t rise rise time txp/n[4:0] 100base-tx 3.0 4.0 5.0 ns t fall fall time txp/n[4:0] 100base-tx 3.0 4.0 5.0 ns t rise / t fall symmetry txp/n[4:0] 100base-tx 0 0.5 ns dcd duty cycle distortion txp/n[4:0] 100base-tx 0 0.5 1 1. ansi x3.263-1995 figure 9-3. ns, peak- peak transmit jitter txp/n[4:0] 100base-tx 0 1.4 ns, peak- peak
mechanical drawings 128-pin pqfp package copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classification: proprietary information page 161 section 10. mechanical drawings 10.1 128-pin pqfp package 0.5 basic 0.25 min 1.6 nominal 38 39 1 128 64 65 102 103 pin1 indicator 23.20 0.10 17.20 0.15 3.40 max (all dimensions in mm)
link street ? 88e6060 - unrestricted low power, 6-port, 10/100 ethernet switch doc. no. mv-s100952-u0, rev. -- copyright ? 2008 marvell page 162 document classification: proprietary information january 3, 2008, preliminary section 11. ordering information 11.1 ordering part numbers figure 45 shows the ordering part numbering scheme for the devices. contact marvell ? faes or sales representa- tives for complete or dering information. figure 45: sample part number table 109: part order option - commercial package type part order number 88e6060 128-pin pqfp - commercial - rohs 6/6 88e6060-xx-rcj1c000 table 110: part order option - industrial package type part order number 88e6060 128-pin pqfp - industrial - rohs 6/6 88E6060-XX-RCJ1I000 ? xx ? xxx ? c000 - t123 part number 88e6060 package code rcj = 128-pin pqfp environmental 1 = rohs 6/6 temperature range c = commercial i = industrial custom code custom (optional) 88e6060 custom code
ordering information package markings copyright ? 2008 marvell doc. no. mv-s100952-u0, rev. -- january 3, 2008, preliminary document classification: proprietary information page 163 11.2 package markings figure 46 is an example of the package marking and pin 1 location for the 88e6060 device 128-pin pqfp com- mercial rohs 6/6 package. figure 46: 88e6060 128-pin pqfp commercial rohs 6/6 packag e marking and pin 1 location figure 47 is an example of the package marking and pin 1 location for the 88e6060 device 128-pin pqfp indus- trial rohs 6/6 package. figure 47: 88e6060 128-pin pqfp industrial ro hs 6/6 package marking and pin 1 location 88e6060-rcj1 lot number yyww xx@ country logo part number, package co de, environmental code environmental code - 1 = rohs 6/6 date code, die revision, assembly plant code yyww = date code xx = die revision @ = assembly location code country of origin (contained in the mold id or marked as the last line on the package.) pin 1 location n ote : the above example is not drawn to scale. location of markings is approximate. 88e6060-rcj1 lot number yyww xx@ country logo part number, package co de, environmental code environmental code - 1 = rohs 6/6 date code, die revision, assembly plant code yyww = date code xx = die revision @ = assembly location code industrial grade package marking country of origin (contained in the mold id or marked as the last line on the package.) pin 1 location n ote : the above example is not drawn to scale. location of markings is approximate. i
marvell. moving forward faster marvell semiconductor, inc. 5488 marvell lane santa clara, ca 95054, usa tel: 1.408.222.2500 fax: 1.408.752.9028 www.marvell.com back cover

▲Up To Search▲

Price & Availability of 88E6060-XX-RCJ1I000

	To Download 88E6060-XX-RCJ1I000 Datasheet File
If you can't view the Datasheet, Please click here to try to view without PDF Reader .