82583V [INTEL]
Intel® 82583V GbE Controller; 英特尔82583V千兆以太网控制器
| 型号： | 82583V |
| 厂家： | 
INTEL |
| 描述： | Intel® 82583V GbE Controller |
| 文件： | 总374页 (文件大小：5578K) |
| 中文： | 中文翻译 | 下载： | 下载PDF数据表文档文件 |
Intel® 82583V GbE Controller
Datasheet
Product Features
PCI Express* (PCIe*)
High Performance
— 64-bit address master support for systems
using more than 4 GB of physical memory
— Programmable host memory receive buffers
(256 bytes to 16 KB)
— TCP segmentation capability compatible with
Large Send offloading features
— Support up to 256 KB TCP segmentation (TSO
v2)
— Intelligent interrupt generation features to
— Fragmented UDP checksum offload for packet
enhance driver performance
reassemble
— Descriptor ring management hardware for
transmit and receive software controlled reset
(resets everything except the configuration
space)
— Message Signaled Interrupts (MSI)
— Configurable receive and transmit data FIFO,
programmable in 1 KB increments
— IPv4 and IPv6 checksum offload support
(receive, transmit, and large send)
— Split header support
— 40 KB packet buffer size
Low Power
— Magic Packet* wake-up enable with unique
MAC address
MAC
— ACPI register set and power down functionality
supporting D0 andD3 states
— Flow Control Support compliant with the
802.3X Specification
— Full wake up support (APM and ACPI 2.0)
— Smart power down at S0 no link and Sx no link
— LAN disable function
— VLAN support compliant with the 802.1Q
Specification
— MAC Address filters: perfect match unicast
— filters; multicast hash filtering, broadcast filter
— and promiscuous mode
— Statistics for RMOM
— MAC loopback
Technology
— 9 mm x 9 mm 64-pin QFN package with
Exposed Pad*
— Configurable LED operation for customization
of LED displays
PHY
— Compliant with the 1 Gb/s IEEE 802.3 802.3u
802.3ab Specifications
— IEEE 802.3ab auto negotiation support
— Full duplex operation at 10/100/1000 Mb/s
— Half duplex at 10/100 Mb/s
— Auto MDI, MDI-X crossover at all speeds
June 2012
Revision 2.5
INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL PRODUCTS. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR
OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL'S TERMS AND CONDITIONS
OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING
TO SALE AND/OR USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE,
MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.
A "Mission Critical Application" is any application in which failure of the Intel Product could result, directly or indirectly, in personal injury or death.
SHOULD YOU PURCHASE OR USE INTEL'S PRODUCTS FOR ANY SUCH MISSION CRITICAL APPLICATION, YOU SHALL INDEMNIFY AND HOLD INTEL AND
ITS SUBSIDIARIES, SUBCONTRACTORS AND AFFILIATES, AND THE DIRECTORS, OFFICERS, AND EMPLOYEES OF EACH, HARMLESS AGAINST ALL
CLAIMS COSTS, DAMAGES, AND EXPENSES AND REASONABLE ATTORNEYS' FEES ARISING OUT OF, DIRECTLY OR INDIRECTLY, ANY CLAIM OF PRODUCT
LIABILITY, PERSONAL INJURY, OR DEATH ARISING IN ANY WAY OUT OF SUCH MISSION CRITICAL APPLICATION, WHETHER OR NOT INTEL OR ITS
SUBCONTRACTOR WAS NEGLIGENT IN THE DESIGN, MANUFACTURE, OR WARNING OF THE INTEL PRODUCT OR ANY OF ITS PARTS.
Intel may make changes to specifications and product descriptions at any time, without notice. Designers must not rely on the absence or characteristics
of any features or instructions marked "reserved" or "undefined". Intel reserves these for future definition and shall have no responsibility whatsoever
for conflicts or incompatibilities arising from future changes to them. The information here is subject to change without notice. Do not finalize a design
with this information.
The products described in this document may contain design defects or errors known as errata which may cause the product to deviate from published
specifications. Current characterized errata are available on request.
Contact your local Intel sales office or your distributor to obtain the latest specifications and before placing your product order.
Copies of documents which have an order number and are referenced in this document, or other Intel literature, may be obtained by calling 1-800-548-
4725, or go to: http://www.intel.com/design/literature.htm.
Intel and Intel logo are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries.
*Other names and brands may be claimed as the property of others.
Copyright © 2012, Intel Corporation. All Rights Reserved.
2
Datasheet—82583V GbE Controller
Contents
1.0 Introduction............................................................................................................ 10
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
Scope.............................................................................................................. 10
Number Conventions ......................................................................................... 10
Acronyms......................................................................................................... 11
Reference Documents........................................................................................ 11
82583V Architecture Block Diagram..................................................................... 12
System Interface............................................................................................... 12
Features Summary............................................................................................ 12
Product Codes................................................................................................... 15
2.0 Pin Interface ........................................................................................................... 16
2.1
2.2
2.3
Pin Assignments................................................................................................ 16
Pull-Up/Pull-Down Resistors and Strapping Options................................................ 17
Signal Type Definition........................................................................................ 17
2.3.1 PCIe..................................................................................................... 17
2.3.2 NVM Port............................................................................................... 18
2.3.3 LEDs .................................................................................................... 18
2.3.4 PHY Pins ............................................................................................... 19
2.3.5 Miscellaneous and Test Pins..................................................................... 20
2.3.6 Power Supplies and Support Pins.............................................................. 21
2.3.7 Reserved Pins........................................................................................ 21
Package........................................................................................................... 22
2.4
3.0 Electrical Specifications........................................................................................... 24
3.1
3.2
Introduction ..................................................................................................... 24
Voltage Regulator Power Supply Specification ....................................................... 24
3.2.1 3.3 V dc Rail.......................................................................................... 24
3.2.2
3.2.3 1.05 V dc Rail ........................................................................................ 25
3.2.4 PNP Specifications ................................................................................. 25
1.9 V dc Rail......................................................................................... 24
3.3
3.4
3.5
3.6
3.7
3.8
3.9
Power Sequencing............................................................................................. 26
Power-On Reset ................................................................................................ 26
Power Scheme Solutions .................................................................................... 27
Flash AC Specifications....................................................................................... 30
EEPROM AC Specifications .................................................................................. 31
Discrete/Integrated Magnetics Specifications......................................................... 32
Oscillator/Crystal Specifications........................................................................... 33
3.10 I/O DC Parameters............................................................................................ 34
3.10.1 LEDs .................................................................................................... 35
4.0 Initialization............................................................................................................ 36
4.1
4.2
4.3
Introduction ..................................................................................................... 36
Reset Operation................................................................................................ 36
Power Up ......................................................................................................... 38
4.3.1 Power-Up Sequence................................................................................ 38
4.3.2 Timing Diagram ..................................................................................... 46
Global Reset (PE_RST_N, PCIe In-Band Reset)...................................................... 47
4.4.1 Reset Sequence ..................................................................................... 47
4.4.2 Timing Diagram ..................................................................................... 48
Timing Parameters ............................................................................................ 50
4.5.1 Timing Requirements.............................................................................. 50
4.5.2 MDIO and NVM Semaphore...................................................................... 50
4.4
4.5
3
82583V GbE Controller—Datasheet
4.6
Software Initialization Sequence ..........................................................................51
4.6.1 Interrupts During Initialization..................................................................51
4.6.2 Global Reset and General Configuration .....................................................51
4.6.3 Link Setup Mechanisms and Control/Status Bit Summary .............................52
4.6.4 Initialization of Statistics..........................................................................53
4.6.5 Receive Initialization ...............................................................................54
4.6.6 Transmit Initialization..............................................................................54
5.0 Non-Volatile Memory (NVM) Map .............................................................................56
5.1
Basic Configuration Table....................................................................................56
5.1.1 Hardware Accessed Words .......................................................................58
5.1.2 Software Accessed Words ........................................................................72
6.0 Interconnects ..........................................................................................................80
6.1
PCIe ................................................................................................................80
6.1.1 Architecture, Transaction, and Link Layer Properties....................................81
6.1.2 General Functionality ..............................................................................82
6.1.3 Transaction Layer ...................................................................................82
6.1.4 Flow Control...........................................................................................87
6.1.5 Host I/F.................................................................................................89
6.1.6 Error Events and Error Reporting ..............................................................90
6.1.7 Link Layer..............................................................................................93
6.1.8 PHY ......................................................................................................94
6.1.9 Performance Monitoring...........................................................................95
Ethernet Interface .............................................................................................95
6.2.1 MAC/PHY GMII/MII Interface....................................................................95
6.2.2 Duplex Operation for Copper PHY/GMII/MII Operation .................................96
6.2.3 Auto-Negotiation & Link Setup Features.....................................................97
6.2.4 Loss of Signal/Link Status Indication .......................................................100
6.2.5 10/100 Mb/s Specific Performance Enhancements.....................................101
6.2.6 Flow Control.........................................................................................102
SPI Non-Volatile Memory Interface.....................................................................105
6.3.1 General Overview .................................................................................105
6.3.2 Supported NVM Devices.........................................................................105
6.3.3 NVM Device Detection ...........................................................................106
6.3.4 Device Operation with an External EEPROM..............................................107
6.3.5 Device Operation with Flash ...................................................................107
6.3.6 Shadow RAM........................................................................................107
6.3.7 NVM Clients and Interfaces ....................................................................109
6.3.8 NVM Write and Erase Sequence ..............................................................110
6.2
6.3
7.0 Inline Functions.....................................................................................................114
7.1
Packet Reception .............................................................................................114
7.1.1 Packet Address Filtering.........................................................................114
7.1.2 Receive Data Storage............................................................................115
7.1.3 Legacy Receive Descriptor Format...........................................................115
7.1.4 Extended Rx Descriptor .........................................................................118
7.1.5 Packet Split Receive Descriptor...............................................................123
7.1.6 Receive Descriptor Fetching ...................................................................127
7.1.7 Receive Descriptor Write Back ................................................................127
7.1.8 Receive Descriptor Queue Structure ........................................................128
7.1.9 Receive Interrupts ................................................................................130
7.1.10 Receive Packet Checksum Offloading.......................................................133
4
Datasheet—82583V GbE Controller
7.2
Packet Transmission ........................................................................................ 135
7.2.1 Transmit Functionality........................................................................... 135
7.2.2 Transmission Flow Using Simplified Legacy Descriptors.............................. 136
7.2.3 Transmission Process Flow Using Extended Descriptors.............................. 136
7.2.4 Transmit Descriptor Ring Structure......................................................... 137
7.2.5 Overview of On-Chip Transmit Modes...................................................... 139
7.2.6 Pipelined Tx Data Read Requests............................................................ 140
7.2.7 Transmit Interrupts .............................................................................. 141
7.2.8 Transmit Data Storage.......................................................................... 141
7.2.9 Transmit Descriptor Formats.................................................................. 142
7.2.10 Extended Data Descriptor Format........................................................... 150
TCP Segmentation........................................................................................... 154
7.3.1 TCP Segmentation Performance Advantages ............................................ 154
7.3.2 Ethernet Packet Format......................................................................... 154
7.3.3 TCP Segmentation Data Descriptors........................................................ 155
7.3.4 TCP Segmentation Source Data.............................................................. 156
7.3.5 Hardware Performed Updating for Each Frame ......................................... 156
7.3.6 TCP Segmentation Use of Multiple Data Descriptors .................................. 157
Interrupts ...................................................................................................... 160
7.4.1 Legacy and MSI Interrupt Modes ............................................................ 160
7.4.2 Registers............................................................................................. 160
7.4.3 Interrupt Moderation ............................................................................ 161
7.4.4 Clearing Interrupt Causes...................................................................... 164
802.1q VLAN Support ...................................................................................... 165
7.5.1 802.1q VLAN Packet Format .................................................................. 165
7.5.2 Transmitting and Receiving 802.1q Packets ............................................. 166
7.5.3 802.1q VLAN Packet Filtering ................................................................. 166
LEDs.............................................................................................................. 167
7.3
7.4
7.5
7.6
8.0 Power Management and Delivery........................................................................... 170
8.1
8.2
8.3
Assumptions................................................................................................... 170
Power Consumption......................................................................................... 170
Power Delivery................................................................................................ 171
8.3.1 The 1.9 V dc Rail.................................................................................. 171
8.3.2 The 1.05 V dc Rail................................................................................ 171
Power Management ......................................................................................... 171
8.4.1 82583V Power States............................................................................ 171
8.4.2 Auxiliary Power Usage........................................................................... 172
8.4.3 Power Limits by Certain Form Factors ..................................................... 173
8.4.4 Power States ....................................................................................... 173
8.4.5 Timing of Power-State Transitions .......................................................... 177
Wake Up........................................................................................................ 180
8.5.1 Advanced Power Management Wake Up .................................................. 180
8.5.2 PCIe Power Management Wake Up.......................................................... 181
8.5.3 Wake-Up Packets ................................................................................. 181
8.4
8.5
9.0 Driver Programing Interface.................................................................................. 188
9.1
Introduction ................................................................................................... 188
9.1.1 Memory and I/O Address Decoding......................................................... 188
9.1.2 Registers Byte Ordering ........................................................................ 191
9.1.3 Register Conventions............................................................................ 191
5
82583V GbE Controller—Datasheet
9.2
Configuration and Status Registers - CSR Space ..................................................192
9.2.1 Register Summary Table........................................................................192
9.2.2 General Register Descriptions.................................................................197
9.2.3 PCIe Register Descriptions .....................................................................216
9.2.4 Interrupt Register Descriptions...............................................................224
9.2.5 Receive Register Descriptions.................................................................230
9.2.6 Transmit Register Descriptions ...............................................................245
9.2.7 Statistic Register Descriptions ................................................................252
9.2.8 PHY Registers.......................................................................................267
9.2.9 Diagnostic Register Descriptions .............................................................296
10.0 Programing Interface.............................................................................................302
10.1 PCIe Configuration Space..................................................................................302
10.1.1 PCIe Compatibility ................................................................................302
10.1.2 Mandatory PCI Configuration Registers ....................................................303
10.1.3 PCI Power Management Registers ...........................................................308
10.1.4 Message Signaled Interrupt (MSI) Configuration Registers..........................311
10.1.5 PCIe Configuration Registers ..................................................................312
11.0 Design Considerations ...........................................................................................324
11.1 PCIe ..............................................................................................................324
11.1.1 Port Connection to the 82583V ...............................................................324
11.1.2 PCIe Reference Clock ............................................................................324
11.1.3 Other PCIe Signals................................................................................324
11.1.4 PCIe Routing........................................................................................325
11.2 Clock Source...................................................................................................325
11.2.1 Frequency Control Device Design Considerations.......................................325
11.2.2 Frequency Control Component Types.......................................................325
11.3 Crystal Support ...............................................................................................327
11.3.1 Crystal Selection Parameters..................................................................327
11.3.2 Crystal Placement and Layout Recommendations ......................................330
11.4 Oscillator Support............................................................................................331
11.4.1 Oscillator Placement and Layout Recommendations...................................332
11.5 Ethernet Interface ...........................................................................................333
11.5.1 Magnetics for 1000 BASE-T....................................................................333
11.5.2 Magnetics Module Qualification Steps ......................................................333
11.5.3 Third-Party Magnetics Manufacturers.......................................................333
11.5.4 Designing the 82583V as a 10/100 Mb/s Only Device ................................334
11.5.5 Layout Considerations for the Ethernet Interface.......................................335
11.5.6 Physical Layer Conformance Testing........................................................341
11.5.7 Troubleshooting Common Physical Layout Issues ......................................341
11.6 82583V Power Supplies ....................................................................................342
11.6.1 82583V GbE Controller Power Sequencing................................................342
11.6.2 Power and Ground Planes ......................................................................344
11.7 Device Disable.................................................................................................345
11.7.1 BIOS Handling of Device Disable.............................................................345
11.8 82583V Exposed Pad*......................................................................................346
11.8.1 Introduction.........................................................................................346
11.8.2 Component Pad, Solder Mask and Solder Paste.........................................346
11.8.3 Landing Pattern A (No Via In Pad)...........................................................348
11.8.4 Landing Pattern B (Thermal Relief; No Via In Pad).....................................349
11.9 Assembly Process Flow.....................................................................................350
11.10 Reflow Guidelines ............................................................................................350
11.11 XOR Testing....................................................................................................352
6
Datasheet—82583V GbE Controller
12.0 Thermal Design Considerations ............................................................................. 354
12.1 Introduction ................................................................................................... 354
12.2 Intended Audience .......................................................................................... 354
12.3 Measuring the Thermal Conditions..................................................................... 354
12.4 Thermal Considerations.................................................................................... 354
12.5 Packaging Terminology .................................................................................... 355
12.6 Product Package Thermal Specification............................................................... 355
12.7 Thermal Specifications ..................................................................................... 356
12.7.1 Case Temperature................................................................................ 356
12.7.2 Designing for Thermal Performance ........................................................ 356
12.8 Thermal Attributes .......................................................................................... 357
12.8.1 Typical System Definitions..................................................................... 357
12.9 82583V Package Thermal Characteristics............................................................ 358
12.10 Reliability....................................................................................................... 358
12.11 Measurements for Thermal Specifications ........................................................... 359
12.12 Case Temperature Measurements...................................................................... 359
12.12.1Attaching the Thermocouple .................................................................. 360
12.13 Conclusion ..................................................................................................... 360
12.14 PCB Guidelines................................................................................................ 361
13.0 Diagnostics............................................................................................................ 362
13.1 Introduction ................................................................................................... 362
13.2 FIFO Pointer Accessibility.................................................................................. 362
13.3 FIFO Data Accessibility..................................................................................... 362
13.4 Loopback Operations ....................................................................................... 363
14.0 Board Layout and Schematic Checklists................................................................. 364
15.0 Models................................................................................................................... 374
7
82583V GbE Controller—Datasheet
Revision History
Date
Revision Description
•
•
Revised table 24 - NVM Map of Address Range 0x00-0x3F (Word 0x05).
Revised section 9.2.5.11 - Receive Descriptor Control - RXDCTL (0x02828; RW).
June 2012
2.5
2.4
•
•
Revised section 9.2.2.3 (EEPROM/FLASH Control Register; bit 23 footnote).
Revised section 9.2.2.15 (Extended Configuration Control; bits 7:5).
January 2012
February 2011
October 2009
•
•
•
Updated table 18 (Cload value).
Updated section 3.9 “Oscillator/Crystal Specifications” (added Cload note).
Updated section 11.3.1.8 “Load Capacitance and Discrete Capacitors” (new crystal load
capacitance formula).
Added new section 11.5.4 “Designing the 82583V as a 10/100 Mb/s Only Device”.
Removed section 11.5.5.7.1 “Signal Detect”.
Removed all references to “heat sinks” in section 12.0 “Thermal Design Considerations”.
Updated sections 5.2.1.12 (bits 15:13), 6.1.1.13 (bits 6:5), 6.1.1.15 (bits 15:8) - changed bit
values.
Added section 3.6 “Flash AC Specifications” and section 3.7 “EEPROM AC Specifications”.
Added MDIO and NVM semaphore information to section 4.5.1.
Removed section 5.1.
Added new hardware defaults and NVM image settings to section 5.0 “Non-Volatile Memory
(NVM) Map”.
Revised section 6.3.2 “Supported NVM Devices”.
Revised section 9.2.3.11 (bit 1:0 descriptions).
Revised section 12.6 “Product Package Thermal Specification” - added a psi JT note after table
60.
Revised section 10.1.2.2 “Device ID” - changed to 0x150C.
Revised section 5.1.1.1.6 (PCIe Init Configuration 3 Word (Word 0x1A); bits 3:2).
Revised section 10.1.5.1.7 (Link CAP, Offset 0xEC, (RO); bits 11:10).
