2017-2019 Microchip Technology Inc. DS00002330D-page 1
Highlights
• Non-blocking wire-speed Ethernet switching fabric
• Full-featured forwarding and filtering control, includ-
ing Access Control List (ACL) filtering
• Full VLAN and QoS support
• Five ports with integrated 10/100/1000BASE-T PHY
transceivers
• Two ports with 10/100/1000 Ethernet MACs and con-
figurable RGMII/MII/RMII interfaces
• IEEE 802.1X access control support
• EtherGreen™ power management features,
including low power standby
• Flexible management interface options: SPI, I2C,
MIIM, and in-band management via any port
• Commercial/Industrial temperature range support
• 128-pin TQFP-EP (14 x 14mm) RoHS compliant pkg
Target Applications
• Stand-alone 10/100/1000Mbps Ethernet switches
• VoIP infrastructure switches
• Broadband gateways/firewalls
• Wi-Fi access points
• Integrated DSL/cable modems
• Security/surveillance systems
• Industrial control/automation switches
• Networked measurement and control systems
Features
• Switch Management Capabilities
- 10/100/1000Mbps Ethernet switch basic functions:
frame buffer management, address look-up table,
queue management, MIB counters
- Non-blocking store-and-forward switch fabric assures
fast packet delivery by utilizing 4096 entry forwarding
table with 256kByte frame buffer
- Jumbo packet support up to 9000 bytes
- Port mirroring/monitoring/sniffing:
ingress and/or egress traffic to any port
- MIB counters for fully-compliant statistics gathering
34 counters per port
- Tail tagging mode (one byte added before FCS) sup-
port at host port to inform the processor which ingress
port receives the packet and its priority
- Loopback modes for remote failure diagnostics
- Rapid spanning tree protocol (RSTP) support for topol-
ogy management and ring/linear recovery
- Multiple spanning tree protocol (MSTP) support
• Five Integrated PHY Ports
- 1000BASE-T/100BASE-TX/10BASE-Te IEEE 802.3
- Fast Link-up option significantly reduces link-up time
- Auto-negotiation and Auto-MDI/MDI-X support
- On-chip termination resistors and internal biasing for
differential pairs to reduce power
- LinkMD® cable diagnostic capabilities for determining
cable opens, shorts, and length
• Two Configurable External MAC Ports
- Reduced Gigabit Media Independent Interface
(RGMII) v2.0
- Reduced Media Independent Interface (RMII) v1.2
with 50MHz reference clock input/output option
- Media Independent Interface (MII) in PHY/MAC mode
• Advanced Switch Capabilities
- IEEE 802.1Q VLAN support for 128 active VLAN
groups and the full range of 4096 VLAN IDs
- IEEE 802.1p/Q tag insertion/removal on per port basis
- VLAN ID on per port or VLAN basis
- IEEE 802.3x full-duplex flow control and half-duplex
back pressure collision control
- IEEE 802.1X access control
(Port-based and MAC address based)
- IGMP v1/v2/v3 snooping for multicast packet filtering
- IPv6 multicast listener discovery (MLD) snooping
- IPv4/IPv6 QoS support, QoS/CoS packet prioritization
- 802.1p QoS packet classification with 4 priority queues
- Programmable rate limiting at ingress/egress ports
- Broadcast storm protection
- Four priority queues with dynamic packet mapping for
IEEE 802.1p, IPv4 DIFFSERV, IPv6 Traffic Class
- MAC filtering function to filter or forward unknown uni-
cast, multicast and VLAN packets
- Self-address filtering for implementing ring topologies
• Comprehensive Configuration Registers Access
- High-speed 4-wire SPI (up to 50MHz), I2C interfaces
provide access to all internal registers
- MII Management (MIIM, MDC/MDIO 2-wire) Interface
provides access to all PHY registers
- In-band management via any of the data ports
- I/O pin strapping facility to set certain register bits from
I/O pins at reset time
- On-the-fly configurable control registers
• Power Management
- Energy detect power-down mode on cable disconnect
- Dynamic clock tree control
- Unused ports can be individually powered down
- Full-chip software power-down
- Wake-on-LAN (WoL) standby power mode with PME
interrupt output for system wake upon triggered events
KSZ9897R
7-Port Gigabit Ethernet Switch
with Two RGMII/MII/RMII Interfaces
KSZ9897R
DS00002330D-page 2 2017-2019 Microchip Technology Inc.
TO OUR VALUED CUSTOMERS
It is our intention to provide our valued customers with the best documentation possible to ensure successful use of your Microchip
products. To this end, we will continue to improve our publications to better suit your needs. Our publications will be refined and
enhanced as new volumes and updates are introduced.
If you have any questions or comments regarding this publication, please contact the Marketing Communications Department via
E-mail at docerrors@microchip.com. We welcome your feedback.
Most Current Documentation
To obtain the most up-to-date version of this documentation, please register at our Worldwide Web site at:
http://www.microchip.com
You can determine the version of a data sheet by examining its literature number found on the bottom outside corner of any page.
The last character of the literature number is the version number, (e.g., DS30000000A is version A of document DS30000000).
Errata
An errata sheet, describing minor operational differences from the data sheet and recommended workarounds, may exist for cur-
rent devices. As device/documentation issues become known to us, we will publish an errata sheet. The errata will specify the
revision of silicon and revision of document to which it applies.
To determine if an errata sheet exists for a particular device, please check with one of the following:
• Microchip’s Worldwide Web site; http://www.microchip.com
• Your local Microchip sales office (see last page)
When contacting a sales office, please specify which device, revision of silicon and data sheet (include -literature number) you are
using.
Customer Notification System
Register on our web site at www.microchip.com to receive the most current information on all of our products.
2017-2019 Microchip Technology Inc. DS00002330D-page 3
KSZ9897R
Table of Contents
1.0 Preface ............................................................................................................................................................................................ 4
2.0 Introduction ..................................................................................................................................................................................... 8
3.0 Pin Descriptions and Configuration ................................................................................................................................................. 9
4.0 Functional Description .................................................................................................................................................................. 18
5.0 Device Registers ........................................................................................................................................................................... 63
6.0 Operational Characteristics ......................................................................................................................................................... 168
7.0 Design Guidelines ....................................................................................................................................................................... 183
8.0 Package Information ................................................................................................................................................................... 186
Appendix A: Data Sheet Revision History ......................................................................................................................................... 190
The Microchip Web Site .................................................................................................................................................................... 193
Customer Change Notification Service ............................................................................................................................................. 193
Customer Support ............................................................................................................................................................................. 193
Product Identification System ........................................................................................................................................................... 194
KSZ9897R
DS00002330D-page 4 2017-2019 Microchip Technology Inc.
1.0 PREFACE
1.1 Glossary of Terms
TABLE 1-1: GENERAL TERMS
Term Description
10BASE-Te 10 Mbps Ethernet, 2.5V signaling, IEEE 802.3 compliant
100BASE-TX 100 Mbps Fast Ethernet, IEEE 802.3u compliant
1000BASE-T 1000 Mbps Gigabit Ethernet, IEEE 802.3ab compliant
ADC Analog-to-Digital Converter
AN Auto-Negotiation
BLW Baseline Wander
BPDU Bridge Protocol Data Unit. Messages which carry the Spanning Tree Protocol informa-
tion.
Byte 8 bits
CRC Cyclic Redundancy Check. A common technique for detection data transmission
errors. CRC for Ethernet is 32 bits long.
CSR Control and Status Registers
DA Destination Address
DWORD 32 bits
FCS Frame Check Sequence. The extra checksum characters added to the end of an
Ethernet frame, used for error detection and correction.
FID Frame or Filter ID. Specifies the frame identifier. Alternately is the filter identifier.
FIFO First In First Out buffer
FSM Finite State Machine
GPIO General Purpose I/O
Host External system (Includes processor, application software, etc.)
IGMP Internet Group Management Protocol. Defined by RFC 1112, RFC 2236, and RFC
4604 to establish multicast group membership in IPv4 networks.
IPG Inter-Packet Gap. A time delay between successive data packets mandated by the
network standard for protocol reasons.
Jumbo Packet A packet larger than the standard Ethernet packet (1518 bytes). Large packet sizes
allow for more efficient use of bandwidth, lower overhead, less processing, etc..
lsb Least Significant Bit
LSB Least Significant Byte
MAC Media Access Controller. A functional block responsible for implementing the media
access control layer, which is a sublayer of the data link layer.
MDI Medium Dependent Interface. An Ethernet port connection that allows network hubs or
switches to connect to other hubs or switches without a null-modem, or crossover,
cable.
MDIX Media Independent Interface with Crossover. An Ethernet port connection that allows
networked end stations (i.e., PCs or workstations) to connect to each other using a
null-modem, or crossover, cable.
MIB Management Information Base. The MIB comprises the management portion of net-
work devices. This can include monitoring traffic levels and faults (statistical), and can
also change operating parameters in network nodes (static forwarding addresses).
MII Media Independent Interface. The MII accesses PHY registers as defined in the IEEE
802.3 specification.
MIIM Media Independent Interface Management
2017-2019 Microchip Technology Inc. DS00002330D-page 5
KSZ9897R
MLD Multicast Listening Discovery. This protocol is defined by RFC 3810 and RFC 4604 to
establish multicast group membership in IPv6 networks.
MLT-3 Multi-Level Transmission Encoding (3-Levels). A tri-level encoding method where a
change in the logic level represents a code bit “1” and the logic output remaining at the
same level represents a code bit “0”.
msb Most Significant Bit
MSB Most Significant Byte
NRZ Non Return to Zero. A type of signal data encoding whereby the signal does not return
to a zero state in between bits.
NRZI Non Return to Zero Inverted. This encoding method inverts the signal for a “1” and
leaves the signal unchanged for a “0”
N/A Not Applicable
NC No Connect
OUI Organizationally Unique Identifier
PHY A device or function block which performs the physical layer interface function in a net-
work.
PLL Phase Locked Loop. A electronic circuit that controls an oscillator so that it maintains a
constant phase angle (i.e., lock) on the frequency of an input, or reference, signal.
RESERVED Refers to a reserved bit field or address. Unless otherwise noted, reserved bits must
always be zero for write operations. Unless otherwise noted, values are not guaran-
teed when reading reserved bits. Unless otherwise noted, do not read or write to
reserved addresses.
RTC Real-Time Clock
SA Source Address
SFD Start of Frame Delimiter. The 8-bit value indicating the end of the preamble of an
Ethernet frame.
SQE Signal Quality Error (also known as “heartbeat”)
SSD Start of Stream Delimiter
TCP Transmission Control Protocol
UDP User Datagram Protocol - A connectionless protocol run on top of IP networks
UTP Unshielded Twisted Pair. Commonly a cable containing 4 twisted pairs of wire.
UUID Universally Unique IDentifier
VLAN Virtual Local Area Network
WORD 16 bits
TABLE 1-1: GENERAL TERMS (CONTINUED)
Term Description
KSZ9897R
DS00002330D-page 6 2017-2019 Microchip Technology Inc.
1.2 Buffer Types
TABLE 1-2: BUFFER TYPES
Buffer Type Description
I Input
IPU Input with internal pull-up (58 k ±30%)
IPU/O Input with internal pull-up (58 k ±30%) during power-up/reset;
output pin during normal operation
IPD Input with internal pull-down (58 k ±30%)
IPD/O Input with internal pull-down (58 k ±30%) during power-up/reset;
output pin during normal operation
O8 Output with 8 mA sink and 8 mA source
O24 Output with 24 mA sink and 24 mA source
OPU Output (8mA) with internal pull-up (58 k ±30%)
OPD Output (8mA) with internal pull-down (58 k ±30%)
A Analog
AIO Analog bidirectional
ICLK Crystal oscillator input pin
OCLK Crystal oscillator output pin
P Power
GND Ground
Note: Refer to Section 6.3, "Electrical Characteristics," on page 169 for the electrical characteristics of the vari-
ous buffers.
2017-2019 Microchip Technology Inc. DS00002330D-page 7
KSZ9897R
1.3 Register Nomenclature
1.4 References
•NXP I
2C-Bus Specification (UM10204, April 4, 2014): www.nxp.com/documents/user_manual/UM10204.pdf
TABLE 1-3: REGISTER NOMENCLATURE
Register Bit Type Notation Register Bit Description
RRead: A register or bit with this attribute can be read.
WWrite: A register or bit with this attribute can be written.
RO Read only: Read only. Writes have no effect.
RC Read to Clear: Contents is cleared after the read. Writes have no effect.
WO Write only: If a register or bit is write-only, reads will return unspecified data.
WC Write One to Clear: Writing a one clears the value. Writing a zero has no effect.
LL Latch Low: Applies to certain RO status bits. If a status condition causes this bit to go
low, it will maintain the low state until read, even if the status condition changes. A read
clears the latch, allowing the bit to go high if dictated by the status condition.
LH Latch High: Applies to certain RO status bits. If a status condition causes this bit to go
high, it will maintain the high state until read, even if the status condition changes. A
read clears the latch, allowing the bit to go low if dictated by the status condition.
SC Self-Clearing: Contents are self-cleared after the being set. Writes of zero have no
effect. Contents can be read.
RESERVED Reserved Field: Reserved fields must be written with zeros, unless otherwise indi-
cated, to ensure future compatibility. The value of reserved bits is not guaranteed on a
read.
KSZ9897R
DS00002330D-page 8 2017-2019 Microchip Technology Inc.
2.0 INTRODUCTION
2.1 General Description
The KSZ9897R is a highly-integrated, IEEE 802.3 compliant networking device that incorporates a layer-2 managed
Gigabit Ethernet switch, five 10BASE-Te/100BASE-TX/1000BASE-T physical layer transceivers (PHYs) and associated
MAC units, and two MAC ports with individually configurable RGMII/MII/RMII interfaces for direct connection to a host
processor/controller, another Ethernet switch, or an Ethernet PHY transceiver.
The KSZ9897R is built upon industry-leading Ethernet technology, with features designed to offload host processing
and streamline the overall design:
• Non-blocking wire-speed Ethernet switch fabric supports 1 Gbps on RGMII
• Full-featured forwarding and filtering control, including port-based Access Control List (ACL) filtering
• Full VLAN and QoS support
• Traffic prioritization with per-port ingress/egress queues and by traffic classification
• Spanning Tree support
• IEEE 802.1X access control support
A host processor can access all KSZ9897R registers for control over all PHY, MAC, and switch functions. Full register
access is available via the integrated SPI or I2C interfaces, and by in-band management via any one of the data ports.
PHY register access is provided by a MIIM interface. Flexible digital I/O voltage allows the MAC port to interface directly
with a 1.8/2.5/3.3V host processor/controller/FPGA.
Additionally, a robust assortment of power-management features including Wake-on-LAN (WoL) for low power standby
operation, have been designed to satisfy energy-efficient system requirements.
The KSZ9897R is available in commercial (0°C to +70°C) and industrial (-40°C to +85°C) temperature ranges. An inter-
nal block diagram of the KSZ9897R is shown in Figure 2-1.
FIGURE 2-1: INTERNAL BLOCK DIAGRAM
KSZ9897R
Port110/100/1000
PHY1
10/100/1000
PHY2
10/100/1000
PHY3
10/100/1000
PHY4
10/100/1000
PHY5
Port2
Port3
Port4
Port5
GMAC1
GMAC2
GMAC3
GMAC4
GMAC5
SwitchEngine
QueueManagement,QOS,Etc.
Control
Registers
GMAC6
GMAC7
RGMII/MII/RMII
RGMII/MII/RMII
Address
Lookup
MIB
Counters
Frame
Buffers
Queue
Mgmt.
SPI/I2C/MIIM
2017-2019 Microchip Technology Inc. DS00002330D-page 9
KSZ9897R
3.0 PIN DESCRIPTIONS AND CONFIGURATION
3.1 Pin Assignments
The device pin diagram for the KSZ9897R can be seen in Figure 3-1. Table 3-1 provides a KSZ9897R pin assignment
table. Pin descriptions are provided in Section 3.2, "Pin Descriptions".
FIGURE 3-1: PIN ASSIGNMENTS (TOP VIEW)
Note: When an “_N” is used at the end of the signal name, it indicates that the signal is active low. For example,
RESET_N indicates that the reset signal is active low.
The buffer type for each signal is indicated in the “Buffer Type” column of the pin description tables in Sec-
tion 3.2, "Pin Descriptions". A description of the buffer types is provided in Section 1.2, "Buffer Types".
KSZ9897R
128-TQFP-EP
(Top View)
TXRX1P_A
TXRX1M_A
TXRX1P_B
TXRX1M_B
TXRX1P_C
TXRX1M_C
TXRX1P_D
TXRX1M_D
TXRX2P_A
TXRX2M_A
AVDDL
TXRX2P_B
TXRX2M_B
TXRX2P_C
TXRX2M_C
TXRX2P_D
TXRX2M_D
TXRX3P_A
TXRX3M_A
TXRX3P_B
TXRX3M_B
TXRX3P_C
TXRX3M_C
TXRX3P_D
TXRX3M_D
RESET_N
CLKO_25_125
INTRP_N
PME_N
LED2_1
LED2_0
NC
LED3_1
LED3_0
LED4_1
LED4_0
RXD7_0
RXD7_1
RXD7_2
RXD7_3
CRS7
RX_ER7
RX_DV7/CRS_DV7/RX_CTL7
RX_CLK7/REFCLKO7
TXD7_0
TXD7_1
TXD7_2
TXD7_3
COL7
TX_ER7
TX_EN7/TX_CTL7
TX_CLK7/REFCLKI7
RXD6_0
ISET
XI
XO
GND
TXRX5M_D
TXRX5P_D
AVDDL
TXRX5M_C
TXRX5P_C
TXRX5M_B
TXRX5P_B
TXRX5M_A
TXRX5P_A
NC
LED1_1
LED1_0
LED5_1
LED5_0
SCL/MDC
SCS_N
SDI/SDA/MDIO
SDO
TXRX4P_A
TXRX4M_A
TXRX4P_B
TXRX4M_B
TXRX4P_C
TXRX4M_C
TXRX4P_D
TXRX4M_D
TX_CLK6/REFCLKI6
TX_EN6/TX_CTL6
TX_ER6
COL6
TXD6_3
TXD6_2
TXD6_1
TXD6_0
RX_CLK6/REFCLKO6
RX_DV6/CRS_DV6/RX_CTL6
RX_ER6
CRS6
RXD6_3
RXD6_2
RXD6_1
33
34
35
36
37
38
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
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
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
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
AVDDL
AVDDL
AVDDL
AVDDH
AVDDH
AVDDL
AVDDL
AVDDH
AVDDH
AVDDH
AVDDL
AVDDH
AVDDH
AVDDL
DVDDL
DVDDL
DVDDL
DVDDL
DVDDL
DVDDL
GND
GND
GND
GND
GND
VDDIO
DVDDL
DVDDL
VDDIO
VDDIO
(Connect exposed pad to ground with a via field)
GND
KSZ9897R
DS00002330D-page 10 2017-2019 Microchip Technology Inc.
Note 3-1 This pin also provides configuration strap functions during hardware/software resets. Refer to Section
3.2.1, "Configuration Straps" for additional information.
TABLE 3-1: PIN ASSIGNMENTS
Pin Pin Name Pin Pin Name Pin Pin Name Pin Pin Name
1TXRX1P_A 33 AVDDH 65 RXD6_0 (Note 3-1)97 SDO
2TXRX1M_A 34 TXRX4P_A 66 TX_CLK7/REFCLKI7 98 SDI/SDA/MDIO
3AVDDL 35 TXRX4M_A 67 TX_EN7/TX_CTL7 99 VDDIO
4TXRX1P_B 36 AVDDL 68 TX_ER7 100 SCS_N
5TXRX1M_B 37 TXRX4P_B 69 COL7 101 SCL/MDC
6TXRX1P_C 38 TXRX4M_B 70 TXD7_3 102 LED5_0
7TXRX1M_C 39 TXRX4P_C 71 TXD7_2 103 LED5_1 (Note 3-1)
8TXRX1P_D 40 TXRX4M_C 72 TXD7_1 104 DVDDL
9TXRX1M_D 41 AVDDL 73 TXD7_0 105 LED1_0
10 AVDDH 42 TXRX4P_D 74 DVDDL 106 LED1_1 (Note 3-1)
11 DVDDL 43 TXRX4M_D 75 RX_CLK7/REFCLKO7 107 GND
12 TXRX2P_A 44 AVDDH 76 RX_DV7/CRS_DV7/
RX_CTL7 (Note 3-1)
108 NC
13 TXRX2M_A 45 DVDDL 77 VDDIO 109 GND
14 AVDDL 46 GND 78 RX_ER7 110 DVDDL
15 TXRX2P_B 47 GND 79 CRS7 111 AVDDH
16 TXRX2M_B 48 TX_CLK6/REFCLKI6 80 RXD7_3 (Note 3-1)112 TXRX5P_A
17 TXRX2P_C 49 TX_EN6/TX_CTL6 81 RXD7_2 (Note 3-1)113 TXRX5M_A
18 TXRX2M_C 50 TX_ER6 82 RXD7_1 (Note 3-1)114 AVDDL
19 AVDDL 51 COL6 83 RXD7_0 (Note 3-1)115 TXRX5P_B
20 TXRX2P_D 52 TXD6_3 84 GND 116 TXRX5M_B
21 TXRX2M_D 53 TXD6_2 85 LED4_0 (Note 3-1)117 TXRX5P_C
22 AVDDH 54 TXD6_1 86 LED4_1 (Note 3-1)118 TXRX5M_C
23 DVDDL 55 TXD6_0 87 DVDDL 119 AVDDL
24 TXRX3P_A 56 DVDDL 88 LED3_0 120 TXRX5P_D
25 TXRX3M_A 57 RX_CLK6/REFCLKO6 89 LED3_1 (Note 3-1)121 TXRX5M_D
26 TXRX3P_B 58 RX_DV6/CRS_DV6/
RX_CTL6 90 NC 122 AVDDH
27 TXRX3M_B 59 RX_ER6 91 LED2_0 (Note 3-1)123 GND
28 TXRX3P_C 60 CRS6 92 LED2_1 (Note 3-1)124 AVDDL
29 TXRX3M_C 61 VDDIO 93 PME_N 125 XO
30 AVDDL 62 RXD6_3 (Note 3-1)94 INTRP_N 126 XI
31 TXRX3P_D 63 RXD6_2 (Note 3-1)95 CLKO_25_125 127 ISET
32 TXRX3M_D 64 RXD6_1 (Note 3-1)96 RESET_N 128 AVDDH
Exposed Pad Must be Connected to GND
2017-2019 Microchip Technology Inc. DS00002330D-page 11
KSZ9897R
3.2 Pin Descriptions
This sections details the functions of the various device signals.
TABLE 3-2: PIN DESCRIPTIONS
NAME SYMBOL BUFFER
TYPE DESCRIPTION
Ports 5-1 Gigabit Ethernet Pins
Port 5-1
Ethernet TX/RX
Pair A +
TXRX[5:1]P_A AIO Port 5-1 1000BASE-T Differential Data Pair A (+)
Note: 100BASE-TX and 10BASE-Te are also sup-
ported on the A and B pairs.
Port 5-1
Ethernet TX/RX
Pair A -
TXRX[5:1]M_A AIO Port 5-1 1000BASE-T Differential Data Pair A (-)
Note: 100BASE-TX and 10BASE-Te are also sup-
ported on the A and B pairs.
Port 5-1
Ethernet TX/RX
Pair B +
TXRX[5:1]P_B AIO Port 5-1 1000BASE-T Differential Data Pair B (+)
Note: 100BASE-TX and 10BASE-Te are also sup-
ported on the A and B pairs.
Port 5-1
Ethernet TX/RX
Pair B -
TXRX[5:1]M_B AIO Port 5-1 1000BASE-T Differential Data Pair B (-)
Note: 100BASE-TX and 10BASE-Te are also sup-
ported on the A and B pairs.
Port 5-1
Ethernet TX/RX
Pair C +
TXRX[5:1]P_C AIO Port 5-1 1000BASE-T Differential Data Pair C (+)
Port 5-1
Ethernet TX/RX
Pair C -
TXRX[5:1]M_C AIO Port 5-1 1000BASE-T Differential Data Pair C (-)
Port 5-1
Ethernet TX/RX
Pair D +
TXRX[5:1]P_D AIO Port 5-1 1000BASE-T Differential Data Pair D (+)
Port 5-1
Ethernet TX/RX
Pair D -
TXRX[5:1]M_D AIO Port 5-1 1000BASE-T Differential Data Pair D (-)
Ports 7-6 RGMII/MII/RMII Pins
Port 7-6
Transmit/
Reference
Clock
TX_CLK[7:6]/
REFCLKI[7:6] I/O8 MII Mode: TX_CLK[7:6] is the Port 7-6 25/2.5MHz Trans-
mit Clock. In PHY mode this pin is an output, in MAC mode
it is an input.
RMII Mode: REFCLKI[7:6] is the Port 7-6 50MHz Refer-
ence Clock input when in RMII Normal mode. This pin is
unused when in RMII Clock mode.
RGMII Mode: TX_CLK[7:6] is the Port 7-6 125/25/2.5MHz
Transmit Clock input.
Port 7-6
Transmit
Enable/Control
TX_EN[7:6]/
TX_CTL[7:6] IPD MII/RMII Modes: TX_EN[7:6] is the Port 7-6 Transmit
Enable.
RGMII Mode: TX_CTL[7:6] is the Port 7-6 Transmit Con-
trol.
Port 7-6
Transmit Error
TX_ER[7:6] IPD MII Mode: Port 7-6 Transmit Error input.
RMII/RGMII Modes: Not used. Do not connect this pin in
these modes of operation.
KSZ9897R
DS00002330D-page 12 2017-2019 Microchip Technology Inc.
Port 7-6
Collision Detect
COL[7:6] IPD/O8 MII Mode: Port 7-6 Collision Detect. In PHY mode this pin
is an output, in MAC mode it is an input.
RMII/RGMII Modes: Not used. Do not connect this pin in
these modes of operation.
Port 7-6
Transmit Data 3
TXD[7:6]_3 IPD MII/RGMII Modes: Port 7-6 Transmit Data bus bit 3.
RMII Mode: Not used. Do not connect this pin in this mode
of operation.
Port 7-6
Transmit Data 2
TXD[7:6]_2 IPD MII/RGMII Modes: Port 7-6 Transmit Data bus bit 2.
RMII Mode: Not used. Do not connect this pin in this mode
of operation.
Port 7-6
Transmit Data 1
TXD[7:6]_1 IPD MII/RMII/RGMII Modes: Port 7-6 Transmit Data bus bit 1.
Port 7-6
Transmit Data 0
TXD[7:6]_0 IPD MII/RMII/RGMII Modes: Port 7-6 Transmit Data bus bit 0.
Port 7-6
Receive/
Reference
Clock
RX_CLK[7:6]/
REFCLKO[7:6] I/O24 MII Mode: RX_CLK[7:6] is the Port 7-6 25/2.5MHz
Receive Clock. In PHY mode this pin is an output, in MAC
mode it is an input.
RMII Mode: REFCLKO[7:6] is the Port 7-6 50MHz Refer-
ence Clock output when in RMII Clock mode. This pin is
unused when in RMII Normal mode.
RGMII Mode: RX_CLK[7:6] is the Port 7-6 125/25/2.5MHz
Receive Clock output.
Port 7-6
Receive Data
Valid / Carrier
Sense / Control
RX_DV[7:6]/
CRS_DV[7:6]/
RX_CTL[7:6]
IPD/O24 MII Mode: RX_DV[7:6] is the Port 7-6 Received Data Valid
output.
RMII Mode: CRS_DV[7:6] is the Carrier Sense / Receive
Data Valid output.
RGMII Mode: RX_CTL[7:6] is the Receive Control output.
Note: The RX_DV7/CRS_DV7/RX_CTL7 pin also pro-
vides configuration strap functions during hard-
ware/software resets. Refer to Section 3.2.1,
"Configuration Straps" for additional information.
Port 7-6
Receive Error
RX_ER[7:6] IPD/O24 MII Mode: Port 7-6 Receive Error output.
RMII/RGMII Modes: Not used. Do not connect this pin in
these modes of operation.
Port 7-6
Carrier Sense
CRS[7:6] IPD/O8 MII Mode: Port 7-6 Carrier Sense. In PHY mode this pin is
an output, in MAC mode it is an input.
RMII/RGMII Modes: Not used. Do not connect this pin in
these modes of operation.
TABLE 3-2: PIN DESCRIPTIONS (CONTINUED)
NAME SYMBOL BUFFER
TYPE DESCRIPTION
2017-2019 Microchip Technology Inc. DS00002330D-page 13
KSZ9897R
Port 7-6
Receive Data 3
RXD[7:6]_3 IPD/O24 MII/RGMII Modes: Port 7-6 Receive Data bus bit 3.
RMII Mode: Not used. Do not connect this pin in this mode
of operation.
Note: These pins also provide configuration strap
functions during hardware/software resets.
Refer to Section 3.2.1, "Configuration Straps"
for additional information.