•
•
•
•
•
•
•
•
2.3
•
•
•
•
•
•
•
•
Changed the pull-up value of AUX_PWR from 1 KΩ to 10 KΩ. in the schematic checklist.
Changed “calibration load” to “Cload” in the schematic checklist.
2.2
August 2009
June 2009
April 2009
2.1
2.0
1.0
Updated section 11.3.1.6 “Load Capacitance”.
Initial Public Release.
Initial Release (Intel Confidential).
•
•
Updated section 6.1.2.1 (NVM words 0x03 through 0x07).
Added new section 13.10 (Assembly Process Flow).
June 2010
April 2010
3.0
2.9
Updated sections 6.1.1.12 (bits 15:13), 6.1.1.13 (bits 6:5), 6.1.1.15 (bits 15:8) - changed bit
values.
8
Datasheet—82583V GbE Controller
Note:
This page intentionally left blank.
9
82583V GbE Controller—Introduction
1.0
Introduction
The 82583V is a single, compact, low power components that offer a fully-integrated
Gigabit Ethernet Media Access Control (MAC) and Physical Layer (PHY) port. The
82583V uses the PCI Express* (PCIe*) architecture and provides a single-port
implementation in a relatively small area so it can be used for server and client
configurations as a LAN on Motherboard (LOM) design. The 82583V can also be used in
embedded applications such as switch add-on cards and network appliances.
External interfaces provided on the 82583V:
• PCIe Rev. 1.1 (2.5 GHz) x1
• MDI (Copper) standard IEEE 802.3 Ethernet interface for 1000BASE-T, 100BASE-
TX, and 10BASE-T applications (802.3, 802.3u, and 802.3ab)
• IEEE 1149.1 JTAG (note that BSDL testing is NOT supported)
Additional product details:
• 9 mm x 9 mm 64-pin QFN package
• Support for PCI 3.0 Vital Product Data (VPD)
1.1
1.2
Scope
This document presents the architecture (including device operation, pin descriptions,
register definitions, etc.) for the 82583V. This document is intended to be a reference
for software device driver developers, board designers, test engineers, or others who
might need specific technical or programming information about the 82583V.
Number Conventions
Unless otherwise specified, numbers are represented as follows:
• Hexadecimal numbers are identified by an "0x" suffix on the number (0x2A, 0x12).
• Binary numbers are identified by a "b" suffix on the number (0011b).
Any other numbers without a suffix are intended as decimal numbers.
10
Introduction—82583V GbE Controller
1.3
Acronyms
Following are a list of acronyms that are used throughout this document.
Acronym
Definition
CSR
Control and Status Register. Usually refers to a hardware register.
Dynamic Host Configuration Protocol. A TCP/IP protocol that enables a client to
receive a temporary IP address over the network from a remote server.
DHCP
IEEE
IP
Institute of Electrical and Electronics Engineers.
Internet Protocol. The protocol within TCP/IP that governs the breakup and
reassembly of data messages into packets and the packet routing within the network.
The 4-byte or 16-byte address that designates the Ethernet controller within the IP
communication protocol. This address is dynamic and can be updated frequently
during runtime.
IP Address
LAN
Local Area Network. Also known as the Ethernet.
The 6-byte address that designates Ethernet controller within the Ethernet protocol.
This address is constant and unique per Ethernet controller.
MAC Address
NA
Not Applicable.
Network Interface Card. Generic name for a Ethernet controller that resides on a
Printed Circuit Board (PCB).
NIC
OS
Operating System. Usually designates the PC system’s software.
To Be Defined.
TBD
1.4
Reference Documents
• Intel® 82583V GbE Controller Specification Update, Intel Corporation.
• PCI Express* Specification v1.1 (2.5 GT/s)
• Advanced Configuration and Power Interface Specification
• PCI Bus Power Management Interface Specification
11
82583V GbE Controller—Introduction
1.5
82583V Architecture Block Diagram
Figure 1 shows a high-level architecture block diagram for the 82583V.
PCIe
PCIe I/F
Rx/Tx DMA
Rx/Tx FIFO
Transmit
Switch
Filter
MAC
PHY
Link
82583V Architecture Block Diagram
System Interface
Figure 1.
1.6
The 82583V provides one PCIe lane operating at 2.5 GHz with sufficient bandwidth to
support 1000 Mb/s transfer rate. On-chip buffering mitigates instantaneous receive
bandwidth demands and eliminates transmit under–runs by buffering the entire
outgoing packet prior to transmission.
1.7
Features Summary
This section describes the 82583V’s features that were present in previous Intel client
GbE controllers and those features that are new to the 82583V.
12
Introduction—82583V GbE Controller
Table 1.
Network Features
Feature
82583V
82573L
Compliant with the 1 Gb/s Ethernet 802.3
802.3u 802.3ab specifications
Y
Y
Multi-speed operation: 10/100/1000 Mb/s
Full-duplex operation at 10/100/1000 Mb/s
Half-duplex operation at 10/100 Mb/s
Y
Y
Y
Y
Y
Y
Flow control support compliant with the 802.3X
specification
Y
Y
Y
Y
VLAN support compliant with the 802.3q
specification
MAC address filters: perfect match unicast
filters; multicast hash filtering, broadcast filter
and promiscuous mode
Y
Y
Y
Y
Configurable LED operation for OEM
customization of LED displays
Statistics for RMON
MAC loopback
Y
Y
Y
Y
Table 2.
Host Interface Features
Feature
82583V
82573L
PCIe interface to chipset
Y
Y
64-bit address master support for systems using
more than 4 GB of physical memory
Y
Y
Y
Y
Y
Y
Y
Y
Programmable host memory receive buffers (256
bytes to 16 KB)
Intelligent interrupt generation features to
enhance software device driver performance
Descriptor ring management hardware for
transmit and receive
Software controlled reset (resets everything
except the configuration space)
Y
Y
Y
Y
Message Signaled Interrupts (MSI)
13
82583V GbE Controller—Introduction
Table 3.
Performance Features
Feature
82583V
82573L
Configurable receive and transmit data FIFO;
programmable in 1 KB increments
Y
Y
TCP segmentation capability compatible with NT
5.x TCP Segmentation Offload (TSO) features
Y
Y
Y
Y
N
Y
Supports up to 256 KB TSO (TSO v2)
Fragmented UDP checksum offload for packet re-
assembly
IPv4 and IPv6 checksum offload support (receive,
transmit, and TSO)
Y
Y
Split header support
Packet buffer size
Y
Y
40 KB
32 KB
Table 4.
Power Management Features
Feature
82583V
82573L
Magic packet wake-up enable with unique MAC
address
Y
Y
ACPI register set and power down functionality
supporting D0 and D3 states
Y
Y
Full wake-up support (APM and ACPI 2.0)
Smart power down at S0 no link and Sx no link
LAN disable functionality
Y
Y
Y
Y
Y
Y
14
Introduction—82583V GbE Controller
1.8
Product Codes
Table 5 lists the product ordering codes for the 82583V.
Product Ordering Codes
Table 5.
Part Number
Product Name
Description
•
•
Embedded and Entry Server GbE LAN.
Operates using a standard temperature
range (0 °C to 85 °C).
Intel® 82583V Gigabit Network
Connection
WG82583V
15
82583V GbE Controller—Pin Interface
2.0
Pin Interface
2.1
Pin Assignments
The 82583V supports a 64-pin, 9 x 9 QFN package with an Exposed Pad* (e-Pad*).
Note that the e-Pad is ground.
42
46 45 44 43
41 40 39 38 37 36 35 34 33
48 47
VDD3p3
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
32
31
30
29
28
27
26
25
24
23
22
21
20
MDI_MINUS[3]
MDI_PLUS[3]
AVDD1p9
LED0
LED1
TEST_EN
DEV_OFF_N
MDI_MINUS[2]
MDI_PLUS[2]
MDI_MINUS[1]
MDI_PLUS[1]
AVDD1p9
VDD1p0
PECLKp
PECLKn
PE_Rp
82583V
64 Pin QFN
9 mm x 9 mm
0.5 mm pin pitch
with Exposed Pad*
MDI_MINUS[0]
MDI_PLUS[0]
DIS_REG10
VDD1p0
PE_Rn
AVDD1p9
PE_Tp
PE_Tn
AVDD1p9
19 VDD1p9
CTRL10
18
AVDD1p9
VDD1p0
17 PE_RST_N
CTRL19
1
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Figure 2.
82583V 64-Pin, 9 x 9 QFN Package With e-Pad
16
Pin Interface—82583V GbE Controller
2.2
2.3
Pull-Up/Pull-Down Resistors and Strapping Options
• As stated in the Name and Function table columns, the internal Pull-Up/Pull-Down
(PU/PD) resistor values are 30 KΩ ± 50%.
• Only relevant (digital) pins are listed; analog or bias and power pins have specific
considerations listed in Section 3.0.
• NVMT and AUX_PWR are used for a static configuration. They are sampled while
PE_RST_N is active and latched when PE_RST_N is deasserted. At other times,
they revert to their standard usage.
Signal Type Definition
In
Input is a standard input-only signal.
Out (O)
T/s
Totem pole output is a standard active driver.
Tri-State is a bi-directional, tri-state input/output pin.
Sustained tri-state is an active low tri-state signal owned and driven by one and only one agent
at a time. The agent that drives an s/t/s pin low must drive it high for at least one clock before
letting it float. A new agent cannot start driving an s/t/s signal any sooner than one clock after
the previous owner tri-states it.
S/t/s
O/d
A-in
A-out
B
Open drain enables multiple devices to share as a wire-OR.
Analog input signals.
Analog output signals.
Input bias.
2.3.1
PCIe
Table 6.
PCIe
Op
Mode
Symbol
Lead #
Type
Name and Function
PCIe Differential Reference Clock In
This pin receives a 100 MHz differential clock input. This clock
is used as the reference clock for the PCIe Tx/Rx circuitry and
by the PCIe core PLL to generate a 125 MHz clock and 250
MHz clock for the PCIe core logic.
PECLKp
PECLKn
26
25
A-in
Input
PCIe Serial Data Output
Serial differential output link in the PCIe interface running at
2.5 Gb/s. This output carries both data and an embedded 2.5
GHz clock that is recovered along with data at the receiving
end.
PE_Tp
PE_Tn
21
20
A-out
Output
17
82583V GbE Controller—Pin Interface
Table 6.
PCIe
Op
Mode
Symbol
Lead #
Type
Name and Function
PCIe Serial Data Input
PE_Rp
PE_Rn
24
23
Serial differential input link in the PCIe interface running at
2.5 Gb/s. The embedded clock present in this input is
recovered along with the data.
A-in
Input
Wake
The 82583V drives this signal to zero when it detects a wake-
up event and either:
PE_WAKE_N/
JTAG_TDO
16
17
O/d
In
Output
Input
•
•
The PME_en bit in PMCSR is 1b or
The APME bit of the Wake Up Control (WUC) register is
1b.
JTAG TDO Output.
Power and Clock Good Indication
The PE_RST_N signal indicates that both PCIe power and
clock are available.
PE_RST_N
2.3.2
NVM Port
Table 7.
NVM Port
Op
Mode
Symbol
Lead #
Type
Name and Function
Serial Data Output
Connect this lead to the input of the Non-Volatile Memory
NVM_SI
NVM_SO
12
T/s
Output (NVM).
Note: The NVM_SI port pin includes an internal pull-up
resistor.
Serial Data Input
Connect this lead to the output of the NVM.
Note: The NVM_SO port pin includes an internal pull-up
resistor.
14
T/s
Input
Non-Volatile Memory Serial Clock
NVM_SK
NVM_CS_N
LEDs
13
15
T/s
T/s
Output
Output
Note: The NVM_SK port pin includes an internal pull-up
resistor.
Non-Volatile Memory Chip Select Output
Note: The NVM_CS port pin includes an internal pull-up
resistor.
2.3.3
Table 8 lists the functionality of each LED output pin. The default activity of each LED
can be modified in the NVM. The LED functionality is reflected and can be further
modified in the configuration registers (LEDCTL).
18
Pin Interface—82583V GbE Controller
Table 8.
LEDs
Op
Mode
Symbol
Lead #
Type
Name and Function
LED0
Programmable LED.
LED0
LED1
LED2
31
Out
Output
Output
Output
LED1
Programmable LED.
30
33
Out
Out
LED2
Programmable LED.
2.3.4
PHY Pins
Note:
The 82583V has built in termination resistors. As a result, external termination
resistors should not be used.
Table 9.
PHY Pins
Op
Mode
Symbol
Lead #
Type
Name and Function
Media Dependent Interface[0]:
1000BASE-T:
In MDI configuration, MDI[0]+/-corresponds to BI_DA+/-
and in MDI-X configuration MDI[0]+/- corresponds to
BI_DB+/-.
100BASE-TX:
MDI_PLUS[0]
MDI_MINUS[0]
58
57
A
Bi-dir
In MDI configuration, MDI[0]+/- is used for the transmit
pair and in MDIX configuration MDI[0]+/- is used for the
receive pair.
10BASE-T:
In MDI configuration, MDI[0]+/- is used for the transmit
pair and in MDI-X configuration MDI[0]+/- is used for the
receive pair.
Media Dependent Interface[1]:
1000BASE-T:
In MDI configuration, MDI[1]+/- corresponds to BI_DB+/-
and in MDI-X configuration MDI[1]+/- corresponds to
BI_DA+/-.
100BASE-TX:
MDI_PLUS[1]
MDI_MINUS[1]
55
54
A
Bi-dir
In MDI configuration, MDI[1]+/- is used for the receive
pair and in MDI-X configuration MDI[1]+/- is used for the
transmit pair.
10BASE-T:
In MDI configuration, MDI[1]+/- is used for the receive
pair and in MDI-X configuration MDI[1]+/- is used for the
transmit pair.
Media Dependent Interface[3:2]:
1000BASE-T:
In MDI and in MDI-X configuration, MDI[2]+/-
corresponds to BI_DC+/- and MDI[3]+/- corresponds to
BI_DD+/-.
100BASE-TX: Unused.
10BASE-T:Unused.
MDI_PLUS[2]
MDI_MINUS[2]
MDI_PLUS[3]
MDI_MINUS[3]
53
52
50
49
A
Bi-dir
19
82583V GbE Controller—Pin Interface
Table 9.
PHY Pins
Op
Mode
Symbol
Lead #
Type
Name and Function
XTAL In/Out
A-In
XTAL1
XTAL2
43
42
Input/
Output
These pins can be driven by an external 25 MHz crystal or
driven by an external MOS level 25 MHz oscillator. Used to
drive the PHY.
A-Out
A-out
A
ATEST_P
ATEST_N
45
46
Positive side of the high speed differential debug port for
the PHY.
Output
Bias
PHY Termination
This pin should be connected through a 4.99 KΩ +-1%
resister to ground.
RSET
48
2.3.5
Miscellaneous and Test Pins
Table 10.
Miscellaneous and Test Pins
Op
Mode
Symbol
Lead #
Type
Name and Function
This is a 3.3 V dc input signal. Asserting DEV_OFF_N
puts the 82583V in device disable mode. Note that
this pin is asynchronous.
DEV_OFF_N
28
29
In
Input
Enables Test Mode
Test pins are overloaded on the functional signals as
described in the pin description text of this section.
The pin is active high.
Note: This pin should be externally pulled down for
normal operation.
TEST_EN
In
In
Input
Input
Auxiliary Power Indication.
AUX_PWR is supported when sampled high and
should be connected using a resistor
AUX_PWR/
JTAG_TCK
39
JTAG Clock Input
Note: The AUX_PWR/JTAG_TCK port pin includes an
internal pull-down resistor.
NVM Type
The NVM is Flash when sampled LOW and EEPROM
when sampled HIGH.
JTAG TMS Input.
NVMT/JTAG_TMS 38
In
In
Input
Input
Note: The NVMT/JTAG_TMS port pin includes an
internal pull-up resistor. Also note that the internal
pull-up is disconnected during startup. As a result,
NVMT MUST be connected externally.
JTAG TDI Input
Note: The JTAG_TDI port pin includes an internal
JTAG_TDI
40
pull-up resistor.
20
Pin Interface—82583V GbE Controller
2.3.6
Power Supplies and Support Pins
2.3.6.1
Power Support
Power Support
Table 11.
Type /
Voltage
Symbol
Lead #
Name and Function
1.05 V dc Control
Voltage control for an external 1.05 V dc PNP.
CTRL10
62
64
A-out
A-out
1.9 V dc Control
Voltage control for an external 1.9 V dc PNP.
CTRL19
Disable 1.05 V dc Regulator
When high, the internal 1.05 V dc regulator is disabled and the
CTRL10 signal is active. When low, the internal 1.05 V dc
regulator is enabled using its internal power transistor. In this
case, the CTRL10 signal is inactive.
DIS_REG10
59
A-in
2.3.6.2
Power Supply
Power Supply
Table 12.
Type /
Voltage
Symbol
Lead #
Name and Function
4, 11, 18, 27, 1.05 V
VDD1p0
1.05 V dc power supply (7).
37, 41, 60
dc
22, 44, 47,
51, 56, 61, 63
AVDD1p9
VDD3p3
1.9 V dc 1.9 V dc power supply (7).
3.3 V dc 3.3 V dc power supply (2).
3.3 V dc 3.3 V dc power supply (1).
10, 32
1
AVDD3p3/
VDD3p3
Fuse voltage for programming on-die fuses. Connect to 1.9 V dc for
VDD1p9
GND
19
1.9 V dc
Ground
normal operation.
The e-Pad metal connection on the bottom of the package. Should be
connected to ground.
e-Pad
2.3.7
Reserved Pins
Symbol
Lead #
Name and Function
RSVD2_PD
RSVD3_PD
RSVD7_PD
2
3
7
These pins need to be pulled down using 10 KΩ resistors.
RSVD5_PU
RSVD6_PU
RSVD8_PU
RSVD9_PU
RSVD34_PU
RSVD35_PU
RSVD36_PU
5
6
8
9
34
35
36
These pins need to be pulled up using 10 KΩ resistors.
21
82583V GbE Controller—Pin Interface
2.4
Package
The 82583V supports a 64-pin, 9 x 9 QFN package with e-Pad. Figure 3 shows the
package schematics.
Figure 3.
82583V QFN 9 x 9 mm Package
22
Pin Interface—82583V GbE Controller
Note:
This page intentionally left blank.
23
82583V GbE Controller—Electrical Specifications
3.0
Electrical Specifications
3.1
Introduction
This chapter describes the 82583V's electrical properties.
Voltage Regulator Power Supply Specification
3.3 V dc Rail
3.2
3.2.1
Title
Rise Time
Description
Time from 10% to 90% mark
Min
Max
100
Units
ms
1
Mononotonicity
Slope
Voltage dip allowed in ramp
0
mV dc
V dc/s
V dc
mV
Ramp rate at any given time between 10% and 90%
Voltage range for normal operating conditions
Maximum voltage ripple @ BW = 50 MHz
Maximum voltage allowed
2880
3.6
70
4
Operational Range
Ripple
3
Overshoot
Capacitance
V dc
μF
Minimum capacitance
25
3.2.2
1.9 V dc Rail
Title
Description
Min
Max
100
Units
Rise Time
Time from 10% to 90% mark
1
ms
Mononotonicity
Slope
Voltage dip allowed in ramp
0
mV dc
V dc/s
V dc
mV dc
V dc
μF
Ramp rate at any given time between 10% and 90%
Voltage range for normal operating conditions
Maximum voltage ripple @ BW = 50 MHz
Maximum voltage allowed
1440
2
Operational Range
Ripple
1.8
50
2.7
40
Overshoot
Output Capacitance
Input Capacitance
Capacitance ESR
Ictrl
Capacitance range when using PNP circuit
Capacitance range when using PNP circuit
20
20
5
μF
1
Equivalent series resistance of output capacitance
100
10
mΩ
Maximum output current rating to CTRL19
mA
1. Do not use tantalum capacitors.
24
Electrical Specifications—82583V GbE Controller
3.2.3
1.05 V dc Rail
Title
Description
Time from 10% to 90% mark
Min
Max
Units
Rise Time
1
100
0
ms
Mononotonicity
Slope
Voltage dip allowed in ramp
mV dc
V dc/s
%
Ramp rate at any given time between 10% and 90%
Voltage range for normal operating conditions
Maximum voltage ripple @ BW = 50 MHz
Maximum voltage allowed
800
+5
50
Operational Range
Ripple
-5
mV dc
V dc
μF
Overshoot
1.5
40
Output Capacitance
Input Capacitance
Capacitance ESR
Ictrl
Capacitance range when using PNP circuit
Capacitance range when using PNP circuit
20
20
μF
1
Equivalent series resistance of output capacitance
10
10
mΩ
Maximum output current rating to CTRL10
mA
1. Do not use tantalum capacitors.
3.2.4
PNP Specifications
Table 13.
External Power Supply Specification
Title
Description
Min Max Units
VCBO
20
20
1
V dc
V dc
A
VCEO
IC(max)
IC(peak)
1.2
A
Ptot
hFE
hfe
Cc
Minimum total dissipated power @ 25 °C ambient temperature 1.5
W
DC current gain @ Vce=-10 V dc, Ic=500 mA
85
AC current gain @ Ic=50mA VCE=-10 V dc, f=20 MHz
collector capacitance @ VCB=-5V, f=1MHz
2.5
50
pF
fT
Transition frequency @ Ic=10mA, VCE=-5 V dc, f=100 MHz
40
MHz
Recommended transistor BCP69
25
82583V GbE Controller—Electrical Specifications
3.3
Power Sequencing
For proper and safe operation, the power supplies must follow the following rule:
VDD3p3 (3.3 V dc) ≥ AVDD1p9 (1.9 V dc) ≥ VDD1p0 (1.05 V dc)
This means that VDD3p3 MUST start ramping before AVDD1p8 and VDD1p0, but
VDD1p0 MIGHT reach its nominal operating range before AVDD1p8 and VDD3p3.
Basically, the higher voltages must be greater than or equal to the lower voltages. This
is necessary to avoid low impedance paths through clamping diodes and to eliminate
back-powering.
The same requirements apply to the power-down sequence.
Internal Power On Reset must be low throughout the time that the power supplies are
ramping. This guarantees that the MAC and PHY resets cleanly. While Internal Power
On Reset is low, reset to the PHY is also asserted. After the power supplies are valid,
Internal Power On Reset must remain low for at least tCLK125START to guarantee that the
CLK125 clock from the PHY is running.
3.4
Power-On Reset
• Power up sequence – 3.3 V dc -> 1.9 V dc -> 1.05 V dc
• Power down sequence 1.05 V dc -> 1.9 V dc->3.3 V dc
Table 14.
Power Detection Thresholds
Symbol
Parameter
Specifications
Min Typ Max
1.7
Units
V1a
High threshold for 3.3 V dc supply
Low threshold for 3.3 V dc supply
High threshold for 1.05 V dc supply
Low threshold for 1.05 V dc supply
1.35
1.35
0.6
2.0
1.9
0.75
0.6
V dc
V dc
V dc
V dc
V2a
V1b
V2b
1.6
0.7
0.35
0.45
26
Electrical Specifications—82583V GbE Controller
3.5
Power Scheme Solutions
Figure 4 shows the intended design options for power solutions. The values for the
various components in Figure 4 are listed in Table 15; Table 16 and Table 17 list the
power consumption values.
3.3 V dc
3.3 V dc
3.3 V dc
C4
R
VDD3p3
VDD3p3
CTRL10
VDD1p0
CTRL10
VDD1p0
X
C1
C1
R2
C2
C5
C3
82583V
82583V
3.3 V dc
3.3 V dc
3.3 V dc
C3
C4
R
R
1 K ohm
CTRL19
CTRL19
DIS_REG10
DIS_REG10
R1
R1
1 K ohm
AVDD1p9
AVDD1p9
C4
OPTION A:
Fully Integrated 1.05 V dc Regulator
1.9 V dc PnP Transistor Regulator
OPTION B:
External 1.05 V dc
1.9 V dc PnP Transistor Regulator
3.3 V dc
3.3 V dc
1.05 V dc
VDD3p3
CTRL10
VDD1p0
X
VDD3p3
CTRL10
VDD1p0
X
C1
C1
C2
C2
82583V
82583V
3.3 V dc
1.9 V dc
1 K ohm
X
3.3 V dc
CTRL19
CTRL19
X
DIS_REG10
DIS_REG10
1 K ohm
External
1.9 V dc
AVDD1p9
AVDD1p9
C4
Regulator
C4
OPTION C:
All External Power Supplies
OPTION D:
Fully Integrated 1.05 V dc
External 1.9 V dc Regulator
Figure 4.
Power Scheme Schematics
27
82583V GbE Controller—Electrical Specifications
Table 15.
Parameters For Power Scheme Options
1
Option A
Option B
Option C
Option D
C1
C2
C3
C4
10 μF
10 μF
10 μF
10 μF
10 μF
10 μF
22 μF + 0.1 μF
(multiple)
22 μF + 0.1 μF
(multiple)
22 μF + 0.1 μF
(multiple)
10μF
10 μF +0.1 μF (multiple 22 μF + 0.1 μF
near pins)
10 μF +0.1 μF
(multiple near pins)
(multiple near pins)
10 μF +0.1 μF
(multiple near pins)
C5
R1
R2
R
0 Ω
0 Ω
0 Ω
5 KΩ
5 KΩ
1. 1.05 V dc PNP uses 1.9 V dc from PNP.
Notes:
1.
2.
3.
4.
5.
6.
All capacitors are ceramic type.
10 μF capacitance can be 2 x 4.7 μF.
22 μF can be 2 x 10 μF or 4 x 4.7 μF for 1.9 V dc bypass.
Place 0.1 μF capacitors near pins.
PNP must be placed 0.5-inch (10 mm) from the 82583V.
VDD1p0 pins are connected together by a plane.
Note:
The following numbers apply to device current and power and do not include power
losses on external components.
Table 16.
Options B and C Power Consumption (External 1.05 V dc Regulator)
3.3
[mA]
1.9
[mA]
1.05
[mA]
Power
[mW]
State
Mode
S0 - Maximum
1000Base-T active, 90 °C
1000Base-T active
1000Base-T idle
5
4
4
4
4
4
4
4
4
4
4
4
266
261
217
116
71
195
184
108
60
22
48
11
5
727
702
539
296
171
372
157
45
100Base-T active
100Base-T idle
S0 - Typical
10Base-T active
162
70
10Base-T idle
Cable disconnect
14
LAN disable
13
2
40
D3 cold with WOL 100 Mb/s
D3 cold with WOL 10 Mb/s
D3 cold without WOL
71
22
11
5
171
157
34
SX
70
8
28
Electrical Specifications—82583V GbE Controller
Table 17.
Options A and D Power Consumption (Fully Integrated 1.05 V dc Regulator)
3.3
[mA]
1.9
[mA]
Power
[mW]
State
Mode
S0 - Maximum
1000Base-T active, 90 °C
1000Base-T active
1000Base-T idle
5
4
4
4
4
4
4
4
4
4
4
4
471
455
331
178
93
911
878
642
351
190
416
167
44
100Base-T active
100Base-T idle
S0 - Typical
10Base-T active
212
81
10Base-T idle
Cable disconnect
18
LAN disable
12
36
D3 cold with WOL 100 Mb/s
D3 cold with WOL 10 Mb/s
D3 cold without WOL
92
188
167
35
SX
81
13
29
82583V GbE Controller—Electrical Specifications
3.6
Flash AC Specifications
The 82583V is designed to support a serial flash. Applicable over the recommended
operating range from Ta = -40 °C to +85 °C, VCC3P3 = 3.3 Vdc, Cload = 1 TTL Gate
and 16 pF (unless otherwise noted).
Symbol
Parameter
Min
Typ
Max
Units
MHz
Note
t
t
t
t
t
t
t
t
t
t
t
t
t
SCK clock frequency
Input rise time
Input fall time
SCK high time
SCK low time
0
15.625
2.5
20
20
20
[1]
SCK
RI
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
2.5
FI
20
20
25
25
25
5
32
[2]
[2]
WH
WL
CS
CSS
CSH
SU
H
32
CS high time
CS setup time
CS hold time
Data-in setup time
Data-in hold time
Output valid
5
20
V
Output hold time
Output disable time
0
HO
DIS
100
Notes:
1.
2.
Clock is 62.5 MHz divided by 4. In bit banging mode maximum allowable frequency is 20 MHz.
45% to 55% duty cycle.
tCS
VIH
CS
VIL
t
tCSH
CSS
VIH
Sck
tWH
tWL
VIL
tSU
tH
VIH
VIL
VOH
SI
VALID IN
tHO
t
tDIS
val
SO
HI-Z
HI-Z
VOL
Figure 5.
Flash Timing Diagram
30
Electrical Specifications—82583V GbE Controller
3.7
EEPROM AC Specifications
The 82583V is designed to support a standard serial EEPROM. Applicable over
recommended operating range from Ta = -40 °C to +85 °C, VCC3P3 = 3.3 Vdc,
Cload = 1 TTL Gate and 16 pF (unless otherwise noted).
Symbol
Parameter
Min
Typ
Max
2.1
Units
MHz
Note
t
t
t
t
t
t
t
t
t
t
t
t
t
SCK clock frequency
Input rise time
Input fall time
SCK high time
SCK low time
0
2
[1]
[2]
SCK
2
2
µs
µs
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
RI
FI
200
200
250
250
250
50
250
250
WH
WL
CS
CSS
CSH
SU
H
CS high time
CS setup time
CS hold time
Data-in setup time
Data-in hold time
Output valid
50
0
200
250
V
Output hold time
Output disable time
0
HO
DIS
Notes:
1.
2.
Clock is 2 MHz.
45% to 55% duty cycle.
tCS
VIH
CS
t
tCSH
CSS
VIH
Sck
tWH
t H
tWL
VIL
tSU
VIH
VIL
SI
VALID IN
t
tHO
tDIS
val
SOVOH HI-Z
VOL
HI-Z
Figure 6.
EEPROM Timing Diagram
31
82583V GbE Controller—Electrical Specifications
3.8
Discrete/Integrated Magnetics Specifications
Criteria
Condition
Values (Min/Max)
1500 Vrms (min)
At 50 to 60 Hertz for 60 seconds
For 60 seconds
Voltage
Isolation
2250 V dc (min)
Open Circuit
Inductance
(OCL) or OCL
(alternate)
With 8 mA DC bias at 25 °C
400 μH (min)
With 8 mA DC bias at 0 °C to 70 °C
350 μH (min)
100 kHz through 999 kHz
1.0 MHz through 60 MHz
60.1 MHz through 80 MHz
80.1 MHz through 100 MHz
100.1 MHz through 125 MHz
1 dB (max)
0.6 dB (max)
0.8 dB (max)
1.0 dB (max)
2.4 dB (max)
Insertion Loss
1.0 MHz through 40 MHz
40.1 MHz through 100 MHz
When reference impedance si 85 Ω,
100 Ω, and 115 Ω.
18 dB (min)
Return Loss
12 to 20 * LOG (frequency in MHz / 80) dB (min)
Note that return loss values might
vary with MDI trace lengths. The
LAN magnetics might need to be
measured in the platform where it
is used.
Crosstalk
Isolation
Discrete
Modules
1.0 MHz through 29.9 MHz
30 MHz through 250 MHz
250.1 MHz through 375 MHz
-50.3+(8.8*(freq in MHz / 30)) dB (max)
-26-(16.8*(LOG(freq in MHz / 250)))) dB (max)
-26 dB (max)
Crosstalk
Isolation
Integrated
Modules
1.0 MHz through 10 MHz
10.1 MHz through 100 MHz
100.1 MHz through 375 MHz
-50.8+(8.8*(freq in MHz / 10)) dB (max)
-26-(16.8*(LOG(freq in MHz / 100)))) dB (max)
-26 dB (max)
1.0 MHz through 29.9 MHz
30 MHz through 500 MHz
-40.2+(5.3*((freq in MHz / 30)) dB (max)
-22-(14*(LOG((freq in MHz / 250)))) dB (max)
Diff to CMR
CM to CMR
1.0 MHz through 270 MHz
270.1 MHz through 300 MHz
300.1 MHz through 500 MHz
-57+(38*((freq in MHz / 270)) dB (max)
-17-2*((300-(freq in MHz) / 30) dB (max)
-17 dB (max)
32
Electrical Specifications—82583V GbE Controller
3.9
Oscillator/Crystal Specifications
See Figure 7 for recommended crystal placement and layout instructions.
External Crystal Specifications
Table 18.
Recommended
Parameter Name
Frequency
Symbol
Max/Min Range
Conditions
Value
f
25 [MHz]
@25 [°C]
o
Vibration Mode
Fundamental
±30 [ppm]
±30 [ppm]
Frequency Tolerance @25 °C
Temperature Tolerance
Series Resistance (ESR)
Crystal Load Capacitance
Shunt Capacitance
Drive Level
Df/f @25°C
@25 [°C]
o
Df/f
o
R
s
50 [Ω] max
@25 [MHz]
C
C
18 [pF]
load
o
6 [pF] max
D
300 [μW] max
L
Aging
Df/f
±5 ppm per year
Parallel
±5 ppm per year max
o
Calibration Mode
Insulation Resistance
500 [MΩ] min
@ 100 V dc
Note:
Intel does not recommend a significantly lower Cload capacitance then 18 pF due to
crystal trimming sensitivity (ppm/pF) versus Cload rating. Lower Cload crystals are
more likely to cause the LAN reference frequency to be out of specification (30 ppm).
Also, a ≥ 5x gain or ≥ 5x negative resistance is NOT necessary from Intel’s LAN
component’s crystal driver. A ≥ 5x gain or ≥ 5x negative resistance can cause crystal
overdrive and accelerate crystal aging (ppm/year). As a result, the crystal circuit might
shift off frequency more quickly than it would with lower driver levels and normal
aging.
33
82583V GbE Controller—Electrical Specifications
Table 19.
Clock Oscillator Specifications
Parameter Name
Frequency
Symbol/Parameter
Conditions
@25 [°C]
Min
Typ
Max
Unit
MHz
f
25.0
3.3
o
Swing
Vp-p1
f/f
3
3.6
V
Frequency Tolerance
Operating Temperature
Aging
-20 to +70
±50
[ppm]
o
T
-20 to +70 [°C]
opr
f/f
±5 ppm per year
[ppm]
[pF]
nS
o
Coupling capacitor
TH_XTAL_IN
Coupling
12
13
13
15
20
20
18
XTAL_IN High Time
XTAL_IN Low Time
XTAL_IN Total Jitter
TL_XTAL_IN
nS
1
TJ_XTAL_IN
200
pS
1. Broadband peak-to-peak = 200 pS, Broadband rms = 3 pS, 12 KHz to 20 MHz rms = 1 ps.
Note:
Peak-to-peak voltage presented at the XTAL1 input cannot exceed 1.9 V dc.
Figure 7.
XTAL Timing Diagram
3.10
I/O DC Parameters
This section specifies the timing and electrical parameters for the various I/O
interfaces.
34
Electrical Specifications—82583V GbE Controller
3.10.1
LEDs
Symbol/Parameter
Conditions
Min
3.0
Typ
Max
Unit
VDD3p3
3.3
3.6
10
V dc
μA
Input leakage
Iol @ VOL=0.4 V dc
Ioh @ VOH=VDDO-0.4 V dc
Cin
0<Vin<VDD3p3
12
12
mA
mA
pF
5
35
82583V GbE Controller—Initialization
4.0
Initialization
4.1
Introduction
This chapter discusses initialization steps. This includes:
• General hardware power-up state
• Basic device configuration
• Initialization of transmit and receive operation
• Link configuration and software reset capability
• Statistics initialization
4.2
Reset Operation
The 82583V reset sources are as follows:
• Internal Power On Reset- The 82583V has an internal mechanism for sensing the
power pins. Once power is up and stable, the 82583V implements an internal reset.
This reset acts as a master reset of the entire chip. It is level sensitive, and while it
is 0b holds all of the registers in reset. Internal Power On Reset is an indication that
device power supplies are all stable. Internal Power On Reset changes state during
system power up.
• PE_RST_N - Indicates that both the power and the PCIe clock sources are stable; a
value of 0b indicates reset active. This pin asserts an internal reset also after a
D3cold exit. Most units are reset on the rising edge of PE_RST_N. The only
exception is the PCIe unit, which is kept in reset while PE_RST_N is active.
• Device Disable/Dr Disable - The 82583V enters a device disable mode when the
DEV_OFF_N pin is asserted without shutdown (see Section 8.4.4.4). The 82583V
enters Dr disable mode when certain conditions are met in the Dr state (see
Section 8.4.4.3).
• In-band PCIe reset - The 82583V generates an internal reset in response to a
Physical Layer (PHY) message from PCIe or when the PCIe link goes down (entry to
polling or detect state). This reset is equivalent to PCI reset in previous (PCI) GbE
controllers.
• D3hotD0 transition - This is also known as ACPI reset. The 82583V generates an
internal reset on the transition from D3hot power state to D0 (caused after
configuration writes from D3 to D0 power state). Note that this reset is per function
and resets only the function that transitioned from D3hot to D0.
• Software Reset - Software can reset the 82583V by writing the Device Reset bit of
the Device Control (CTRL.RST) register. The 82583V re-reads the per-function NVM
fields after a software reset. Bits that are normally read from the NVM are reset to
their default hardware values. Note that this reset is per function and resets only
the function that received the software reset. PCI configuration space
(configuration and mapping) of the device is unaffected.
36
Initialization—82583V GbE Controller
• EEPROM Reset - Writing a 1b to the EEPROM Reset bit of the Extended Device
Control (CTRL_EXT.EE_RST) register causes the 82583V to re-read the per-function
configuration from the NVM, setting the appropriate bits in the registers loaded by
the NVM.
• PHY Reset - Software can write a 1b to the PHY Reset bit of the Device Control
(CTRL.PHY_RST) register to reset the internal PHY.
The resets affect the following registers and logic:
Table 20.
82583V Resets
Reset Name
Internal
PE_
RST_
N
In-band
PCIe
Reset
Reset
activation
Power
On
Reset
Device/Dr
Disable
D3hot
D0
SW
EE
PHY
Notes
Reset Reset Reset
PCIe Data Path
Load NVM
√
√
√
√
√
√
√
√
6
√
√
√
PCI Config
Registers RO
√
√
√
√
PCI Config
Registers RW
√
√
√
√
√
1
√
√
√
√
√
√
√
Data path
4
Wake Up (PM)
Context
Wake Up
Control
√
√
√
√
2
3
Register
Wake Up
Status
Registers
PHY
√
√
√
√
√
√
√
√
√
√
Strapping Pins
Notes:
1. If D3cold is not supported, the wake-up context is reset (PME_Status and PME_En
bits).
2. Refers to bits in the Wake-Up Control (WUC) register that are not part of the wake-
up context (the PME_En and PME_Status bits).
3. The Wake-Up Status (WUS) registers include the following:
— WUS register.
— Wake-up packet length.
— Wake-up packet memory.
37
82583V GbE Controller—Initialization
4. The following register fields do not follow the previously mentioned general rules:
— Packet Buffer Allocation (PBA) - reset on Internal Power On Reset only.