Port 7-6
Receive Data 2
RXD[7:6]_2 IPD/O24 MII/RGMII Modes: Port 7-6 Receive Data bus bit 2.
RMII Mode: Not used. Do not connect this pin in this mode
of operation.
Note: These pins also provide configuration strap
functions during hardware/software resets.
Refer to Section 3.2.1, "Configuration Straps"
for additional information.
Port 7-6
Receive Data 1
RXD[7:6]_1 IPD/O24 MII/RMII/RGMII Modes: Port 7-6 Receive Data bus bit 1.
Note: These pins also provide configuration strap
functions during hardware/software resets.
Refer to Section 3.2.1, "Configuration Straps"
for additional information.
Port 7-6
Receive Data 0
RXD[7:6]_0 IPD/O24 MII/RMII/RGMII Modes: Port 7-6 Receive Data bus bit 0.
Note: These pins also provide configuration strap
functions during hardware/software resets.
Refer to Section 3.2.1, "Configuration Straps"
for additional information.
SPI/I2C/MIIM Interface Pins
SPI/I2C/MIIM
Serial Clock
SCL/MDC IPU SPI/I2C Modes: SCL serial clock.
MIIM Mode: MDC serial clock.
SPI Data Out SDO O8 SPI Mode: Data out (also known as MISO).
I2C/MIIM Modes: Not used.
SPI Data In /
I2C/MIIM Data
In/Out
SDI/SDA/MDIO IPU/O8 SPI Mode: SDI Data In (also known as MOSI).
I2C Mode: SDA Data In/Out.
MIIM Mode: MDIO Data In/Out.
SDI and MDIO are open-drain signals when in the output
state. An external pull-up resistor to VDDIO (1.0kΩ to
4.7kΩ) is required.
SPI Chip Select SCS_N IPU SPI Mode: Chip Select (active low).
I2C/MIIM Modes: Not used.
TABLE 3-2: PIN DESCRIPTIONS (CONTINUED)
NAME SYMBOL BUFFER
TYPE DESCRIPTION
KSZ9897R
DS00002330D-page 14 2017-2019 Microchip Technology Inc.
LED Pins
Port 1
LED Indicator 0
LED1_0 IPU/O8 Port 1 LED Indicator 0.
Active low output sinks current to light an external LED.
Port 1
LED Indicator 1
LED1_1 IPU/O8 Port 1 LED Indicator 1.
Active low output sinks current to light an external LED.
Note: This pin also provides configuration strap func-
tions during hardware/software resets. Refer to
Section 3.2.1, "Configuration Straps" for addi-
tional information.
Port 2
LED Indicator 0
LED2_0 IPU/O8 Port 2 LED Indicator 0.
Active low output sinks current to light an external LED.
Note: This pin also provides configuration strap func-
tions during hardware/software resets. Refer to
Section 3.2.1, "Configuration Straps" for addi-
tional information.
Port 2
LED Indicator 1
LED2_1 IPU/O8 Port 2 LED Indicator 1.
Active low output sinks current to light an external LED.
Note: This pin also provides configuration strap func-
tions during hardware/software resets. Refer to
Section 3.2.1, "Configuration Straps" for addi-
tional information.
Port 3
LED Indicator 0
LED3_0 IPU/O8 Port 3 LED Indicator 0.
Active low output sinks current to light an external LED.
Port 3
LED Indicator 1
LED3_1 IPU/O8 Port 3 LED Indicator 1.
Active low output sinks current to light an external LED.
Note: This pin also provides configuration strap func-
tions during hardware/software resets. Refer to
Section 3.2.1, "Configuration Straps" for addi-
tional information.
Port 4
LED Indicator 0
LED4_0 IPU/O8 Port 4 LED Indicator 0.
Active low output sinks current to light an external LED.
Note: This pin also provides configuration strap func-
tions during hardware/software resets. Refer to
Section 3.2.1, "Configuration Straps" for addi-
tional information.
Port 4
LED Indicator 1
LED4_1 IPU/O8 Port 4 LED Indicator 1.
Active low output sinks current to light an external LED.
Note: This pin also provides configuration strap func-
tions during hardware/software resets. Refer to
Section 3.2.1, "Configuration Straps" for addi-
tional information.
Port 5
LED Indicator 0
LED5_0 IPU/O8 Port 5 LED Indicator 0.
Active low output sinks current to light an external LED.
Port 5
LED Indicator 1
LED5_1 IPU/O8 Port 5 LED Indicator 1.
Active low output sinks current to light an external LED.
Note: This pin also provides configuration strap func-
tions during hardware/software resets. Refer to
Section 3.2.1, "Configuration Straps" for addi-
tional information.
TABLE 3-2: PIN DESCRIPTIONS (CONTINUED)
NAME SYMBOL BUFFER
TYPE DESCRIPTION
2017-2019 Microchip Technology Inc. DS00002330D-page 15
KSZ9897R
Miscellaneous Pins
Interrupt INTRP_N OPU Active low, open-drain interrupt.
Note: This pin requires an external pull-up resistor.
Power
Management
Event
PME_N O8 Power Management Event.
This output signal indicates that an energy detect event has
occurred. It is intended to wake up the system from a low
power mode.
Note: The assertion polarity is programmable (default
active low). An external pull-up resistor is
required for active-low operation; an external
pull-down resistor is required for active-high
operation.
System Reset RESET_N IPU Active low system reset.
The device must be reset either during or after power-on.
An RC circuit is suggested for power-on reset.
Crystal Clock /
Oscillator Input
XI ICLK Crystal clock / oscillator input.
When using a 25MHz crystal, this input is connected to one
lead of the crystal. When using an oscillator, this pin is the
input from the oscillator. The crystal oscillator should have a
tolerance of ±50ppm.
Crystal Clock
Output
XO OCLK Crystal clock / oscillator output.
When using a 25MHz crystal, this output is connected to
one lead of the crystal. When using an oscillator, this pin is
left unconnected.
25/125MHz
Reference
Clock Output
CLKO_25_125 O24 25/125MHz reference clock output, derived from the crystal
input.
Transmit
Output Current
Set Resistor
ISET A Transmit output current set resistor.
This pin configures the physical transmit output current. It
must be connected to GND through a 6.04kΩ 1% resistor.
No Connect NC - No Connect. For proper operation, this pin must be left
unconnected.
Power/Ground Pins
+3.3/2.5/1.8V
I/O Power
VDDIO P +3.3V / +2.5V / +1.8V I/O Power
+2.5V
Analog Power
AVDDH P +2.5V Analog Power
+1.2V
Analog Power
AVDDL P +1.2V Analog Power
+1.2V
Digital Power
DVDDL P +1.2V Digital Power
Ground GND GND Ground (pins and pad)
TABLE 3-2: PIN DESCRIPTIONS (CONTINUED)
NAME SYMBOL BUFFER
TYPE DESCRIPTION
KSZ9897R
DS00002330D-page 16 2017-2019 Microchip Technology Inc.
3.2.1 CONFIGURATION STRAPS
The KSZ9897R utilizes configuration strap pins to configure the device for different modes. While RESET_N is low,
these pins are hi-Z. Pull-up/down resistors are used to create high or low states on these pins, which are internally sam-
pled at the rising edge of RESET_N. All of these pins have a weak internal pull-up or pull-down resistor which provides
a default level for strapping. To strap an LED pin low, use a 750Ω to 1kΩ external pull-down resistor. To strap a non-LED
pin high, use an external 1kΩ to 10kΩ pull-up resistor to VDDIO. Once RESET_N is high, all of these pins become driven
outputs.
Because the internal pull-up/down resistors are not strong, consideration must be given to any other pull-up/down resis-
tors which may reside on the board or inside a device connected to these pins.
When an LED pin is directly driving an LED, the effect of the LED and LED load resistor on the strapping level must be
considered. This is the reason for using a small value resistor to pull an LED pin low. This is especially true when an
LED is powered from a voltage that is higher than VDDIO.
The configuration strap pins and their associated functions are detailed in Ta b l e 3 - 3 .
TABLE 3-3: CONFIGURATION STRAP DESCRIPTIONS
CONFIGURATION
STRAP PIN DESCRIPTION
LED1_1 Flow Control (All Ports)
0: Flow control disabled
1: Flow control enabled (Default)
LED2_1 Link-up Mode (All PHYs)
0: Fast Link-up: Auto-negotiation and auto MDI/MDI-X are disabled
1: Normal Link-up: Auto-negotiation and auto MDI/MDI-X are enabled (Default)
Note: Since Fast Link-up disables auto-negotiation and auto-crossover, it is suitable only
for specialized applications.
LED4_0, LED2_0 When LED2_1 = 1 at strap-in (Normal Link-up):
[LED4_0, LED2_0]: Auto-Negotiation Enable (All PHYs) / NAND Tree Test Mode
00: Reserved
01: Auto-negotiation disabled, forced as 100 Mbps and half duplex. Auto-MDI-X is on.
10: NAND Tree test mode
11: Auto-negotiation enabled (Default)
When LED2_1 = 0 at strap-in (Fast Link-up; All PHYs Full-Duplex; Auto-negotiation and
Auto-MDI-X are off):
LED2_0: 1000BASE-T Master/Slave Mode, 100BASE-T MDI/MDI-X Mode (All PHYs)
0: 1000BASE-T: Slave Mode
100BASE-T: MDI-X
1: 1000BASE-T: Master Mode (Default)
100BASE-T: MDI (Default)
LED4_0: PHY Speed Select (All PHYs)
0: 1000BASE-T
1: 100BASE-TX (Default)
LED4_1, LED3_1 [LED4_1, LED3_1]: Management Interface Mode
00: MIIM (MDIO)
01: I2C
1x: SPI (Default)
LED5_1 Switch Enable at Startup
0: Start Switch is disabled. The switch will not forward packets until the Start Switch bit is set
in the Switch Operation Register.
1: Start Switch is enabled. The switch will forward packets immediately after reset. (Default)
RXD6_3, RXD6_2 [RXD6_3, RXD6_2]: Port 6 Mode
00: RGMII (Default)
01: RMII
10: Reserved
11: MII
2017-2019 Microchip Technology Inc. DS00002330D-page 17
KSZ9897R
RXD6_1 Port 6 MII/RMII Mode
0: MII: PHY Mode (Default)
RMII: Clock Mode. RMII 50MHz reference clock is output on REFCLKO6. (Default)
RGMII: No effect
1: MII: MAC Mode
RMII: Normal Mode. RMII 50MHz reference clock is input on REFCLKI6.
RGMII: No effect
RXD6_0 Port 6 Speed Select
0: 1000Mbps Mode (Default)
1: 100Mbps Mode
Note: If Port 6 is configured for MII or RMII, set the speed to 100Mbps.
RXD7_3, RXD7_2 [RXD7_3, RXD7_2]: Port 7 Mode
00: RGMII (Default)
01: RMII
10: Reserved
11: MII
RXD7_1 Port 7 MII/RMII Mode
0: MII: PHY Mode (Default)
RMII: Clock Mode. RMII 50MHz reference clock is output on REFCLKO7. (Default)
RGMII: No effect
1: MII: MAC Mode
RMII: Normal Mode. RMII 50MHz reference clock is input on REFCLKI7.
RGMII: No effect
RXD7_0 Port 7 Speed Select
0: 1000Mbps Mode (Default)
1: 100Mbps Mode
Note: If Port 7 is configured for MII or RMII, set the speed to 100Mbps.
RX_DV7/CRS_DV7/
RX_CTL7 In-Band Management
0: Disable In-Band Management (Default)
1: Enable In-Band Management
TABLE 3-3: CONFIGURATION STRAP DESCRIPTIONS (CONTINUED)
CONFIGURATION
STRAP PIN DESCRIPTION
KSZ9897R
DS00002330D-page 18 2017-2019 Microchip Technology Inc.
4.0 FUNCTIONAL DESCRIPTION
This section provides functional descriptions for the following:
•Physical Layer Transceiver (PHY)
•LEDs
•Media Access Controller (MAC)
•Switch
•NAND Tree Support
•Clocking
•Power
•Power Management
•Management Interface
•In-Band Management
•MAC Interface (RGMII/MII/RMII Port 6-7)
4.1 Physical Layer Transceiver (PHY)
Ports 1 through 5 include completely integrated triple-speed (10BASE-Te, 100BASE-TX, 1000BASE-T) Ethernet phys-
ical layer transceivers for transmission and reception of data over standard four-pair unshielded twisted pair (UTP), CAT-
5 or better Ethernet cable.
The device reduces board cost and simplifies board layout by using on-chip termination resistors for the four differential
pairs, eliminating the need for external termination resistors. The internal chip termination and biasing provides signifi-
cant power savings when compared with using external biasing and termination resistors.
The device can automatically detect and correct for differential pair misplacements and polarity reversals, and correct
for propagation delay differences between the four differential pairs, as specified in the IEEE 802.3 standard for
1000BASE-T operation.
4.1.1 1000BASE-T TRANSCEIVER
The 1000BASE-T transceiver is based on a mixed-signal/digital signal processing (DSP) architecture, which includes
the analog front-end, digital channel equalizers, trellis encoders/decoders, echo cancelers, cross-talk cancelers, a pre-
cision clock recovery scheme, and power-efficient line drivers.
4.1.1.1 Analog Echo Cancellation Circuit
In 1000BASE-T mode, the analog echo cancellation circuit helps to reduce the near-end echo. This analog hybrid circuit
relieves the burden of the ADC and the adaptive equalizer. This circuit is disabled in 10BASE-Te/100BASE-TX mode.
4.1.1.2 Automatic Gain Control (AGC)
In 1000BASE-T mode, the automatic gain control circuit provides initial gain adjustment to boost up the signal level. This
pre-conditioning circuit is used to improve the signal-to-noise ratio of the receive signal.
4.1.1.3 Analog-to-Digital Converter (ADC)
In 1000BASE-T mode, the analog-to-digital converter digitizes the incoming signal. ADC performance is essential to the
overall performance of the transceiver. This circuit is disabled in 10BASE-Te/100BASE-TX mode.
4.1.1.4 Timing Recovery Circuit
In 1000BASE-T mode, the mixed signal clock recovery circuit, together with the digital phase locked loop (PLL), is used
to recover and track the incoming timing information from the received data. The digital PLL has very low long-term jitter
to maximize the signal-to-noise ratio of the receive signal.
The 1000BASE-T slave PHY must transmit the exact receive clock frequency recovered from the received data back to
the 1000BASE-T master PHY. Otherwise, the master and slave will not be synchronized after long transmission. This
also helps to facilitate echo cancellation and NEXT removal.
2017-2019 Microchip Technology Inc. DS00002330D-page 19
KSZ9897R
4.1.1.5 Adaptive Equalizer
In 1000BASE-T mode, the adaptive equalizer provides the following functions:
• Detection for partial response signaling
• Removal of NEXT and ECHO noise
• Channel equalization
Signal quality is degraded by residual echo that is not removed by the analog hybrid because of impedance mismatch.
The device uses a digital echo canceler to further reduce echo components on the receive signal.
In 1000BASE-T mode, data transmission and reception occurs simultaneously on all four pairs of wires (four channels).
This results in high-frequency cross-talk coming from adjacent wires. The device uses three NEXT cancelers on each
receive channel to minimize the cross-talk induced by the other three channels.
In 10BASE-Te/100BASE-TX mode, the adaptive equalizer needs only to remove the inter-symbol interference and
recover the channel loss from the incoming data.
4.1.1.6 Trellis Encoder and Decoder
In 1000BASE-T mode, the transmitted 8-bit data is scrambled into 9-bit symbols and further encoded into 4D-PAM5
symbols. On the receiving side, the idle stream is examined first. The scrambler seed, pair skew, pair order and polarity
must be resolved through the logic. The incoming 4D-PAM5 data is then converted into 9-bit symbols and de-scrambled
into 8-bit data.
4.1.2 100BASE-TX TRANSCEIVER
4.1.2.1 100BASE-TX Transmit
The 100BASE-TX transmit function performs parallel-to-serial conversion, 4B/5B coding, scrambling, NRZ-to-NRZI con-
version, and MLT3 encoding and transmission.
The circuitry starts with a parallel-to-serial conversion, which converts the MII data from the MAC into a 125MHz serial
bit stream. The data and control stream is then converted into 4B/5B coding, followed by a scrambler. The serialized
data is further converted from NRZ-to-NRZI format, and then transmitted in MLT3 current output. An external ISET resis-
tor sets the output current for the 1:1 transformer ratio.
The output signal has a typical rise/fall time of 4ns and complies with the ANSI TP-PMD standard regarding amplitude
balance, overshoot, and timing jitter. The wave-shaped 10BASE-Te output driver is also incorporated into the 100BASE-
TX driver.
4.1.2.2 100BASE-TX Receive
The 100BASE-TX receiver function performs adaptive equalization, DC restoration, MLT3-to-NRZI conversion, data and
clock recovery, NRZI-to-NRZ conversion, de-scrambling, 4B/5B decoding, and serial-to-parallel conversion.
The receiving side starts with the equalization filter to compensate for inter-symbol interference (ISI) over the twisted
pair cable. Since the amplitude loss and phase distortion is a function of the cable length, the equalizer has to adjust its
characteristics to optimize performance. In this design, the variable equalizer makes an initial estimation based on com-
parisons of incoming signal strength against some known cable characteristics, and then tunes itself for optimization.
This is an ongoing process and self-adjusts against environmental changes such as temperature variations.
Next, the equalized signal goes through a DC restoration and data conversion block. The DC restoration circuit is used
to compensate for the effect of baseline wander and to improve the dynamic range. The differential data conversion
circuit converts the MLT3 format back to NRZI. The slicing threshold is also adaptive.
The clock recovery circuit extracts the 125MHz clock from the edges of the NRZI signal. This recovered clock is then
used to convert the NRZI signal into the NRZ format. This signal is sent through the de-scrambler followed by the 4B/
5B decoder. Finally, the NRZ serial data is converted to an MII format and provided as the input data to the MAC.
4.1.2.3 Scrambler/De-Scrambler
The purpose of the scrambler is to spread the power spectrum of the signal to reduce electromagnetic interference (EMI)
and baseline wander. The scrambler is used only for 100BASE-TX.
Transmitted data is scrambled through the use of an 11-bit wide linear feedback shift register (LFSR). The scrambler
generates a 2047-bit non-repetitive sequence. Then the receiver de-scrambles the incoming data stream using the
same sequence as at the transmitter.
KSZ9897R
DS00002330D-page 20 2017-2019 Microchip Technology Inc.
4.1.3 10BASE-Te TRANSCEIVER
10BASE-Te is an energy-efficient version of 10BASE-T which is powered from a 2.5V supply. It has a reduced transmit
signal amplitude and requires Cat5 cable. It is inter-operable to 100m with 10BASE-T when Cat5 cable is used.
4.1.3.1 10BASE-Te Transmit
The 10BASE-Te driver is incorporated with the 100BASE-TX driver to allow for transmission using the same magnetics.
They are internally wave-shaped and pre-emphasized into outputs with typical 1.75V amplitude (compared to the typical
transmit amplitude of 2.5V for 10BASE-T). The harmonic contents are at least 27dB below the fundamental frequency
when driven by an all-ones Manchester-encoded signal.
4.1.3.2 10BASE-Te Receive
On the receive side, input buffers and level detecting squelch circuits are employed. A differential input receiver circuit
and a phase-locked loop (PLL) perform the decoding function.
The Manchester-encoded data stream is separated into clock signal and NRZ data. A squelch circuit rejects signals with
levels less than 400mV or with short pulse widths to prevent noise at the RXP1 or RXM1 input from falsely triggering
the decoder. When the input exceeds the squelch limit, the PLL locks onto the incoming signal and the device decodes
a data frame. The receiver clock is maintained active during idle periods in between data reception.
4.1.4 AUTO MDI/MDI-X
The automatic MDI/MDI-X feature, also known as auto crossover, eliminates the need to determine whether to use a
straight cable or a crossover cable between the device and its link partner. The auto-sense function detects the MDI/
MDI-X pair mapping from the link partner, and assigns the MDI/MDI-X pair mapping of the device accordingly. Table 4 -
1 shows the device’s 10/100/1000 Mbps pin configuration assignments for MDI and MDI-X pin mapping.
Auto MDI/MDI-X is enabled by default. It can be disabled through the port control registers. If Auto MDI/MDI-X is dis-
abled, the port control register can also be used to select between MDI and MDI-X settings.
An isolation transformer with symmetrical transmit and receive data paths is recommended to support Auto MDI/MDI-X.
4.1.5 PAIR-SWAP, ALIGNMENT, AND POLARITY CHECK
In 1000Base-T mode, the device:
• Detects incorrect channel order and automatically restores the pair order for the A and B pairs. This is also done
separately for the C and D pairs. Crossing of A or B pairs to C or D pairs is not corrected.
• Supports 50±10ns difference in propagation delay between pairs of channels in accordance with the IEEE 802.3
standard, and automatically corrects the data skew so the corrected four pairs of data symbols are synchronized.
Incorrect pair polarities of the differential signals are automatically corrected for all speeds.
4.1.6 WAVE SHAPING, SLEW-RATE CONTROL, AND PARTIAL RESPONSE
In communication systems, signal transmission encoding methods are used to provide the noise-shaping feature and
to minimize distortion and error in the transmission channel.
• For 1000BASE-T, a special partial-response signaling method is used to provide the bandwidth-limiting feature for
the transmission path.
• For 100BASE-TX, a simple slew-rate control method is used to minimize EMI.
• For 10BASE-Te, pre-emphasis is used to extend the signal quality through the cable.
TABLE 4-1: MDI/MDI-X PIN DEFINITIONS
Pin (RJ45 pair)
MDI MDI-X
1000BASE-T 100BASE-TX 10BASE-Te 1000BASE-T 100BASE-TX 10BASE-Te
TXRXxP/M_A (1,2)A+/-TX+/-TX+/-B+/-RX+/-RX+/-
TXRXxP/M_B (3,6) B+/- RX+/- RX+/- A+/- TX+/- TX+/-
TXRXxP/M_C (4,5) C+/- Not used Not used D+/- Not used Not used
TXRXxP/M_D (7,8) D+/- Not used Not used C+/- Not used Not used
2017-2019 Microchip Technology Inc. DS00002330D-page 21
KSZ9897R
4.1.7 AUTO-NEGOTIATION
The device conforms to the auto-negotiation protocol as described by IEEE 802.3. Auto-negotiation allows each port to
operate at either 10BASE-Te, 100BASE-TX or 1000BASE-T by allowing link partners to select the best common mode
of operation. During auto-negotiation, the link partners advertise capabilities across the link to each other and then com-
pare their own capabilities with those they received from their link partners. The highest speed and duplex setting that
is common to the two link partners is selected as the mode of operation.
The following list shows the speed and duplex operation mode from highest to lowest priority.
• Priority 1: 1000BASE-T, full-duplex
• Priority 2: 1000BASE-T, half-duplex
• Priority 3: 100BASE-TX, full-duplex
• Priority 4: 100BASE-TX, half-duplex
• Priority 5: 10BASE-Te, full-duplex
• Priority 6: 10BASE-Te, half-duplex
If the KSZ9897R link partner doesn’t support auto-negotiation or is forced to bypass auto-negotiation for 10BASE-Te
and 100BASE-TX modes, the KSZ9897R port sets its operating mode by observing the signal at its receiver. This is
known as parallel detection, and allows the KSZ9897R to establish a link by listening for a fixed signal protocol in the
absence of the auto-negotiation advertisement protocol.
The auto-negotiation link-up process is shown in Figure 4-1.
For 1000BASE-T mode, auto-negotiation is always required to establish a link. During 1000BASE-T auto-negotiation,
the master and slave configuration is first resolved between link partners. Then the link is established with the highest
common capabilities between link partners.
FIGURE 4-1: AUTO-NEGOTIATION AND PARALLEL OPERATION
KSZ9897R
DS00002330D-page 22 2017-2019 Microchip Technology Inc.
Auto-negotiation is enabled by default after power-up or hardware reset. Afterwards, auto-negotiation can be enabled
or disabled via bit 12 of the PHY Basic Control Register. If auto-negotiation is disabled, the speed is set by bits 6 and
13 of the PHY Basic Control Register, and the duplex is set by bit 8.
If the speed is changed on the fly, the link goes down and either auto-negotiation or parallel detection initiate until a
common speed between the KSZ9897R and its link partner is re-established for a link.
If link is already established and there is no change of speed on the fly, the changes (for example, duplex and pause
capabilities) will not take effect unless either auto-negotiation is restarted through bit 9 of the PHY Basic Control Regis-
ter, or a link-down to link-up transition occurs (i.e. disconnecting and reconnecting the cable).
After auto-negotiation is completed, the link status is updated in the PHY Basic Status Register, and the link partner
capabilities are updated in the PHY Auto-Negotiation Link Partner Ability Register, PHY Auto-Negotiation Expansion
Status Register, and PHY 1000BASE-T Status Register.
4.1.8 FAST LINK-UP
Link up time is normally determined by the time it takes to complete auto-negotiation. Additional time may be added by
the auto MDI/MDI-X feature. The total link up time from power-up or cable connect is typically a second or more.
Fast Link-up mode significantly reduces 100BASE-TX link-up time by disabling both auto-negotiation and auto MDI/
MDI-X, and fixing the TX and RX channels. This mode is enabled or disabled by the LED2_1 strapping option. It is not
set by registers, so fast link-up is available immediately upon power-up. Fast Link-up is available at power-up only for
100BASE-TX link speed, which is selected by strapping the LED4_0 pin high. Fast Link-up is also available for 10BASE-
Te, but this link speed must first be selected via a register write.
Fast Link-up is intended for specialized applications where both link partners are known in advance. The link must also
be known so that the fixed transmit channel of one device connects to the fixed receive channel of the other device, and
vice versa. The TX and RX channel assignments are determined by the MDI/MDI-X strapping option on LED2_0.
If a device in Fast Link-up mode is connected to a normal device (auto-negotiate and auto-MDI/MDI-X), there will be no
problems linking, but the speed advantage of Fast Link-up will not be realized.
For more information on configuration straps, refer to Section 3.2.1, "Configuration Straps," on page 16.
4.1.9 LinkMD® CABLE DIAGNOSTICS
The LinkMD® function utilizes Time Domain Reflectometry (TDR) to analyze the cabling for common cabling problems,
such as open circuits, short circuits and impedance mismatches.
LinkMD® works by sending a pulse of known amplitude and duration down the MDI or MDI-X pair, and then analyzing
the shape of the reflected signal to determine the type of fault. The time duration for the reflected signal to return pro-
vides the approximate distance to the cabling fault. The LinkMD® function processes this TDR information and presents
it as a numerical value that can be translated to a cable distance.
A LinkMD test is initiated individually for each PHY and for a specific PHY differential pair.
4.1.9.1 Usage
To run a LinkMD test on all four pairs of one PHY, follow this flow.
1. Disable auto-negotiation: Write 0 to of register 0xN100-0xN101 bit 12.
2. Configure register 0xN112-0xN113 to enable master-slave manual configuration mode.
3. Start cable diagnostic by writing 1 to register 0xN124-0xN125 bit 15. This enable bit is self-clearing.
4. Wait (poll) for register 0xN124-0xN125 bit 15 to return 0, which indicates that the cable diagnostic test is com-
pleted. Alternatively, wait 250ms.
5. Read cable diagnostic test status in register 0xN124-0xN125 bits [9-8]. The results are:
a) 00 = normal operation
b) 01 = open condition detected in cable (valid result)
c) 10 = short condition detected in cable (valid result)
d) 11 = cable diagnostic test invalid (test failed)
The ‘11’ case occurs when the PHY is unable to shut down the link partner. In this instance, the test is not run because
it would be impossible for the PHY to determine if the detected signal is a reflection of the signal generated or a signal
from another source.
2017-2019 Microchip Technology Inc. DS00002330D-page 23
KSZ9897R
6. For status 01 or 10, read the Cable Diagnostic Result in register 0xN124-0xN125 bits [7:0]. Get distance to fault
by the following formula:
7. Distance to fault (meters) = 0.8 * (Cable Diagnostic Result – 22).
8. To test another differential pair on this PHY, change the value of register 0xN124-0xN125 bits [13:12] when initi-
ating the test.
9. Return the registers to their original values and restart auto-negotiation.
The following script will test the four pairs of port 1. For other ports, change the register addresses accordingly.
“ww” = write word (16-bits) [register] [data]
“rw” = read word (16-bits) [register]
Values are hexadecimal.
ww 1100 0140 # initialization
ww 1112 1000 # initialization
ww 1124 8000 # initiate test for pair A
sleep 250 msec
rw 1124 # read result for pair A
ww 1124 9000 # initiate test for pair B
sleep 250 msec
rw 1124 # read result for pair B
ww 1124 a000 # initiate test for pair C
sleep 250 msec
rw 1124 # read result for pair C
ww 1124 b000 # initiate test for pair D
sleep 250 msec
rw 1124 # read result for pair D
ww 1112 0700 # return register to default setting
ww 0 1340 # return register to default setting (may vary by application)
4.1.10 REMOTE PHY LOOPBACK
This loopback mode checks the line (differential pairs, transformer, RJ-45 connector, Ethernet cable) transmit and
receive data paths between the KSZ9897R and its Ethernet PHY link partner, and is supported for 10/100/1000 Mbps
at full-duplex.