— Packet Buffer Size (PBS) - reset on Internal Power On Reset only.
— LED configuration registers.
— The Aux Power Detected bit in the PCIe Device Status register is reset on
Internal Power On Reset and PCIe Power Good only.
— FLA - reset on Internal Power On Reset only.
5. The NVM is loaded only when the LAN function exits D3hot state.
In situations where the device is reset using the software reset CTRL.RST, the TX data
lines will be forced to all zeros. This causes a substantial number of symbol errors to be
detected by the link partner.
4.3
Power Up
4.3.1
Power-Up Sequence
Figure 8 through Figure 14 shows the 82583V’s power-up sequencing.
Figure 8 shows a high-level view of the power sequence, while Figure 9 through
Figure 14 provides a more detailed description of each state.
38
Initialization—82583V GbE Controller
Start
Power-On-Reset
Flash
EEPROM
A
B
Load EEPROM
Load Flash
C
Initialize PHY
D
Read NVM after PERST#
de-assertion
E
Initialize PCIe and PHY
Bring up PCIe link
Figure 8.
82583V Power Up - General Flow
39
82583V GbE Controller—Initialization
Start
Legend
Stage
Power ramp up
(3.3 V dc, 1.9 V dc,
1.05 V dc)
Comments
Duration (ms) Note
Start
Xosc stabe
From power-up
<10
Internal power-on-
reset triggers
From power-up
<50
82583V samples
NVMT strapping
Determine NVM type
0
Flash
EEPROM
A
B
Figure 9.
82583V Initialization - Power-On Reset
40
Initialization—82583V GbE Controller
A
Read signature at word
0x12
~0
Bad
signature
Good
signature
Read signature at word
2K+0x12
Load sector 0 to Shadow
RAM
Set EEC.SHADV & clear
EEC.SEL1VAL
~0
~2.1
1
Load base area (0x00-
0x40) from Shadow
RAM
Bad
signature
Good
signature
82583V set to default
values
Load sector 1 to Shadow
RAM
Set EEC.SHADV & set
EEC.SEL1VAL
Set EEC.Auto_RD
0
~0.0032
2
~2.1
1
Clear Write Protection
Set Flash write status
enable and write status
0.008
3
C
Figure 10.
82583V Initialization - Flash Load
Notes:
1. A 4 KB sector is read in a single burst, so the packet overhead is negligible. The
rate is 4 KB x 8 bits / 15.625 Mb/s = 2.1 ms.
2. The shadow RAM is read at the rate of one word every ~3 clocks of 62.5 MHz, or
~50 ns per word. The 64 words are read in 3.2 ms.
3. Clear write protection is required for an SST* Flash only. The instruction codes that
are required to initiate are hardwired in the design as defined by SST 25xxx Flash
family: code 0x50 for write status enable and code 0x01 for status write. The
82583V writes a data of 0x00 to the status word which clears all protection.
Software accesses to the Flash are not executed until this step completes.
41
82583V GbE Controller—Initialization
B
Detect Address length of
1B or 2B based on
signature
~0
Bad
signature
Good
signature
82583V set to default
values
Load base area (0x00-
0x40) from EEPROM
Set EEC.Auto_RD
0
Set EEC.Auto_RD
~1.28
3
C
Figure 11.
82583V Initialization - EEPROM Load
Each word is read separately using a 5-byte command (1 byte instruction, 2 byte
address, and 2 byte data). Total time at 2 Mb/s is 64 words x 5 bytes x 8 bits/2 Mb/s =
1.28 ms. The rate is 20 μs per word.
42
Initialization—82583V GbE Controller
C
Enable wake up based
on NVM configuration
~0
Enable the PHY if
needed
PHY was inactive up to
now
11
D
Figure 12.
82583V PHY Initialization
Each PCIe register write takes ~20 PCIe clocks (31.25 MHz) per table entry <=>
640 ns per Dword. Each PHY register write takes those 20 clocks + 64 MDC cycles on
the MDIO interface (2.5 MHz) => 26.24 ms per Dword. Therefore, the total is 640 ns x
4 + 26.24 ms x 16 = 422 ms.
Each PCIe register write takes ~20 PCIe clocks (31.25 MHz) per table entry <=>
640 ns per Dword. Therefore, the bottleneck is the EEPROM at 40 ms per Dword. Each
PHY register write takes those 20 clocks + 64 MDC cycles on the MDIO interface (2.5
MHz) => 26.24 ms per Dword. Therefore, the bottleneck is the EEPROM at 40 ms per
Dword. The 16+4 entries take 20 Dwords x 40 ms = 0.8 s.
43
82583V GbE Controller—Initialization
D
PERST# is de-asserted
by the platform
PHY is powered down
~0
NVMT strapping is
sampled
Determine NVM type
~0
Flash
No NVM
EEPROM
Detect Address length of
1B or 2B based on
signature
Check valid Shadow and
signature
82583V set to default
values
Set EEC.Auto_RD
0
~0
~0
Load base area (0x00-
0x40) fromShadow
RAM
Load base area (0x00-
0x40) from EEPROM
Set EEC.Auto_RD
Set EEC.Auto_RD
~0.0032
2
~1.28
3
E
Figure 13.
82583V Initialization - NVM Load After PE_RST_N
44
Initialization—82583V GbE Controller
E
Enable the PHY
PHY was in power-down
during NVM load
11
Start PCIe link training
Must start < 20 µs after
PERST# de-assertion
PCIe link ready to
accept configuration
requests
Must start < 100 µs
after PERST#
Figure 14.
82583V Initialization - PHY and PCIe
45
82583V GbE Controller—Initialization
4.3.2
Timing Diagram
Power
txo
1
g
Xosc
tpp
g
2
Power-On-Reset
(internal)
tPVP
GL
7
PCIe reference clock
PERST#
tPWRGD
-CLK
6
8
tee
Ext.
Conf.
tee
3
Auto
Read
Auto
Read
Ext.
NVM Load
Conf.
4
tpgtrn
9
PHY State
tpgcfg
11
tpgres
Powered-down
Active / Down
13
12
10
PCIe Link up
Wake
L0
5
D-State
Dr
D0u
D0a
Figure 15.
Table 21.
Power-Up Timing Diagram
Notes to Power-Up Timing Diagram
Note
1
2
Xosc is stable txog after power is stable
Internal reset is released after all power supplies are good and tppg after Xosc
is stable.
An NVM read starts on the rising edge of the internal reset or Internal Power
On Reset#.
3
4
5
After reading the NVM, PHY might exit power down mode.
APM wake up might be enabled based on NVM contents.
The PCIe reference clock is valid tPWRGD-CLK before the de-assertion of
PE_RST_N (according to PCIe specification).
6
7
8
PE_RST_N is de-asserted tPVPGL after power is stable (according to PCIe
specification).
De-assertion of PE_RST_N causes the NVM to be re-read, asserts PHY power-
down, and disables Wake Up.
9
After reading the NVM, PHY exits power-down mode.
10
Link training starts after tpgtrn from PE_RST_N de-assertion.
A first PCIe configuration access might arrive after tpgcfg from PE_RST_N de-
assertion.
11
12
13
A first PCI configuration response can be sent after tpgres from PE_RST_N de-
assertion
Writing a 1b to the Memory Access Enable bit in the PCI Command register
transitions the device from D0u to D0 state.
46
Initialization—82583V GbE Controller
4.4
Global Reset (PE_RST_N, PCIe In-Band Reset)
4.4.1
Reset Sequence
Figure 16 and Figure 17 show the 82583V's sequence following global reset (PE_RST_N
de-assertion or PCIe in-band reset) and until the device is ready to accept host
commands.
Reset (PE_RST_# de-
assertion or in-band)
PHY is powered down
~0
NVMT strapping is
sampled
Determine NVM type
~0
Flash
No NVM
EEPROM
Detect Address length of
1B or 2B based on
signature
Check valid Shadow and
signature
82583V set to default
values
Set EEC.Auto_RD
0
~0
~0
Load base area (0x00-
0x40) from Shadow
RAM
Load base area (0x00-
0x40) from EEPROM
Set EEC.Auto_RD
Set EEC.Auto_RD
~0.0032
2
~1.28
3
A
Figure 16.
82583V Global Reset - NVM Load
47
82583V GbE Controller—Initialization
A
Enable the PHY
PHY was in power-down
during NVM load
11
Start PCIe link training
Must start < 80 µs after
PERST# de-assertion
PCIe link ready to
accept configuration
requests
Must start < 100 µs
after PERST#
Figure 17.
82583V Global Reset - PHY and PCIe
4.4.2
Timing Diagram
The following timing diagram shows the 82583V’s behavior through a PE_RST_N reset.
48
Initialization—82583V GbE Controller
PCIe reference
clock
tPWRGD-CLK
tclkpg
1
3
PERST#
4
tee
Ext.
Conf.
Auto
Read
2
NVM Load
tpgtrn
5
PHY State
Active
tpgcfg
tpgres
Active / Down
6
7
8
9
10
PCIe Link up
L0
L0
Wake
Any mode
APM
D0a
D-State
Dr
D0u
D0a
Figure 18.
Table 22.
Global Reset Timing Diagram
Notes to Global Reset Timing Diagram
Note
The system must assert PE_RST_N before stopping the PCIe reference clock. It
must also wait tl2clk after link transition to L2/L3 before stopping the reference
clock.
1
On assertion of PE_RST_N, the 82583V transitions to Dr state and the PCIe link
transition to electrical idle. The PHY state is defined by the wake configuration.
2
3
4
5
The system starts the PCIe reference clock tPWRGD-CLK before de-assertion
PE_RST_N.
De-assertion of PE_RST_N causes the NVM to be re-read, asserts PHY power-
down, and disables wake up.
After reading the NVM base area, PHY reset is de-asserted. APM wake might be
enabled.
Link training starts after the NVM was fully read (including extended
configuration if needed).
6
7
8
Link training starts after tpgtrn from PE_RST_N de-assertion.
A first PCIe configuration access might arrive after tpgcfg from PE_RST_N de-
assertion.
A first PCI configuration response can be sent after tpgres from PE_RST_N de-
assertion.
9
Writing a 1b to the Memory Access Enable bit in the PCI Command register
transitions the device from D0u to D0 state.
10
49
82583V GbE Controller—Initialization
4.5
Timing Parameters
Timing Requirements
4.5.1
The 82583V requires the following start-up and power state transitions.
Table 23.
Timing Requirements
Parameter
Description
Min
Max
Notes
txog
Xosc stable from power stable
10 ms
tPWRGD-
CLK
PCIe clock valid to PCIe power good
100 μs
100 ms
100 ms
10 ms
0 ns
-
-
According to PCIe specification.
According to PCIe specification.
According to PCIe specification.
Power rails stable to PCIe PE_RST_N
inactive
tPVPGL
Tpgcfg
td0mem
tl2pg
External PE_RST_N signal to first
configuration cycle.
Device programmed from D3h to D0
state to next device access
According to PCI power
management specification.
L2 link transition to PE_RST_N
assertion
According to PCIe specification.
According to PCIe specification.
L2 link transition to removal of PCIe
reference clock
tl2clk
100 ns
PE_RST_N assertion to removal of PCIe
reference clock
Tclkpg
Tpgdl
0 ns
According to PCIe specification.
According to PCIe specification.
PE_RST_N assertion time
100 μs
4.5.2
MDIO and NVM Semaphore
The MDIO and NVM semaphore mechanism resolved possible conflicts between
software and hardware access to the MDIO and NVM (the latter applies only to software
accesses through the EERD register). The mechanism does not block software accesses
to MDIO or the NVM, therefore programmers can enable software to use or ignore this
process at will. For example, software might track the hardware state through other
means (such as, a software state machine) and avoid any MDIO and NVM accesses
when hardware is in configuration states. However, hardware must comply with the
protocol.The EXTCNF_CTRL.MDIO/NVM SW Ownership bit, EXTCNF_CTRL.MDIO MNG
Ownership bit and the EXTCNF_CTRL.MDIO/NVM HW Ownership bit provide a
mechanism for software, manageability and hardware entities to arbitrate for accesses
to MDIO and NVM. Software arbitration for NVM accesses is only required when done
through the EERD register. A request for ownership is registered by writing a 1b into
the respective bit (software writes to the MDIO/NVM SW Ownership bit, manageability
writes to the MDIO MNG Ownership bit and hardware writes to the MDIO/NVM HW
Ownership). The requesting agent is granted access when the same bit is read as 1b
(access is not granted as long as the bit is 0b). The MDIO/NVM SW Ownership and the
MDIO/NVM HW Ownership bits are cleared on reset, while the MDIO MNG Ownership bit
is reset only by LAN_PWR_GOOD (or if the firmware clears it). The 82583 guarantees
that at any given time at most only one bit is 1b. Access is granted when a bit is
actually written with 1b and the other bits are 0b. Once the access completes, the
controlling agent must write a 0b to its ownership bit to enable accesses by the other
agents.
The 82583’s hardware sets the bit while loading the extended configuration area.
50
Initialization—82583V GbE Controller
4.6
Software Initialization Sequence
The following sequence of commands is typically issued to the device by the software
device driver in order to initialize the 82583V to normal operation. The major
initialization steps are:
1. Disable Interrupts - see Interrupts during initialization.
2. Issue Global Reset and perform General Configuration - see Global Reset and
General Configuration.
3. Setup the PHY and the link - see Link Setup Mechanisms and Control/Status Bit
Summary.
4. Initialize all statistical counters - see Initialization of Statistics.
5. Initialize Receive - see Receive Initialization.
6. Initialize Transmit - see Transmit Initialization.
7. Enable Interrupts - see Interrupts during initialization.
4.6.1
Interrupts During Initialization
Most drivers disable interrupts during initialization to prevent re-entrancy. Interrupts
are disabled by writing to the IMC register. Note that the interrupts need to be disabled
also after issuing a global reset, so a typical driver initialization flow is:
1. Disable interrupts
2. Issue a global reset
3. Disable interrupts (again)
4. …
After the initialization completes, a typical driver enables the desired interrupts by
writing to the IMS register.
4.6.2
Global Reset and General Configuration
Device initialization typically starts with a global reset that puts the device into a known
state and enables the software device driver to continue the initialization sequence.
Several values in the Device Control (CTRL) register need to be set at power up or after
a device reset for normal operation.
• Full duplex should be set per interface negotiation (if done in software), or is set by
the hardware if the interface is auto-negotiating. This is reflected in the Device
Status register in the auto-negotiating case. A default value is loaded from the
NVM.
• Speed is determined via auto-negotiation by the PHY, or forced by software if the
link is forced. Status information for speed is also readable in STATUS.
• ILOS should normally be set to 0b.
If using XOFF flow control, program the FCAH, FCAL, and FCT registers. If not, they
should be written with 0x0.
GCR bit 22 should be set to 1b by software during initialization.
51
82583V GbE Controller—Initialization
4.6.3
Link Setup Mechanisms and Control/Status Bit Summary
4.6.3.1
PHY Initialization
Refer to the PHY documentation for the initialization and link setup steps. The device
driver uses the MDIC register to initialize the PHY and setup the link.
4.6.3.2
MAC/PHY Link Setup
This section summarizes the various means of establishing proper MAC/PHY link
setups, differences in MAC CTRL register settings for each mechanism, and the relevant
MAC status bits. The methods are ordered in terms of preference (the first mechanism
being the most preferred).
• MAC settings automatically based on duplex and speed resolved by PHY.
(CTRL.FRCDPLX = 0b, CTRL.FRCSPD = 0b, CTRL.ASDE = 0b)
— CTRL.FD - Don't care; duplex setting is established from PHY's internal
indication to the MAC (FDX) after PHY has auto-negotiated a successful link-up.
— CTRL.SLU - Must be set to 1b by software to enable communications between
MAC and PHY.
— CTRL.RFCE - Must be set by software after reading flow control resolution from
PHY registers.
— CTRL.TFCE - Must be set by software after reading flow control resolution from
PHY registers.
— CTRL.SPEED - Don't care; speed setting is established from PHY's internal
indication to the MAC (SPD_IND) after PHY has auto-negotiated a successful
link-up.
— STATUS.FD - Reflects the actual duplex setting (FDX) negotiated by the PHY
and indicated to the MAC.
— STATUS.LU - Reflects link indication (LINK) from the PHY qualified with
CTRL.SLU (set to 1b).
— STATUS.SPEED - Reflects actual speed setting negotiated by the PHY and
indicated to the MAC (SPD_IND).
• MAC duplex setting automatically based on resolution of PHY, software-
forced MAC/PHY speed. (CTRL.FRCDPLX = 0b, CTRL.FRCSPD = 1b,
CTRL.ASDE = don't care)
— CTRL.FD - Don't care; duplex setting is established from PHY's internal
indication to the MAC (FDX) after PHY has auto-negotiated a successful link-up.
— CTRL.SLU - Must be set to 1b by software to enable communications between
the MAC and PHY.
— CTRL.RFCE - Must be set by software after reading flow control resolution from
PHY registers.
— CTRL.TFCE - Must be set by software after reading flow control resolution from
the PHY registers.
— CTRL.SPEED - Set by software to desired link speed (must match speed setting
of PHY).
— STATUS.FD - Reflects the actual duplex setting (FDX) negotiated by the PHY
and indicated to MAC.
— STATUS.LU - Reflects link indication (LINK) from the PHY qualified with
CTRL.SLU (set to 1b).
— STATUS.SPEED - Reflects MAC forced speed setting written in CTRL.SPEED.
52
Initialization—82583V GbE Controller
• MAC duplex and speed settings forced by software based on resolution of
PHY. (CTRL.FRCDPLX = 1b, CTRL.FRCSPD = 1b, CTRL.ASDE = don't care)
— CTRL.FD - Set by software based on reading PHY status register after the PHY
has auto-negotiated a successful link-up.
— CTRL.SLU. - Must be set to 1b by software to enable communications between
the MAC and PHY.
— CTRL.RFCE - Must be set by software after reading flow control resolution from
the PHY registers.
— CTRL.TFCE - Must be set by software after reading flow control resolution from
the PHY registers.
— CTRL.SPEED - Set by software based on reading PHY status register after the
PHY has auto-negotiated a successful link-up.
— STATUS.FD - Reflects the MAC forced duplex setting written to CTRL.FD.
— STATUS.LU - Reflects link indication (LINK) from the PHY qualified with
CTRL.SLU (set to 1b).
— STATUS.SPEED - Reflects MAC forced speed setting written in CTRL.SPEED.
• MAC/PHY duplex and speed settings both forced by software (fully-forced
link setup). (CTRL.FRCDPLX = 1b, CTRL.FRCSPD = 1b, CTRL.SLU = 1b)
— CTRL.FD - Set by software to desired full-/half- duplex operation (must match
duplex setting of the PHY).
— CTRL.SLU - Must be set to 1b by software to enable communications between
the MAC and PHY. The PHY must also be forced/configured to indicate positive
link indication (LINK) to the MAC.
— CTRL.RFCE - Must be set by software to the desired flow-control operation
(must match flow-control settings of the PHY).
— CTRL.TFCE - Must be set by software to the desired flow-control operation
(must match flow-control settings of the PHY).
— CTRL.SPEED - Set by software to desired link speed (must match speed setting
of the PHY).
— STATUS.FD - Reflects the MAC duplex setting written by software to CTRL.FD.
— STATUS.LU - Reflects 1b (positive link indication LINK from PHY qualified with
CTRL.SLU).
Note:
Since both CTRL.SLU and the PHY link indication LINK are forced, this bit set does not
guarantee that operation of the link has been truly established.
— STATUS.SPEED - Reflects MAC forced speed setting written in CTRL.SPEED.
4.6.4
Initialization of Statistics
Statistics registers are hardware-initialized to values as detailed in each particular
register's description. The initialization of these registers begins at transition to D0
active power state (when internal registers become accessible, as enabled by setting
the Memory Access Enable field of the PCIe Command register), and is guaranteed to
complete within 1 ms of this transition. Access to statistics registers prior to this
interval might return indeterminate values.
All of the statistical counters are cleared on read and a typical software device driver
reads them (thus making them zero) as a part of the initialization sequence.
53
82583V GbE Controller—Initialization
4.6.5
Receive Initialization
Program the receive address register per the station address. This can come from the
NVM or from any other means, for example, on some systems, this comes from the
system EEPROM not the NVM on a Network Interface Card (NIC).
Set up the Multicast Table Array (MTA) per software. This generally means zeroing all
entries initially and adding in entries as requested.
Program the interrupt mask register to pass any interrupt that the software device
driver cares about. Suggested bits include RXT, RXO, RXDMT and LSC. There is no
reason to enable the transmit interrupts.
Program RCTL with appropriate values. If initializing it at this stage, it is best to leave
the receive logic disabled (EN = 0b) until the receive descriptor ring has been
initialized. If VLANs are not used, software should clear the VFE bit. Then there is no
need to initialize the VFTA array. Select the receive descriptor type. Note that if using
the header split RX descriptors, tail and head registers should be incremented by two
per descriptor.
4.6.5.1
Initialize the Receive Control Register
To properly receive packets requires simply that the receiver is enabled. This should be
done only after all other setup is accomplished. If software uses the Receive Descriptor
Minimum Threshold Interrupt, that value should be set.
Do the following for the receive queue:
• Allocate a region of memory for the receive descriptor list.
• Receive buffers of appropriate size should be allocated and pointers to these
buffers should be stored in the descriptor ring.
• Program the descriptor base address with the address of the region.
• Set the length register to the size of the descriptor ring.
• If needed, program the head and tail registers. Note: the head and tail pointers are
initialized (by hardware) to zero after a power-on or a software-initiated device
reset.
• The tail pointer should be set to point one descriptor beyond the end.
4.6.6
Transmit Initialization
Program the TXDCTL register with the desired TX descriptor write-back policy.
Suggested values are:
• GRAN = 1b (descriptors)
• WTHRESH = 1b
• All other fields 0b.
Program the TCTL register. Suggested configuration:
• CT = 0x0F (16d collision)
• COLD: HDX = 511 (0x1FF); FDX = 63 (0x03F)
• PSP = 1b
• EN=1b
• All other fields 0b
54
Initialization—82583V GbE Controller
Do the following for the transmit queue:
• Allocate a region of memory for the transmit descriptor list.
• Program the descriptor base address with the address of the region.
• Set the length register to the size of the descriptor ring.
• If needed, program the head and tail registers.
Note:
The head and tail pointers are initialized (by hardware) to zero after a power-on or a
software-initiated device reset.
Program the TIPG register with the following (decimal) values to get the minimum legal
IPG:
• IPGT = 8
• IPGR1 = 2
• IPGR2 = 10
Note:
IPGR1 and IPGR2 are not needed in full-duplex, but it is easier to always program them
to the values listed.
Initialize the transmit descriptor registers (TDBAL, TDBAH, TDL, TDH, and TDT).
55
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.0
Non-Volatile Memory (NVM) Map
The NVM contains two regions located at fixed addresses and various regions located at
programmable addresses throughout the physical NVM space.
The NVM base area resides at word addresses 0x00-0x3F. All defined fields are fixed,
while reserved words might be used by some programmable areas. The base area is
present in the NVM in all system configurations.
The programmable area is as follows:
Additional configuration for the PHY is located in the extended configuration area. The
extended configuration pointer indicates the location of the extended configuration
area. A value of 0x0000 means that the extended configuration area is disabled. This
should be the case for the 82583V.
Note:
The NVM image must fit the specific NVM part being used. Special attention should be
paid to NVM words and fields that vary, like the examples of NVMTYPE or NVSIZE. For
the latest 82583V NVM images, contact your Intel representative.
5.1
Basic Configuration Table
Table 24 lists the NVM map for the 0x00-0x3F address range.
NVM Map of Address Range 0x00-0x3F
Table 24.
Word
0x00
0x01
0x02
Used By
HW
HW
HW
15
8
7
0
Ethernet Address Byte 2
Ethernet Address Byte 4
Ethernet Address Byte 6
Ethernet Address Byte 1
Ethernet Address Byte 3
Ethernet Address Byte 5
0x03
0x04
SW
SW
SW
Compatibility High
Compatibility Low
0x05
Image Version Information 1
0x06
0x07
Compatibility High
Compatibility Low
0x08
0x09
PBA, Byte 1
PBA, Byte 3
PBA, Byte 2
PBA, Byte 4
SW
0x0A
0x0B
0x0C
0x0D
0x0E
0x0F
0x10
HW
HW
HW
HW
HW
HW
HW
Init Control 1
Subsystem ID
Subsystem Vendor ID
Device ID
Reserved
Init Control 2
NVM Word 0
56
Non-Volatile Memory (NVM) Map—82583V GbE Controller
Word
0x11
Used By
15
8
7
0
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
SW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
HW
SW
SW
NVM Word 1
0x12
0x13
0x14
0x15
0x16
0x17
0x18
0x19
0x1A
0x1B
0x1C
0x1D
0x1E
0x1F
0x20
0x21
0x22
0x23
0x24
0x25
0x26
0x27
0x28
0x29
0x2A
0x2B
0x2C
0x2D
0x2E
0x2F
0x30-0x3E
0x3F
NVM Word 2
Reserved
Reserved
Reserved
Reserved
PCIe Electrical Idle Delay
PCIe Init Configuration 1
PCIe Init Configuration 2
PCIe Init Configuration 3
PCIe Control
PHY Configuration
LEDCTL 1
Reserved
Device REV ID
LEDCTL 0 2
Flash Parameters
Flash LAN Address
LAN Power Consumption
SW Flash Vendor Detection
Init Control 3
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
Least Significant Word of Firmware ID
Most Significant Word of Firmware ID
Reserved
Reserved
VPD Pointer
Software Section
Software Checksum, Words 0x00 Through 0x3F
57
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.1.1
Hardware Accessed Words
This section describes the NVM words that are loaded by the 82583V hardware.
5.1.1.1
Ethernet Address (Words 0x00-0x02)
The Ethernet Individual Address (IA) is a 6-byte field that must be unique for each
Network Interface Card (NIC), and thus unique for each copy of the NVM image. The
first three bytes are vendor specific - for example, the IA is equal to [00 AA 00] or [00
A0 C9] for Intel products. The value from this field is loaded into the Receive Address
Register 0 (RAL0/RAH0).
For the purpose of this specification, the IA byte numbering convention is indicated
below:
IA Byte / Value
Vendor
1
2
3
4
5
6
Intel Original
Intel New
00
00
AA
A0
00
C9
variable
variable
variable
variable
variable
variable
5.1.1.2
Initialization Control Word 1 (word 0x0A)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
15
14
Reserved
Reserved
0b
0b
Reserved.
Reserved.
Reserved.
0b
0b
13:12 Reserved
00b
00b
Default setting for the Force Speed bit in the Device Control
register (CTRL[11]).
11
FRCSPD
1b
0b
10
9
FD
1b
1b
0b
0b
1b
1b
0b
1b
0b
0b
1b
1b
Default setting for duplex setting. Mapped to CTRL[0].
Reserved, must be set to 1b.
Reserved, must be set to 0b.
Reserved, Must be set to 0b.
Reserved.
Reserved
Reserved
Reserved
Reserved
Reserved
8
7
6
5
Reserved.
Default setting for the Loss-of-signal polarity setting for
CTRL[7].
4
ILOS
0b
0b
3
2
Reserved
Reserved
Load
Subsystem 1b
IDs
1b
0b
1b
0b
Reserved, must be set to 1b.
Reserved, must be set to 0b.
This bit, when equal to 1b, indicates that the device is to
load its PCIe subsystem ID and subsystem vendor ID from
the NVM (words 0x0B and 0x0C).
1
0
1b
1b
Load
Device ID
This bit, when equal to 1b, indicates that the device is to
load its PCIe device ID from the NVM (word 0x0D).
1b
58
Non-Volatile Memory (NVM) Map—82583V GbE Controller
5.1.1.3
5.1.1.4
5.1.1.5
5.1.1.6
Subsystem ID (Word 0x0B)
If the load subsystem IDs in word 0x0A is set, this word is loaded to initialize the
subsystem ID. The default value is 0x0.
Subsystem Vendor ID (Word 0x0C)
If the load subsystem IDs in word 0x0A is set, this word is loaded to initialize the
subsystem vendor ID. The default value is 0x8086.
Device ID (Word 0x0D)
If the load vendor/device IDs in word 0x0A is set, this word is loaded to initialize the
device ID of the function. The default value is 0x150C for the 82583V.
Initialization Control Word 2 (Word 0x0F)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
APM PME#
Enable
Initial value of the Assert PME On APM Wakeup bit in
the Wake Up Control register (WUC.APMPME).
15
0b
1b
14:13
12
Reserved
NVMTYPE
00b
0b
00b
0b
Reserved.
0b = EEPROM.
1b = Flash.
NVM size [bytes]
Equals 128 * 2 ** NVSIZE. (When NVM=Flash the
NVSIZE should be >= 9 ‡. Therefore, the minimal
supported Flash size is 64 KB).
Note: A value of 1111b is reserved.
Following are all possible NVSIZE values and their
corresponding NVM sizes (in both bytes and bits):
0000b = 128 B / 1 Kb.
0001b = 256 B / 2 Kb.
0010b = 0.5 KB / 4 Kb.
0011b = 1 KB / 8 Kb.
0100b = 2 KB / 16 Kb.
0101b = 4 KB / 32 Kb.
0110b = 8 KB / 64 Kb.
0111b = 16 KB / 128 Kb.
1000b = 32 KB / 256 Kb.
1001b = 64 KB / 0.5 Mb.
1010b = 128 KB / 1 Mb.
1011b = 265 KB / 2 Mb.
1100b = 0.5 MB / 4 Mb.
1101b = 1 MB / 8 Mb.
1110b = 2 MB / 16 Mb.
1111b = Reserved.
11:8
NVSIZE
0000b
0000b
7
6
Reserved
Reserved
0b
1b
0b
1b
Reserved.
Reserved.
59
82583V GbE Controller—Non-Volatile Memory (NVM) Map
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
5
4
3
2
1
0
Reserved
0b
1b
1b
0b
0b
0b
0b
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved
Reserved
Reserved
Reserved
Reserved
1b
1b
0b
0b
0b
5.1.1.7
NVM Protected Word 0 - NVP0 (Word 0x10)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
15:8
7:0
Reserved
Reserved
0x0
0x0
0x0
0x0
Reserved.
Reserved.
5.1.1.8
NVM Protected Word 1 - NVP1 (Word 0x11)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
Defines the instruction code for the block erase used by the
82583V. The erase block size is defined by the SECSIZE
field in address 0x12.
15:8
FSECER
0x20
0x20
7:1
0
Reserved
0x0
1b
0x0
1b
Reserved.
RAM_PWR_
SAVE_EN
When set to 1b, enables reducing power consumption by
clock gating the 82583V RAMs.
60
Non-Volatile Memory (NVM) Map—82583V GbE Controller
5.1.1.9
NVM Protected Word 2 - NVP2 (Word 0x12)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
Signature
The 8-bit Signature field indicates to the device that
there is a valid NVM present. If the Signature field does
not equal 0x7E then the default values are used for the
device configuration.
15:8
SIGN
0x7E
0x7E
7
6
5
4
Reserved
Reserved
Reserved
Reserved
0b
0b
0b
0b
0b
1b
1b
1b
Reserved, must be set to 0b.
Reserved.
Reserved.
Reserved.
The SECSIZE defines the Flash sector erase size as
follows:
00b = 256 bytes.
01b = 4 KB.
10b = Reserved.
11b = Reserved.
3:2
1:0
SECSIZE
Reserved
01b
00b
01b
00b
Reserved, must be set to 00b.
5.1.1.10
Extended Configuration Word 1 (Word 0x14)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
15:13
12
Reserved
Reserved
Reserved
100b
100b
1b
Reserved.
Reserved.
Reserved.
0b
11:0
0x0
0x0
5.1.1.11
Extended Configuration Word 2 (Word 0x15)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
15:8
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
0x0
0x0
1b
1b
0b
0b
1b
0b
0b
0b
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
7
6
5
4
3
2
1
0
1b
1b
0b
0b
1b
0b
0b
0b
61
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.1.1.12
Extended Configuration Word 3 (Word 0x16)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
15:8
7:0
Reserved
Reserved
0x0
0x0
0x0
0x0
Reserved.
Reserved.
5.1.1.13
PCIe Electrical Idle Delay (Word 0x17)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
15:14 Reserved
00b
00b
1b
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
13
12:8
7:3
2
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
1b
0x7
0x0
1b
0x7
0x0
1b
1
0b
0b
0
0b
0b
5.1.1.14
PCIe Init Configuration 1 Word (Word 0x18)
NVM Image
Bit
15
Name
Reserved
Hardware Default
Description
Setting
0b
0b
110b (32 μs-64 μs)
L1_Act_Acc_Latency 110b (32 μs-64 μs) 110b (32 μs-64 μs)
Reserved, must be set to 0b.
L1 active exit latency for the
configuration space.
14:12
11:9
8:6
L1_Act_Ext_Latency 110b (32μs-64 μs)
L1 active acceptable latency for
the configuration space.
L0s acceptable latency for the
configuration space.
L0s_Acc_Latency
011b (512 ns)
001b
011b (512 ns)
001b
L0s exit latency for active state
power management (separated
reference clock) – (latency
between 64 ns – 128 ns).
5:3
2:0
L0s_Se_Ext_Latency
L0s exit latency for active state
power management (common
reference clock) – (latency
between 64 ns – 128 ns).
L0s_Co_Ext_Latency
001b
001b
62
Non-Volatile Memory (NVM) Map—82583V GbE Controller
5.1.1.15
PCIe Init Configuration 2 Word (Word 0x19)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
DLLP timer
enable
15
0b
0b
When set, enables the DLLP timer counter.
14
13
12
Reserved
Reserved
SER_EN
0b
1b
0b
0b
1b
1b
Reserved, must be set to 0b.
Reserved, must be set to 1b.
When set to 1b, the serial number capability is enabled.
Extra NFTS (number of fast training signal), which is
added to the original requested number of NFTS (as
requested by the upstream component).
11:8
7:0
ExtraNFTS
NFTS
0x1
0x1
0x50
0x50
Number of special sequence for L0s transition to L0.
5.1.1.16
PCIe Init Configuration 3 Word (Word 0x1A)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
When set to 1b, this bit enables the PHY to be a master
(upstream component/cross link functionality).
15
Master_Enable 0b
0b
0b
Scrambling Disable
When set to 1b, this bit disables the PCIe LFSR scrambling.
14
13
Scram_dis
0b
0b
ACK/NAK Scheme
0b = Scheduled for transmission following any TLP.
1b = Scheduled for transmission according to time outs
specified in the PCIe specification.
Ack_Nak_Sch
0b
0b
Cache Line Size
0b = 64 bytes.
1b = 128 bytes.
Note: The value loaded must be equal to the actual cache
line size used by the platform, as configured by system
software.
12
Cache_Lsize
0b
11:10
9
PCIE_Cap
IO_Sup
01b
1b
01b
1b
PCIe Capability Version
I/O Support (Effect I/O BAR Request)
0b = I/O is not supported.
1b = I/O is supported.
Default Packet Size
0b = 128 bytes.
1b = 256 bytes.
8
Packet_Size
1b
1b
7
6
5
4
Reserved
Reserved
Reserved
Reserved
0b
0b
0b
0b
0b
0b
1b
1b
Reserved, must be set to 0b.
Reserved, must be set to 0b.
Reserved.
Reserved.
63
82583V GbE Controller—Non-Volatile Memory (NVM) Map
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
Determines support for Active State Link Power
Management (ASLPM). Loaded into the PCIe Active State
Link PM Support register.
Note: Changing the default value of this field might affect
certain power savings features of the 82583V. However, in
some applications, it might be necessary to change this
Act_Stat_PM_
Sup
3:2
0x3
0x3
®
value as explained in the Intel 82583V Gigabit Ethernet
Controller Specification Update. Please refer to Erratum #9
for more details.
When set, the 82583V uses the PCIe reference clock
supplied on the connector (for add-in solutions).
1
0
Slot_Clock_Cfg 1b
Loop back
1b
0b
Check Polarity Inversion in Loop-Back Master Entry
During normal operation polarity is adjusted during link up.
When this bit is set, the receiver re-checks the polarity of
Rx-data and then inverts it accordingly, when entering a
near-end loopback. When cleared, polarity is not re-
checked after link up.
polarity
0b
inversion
5.1.1.17
PCIe Control (Word 0x1B)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
PCIE_RX_
Valid
Force receiver presence detection. When set, the 82583V
overrides the receiver (partner) detection status.
15
0b
1b
0b
0b
0b
1b
0b
0b
MSB [2] of period in L0s state before transitioning into an L1
state (lower bits are in bits [1:0].
Recommended setting: {14, 1:0} = 011b – 32 μs.
Latency_To_E
nter_L1
14
13
12
PCIE Down
Reset Disable
Disable a core reset when the PCIe link goes down.
When cleared, LTSSM complies with the SlimPIPE
specification (power mode transition). When set, LTSSM
behaves as in previous generations.
PCIE_LTSSM
When this bit is set, the LTSSM recovery states always
progress towards link up (force a good recovery when a
recovery occurs).
Good
Recovery
11
10
0b
1b
0b
1b
Disable leaky bucket mechanism in the PCIe PHY. Disabling
this mechanism holds the link from going to recovery retrain
in case of disparity errors.
Leaky Bucket
Disable
9:7
6
Reserved
0x0
0b
0x0
0b
Reserved.
Reserved
Reserved.
5
L2 Disable
Skip Disable
0b
0b
Disable the link from entering L2 state.
Disable skip symbol insertion in the elastic buffer.
4
0b
0b
64
Non-Volatile Memory (NVM) Map—82583V GbE Controller
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
3
2
Reserved
0b
1b
Reserved.
Electrical Idle Mask
If set to 1b, disables the check for illegal electrical idle
sequence (such as, eidle ordered set without common mode
and vise versa), and accepts any of them as the correct eidle
sequence.
Note: The specification can be interpreted so that idle
ordered set is sufficient for transition to power management
states. The use of this bit allows an acceptance of such
interpretation and avoids the possibility of correct behavior
to be understood as illegal sequences.
Electrical
IDLE
0b
0b
Period in L0s state before transitioning into an L1 state bits
[1:0].
00b = 64 μs.
01b = 256 μs.
10b = 1 ms.
11b = 4 ms.
Latency_To_E
nter_L1
1:0
0x3
0x3
65
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.1.1.18
LED 1 Configuration Defaults/PHY Configuration (Word 0x1C)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
15
Reserved
0b
0b
0b
1b
Reserved.
When set, 1000 Mb/s operation is disabled in all power
modes.
14
13
Giga Disable
Reserved
0b
1b
Reserved.
When set, the PHY operates in class A mode instead of class
B mode. This mode only applies for 1000BASE-T operation.
10BASE-T and 100BASE-T operation continue to run in Class
B mode by default, regardless of this signal value.
12
Class AB
0b
0b
Disable 1000
in non-D0a
11
10
1b
1b
1b
1b
Disables 1000 Mb/s operation in non-D0a states.
Low Power Link Up
Enables decrease in link speed in non-D0a states when the
power policy and power management state dictate so.