The loopback data path is shown in Figure 4-2 and functions as follows:
• The Ethernet PHY link partner transmits data to the KSZ9897R PHY port.
• Data received at the external pins of the PHY port is looped back without passing through the MAC and internal
switch fabric.
• The same KSZ9897R PHY port transmits data back to the Ethernet PHY link partner.
KSZ9897R
DS00002330D-page 24 2017-2019 Microchip Technology Inc.
The following programming steps and register settings are for remote PHY loopback mode for 1000BASE-T Master
Mode, 1000BASE-T Slave Mode, 100BASE-TX Mode, and 10BASE-T Mode.
• 1000BASE-T Master Mode
- Set Port N (1-5), PHY 1000BASE-T Control Register = 0x1F00
- Set Port N (1-5), PHY Remote Loopback Register = 0x01F0
- Set Port N (1-5), PHY Basic Control Register = 0x1340
• 1000BASE-T Slave Mode
- Set Port N (1-5), PHY 1000BASE-T Control Register = 0x1300
- Set Port N (1-5), PHY Remote Loopback Register = 0x01F0
- Set Port N (1-5), PHY Basic Control Register = 0x1340
• 100BASE-TX Mode
- Set Port N (1-5), PHY Auto-Negotiation Advertisement Register = 0x0181
- Set Port N (1-5), PHY 1000BASE-T Control Register = 0x0C00
- Set Port N (1-5), PHY Remote Loopback Register = 0x01F0
- Set Port N (1-5), PHY Basic Control Register = 0x3300
• 10BASE-T Mode
- Set Port N (1-5), PHY Auto-Negotiation Advertisement Register = 0x0061
- Set Port N (1-5), PHY 1000BASE-T Control Register = 0x0C00
- Set Port N (1-5), PHY Remote Loopback Register = 0x01F0
- Set Port N (1-5), PHY Basic Control Register = 0x3300
FIGURE 4-2: REMOTE PHY LOOPBACK
RJ-45 Switch
Fabric
MAC
10/100/1000
PHY
Device PHY Port N
CAT-5
(UTP)
RJ-45 Ethernet PHY
Link Partner
2017-2019 Microchip Technology Inc. DS00002330D-page 25
KSZ9897R
4.2 LEDs
Each PHY port has two programmable LED output pins, LEDx_0 and LEDx_1, to indicate the PHY link and activity sta-
tus. Two different LED modes are available. The LED mode can be changed individually for each PHY port by writing
to the PHY Mode bit in the PHY indirect register: MMD 2, address 0, bit 4:
•1 = Single-LED Mode
•0 = Tri-Color Dual-LED Mode (Default)
Each LED output pin can directly drive an LED with a series resistor (typically 220Ω to 470Ω). LED outputs are active-
low.
4.2.1 SINGLE-LED MODE
In single-LED mode, the LEDx_1 pin indicates the link status while the LEDx_0 pin indicates the activity status, as shown
in Figure 4-2.
4.2.2 TRI-COLOR DUAL-LED MODE
In tri-color dual-LED mode, the link and activity status are indicated by the LEDx_1 pin for 1000BASE-T; by the LEDx_0
pin for 100BASE-TX; and by both LEDx_1 and LEDx_0 pins, working in conjunction, for 10BASE-T. This behavior is
summarized in Figure 4-3.
4.3 Media Access Controller (MAC)
4.3.1 MAC OPERATION
The device strictly abides by IEEE 802.3 standards to maximize compatibility. Additionally, there is an added MAC fil-
tering function to filter unicast packets. The MAC filtering function is useful in applications, such as VoIP, where restrict-
ing certain packets reduces congestion and thus improves performance.
The transmit MAC takes data from the egress buffer and creates full Ethernet frames by adding the preamble and the
start-of-frame delimiter ahead of the data, and generates the FCS that is appended to the end of the frame. It also sends
flow control packets as needed.
TABLE 4-2: SINGLE-LED MODE PIN DEFINITION
LED Pin Pin State Pin LED Definition Link/Activity
LEDx_1 HOFFLink Off
L ON Link On (any speed)
LEDx_0 H OFF No Activity
Toggle Blinking Activity (RX,TX)
TABLE 4-3: TRI-COLOR DUAL-LED MODE PIN DEFINITION
LED Pin (State) LED Pin (Definition) Link/Activity
LEDx_1 LEDx_0 LEDx_1 LEDx_0
H H OFF OFF Link off
L H ON OFF 1000Mbps Link / No Activity
Toggle H Blinking OFF 1000Mbps Link / Activity (RX,TX)
H L OFF ON 100Mbps Link / No Activity
H Toggle OFF Blinking 100Mbps Link / Activity (RX,TX)
L L ON ON 10Mbps Link / No Activity
Toggle Toggle Blinking Blinking 10Mbps Link / Activity (RX,TX)
KSZ9897R
DS00002330D-page 26 2017-2019 Microchip Technology Inc.
The receive MAC accepts data via the integrated PHY or via the MII/RMII/RGMII interface. It decodes the data bytes,
strips off the preamble and SFD of each frame. The destination and source addresses and VLAN tag are extracted for
use in filtering and address/ID lookup, and the MAC also calculates the CRC of the received frame, which is compared
to the FCS field. The MAC can discard frames that are the wrong size, that have an FCS error, or when the source MAC
address matches the Switch MAC address.
The receive MAC also implements the Wake on LAN (WoL) feature. This system power saving feature is described in
detail in the Section 4.8, "Power Management".
MIB statistics are collected in both receive and transmit directions.
4.3.2 INTER-PACKET GAP (IPG)
If a frame is successfully transmitted, then the minimum 96-bit time for IPG is specified as being between two consec-
utive packets. If the current packet is experiencing collisions, the minimum 96-bit time for IPG is specified as being from
carrier sense (CRS) to the next transmit packet.
4.3.3 BACK-OFF ALGORITHM
The device implements the IEEE standard 802.3 binary exponential back-off algorithm in half-duplex mode. After 16
consecutive collisions, the packet is dropped.
4.3.4 LATE COLLISION
If a transmit packet experiences collisions after 512 bit times of the transmission, the packet is dropped.
4.3.5 LEGAL PACKET SIZE
On all ports, the device discards received packets smaller than 64 bytes (excluding VLAN tag, including FCS) or larger
than the maximum size. The default maximum size is the IEEE standard of 1518 bytes, but the device can be configured
to accept jumbo packets up to 9000 bytes. Jumbo packet traffic on multiple ports can stress switch resources and cause
activation of flow control.
4.3.6 FLOW CONTROL
The device supports standard MAC Control PAUSE (802.3x flow control) frames in both the transmit and receive direc-
tions for full-duplex connections.
In the receive direction, if a PAUSE control frame is received on any port, the device will not transmit the next normal
frame on that port until the timer, specified in the PAUSE control frame, expires. If another PAUSE frame is received
before the current timer expires, the timer will then update with the new value in the second PAUSE frame. During this
period (while it is flow controlled), only flow control packets from the device are transmitted.
In the transmit direction, the device has intelligent and efficient ways to determine when to invoke flow control and send
PAUSE frames. The flow control is based on availability of the system resources, including available buffers and avail-
able transmit queues.
The device issues a PAUSE frame containing the maximum pause time defined in IEEE standard 802.3x. Once the
resource is freed up, the device sends out another flow control frame with zero pause time to turn off the flow control
(turn on transmission to the port). A hysteresis feature is provided to prevent the flow control mechanism from being
constantly activated and deactivated.
4.3.7 HALF-DUPLEX BACK PRESSURE
A half-duplex back pressure option (non-IEEE 802.3 standard) is also provided. The activation and deactivation condi-
tions are the same as in full-duplex mode. If back pressure is required, the device sends preambles to defer the other
stations' transmission (carrier sense deference).
To avoid jabber and excessive deference (as defined in the 802.3 standard), after a certain time, the device discontinues
the carrier sense and then raises it again quickly. This short silent time (no carrier sense) prevents other stations from
sending out packets thus keeping other stations in a carrier sense deferred state. If the port has packets to send during
a back pressure situation, the carrier sense type back pressure is interrupted and those packets are transmitted instead.
If there are no additional packets to send, carrier sense type back pressure is reactivated again until chip resources free
up. If a collision occurs, the binary exponential back-off algorithm is skipped and carrier sense is generated immediately,
thus reducing the chance of further collision and carrier sense is maintained to prevent packet reception.
2017-2019 Microchip Technology Inc. DS00002330D-page 27
KSZ9897R
To ensure no packet loss in 10BASE-Te or 100BASE-TX half-duplex modes, the user must enable the following:
• No excessive collision drop (Switch MAC Control 1 Register)
• Back pressure (Port MAC Control 1 Register)
4.3.8 FLOW CONTROL AND BACK PRESSURE REGISTERS
Table 4-4 provides a list of flow control and back pressure related registers.
4.3.9 BROADCAST STORM PROTECTION
The device has an intelligent option to protect the switch system from receiving too many broadcast packets. As the
broadcast packets are forwarded to all ports except the source port, an excessive number of switch resources (band-
width and available space in transmit queues) may be utilized. The device has the option to include “multicast packets”
for storm control. The broadcast storm rate parameters are programmed globally, and can be enabled or disabled on a
per port basis. The rate is based on a 5ms interval for 1000BASE-T, a 50ms interval for 100BASE-TX and a 500ms
interval for 10BASE-Te. At the beginning of each interval, the counter is cleared to zero and the rate limit mechanism
starts to count the number of bytes during the interval. The rate definition is described in control registers. The default
setting equates to a rate of 1%.
4.3.10 SELF-ADDRESS FILTERING
Received packets can be filtered (dropped) if their source address matches the device's MAC address. This feature is
useful for automatically terminating packets once they have traversed a ring network and returned to their source. It can
be enabled on a per-port basis via the Switch Lookup Engine Control 1 Register and Port Control 2 Register.
4.4 Switch
4.4.1 SWITCHING ENGINE
A high-performance switching engine is used to move data to and from the MAC's packet buffers. It operates in store
and forward mode, while an efficient switching mechanism reduces overall latency. The switching engine has a
256KByte internal frame buffer that is shared between all the ports.
For the majority of switch functions, all of the data ports are treated equally. However, a few functions such as IGMP
snooping, 802.1X, forwarding invalid VLAN packets, etc., give special recognition to the host port. Any port (but most
commonly port 6 or port 7) may be assigned as the host port by enabling tail tagging mode for that port. Only one port
may be a host port.
TABLE 4-4: FLOW CONTROL AND BACK PRESSURE REGISTERS
Registers Description
LED Configuration Strap Register LED configuration strap settings.
(LED1_1 enables flow control and back pressure)
Switch MAC Address 0 Register
through
Switch MAC Address 5 Register
Switch's MAC address, used as source address of PAUSE control
frames
Switch MAC Control 0 Register “Aggressive back-off” enable
Switch MAC Control 1 Register BP mode, “Fair mode” enable, “no excessive collision drop” enable
Switch MAC Control 4 Register Pass PAUSE control frames
Port Status Register Flow control enable (per port)
PHY Auto-Negotiation Advertisement Register PHY - flow control advertisement (per port)
Port MAC Control 1 Register Half-duplex back pressure enable (per port)
Port Ingress Rate Limit Control Register Ingress rate limit flow control enable (per port)
Port Control 0 Register Drop mode (per port)
KSZ9897R
DS00002330D-page 28 2017-2019 Microchip Technology Inc.
When a switch receives a non-error packet, it checks the packet's destination MAC address. If the address is known,
the packet is forwarded to the output port that is associated with the destination MAC address. The following paragraphs
describe the key functions of destination address lookup and source address learning. These processes may be com-
bined with VLAN support and other features, which are described in the subsequent sub-sections.
4.4.2 ADDRESS LOOKUP
Destination address lookup is performed in three separate internal address tables in the device:
1. Address Lookup (ALU) Table: 4K dynamic + static entries
2. Static Address Table: 16 static entries
3. Reserved Multicast Address Table: 8 pre-configured static entries
4.4.2.1 Address Lookup (ALU) Table
The Address Lookup (ALU) Table stores MAC addresses and their associated information. This table holds both
dynamic and static entries. Dynamic entries are created automatically in hardware, as described in Section 4.4.2.4,
"Learning". Static entries are created by management software.
This table is a 4-way associative memory, with 1K buckets, for a total of 4K entries. A hash function translates the
received packet's MAC address (and optionally the FID) into a 10-bit index for accessing the table. At each bucket are
four fully-associative address entries. All four entries are simultaneously compared to the MAC address (plus optional
FID) for a possible match.
Three options are available for the hashing function, as described in Tab l e 4 - 5 . If VLAN is enabled (802.1Q VLAN
Enable bit in the Switch Lookup Engine Control 0 Register), the VLAN group (FID) is included in the hashing function
along with the MAC address. If VLAN is not enabled the hashing function is applied to MAC address and the FID in the
default VLAN (VID=1) which is 0.
4.4.2.2 Static Address Table
The 16-entry Static Address Table is typically used to hold multicast addresses, but is not limited to this. As with static
entries in the ALU table, entries in the Static Address Table are created by management software. It serves the same
function as static entries that are created in the ALU table, so its use is optional.
TABLE 4-5: ADDRESS LOOKUP TABLE HASHING OPTIONS
HASH_OPTION
(Switch Lookup Engine
Control 0 Register)
Description
01b (Default) A hash algorithm based on the CRC of the MAC address plus FID. The hash algorithm uses
the CRC-CCITT polynomial. The input to the hash is reduced to a 16-bit CRC hash value.
Bits [9:0] of the hash value plus (binary addition) 7-bit FID (zero extended on the left) are
used as an index to the table. The CRC-CCITT polynomial is: X16+X12+X5+1.
10b An XOR algorithm based on 16 bits of the XOR of the triple-folded MAC address. Bits [9:0]
of the XOR value plus 7-bit FID (left-extended) are used to index the table.
00b or 11b A direct algorithm. The 10 least significant bits of the MAC address plus 7 bit FID are used
to index the table.
2017-2019 Microchip Technology Inc. DS00002330D-page 29
KSZ9897R
4.4.2.3 Reserved Multicast Address Table
The Reserved Multicast Address Table holds 8 pre-configured address entries, as defined in Table 4-6. This table is an
optional feature that is disabled at power-on. If desired, the forwarding ports may be modified.
If a match is found in one of the tables, then the destination port is read from that table entry. If a match is found in more
than one table, static entries will take priority over dynamic entries.
4.4.2.4 Learning
The internal lookup engine updates the ALU table with a new dynamic entry if the following conditions are met:
• The received packet's source address (SA) does not exist in the lookup table.
• The received packet has no errors, and the packet size is of legal length.
• The received packet has a unicast SA.
The lookup engine inserts the qualified SA into the table, along with the port number and age count. If all four table
entries are valid, the oldest of the (up to four) dynamic entries may be deleted to make room for the new entry. Static
entries are never deleted by the learning process. If all four entries are static entries, the address is not learned but an
interrupt is generated and the table index number is made available to the interrupt service routine.
4.4.2.5 Migration
The internal lookup engine also monitors whether a station has moved. If a station has moved, it updates the ALU table
accordingly. Migration happens when the following conditions are met:
• The received packet's SA is in the table but the associated source port information is different.
• The received packet has no receiving errors, and the packet size is of legal length.
The lookup engine updates the existing record in the table with the new source port information.
TABLE 4-6: RESERVED MULTICAST ADDRESS TABLE
Group Address MAC Group
Address Function
Default PORT FORWARD
Value
(defines forwarding port: P7...P1)
Default Forwarding Action
0 (01-80-C2-00)-00-00 Bridge Group Data 100_0000 Forward only to the highest
numbered port
(default host port)
1 (01-80-C2-00)-00-01 MAC Control Frame
(typically flow control)
000_0000 Drop MAC flow control
2 (01-80-C2-00)-00-03 802.1X Access
Control
100_0000 Forward to highest num-
bered port
3 (01-80-C2-00)-00-10 Bridge Management 111_1111 Flood to all ports
4 (01-80-C2-00)-00-20 GMRP 011_1111 Flood to all ports except high-
est numbered port
5 (01-80-C2-00)-00-21 GVRP 011_1111 Flood to all ports except high-
est numbered port
6 (01-80-C2-00)-00-02,
(01-80-C2-00)-00-04 –
(01-80-C2-00)-00-0F
100_0000 Forward to highest num-
bered port
7 (01-80-C2-00)-00-11 -
(01-80-C2-00)-00-1F,
(01-80-C2-00)-00-22 -
(01-80-C2-00)-00-2F
011_1111 Flood to all ports except high-
est numbered port
KSZ9897R
DS00002330D-page 30 2017-2019 Microchip Technology Inc.
4.4.2.6 Aging
The lookup engine updates the age count information of a dynamic record in the ALU table whenever the corresponding
SA appears. The age count is used in the aging process. If a record is not updated for a period of time, the lookup engine
removes the record from the table. The lookup engine constantly performs the aging process and continuously removes
aging records. The aging period is about 300 seconds (±75 seconds) and can be configured longer or shorter (1 second
to 30 minutes). This feature can be enabled or disabled. Static entries are exempt from the aging process.
4.4.2.7 Forwarding
The device forwards packets using the algorithm that is depicted in Figure 4-3. Figure 4-3 shows stage one of the for-
warding algorithm where the search engine looks up the VLAN ID, static table, and dynamic table for the destination
address, and comes up with “port to forward 1" (PTF1). PTF1 is then further modified by spanning tree, IGMP snooping,
port mirroring, and port VLAN processes.
The ACL process works in parallel with the flow outlined above. The authentication and ACL processes have the highest
priority in the forwarding process, and the ACL result may override the result of the above flow. The output of the ACL
process is the final “port-to-forward 2" (PTF2) destination port(s).
The device will not forward the following packets:
• Error packets: These include framing errors, frame check sequence (FCS) errors, alignment errors, and illegal
size packet errors.
• MAC Control PAUSE frames: The device intercepts these packets and performs full duplex flow control accord-
ingly.
• “Local” packets: Based on destination address (DA) lookup. If the destination port from the lookup table matches
the port from which the packet originated, the packet is defined as “local”.
• In-Band Management packets.
FIGURE 4-3: PACKET FORWARDING PROCESS FLOWCHART
Start
PTF1=NULL VLAN ID Valid?
no
Get PTF1 from
Static Array
Search
Static
Array
found
Get PTF1 from
Address Table
Search
Address
Look-up
Table
found
Get PTF1 from
VLAN Table
PTF1
yes
not found
not found
- Search VLAN table
- Ingress VLAN filtering
-Discard NPVID check
Search based on
DA or DA+FID
Search based on
DA+FID
PTF1
Spanning Tree
Process
Port Mirror
Process
IGMP / MLD
Process
PTF2
- Check receiving port s receive enable bit
- Check destination port s transmit enable bit
- Check whether packets are special (BPDU)
- IGMP / MLD packets are forwarded
to Host port
- Process does not apply to packets
received at Host port
-RX Mirror
-TX Mirror
- RX or TX Mirror
- RX and TX Mirror
Port
Authentication
& ACL
2017-2019 Microchip Technology Inc. DS00002330D-page 31
KSZ9897R
4.4.2.8 Lookup Engine Registers
Table 4-7 provides a list of lookup engine related registers.
4.4.3 IEEE 802.1Q VLAN
Virtual LAN is a means of segregating a physical network into multiple virtual networks whereby traffic may be confined
to specific subsets of the greater network. IEEE 802.1Q defines a VLAN protocol using a 4-byte tag that is added to the
Ethernet frame header. The device supports port-based and tag-based VLANs, including tagging, un-tagging, forward-
ing and filtering.
4.4.3.1 Non-Tag Port-Based VLAN
The simplest VLAN method establishes forwarding restrictions on a port-by-port basis without using VLAN tags. There
is a register for each ingress port that is used to specify the allowed forwarding ports. An incoming packet is restricted
from being forwarded to any egress port that is disallowed for that ingress port. The settings are made in the Port Control
1 Register. This function is always enabled; it is not enabled and disabled by the 802.1Q VLAN Enable bit in the Switch
Lookup Engine Control 0 Register. The default setting is to allow all ingress-to-egress port paths.
4.4.3.2 Tag-Based VLAN
When 802.1Q VLAN is enabled, an internal VLAN Table with 4k entries is used to a store port membership list, VLAN
group ID (FID) and additional information relating to each VLAN. This table must be set up by an administrator prior to
enabling 802.1Q VLAN. Enabling is done by setting the 802.1Q VLAN Enable bit in the Switch Lookup Engine Control
0 Register.
In 802.1Q VLAN mode, the lookup process starts with VLAN Table lookup, using the tag's VID as the address. The first
step is to determine whether the VID is valid. If the VID is not valid, the packet is dropped and its address is not learned.
Alternatively, unknown VID packets may be forwarded to pre-defined ports or to the host port. If the VID is valid, the FID
is retrieved for further lookup. The FID + Destination Address (hashed(DA) + FID) are used to determine the destination
port. The FID + Source Address (hashed(SA) + FID) are used for address learning (see Table 4-9 and Table 4-10).
TABLE 4-7: LOOKUP ENGINE REGISTERS
Registers Description
Global Interrupt Status Register,
Global Interrupt Mask Register
Top level LUE interrupt
Switch Lookup Engine Control 0 Register,
Switch Lookup Engine Control 1 Register,
Switch Lookup Engine Control 2 Register,
Switch Lookup Engine Control 3 Register
Misc.
Address Lookup Table Interrupt Register,
Address Lookup Table Mask Register
Low level LUE interrupts
Address Lookup Table Entry Index 0 Register,
Address Lookup Table Entry Index 1 Register
Access failure address/index
ALU Table Index 0 Register,
ALU Table Index 1 Register,
ALU Table Access Control Register,
Static Address and Reserved Multicast Table
Control Register,
ALU / Static Address Table Entry 1 Register,
ALU / Static Address / Reserved Multicast
Table Entry 2 Register,
ALU / Static Address Table Entry 3 Register,
ALU / Static Address Table Entry 4 Register
Address table access registers
KSZ9897R
DS00002330D-page 32 2017-2019 Microchip Technology Inc.
The hashed(DA) + FID are hashed and used for forwarding lookup in the Address Lookup and Static Address Tables.
For a successful address table lookup, the FID fields must also match. If the match fails, the packet is broadcast to all
the VLAN port members defined in the VLAN Table entry. If there is a match and egress VLAN filtering is enabled, the
packet is forwarded to those ports that are in both the address table port forwarding list and the VLAN table port mem-
bership list.
A similar address table lookup is performed using the hashed(SA) + FID. If the lookup fails, the FID and SA are learned.
If a non-tagged or null-VID-tagged packet is received, the ingress port default VID (Port Default Tag 0 Register and Port
Default Tag 1 Register) is used for lookup.
Table 4-8 details the forwarding and discarding actions that are taken for the various VLAN scenarios. The first entry in
the table is explained by the fact that VLAN Table lookup is enabled even when 802.1Q VLAN is not enabled. Notice
that in the Port Default Tag 0 Register and Port Default Tag 1 Register, the port default VID is 1 for each port. Corre-
spondingly, the VLAN port membership list in the VLAN Table entry for VID=1 is pre-configured at power-on to all ones.
This provides the standard Ethernet switch behavior of broadcasting all packets with unknown destination address. If
the VLAN table entry # 1 is changed, or if the port default VID is changed, this may affect the forwarding action for
“unknown packets” even when VLAN is not enabled.
It should also be noted that the default values of the Egress VLAN Filtering bits are zero. These bits are zero only for
backwards compatibility with previous “KSZ” switches. The resulting switch behavior, in the event of a successful VLAN
and ALU lookups, is to forward the packet to the ports in the address table port forwarding list, without regard to the
VLAN port membership list. It is suggested that the Egress VLAN Filtering bits be set to one so that the VLAN port mem-
bership list from the VLAN Table will be used to qualify the forwarding determined from the address lookup.
TABLE 4-8: VLAN FORWARDING
VLAN
Enable
(Note 4-1)
VLAN
Match/
Valid
(Note 4-2)
Forward
Option
(Note 4-3)
Egress
VLAN
Filtering
(Note 4-4)
Unknown
VID
Forward
(Note 4-5)
Drop
Invalid
VID
(Note 4-6)
ALU
Match/
Valid
(Note 4-7)
Action
0 X X X X X No Forward to port membership list
of default VID in LAN table
0 X X X X X Yes Forward to Address Lookup
port forwarding list
1 No X X 0 0 X Forward to host port
1 No X X 0 (def) 1 (def) X Discard
1 No X X 1 X X Forward to Unknown VID
packet forward port list
1Yes0 X X X NoBroadcast: Forward to VLAN
table port membership list
(PORT FORWARD)
Multicast: Forward to
Unknown Multicast ports if UM
is enabled. Else, forward to
VLAN table port membership
list.
Unicast: Forward to Unknown
Unicast ports if UU is enabled.
Else forward to VLAN table port
membership list.
1 Yes 0 0 (def) X X Yes Forward to address table
lookup port forwarding list
1 Yes 0 1 X X Yes Forward to address table
lookup port forwarding list &
VLAN table port membership
list (bitwise AND)
1 Yes 1 X X X Yes Forward to VLAN table port
membership list
2017-2019 Microchip Technology Inc. DS00002330D-page 33
KSZ9897R
Note 4-1 VLAN Enable is bit 7 in the Switch Lookup Engine Control 0 Register
Note 4-2 VLAN Match/Valid indicates when the VLAN Table entry is valid
Note 4-3 Forward Option is a bit in the VLAN Table Entry 0 Register
Note 4-4 Egress VLAN Filtering are bits 5 and 4 in the Switch Lookup Engine Control 2 Register
Note 4-5 Unknown VID Forwarding is in the Unknown VLAN ID Control Register
Note 4-6 Drop Invalid VID is bit 6 in the Switch Lookup Engine Control 0 Register
Note 4-7 ALU Match/Valid indicates when the Address Lookup is a success
Table 4-9 describes in more detail the address lookup process that follows the VLAN Table lookup. Lookup occurs in
both the Address Lookup Table and the Static Address Table simultaneously, and the resulting action depends on the
results of the two lookups.
A source address (SA) lookup is also performed in the Address Lookup Table. SA lookup also performs SA filtering and
MAC priority when the address is hit. Ta b le 4- 1 0 describes how learning is performed in the Address Lookup Table when
a successful VLAN table lookup has been done and the no matching static entry is found in the Address Lookup Table
or the Static Address Table.
Note: “(def)” indicates the default power-up value.
TABLE 4-9: HASHED(DA) + FID LOOKUP IN VLAN MODE
DA found in
Static MAC
Table?
Use FID Flag?
(Static MAC Table) FID Match? DA+FID found in
ALU Table? Action
No Don’t Care Don’t Care No Lookup has failed. Broadcast to
the membership ports defined in
the VLAN Table
No Don’t Care Don’t Care Yes Send to the destination port
defined in the Address Lookup
(ALU) Table
Yes 0 Don’t Care Don’t Care Send to the destination port(s)
defined in the Static Address Table
Yes 1 No No Lookup has failed. Broadcast to
the membership ports defined in
the VLAN Table.
Yes 1 No Yes Send to the destination port
defined in the Address Lookup
(ALU) Table
Yes 1 Yes Don’t Care Send to the destination port(s)
defined in the Static Address Table
TABLE 4-10: HASHED(SA) + FID LOOKUP IN VLAN MODE
FID + SA found in Address Lookup (ALU) Table? Action
No Learn and add FID + SA to the Address Lookup (ALU) Table
Yes If the static bit is 0, the time stamp and the egress port map
are updated. If the static bit is 1, then nothing is done.
KSZ9897R
DS00002330D-page 34 2017-2019 Microchip Technology Inc.
4.4.3.2.1 Tag Insertion and Removal
Tag insertion is enabled on all ports when the VLAN feature is enabled. At the ingress port, untagged packets are tagged
with the ingress port's default tag. The default tag is separately programmable for each port. The switch does not add
tags to already tagged packets unless double tagging is enabled.
At the egress port, tagged packets will have their 802.1Q VLAN tags removed if un-tagging is enabled in the VLAN table
entry. Untagged packets will not be modified if 802.1Q is enabled.
4.4.3.2.2 Double Tagging
The switch supports double tagging, also known as Q-in-Q or VLAN stacking. This feature can be used for service pro-
viders to append a second VLAN tag in addition to a first VLAN tag applied by the customer. VLAN support can be
enabled either with or without double tagging. When double tagging is enabled, the outer tag is recognized and is used
for VLAN and address lookup instead of the inner tag. The outer tag precedes the inner tag in the frame header: the
outer tag is located immediately after the source address, and contains a different Tag Protocol Identifier (TPID) value
than the inner tag.
Additional controls are available for full control of the VLAN function. Some of these features can be enabled on a per-
port basis, while others are global:
•Ingress VLAN Filtering: Discard packet if VID port membership in VLAN table does not include the ingress port.