LPLU
D0 Low Power Link Up
9
D0LPLU
0b
0b
Enables decrease in link speed in D0a state when the power
policy and power management state dictate so.
8
7
Reserved
1b
1b
1b
1b
Reserved.
Initial Value of LED1_BLINK Field
0b = Non-blinking
LED1 Blink
Initial Value of LED1_IVRT Field
0b = Active-low output
6
5
LED1 Invert
0b
0b
0b
0b
LED1 Blink Mode
0b = Blinks at 200 ms on and 200 ms off.
1b = Blinks at 83 ms on and 83 ms off.
LED1 Blink
Mode
4
Reserved
0b
0b
Reserved.
Initial value of the LED1_MODE field specifying what event/
state/pattern is displayed on the LED1 (ACTIVITY) output. A
value of 0011b (0x3) indicates the ACTIVITY state.
3:0
LED1 Mode
0x4
0x4
66
Non-Volatile Memory (NVM) Map—82583V GbE Controller
5.1.1.19
Reserved Word 0x1D
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
15:9
8
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
0x0
0x0
1b
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
1b
0b
0b
0b
0x0
7
0b
6
0b
5
0b
4:0
0x0
5.1.1.20
Device Rev ID (Word 0x1E)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
15
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
0b
1b
0b
0b
0b
0b
1b
1b
0x0
1b
1b
1b
1b
0b
0b
0b
0b
0x0
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
14
13
12
11
10
9
8
7:0
67
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.1.1.21
LED 0, 2 Configuration Defaults (Word 0x1F)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
Initial Value of LED2_BLINK Field
0b = Non-blinking.
15
LED2 Blink
0b
0b
0b
0b
Initial Value of LED2_IVRT Field
0b = Active-low output.
14
LED2 Invert
LED2 Blink Mode
0b = Blinks at 200 ms on and 200 ms off.
1b = Blinks at 83 ms on and 83 ms off.
LED2 Blink
Mode
1
1
13
12
0b
0b
Reserved
0b
0b
Reserved.
Initial value of the LED2_MODE field specifying what event/
state/pattern is displayed on LED2 (LINK_100) output.
A value of 0110b (0x6) causes this to indicate 100 Mb/s
operation.
11:8
LED2 Mode
0x7
0x7
Initial Value of LED0_BLINK Field
0b = Non-blinking.
7
6
LED0 Blink
0b
0b
0b
0b
Initial Value of LED0_IVRT Field
0b = Active-low output.
LED0 Invert
LED0 Blink Mode
0b = Blinks at 200 ms on and 200 ms off.
1b = Blinks at 83 ms on and 83 ms off.
LED0 Blink
Mode
1
1
5
0b
0b
4
Reserved
0b
0b
Reserved, set to 0b.
Initial value of the LED0_MODE field specifying what event/
state/pattern is displayed on the LED0 (LINK_UP) output. A
value of 0010b (0x2) causes this to indicate LINK_UP state.
3:0
LED0 Mode
0x6
0x6
1. These bits are read from the NVM.
68
Non-Volatile Memory (NVM) Map—82583V GbE Controller
5.1.1.22
Flash Parameters - FLPAR (Word 0x20)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
Defines the instruction code for the Flash device erase. A value
of 0x00 means that the device does not support the device
erase.
15:8
7:6
FDEVER
0x60
0x0
0x60
0x0
Reserved
Reserved.
SST Flash Not
When set to 0b, indicates an SST FLASH type: write access to
the Flash is limited to 1 byte at a time and it is required to clear
write protection at power up. When set to 1b, burst write access
to the Flash is enabled up to 256 bytes and it is not required to
clear write protection at power up.
5
FLSSTn
0b
0b
Very Long Cycle Indication
When set to 1b, the LONGC indicates to the 82583V that a Flash
write instruction is considered a very long instruction. When set
to '0b, the LONGC indicates that a write cycle to the Flash is not
considered a very long cycle.
4
LONGC
0b
0b
3:0
Reserved
0x0
0x0
Reserved.
5.1.1.23
Flash LAN Address - FLANADD (Word 0x21)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
15
DISLFB
0b
0b
1b = Disables the LAN Flash BAR.
14:12
LANSIZE
0x0
0x0
0x0
LAN boot expansion window size = 2 KB * 2 ** LANSIZE.
LAN Flash Address
Defines the location of the LAN boot expansion ROM in the
physical Flash device as defined in the following equation:
11:8
LBADD
0x0
Word Address = 4 KB * (LBADD + PEND).
1b = Disables the LAN expansion boot ROM BAR.
Reserved.
7
DISLEXP
Reserved
Reserved
0b
1b
6:1
0
0x0
0b
0x0
0b
Reserved.
69
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.1.1.24
LAN Power Consumption (Word 0x22)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
The value in this field is reflected in the PCI Power Management
Data register of the function for D0 power consumption and
dissipation (Data_Select = 0 or 4). Power is defined in 100 mW
units. The power also includes the external logic required for the
LAN function.
LAN D0
Power
15:8
7:5
0xF
0x0
0xF
0x0
Reserved
Reserved.
The value in this field is reflected in the PCI Power Management
Data Register of the function for D3 power consumption and
dissipation (Data_Select = 3 or 7). Power is defined in 100 mW
units. The power also includes the external logic required for the
function. The most significant bits in the Data register that
reflects the power values are padded with zeros.
LAN D3
Power
4:0
0x4
0x4
5.1.1.25
Flash Software Detection Word (Word 0x23)
The setting of this word to 0xFFFF enables detection of the flash vendor by software
tools.
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
Checksum Validity Indication
0b = Checksum should be corrected by software tools.
1b = Checksum may be considered valid.
Checksum
Validity
15
0x0
0x0
1b
Enable/disable bit for Deep Smart Power Down
functionality.
0b = Enable Deep Smart Power Down (DSPD).
1b = Disable DSPD (default).
Deep Smart
Power Down
14
1b
13:8
7:0
Reserved
0x3F
0xFF
0x3F
0xFF
Reserved.
Flash Vendor
Detect
This word must be set to 0xFF.
70
Non-Volatile Memory (NVM) Map—82583V GbE Controller
5.1.1.26
Initialization Control 3 (Word 0x24)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
15
Reserved
Reserved
Reserved
Reserved
Reserved
0b
0b
1b
0b
0b
1b
Reserved.
Reserved.
Reserved.
Reserved.
Reserved.
14
13
12
11
1b
1b
0b
1b
Initial value of Advanced Power Management Wake Up
Enable in the Wake Up Control (WUC.APME) register.
Mapped to CTRL[6] and to WUC[0].
10
APM Enable
0b
1b
9
Reserved
Reserved
Reserved
0b
1b
Reserved.
Reserved.
Reserved.
No PHY Reset
8
0b
1b
7:1
0x0
0x0
When set to 1b, this bit prevents the PHY reset signal and
the power changes reflected to the PHY according to the
MANC.Keep_PHY_Link_Up value.
0
No_Phy_Rst
1b
1b
71
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.1.2
Software Accessed Words
5.1.2.1
Compatibility Fields (Words 0x03 - 0x07)
This section describes the meaningful NVM words in the basic configuration space that
are used by software at word addresses 0x03 to 0x07.
5.1.2.2
Compatibility Bytes (Word 0x03)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
15:13 Reserved
000b
000b
0b
Reserved.
Reserved.
12
Reserved
0b
0b
LOM or NIC
0b = NIC.
1b = LOM.
11
LOM
0b
1b
0b
0b
Server NIC
0b = Client.
1b = Server.
10
9
Server NIC 1b
Client NIC
0b = Server.
1b = Client.
Client NIC
1b
Retail Card
0b = Retail.
1b = OEM.
8
Retail Card 0b
7:6
5
Reserved
Reserved
Reserved
Reserved
00b
00b
1b
Reserved.
1b
0b
0b
Reserved.
4
0b
Reserved
3
0b
Reserved.
PCI Bridge NOT Present
2
PCI Bridge
Reserved
1b
0b
0b = PCI bridge NOT present.
1b = PCI bridge present.
1:0
00b
00b
Reserved.
5.1.2.3
Compatibility Bytes (Word 0x04)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
15:12 Reserved
0xF
0xF
Reserved.
11:8
7:4
LED 2 Control
0x7
0x4
0x6
0x7
0x4
0x6
Control for LED 2 - LINK_1000.
Control for LED 1 - LINK/ACTIVITY.
Control for LED 0 - LINK_100.
LED 1Control
LED 0 Control
3:0
72
Non-Volatile Memory (NVM) Map—82583V GbE Controller
5.1.2.3.1
5.1.2.3.2
Compatibility Byte (Word 0x05)
NVM
Image
Setting
Hardware
Default
Bits
Name
Description
15:0
Reserved
0x10A0
0x10A0
Reserved.
Compatibility Bytes (Word 0x06 - 0x07)
NVM
Hardware
Default
Bits
Name
Image
Description
Setting
15:0
Reserved
0xFFFF
0xFFFF
Reserved.
5.1.2.4
PBA Number (Word 0x08 and 0x09)
The nine-digit Printed Board Assembly (PBA) number used for Intel manufactured
Network Interface Cards (NICs) is stored in the EEPROM.
Note that through the course of hardware ECOs, the suffix field is incremented. The
purpose of this information is to enable customer support (or any user) to identify the
revision level of a product.
Network driver software should not rely on this field to identify the product or its
capabilities.
Current PBA numbers have exceeded the length that can be stored as hex values in
these two words. For these PBA numbers the high word is a flag (0xFAFA) indicating
that the PBA is stored in a separate PBA block. The low word is a pointer to a PBA block.
PBA Number
Word 0x08
Word 0x09
Pointer to PBA Block
G23456-003
FAFA
The PBA block is pointed to by word 0x09.
Word Offset
Description
End UserReserved
0x0
Length in words of the PBA block (default 0x6).
PBA number stored in hexadecimal ASCII values.
No
No
0x1 ... 0x5
The PBA block contains the complete PBA number including the dash and the first digit
of the 3-digit suffix. For example:
Word
Offset 0
Word
Offset 1
Word
Offset 2
Word
Offset 3
Word
Offset 4
Word
Offset 5
PBA Number
G23456-003
0006
4732
3334
3536
2D30
3033
Older PBA numbers starting with (A,B,C,D,E) are stored directly in words 0x08 and
0x09. The dash itself is not stored nor is the first digit of the 3-digit suffix, as it is
always 0b for relevant products.
73
82583V GbE Controller—Non-Volatile Memory (NVM) Map
PBA Number
Byte 1
Byte 2
Byte 3
Byte 4
123456-003
12
34
56
03
5.1.2.5
5.1.2.6
PXE Words (Words 0x30:0x3E)
Words 0x30 through 0x3E are reserved for software and are used by IBA/PXE software.
iSCSI Boot Configuration Start Address (Word 0x3D)
Bit
Name
Default
Description
15:0
Address
0x0
NVM word address of the iSCSI boot configuration structure starting point.
5.1.2.6.1
Boot Agent Main Setup Options (Word 0x30)
The boot agent software configuration is controlled by the NVM with the main setup
options stored in word 0x30. These options are those that can be changed by using the
Control-S setup menu or by using the IBA Intel Boot Agent utility. Note that these
settings only apply to Boot Agent software.
74
Non-Volatile Memory (NVM) Map—82583V GbE Controller
Table 19.
Boot Agent Main Setup Options
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
15:13
12
Reserved
000b
Reserved, set to 0x0.
Force Full Duplex.
Set this bit to 0b for half duplex and 1b for full duplex.
Note that this bit is a don’t care unless bits 10 and 11 are set.
FDP
FSP
0b
Force Speed.
These bits determine speed.
01b = 10 Mb/s.
11:10
00b
10b = 100 Mb/s.
11b = Not allowed.
All zeros indicate auto-negotiate (the current bit state).
Note that bit 12 is a don’t care unless these bits are set.
Reserved
Set this bit to 0b.
9
8
Reserved
DSM
0b
1b
Display Setup Message.
If this bit is set to 1b, the "Press Control-S" message appears after the
title message.
1b
Prompt Time. These bits control how long the "Press Control-S" setup
prompt message appears, if enabled by DIM.
00b = 2 seconds (default).
01b = 3 seconds.
10b = 5 seconds.
7:6
PT
00b
11b = 0 seconds.
Note that the Ctrl-S message does not appear if 0 seconds prompt time is
selected.
5
Reserved
DBS
0b
Reserved
Default Boot Selection. These bits select which device is the default boot
device. These bits are only used if the agent detects that the BIOS does
not support boot order selection or if the MODE field of word 0x31 is set
to MODE_LEGACY.
00b = Network boot, then local boot.
01b = Local boot, then network boot.
10b = Network boot only.
4:3
00b
11b = Local boot only.
2
Reserved
PS
0b
Reserved
Protocol Select. These bits select the boot protocol.
00b = PXE (default value).
01b = Reserved.
1:0
00b
Other values are undefined.
75
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.1.2.6.2
Boot Agent Configuration Customization Options (Word 0x31)
Word 0x31 contains settings that can be programmed by an OEM or network
administrator to customize the operation of the software. These settings cannot be
changed from within the Control-S setup menu or the IBA Intel Boot Agent utility. The
lower byte contains settings that would typically be configured by a network
administrator using the Intel Boot Agent utility; these settings generally control which
setup menu options are changeable. The upper byte are generally settings that would
be used by an OEM to control the operation of the agent in a LOM environment,
although there is nothing in the agent to prevent their use on a NIC implementation.
Table 20.
Boot Agent Configuration Customization Options (Word 0x31)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
Signature
15:14
SIG
Reserved
11b
Set these bits to 11b to indicate valid data.
13:12
11
00b
0b
Reserved, must be set to 00b.
Continuous Retry Disabled (0b default).
Selects the agent's boot order setup mode. This field
changes the agent's default behavior in order to make it
compatible with systems that do not completely support
the BBS and PnP Expansion ROM standards. Valid values
and their meanings are:
000b = Normal behavior. The agent attempts to detect BBS
and PnP Expansion ROM support as it normally does.
001b = Force Legacy mode. The agent does not attempt to
detect BBS or PnP Expansion ROM supports in the BIOS
and assumes the BIOS is not compliant. The BIOS boot
order can be changed in the Setup Menu.
010b = Force BBS mode. The agent assumes the BIOS is
BBS-compliant, even though it may not be detected as
such by the agent's detection code. The BIOS boot order
CANNOT be changed in the Setup Menu.
011b = Force PnP Int18 mode. The agent assumes the
BIOS allows boot order setup for PnP Expansion ROMs and
hooks interrupt 18h (to inform the BIOS that the agent is a
bootable device) in addition to registering as a BBS IPL
device. The BIOS boot order CANNOT be changed in the
Setup Menu.
10:8
MODE
000b
100b = Force PnP Int19 mode. The agent assumes the
BIOS allows boot order setup for PnP Expansion ROMs and
hooks interrupt 0x19 (to inform the BIOS that the agent is
a bootable device) in addition to registering as a BBS IPL
device. The BIOS boot order CANNOT be changed in the
Setup Menu.
101b = Reserved for future use. If specified, treated as
value 000b.
110b = Reserved for future use. If specified, treated as
value 000b.
111b = Reserved for future use. If specified, treated as
value 000b.
7:6
5
Reserved
DFU
00b
0b
Reserved, must be set to 00b.
Disable Flash Update
If set to 1b, no updates to the Flash image using PROSet is
allowed.
The default for this bit is 0b; allow Flash image updates
using PROSet.
76
Non-Volatile Memory (NVM) Map—82583V GbE Controller
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
Disable Legacy Wakeup Support
If set to 1b, no changes to the Legacy OS Wakeup Support
menu option is allowed.
4
DLWS
0b
The default for this bit is 0b; allow Legacy OS Wakeup
Support menu option changes.
Disable Boot Selection
If set to 1b, no changes to the boot order menu option is
allowed.
The default for this bit is 0b; allow boot order menu option
changes.
3
2
DBS
DPS
0b
0b
Disable Protocol Select
If set to 1b, no changes to the boot protocol is allowed.
The default for this bit is 0b; allow changes to the boot
protocol.
Disable Title Message
If set to 1b, the title message displaying the version of the
boot agent is suppressed; the Control-S message is also
suppressed. This is for OEMs who do not want the boot
agent to display any messages at system boot.
The default for this bit is 0b; allow the title message that
displays the version of the boot agent and the Control-S
message.
1
0
DTM
DSM
0b
0b
Disable Setup Menu
If set to 1b, no invoking the setup menu by pressing
Control-S is allowed. In this case, the EEPROM can only be
changed via an external program.
The default for this bit is 0b; allow invoking the setup menu
by pressing Control-S.
5.1.2.6.3
Table 21.
Boot Agent Configuration Customization Options (Word 0x32)
Word 0x32 is used to store the version of the boot agent that is stored in the Flash
image. When the Boot Agent loads, it can check this value to determine if any first-time
configuration needs to be performed. The agent then updates this word with its
version. Some diagnostic tools to report the version of the Boot Agent in the Flash also
read this word. This word is only valid if the PPB is set to 0b. Otherwise the contents
might be undefined.
Boot Agent Configuration Customization Options (Word 0x32)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
15:12
11:8
7:0
MAJOR
0x1
0x1
PXE boot agent major version.
PXE boot agent minor version.
PXE boot agent build number..
MINOR
BUILD
0x2
0x2
0x28
0x1C
77
82583V GbE Controller—Non-Volatile Memory (NVM) Map
5.1.2.6.4
Table 22.
IBA Capabilities (Word 0x33)
Word 0x33 is used to enumerate the boot technologies that have been programmed
into the Flash. It is updated by IBA configuration tools and is not updated or read by
IBA.
IBA Capabilities
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
Signature
These bits must be set to 11b to indicate that this
word has been programmed by the agent or other
configuration software.
15:14
SIG
11b
13:5
4
Reserved
0x0
0b
Reserved, must be set to 0x0.
0b
0b
iSCSI boot capability not present.
EFI EBC capability is present in Flash.
0b = The EFI code is not present.
1b = The EFI code is present.
3
2
1
EFI
0b
1b
1b
Reserved
UNDI
Reserved, set to 1b.
PXE/UNDI capability is present in Flash.
1b = The PXE base code is present.
0b = The PXE base code is not present.
1b
0b
PXE base code is present in Flash.
0b = The PXE base code is present.
1b = The PXE base code is not present.
0
BC
1b
5.1.2.7
5.1.2.8
Virtual MAC Data Pointer (Word 0x37)
NVM
Hardware
Default
Bit
Name
Image
Description
Setting
15:0
0x0
0xFFFF
Virtual MAC Data Pointer.
Vital Product Data Pointer (VDP) (Word 0x2F)
NVM
Image
Size
Hardware
Default
Bit
Name
Description
15:0
VDP
0xFFFF
Reserved, set to 0xFFFF.
78
Non-Volatile Memory (NVM) Map—82583V GbE Controller
5.1.2.9
iSCSI Boot Configuration Start Address (Word 0x3D)
NVM
Image
Setting
Hardware
Default
Bit
Name
Description
NVM word address of the iSCSI boot configuration
structure starting point.
15:0
Address
0x0
0x120
5.1.2.10
Checksum Word Calculation (Word 0x3F)
The checksum word (0x3F) is used to ensure that the base NVM image is a valid image.
The value of this word should be calculated such that after adding all the words (0x00-
0x3F), including the checksum word itself, the sum should be 0xBABA. The initial value
in the 16-bit summing register should be 0x0000 and the carry bit should be ignored
after each addition.
Note:
Hardware does not calculate the word 0x3F checksum during an NVM write or read. It
must be calculated by software independently and included in the NVM write data. This
field is provided strictly for software verification of NVM validity. All hardware
configuration based on word 0x00-0x3F content is based on the validity of the
Signature field of the NVM.