•Discard non PVID Packet: Discard packet if VID does not match the ingress port default VID.
•Discard un-tagged Packet: Discard any received packet without a tag.
•Drop tag: Drops the packet if it is VLAN tagged.
•Unknown VID Forward: Forward to a fixed set of ports if VLAN lookup fails.
•Drop unknown VID: Additional options for unknown VID packets: discard or forward to the host port.
•Null VID Replacement: Replace a null VID with the ingress port default VID.
•PVID Replacement: Replace a non-null VID with the ingress port default VID.
•Double Tag Mcast Trap: In double tag mode, trap all reserved multicast packets and forward to the host port.
4.4.3.3 VLAN Registers
Table 4-11 provides a list of VLAN related registers.
TABLE 4-11: VLAN REGISTERS
Registers Description
Switch Operation Register Double tag enable
Switch Lookup Engine Control 0 Register VLAN enable; Drop invalid VID frames
Switch Lookup Engine Control 2 Register Trap double tagged MC frames;
Dynamic & status egress VLAN filtering
Unknown VLAN ID Control Register Forward unknown VID
Switch MAC Control 2 Register Null VID replacement with PVID at egress
VLAN Table Entry 0 Register,
VLAN Table Entry 1 Register,
VLAN Table Entry 2 Register,
VLAN Table Index Register,
VLAN Table Access Control Register
Read/write access to the VLAN table
Port Default Tag 0 Register,
Port Default Tag 1 Register
Port default tag
Port Ingress MAC Control Register Drop non-VLAN frames; Tag drop
Port Transmit Queue PVID Register PVID replacement at egress
Port Control 2 Register VLAN table lookup for VID=0;
Ingress VLAN filtering; PVID mismatch discard
2017-2019 Microchip Technology Inc. DS00002330D-page 35
KSZ9897R
4.4.4 QUALITY-OF-SERVICE (QOS) PRIORITY SUPPORT
The device provides quality-of-service (QoS) for applications such as VoIP. There are multiple methods for assigning
priority to ingress packets. Depending on the packet prioritization method, the packet priority levels are mapped to the
egress queues for each port. Each port can be configured for 1, 2, and 4 egress queues, which are prioritized. The
default is 1 queue per port.
When configured for 4 priority queues, Queue 3 is the highest priority queue and Queue 0 is the lowest priority. Likewise,
for a 2-queue configuration, Queue 1 is the highest priority queue. If a port is not configured as 2 or 4 queues, then high
priority and low priority packets have equal priority in the single transmit queue.
There is an additional option for every port to select either to always deliver packets from the highest priority queue first,
or use weighted round robin queuing amongst the multiple queues. This is described later in Section 4.4.13, "Scheduling
and Rate Limiting".
4.4.4.1 Port-Based Priority
With port-based priority, each ingress port is individually classified as a specific priority level. All packets received at the
high-priority receiving port are marked as high priority and are sent to the high-priority transmit queue if the correspond-
ing transmit queue is split into 2 or 4 queues.
4.4.4.2 IEEE 802.1p-Based Priority
For IEEE 802.1p-based priority, the device examines the ingress packets to determine whether they are tagged. If
tagged, the 3-bit PCP priority field in the VLAN tag is retrieved and used to look up the “priority mapping” value. The
“priority mapping” value is programmable.
Figure 4-4 illustrates how the 802.1p priority field is embedded in the 802.1Q VLAN tag.
4.4.4.3 IEEE 802.1p Priority Field Re-Mapping
This is a QoS feature that allows the device to set the “User Priority Ceiling” at any ingress port. If the ingress packet's
priority field has a higher priority value than the default tag's priority field of the ingress port, the packet's priority field is
replaced with the default tag's priority field.
4.4.4.4 DiffServ (DSCP) Priority (IP)
DiffServ-based priority from the DSCP field in the IP header can be used to determine packet priority. The 6-bit DSCP
value is used as an index to a set of registers which translate the 6-bit DSCP value to a 2-bit value that specifies one of
the 4 (or 2) queues. These registers are fully programmable.
4.4.4.5 ACL Priority
The Access Control List (ACL) Filtering feature can also be used to assign priority to received packets. This is discussed
in Section 4.4.16, "Access Control List (ACL) Filtering".
FIGURE 4-4: 802.P PRIORITY FIELD FORMAT
KSZ9897R
DS00002330D-page 36 2017-2019 Microchip Technology Inc.
4.4.5 TRAFFIC CONDITIONING & POLICING
4.4.5.1 Two Rate Three Color Marker
The Two Rate Three Color Marker meters an IP packet stream and marks its packets green, yellow, or red. A packet is
marked red if it exceeds the Peak Information Rate (PIR). Otherwise, it is marked either yellow or green depending on
whether it exceeds or doesn't exceed the Committed Information Rate (CIR).
The Meter operates in one of two modes. In the Color-Blind mode, the Meter assumes that the packet stream is uncol-
ored. In the Color-Aware mode, the Meter assumes that some preceding entity has pre-colored the incoming packet
stream so that each packet is green, yellow, or red. The Marker (re)colors an IP packet according to the results of the
Meter.
4.4.5.2 Weighted Random Early Detection (WRED)
The WRED feature monitors the average queue size of packet memory and ingress queue size of each traffic class,
and drops packets based on memory and queue utilization. If the buffers are almost empty, all incoming traffic is
accepted. As the buffer utilization increases, the probability for dropping an incoming packet also increases.
WRED is intended to avoid the problem of global synchronization. Global synchronization can occur when a switch
becomes congested and begins dropping incoming packets all at once. For TCP streams, packet drops invoke the TCP
congestion control mechanism, which reduce the transmission rate until there are no more packet drops. If there are
many TCP streams and their congestion control mechanisms act in unison, this can cause an undesirable oscillation in
traffic rates. By selectively dropping some packets early rather than waiting until the buffer is full, WRED avoids dropping
large numbers of packets at once and minimizes the chances of global synchronization.
The packet drop probability is based on the minimum threshold, maximum threshold, and a probability multiplier. When
the average queue depth is above the minimum threshold, packets start getting dropped. The rate of packet drop
increases linearly as the average queue size increases until the average queue size reaches the maximum threshold.
The probability multiplier is the fraction of packets dropped when the average queue depth is at the maximum threshold.
When the average queue size is above the maximum threshold, all packets are dropped.
4.4.6 SPANNING TREE SUPPORT
To support spanning tree, one port is the designated port for the host processor, which is defined as the port for which
tail tagging is enabled. Each of the other ports can be configured in one of the five spanning tree states via “transmit
enable”, “receive enable” and “learning disable” register bits. Tab l e 4 - 1 2 shows the setting and software actions taken
for each of the five spanning tree states.
TABLE 4-12: SPANNING TREE STATES
Disable State Port Setting Software Action
The port should not forward
or receive any packets.
Learning is disabled.
transmit enable = 0
receive enable = 0
learning disable = 1
The processor should not send any packets to the
port. The switch may still send specific packets to the
processor (packets that match some entries in the
“Static MAC Table” with “overriding bit” set) and the
processor should discard those packets. Address
learning is disabled on the port in this state.
Blocking State Port Setting Software Action
Only packets to the proces-
sor are forwarded.
Learning is disabled.
transmit enable = 0
receive enable = 0
learning disable = 1
The processor should not send any packets to the
port(s) in this state. The processor should program
the “Static MAC Table” with the entries that it needs
to receive (for example, BPDU packets). The “over-
riding” bit should also be set so that the switch will
forward those specific packets to the processor.
Address learning is disabled on the port in this state.
Listening State Port Setting Software Action
2017-2019 Microchip Technology Inc. DS00002330D-page 37
KSZ9897R
4.4.7 RAPID SPANNING TREE SUPPORT
There are three operational states assigned to each port for the Rapid Spanning Tree Protocol (RSTP):
1. Discarding State
2. Learning State
3. Forwarding State
4.4.7.1 Discarding State
Discarding ports do not participate in the active topology and do not learn MAC addresses.
• Discarding state: the state includes three states of the disable, blocking and listening of STP.
• Port setting: transmit enable = “0”, receive enable = “0”, learning disable = “1”.
• Software action: The host processor should not send any packets to the port. The switch may still send specific
packets to the processor (packets that match some entries in the static table with “overriding bit” set) and the pro-
cessor should discard those packets. When the port's learning capability (learning disable = '1') is disabled, port
related entries in the ALU table and static MAC table can be rapidly flushed.
4.4.7.2 Learning State
Ports in “learning state” learn MAC addresses, but do not forward user traffic.
• Learning State: Only packets to and from the host processor are forwarded. Learning is enabled.
• Port setting for Learning State: transmit enable = “0”, receive enable = “0”, learning disable = “0”.
• Software action: The processor should program the Static Address Table with the entries that it needs to receive
(e.g., BPDU packets). The “overriding” bit should be set so that the switch will forward those specific packets to
the processor. The processor may send packets to the port(s) in this state (see Section 4.4.9, "Tail Tagging Mode"
for details). Address learning is enabled on the port in this state.
4.4.7.3 Forwarding State
Ports in “forwarding states” fully participate in both data forwarding and MAC learning.
• Forwarding state: Packets are forwarded and received normally. Learning is enabled.
• Port setting: transmit enable = “1”, receive enable = “1”, learning disable = “0”.
Only packets to and from the
processor are forwarded.
Learning is disabled.
transmit enable = 0
receive enable = 0
learning disable = 1
The processor should program the “Static MAC
Table” with the entries that it needs to receive (for
example, BPDU packets). The “overriding” bit should
be set so that the switch will forward those specific
packets to the processor. The processor may send
packets to the port(s) in this state. Address learning
is disabled on the port in this state.
Learning State Port Setting Software Action
Only packets to and from the
processor are forwarded.
Learning is enabled.
transmit enable = 0
receive enable = 0
learning disable = 0
The processor should program the “Static MAC
Table” with the entries that it needs to receive (for
example, BPDU packets). The “overriding” bit should
be set so that the switch will forward those specific
packets to the processor. The processor may send
packets to the port(s) in this state. Address learning
is enabled on the port in this state.
Forwarding State Port Setting Software Action
Packets are forwarded and
received normally.
Learning is enabled.
transmit enable = 1
receive enable = 1
learning disable = 0
The processor programs the “Static MAC Table” with
the entries that it needs to receive (for example,
BPDU packets). The “overriding” bit is set so that the
switch forwards those specific packets to the proces-
sor. The processor can send packets to the port(s) in
this state. Address learning is enabled on the port in
this state.
TABLE 4-12: SPANNING TREE STATES (CONTINUED)
KSZ9897R
DS00002330D-page 38 2017-2019 Microchip Technology Inc.
• Software action: The host processor should program the Static Address Table with the entries that it needs to
receive (e.g., BPDU packets). The “overriding” bit should be set so that the switch will forward those specific pack-
ets to the processor. The processor may send packets to the port(s) in this state (see Section 4.4.9, "Tail Tagging
Mode" for details). Address learning is enabled on the port in this state.
RSTP uses only one type of BPDU called RSTP BPDUs. They are similar to STP configuration BPDUs with the excep-
tion of a type field set to “version 2" for RSTP and “version 0" for STP, and a flag field carrying additional information.
4.4.8 MULTIPLE SPANNING TREE SUPPORT
Multiple Spanning Tree Protocol (MSTP) is an extension of RSTP that allows different VLANs to have different spanning
tree configurations. The VLAN Table, Address Lookup Table and Static Address Table all contain a 3-bit field which can
be used to specify one of eight spanning trees. Each port contains state registers for specifying unique states for each
of the spanning trees.
4.4.9 TAIL TAGGING MODE
Tail tagging is a method to communicate ingress and egress port information between the host processor and the switch.
It is useful for spanning tree protocol, IGMP/MLD snooping, and other applications. As shown in Figure 4-5, the tail tag
is inserted at the end of the packet, between the payload and the 4-byte CRC / FCS.
When the switch forwards a received packet to the host port, one tail tagging byte is added to the packet by the
switch to indicate to the host processor the port that the packet was received on. The format is shown in
Table 4-13.
In the opposite direction, the host processor must add two tail tag bytes to each packet that it sends to the switch to
indicate the intended egress ports. When multiple priority queues are enabled, the tail tag is also used to indicate the
priority queue. The format is shown in Table 4-14. This tail tag is removed by the switch before the packet leaves the
switch. If the Lookup bit (bit 10) is set, packet forwarding follows the standard forwarding process, and bits [9:0] are
ignored. When the Lookup bit is not set, bits [8:0] determine the forwarding ports and priority queue, while the Override
bit (bit 9) determines whether port blocking is overridden.
Tail tagging applies only to the host port, never to any other ports of the switch.
FIGURE 4-5: TAIL TAG FRAME FORMAT
TABLE 4-13: RECEIVE TAIL TAG FORMAT (FROM SWITCH TO HOST)
Bits Description
7 PTP Message Indication
0 = Is not a PTP message. A 4-byte receive timestamp has not been added.
1 = Is a PTP message. A 4-byte timestamp has been added before the tail tag.
6:3 Reserved
2:0 Received Port
000 = Packet received at Port 1
001 = Packet received at Port 2
010 = Packet received at Port 3
011 = Packet received at Port 4
100 = Packet received at Port 5
101 = Packet received at Port 6
110 = Packet received at Port 7
DEST
ADDRESS
SOURCE
ADDRESS
66
802.1Q
TAG
(4)
ETYPE or
LENGTH
2
TAIL
TAG
2 (From Host)
1 (To Host)
PAYLOAD
46 (42) - 1500
FCS
4BYTES
2017-2019 Microchip Technology Inc. DS00002330D-page 39
KSZ9897R
By default, tail tagging is disabled. To enable it, set the Tail Tag Enable bit in one of the Port Operation Control 0 Register
at address 0xN020 for port “N”. When this bit is set for one port, that port is referred to as the “host” port. Do not set the
Tail Tag Enable bit for more than one port.
4.4.10 IGMP SUPPORT
For Internet Group Management Protocol (IGMP) support in Layer 2, the device provides two components:
•“IGMP” Snooping
•“Multicast Address Insertion” in the Static MAC Table
4.4.10.1 “IGMP” Snooping
The device traps IGMP packets and forwards them only to the processor (host port). The IGMP packets are identified
as IP packets (either Ethernet IP packets, or IEEE 802.3 SNAP IP packets) with IP version = 0x4 and protocol version
number = 0x2.
4.4.10.2 “Multicast Address Insertion” in the Static MAC Table
Once the multicast address is programmed in the Static Address Table or Address Lookup Table, the multicast session
is trimmed to the subscribed ports, instead of broadcasting to all ports.
So that the host processor knows which port the IGMP packet was received on, Tail Tagging Mode must be enabled.
4.4.11 IPV6 MLD SNOOPING
The device traps IPv6 Multicast Listener Discovery (MLD) packets and forwards them only to the processor (host port).
4.4.12 PORT MIRRORING
the device supports “port mirroring” comprehensively as:
•“Receive Only” Mirror-on-a-Port
•“Transmit Only” Mirror-on-a-Port
•“Receive and Transmit” Mirror-on-a-Port
TABLE 4-14: TRANSMIT TAIL TAG FORMAT (FROM HOST TO SWITCH)
Bits Description
15:11 Reserved
10 Lookup
0 = Port forwarding is determined by tail tag bits [9:0] below.
1 = Tail tag bits [9:0] are ignored and port forwarding is determined by the standard
switch forwarding process (address lookup, VLAN, etc.)
9 Port Blocking Override
When set, packets will be sent from the specified port(s) regardless, and any port
blocking (see Port Transmit Enable in Port MSTP State Register) is ignored.
8:7 Egress priority (0 to 3)
6 Forward to Port 7
5 Forward to Port 6
4 Forward to Port 5
3 Forward to Port 4
2 Forward to Port 3
1 Forward to Port 2
0 Forward to Port 1
Note: The port for which Tail Tagging Mode is enabled is the host port.
KSZ9897R
DS00002330D-page 40 2017-2019 Microchip Technology Inc.
4.4.12.1 “Receive Only” Mirror-on-a-Port
All the packets received on the port are mirrored on the sniffer port. For example, 1 is programmed to be “receive sniff”
and the host port is programmed to be the “sniffer”. A packet received on port 1 is destined to port 2 after the internal
lookup. The packet is forwarded to both port 2 and the host port. The device can optionally even forward “bad” received
packets to the “sniffer port”.
4.4.12.2 “Transmit Only” Mirror-on-a-Port
All the packets transmitted on the port are mirrored on the sniffer port. For example, port 1 is programmed to be “transmit
sniff” and the host port is programmed to be the “sniffer port”. A packet received on port 2 is destined to port 1 after the
internal lookup. The device forwards the packet to both port 1 and the host port.
4.4.12.3 “Receive and Transmit” Mirror-on-a-Port
All the packets received on port A and transmitted on port B are mirrored on the sniffer port. For example, port 1 is pro-
grammed to be “receive sniff”, port 2 is programmed to be “transmit sniff”, and the host port is programmed to be the
“sniffer port”. A packet received on port 1 is destined to port 2 after the internal lookup. The device forwards the packet
to both port 2 and the host port.
Multiple ports can be selected as “receive sniff” or “transmit sniff”. In addition, any port can be selected as the “sniffer
port”.
4.4.13 SCHEDULING AND RATE LIMITING
Each device port has two egress packet scheduling options, which can be applied when the port is configured for two
or four queues. Additionally, each port has ingress and egress rate limiter features.
4.4.13.1 Strict Priority Scheduling
When an egress port is configured as two or four queues, and strict priority scheduling is selected, each queue will take
absolute priority over all lower priority queues. If a packet is available to transmit from queue 3 (the highest priority
queue), then it will take priority for transmission over any packet that will also be available in any of the other queues.
A packet in queue 2 will be transmitted only if no packet is available in queue 3. Weighted round robin is an alternative
to strict priority scheduling.
4.4.13.2 Weighted Round Robin (WRR) Scheduling
WRR scheduling is an alternative to strict priority scheduling for egress queues. It is referred to as fair queuing because
it gives proportionally higher priority to the highest priority queue, but not absolute priority.
4.4.13.3 Rate Limiting
The device supports independent ingress and egress hardware rate limiting on each port. Normally these two features
are considered mutually exclusive, and users are discouraged from using both on the same port.
For 10BASE-Te, a rate setting above 10Mbps means the rate is not limited. Likewise, for 100BASE-TX, a rate setting
above 100Mbps means the rate is not limited. On the receive side, the data receive rate for each priority at each port
can be limited by setting up ingress rate control registers. On the transmit side, the data transmit rate for each priority
queue at each port can be limited by setting up egress rate control registers. The size of each frame has options to
include minimum inter-frame gap (IFG) or preamble byte, in addition to the data field (from packet DA to FCS).
For ingress rate limiting, the device provides options to selectively choose frames from all types, multicast, broadcast,
and flooded unicast frames. The data rate from those selected type of frames is counted. Packets are dropped at the
ingress port when the data rate exceeds the specified rate limit.
For egress rate limiting, the leaky bucket algorithm is applied to each output priority queue for shaping output traffic.
Inter-frame gap is stretched on a per frame base to generate smooth, non-burst egress traffic. The throughput of each
output priority queue is limited by the egress rate specified.
If any egress queue receives more traffic than the specified egress rate throughput, packets may be accumulated in the
output queue and packet memory. After the memory of the queue or the port is used up, packet dropping or flow control
will be triggered. As a result of congestion, the actual egress rate may be dominated by flow control/dropping at the
ingress end, and may be therefore slightly less than the specified egress rate.
To reduce congestion, it is a good practice to ensure that the egress bandwidth exceeds the ingress bandwidth.
2017-2019 Microchip Technology Inc. DS00002330D-page 41
KSZ9897R
4.4.14 INGRESS MAC ADDRESS FILTERING FUNCTION
When a packet is received, the destination MAC address is looked up in both the static and dynamic MAC address
tables. If the address is not found in either of these tables, then the destination MAC address is “unknown”. By default,
an unknown packet is forwarded to all ports except the port at which it was received. An optional feature makes it pos-
sible to specify the port or ports to which to forward unknown packets. It is also possible to specify no ports, meaning
that unknown packets will be discarded. This feature is implemented separately for unknown unicast, unknown multicast
and unknown VID packets.
4.4.15 802.1X ACCESS CONTROL
IEEE 802.1X is a Port-based authentication protocol. EAPOL is the protocol normally used by the authentication pro-
cess as uncontrolled Port. By receiving and extracting special EAPOL frames, the host processor can control whether
the ingress and egress ports should forward packets or not. If a user port wants service from another port (authentica-
tor), it must get approved by the authenticator. The device detects EAPOL frames by checking the destination address
of the frame. The destination addresses should be either a multicast address as defined in IEEE 802.1x (01-80-C2-00-
00-03) or an address used in the programmable reserved multicast address domain with offset -00-03. Once EAPOL
frames are detected, the frames are forwarded to the host port so it can send the frames to the authenticator server.
Eventually, the CPU determines whether the requester is qualified or not based on its source MAC address, and frames
are either accepted or dropped.
When the device is configured as an authenticator, the ports of the switch must then be configured for authorization. In
an authenticator-initiated port authorization, a client is powered up or plugs into the port, and the authenticator port
sends an Extensible Authentication Protocol (EAP) PDU to the supplicant requesting the identification of the supplicant.
At this point in the process, the port on the switch is connected from a physical standpoint; however, the 802.1X process
has not authorized the port and no frames are passed from the port on the supplicant into the switching fabric. If the
supplicant attached to the switch (KSZ9897R) did not understand the EAP PDU that it was receiving from the switch, it
would not be able to send an ID and the port would remain unauthorized. In this state, the port would be blocked from
passing any user traffic. If the supplicant is running the 802.1X EAP, it would respond to the request with its configured
ID. (This could be a user name/password combination or a certificate.)
After the device receives the ID from the supplicant, it passes the ID information to an authentication server (RADIUS
server) that can verify the identification information. The RADIUS server responds to the switch with either a success
or failure message. If the response is a success, the port will be authorized and user traffic will be allowed to pass
through the port like any switch port connected to an access device. If the response is a failure, the port will remain
unauthorized and, therefore, unused. If there is no response from the server, the port will also remain unauthorized and
will not pass any traffic.
Port control can be performed via the Access Control List (ACL) Filtering feature.
4.4.16 ACCESS CONTROL LIST (ACL) FILTERING
An Access Control List (ACL) can be created for each port to perform filtering on incoming layer 2 MAC, layer 3 IP or
layer 4 TCP/UDP packets. Multicast filtering is handled in the Static Address Table and the Reserved Multicast Address
Table, but the ACL provides additional capabilities for filtering routed network protocols. As shown in Figure 4-3, ACL
filtering may take precedence over other forwarding functions.
The ACL allows the switch to filter ingress traffic based on the following header fields:
• Source or destination MAC address and/or EtherType
• Source or destination IPv4 address with programmable mask
• IPv4 protocol
• Source or destination UDP port
• Source or destination TCP port
• TCP Flag with programmable mask
The ACL is implemented as an ordered list of up to 16 access control rules which are programmed into the ACL Table.
Each entry specifies certain rules (a set of matching conditions and action rules) to control the forwarding and priority
of packets. When a packet is received on an interface, the switch compares the fields in the packet against any applied
ACLs to verify that the packet has the permissions required to be forwarded, based on the conditions specified in the
lists. Multiple match conditions can be either AND'ed or OR'ed together.
KSZ9897R
DS00002330D-page 42 2017-2019 Microchip Technology Inc.
The ACL can also implement a count function that generates an interrupt rather than a forwarding action. The counter
can be either a watchdog timer or an event counter. As a watchdog timer, an interrupt is generated if a packet with a
specific MAC address and EtherType is not received within a specified time interval. As an event counter, an interrupt
is generated once a specified number of packets with a specific MAC address and EtherType have been received.
The ACL consists of three parts: matching rules, action rules, and processing entries. A matching rule specifies what
comparison test shall be performed on the incoming packet. It can also enable a counter function. An action rule spec-
ifies the forwarding action to be taken if the matching test succeeds. Alternatively, when a count function is enabled in
a matching rule, the 11-bit count value is stored in the corresponding action rule field and there is no forwarding action.
In general, the 16 matching rules are not directly linked to the 16 action rules. For example, matching entry #0 is not
necessarily related to action entry #0. The exception is when the counter function is enabled in a matching rule, whereby
the matching rule and action rule fields at the same ACL table entry will function together and are no longer independent.
Each of the 16 processing entries is used to link any number of matching rules (specified in RuleSet) to any one action
rule (specified in FRN). When there are multiple matching rules in a RuleSet, those rules are AND'ed together. Only if
all of those matching results are true will the FRN action be taken.
It is also possible to configure the ACL table so that multiple processing entries specify the same action rule. In this way,
the final matching result is the OR of the matching results from each of the multiple RuleSets.
The 16 ACL rules represent an ordered list, with entry #0 having the highest priority and entry #15 having the lowest
priority. All matching rules are evaluated. If there are multiple true match results and multiple corresponding actions, the
highest priority (lowest numbered) of those actions will be the one taken.
4.4.16.1 Processing Entry Description
The Processing Entry consists of two parameters as described in Ta b l e 4 - 1 5 .
TABLE 4-15: ACL PROCESSING ENTRY PARAMETERS
Parameter Description
FRN[3:0] First Rule Number
Pointer to an Action rule entry. Possible values are 0 to 15. If all Matching rules speci-
fied in the RuleSet are evaluated true, then this is the resulting Action rule.
RuleSet[15:0] Specifies a set of one or more Matching rule entries.
RuleSet has one bit for each of the 16 Matching rule entries. If multiple Matching rules
are selected, then all conditions will be AND'ed to produce a final match result.
0 = Matching rule not selected
1 = Matching rule selected
2017-2019 Microchip Technology Inc. DS00002330D-page 43
KSZ9897R
The examples in Figure 4-6 are interpreted as follows:
• Rule #0: Test the matching rule entry #0. If true, apply action rule entry #0.
• Rule #1: Test the matching rule entry #1. If true, apply action rule entry #0.
• Rule #2: Test the matching rule entry #2. If true, apply action rule entry #1.
• Matching rule entry #3 is configured for the counter function. Action entry #3 is used to hold the corresponding
count value.
• Rule #4: Test the matching rule entries #4, 5 and 6. If all are true, apply action rule entry #4.
• Rule #6: Test the matching rule entry #6. If true, apply action rule entry #5.
• Rule #8: Test the matching rule entries #7 and 11. If both are true, apply action rule entry #9.
No more than one action can be taken for any packet. If the matching conditions are true for multiple RuleSets, then the
corresponding FRN field with the lowest value (highest priority) determines the action to be taken.
Note that processing entries #0 and 1 produce an OR function: action #0 is taken if RuleSet #0 or RuleSet #1 is true.
Notice that processing entries #4 and 6 have overlapping RuleSets, but different FRNs. This can be summarized as:
If match #4, 5 and 6 are all true, then apply action #4,
Else if match #6 is true, then apply action #5.
Table 4-16 summarizes the available matching options. The MD and ENB fields are used to select the desired matching
option. More configuration details are given in the following section.
FIGURE 4-6: ACL STRUCTURE AND EXAMPLE RULE VALUES
FRN
0
0
1
4
5
9
#0
#15
#14
#13
#12
#11
#10
#9
#8
#7
#6
#5
#4
#3
#2
#1
0
1
2
none
4, 5, 6
none
6
none
7, 11
none
none
none
none
none
none
none
RuleSet
Processing Field
Entry Number
Entry #0
Entry #1
Entry #2
Entry #3
Entry #4
Entry #5
Entry #6
Entry #7
Entry #8
Entry #9
Entry #10
Entry #11
Entry #12
Entry #13
Entry #14
Entry #15
Entry #0
Entry #1
Entry #2
Entry #3
Entry #4
Entry #5
Entry #6
Entry #7
Entry #8
Entry #9
Entry #10
Entry #11
Entry #12
Entry #13
Entry #14
Entry #15
Action Rule Matching Rule
Action Field
OR
AND
When counter
function is enabled
in Matching Rule.
KSZ9897R
DS00002330D-page 44 2017-2019 Microchip Technology Inc.
4.4.16.2 Matching Rule Description
The Matching Rule consists of several parameters. The first two parameters, MD[1:0] and ENB[1:0], determine the orga-
nization of the remainder of each Matching Rule.
When MD = 00, the Matching Rule is disabled.
Details for MD = 01, ENB = 00:
The 11 bits of the aggregated bit fields from PM, P, RPE, RP and MM in the Action rule entry specify a count value for
packets matching MAC Address and TYPE in the Matching Field.
The count unit is determined by the TU bit (located in the Action rule).