79
82583V GbE Controller—Interconnects
6.0
Interconnects
6.1
PCIe
PCIe is a third generation I/O architecture that enables cost competitive, next
generation I/O solutions providing industry leading price/performance and feature
richness. It is an industry-driven specification.
PCIe defines a basic set of requirements that comprehends the majority of the targeted
application classes. High-end application requirements such as Enterprise class servers
and high-end communication platforms are delivered by a set of advanced extensions
that compliment the baseline requirements.
To guarantee headroom for future applications of PCIe, a software-managed
mechanism for introducing new, enhanced capabilities in the platform is provided.
Figure 23 shows the PCIe architecture.
Config/OS
S/W
PCI.sys Compliant
Preserve Driver Model
Advanced Xtensions
Protocol
Common Base Protocol
Configurable widths 1 .. 32
Link
Physical
Point to point, serial, differential,
(electrical
hot-plug, inter-op formfactors
Mechanical)
Figure 23.
PCIe Stack Structure
The PCIe physical layer consists of a differential transmit pair and a differential receive
pair. Full-duplex data on these two point-to-point connections is self-clocked such that
no dedicated clock signals are required.
Note:
The bandwidth of this interface increases linearly with frequency.
80
Interconnects—82583V GbE Controller
A packet is the fundamental unit of information exchange and the protocol includes a
message space to replace the number of side-band signals found on many of today’s
buses. This movement of hard-wired signals from the physical layer to messages within
the transaction layer enables easy and linear physical layer width expansion for
increased bandwidth.
The common base protocol uses split transactions along with several mechanisms that
are included to eliminate wait states and to optimize the reordering of transactions to
further improve system performance.
6.1.1
Architecture, Transaction, and Link Layer Properties
• Split transaction, packet-based protocol
• Common flat address space for load/store access (such as a PCI addressing
model):
— Memory address space of 32 bits to enable compact packet header (must be
used to access addresses below 4 GB)
— Memory address space of 64 bits using extended packet header
• Transaction layer mechanisms:
— PCI-X style relaxed ordering
— Optimizations for no-snoop transactions
• Credit-based flow control
• Packet sizes/formats:
— Maximum packet size supports 128- and 256-byte data payload
— Maximum read request size of 4 KB
• Reset/initialization:
— Frequency/width/profile negotiation performed by hardware
• Data integrity support:
— Using CRC-32 for transaction layer packets
• Link layer retry for recovery following error detection:
— Using CRC-16 for link layer messages
• No retry following error detection:
— 8b/10b encoding with running disparity
• Software configuration mechanism:
— Uses PCI configuration and bus enumeration model
— PCIe-specific configuration registers mapped via PCI extended capability
mechanism
• Baseline messaging:
— In-band messaging of formerly side-band legacy signals (such as interrupts)
— System-level power management supported via messages
• Power Management (PM):
— Full PCI PM support
— Wake capability from D3cold state
— Compliant with ACPI 2.0, PCI PM software model
— Active state power management (transparent to software including ACPI)
81
82583V GbE Controller—Interconnects
6.1.1.1
Physical Interface Properties
• Point to point interconnect
— Full-duplex; no arbitration
• Signaling technology:
— Low voltage differential
— Embedded clock signaling using 8b/10b encoding scheme
• Serial frequency of operation: 2.5 GHz.
• Interface width of one lane per direction
• DFT and DFM support for high volume manufacturing
6.1.1.2
6.1.2
Advanced Extensions
PCIe defines a set of optional features to enhance platform capabilities for specific
usage modes. The 82583V supports the following optional features:
• Extended error reporting – messaging support to communicate multiple types/
severity of errors
• Serial number
General Functionality
• Native/legacy:
— The PCIe capability register states the device/port type.
— The 82583V is a native device by default.
• Locked transactions:
— The 82583V does not support locked requests as a target or master.
• End to End CRC (ECRC):
— Not supported by the 82583V
6.1.3
Transaction Layer
The upper layer of the PCIe architecture is the transaction layer. The transaction layer
connects to the 82583V’s core using an implementation-specific protocol. Through this
core-to-transaction-layer protocol, the application-specific parts of the 82583V interact
with the PCIe subsystem and transmit and receive requests to or from the remote PCIe
agent, respectively.
6.1.3.1
Transaction Types Received by the Transaction Layer
Transaction Types at the Rx Transaction Layer
Table 25.
Tx Later
Reaction
Hardware Should Keep
Data From Original Packet
Transaction Type
FC Type
For Client
Configuration Read
Request
NPH
CPLH + CPLD Requester ID, TAG, Attribute Configuration space
CPLH Requester ID, TAG, Attribute Configuration space
CPLH + CPLD Requester ID, TAG, Attribute CSR
Configuration Write
Request
NPH +
NPD
Memory Read
Request
NPH
82
Interconnects—82583V GbE Controller
Tx Later
Reaction
Hardware Should Keep
Data From Original Packet
Transaction Type
FC Type
For Client
PH +
PD
Memory Write
Request
-
-
CSR
I/O Read Request
I/O Write Request
NPH
CPLH + CPLD Requester ID, TAG, Attribute CSR
NPH +
NPD
CPLH
Requester ID, TAG, Attribute CSR
CPLH +
CPLD
Read Completions
Message
-
-
-
-
DMA
Message Unit / INT / PM
/ Error Unit
PH
Flow control types:
• PH - Posted request headers
• PD - Posted request data payload
• NPH - Non-posted request headers
• NPD - Non-posted request data payload
• CPLH - Completion headers
• CPLD - Completion data payload
6.1.3.2
Transaction Types Initiated by The 82583V
Transaction Types at the Tx Transaction Layer
Table 26.
Transaction Type
Payload Size
FC Type
From Client
Configuration Read Request
Completion
Dword
-
CPLH + CPLD
Configuration space
Configuration Write Request
Completion
CPLH
Configuration space
I/O Read Request Completion
I/O Write Request Completion
Read Request Completion
Memory Read Request
Dword
-
CPLH + CPLD
CPLH
CSR
CSR
CSR
DMA
DMA
Dword/Qword
CPLH + CPLD
NPH
-
1
Memory Write Request
<= MAX_PAYLOAD_SIZE
PH + PD
Message Unit / INT /
PM / Error Unit
Message
-
PH
1. The MAX_PAYLOAD_SIZE supported is loaded from the NVM (either 128 bytes or 256 bytes). Effective
MAX_PAYLOAD_SIZE is according to configuration space register.
6.1.3.3
Message Handling by The 82583V (as a Receiver)
Message packets are special packets that carry a message code.
The upstream device transmits special messages to the 82583V by using this
mechanism.
The transaction layer decodes the message code and responds to the message
accordingly.
83
82583V GbE Controller—Interconnects
Table 27.
Supported Message in The 82583V (As a Receiver)
Message
Routing
r2r1r0
Message
PM_Active_State_NAK
Device’s Later Response
code [7:0]
0x14
0x19
0x41
0x43
0x40
0x45
0x47
0x44
0x50
0x7E
0x7E
0x7F
0x7F
0x00
100
Internal signal set
Internal signal set
Silently drop
011
PME_Turn_Off
100
Attention_Indicator_On
Attention_Indicator_Blink
Attention_Indicator_Off
Power_Indicator_On
Power_Indicator_Blink
Power_Indicator_Off
100
Silently drop
100
Silently drop
100
Silently drop
100
Silently drop
100
Silently drop
100
Slot power limit support (has one Dword data)
Vendor_defined Type 0 no data
Vendor_defined Type 0 data
Vendor_defined Type 1 no data
Vendor_defined Type 1 data
Unlock
Silently drop
010,011,100
010,011,100
010,011,100
010,011,100
011
Unsupported request - NEC*
Unsupported request - NEC*
Silently drop
Silently drop
Silently drop
6.1.3.4
Message Handling by The 82583V (As a Transmitter)
The transaction layer is also responsible for transmitting specific messages to report
internal/external events (such as interrupts and PMEs).
Table 28.
Supported Message in The 82583V (As a Transmitter)
Message
code [7:0]
Routing
r2r1r0
Message
0x20
100
100
100
100
100
100
100
100
000
000
000
000
101
Assert INT A
Assert INT B
Assert INT C
Assert INT D
DE- Assert INT A
DE- Assert INT B
DE- Assert INT C
DE- Assert INT D
ERR_COR
0x21
0x22
0x23
0x24
0x25
0x26
0x27
0x30
0x31
0x33
0x18
0x1B
ERR_NONFATAL
ERR_FATAL
PM_PME
PME_TO_Ack
84
Interconnects—82583V GbE Controller
6.1.3.5
Data Alignment
4 KB Boundary:
Requests must never specify an address/length combination that causes a memory
space access to cross a 4 KB boundary. It is hardware’s responsibility to break requests
into 4 KB-aligned requests (if needed). This does not pose any requirement on
software. However, if software allocates a buffer across a 4 KB boundary, hardware
then issues multiple requests for the buffer. Software should consider aligning buffers
to a 4 KB boundary in cases where it improves performance.
The alignment to the 4 KB boundaries is done in the core. The transaction layer does
not do any alignment according to these boundaries.
64 Bytes:
It is also recommended that requests are multiples of 64 bytes and aligned to make
better use of memory controller resources. This is also done in the core.
6.1.3.6
Configuration Request Retry Status
The 82583V might have a delay in initialization due to an NVM read. The PCIe defined a
mechanism for devices that require completion of a lengthy self-initialization sequence
before being able to service configuration requests.
If the read of the PCIe section in the NVM was not completed before the 82583V
received a configuration request, then the 82583V responds with a configuration
request retry completion status to terminate the request, and effectively stalls the
configuration request until such time that the subsystem has completed local
initialization and is ready to communicate with the host.
6.1.3.7
Ordering Rules
The 82583V meets the PCIe ordering rules (PCI-X rules) by following the PCI simple
device model:
• Deadlock avoidance - Master and target accesses are independent - The response
to a target access does not depend on the status of a master request to the bus. If
master requests are blocked (such as due to no credits), target completions can
still proceed (if credits are available).
• Descriptor/data ordering - the 82583V does not proceed with some internal actions
until respective data writes have ended on the PCIe link:
— The 82583V does not update an internal header pointer until the descriptors
that the header pointer relates to are written to the PCIe link.
— The 82583V does not issue a descriptor write until the data that the descriptor
relates to is written to the PCIe link.
The 82583V can issue the following master read request from each of the following
clients:
• Rx descriptor read queue
• Tx descriptor read queue
Completed separate read requests are not guaranteed to return in order. Completions
for a single read request are guaranteed to return in address order.
85
82583V GbE Controller—Interconnects
6.1.3.8
Transaction Attributes
6.1.3.8.1
Traffic Class (TC) and Virtual Channels (VC)
The 82583V supports only TC = 0 and VC = 0 (default).
6.1.3.8.2
Relaxed Ordering
The 82583V takes advantage of the relaxed ordering rules in PCIe by setting the
relaxed ordering bit in the packet header. The 82583V also enables the system to
optimize performance in the following cases:
• Relaxed ordering for descriptor and data reads: When the 82583V is a master in a
read transaction, its split completion has no relationship with the writes from the
CPUs (same direction). It should be allowed to bypass the writes from the CPUs.
• Relaxed ordering for receiving data writes: When the 82583V masters receive data
writes, it also enables them to bypass each other in the path to system memory
because the software does not process this data until their associated descriptor
writes have been completed.
• The 82583V cannot perform relax ordering for descriptor writes or an MSI write.
Relaxed ordering can be used in conjunction with the no-snoop attribute to enable the
memory controller to advance non-snoop writes ahead of earlier snooped writes.
Relaxed ordering is enabled in the 82583V by setting the RO_DIS bit to 0b in the
CTRL_EXT register.
6.1.3.8.3
Snoop Not Required
The 82583V sets the Snoop Not Required attribute bit for master data writes. System
logic can provide a separate path into system memory for non-coherent traffic. The
non-coherent path to system memory provides higher, more uniform, bandwidth for
write requests.
The Snoop Not Required attribute bit does not alter transaction ordering. Therefore, to
achieve maximum benefit from snoop not required transactions, it is advisable to set
the relaxed ordering attribute as well (assuming that system logic supports both
attributes).
Software configures no-snoop support through the 82583V’s control register and a set
of NONSNOOP bits in the GCR register in the CSR space. The default value for all bits is
disabled.
The 82583V supports a No-Snoop bit for each relevant DMA client:
1. TXDSCR_NOSNOOP - Transmit descriptor read.
2. TXDSCW_NOSNOOP - Transmit descriptor write.
3. TXD_NOSNOOP - Transmit data read.
4. RXDSCR_NOSNOOP - Receive descriptor read.
5. RXDSCW_NOSNOOP - Receive descriptor write.
6. RXD_NOSNOOP - Receive data write.
All PCIe functions in the 82583V are controlled by this register.
86
Interconnects—82583V GbE Controller
6.1.3.9
Error Forwarding
If a Transaction Layer Protocol (TLP) is received with an error-forwarding trailer, the
packet is dropped and not delivered to its destination. The 82583V does not initiate any
additional master requests for that PCI function until it detects an internal reset or
software. Software is able to access device registers after such a fault.
System logic is expected to trigger a system-level interrupt to inform the operating
system of the problem. The operating system can then stop the process associated
with the transaction, re-allocate memory instead of the faulty area, etc.
6.1.3.10
Master Disable
System software can disable master accesses on the PCIe link by either clearing the
PCI Bus Master bit or by bringing the function into a D3 state. From that time on, the
82583V must not issue master accesses for this function. Due to the full-duplex nature
of PCIe, and the pipelined design in the 82583V, it might happen that multiple requests
from several functions are pending when the master disable request arrives. The
protocol described in this section insures that a function does not issue master requests
to the PCIe link after its master enable bit is cleared (or after entry to D3 state).
Two configuration bits are provided for the handshake between the device function and
its driver:
• PCIe Master Disable bit in the Device Control (CTRL) register - When the PCIe
Master Disable bit is set, the 82583V blocks new master requests. The 82583V then
proceeds to issue any pending requests by this function. This bit is cleared on
master reset (Internal Power On Reset all the way to a software reset) to enable
master accesses.
• PCIe Master Enable Status bits in the Device Status register - Cleared by the
82583V when the PCIe Master Disable bit is set and no master requests are
pending by the relevant function, set otherwise.
Software Note:
— The software device driver sets the PCIe Master Disable bit when notified of a
pending master disable (or D3 entry). The 82583V then blocks new requests
and proceeds to issue any pending requests by this function. The software
device driver then polls the PCIe Master Enable Status bit. Once the bit is
cleared, it is guaranteed that no requests are pending from this function. The
software device driver might time out if the PCIe Master Enable Status bit is not
cleared within a given time.
— The PCIe Master Disable bit must be cleared to enable a master request to the
PCIe link. This can be done either through reset or by the software device
driver.
6.1.4
Flow Control
6.1.4.1
Flow Control Rules
The 82583V only implements the default Virtual Channel (VC0). A single set of credits
is maintained for VC0.
87
82583V GbE Controller—Interconnects
Table 29.
Allocation of FC Credits
Credit Type
Operations
Target write (1 unit)
Number Of Credits
Posted Request Header (PH)
2 units
Message (1 unit)
Target write (Length/16B=1)
Message (1 unit)
Posted Request Data (PD)
16 credits (for 256 bytes)
2 units
Target read (1 unit)
Non-Posted Request Header (NPH) Configuration read (1 unit)
Configuration write (1 unit)
Non-Posted Request Data (NPD)
Completion Header (CPLH)
Completion Data (CPLD)
Configuration write (1 unit)
Read completion (N/A)
Read completion (N/A)
2 units
Infinite (accepted immediately)
Infinite (accepted immediately)
Rules for FC updates:
• The 82583V maintains two credits for NPD at any given time. It increments the
credit by one after the credit is consumed and sends an UpdateFC packet as soon
as possible. UpdateFC packets are scheduled immediately after a resource is
available.
• The 82583V provides two credits for PH (such as for two concurrent target writes)
and two credits for NPH (such as for two concurrent target reads). UpdateFC
packets are scheduled immediately after a resource becomes available.
• The 82583V follows the PCIe recommendations for frequency of UpdateFC FCPs.
6.1.4.2
6.1.4.3
Upstream Flow Control Tracking
The 82583V issues a master transaction only when the required FC credits are
available. Credits are tracked for posted, non-posted, and completions (the later to
operate against a switch).
Flow Control Update Frequency
In any case, UpdateFC packets are scheduled immediately after a resource becomes
available.
When the link is in the L0 or L0s link state, update FCPs for each enabled type of non-
infinite FC credit must be scheduled for transmission at least once every 30 µs (-0%/
+50%), except when the Extended Sync bit of the Control Link register is set, in which
case the limit is 120 µs (-0%/+50%).
6.1.4.4
Flow Control Timeout Mechanism
The 82583V implements the optional FC update timeout mechanism. The mechanism is
activated when the link is in L0 or L0s link state. It uses a timer with a limit of 200 µs (-
0%/+50%), where the timer is reset by the receipt of any init or update FCP.
Alternately, the timer can be reset by the receipt of any DLLP.
After timer expiration, the mechanism instructs the PHY to retrain the link (via the
LTSSM recovery state).
88
Interconnects—82583V GbE Controller
6.1.5
Host I/F
Tag IDs
6.1.5.1
PCIe device numbers identify logical devices within the physical device (the 82583V is a
physical device). The 82583V implements a single logical device with one PCI function -
LAN. The device number is captured from each type 0 configuration write transaction.
Each of the PCIe functions interface with the PCIe unit through one or more clients. A
client ID identifies the client and is included in the Tag field of the PCIe packet header.
Completions always carry the tag value included in the request to enable routing of the
completion to the appropriate client.
Client IDs are assigned as follows:
Table 30.
Assignment of Client IDs
TAG Code
Flow: TLP TYPE – Usage
in Hex
00
RX: WR REQ (data from Ethernet to main memory)
01
RX: RD REQ to read descriptor to core
02
RX: WR REQ to write back descriptor from core to memory
04
TX: RD REQ to read descriptor to core
05
TX: WR REQ to write back descriptor from core to memory
07:06
08
Reserved
TX: RD REQ data 0 from main memory to Ethernet
09
TX: RD REQ data 1 from main memory to Ethernet
0A
TX: RD REQ data 2 from main memory to Ethernet
0B
TX: RD REQ data 3 from main memory to Ethernet
0C
Reserved
Reserved
Reserved
MSI
0E
11:10
1E
1F
Message unit
Reserved
Others
89
82583V GbE Controller—Interconnects
6.1.5.1.1
Completion Timeout Mechanism
In any split transaction protocol, there is a risk associated with the failure of a
requester to receive an expected completion. To enable requesters to attempt recovery
from this situation in a standard manner, the completion timeout mechanism is defined.
• The completion timeout mechanism is activated for each request that requires one
or more completions when the request is transmitted.
• The completion timeout timer should not expire in less than 10 ms.
• The completion timeout timer must expire if a request is not completed in 50 ms.
• A completion timeout is a reported error associated with the requestor device/
function.
A Memory Read Request for which there are multiple completions are considered
completed only when all completions are received by the requester. If some, but not all,
requested data is returned before the completion timeout timer expires, the requestor
is permitted to keep or discard the data that was returned prior to timer expiration.
6.1.5.1.2
Out of Order Completion Handling
In a split transaction protocol, when using multiple read requests in a multi processor
environment, there is a risk that the completions might arrive from the host memory
out of order and interleave. In this case the host interface role is to sort the request
completions and transfer them to the Ethernet core in the correct order.
6.1.6
Error Events and Error Reporting
Mechanism in General
6.1.6.1
PCIe defines two error reporting paradigms: the baseline capability and the Advanced
Error Reporting (AER) capability. The baseline error reporting capabilities are required
of all PCIe devices and define the minimum error reporting requirements. The AER
capability is defined for more robust error reporting and is implemented with a specific
PCIe capability structure.