TABLE 4-16: MATCHING RULE OPTIONS
MD[1:0] ENB[1:0] Matching Rule
00 XX Matching rule disabled
01
(Layer 2 matching:
MAC address,
EtherType)
00 Action field is used as count value for packets matching MAC address and
EtherType
01 Compare EtherType only
10 Compare MAC address only
11 Compare both MAC address and EtherType
10
(Layer 3 matching:
IP address)
00 Reserved
01 Compare IPv4 source and destination address (with mask)
10 Compare both source and destination IPv4 addresses (without mask)
11 Reserved
11
(Layer 4 matching:
TCP, UDP,
IP protocol)
00 Compare IPv4 protocol
01 Compare TCP source port or destination port
10 Compare UDP source port or destination port
11 Compare TCP sequence number
TABLE 4-17: ACL MATCHING RULE PARAMETERS FOR MD = 01
Parameter Description
MD[1:0] MODE
00 = Matching rule is disabled
01 = Layer 2 MAC header or counter filtering
10 = Layer 3 IP header filtering
11 = Layer 4 TCP header (and IP protocol) filtering
ENB[1:0] 00 = Count Mode. Both the MAC Address and TYPE are tested. A count value (either
time or packet count) is also incorporated. Details are given below this table.
01 = Comparison is performed only on the TYPE value
10 = Comparison is performed only on the MAC Address value
11 = Both the MAC Address and TYPE are tested
S/D Source / Destination
0 = Destination address
1 = Source address
EQ Equal / Not Equal
0 = Not Equal produces true result
1 = Equal produces true result
MAC ADDRESS[47:0] 48-bit MAC address
TYPE[15:0] EtherType
2017-2019 Microchip Technology Inc. DS00002330D-page 45
KSZ9897R
• When 0, the unit is microsecond.
• When 1, the unit is millisecond.
The CA bit (located in the Action rule) determines the algorithm used to generate an interrupt when the count terminates.
• When 0, an 11-bit counter will be loaded with the count value from the list and start counting down every unit time.
An interrupt will be generated when the timer expires, i.e. the next qualified packet has not been received within
the period specified by the value.
• When 1, the counter is incremented with every matched packet received. An interrupt is generated when the ter-
minal count is reached. The count resets thereafter. Time units are not used in this mode.
TABLE 4-18: ACL MATCHING RULE PARAMETERS FOR MD = 10
Parameter Description
MD[1:0] MODE
00 = Matching rule is disabled
01 = Layer 2 MAC header or counter filtering
10 = Layer 3 IP header filtering
11 = Layer 4 TCP header (and IP protocol) filtering
ENB[1:0] 00 = Reserved
01 = IPv4 source or destination address (with mask)
10 = IPv4 source and destination address (without mask)
11 = Reserved
S/D Source / Destination
0 = Destination address
1 = Source address
EQ Equal / Not Equal
0 = Not Equal produces true result
1 = Equal produces true result
IP ADDRESS[31:0] IPv4 address
Source or destination address (determined by S/D) when ENB = 01,
Source address when ENB = 10
IP MASK[31:0] Mask bits for the IPv4 address when ENB = 01:
0 = This bit of the address is compared
1 = This bit of the address is not compared
Destination IPv4 address when ENB = 10
TABLE 4-19: ACL MATCHING RULE PARAMETERS FOR MD = 11
Parameter Description
MD[1:0] MODE
00 = Matching rule is disabled
01 = Layer 2 MAC header or counter filtering
10 = Layer 3 IP header filtering
11 = Layer 4 TCP header (and IP protocol) filtering
ENB[1:0] 00 = IP Protocol comparison is enabled
01 = TCP source/destination port comparison is enabled
10 = UDP source/destination port comparison is enabled
11 = TCP sequence number is compared
S/D Source / Destination
0 = Destination address
1 = Source address
EQ Equal / Not Equal
0 = Not Equal produces true result
1 = Equal produces true result
KSZ9897R
DS00002330D-page 46 2017-2019 Microchip Technology Inc.
4.4.16.3 Action Rule Description
MAX PORT[15:0]
MIN PORT[15:0]
Max and Min Ports for TCP/UDP
or
TCP Sequence Number[31:0]
PC[1:0] Port Comparison
00 = Port comparison is disabled
01 = Port matches either one of MAX or MIN
10 = Match if port number is in the range of MIN to MIN
11 = Match if port number is out of the range
PRO[7:0] IPv4 protocol to be matched
FME TCP Flag Match Enable
0 = TCP FLAG matching disabled
1 = TCP FLAG matching enabled
FMASK[7:0] TCP FLAG Mask
0 = This bit of the Flag field is compared
1 = This bit of the Flag field is not compared
FLAG[7:0] TCP Flag to be matched
TABLE 4-20: ACL ACTION RULE PARAMETERS FOR NON-COUNT MODES (MD ≠ 01 OR ENB ≠
00)
Parameter Description
PM[1:0] Priority Mode
00 = ACL does not specify the packet priority. Priority is determined by standard QoS
functions.
01 = Change packet priority to P[2:0] if it is greater than QoS result.
10 = Change packet priority to P[2:0] if it is smaller than the QoS result.
11 = Always change packet priority to P[2:0].
P[2:0] Priority value
RPE Remark Priority Enable
0 = Disable priority remarking
1 = Enable priority remarking. VLAN tag priority (PCP) bits are replaced by RP[2:0].
RP[2:0] Remarked Priority value
MM[1:0] Map Mode
00 = No forwarding remapping
01 = The forwarding map in FORWARD is OR'ed with the forwarding map from the
Address Lookup Table.
10 = The forwarding map in FORWARD is AND'ed with the forwarding map from the
Address Lookup Table.
11 = The forwarding map in FORWARD replaces the forwarding map from the Address
Lookup Table.
FORWARD[N-1:0] Forwarding Ports
Bit 0 corresponds to port 1
Bit 1 corresponds to port 2, etc.
0 = Do not forward to this port
1 = Forward to this port
TABLE 4-19: ACL MATCHING RULE PARAMETERS FOR MD = 11 (CONTINUED)
Parameter Description
2017-2019 Microchip Technology Inc. DS00002330D-page 47
KSZ9897R
Figure 4-7 shows basic organization of the ACL Table. The table has 16 entries, and each entry includes a matching
field, action field and process field. Although these fields are stored together in one table, it is important to note that for
a given table entry, the Matching, Action and Process fields generally do not form an associated group. The one excep-
tion is when the Matching Rule is in Count Mode (MD = 01 and ENB = 00). In that case, the Matching and Action fields
are used in tandem.
4.4.16.4 ACL Interrupts
The ACL filtering functions do not generate interrupts. Interrupts apply only for the Count Mode (MD = 01, ENB = 00).
The Matching Rule can be configured either to timeout if the interval between packets of a specific type (MAC address
and EtherType), or when a set number of these packets are received. There is a separate interrupt for each port. Port
specific interrupt status and masks are located in the Port Interrupt Status Register and Port Interrupt Mask Register.
The top level interrupt registers for each port are in the Global Port Interrupt Status Register and Global Port Interrupt
Mask Register.
TABLE 4-21: ACL ACTION RULE PARAMETERS FOR COUNT MODE (MD = 01 OR ENB = 00)
Parameter Description
COUNT[10:0] Count value
TU Time unit for counter.
0 = Microseconds
1 = Milliseconds
CA Counter Algorithm.
0 = An 11-bit counter will be loaded with the count value from the list and start counting
down every unit time. An interrupt will be generated when the timer expires, i.e. the
next qualified packet has not been received within the period specified by the value.
1 = The counter is incremented with every matched packet received. An interrupt is
generated when the terminal count is reached. The count resets thereafter. Time units
are not used in this mode.
FIGURE 4-7: ACL TABLE FORMAT
F
R
N
MD
[1:0]
ENB
[1:0]
S
/
D
E
QMAC ADDRESS [47:0] TYPE [15:0]
F
R
N
MD
[1:0]
ENB
[1:0]
S
/
D
E
QIP ADDRESS [31:0] IP MASK [31:0]
F
R
N
MD
[1:0]
ENB
[1:0]
S
/
D
E
Q
MAX
PORT
[15:0]
MIN
PORT
[15:0]
Resvd
(5)
PC
[1:0]
PRO
[7:0]
F
M
E
FMSK
[7:0]
FLAG
[7:0]
PM
[1:0]
PM
[1:0]
PM
[1:0]
P
[2:0]
P
[2:0]
P
[2:0]
R
P
E
R
P
E
R
P
E
RP
[2:0]
RP
[2:0]
RP
[2:0]
MM
[1:0]
MM
[1:0]
MM
[1:0]
FORWARD
[# ports]
FORWARD
[# ports]
FORWARD
[# ports]
RULESET
[15:0]
RULESET
[15:0]
RULESET
[15:0]
ENTRY # 0
ENTRY # 2
ENTRY # 1
ENTRY # 4
ENTRY # 3
ENTRY # 5
ENTRY # 6
ENTRY # 8
ENTRY # 7
ENTRY # 9
ENTRY # 11
ENTRY # 10
ENTRY # 12
ENTRY # 13
ENTRY # 15
ENTRY # 14
MATCHING Rule ACTION Rule PROCESS
Field
PROCESS
Field
MD = 01
ENB = 00
MD = 10
MD = 11
F
R
N
MD
[1:0]
ENB
[1:0]
S
/
D
E
QMAC ADDRESS [47:0] TYPE [15:0]
COUNT[10:0] TU
RULESET
[15:0]
MD = 01
ENB != 00
CA unu
sed
KSZ9897R
DS00002330D-page 48 2017-2019 Microchip Technology Inc.
4.4.16.5 ACL Registers
Table 4-22 provides a list of ACL related registers.
4.5 NAND Tree Support
The KSZ9897R provides parametric NAND tree support for fault detection between chip I/Os and board. The NAND tree
is a chain of nested NAND gates in which each KSZ9897R digital I/O (NAND tree input) pin is an input to one NAND
gate along the chain. At the end of the chain, the INTRP_N pin provides the output for the last NAND gate.
The NAND tree test process includes:
• Enabling NAND tree mode
• Pulling all NAND tree input pins high
• Driving low each NAND tree input pin sequentially per the NAND tree pin order, starting with the first row of
Table 4-23.
• Checking the NAND tree output to ensure there is a toggle high-to-low or low-to-high for each NAND tree input
driven low.
TABLE 4-22: ACL REGISTERS
Registers Description
Port Interrupt Status Register,
Port Interrupt Mask Register
ACL interrupt
Port ACL Access 0 Register through
Port ACL Access F Register,
Port ACL Byte Enable MSB Register,
Port ACL Byte Enable LSB Register,
Port ACL Access Control 0 Register
ACL Table access
Port Priority Control Register Priority classification
Port Authentication Control Register ACL enable
TABLE 4-23: NAND TREE TEST PIN ORDER
NAND Tree Sequence Pin Number Pin Name NAND Tree Description
148TX_CLK6/REFCLKI6 Input
249TX_EN6/TX_CTL6 Input
350TX_ER6 Input
451COL6 Input
552TXD6_3 Input
653TXD6_2 Input
754TXD6_1 Input
855TXD6_0 Input
957RX_CLK6/REFCLKO6 Input
10 58 RX_DV6/CRS_DV6/RX_CTL6 Input
11 59 RX_ER6 Input
12 60 CRS6 Input
13 62 RXD6_3 Input
14 63 RXD6_2 Input
15 64 RXD6_1 Input
16 65 RXD6_0 Input
17 66 TX_CLK7/REFCLKI7 Input
18 67 TX_EN7/TX_CTL7 Input
2017-2019 Microchip Technology Inc. DS00002330D-page 49
KSZ9897R
The following procedure can be used to check for faults on the KSZ9897R digital I/O pin connections to the board:
1. Enable NAND tree mode via the LED2_1, LED2_0, and LED4_0 configuration strap pins option.
2. Use board logic to drive all KSZ9897R NAND tree input pins high and verify that the INTRP_N pin output is high.
3. Use board logic to drive each NAND tree input pin, per the NAND Tree pin order, as follows:
a) Toggle the first pin in the NAND tree sequence (TX_CLK6/REFCLKI6) from high to low, and verify the
INTRP_N pin switches from high to low to indicate that the first pin is connected properly.
b) Leave the first pin (TX_CLK6/REFCLKI6) low.
c) Toggle the second pin in the NAND tree sequence (TX_EN6/TX_CTL6) from high to low, and verify the
INTRP_N pin switches from low to high to indicate that the second pin is connected properly.
d) Leave the first pin (TX_CLK6/REFCLKI6) and the second pin (TX_EN6/TX_CTL6) low.
e) Toggle the third pin in the NAND tree sequence (TX_ER6) from high to low, and verify the INTRP_N pin
switches from high to low to indicate that the third pin is connected properly.
f) Continue with this sequence until all KSZ9897R NAND tree input pins have been toggled.
19 68 TX_ER7 Input
20 69 COL7 Input
21 70 TXD7_3 Input
22 71 TXD7_2 Input
23 72 TXD7_1 Input
24 73 TXD7_0 Input
25 75 RX_CLK7/REFCLKO7 Input
26 76 RX_DV7/CRS_DV7/RX_CTL7 Input
27 78 RX_ER7 Input
28 79 CRS7 Input
29 80 RXD7_3 Input
30 81 RXD7_2 Input
31 82 RXD7_1 Input
32 83 RXD7_0 Input
33 85 LED4_0 Input
34 86 LED4_1 Input
35 88 LED3_0 Input
36 89 LED3_1 Input
37 90 NC Input
38 91 LED2_0 Input
39 92 LED2_1 Input
40 93 PME_N Input
41 96 RESET_N Input
42 97 SDO Input
43 98 SDI/SDA/MDIO Input
44 100 SCS_N Input
45 101 SCL/MDC Input
46 102 LED5_0 Input
47 103 LED5_1 Input
48 105 LED1_0 Input
49 106 LED1_1 Input
50 94 INTRP_N Output
TABLE 4-23: NAND TREE TEST PIN ORDER (CONTINUED)
NAND Tree Sequence Pin Number Pin Name NAND Tree Description
KSZ9897R
DS00002330D-page 50 2017-2019 Microchip Technology Inc.
Each KSZ9897R NAND tree input pin must cause the INTRP_N output pin to toggle high-to-low or low-to-high to indi-
cate a good connection. If the INTRP_N pin fails to toggle when the KSZ9897R input pin toggles from high to low, the
input pin has a fault.
4.6 Clocking
4.6.1 PRIMARY CLOCK
The device requires a 25MHz reference clock input at the XI pin. This clock is internally multiplied up and used to clock
all of the internal logic and switching functions. It is also normally used as to clock the PHY transmit paths. This clock
may be supplied by connecting a crystal between the XI and XO pins (and appropriate load capacitors to ground). Alter-
natively, an external CMOS clock signal may drive XI, while XO is left unconnected. The XI/XO block is powered from
AVDDH.
4.6.2 MAC INTERFACE CLOCKS
The MII interface is clocked asymmetrically, with the PHY device driving both the RX_CLKx receive clock and the TX_-
CLKx transmit clock to the MAC device. Each MII port may be configured at reset by a strapping option to take the role
of either the PHY or the MAC. RX_CLKx and TX_CLKx are therefore either both inputs or both outputs, depending on
the MII mode.
The RMII interface uses a single 50MHz clock. This REFCLK may be sourced either from the KSZ9897R or from the
connected device. A strapping option is used to select the mode for each port. “Normal Mode” is the mode where the
other device supplies the clock, and the clock is an input to the REFCLKIx pin of the device. “Clock Mode” is the mode
where the KSZ9897R generates the 50MHz clock on the REFCLKOx pin.
The RGMII interface employs source synchronous clocking, so it is symmetrical and does not require a mode selection.
An output clock is generated on the RX_CLKx pin, while an input clock is received on the TX_CLKx pin. The clock
speed scales with the interface data rate - either 10, 100 or 1000 Mbps. A strapping option is used to select between
the 100 and 1000 Mbps speeds. If the 10 Mbps rate is required, then a register setting is used to set that speed.
The MAC interfaces are powered from VDDIO.
4.6.3 SERIAL MANAGEMENT INTERFACE CLOCK
Whether configured to be SPI, I2C or MIIM, the KSZ9897R is always a slave and receives the clock as an input. The
serial management interface is powered from VDDIO.
4.6.4 CLKO_25_125
An output clock, derived from the local 25MHz reference at XI, is provided on the CLKO_25_125 pin. The output fre-
quency choices are 25MHz (default) and 125MHz. If not needed, this output clock can also be disabled. CLKO_25_125
is controlled via the Output Clock Control Register, and is powered from VDDIO.
Note: Refer to Section 3.2.1, "Configuration Straps," on page 16 for additional information on using configuration
straps.
2017-2019 Microchip Technology Inc. DS00002330D-page 51
KSZ9897R
4.7 Power
The KSZ9897R requires two to three supply voltages. The device core operates from a 1.2V supply (DVDDL and
AVDDL). The PHY transceivers and XI/XO crystal/clock interface operate from a 2.5V supply (AVDDH). The digital I/
O's can be operated from 1.8V, 2.5V or 3.3V (VDDIO). The digital I/Os powered from VDDIO include RGMII, RMII, MII,
SPI, I2C, MIIM, LED, RESET_N, PME_N, INTRP_N and CLKO_25_125. An example power connection diagram can be
seen in Figure 4-8.
4.8 Power Management
The device supports enhanced power management features in a low-power state with energy detection to ensure low-
power dissipation during device idle periods. There are three operation modes under the power management function
which are implemented globally (i.e., applying to all ports):
• Normal Operation Mode
• Energy Detect Mode
• Global Soft Power Down Mode
Table 4-24 summarizes all internal function blocks status under the three power-management operation modes.
FIGURE 4-8: POWER CONNECTION DIAGRAM
0.1µF
22µF
0.1µF
VDDIO
VDDIO
VDDIO
AVDDH
AVDDH
AVDDH
AVDDH
AVDDH
AVDDH
AVDDH
DVDDL
AVDDL
AVDDL
AVDDL
AVDDL
AVDDL
AVDDL
AVDDL
AVDDL
AVDDL
DVDDL
DVDDL
DVDDL
DVDDL
DVDDL
DVDDL
DVDDL
GND
GND
GND
GND
GND
GND
GND
(exposed pad)
+1.8V, 2.5V or 3.3V
+2.5V
+1.2V
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
22µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
22µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
0.1µF
10µF
KSZ9897R
DS00002330D-page 52 2017-2019 Microchip Technology Inc.
There is one additional power saving mode that may be implemented on a per-port basis:
• Port-Based Power Down
The first three global power modes are mutually exclusive; only one mode may be selected at a time. Port-based power
down may be enabled independent of the global power mode.
4.8.1 NORMAL OPERATION MODE
At power-up, the device enters into Normal operation mode. It is also selected via bits [4:3] = 00 in the Power Down
Control 0 Register. When the device is in normal operation mode, all PLL clocks are running, PHYs and MACs are on,
and the CPU is ready to read or write the device registers through the serial interface (SPI, I2C or MIIM).
During normal operation mode, the host processor can change the power management mode bits in the Power Down
Control 0 Register to transition to any of the other power management modes.
4.8.2 ENERGY-DETECT MODE
Energy-detect mode, also known as energy-detect power down (EDPD) mode, is enabled by setting bits [4:3] to 01 in
the Power Down Control 0 Register. Energy-detect mode provides a mechanism to save power when the device is not
connected to an active link partner. Auto-negotiation must be enabled when in energy-detect mode.
Energy-detect mode consists of two states, normal-power state and low-power state. When the device is in this mode,
it will monitor the cable energy. If there is no energy on the cable for a time longer than a pre-configured value, the device
will go into the low-power state. While in low-power state, the device reduces power consumption by disabling all cir-
cuitry except the energy-detect circuitry of the receiver, which consumes minimal power. When the device is in the low-
power state, it will transmit link pulses at long intervals, with a very low duty cycle. At the same time, it continuously
monitors for energy on the cable. Once energy is detected from the cable and is present for a time longer than 100ns,
the device will enter the normal-power state.
4.8.3 GLOBAL SOFT POWER-DOWN MODE
Soft power-down mode is used to power down the device when it is not in use after power-up. This mode disables all
internal functions except for the serial (SPI or I2C) management interface.
When soft power-down mode is exited, all registers are reset to their default values, and all configuration strap pins are
sampled to set the device settings.
4.8.4 PORT-BASED POWER DOWN
Unused ports may be powered down individually to save power.
4.8.5 WAKE ON LAN (WOL)
Wake on LAN allows a computer to be turned on or woken up by a network message. The message is usually sent by
a program executed on another computer on the same local area network. Wake-up frame events are used to awaken
the system whenever meaningful data is presented to the system over the network. Examples of meaningful data
include the reception of a Magic Packet, a management request from a remote administrator, or simply network traffic
directly targeted to the local system. The device can be programmed to notify the host of the Wake-Up frame detection
with the assertion of the power management event signal (PME_N).
The device’s MACs support the detection of the following Wake-Up events:
• Detection of energy signal over a pre-configured value
• Detection of a linkup in the network link state
TABLE 4-24: MDI/MDI-X PIN DEFINITIONS
Functional Blocks
Power Management Operation Modes
Normal Mode Energy Detect Mode Soft Power Down Mode
Internal PLL Clock Enabled Disabled Disabled
TX/RX PHYs Enabled Energy Detect at RX Disabled
MACs Enabled Disabled Disabled
Host Interface Enabled Disabled Disabled
2017-2019 Microchip Technology Inc. DS00002330D-page 53
KSZ9897R
• Receipt of a Magic Packet
There are also other types of Wake-Up events that are not listed here as manufacturers may choose to implement these
in their own way.
4.8.5.1 Direction of Energy
The energy is detected from the cable and is continuously presented for a time longer than pre-configured value, espe-
cially when this energy change may impact the level at which the system should re-enter to the normal power state.
4.8.5.2 Direction of Link-up
Link status wake events are useful to indicate a linkup in the network's connectivity status.
4.8.5.3 Magic PacketTM
The Magic Packet is a broadcast frame containing anywhere within its payload 6 bytes of all 1s (FF FF FF FF FF FF)
followed by sixteen repetitions of the target computer's 48-bit DA MAC address. Since the magic packet is only scanned
for the above string, and not actually parsed by a full protocol stack, it may be sent as any network- and transport-layer
protocol.
Magic Packet technology is used to remotely wake up a sleeping or powered-off PC on a LAN. This is accomplished by
sending a specific packet of information, called a Magic Packet frame, to a node on the network. When a PC capable
of receiving the specific frame goes to sleep, it enables the Magic Packet RX mode in the LAN controller, and when the
LAN controller receives a Magic Packet frame, it will alert the system to wake up. Once the device has been enabled
for Magic Packet Detection, it scans all incoming frames addressed to the node for a specific data sequence, which
indicates to the controller this is a Magic Packet frame.
A Magic Packet frame must also meet the basic requirements for the LAN technology chosen, such as Source Address
(SA), Destination Address (DA), which may be the receiving station's IEEE MAC address, or a multicast or broadcast
address and CRC. The specific sequence consists of 16 duplications of the MAC address of this node, with no breaks
or interruptions. This sequence can be located anywhere within the packet, but must be preceded by a synchronization
stream. The synchronization stream is defined as 6 bytes of 0xFF. The device will also accept a broadcast frame, as
long as the 16 duplications of the IEEE address match the address of the machine to be awakened.
4.8.5.4 Interrupt Generation on Power Management Related Events
There are two ways an interrupt can be generated to the host whenever a power management related event takes place.
The resulting interrupts are via the PME_N signal pin or via the INTRP_N signal pin.
4.9 Management Interface
The management interface may be used by an external host processor to read and write the device’s registers. This
interface has three available modes of operation: SPI, I2C or MIIM. The interface mode is selected at the deassertion
of reset by a strapping option (refer to Section 3.2.1, "Configuration Straps," on page 16 for additional information).
Of the three interface options, SPI provides the highest performance, while MIIM performance is the lowest. Most impor-
tantly, MIIM provides access to the PHY control and status registers, but not to any of the switch registers. The vast
majority of applications therefore can use SPI or I2C, but not MIIM.
Register access is also available through the high-performance in-band management interface as described in Section
4.10, "In-Band Management," on page 57.
4.9.1 SPI SLAVE BUS
The KSZ9897R supports a slave mode SPI interface that provides complete access to all device registers via an SPI
master device. The SPI master device supplies the clock (SCL), select (SCS_N), and serial input data (SDI). Serial out-
put data (SDO) is driven by the KSZ9897R.
SCL is expected to stay low when SPI operation is idle. SPI operations start with the falling edge of SCS_N and end with
the rising edge of SCS_N. A single read or write access consists of a 27-bit command/address phase, then a 5-bit turn-
around (TA) phase, then an 8-bit data phase. For burst read or write access, SCS_N is held low while SCL continues to
toggle. For every 8 cycles of SCL, the device will increment the address counter, and the corresponding data byte will
be transferred on SDI or SDO in succession.
All commands, addresses and data are transferred most significant bit first. Input data on SDI is latched on the rising
edge of clock SCL. Output data on SDO is clocked on the falling edge of SCL.
KSZ9897R
DS00002330D-page 54 2017-2019 Microchip Technology Inc.
As shown in Figure 4-25, there are two commands: register read and register write. Figure 4-9 and Figure 4-10 show
the timing for these two operations.
Note 4-8 TA bits are turn-around bits. They are “don't care” bits.
4.9.2 I2C BUS
The management interface may be configured to be an I2C slave. In this mode, an I2C master has complete program-
ming access to the device's internal control and status registers, including all MIB counters, address lookup tables,
VLAN table and ACL table.
The 7-bit device address is fixed as 1011_111. Because of the fixed address, only one KSZ9897R may be on the I2C
bus at a time. The R/W control bit is then appended as the least significant bit to form these 8-bit address/control words:
1011_1110 <write>
1011_1111 <read>
TABLE 4-25: REGISTER ACCESS USING THE SPI INTERFACE
SPI Operation
Command/Address Phase (SDI pin) TA bits
(Note 4-8)
Data Phase
(SDO or SDI pins)
Command Register Address
Register Read 011 A23 A22 A21 A20 … A7 A6 A5 A4 A3 A2 A1 A0 XXXXX D7 D6 D5 D4 D3 D2 D1 D0
Register Write 010 A23 A22 A21 A20 … A7 A6 A5 A4 A3 A2 A1 A0 XXXXX D7 D6 D5 D4 D3 D2 D1 D0
Note: The actual device address space is 16 bits (A15 - A0), so the values of address bits A23 - A16 in the SPI
command/address phase are “don't care”.
FIGURE 4-9: SPI REGISTER READ OPERATION
FIGURE 4-10: SPI REGISTER WRITE OPERATION
SCS_N
SCL
SDI
(MOSI)
SDO
(MISO)
Read
Command Read Address TurnAround
D0D1D2D3D4D5D6D7
TATATA TATAA1 A0A2A3A4A5A6A7A9A10A11 A8A12A13A14A15A17 A16A18011A20 A19A22 A21A23
1234567891011121314151617 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
Read Data
SCS_N
SCL
SDI
(MOSI)
SDO
(MISO)
Write
Command Write Address Write DataTurnAround
1234567891011121314151617 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
D0D1D2D3D4D5D6D7
TATATA TATAA1 A0A2A3A4A5A6A7A9A10A11 A8A12A13A14A15A17 A16A18001A20 A19A22 A21A23
2017-2019 Microchip Technology Inc. DS00002330D-page 55
KSZ9897R
The internal registers and tables of the device are accessed using 16-bit addressing and 8-bit data. The access formats
are as follows:
4.9.3 MII MANAGEMENT (MIIM) INTERFACE
The device supports the IEEE 802.3 MII management interface, also known as the management data input/output
(MDIO) interface. This interface allows upper-layer devices to monitor and control the states of the KSZ9897R PHY
blocks, but it does not provide access to the switch registers. An external device with MDC/MDIO capability can read
the PHY status or configure the PHY settings. Details on the MIIM interface can be found in Clauses 22 and 45 of the
IEEE 802.3 Specification.
Use of MIIM conflicts with use of the In-Band Management interface. These interfaces cannot be used simultaneously.
The MIIM interface consists of the following:
• A physical connection that uses a data signal (MDIO) and a clock signal (MDC) for communication between an
external controller and the KSZ9897R. Note that the MDIO signal is open-drain.
• A specific protocol that operates across the two signal physical connection that allows an external controller to
communicate with the internal PHY devices.
• Access to a set of standard, vendor-specific and extended (MMD) 16-bit registers. These registers are also
directly accessible via the SPI and I2C interface options.
FIGURE 4-11: SINGLE BYTE REGISTER WRITE
FIGURE 4-12: SINGLE BYTE REGISTER READ
FIGURE 4-13: BURST REGISTER WRITE
FIGURE 4-14: BURST REGISTER READ
KSZ9897R
DS00002330D-page 56 2017-2019 Microchip Technology Inc.
The MIIM Interface can operate up to a maximum clock speed of 5MHz. Access is limited to only the registers in the
PHY blocks of ports 1 through 5. Tab l e 4-26 summarizes the MII management interface frame format.
The MIIM PHY address to PHY port mapping is as follows:
• PHY Address 1h to PHY port 1
• PHY Address 2h to PHY port 2
• PHY Address 3h to PHY port 3
• PHY Address 4h to PHY port 4
• PHY Address 5h to PHY port 5
The MIIM register address space consists of two distinct areas.
•Standard MIIM Registers (Direct)
•MDIO Manageable Device (MMD) Registers (Indirect)
4.9.3.1 Standard MIIM Registers (Direct)
Standard registers provide direct read/write access to a 32-register address space, as defined in Clause 22 of the IEEE
802.3 Specification. Within this address space, the first 16 registers (Registers 0h to Fh) are defined according to the
IEEE specification, while the remaining 16 registers (Registers 10h to 1Fh) are defined specific to the PHY vendor.
The KSZ9897R supports the standard registers listed in Ta ble 4 - 2 7 for each PHY port. Each 16-bit MIIM Standard Reg-
ister Address maps to two corresponding 8-bit Port N Register Addresses. The register bit map and description are
located at the 8-bit Port N Register Addresses.
TABLE 4-26: MII MANAGEMENT INTERFACE FRAME FORMAT
Operation
Mode
Preamble
(32-bit)
Start of
Frame
(2-bit)
Operation
Code
(2-bit)
PHY
Address
(5-bit)
Register
Address
(5-bit)
Turn
Around
(2-bit)
Register
Data
(16-bit)
Idle
Read All 1s 01 10 A[4:0] Reg[4:0] Z0 D[15:0] Z
Write All 1s 01 01 A[4:0] Reg[4:0] 10 D[15:0] Z
TABLE 4-27: STANDARD MIIM REGISTERS
MIIM Standard
Register Address
(hex)
Port N
Register Address
(hex)
Description
IEEE-Defined Registers
0h 0xN100 - 0xN101 PHY Basic Control Register
1h 0xN102 - 0xN103 PHY Basic Status Register
2h 0xN104 - 0xN105 PHY ID High Register
3h 0xN106 - 0xN107 PHY ID Low Register
4h 0xN108 - 0xN109 PHY Auto-Negotiation Advertisement Register
5h 0xN10A - 0xN10B PHY Auto-Negotiation Link Partner Ability Register
6h 0xN10C - 0xN10D PHY Auto-Negotiation Expansion Status Register
7h 0xN10E - 0xN10F PHY Auto-Negotiation Next Page Register
8h 0xN110 - 0xN111 PHY Auto-Negotiation Link Partner Next Page Ability Register
9h 0xN112 - 0xN113 PHY 1000BASE-T Control Register
Ah 0xN114 - 0xN115 PHY 1000BASE-T Status Register
Bh-Ch - RESERVED
Dh 0xN11A - 0xN11B PHY MMD Setup Register
Eh 0xN11C - 0xN11D PHY MMD Data Register
Fh 0xN11E - 0xN11F PHY Extended Status Register
2017-2019 Microchip Technology Inc. DS00002330D-page 57
KSZ9897R
4.9.3.2 MDIO Manageable Device (MMD) Registers (Indirect)
The MIIM interface provides indirect access to a set of MMD registers as defined in Section 5.4, "MDIO Manageable
Device (MMD) Registers (Indirect)," on page 166.
4.10 In-Band Management
The in-band management access (IBA) is a feature that provides full register read and write access via any one of the
seven data ports. Port 7 is the default IBA port. The in-band feature is enabled or disabled by a strapping option at
power-up and reset. To use a different port instead of port 7 for IBA, the SPI or I2C interface or IBA must be used to
write to a control register. IBA may not be used on more than one port at a time, but the IBA port can still be used for
sending and receiving non-IBA traffic.
In-band management frames are processed differently from normal network frames. They are recognized as special
frames, so address and VID lookup, VLAN tagging, source address filtering, un-tag discard, tagged frame drop, etc. are
not applied to them. Received in-band management frames are never forwarded to the switch fabric or to any other port.
The In-Band Management (IBA) Control Register is used to enable and control the IBA feature and to specify one of the
seven ports as the IBA port.
The IBA frame format is shown in Figure 4-15. The layer 2 portion of the IBA frame contains normal destination address
(DA) and source address (SA) fields. The DA of the frames are defined to be the switch MAC address (default 00-10-
A1-FF-FF-FF), and the SA is the MAC address of the source device. The DA and SA will be swapped in the response
frame. A special 4-byte IBA tag follows the SA. This is then followed by the 2-byte EtherType/Length field that serves to
identify this as an IBA frame.
Only one IBA frame can be processed at a time. Any subsequent IBA frames received by the device will be dropped
unless the most recent response frame has been fully transmitted.
There are six types of read/write commands: READ, WRITE, WAIT on 0, WAIT on 1, MODIFY to 0 and MODIFY to 1.
The minimum IBA read or write size is 32-bits. There is no IBA option for 8-bit or 16-bit transfers.
The burst commands offer fast and bundled data return, up to the capacity of the IBA frame buffer. There are two types
of operations in burst command: READ burst and Write burst.
Vendor-Specific Registers
10h - RESERVED
11h 0xN122 - 0xN123 PHY Remote Loopback Register
12h 0xN124 - 0xN125 PHY LinkMD Register
13h 0xN126 - 0xN127 PHY Digital PMA/PCS Status Register
14h - RESERVED
15h 0xN12A - 0xN12B Port RXER Count Register
16h-1Ah - RESERVED
1Bh 0xN136 - 0xN137 Port Interrupt Control / Status Register
1Ch 0xN138 - 0xN139 PHY Auto MDI / MDI-X Register
1Dh-1Eh - RESERVED
1Fh 0xN13E - 0xN13F PHY Control Register
TABLE 4-27: STANDARD MIIM REGISTERS (CONTINUED)
MIIM Standard
Register Address
(hex)
Port N
Register Address
(hex)
Description
KSZ9897R
DS00002330D-page 58 2017-2019 Microchip Technology Inc.
4.11 MAC Interface (RGMII/MII/RMII Port 6-7)
Strapping options are used to individually select any of these MAC interface options for ports 6 and 7:
•Media Independent Interface (MII): Supports 100 and 10 Mbps data rates
•Reduced Media Independent Interface (RMII): Supports 100 and 10 Mbps data rates
•Reduced Gigabit Media Independent Interface (RGMII): Supports 1000, 100 and 10 Mbps data rates
Note that the signals on the KSZ9897R MAC interfaces are named as they would be for a PHY: the TX direction is into
the KSZ9897R, while the RX direction is out of the KSZ9897R, as if to a host processor with integrated MAC. Signal
connection to such a “MAC” device is TX-to-TX, and RX-to-RX.
An external PHY (such as the Microchip KSZ9031RNX) may be connected to either port, but in that case the signal
connection will be RX-to-TX, and TX-to-RX.
The RGMII/MII/RMII interfaces are powered by the VDDIO power supply.
FIGURE 4-15: IN-BAND MANAGEMENT FRAME FORMAT
LSB 0 1 654327MSB
PREAMBLE
SFD
7 Bytes
1 Byte
DES. ADDRESS
6 Bytes
IBAF TAG TYPE
2 Bytes
SRC. ADDRESS
6 Bytes
IBAF TAG CONTROL INFO.
2 Bytes
2 Bytes
46 bit addressi/g u/l i/g=0, Individual Address; i/g=1, Group Address
u/l=0, Globally Admin. Addr. u/l=1, Locally Admin. Addr.
2 Bytes
MAC LENGTH/TYPE
2 Bytes
TPID
10011000
00000000
00000000
00000000
0x9800 for 98xx Family
Reserved
010xxxxx
11111110
Sequence #
pcfi 3 bits of Priority, 1 bit of Canonical Format Indicator = 0
4 bits of mode, 8 bits of sequence #
IBAF TAG
2 Bytes
ACCESS COMMAND
4 Bytes
DATA
4 Bytes
4 Bytes
4 Bytes
4 Bytes
4 Bytes
4 Bytes
4 Bytes
4 Bytes
ACCESS CODE
4 Bytes
EXTENSION
Variable
FRAME CHECK SEQUENCE
4 Bytes
PAD
Variable
IBAF
Layer 3
Data
–
MAC
DATA
0eeeeAAAAAAAA
OP.
Code
Bits Transmitted from Left to Right
mode
ACCESS FORMAT
Reserved 0x0001 = Rd/Wr,
0x0002 = Burst
ACCESS COMMAND/DATA
ACCESS COMMAND/DATA
ACCESS COMMAND/DATA
ACCESS COMMAND/DATA
DATA
DATA
DATA
DATA
AAAAAAAAAAAAAAAA
COMMAND = 0x0001 (Rd/Wr)
0000000000000
OP.
Code
00000000000aaaaa
COMMAND = 0x0002 (Dump)
0xdddddddd
COMMAND = 0x0001 (Rd/Wr)
3 bits of OP code:
001 = READ, 010 = WRITE,
100 = WAIT on 0, 101 = WAIT on 1,
110 = MODIFY to 0, 111 = MODIFY to 1,
000 = end of command list
eeee = Byte enable
AAAA..AAAA[23:0] = Register Address
000000AAAAAAAADir
AAAAAAAAAAAAAAAA
COMMAND = 0x0002 (Burst)
2 bits of Direction:
01 = READ, 10 = WRITE,
AAAA..AAAA[23:0] = Starting Register Address
0xmaskmask
COMMAND = 0x0001 (WAIT/
MODIFY)
Mask (1) = bit in reg. to be tested for OP code 100/101
OR
set (1) = bit to be set to 0/1 based on OP code 110/111
dddddddd = data written/read
26 0's and 6 bits of burst count
COMMAND = 0x0002 (Burst)
2017-2019 Microchip Technology Inc. DS00002330D-page 59
KSZ9897R
4.11.1 MEDIA INDEPENDENT INTERFACE (MII)
The media independent interface (MII) is specified in Clause 22 of the IEEE 802.3 standard. It provides a common inter-
face between PHY layer and MAC layer devices. The data interface is 4-bits wide and runs at one quarter the network
bit rate; either 2.5MHz in 10BASE-Te or 25MHz in 100BASE-TX (not encoded). Additional signals on the transmit side
indicate when data is valid or when an error occurs during transmission. Similarly, the receive side provides signals that
convey when the data is valid and without physical layer errors. For half duplex operation, the COL signal indicates if a
collision has occurred during transmission.
Each MII interface operates in either PHY Mode or MAC Mode. Select PHY Mode when the port is connected to a pro-
cessor or other device with a MAC function; select MAC Mode when connecting to an external PHY. Note that the direc-
tion of the TX_CLKx, RX_CLKx, COLx and CRSx signals is affected by the PHY mode or MAC mode setting, while
other MII signals do not change direction.
MII mode is selected at reset by a configuration strap option on pins RXD6_3 and RXD6_2 for port 6, and pins RXD7_3
and RXD7_2 for port 7. The Speed strapping option (on pin RXD6_0 for port 6 and RXD7_0 for port 7) should be set for
100/10 Mbps Mode. PHY Mode or MAC Mode is selected by strapping option on pins RXD6_1 (port 6) and RXD7_1
(port 7). Refer to Section 3.2.1, "Configuration Straps," on page 16 for additional information.
The interface contains two distinct groups of signals, one for transmission and the other for reception. Ta b l e 4 - 28 and
Table 4-29 describe the signals used by the MII interface to connect to an external MAC or to an external PHY, respec-
tively.
4.11.2 REDUCED MEDIA INDEPENDENT INTERFACE (RMII)
The reduced media independent interface (RMII) specifies a low pin count interface, which is based on MII, that provides
communication with a MAC attached to the port. As with MII, RMII provides a common interface between physical layer
and MAC layer devices, or between two MAC layer devices, and has the following key characteristics:
TABLE 4-28: MII (PHY MODE) CONNECTION TO EXTERNAL MAC
MII Interface Signals Description KSZ9897R Signals in PHY Mode External MAC Device Signals
Transmit Enable TX_ENx (input) TX_EN (output)
Transit Error TX_ERx (input) TX_ER (output)
Transmit Data Bits [3:0] TXDx_[3:0] (input) TXD[3:0] (output)
Transmit Clock TX_CLKx (output) TX_CLK (input)
Collision Detection COLx (output) COL (input)
Carrier Sense CRSx (output) CRS (input)
Received Data Valid RX_DVx (output) RX_DV (input)
Receive Error RX_ERx (output) RX_ER (input)
Receive Data Bits [3:0] RXDx_[3:0] (output) RXD[3:0] (input)
Receive Clock RX_CLKx (output) RX_CLK (input)
TABLE 4-29: MII (MAC MODE) CONNECTION TO EXTERNAL PHY
MII Interface Signals Description KSZ9897R Signals in MAC Mode External PHY Device Signals
Transmit Enable RX_DVx (output) TX_EN (input)
Transit Error RX_ERx (output) TX_ER (input)
Transmit Data Bits [3:0] RXDx_[3:0] (output) TXD[3:0] (input)
Transmit Clock RX_CLKx (input) TX_CLK (output)
Collision Detection COLx (input) COL (output)
Carrier Sense CRSx (input) CRS (output)
Received Data Valid TX_ENx (input) RX_DV (output)
Receive Error TX_ERx (input) RX_ER (output)
Receive Data Bits [3:0] TXDx_[3:0] (input) RXD[3:0] (output)
Receive Clock TX_CLKx (input) RX_CLK (output)
KSZ9897R
DS00002330D-page 60 2017-2019 Microchip Technology Inc.
• Supports network data rates of either 10Mbps or 100Mbps.
• Uses a single 50MHz clock reference (provided internally or externally) for both transmit and receive data.
• Uses independent 2-bit wide transmit and receive data paths.
• Contains two distinct groups of signals: one for transmission and the other for reception.
The user selects one of the two RMII clocking modes by setting the appropriate strapping option. The clocking mode is
selected separately for ports 6 and 7.
While in RMII Normal Mode, the port will require an external 50MHz signal to be input to TX_CLKx/REFCLKIx from an
external source. This mode is selected by strapping the appropriate pin (RXD6_1 for port 6; RXD7_1 for port 7) high
during reset.
While in RMII Clock Mode, the port will output a 50MHz clock on RX_CLKx/REFCLKOx, which is derived from the
25MHz crystal or oscillator attached to the XI clock input. The TX_CLKx/REFCLKIx input is unused in this mode. This
mode is selected by strapping the appropriate pin (RXD6_1 for port 6; RXD7_1 for port 7) low during reset.
Refer to Section 3.2.1, "Configuration Straps," on page 16 for additional configuration strap information.
Table 4-30 describes the signals used by the RMII interface. Refer to the RMII specification for full details on the signal
descriptions.
A device port in RMII mode may connect to either an external MAC device (such as a host processor) or to an external
PHY; but unlike MII, RMII does not provide separate PHY and MAC modes of operation. However, it is necessary to
connect the pins properly.
TABLE 4-30: RMII SIGNAL DESCRIPTIONS
RMII Signal
Name
(per spec)
RMII Signal
(per KSZ9897R)
Pin Direction
(with respect to
PHY, KSZ9897R)
Pin Direction
(with respect to
MAC)
RMII Signal Description
REF_CLK REFCLKI6
REFCLKI7 Input Input or Output Synchronous 50MHz reference clock,
when port is in RMII Normal Mode
n/a REFCLKO6
REFCLKO7 Output Input Synchronous 50MHz reference clock,
when port is in RMII Clock Mode
TX_EN TX_EN6
TX_EN7 Input Output Transmit Enable
TXD[1:0] TXD6_[1:0]
TXD7_[1:0] Input Output Transmit Data Bit [1:0]
CRS_DV RX_DV6
RX_DV7 Output Input Carrier Sense / Receive Data Valid
RX_ER RX_ER6
RX_ER7 Output Input or
not required Receive Error
RXD[1:0] RXD6_[1:0]
RXD7_[1:0] Output Input Receive Data Bit [1:0]
TABLE 4-31: RMII CONNECTION TO EXTERNAL MAC
RMII Interface Signals Description KSZ9897R Signals External MAC Device Signals
Transmit Enable TX_ENx (input) TX_EN (output)
Transmit Data Bits [1:0] TXDx_[1:0] (input) TXD[1:0] (output)
Reference Clock REFCLKIx (input)
or REFCLKOx (output)
REF_CLK
(input or output)
Carrier Sense Data Valid RX_DVx (output) CRS_DV (input)
Receive Error RX_ERx (output) RX_ER (input)
Receive Data Bits [1:0] RXDx_[1:0] (output) RXD[1:0] (input)
2017-2019 Microchip Technology Inc. DS00002330D-page 61
KSZ9897R
4.11.3 REDUCED GIGABIT MEDIA INDEPENDENT INTERFACE (RGMII)
RGMII provides a common interface between RGMII PHYs and MACs, and has the following key characteristics:
• Pin count is reduced from 24 pins for GMII to 12 pins for RGMII.
• All speeds (10Mbps, 100Mbps and 1000Mbps) are supported at both half- and full-duplex.
• Data transmission and reception are independent and belong to separate signal groups.
• Transmit data and receive data are each four bits wide - a nibble.
In RGMII operation, the RGMII pins function as follows:
• The MAC sources the transmit reference clock, TX_CLKx, at 125MHz for 1000Mbps, 25MHz for 100Mbps, and
2.5MHz for 10Mbps.
• The PHY recovers and sources the receive reference clock, RX_CLKx, at 125MHz for 1000Mbps, 25MHz for
100Mbps, and 2.5MHz for 10Mbps.
• For 1000BASE-T, the transmit data, TXDx_[3:0], is presented on both edges of TX_CLKx, and the received data,
RXDx_[3:0], is clocked out on both edges of the recovered 125MHz clock, RX_CLKx.
• For 10BASE-T/100BASE-TX, the MAC holds TX_CTLx low until both the PHY and MAC operate at the same
speed. During the speed transition, the receive clock is stretched on either a positive of neagative pulse to ensure
that no clock glitch is presented to the MAC.
•TX_ERx and RX_ERx are combined with TX_ENx and RX_DVx, respectively, to form TX_CTLx and RX_CTLx.
These two RGMII control signals are valid at the falling clock edge.
After power-up or reset, the device is configured to RGMII mode if the appropriate configuration strap pins are set to
one of the RGMII mode capability options. Refer to Section 3.2.1, "Configuration Straps," on page 16 for available
options. Note that there is no mechanism for the RGMII interface to adapt its speed automatically to the speed of the
connected RGMII device. A configuration strap option sets the speed of each RGMII interface at power-up to either
1000Mbps or 100Mbps. For each port, a control register can override the configuration strap option and set the RGMII
speed to either 1000, 100 or 10Mbps. If a PHY is connected to an RGMII port, it should be ensured that the PHY link
speed is fixed in order to avoid a mismatch to the RGMII speed.
The device provides the option to add a minimum of 1.5ns internal delay to either TX_CLKx or RX_CLKx, via the RGMII
Internal Delay control bits in the XMII Port Control 1 Register. This can reduce or eliminate the need to add trace delay
to the clock signals on the printed circuit board. RGMII_ID_ig enables delay on TX_CLKx, and the default is off.
RGMII_ID_eg enables delay on RX_CLKx, and the default is on. Users should also be aware of any internal clock delay
that may be added by the connected RGMII device.
TABLE 4-32: RMII CONNECTION TO EXTERNAL PHY
RMII Interface Signals Description KSZ9897R Signals External PHY Device Signals
Transmit Enable RX_DVx (output) TX_EN (input)
Transmit Data Bits [1:0] RXDx_[1:0] (output) TXD[1:0] (input)
Reference Clock REFCLKIx (input)
or REFCLKOx (output)
REF_CLK
(input or output)
Carrier Sense Data Valid TX_ENx (input) CRS_DV (output)
Receive Error No connection RX_ER (output)
Receive Data Bits [1:0] TXDx_[1:0] (input) RXD[1:0] (output)
KSZ9897R
DS00002330D-page 62 2017-2019 Microchip Technology Inc.
TABLE 4-33: RGMII SIGNAL DESCRIPTIONS
RGMII Signal
Name
(per spec)
RGMII Signal
(per KSZ9897R)
Pin Direction
(with respect to
PHY, KSZ9897R)
Pin Direction
(with respect to
MAC)
RGMII Signal Description
TXC TX_CLK6
TX_CLK7 Input Output
Transmit Reference Clock
(125MHz for 1000Mbps, 25MHz for
100Mbps, 2.5MHz for 10Mbps)
TX_CTL TX_CTL6
TX_CTL7 Input Output Transmit Control
TXD[3:0] TXD6_[3:0]
TXD7_[3:0] Input Output Transmit Data [3:0]
RXC RX_CLK6
RX_CLK7 Output Input
Receive Reference Clock
(125MHz for 1000Mbps, 25MHz for
100Mbps, 2.5MHz for 10Mbps)
RX_CTL RX_CTL6
RX_CTL7 Output Input Receive Control
RXD[3:0] RXD6_[3:0]
RXD7_[3:0] Output Input Receive Data [3:0]
2017-2019 Microchip Technology Inc. DS00002330D-page 63
KSZ9897R
5.0 DEVICE REGISTERS
The KSZ9897R has a rich set of registers for device management. The registers are accessed by the SPI or I2C inter-
faces, or by in-band management. Alternatively, the MIIM interface can be used to access the PHY registers only. The
MIIM interface cannot access the switch registers.
A 16-bit address is used to access the device registers. This address is split into three hierarchical spaces, as shown
in Figure 5-1. These three spaces are used to designate the port/channel (4-bits), function (page) of the port (4-bits),
and register of function (8-bits). The individual ports are numbered 1 through 7. In the port space, a value of 0 is used
for global registers. Address bit 15 is always 0.
FIGURE 5-1: REGISTER ADDRESS MAPPING
15 14 13 12 11 10 9 8 7 0123456
= 0x0312
0 = Global
Register Space
Function Space
(Page)
Port Space
0
Examples:
000001100010010
N = Port #
= 0xN4080NNN010000001000
Control,
Status,
Storage,
Etc.
0 = Operation
1 = I/O Interface
2 = PHY
3 = General
4 = Look-up Tables
5-F = Reserved
0 = Operation
1 = PHY
2 = Reserved
3 = RGMII / MII / RMII
4 = MAC
5 = MIB Counters
6 = ACL
7 = Reserved
8 = Ingress – Classification, Policing
9 = Egress – Shaping
A = Queue management
B = Address lookup engine
C-F = Reserved
Register
Address
Values of N
are 1 - 7
KSZ9897R
DS00002330D-page 64 2017-2019 Microchip Technology Inc.
Register addressing is by bytes, and the management interface (SPI, I2C or in-band) transfers data by bytes. Where
registers are shown as 16-bits or 32-bits, this is for descriptive purposes only. Data can always be written and read as
individual bytes and in any order.
For multi-byte registers, the data is addressed in a big-endian format, with the most significant byte at the lowest
address, and the least significant byte at the highest address, as shown in Figure 5-2.
The global and port register address maps are detailed in Table 5-1 and Ta b l e 5 -2, respectively. Table 1-3, “Register
Nomenclature,” on page 7 provides a list of register bit type notations.
The remainder of this chapter is organized as follows:
•Global Registers
•Port Registers
•Tables and MIB Counters (Access)
•MDIO Manageable Device (MMD) Registers (Indirect)
FIGURE 5-2: BYTE ORDERING
TABLE 5-1: GLOBAL REGISTER ADDRESS MAP
Address Functional Group
0x0000 - 0x00FF Global Operation Control Registers (0x0000 - 0x00FF)
0x0100 - 0x01FF Global I/O Control Registers (0x0100 - 0x01FF)
0x0200 - 0x02FF Global PHY Control and Status Registers (0x0200 - 0x02FF)
0x0300 - 0x03FF Global Switch Control Registers (0x0300 - 0x03FF)
0x0400 - 0x04FF Global Switch Look Up Engine (LUE) Control Registers (0x0400 - 0x04FF)
0x0500 - 0x0FFF RESERVED
0A0B0C0D0A0B0
16-bit register 32-bit register
0A
0B
0C
0D
0A
0B
memory
memory
address a
address a+1
address a
address a+1
address a+2
address a+3
2017-2019 Microchip Technology Inc. DS00002330D-page 65
KSZ9897R
TABLE 5-2: PORT N (1-7) REGISTER ADDRESS MAP
Address Functional Group
0xN000 - 0xN0FF Port N: Port Operation Control Registers (0xN000 - 0xN0FF)
0xN100 - 0xN1FF Port N: Port Ethernet PHY Registers (0xN100 - 0xN1FF)
0xN200 - 0xN2FF RESERVED
0xN300 - 0xN3FF Port N: Port RGMII/MII/RMII Control Registers (0xN300 - 0xN3FF)
0xN400 - 0xN4FF Port N: Port Switch MAC Control Registers (0xN400 - 0xN4FF)
0xN500 - 0xN5FF Port N: Port Switch MIB Counters Registers (0xN500 - 0xN5FF)
0xN600 - 0xN6FF Port N: Port Switch ACL Control Registers (0xN600 - 0xN6FF)
0xN700 - 0xN7FF RESERVED
0xN800 - 0xN8FF Port N: Port Switch Ingress Control Registers (0xN800 - 0xN8FF)
0xN900 - 0xN9FF Port N: Port Switch Egress Control Registers (0xN900 - 0xN9FF)
0xNA00 - 0xNAFF Port N: Port Switch Queue Management Control Registers (0xNA00 - 0xNAFF)
0xNB00 - 0xNBFF Port N: Port Switch Address Lookup Control Registers (0xNB00 - 0xNBFF)
0xNC00 - 0xNFFF RESERVED
Note: RESERVED address space must not be written under any circumstances. Failure to heed this warn-
ing may result in untoward operation and unexpected results. If it is necessary to write to registers
which contain both writable and reserved bits in the same register, the user should first read back
the reserved bits (RO or R/W), “OR” the desired settable bits with the value read, and then write
back the “ORed” value to the register.
KSZ9897R
DS00002330D-page 66 2017-2019 Microchip Technology Inc.
5.1 Global Registers
This section details the device’s global registers. For an overview of the device’s entire register map, refer to Section
5.0, "Device Registers". For details on the device’s port registers, refer to Section 5.2, "Port Registers".
5.1.1 GLOBAL OPERATION CONTROL REGISTERS (0x0000 - 0x00FF)
5.1.1.1 Global Chip ID 0 Register
5.1.1.2 Global Chip ID 1 Register
5.1.1.3 Global Chip ID 2 Register
5.1.1.4 Global Chip ID 3 Register
Address: 0x0000 Size: 8 bits
Bits Description Type Default
7:0 Fixed Value RO 0x00
Address: 0x0001 Size: 8 bits
Bits Description Type Default
7:0 Chip ID (MSB) RO 0x98
Address: 0x0002 Size: 8 bits
Bits Description Type Default
7:0 Chip ID (LSB) RO 0x97
Address: 0x0003 Size: 8 bits
Bits Description Type Default
7:4 Revision ID RO -
3:1 RESERVED RO -
0Global Software Reset
This bit dos not self-clear.
Refer to the Switch Operation Register for another reset control bit.
0 = Normal operation
1 = Resets the data path and state machines, but not register values.
R/W 0b
2017-2019 Microchip Technology Inc. DS00002330D-page 67
KSZ9897R
5.1.1.5 PME Pin Control Register
5.1.1.6 Global Interrupt Status Register
This register provides the top level interrupt status for the LUE. These interrupts are enabled in the Global Interrupt Mask
Register. For port specific interrupts, refer to the Port Interrupt Status Register.
5.1.1.7 Global Interrupt Mask Register
This register enables the interrupts in the Global Interrupt Status Register.
Address: 0x0006 Size: 8 bits
Bits Description Type Default
7:2 RESERVED RO -
1PME Pin Output Enable
0 = Disabled
1 = Enabled
R/W 0b
0PME Pin Output Polarity
0 = PME is active low
1 = PME is active high
R/W 0b
Address: 0x0010 - 0x0013 Size: 32 bits
Bits Description Type Default
31 Lookup Engine (LUE) Interrupt Status
Refer to the Address Lookup Table Interrupt Register for detailed LUE
interrupt status bits.
0 = No interrupt
1 = Interrupt request
RO 0b
30:0 RESERVED RO -
Address: 0x0014 - 0x0017 Size: 32 bits
Bits Description Type Default
31 Lookup Engine (LUE) Interrupt Mask
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
30:0 RESERVED RO -
KSZ9897R
DS00002330D-page 68 2017-2019 Microchip Technology Inc.
5.1.1.8 Global Port Interrupt Status Register
This register provides the top level interrupt status for the individual ports. These interrupts are enabled in the Global
Port Interrupt Mask Register. Refer to the Port Interrupt Status Register for detailed port interrupt status.
Address: 0x0018 - 0x001B Size: 32 bits
Bits Description Type Default
31:7 RESERVED RO -
6Port 7 Interrupt Status
0 = No interrupt
1 = Interrupt request
RO 0b
5Port 6 Interrupt Status
0 = No interrupt
1 = Interrupt request
RO 0b
4Port 5 Interrupt Status
0 = No interrupt
1 = Interrupt request
RO 0b
3Port 4 Interrupt Status
0 = No interrupt
1 = Interrupt request
RO 0b
2Port 3 Interrupt Status
0 = No interrupt
1 = Interrupt request
RO 0b
1Port 2 Interrupt Status
0 = No interrupt
1 = Interrupt request
RO 0b
0Port 1 Interrupt Status
0 = No interrupt
1 = Interrupt request
RO 0b
2017-2019 Microchip Technology Inc. DS00002330D-page 69
KSZ9897R
5.1.1.9 Global Port Interrupt Mask Register
This register enables the interrupts in the Global Port Interrupt Status Register.
Address: 0x001C - 0x001F Size: 32 bits
Bits Description Type Default
31:7 RESERVED RO -
6Port 7 Interrupt Mask
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
5Port 6 Interrupt Mask
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
4Port 5 Interrupt Mask
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
3Port 4 Interrupt Mask
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
2Port 3 Interrupt Mask
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
1Port 2 Interrupt Mask
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
0Port 1 Interrupt Mask
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
KSZ9897R
DS00002330D-page 70 2017-2019 Microchip Technology Inc.
5.1.2 GLOBAL I/O CONTROL REGISTERS (0x0100 - 0x01FF)
5.1.2.1 Serial I/O Control Register
5.1.2.2 Output Clock Control Register
Address: 0x0100 Size: 8 bits
Bits Description Type Default
7:3 RESERVED R/W 0100_0b
2MIIM Preamble Suppression
This feature affects only the MIIM (MDIO / MDC) interface. When using SPI
or I2C, this bit has no effect.
0 = Normal operation. The switch always expects the MIIM preamble.
1 = The switch will respond to MIIM commands even in the absence of a
preamble.
R/W 0b
1Automatic SPI Data Out Edge Select
When enabled, this feature automatically determines the edge of SCL that is
used to clock out the SPI data on SDO. If SCL ≥ ~25MHz, SDO data is
clocked by the rising edge of SCL. If SCL < ~25 MHz, SDO data is clocked
by the falling edge of SCL.
0 = The automatic feature is disabled, and bit 0 determines the SCL clock
edge used for SDO.
1 = The automatic feature is enabled, and bit 0 is ignored.
R/W 1b
0SPI Data Out Edge Select
When bit 1 is zero, then this bit determines the clock edge used for SPI data
out. When bit 1 is set to 1, this bit is ignored.
0 = SDO data is clocked by the falling edge of SCL
1 = SDO data is clocked by the rising edge of SCL
R/W 0b
Address: 0x0103 Size: 8 bits
Bits Description Type Default
7:5 RESERVED RO 000b
4:2 RESERVED R/W 000b
1CLKO_25_125 Output Pin Enable
0 = Disabled
1 = Enabled
R/W 1b
0CLKO_25_125 Frequency
0 = 25 MHz
1 = 125 MHz
R/W 0b
2017-2019 Microchip Technology Inc. DS00002330D-page 71
KSZ9897R
5.1.2.3 In-Band Management (IBA) Control Register
This register controls the In-Band Access (IBA) feature.
Note 5-1 The default value of this field is determined by the associated configuration strap value. Refer to
Section 3.2.1, "Configuration Straps," on page 16 for additional information.
Address: 0x0104 - 0x0107 Size: 32 bits
Bits Description Type Default
31 IBA Enable
The initial value is strapped in from the RX_DV7/CRS_DV7/RX_CTL7 pin.
0 = Disabled
1 = Enabled
R/W Note 5-1
30 IBA Destination MAC Address Match Enable
Set this bit to enable checking of the destination MAC address in received
IBA frames against the switch MAC address in the Switch MAC Address 0
Register through Switch MAC Address 5 Register. Non-matching frames are
discarded.
When not enabled, the MAC address is not checked.
R/W 0b
29 IBA Reset
Set this bit to initialize the IBA state machine. This bit is self-clearing.
R/W
SC
0b
28:24 RESERVED RO 0x00
23:22 Priority Queue for IBA response
Specifies the transmit priority queue for the IBA response frame. Typically
this value is not changed.
R/W 01b
21:19 RESERVED RO 00_0b
18:16 Port used for IBA communication
000 = Port 1
001 = Port 2
010 = Port 3
011 = Port 4
100 = Port 5
101 = Port 6
110 = Port 7
111 = Reserved
R/W 110
15:0 TPID (EtherType) value for IBA frame header R/W 0x40FE
KSZ9897R
DS00002330D-page 72 2017-2019 Microchip Technology Inc.
5.1.2.4 I/O Drive Strength Register
5.1.2.5 In-Band Management (IBA) Operation Status 1 Register
Address: 0x010D Size: 8 bits
Bits Description Type Default
7RESERVED R/W 0b
6:4 High Speed Drive Strength (24mA)
Controls drive strength of RGMII / MII / RMII (except TX_CLK / REFCLKI,
COL and CRS) and CLKO_25_125.
R/W 110b
3RESERVED R/W 0b
2:0 Low Speed Drive Strength (8mA)
Controls drive strength of TX_CLK / REFCLKI, COL, CRS, LEDs, PME_N,
INTRP_N, SDO and SDI/SDA/MDIO.
R/W 10b
Address: 0x0110 - 0x0113 Size: 32 bits
Bits Description Type Default
31 Good IBA Packet Detect
1 = A good IBA packet is received.
RO 0b
30 IBA Response Packet Transmit Done
1 = An IBA response packet is sent out.
This bit is cleared when a packet with a matching IBA tag field is received.
RO 0b
29 IBA Execution Done
1 = All the commands in one IBA packet are completely executed.
This bit is cleared when a packet with a matching IBA tag field is received.
RO 0b
28:15 RESERVED RO 0x0000
14 IBA MAC Address Mismatch Error
This bit is active only when IBA_ENABLE (In-Band Management (IBA)
Control Register, bit 30) is set.
1 = An IBA packet is received with an unmatched MAC address, unequal to
the switch’s MAC address.
This bit is cleared when a packet with a matching IBA tag field is received.
RO 0b
13 IBA Access Format Error
1 = An IBA packet with a wrong access format (not equal to 0x9800) is
received.
This bit is cleared when a packet with a matching IBA tag field is received.
RO 0b
12 IBA Access Code Error
1 = An IBA packet with an unrecognized access code is received. (Valid
access codes are 0x0001 and 0x0002.)
This bit is cleared when a packet with a matching IBA tag field is received.
RO 0b
2017-2019 Microchip Technology Inc. DS00002330D-page 73
KSZ9897R
5.1.2.6 LED Override Register
5.1.2.7 LED Output Register
11 IBA Access Command Error
1 = An IBA packet with an unrecognized command code is received.
This bit is cleared when a packet with a matching IBA tag field is received.
RO 0b
10 IBA Oversize Packet Error
1 = An oversized IBA packet is received. The maximum IBA packet size is
320 bytes, including 8-byte zeros before FCS and the 4-byte FCS. No
response packet is sent.
This bit is cleared when a packet with a matching IBA tag field is received.
RO 0b
9:7 RESERVED RO 000b
6:0 IBA Access Code Error Location
When IBA Access Command Error (bit 11) is set, these bits indicate the
address location of the wrong command code within the IBA packet.
RO 0x000
Address: 0x0120 - 0x0123 Size: 32 bits
Bits Description Type Default
31:10 RESERVED RO 0x00000
9:0 Override LED
These bits select whether each LEDx_0 and LEDx_1 pin will function as an
LED or General Purpose Output (GPO). The LSB bit of this field represents
LED1_0, followed by LED1_1, LED2_0, etc.. When configured as a GPO, the
GPO output is controlled via the LED Output Register.
0 = LEDx_y pin functions as an LED
1 = LEDx_y pin functions as a GPO
0000000000b
Address: 0x0124 - 0x0127 Size: 32 bits
Bits Description Type Default
31:10 RESERVED RO 0x00000
9:0 GPO Output Control
When configured as a GPO via the LED Override Register, the GPO output
is controlled via this field. The LSB bit of this field represents LED1_0,
followed by LED1_1, LED2_0, etc.
0 = LEDx_y pin outputs low
1 = LEDx_y pin outputs high
R/W 0000000000b
Bits Description Type Default
KSZ9897R
DS00002330D-page 74 2017-2019 Microchip Technology Inc.
5.1.3 GLOBAL PHY CONTROL AND STATUS REGISTERS (0x0200 - 0x02FF)
5.1.3.1 Power Down Control 0 Register
5.1.3.2 LED Configuration Strap Register
Note 5-2 The default value of this field is determined by the associated configuration strap values. Refer to
Section 3.2.1, "Configuration Straps," on page 16 for additional information.
Address: 0x0201 Size: 8 bits
Bits Description Type Default
7:6 RESERVED RO 00b
5PLL Power Down
0 = Normal operation.
1 = Disable PLL. This may be used in combination with EDPD mode – see
below.
R/W 0b
4:3 Power Management Mode
00 = Normal operation
01 = Energy Detect Power Down (EDPD) Mode
10 = Soft Power Down Mode
11 = invalid
R/W 00b
2:0 RESERVED RO 000b
Address: 0x0210 - 0x0213 Size: 32 bits
Bits Description Type Default
31:10 RESERVED RO 0x000000
9:0 Configuration strap values of LED pins
[LED4_1, LED4_0, LED3_1, LED3_0, LED2_1, LED2_0, LED1_1, LED1_0]
RO Note 5-2
2017-2019 Microchip Technology Inc. DS00002330D-page 75
KSZ9897R
5.1.4 GLOBAL SWITCH CONTROL REGISTERS (0x0300 - 0x03FF)
5.1.4.1 Switch Operation Register
Note 5-3 The default value of this field is determined by the LED5_1 configuration strap value. Refer to Section
3.2.1, "Configuration Straps," on page 16 for additional information.
5.1.4.2 Switch MAC Address 0 Register
5.1.4.3 Switch MAC Address 1 Register
Address: 0x0300 Size: 8 bits
Bits Description Type Default
7Double Tag Enable
1 = Double tagging is enabled
0 = Double tagging is disabled
R/W 0b
6:2 RESERVED RO 0x00
1Soft Hardware Reset
When set to 1, all register settings, except configuration strap options, are
reset to default values.
R/W
SC
0b
0Start Switch
1 = Switch function is enabled
0 = Switch function is disabled; no traffic will be passed until this bit is set
R/W Note 5-3
Address: 0x0302 Size: 8 bits
Bits Description Type Default
7:0 MAC Address [47:40]
This register, along with the Switch MAC Address 1-5 Registers, define the
switch’s MAC address to be used as the source address in MAC pause
control frames, and for self-address filtering.
R/W 0x00
Address: 0x0303 Size: 8 bits
Bits Description Type Default
7:0 MAC Address [39:32] R/W 0x10
KSZ9897R
DS00002330D-page 76 2017-2019 Microchip Technology Inc.
5.1.4.4 Switch MAC Address 2 Register
5.1.4.5 Switch MAC Address 3 Register
5.1.4.6 Switch MAC Address 4 Register
5.1.4.7 Switch MAC Address 5 Register
Address: 0x0304 Size: 8 bits
Bits Description Type Default
7:0 MAC Address [31:24] R/W 0xA1
Address: 0x0305 Size: 8 bits
Bits Description Type Default
7:0 MAC Address [23:16] R/W 0xFF
Address: 0x0306 Size: 8 bits
Bits Description Type Default
7:0 MAC Address [15:8] R/W 0xFF
Address: 0x0307 Size: 8 bits
Bits Description Type Default
7:0 MAC Address [7:0] R/W 0xFF
2017-2019 Microchip Technology Inc. DS00002330D-page 77
KSZ9897R
5.1.4.8 Switch Maximum Transmit Unit Register
5.1.4.9 Switch ISP TPID Register
5.1.4.10 Switch Lookup Engine Control 0 Register
Address: 0x0308 - 0x0309 Size: 16 bits
Bits Description Type Default
15:14 RESERVED R/W 00b
13:0 Maximum Frame Length (MTU)
Specifies the maximum transmission unit (MTU), which is the maximum
frame payload size. Frames which exceed this maximum are truncated. This
value can be set as high as 9000 (= 0x2328) if jumbo frame support is
required. Also refer to the Switch MAC Control 1 Register and Port MAC
Control 0 Register.
R/W 0x07D0
Address: 0x030A - 0x030B Size: 16 bits
Bits Description Type Default
15:0 ISP Tag TPID
Default tag TPID (EtherType) for untagged incoming frames or the ISP frame
tag TPID for the double tagging function.
R/W 0x9100
Address: 0x0310 Size: 8 bits
Bits Description Type Default
7802.1Q VLAN Enable
This is the master enable for VLAN forwarding and filtering. Note that the
VLAN Table must be set up before VLAN mode is enabled.
1 = VLAN mode enabled
0 = VLAN mode disabled
R/W 0b
6Drop Invalid VID
1 = All received packets with invalid VLAN ID are dropped.
0 = Received packets with invalid VLAN ID are forwarded to the host port.
Note that the Unknown VID Forwarding feature (Unknown VLAN ID Control
Register), if enabled, takes precedence over this bit.
R/W 1b
5:3 Age Count
This bit, in combination with the Age Period value (Switch Lookup Engine
Control 3 Register), determines the aging time of dynamic entries in the
address lookup table. This value is used for the Age Count field whenever a
dynamic table entry is updated.
R/W 10_0b
KSZ9897R
DS00002330D-page 78 2017-2019 Microchip Technology Inc.
5.1.4.11 Switch Lookup Engine Control 1 Register
2Reserved Multicast Lookup Enable
1 = Enable Reserved Multicast Table
0 = Disable Reserved Multicast Table
R/W 0b
1:0 HASH_OPTION
Defines the hashing option for mapping entries to the dynamic lookup table.
00, 11 = Entry is mapped directly using the 10 least significant bits of the
destination address.
01 = The CRC hashing function is used.
10 = The XOR hashing function is used.
Refer to Section 4.4.2.1, "Address Lookup (ALU) Table," on page 28 for
additional information.
R/W 01b
Address: 0x0311 Size: 8 bits
Bits Description Type Default
7Unicast Learning Disable
1 = Unicast address learning is disabled
0 = Unicast address learning is enabled
R/W 0b
6Self-Address Filtering – Global Enable
The source address of received packets is compared to the MAC address in
registers Switch MAC Address 0 Register through Switch MAC Address 5
Register, and the packet is dropped if there is a match.
Self-address filtering can be enabled on a port-by-port basis by setting the
port enable bit in the Port Control 2 Register in addition to setting this bit.
1 = Enable self-address filtering globally for those ports whose port enable
bit (Port Control 2 Register) is set.
0 = Do not filter self-addressed packets on any port.
R/W 0b
5Flush Address Lookup Table
The Flush Option bit in the Switch Lookup Engine Control 2 Register
determines whether flushing is performed on dynamic entries, static entries,
or both.
1 = Trigger a flush of the entire address lookup table. The static address
table is not flushed.
0 = Normal operation
R/W
SC
0b
4Flush MSTP Address Entries (Address Lookup Table)
The Flush Option bit in the Switch Lookup Engine Control 2 Register
determines whether flushing is performed on dynamic entries, static entries,
or both.
1 = Trigger a flush of the matched MSTP entries
0 = Normal operation
R/W
SC
0b
3Multicast Source Address Filtering
1 = Forward packets with a multicast source address
0 = Drop packets with a multicast source address
R/W 1b
Bits Description Type Default
2017-2019 Microchip Technology Inc. DS00002330D-page 79
KSZ9897R
5.1.4.12 Switch Lookup Engine Control 2 Register
2Aging Enable
1 = Enable address table aging
0 = Disable address table aging
R/W 1b
1Fast Aging
1 = Enable fast aging
0 = Disable fast aging
R/W 0b
0Link Down Flush
1 = Link down will cause the entries of any link down port to be flushed
0 = Link down flush is disabled
R/W 0b
Address: 0x0312 Size: 8 bits
Bits Description Type Default
7RESERVED R/W 0b
6Double Tag Multicast Trap
1 = When double tagging mode is enabled, forward all reserved multicast
packets to the host port only.
0 = Normal forwarding
R/W 0b
5Dynamic Entry Egress VLAN Filtering
Egress VLAN filtering uses the forwarding port map from the VLAN table to
restrict the forwarding ports determined from the address lookup. This is the
recommended mode of operation when VLAN is enabled. The default value
is 0 only for backwards compatibility with previous switches.
1 = Enable. For successful lookup of a dynamic entry in the address table,
the forwarding ports are determined from the AND function of the address
table port map and the VLAN table port map.
0 = Disable. For successful lookup of a dynamic entry in the address table,
the forwarding ports are determined from the address table only.
R/W 0b
4Static Entry Egress VLAN Filtering
Egress VLAN filtering uses the forwarding port map from the VLAN table to
restrict the forwarding ports determined from the address lookup. This is the
recommended mode of operation when VLAN is enabled. The default value
is 0 only for backwards compatibility with previous switches.
1 = Enable. For successful lookup of a static entry in the address table, the
forwarding ports are determined from the AND function of the address table
port map and the VLAN table port map.
0 = Disable. For successful lookup of a static entry in the address table, the
forwarding ports are determined from the address table only.
R/W 0b
Bits Description Type Default
KSZ9897R
DS00002330D-page 80 2017-2019 Microchip Technology Inc.
5.1.4.13 Switch Lookup Engine Control 3 Register
3:2 Flush Option
Determines which address lookup table entries may be flushed by either of
the flush operations in the Switch Lookup Engine Control 1 Register.
00 = No flush or flush is done
01 = Flush only dynamic table entries
10 = Flush only static table entries
11 = Flush both static and dynamic table entries
R/W 00b
1:0 MAC Address Priority
00 = MAC Address (MACA) priority for a packet is determined from the
destination address (DA) lookup
01 = MACA priority for a packet is determined from the source address (SA)
lookup
10 = MACA priority for a packet is determined from the higher of the DA and
SA lookups
11 = MACA priority for a packet is determined from the lower of the DA and
SA lookups
R/W 00b
Address: 0x0313 Size: 8 bits
Bits Description Type Default
7:0 Age Period
This value, multiplied by the Age Count value in the entries of the Address
Lookup Table, determines the aging time of dynamic entries in that table.
The unit is seconds.
R/W 0x4B
Bits Description Type Default
2017-2019 Microchip Technology Inc. DS00002330D-page 81
KSZ9897R
5.1.4.14 Address Lookup Table Interrupt Register
This register provides the detailed interrupt status for the Address Lookup Table. These interrupts are enabled in the
Address Lookup Table Mask Register. The LUE interrupt status bit in the Global Interrupt Status Register is the OR of
the status bits in this register.
5.1.4.15 Address Lookup Table Mask Register
This register masks the Address Lookup Table interrupts in the Address Lookup Table Interrupt Register.
Address: 0x0314 Size: 8 bits
Bits Description Type Default
7:3 RESERVED RO 0x00
2Learn Fail Interrupt Status
An Address Lookup Table entry was not learned because all entries in the
bucket are static
R/WC 0b
1Almost Full Interrupt Status
Interrupt indicates that the Address Lookup Table bucket was almost full (2
or 3 valid entries) when a new static entry was written.
R/WC 0b
0Write Fail Interrupt Status
Interrupt indicates that the Address Lookup Table bucket is full and a write
failed
R/WC 0b
Address: 0x0315 Size: 8 bits
Bits Description Type Default
7:3 RESERVED RO 0x00
2Learn Fail Interrupt Mask
1 = Interrupt is disabled
0 = Interrupt is enabled
R/W 1b
1Almost Full Interrupt Mask
1 = Interrupt is disabled
0 = Interrupt is enabled
R/W 1b
0Write Fail Interrupt Mask
1 = Interrupt is disabled
0 = Interrupt is enabled
R/W 1b
KSZ9897R
DS00002330D-page 82 2017-2019 Microchip Technology Inc.
5.1.4.16 Address Lookup Table Entry Index 0 Register
5.1.4.17 Address Lookup Table Entry Index 1 Register
5.1.4.18 Address Lookup Table Entry Index 2 Register
Address: 0x0316 - 0x0317 Size: 16 bits
Bits Description Type Default
15:12 RESERVED RO 0x0
11:0 /
9:0
Almost Full Entry Index [11:0]
When a static entry is successfully written into the Address Lookup Table,
but the table bucket is almost full (contains 2 or 3 static entries prior to the
write), the entry address is reported here.
Fail Write Index [9:0]
When a static entry write failure occurs in the Address Lookup Table, the
bucket address is reported here.
RO 0x000
Address: 0x0318 - 0x0319 Size: 16 bits
Bits Description Type Default
15:10 RESERVED RO 0000_00
9:0 Fail Learn Index
When a destination address fails to be learned in the Address Lookup Table
because the bucket contains 4 static entries, the bucket address is reported
here.
RO 0x000
Address: 0x031A - 0x031B Size: 16 bits
Bits Description Type Default
15:10 RESERVED RO 0000_00
9:0 CPU Access Index
Whenever there is an external read or write to the Address Lookup Table,
the bucket address of the access is reported here.
RO 0x000
2017-2019 Microchip Technology Inc. DS00002330D-page 83
KSZ9897R
5.1.4.19 Unknown Unicast Control Register
The following three registers control forwarding of packets with 1) unknown unicast destination address, 2) unknown
multicast destination address, and 3) unknown VLAN ID.
If a received packet falls into more than one of these categories, the precedence is:
1. Unknown VID
2. Unknown Unicast
3. Unknown Multicast
5.1.4.20 Unknown Multicast Control Register
Address: 0x0320 - 0x0323 Size: 32 bits
Bits Description Type Default
31 Unknown Unicast Packet Forward
1 = Enable forwarding of unknown unicast packets to the ports specified
below
0 = Disable unknown unicast packet forwarding
R/W 0b
30:7 RESERVED RO 0x000000
6:0 Unknown Unicast Forwarding Ports
Bit 0 is for port 1
Bit 1 is for port 2, etc.
1 = Forward unknown unicast packets to that port
0 = Do not forward to that port
All ones = Forwarded to all ports
All zeros = Forwarded to no ports
R/W 000_0000b
Address: 0x0324 - 0x0327 Size: 32 bits
Bits Description Type Default
31 Unknown Multicast Packet Forward
1 = Enable forwarding of unknown multicast packets to the ports specified
below
0 = Disable unknown multicast packet forwarding
R/W 0b
30:7 RESERVED RO 0x000000
6:0 Unknown Multicast Forwarding Ports
Bit 0 is for port 1
Bit 1 is for port 2, etc.
1 = Forward unknown multicast packets to that port
0 = Do not forward to that port
All ones = Forwarded to all ports
All zeros = Forwarded to no ports
R/W 000_0000b
KSZ9897R
DS00002330D-page 84 2017-2019 Microchip Technology Inc.
5.1.4.21 Unknown VLAN ID Control Register
5.1.4.22 Switch MAC Control 0 Register
Address: 0x0328 - 0x032B Size: 32 bits
Bits Description Type Default
31 Unknown VID Packet Forward
1 = Enable forwarding of unknown VLAN ID (VID) packets to the ports
specified below
0 = Disable unknown VID packet forwarding
R/W 0b
30:7 RESERVED RO 0x000000
6:0 Unknown VID Forwarding Ports
Bit 0 is for port 1
Bit 1 is for port 2, etc.
1 = Forward unknown VID packets to that port
0 = Do not forward to that port
All ones = Forwarded to all ports
All zeros = Forwarded to no ports
R/W 000_0000b
Address: 0x0330 Size: 8 bits
Bits Description Type Default
7Alternate Back-off Mode
The back-off mode applies to half-duplex only. This bit should be set if the
No Excessive Collision Drop bit in the Switch MAC Control 1 Register is
enabled.
1 = Enable alternate back-off mode
0 = Disable
R/W 0b
6:4 RESERVED R/W 000b
3Frame Length Field Check
This applies only when the EtherType/Length field is <1500.
1 = Discard any packet if the actual packet length does not match the frame
length field.
0 = Do not check the packet length.
R/W 0b
2RESERVED R/W 1b
2017-2019 Microchip Technology Inc. DS00002330D-page 85
KSZ9897R
5.1.4.23 Switch MAC Control 1 Register
1Flow Control Packet Drop Mode
This bit controls which flow control packets may be forwarded or dropped. To
enable forwarding of all flow control packets, refer to bit 0 of the Switch MAC
Control 4 Register. Enabling of flow control is managed in the PHY Auto-
Negotiation Advertisement Register (for PHY ports) and the XMII Port
Control 0 Register (for MAC ports).
1 = The switch will drop received packets with either EtherType = 0x8808 or
destination address (DA) = 01-80-C2-00-00-01.
0 = The switch will drop received packets with both EtherType = 0x8808 and
DA = 01-80-C2-00-00-01.
R/W 0b
0Aggressive Back-off Enable
For use with half-duplex back pressure. This is not an IEEE standard.
1 = Enable aggressive back-off algorithm in half-duplex mode to enhance
performance.
0 = Disable
R/W 0b
Address: 0x0331 Size: 8 bits
Bits Description Type Default
7RESERVED R/W 1b
6Multicast Storm Protection Disable
1 = Multicast packets are not included in Broadcast Storm Protection
0 = Multicast packets are included in Broadcast Storm Protection
R/W 1b
5Back Pressure Mode
1 = Use carrier sense based back pressure mode
0 = Use collision based back pressure mode. (This is the recommended
mode.)
R/W 1b
4Flow Control and Back Pressure Fair Mode
1 = Enable fair mode. If a flow controlled ingress port and a non-flow
controlled ingress port forward traffic to the same egress port, packets from
the non-flow controlled port may be dropped.
0 = Disable fair mode. In this scenario the flow controlled port will be flow
controlled, and the non-flow controlled port will be neither flow controlled nor
packets dropped.
R/W 1b
3No Excessive Collision Drop
If this bit is set, Alternate Back-odd Mode (bit 7 in the Switch MAC Control 0
Register) should also be set.
1 = The switch will not drop packets when 16 or more collisions occur.
0 = The switch will drop packets when 16 or more collisions occur
R/W 0b
Bits Description Type Default
KSZ9897R
DS00002330D-page 86 2017-2019 Microchip Technology Inc.
5.1.4.24 Switch MAC Control 2 Register
2Jumbo Packet Support
The programmable packet payload size limit is specified in register 0x0308 –
0x0309, up to a maximum of 9000 bytes. This bit overrides bit 1 of this
register.
1 = Enable support for jumbo packets
0 = Disable
R/W 0b
1Legal Maximum Packet Size Check Disable
1 = Accept packets up to 2000 bytes in size.
0 = Accept only standard size packets, up to 1522 bytes for tagged packets,
or 1518 bytes for untagged packets. Larger packets will be dropped.
R/W 0b
0Pass Short Packet
1 = Accept packets between 32 and 64 bytes in size.
0 = Accept only standard size packets, at least 64 bytes. Smaller packets are
dropped.
R/W 0b
Address: 0x0332 Size: 8 bits
Bits Description Type Default
7:4 RESERVED R/W 0x0
3Null VID Replacement
Describes the behavior when a packet is received with a null (zero) VID.
1 = Replace a null VID with the Port Default VID as defined in the Port
Default Tag 0 Register and Port Default Tag 1 Register
0 = No replacement of null VID
R/W 0b
2:0 Broadcast Storm Protection Rate bits [10:8]
The remainder of this fields bits are in the Switch MAC Control 3 Register.
Multiply this value by 64 to determine how many bytes of packet data are
allowed on an input port in a preset period. The period is 5ms for a
1000Mbps port, 50ms for a 100Mbps port, or 500ms for a 10Mbps. The
default is 1%.
R/W 000b
Bits Description Type Default
2017-2019 Microchip Technology Inc. DS00002330D-page 87
KSZ9897R
5.1.4.25 Switch MAC Control 3 Register
5.1.4.26 Switch MAC Control 4 Register
5.1.4.27 Switch MAC Control 5 Register
Address: 0x0333 Size: 8 bits
Bits Description Type Default
7:0 Broadcast Storm Protection Rate bits [7:0]
The remainder of this fields bits are in the Switch MAC Control 2 Register.
Multiply this value by 64 to determine how many bytes of packet data are
allowed on an input port in a preset period. The period is 5ms for a
1000Mbps port, 50ms for a 100Mbps port, or 500ms for a 10Mbps. The
default is 1%.
R/W 0x4A
Address: 0x0334 Size: 8 bits
Bits Description Type Default
7:1 RESERVED RO 0000_000b
0Pass Flow Control Packets
1 = Switch will forward 802.3x PAUSE flow control frames.
0 = Switch will filter PAUSE frames.
R/W 0b
Address: 0x0335 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6RESERVED R/W 0b
5:4 Ingress Rate Limit Period
00 = 16ms
01 = 64ms
1x = 256ms
R/W 01b
3Queue Based Egress Rate Limit Enable
1 = Queue-based egress rate limiting
0 = Port-based egress rate limiting
R/W 0b
2:0 RESERVED RO 000b
KSZ9897R
DS00002330D-page 88 2017-2019 Microchip Technology Inc.
5.1.4.28 Switch MIB Control Register
MIB counters are provided on a per-port basis. They are read and controlled via the Port N: Port Switch MIB Counters
Registers (0xN500 - 0xN5FF).
5.1.4.29 802.1p Priority Mapping 0 Register
5.1.4.30 802.1p Priority Mapping 1 Register
Address: 0x0336 Size: 8 bits
Bits Description Type Default
7Flush MIB Counters
1 = Flush all MIB counters of enabled ports. Refer to the Port MIB Control
and Status Register.
0 = Normal counter operation
R/W
SC
0b
6Freeze MIB Counters
1 = Freeze MIB counters of enabled ports. Refer to the Port MIB Control and
Status Register.
0 = Normal counter operation
R/W 0b
5:0 RESERVED RO 00_0000b
Address: 0x0338 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When the tag PCP field is 0x1, use this value for priority R/W 001b
3RESERVED RO 0b
2:0 When the tag PCP field is 0x0, use this value for priority R/W 000b
Address: 0x0339 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When the tag PCP field is 0x3, use this value for priority R/W 011b
3RESERVED RO 0b
2:0 When the tag PCP field is 0x2, use this value for priority R/W 010b
2017-2019 Microchip Technology Inc. DS00002330D-page 89
KSZ9897R
5.1.4.31 802.1p Priority Mapping 2 Register
5.1.4.32 802.1p Priority Mapping 3 Register
5.1.4.33 IP DiffServ Priority Enable Register
Address: 0x033A Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When the tag PCP field is 0x5, use this value for priority R/W 101b
3RESERVED RO 0b
2:0 When the tag PCP field is 0x4, use this value for priority R/W 100b
Address: 0x033B Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When the tag PCP field is 0x7, use this value for priority R/W 111b
3RESERVED RO 0b
2:0 When the tag PCP field is 0x6, use this value for priority R/W 110b
Address: 0x033E Size: 8 bits
Bits Description Type Default
7:2 RESERVED RO 0000_00b
1RESERVED R/W 0b
0DiffServ Priority Remap Enable
1 = Use the following registers to remap the DSCP (DiffServ) priority to a 3-
bit priority value
0 = Use DSCP bits [5:3] for priority
R/W 0b
KSZ9897R
DS00002330D-page 90 2017-2019 Microchip Technology Inc.
5.1.4.34 IP DiffServ Priority Mapping 0 Register
5.1.4.35 IP DiffServ Priority Mapping 1 Register
5.1.4.36 IP DiffServ Priority Mapping 2 Register
5.1.4.37 IP DiffServ Priority Mapping 3 Register
Address: 0x0340 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x01, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x00, use this value for priority R/W 000b
Address: 0x0341 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x03, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x02, use this value for priority R/W 000b
Address: 0x0342 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x05, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x04, use this value for priority R/W 000b
Address: 0x0343 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x07, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x06, use this value for priority R/W 000b
2017-2019 Microchip Technology Inc. DS00002330D-page 91
KSZ9897R
5.1.4.38 IP DiffServ Priority Mapping 4 Register
5.1.4.39 IP DiffServ Priority Mapping 5 Register
5.1.4.40 IP DiffServ Priority Mapping 6 Register
5.1.4.41 IP DiffServ Priority Mapping 7 Register
Address: 0x0344 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x09, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x08, use this value for priority R/W 000b
Address: 0x0345 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x0B, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x0A, use this value for priority R/W 000b
Address: 0x0346 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x0D, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x0C, use this value for priority R/W 000b
Address: 0x0347 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x0F, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x0E, use this value for priority R/W 000b
KSZ9897R
DS00002330D-page 92 2017-2019 Microchip Technology Inc.
5.1.4.42 IP DiffServ Priority Mapping 8 Register
5.1.4.43 IP DiffServ Priority Mapping 9 Register
5.1.4.44 IP DiffServ Priority Mapping 10 Register
5.1.4.45 IP DiffServ Priority Mapping 11 Register
Address: 0x0348 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x11, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x10, use this value for priority R/W 000b
Address: 0x0349 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x13, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x12, use this value for priority R/W 000b
Address: 0x034A Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x15, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x14, use this value for priority R/W 000b
Address: 0x034B Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x17, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x16, use this value for priority R/W 000b
2017-2019 Microchip Technology Inc. DS00002330D-page 93
KSZ9897R
5.1.4.46 IP DiffServ Priority Mapping 12 Register
5.1.4.47 IP DiffServ Priority Mapping 13 Register
5.1.4.48 IP DiffServ Priority Mapping 14 Register
5.1.4.49 IP DiffServ Priority Mapping 15 Register
Address: 0x034C Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x19, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x18, use this value for priority R/W 000b
Address: 0x034D Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x1B, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x1A, use this value for priority R/W 000b
Address: 0x034E Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x1D, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x1C, use this value for priority R/W 000b
Address: 0x034F Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x1F, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x1E, use this value for priority R/W 000b
KSZ9897R
DS00002330D-page 94 2017-2019 Microchip Technology Inc.
5.1.4.50 IP DiffServ Priority Mapping 16 Register
5.1.4.51 IP DiffServ Priority Mapping 17 Register
5.1.4.52 IP DiffServ Priority Mapping 18 Register
5.1.4.53 IP DiffServ Priority Mapping 19 Register
Address: 0x0350 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x21, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x20, use this value for priority R/W 000b
Address: 0x0351 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x23, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x22, use this value for priority R/W 000b
Address: 0x0352 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x25, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x24, use this value for priority R/W 000b
Address: 0x0353 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x27, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x26, use this value for priority R/W 000b
2017-2019 Microchip Technology Inc. DS00002330D-page 95
KSZ9897R
5.1.4.54 IP DiffServ Priority Mapping 20 Register
5.1.4.55 IP DiffServ Priority Mapping 21 Register
5.1.4.56 IP DiffServ Priority Mapping 22 Register
5.1.4.57 IP DiffServ Priority Mapping 23 Register
Address: 0x0354 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x29, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x28, use this value for priority R/W 000b
Address: 0x0355 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x2B, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x2A, use this value for priority R/W 000b
Address: 0x0350 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x2D, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x2C, use this value for priority R/W 000b
Address: 0x0357 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x2F, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x2E, use this value for priority R/W 000b
KSZ9897R
DS00002330D-page 96 2017-2019 Microchip Technology Inc.
5.1.4.58 IP DiffServ Priority Mapping 24 Register
5.1.4.59 IP DiffServ Priority Mapping 25 Register
5.1.4.60 IP DiffServ Priority Mapping 26 Register
5.1.4.61 IP DiffServ Priority Mapping 27 Register
Address: 0x0358 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x31, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x30, use this value for priority R/W 000b
Address: 0x0359 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x33, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x32, use this value for priority R/W 000b
Address: 0x035A Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x35, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x34, use this value for priority R/W 000b
Address: 0x035B Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x37, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x36, use this value for priority R/W 000b
2017-2019 Microchip Technology Inc. DS00002330D-page 97
KSZ9897R
5.1.4.62 IP DiffServ Priority Mapping 28 Register
5.1.4.63 IP DiffServ Priority Mapping 29 Register
5.1.4.64 IP DiffServ Priority Mapping 30 Register
5.1.4.65 IP DiffServ Priority Mapping 31 Register
Address: 0x035C Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x39, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x38, use this value for priority R/W 000b
Address: 0x035D Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x3B, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x3A, use this value for priority R/W 000b
Address: 0x035E Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x3D, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x3C, use this value for priority R/W 000b
Address: 0x035F Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6:4 When IPv4 / IPv6 DSCP field is 0x3F, use this value for priority R/W 000b
3RESERVED RO 0b
2:0 When IPv4 / IPv6 DSCP field is 0x3E, use this value for priority R/W 000b
KSZ9897R
DS00002330D-page 98 2017-2019 Microchip Technology Inc.
5.1.4.66 Global Port Mirroring and Snooping Control Register
This register contains global controls for port mirroring and IGMP and MLD snooping. Mirroring also requires additional
register settings for the individual ports. Refer to the Port N: Port Switch Ingress Control Registers (0xN800 - 0xN8FF).
5.1.4.67 WRED DiffServ Color Mapping Register
Address: 0x0370 Size: 8 bits
Bits Description Type Default
7RESERVED RO 0b
6IGMP Snooping Enable (IPv4)
1 = IGMP snooping is enabled. All IGMP packets will be forwarded to the
host port.
0 = IGMP snooping is disabled.
R/W 0b
5:4 RESERVED RO 00b
3MLD Snooping Option
1 = Snooping criteria: IPv6 next header = 43, 44, 50, 51 or 60; or next header
= 0 and hop-by-hop next header = 43, 44, 50, 51 or 60.
0 = Snooping criteria: IPv6 next header = 1 or 58; or next header = 0 and
hop-by-hop next header = 1 or 58.
R/W 0b
2MLD Snooping Enable (IPv6)
1 = MLD snooping is enabled. All MLD packets will be forwarded to the host
port.
0 = MLD snooping is disabled.
R/W 0b
1RESERVED RO 0b
0Sniff Mode Select
1 = Sniffing filter is “Rx AND Tx”. Both the source port and destination ports
need to match.
0 = Sniffing filter is “Rx OR Tx”. Either the source port of the destination port
needs to match. This mode is used to implement Rx-only sniffing.
R/W 0b
Address: 0x0378 Size: 8 bits
Bits Description Type Default
7:6 RESERVED RO 00b
5:4 Red
This field specifies the red Differentiated Services Code Point (DSCP) value.
R/W 11b
3:2 Yellow
This field specifies the yellow DSCP value.
R/W 10b
1:0 Green
This field specifies the green DSCP value.
R/W 01b
2017-2019 Microchip Technology Inc. DS00002330D-page 99
KSZ9897R
5.1.4.68 Queue Management Control 0 Register
Address: 0x0390 - 0x0393 Size: 32 bits
Bits Description Type Default
31:8 RESERVED RO 0x000000
7:6 Priority_2Q
When the 2 queue configuration is selected, this determines how to map the
2-bit priority regeneration result from the Port Priority to Queue Mapping
Register into 2 queues.
00 = Priorities 0, 1, 2 map to the Low priority queue. Priority 3 maps to the
High priority queue.
01 = Not used.
10 = Priorities 0, 1 map to the Low priority queue. Priorities 2, 3 map to the
High priority queue.
11 = Priority 0 maps to the Low priority queue. Priorities 1, 2, 3 map to the
High priority queue.
R/W 10b
5:2 RESERVED R/W 00_00b
1Unicast Port VLAN Membership Discard
This bit applies to the Port VLAN Membership function in registers 0xNA04-
07.
1 = All frames are restricted to the forwarding ports as defined in the Port
Control 1 Register.
0 = Frames forwarding to a single destination port are not limited to the
forwarding ports defined in the Port Control 1 Register. Note that when
mirroring is enabled, a single-destination frames will be dropped if it is
mirrored to another port.
R/W 1b
0RESERVED R/W 0b
KSZ9897R
DS00002330D-page 100 2017-2019 Microchip Technology Inc.
5.1.5 GLOBAL SWITCH LOOK UP ENGINE (LUE) CONTROL REGISTERS (0x0400 - 0x04FF)
The following registers are used for accessing the VLAN Table, Address Lookup Table, Static Address Table and the
Reserved Multicast Address Table. The organizations of these tables, and instructions for accessing them, are provided
in Section 5.3, "Tables and MIB Counters (Access)," on page 154.
5.1.5.1 VLAN Table Entry 0 Register
Address: 0x0400 - 0x403 Size: 32 bits
Bits Description Type Default
31 VALID
This field in the VLAN Table specifies if the table entry is valid.
1 = Table entry is valid
0 = Table entry is invalid
R/W 0b
30:28 RESERVED RO 000b
27 FORWARD OPTION
This field in the VLAN Table specifies how the forwarding ports are
determined.
1 = Forward to the VLAN Table port map (PORT FORWARD field).
0 = Forwarding ports are determined by other variables. It may be the ALU
port map, the VLAN port map, the combination of the two, or the Unknown
Unicast/Multicast feature.
R/W 0b
26:24 PRIORITY
This field in the VLAN Table specifies the priority level.
R/W 000b
23:15 RESERVED RO 0000_0000_0b
14:12 MSTP INDEX
This field in the VLAN Table specifies the Multiple Spanning Tree Protocol
index.
R/W 000b
11:7 RESERVED RO 0000_0b
6:0 FID
This field in the VLAN Table specifies the Filter ID. The FID value is normally
combined with the destination address and then hashed to index the address
lookup table.
R/W 000_0000b
Note: Refer to Section 5.3.4, "VLAN Table," on page 161 for additional information on VLAN tables.
2017-2019 Microchip Technology Inc. DS00002330D-page 101
KSZ9897R
5.1.5.2 VLAN Table Entry 1 Register
5.1.5.3 VLAN Table Entry 2 Register
5.1.5.4 VLAN Table Index Register
Address: 0x0404 - 0x407 Size: 32 bits
Bits Description Type Default
31:7 RESERVED R/O 0x0000000
6:0 PORT UNTAG
This field in the VLAN Table specifies the untagging policy for each egress
port.
Bits [6:0] correspond to ports [7:1].
1 = Untag packets upon egress at this port
0 = Do not untag upon egress at this port
R/W 0x00
Note: Refer to Section 5.3.4, "VLAN Table," on page 161 for additional information on VLAN tables.
Address: 0x0408 - 0x040B Size: 32 bits
Bits Description Type Default
31:7 RESERVED R/O 0x0000000
6:0 PORT FORWARD
This field specifies the forwarding policy to each port. The policy is applied if
the FO bit is set.
Bits [6:0] correspond to ports [7:1].
1 = Forward to this port
0 = Do not forward to this port
R/W 0x00
Note: Refer to Section 5.3.4, "VLAN Table," on page 161 for additional information on VLAN tables.
Address: 0x040C - 0x040D Size: 16 bits
Bits Description Type Default
15:12 RESERVED RO 0x0
11:0 VLAN Index
This value addresses the VLAN table, in combination with the VLAN table
action in the following register.
R/W 0x000
Note: Refer to Section 5.3.4, "VLAN Table," on page 161 for additional information on VLAN tables.
KSZ9897R
DS00002330D-page 102 2017-2019 Microchip Technology Inc.
5.1.5.5 VLAN Table Access Control Register
5.1.5.6 ALU Table Index 0 Register
The ALU Table Index 0 Register and ALU Table Index 1 Register contain the index values for searching and reading/
writing the address lookup table as specified by the action in the ALU Table Access Control Register. For additional infor-
mation on ALU tables, refer to Section 5.3.1, "Address Lookup (ALU) Table," on page 154.
Address: 0x040E Size: 8 bits
Bits Description Type Default
7Start VLAN Table Action
1 = Start the action defined below
0 = Action finished
R/W
SC
0b
6:2 RESERVED RO 000_00b
1:0 Action
Specify the action to be taken for the VLAN table entry addressed in the
index register
00 = No operation
01 = Write
10 = Read
11 = Clear all entries to zero
R/W 00b
Note: Refer to Section 5.3.4, "VLAN Table," on page 161 for additional information on VLAN tables.
Address: 0x0410 - 0x0413 Size: 32 bits
Bits Description Type Default
31 RESERVED R/W 0b
30:23 RESERVED RO 000_0000_0b
22:16 FID Index [6:0]
This is the FID value used to hash index the table
R/W 000_0000b
15:0 MAC Index [47:32]
These are the upper 16 bits of the MAC addressed used to hash index the
table
R/W 0x0000
2017-2019 Microchip Technology Inc. DS00002330D-page 103
KSZ9897R
5.1.5.7 ALU Table Index 1 Register
The ALU Table Index 0 Register and ALU Table Index 1 Register contain the index values for searching and reading/
writing the address lookup table as specified by the action in the ALU Table Access Control Register. For additional infor-
mation on ALU tables, refer to Section 5.3.1, "Address Lookup (ALU) Table," on page 154.
5.1.5.8 ALU Table Access Control Register
This register provides control and status for searching and reading or writing the ALU Table. The ALU Table Index 0
Register and ALU Table Index 1 Register contain the index values, while the ALU / Static Address Table Entry 1 Register,
ALU / Static Address / Reserved Multicast Table Entry 2 Register, ALU / Static Address Table Entry 3 Register, and ALU
/ Static Address Table Entry 4 Register are used for the entry values. Refer to Section 5.3.1, "Address Lookup (ALU)
Table," on page 154 for additional information on ALU tables.
Address: 0x0414 - 0x0417 Size: 32 bits
Bits Description Type Default
31:0 MAC Index [31:0]
These are the lower 32 bits of the MAC addressed used to hash index the
address lookup table. If direct addressing is enabled, then bits [11:0] are
used to directly index the address lookup table.
R/W 0x00000000
Address: 0x0418 - 0x041B Size: 32 bits
Bits Description Type Default
31:30 RESERVED RO 00b
29:16 VALID_COUNT
Indicates the total number of valid entries in the table after the search
finishes
RO 0x0000
15:8 RESERVED RO 0x00
7START_FINISH
1 = Start the action defined below
0 = Action finished
R/W, SC 0b
6VALID
For search operation. It is cleared when register 0x042F (ALU / Static
Address Table Entry 4 Register) is read.
1 = Indicates that the next valid entry is ready
0 = Next valid entry is not ready
RO 0b
5VALID_ENTRY_OR_SEARCH_END
For search operation. It is intended for added convenience when accessing
registers by in-band management (IBA) rather than SPI or I2C. It combines
bits 6 and 7.
1 = Indicates either that the next valid entry is ready, or that the search has
ended.
0 = Neither next valid entry ready nor search ended.
RO 0b
4:3 RESERVED R/W 00b
KSZ9897R
DS00002330D-page 104 2017-2019 Microchip Technology Inc.
5.1.5.9 Static Address and Reserved Multicast Table Control Register
This register provides control and index for reading and writing the Static Address Table and the Reserved Multicast
Table. The ALU / Static Address Table Entry 1 Register, ALU / Static Address / Reserved Multicast Table Entry 2 Reg-
ister, ALU / Static Address Table Entry 3 Register, and ALU / Static Address Table Entry 4 Register are used for the
Static Address Table entry values, while only the ALU / Static Address / Reserved Multicast Table Entry 2 Register is
used for the Reserved Multicast Table entry value. Refer to Section 5.3.2, "Static Address Table," on page 158 and Sec-
tion 5.3.3, "Reserved Multicast Address Table," on page 160 for additional information on these tables.
2DIRECT
1 = Access by direct addressing the ALU Table. This method is used only for
debugging, if at all.
0 = Hashing function is used to index the table. This is the normal method.
Refer to the Switch Lookup Engine Control 0 Register.
R/W 0b
1:0 ACTION
Specifies the action to be taken for the ALU Table entry access
00 = No operation
01 = Write
10 = Read
11 = Search
R/W 00b
Address: 0x041C - 0x041F Size: 32 bits
Bits Description Type Default
31:22 RESERVED RO 0x000
21:16 TABLE_INDEX
Bits [21:16] used to index the Reserved Multicast Table
Bits [19:16] used to index the Static Address Table
R/W 00_0000b
15:8 RESERVED RO 0x00
7START_FINISH
1 = Start access
0 = Access is finished
R/W, SC 0b
6:2 RESERVED R/W 000_00b
1TABLE_SELECT
Specifies which table is being accessed
1 = Access Reserved Multicast Table
0 = Access Static Address Table
R/W 0b
0ACTION
Specifies the action to be taken for the table
1 = Read
0 = Write
R/W 0b
Bits Description Type Default
2017-2019 Microchip Technology Inc. DS00002330D-page 105
KSZ9897R
5.1.5.10 ALU / Static Address Table Entry 1 Register
This register contains the table entry values for read and write operations to the Address Lookup Table and Static
Address Table. The field definitions of this register differ dependent on the table type used, as defined in the following
sections:
•ALU Table Entry 1 Register on page 157
•Static Address Table Entry 1 Register on page 159
Refer to Section 5.3.1, "Address Lookup (ALU) Table," on page 154 and Section 5.3.2, "Static Address Table," on
page 158 for additional information.
5.1.5.11 ALU / Static Address / Reserved Multicast Table Entry 2 Register
The field definitions of this register differ dependent on the table type used, as defined in the following sections:
•ALU Table Entry 2 Register on page 157
•Static Address Table Entry 2 Register on page 159
•Reserved Multicast Address Table Entry 2 Register on page 161
Refer to Section 5.3.1, "Address Lookup (ALU) Table," on page 154, Section 5.3.2, "Static Address Table," on page 158,
and Section 5.3.3, "Reserved Multicast Address Table," on page 160 for additional information.
5.1.5.12 ALU / Static Address Table Entry 3 Register
The field definitions of this register differ dependent on the table type used, as defined in the following sections:
•ALU Table Entry 3 Register on page 158
•Static Address Table Entry 3 Register on page 160
Refer to Section 5.3.1, "Address Lookup (ALU) Table," on page 154 and Section 5.3.2, "Static Address Table," on
page 158 for additional information.
5.1.5.13 ALU / Static Address Table Entry 4 Register
This register contains the table entry values for read and write operations to the Address Lookup Table and Static
Address Table. The field definitions of this register differ dependent on the table type used, as defined in the following
sections:
•ALU Table Entry 4 Register on page 158
•Static Address Table Entry 4 Register on page 160
Refer to Section 5.3.1, "Address Lookup (ALU) Table," on page 154 and Section 5.3.2, "Static Address Table," on
page 158 for additional information.
Address: 0x0420 - 0x0423 Size: 32 bits
Address: 0x0424 - 0x0427 Size: 32 bits
Address: 0x0428 - 0x042B Size: 32 bits
Address: 0x042C - 0x042F Size: 32 bits
KSZ9897R
DS00002330D-page 106 2017-2019 Microchip Technology Inc.
5.2 Port Registers
This section details the device’s port registers. All settings are on a per-port basis. Address field “N” specifies the port
number. Valid values for “N” are 1 to 7 for some registers, 6 to 7 for MAC port specific registers, and 0 to 5 for PHY
specific registers.
For an overview of the device’s entire register map, refer to Section 5.0, "Device Registers". For details on the device’s
global registers, refer to Section 5.1, "Global Registers".
5.2.1 PORT N: PORT OPERATION CONTROL REGISTERS (0xN000 - 0xN0FF)
5.2.1.1 Port Default Tag 0 Register
5.2.1.2 Port Default Tag 1 Register
5.2.1.3 Port PME_WoL Event Register
Address: 0xN000
Port N: 1-7
Size: 8 bits
Bits Description Type Default
7:0 Port Default 802.1Q Tag [15:8]
Bits [7:5]: Priority Code Point (PCP)
Bit [4]: Drop Eligible Indicator (DEI)
Bits [3:0]: VLAN Identifier (VID) [11:8]
R/W 0x00
Address: 0xN001
Port N: 1-7
Size: 8 bits
Bits Description Type Default
7:0 Port Default 802.1Q tag [7:0]
VLAN Identifier (VID) [7:0]
R/W 0x01
Address: 0xN013
Port N: 1-7
Size: 8 bits
Bits Description Type Default
7:3 RESERVED RO 0000_0b
2Magic Packet Detect
To be detected, the destination address of the Magic Packet must match the
Global Switch MAC Address in the Switch MAC Address 0 Register through
Switch MAC Address 5 Register.
1 = Magic Packet is detected at this port
0 = Not detected
RO/WC 0b
2017-2019 Microchip Technology Inc. DS00002330D-page 107
KSZ9897R
5.2.1.4 Port PME_WoL Enable Register
1Link Up Detect
Applicable only to ports with integrated PHY.
1 = Link up is detected at this port
0 = Not detected
RO/WC 0b
0Energy Detect
Applicable only to ports with integrated PHY.
1 = Cable energy is detected at this port
0 = Not detected
RO/WC 0b
Address: 0xN017
Port N: 1-7
Size: 8 bits
Bits Description Type Default
7:3 RESERVED RO 0000_0b
2Magic Packet Detect Enable
To be detected, the destination address of the Magic Packet must match the
Global Switch MAC Address in the Switch MAC Address 0 Register through
Switch MAC Address 5 Register.
1 = The PME pin will be asserted when a Magic Packet is detected at this
port
0 = The PME pin will not be asserted by Magic Packet detection at this port
RO/WC 0b
1Link Up Detect Enable
Applicable only to ports with integrated PHY.
1 = The PME pin will be asserted when link up is detected at this port
0 = The PME pin will not be asserted by link up detection at this port
RO/WC 0b
0Energy Detect Enable
Applicable only to ports with integrated PHY.
1 = The PME pin will be asserted when cable energy is detected at this port
0 = The PME pin will not be asserted by cable energy detection at this port
RO/WC 0b
Bits Description Type Default
KSZ9897R
DS00002330D-page 108 2017-2019 Microchip Technology Inc.
5.2.1.5 Port Interrupt Status Register
These registers provide interrupt status for the individual ports. these interrupts are enabled in the Port Interrupt Mask
Register. For non-port specific interrupt status, refer to the Global Interrupt Status Register.
5.2.1.6 Port Interrupt Mask Register
This register enables the interrupts in the Port Interrupt Status Register.
Address: 0xN01B
Port N: 1-7
Size: 8 bits
Bits Description Type Default
7:2 RESERVED RO 0000_00b
1PHY Interrupt Status
Applies only to PHY ports, not MAC port(s). Refer to the Port Interrupt
Control / Status Register for individual PHY interrupts.
0 = No interrupt
1 = Interrupt detected
RO 0b
0ACL Interrupt Status
ACL interrupts can be generated by configuring an entry in the ACL Table for
counter mode. All ports have an ACL Table.
To clear this bit, toggle the ACL Interrupt Mask bit in the Port Interrupt Mask
Register.
0 = No interrupt
1 = Interrupt detected
RO 0b
Address: 0xN01F
Port N: 1-7
Size: 8 bits
Bits Description Type Default
7:2 RESERVED RO 0000_00b
1PHY Interrupt Enable
Applies only to PHY ports.
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
0ACL Interrupt Enable
Applies to all ports.
0 = Interrupt enabled
1 = Interrupt disabled
R/W 0b
2017-2019 Microchip Technology Inc. DS00002330D-page 109
KSZ9897R
5.2.1.7 Port Operation Control 0 Register
5.2.1.8 Port Status Register
Address: 0xN020
Port N: 1-7
Size: 8 bits
Bits Description Type Default
7Local MAC Loopback
Data going from the internal switch fabric to an egress port is looped back in
the MAC of that egress port and returned to the internal switch fabric.
1 = Local MAC loopback
0 = Normal operation
R/W 0b
6Remote MAC Loopback
Data received at the external pins of a port is looped back in the MAC of that
port and sent out the same port, without passing through the internal switch
fabric. For PHY ports, data is received on the RX+/- pins and transmitted out
the TX+/- pins. For xMII ports, data is received on the TXD pins and
transmitted on the RXD pins.
This loopback path does not include an RX-to-TX elastic buffer. Use small
packet sizes and only one or two packets at a time to avoid data corruption.
1 = Remote MAC loopback
0 = Normal operation
R/W 0b
5:3 RESERVED RO 000b
2Tail Tag Enable
When tail tagging is enabled for a port, it designates that port to be the “host”
or “CPU” port. Do not enable tail tagging for more than one port.
See the tail tagging description for details.
1 = Enable tail tagging on this port
0 = Disable tail tagging for this port
R/W 0b
1:0 Egress Queue Split Enable
11 = Reserved
10 = Four egress queues. Packets are assigned to a queue based on
priority.
01 = Two egress queues. Packets are assigned to a queue based on priority.
00 = Single egress queue. There is no priority differentiation.
R/W 00b
Address: 0xN030
Port N: 1-7
Size: 8 bits
KSZ9897R
DS00002330D-page 110 2017-2019 Microchip Technology Inc.
5.2.2 PORT N: PORT ETHERNET PHY REGISTERS (0xN100 - 0xN1FF)
The registers in this section are for PHY ports only. Refer to the IEEE802.3 clause 22.3.4 for additional details.
Bits Description Type Default
7:5 RESERVED RO 000
4:3 Port Speed Status
For non-PHY ports, these bits duplicate the speed setting in the XMII Port
Control 1 Register bit 6 and XMII Port Control 0 Register bit 4.
For PHY ports, these bits indicate the actual link speed, which is also
available in the PHY Control Register.
00 = 10 Mb/s
01 = 100 Mb/s
10 = 1000 Mb/s
RO -
2Port Duplex Status
For non-PHY ports, this bit duplicates the duplex setting in the XMII Port
Control 0 Register bit 6.
For PHY ports, this bit indicates the actual link duplex, which is also available
in the PHY Control Register.
1 = Full duplex
0 = Half duplex
RO 1 or -
1Transmit Flow Control Enabled Status
For non-PHY ports, this bit duplicates the Tx FC enable bit 5 in the XMII Port
Control 0 Register.
For PHY ports, this bit is set only when FC is enabled (PHY Auto-Negotiation
Advertisement Register, bits 11:10), link is up and FC is established via auto-
negotiation.
1 = TX flow control is enabled
0 = Disabled
RO -
0Receive Flow Control Enabled Status
For non-PHY ports, this bit duplicates the Rx FC enable bit 3 in the XMII Port
Control 0 Register.
For PHY ports, this bit is set only when FC is enabled (PHY Auto-Negotiation
Advertisement Register, bits 11:10), link is up and FC is established via auto-
negotiation.
1 = RX flow control is enabled
0 = Disabled
RO -