Both mechanisms are supported by the 82583V.
Also the SERR# Enable and the Parity Error bits from the legacy command register take
part in the error reporting and logging mechanism.
Figure 24 shows, in detail, the flow of error reporting in the 82583V.
90
Interconnects—82583V GbE Controller
Error Sources
(Associated with Port)
Device Status ::
Correctable Error Detected
Uncorrectable Error Severity
Status ::
Signaled Target Abort
Device Status ::
Uncorrectable Error Status
Correctable Error Status
Non-Fatal Error Detected
Status ::
Received Target Abort
Device Status ::
Fatal Error Detected
Status ::
Received Master Abort
Device Status ::
Correctable Error Mask
Uncorrectable Error Mask
Unsupported Request Detected
Status ::
Detected Parity Error
Device Control ::
Correctable Error Reporting Enable
Root Error Status
Device Control ::
Unsupported Request Reporting Enable
Device Control ::
Non-Fatal Error Reporting Enable
Root Error Command ::
Correctable Error Reporting Enable
Device Control ::
Fatal Error Reporting Enable
Interrupt
Root Error Command ::
Non-Fatal Error Reporting Enable
Root Error Command ::
Fatal Error Reporting Enable
Command ::
SERR# Enable
Status ::
Signaled System Error
Status ::
Master Data Parity Error
Command ::
Parity Error Response
Bridge Control ::
SERR Enable
Root Control ::
System Error on Correctable Error Enable
Root Control ::
System Error on Non-Fatal Error Enable
System Error
Rcv Msg
Error Message
Processing
Root Control ::
System Error on Fatal Error Enable
Secondary Status ::
Received System Error
(Either implementation acceptable
-
the
Secondary Status ::
unqualified version is more like PCI P2P
bridge spec)
Detected Parity Error
Secondary Status ::
Signaled Target Abort
Secondary Side Error Sources
Secondary Status ::
Received Target Abort
Secondary Status ::
Received Master Abort
Bridge Control ::
Parity Error Response Enable
Secondary Status ::
Master Data Parity Error
Figure 24.
6.1.6.1.1
Error Reporting Flow
Error Events
Table 31 lists error events identified by the 82583V and the response in terms of
logging, reporting, and actions taken. Consult the PCIe specification for the affect on
the PCI Status register.
Table 31.
Response and Reporting of Error Events
Error Name
Error Events
Default Severity
Action
PHY errors
•
•
8b/10b Decode errors
Packet framing error
Correctable
Send ERR_CORR
TLP to initiate NAK, drop data
DLLP to Drop
Receiver error
Data link errors
•
•
•
Bad CRC
Not legal EDB
Wrong sequence number
Correctable
Send ERR_CORR
Bad TLP
TLP to initiate NAK, drop data
Correctable
Send ERR_CORR
Bad DLLP
Bad CRC
DLLP to drop
Follow LL rules
Follow LL rules
Correctable
Send ERR_CORR
Replay timer
timeout
REPLAY_TIMER expiration
REPLAY NUM rollover
Correctable
Send ERR_CORR
REPLAY NUM
rollover
91
82583V GbE Controller—Interconnects
Error Name
Error Events
Default Severity
Action
Uncorrectable
Send ERR_FATAL
Data link layer
protocol error
Violations of Flow Control
initialization protocol
TLP errors
Uncorrectable
ERR_NONFATAL
Log header
Poisoned TLP
received
In case of poisoned completion,
no more requests from this client.
TLP with Error Forwarding
•
•
•
•
Wrong config access
MRdLk
Config Request Type1
Unsupported vendor
defined type 0 message
Uncorrectable
ERR_NONFATAL
Log header
•
•
•
•
•
Not valid MSG code
Not supported TLP type
Wrong function number
Wrong TC/VC
Received target access
with data size > 64-bit
Unsupported
Request (UR)
Send completion with UR
•
Received TLP outside
address range
Uncorrectable
ERR_NONFATAL
Completion
Timeout
Completion timeout timer
expired
Send the read request again
Send completion with CA
Uncorrectable
ERR_NONFATAL
Log header
Attempts to write to the Flash
device when writes are
disabled (FWE=10b)
Completer abort
Uncorrectable
ERR_NONFATAL
Log header
Unexpected
completion
Received completion without
a request for it (tag, ID, etc.)
Discard TLP
Uncorrectable
ERR_FATAL
Receiver
Overflow
Received TLP beyond
allocated credits
Receiver behavior is undefined
•
•
Minimum Initial Flow
Control Advertisements
Flow control update for
Infinite Credit
advertisement
Uncorrectable
ERR_FATAL
Flow control
protocol error
Receiver behavior is undefined
•
•
Data payload exceed
Max_Payload_Size
Received TLP data size
does not match length
field
Uncorrectable
ERR_FATAL
Log header
•
TD field value does not
correspond with the
observed size
Malformed TLP
(MP)
Drop the packet, free FC credits
•
•
Byte enables violations.
PM messages that don’t
use TC0.
•
Usage of unsupported VC
Completion with
unsuccessful
completion status
No action (already
done by originator of Free FC credits
completion)
92
Interconnects—82583V GbE Controller
6.1.6.1.2
Error Pollution
Error pollution can occur if error conditions for a given transaction are not isolated to
the error's first occurrence. If the PHY detects and reports a receiver error, to avoid
having this error propagate and cause subsequent errors at upper layers, the same
packet is not signaled at the data link or transaction layers.
Similarly, when the data link layer detects an error, subsequent errors that occur for the
same packet is not signaled at the transaction layer.
6.1.6.1.3
Completion With Unsuccessful Completion Status
A completion with unsuccessful completion status is dropped and not delivered to its
destination. The request that corresponds to the unsuccessful completion is retried by
sending a new request for the undeliverable data.
6.1.7
Link Layer
6.1.7.1
ACK/NAK Scheme
The 82583V supports two alternative schemes for ACK/NAK rate:
1. ACK/NAK is scheduled for transmission following any TLP.
2. ACK/NAK is scheduled for transmission according to timeouts specified in the PCIe
specification.
The PCIe Error Recovery bit, loaded from NVM, determines which of the two schemes is
used.
6.1.7.2
Supported DLLPs
The following DLLPs are supported by the 82583V as a receiver:
DLLPs Received by The 82583V
Table 32.
Remarks
Remarks
ACK
NAK
PM_Request_Ack
InitFC1-P
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
InitFC1-NP
InitFC1-Cpl
InitFC2-P
InitFC2-NP
InitFC2-Cpl
UpdateFC-P
UpdateFC-NP
UpdateFC-Cpl
The following DLLPs are supported by the 82583V as a transmitter:
93
82583V GbE Controller—Interconnects
Table 33.
DLLPs initiated by The 82583V
1
Remarks
Remarks
ACK
NAK
PM_Enter_L1
PM_Enter_L23
PM_Active_State_Request_L1
InitFC1-P
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
v2v1v0 = 000
InitFC1-NP
InitFC1-Cpl
InitFC2-P
InitFC2-NP
InitFC2-Cpl
UpdateFC-P
UpdateFC-NP
1. UpdateFC-Cpl is not sent because of the infinite FC-Cpl allocation.
6.1.7.3
Transmit EDB Nullifying
In case of a retrain necessity, there is a need to guarantee that no abrupt termination
of the Tx packet happens. For this reason, early termination of the transmitted packet
is possible. This is done by appending the EDB to the packet.
6.1.8
PHY
6.1.8.1
Link Width
The 82583V supports a link width of x1 only.
6.1.8.2
Polarity Inversion
If polarity inversion is detected, the receiver must invert the received data.
During the training sequence, the receiver looks at Symbols 6-15 of TS1 and TS2 as the
indicator of lane polarity inversion (D+ and D- are swapped). If lane polarity inversion
occurs, the TS1 Symbols 6-15 received are D21.5 as opposed to the expected D10.2.
Similarly, if lane polarity inversion occurs, Symbols 6-15 of the TS2 ordered set are
D26.5 as opposed to the expected 5D5.2. This provides the clear indication of lane
polarity inversion.
6.1.8.3
L0s Exit Latency
The number of FTS sequences (N_FTS), sent during L1 exit, is loaded from the NVM
into an 8-bit read-only register.
94
Interconnects—82583V GbE Controller
6.1.8.4
Reset
The PCIe PHY can initiate core reset to the 82583V. The reset can be caused by three
sources:
• Upstream move to hot reset - Inband Mechanism (LTSSM).
• Recovery failure (LTSSM returns to detect).
• Upstream component move to disable.
6.1.8.5
6.1.9
Scrambler Disable
The Scrambler/de-scrambler functionality in the 82583V can be eliminated by two
mechanisms:
• Upstream according to the PCIe specification.
• NVM bit.
Performance Monitoring
The 82583V incorporates PCIe performance monitoring counters to provide common
capabilities to evaluate performance. The 82583V implements four 32-bit counters to
correlate between concurrent measurements of events as well as the sample delay and
interval timers. The four 32-bit counters can also operate in a two 64-bit mode to count
long intervals or payloads.
The list of events supported by the 82583V and the counters control bits are described
in the memory register map.
6.2
Ethernet Interface
The 82583V MAC provides a complete CSMA/CD function, supporting IEEE 802.3
(10 Mb/s), 802.3u (100 Mb/s), 802.3z, and 802.3ab (1000 Mb/s) implementations. The
82583V performs all of the functions required for transmission, reception, and collision
handling called out in the standards.
The GMII/MII mode used to communicate between the MAC and the PHY supports
10/100/1000 Mb/s operation, with both half- and full-duplex operation at 10/100 Mb/s,
and only full-duplex operation at 1000 Mb/s.
Note:
The 82583V MAC is optimized for full-duplex operation in 1000 Mb/s mode. Half-duplex
1000 Mb/s operation is not supported.
The PHY features 10/100/1000-BaseT signaling and is capable of performing intelligent
power-management based on both the system power-state and LAN energy-detection
(detection of unplugged cables). Power management includes the ability to shutdown
to an extremely low (powered-down) state when not needed as well as ability to auto-
negotiate to a lower-speed 10/100 Mb/s operation when the system is in low power-
states.
6.2.1
MAC/PHY GMII/MII Interface
The 82583V MAC and PHY communicate through an internal GMII/MII interface that
can be configured for either 1000 Mb/s operation (GMII) or 10/100 Mb/s (MII) mode of
operation. For proper network operation, both the MAC and PHY must be properly
configured (either explicitly via software or via hardware auto-negotiation) to identical
speed and duplex settings. All MAC configuration is performed using device control
registers mapped into system memory or I/O space; an internal MDIO/MDC interface,
accessible via software, is used to configure the PHY operation.
95
82583V GbE Controller—Interconnects
The internal Gigabit Media Independent Interface (GMII) mode of operation is similar to
MII mode of operation. GMII mode uses the same MDIO/MDC management interface
and registers for PHY configuration as MII mode. These common elements of operation
enable the 82583V MAC and PHY to cooperatively determine a link partner's operational
capability and configure the hardware based on those capabilities.
6.2.1.1
MDIO/MDC
The 82583V implements an internal IEEE 802.3 MII Management Interface (also known
as the Management Data Input/Output or MDIO Interface) between the MAC and PHY.
This interface provides the MAC and software the ability to monitor and control the
state of the PHY. The internal MDIO interface defines a physical connection, a special
protocol that runs across the connection, and an internal set of addressable registers.
The internal interface consists of a data line (MDIO) and clock line (MDC), which are
accessible by software via the MAC register space.
Software can use MDIO accesses to read or write registers in either GMII or MII mode
by accessing the 82583V's MDIC register (see section 9.2.2.7).
6.2.1.2
Other MAC/PHY Control and Status
In addition to the internal GMII/MII communication and MDIO interface between the
MAC and the PHY, the 82583V implements a handful of additional internal signals
between MAC and PHY, which provide richer control and features.
• PHY reset - The MAC provides an internal reset to the PHY. This signal combines the
PCI_RST_N input from the PCI bus and the PHY Reset bit of the Device Control
register (CTRL.PHY_RST).
• PHY link status indication - The PHY provides a direct internal indication of link
status (LINK) to the MAC to indicate whether it has sensed a valid link partner.
Unless the PHY has been configured via its MII management registers to assert this
indication unconditionally, this signal is a valid indication of whether a link is
present. The MAC relies on this internal indication to reflect the STATUS.LU status
as well as to initiate actions such as generating interrupts on link status changes,
re-initiating link speed sense, etc.
• PHY duplex indication - The PHY provides a direct internal indication to the MAC of
its resolved duplex mode (FDX). Normally, auto-negotiation by the PHY enables the
PHY to resolve full/duplex communications with the link partner (except when the
PHY is forced through MII register settings). The MAC normally uses this signal
after a link loss/restore to ensure that the MAC is configured consistently with the
re-linked PHY settings. This indication is effectively visible through the MAC register
bit STATUS.FD, each time MAC speed has not been forced.
• PHY speed indication(s) - The PHY provides direct internal indications (SPD_IND) to
the MAC of its negotiated speed (10/100/1000 Mb/s). The result of this indication is
effectively visible through the MAC register bits STATUS.SPEED each time MAC
speed has not been forced.
• MAC Dx power state indication - The MAC indicates its ACPI power state
(PWR_STATE) to the PHY to enable it to perform intelligent power-management
(provided that the PHY power-management is enabled in the MAC CTRL register).
6.2.2
Duplex Operation for Copper PHY/GMII/MII Operation
The 82583V supports half-duplex and full-duplex 10/100 Mb/s MII mode or 1000 Mb/s
GMII mode.
Configuring the duplex operation of the 82583V can either be forced or determined via
the auto-negotiation process. See section 6.2.3 for details on link configuration setup
and resolution.
96
Interconnects—82583V GbE Controller
6.2.2.1
Full Duplex
All aspects of the IEEE 802.3, 802.3u, 802.3z, and 802.3ab specifications are
supported in full duplex operation. Full duplex operation is enabled by several
mechanisms, depending on the speed configuration of the 82583V and the specific
capabilities of the link partner used in the application. During full duplex operation, the
82583V might transmit and receive packets simultaneously across the link interface.
In full-duplex GMII/MII mode, transmission and reception are delineated independently
by the GMII/MII control signals. Transmission starts at the assertion of TX_EN, which
indicates there is valid data on the TX_DATA bus driven from the MAC to the PHY.
Reception is signaled by the PHY by the assertion of the RX_DV signal, which indicates
valid receive data on the RX_DATA lines to the MAC.
6.2.2.2
Half Duplex
The 82583V MAC can operate in half duplex.
In half duplex operation, the MAC attempts to avoid contention with other traffic on the
link by monitoring the CRS signal provided by the PHY and deferring to passing traffic.
When the CRS signal is de-asserted or after a sufficient Inter-Packet Gap (IPG) has
elapsed after a transmission, frame transmission begins. The MAC signals the PHY with
TX_EN at the start of transmission.
If a collision occurs, the PHY detects the collision and asserts the COL signal to the
MAC. Transmitting the frame stops within four link clock times and the 82583V sends a
JAM sequence onto the link. After the end of a collided transmission, the 82583V backs
off and attempts to re-transmit per the standard CSMA/CD method.
Note:
The re-transmissions are done from the data stored internally in the 82583V MAC
transmit packet buffer (no re-access to the data in host memory is performed).
After a successful transmission, the 82583V is ready to transmit any other frame(s)
queued in the MAC's transmit FIFO, after the minimum Inter-Frame Spacing (IFS) of
the link has elapsed.
During transmit, the PHY is expected to signal a carrier-sense (assert the CRS signal)
back to the MAC before one slot time has elapsed. The transmission completes
successfully even if the PHY fails to indicate CRS within the slot time window; if this
situation occurs, the PHY can either be configured incorrectly or be in a link down
situation. Such an event is counted in the Transmit Without CRS statistic register (see
section 9.2.7.11).
6.2.3
Auto-Negotiation & Link Setup Features
The method for configuring the link between two link partners is highly dependent on
the mode of operation.
Configuration of the link can be accomplished by several methods ranging from:
• software's forcing link settings
• software-controlled negotiation
• MAC-controlled auto-negotiation
• auto-negotiation initiated by a PHY.
The following sections describe processes of bringing the link up including configuration
of the 82583V and the transceiver, as well as the various methods of determining
duplex and speed configuration.
97
82583V GbE Controller—Interconnects
The PHY performs auto-negotiation per 802.3ab clause 40 and extensions to clause 28.
Link resolution is obtained by the MAC from the PHY after the link has been established.
The MAC accomplishes this via the MDIO interface, via specific signals from the PHY to
the MAC, or by MAC auto-detection functions.
6.2.3.1
Link Configuration
Link configuration is generally determined by PHY auto-negotiation. The software
device driver must intervene in cases where a successful link is not negotiated or a user
desires to manually configure the link. The following sections discuss the methods of
link configuration for copper PHY operation.
6.2.3.1.1
PHY Auto-Negotiation (Speed, Duplex, Flow-Control)
The PHY performs the auto-negotiation function. The details of this operation are
described in the IEEE P802.3ab draft standard and are not included here.
Auto-negotiation provides a method for two link partners to exchange information in a
systematic manner in order to establish a link configuration providing the highest
common level of functionality supported by both partners. Once configured, the link
partners exchange configuration information to resolve link settings such as:
• Speed: 10/100/1000 Mb/s
• Duplex: full or half
• Flow control operation
PHY specific information required for establishing the link is also exchanged.
Note:
If flow control is enabled in the 82583V, the settings for the desired flow control
behavior must be set by software in the PHY registers and auto-negotiation restarted.
After auto-negotiation completes, the software device driver must read the PHY
registers to determine the resolved flow control behavior of the link and reflect these in
the MAC register settings (CTRL.TFCE and CTRL.RFCE). If no software device driver is
loaded and auto-negotiation is enabled, then hardware sets these bits in accordance
with the auto-negotiation results.
Note:
Note:
By default, the PHY advertises flow control support.
Once PHY auto-negotiation completes, the PHY asserts a link indication (LINK) to the
MAC. Software must set the Set Link Up bit in the Device Control register (CTRL.SLU)
before the MAC recognizes the link indication from the PHY and can consider the link to
be up.
98
Interconnects—82583V GbE Controller
6.2.3.1.2
MAC Speed Resolution
For proper link operation, both the MAC and PHY must be configured for the same
speed of link operation. The speed of the link can be determined and set by several
methods with the 82583V. These include:
• Software-forced configuration of the MAC speed setting based on PHY indications,
which can be determined as follows:
— Software reads of PHY registers directly to determine the PHY's auto-negotiated
speed
— Software reads the PHY's internal PHY-to-MAC speed indication (SPD_IND)
using the MAC STATUS.SPEED register
— Software signals the MAC to attempt to auto-detect the PHY speed from the
PHY-to-MAC RX_CLK, then programs the MAC speed accordingly
• The MAC automatically detecting and setting the link speed of the MAC based on
PHY indications by:
— Using the PHY's internal PHY-to-MAC speed indication (SPD_IND), setting the
MAC speed automatically
— Attempting to auto-detect the PHY speed from the PHY-to-MAC RX_CLK and
setting the MAC speed automatically
Aspects of these methods are discussed in the sections that follow.
6.2.3.1.2.1 Forcing MAC Speed
There might be circumstances when the software device driver must forcibly set the
link speed of the MAC. This can occur when the link is manually configured. To force the
MAC speed, the software device driver must set the CTRL.FRCSPD (force-speed) bit to
1b and then write the speed bits in the Device Control register (CTRL.SPEED) to the
desired speed setting. See section 9.2.2.1 for details.
Note:
Note:
Forcing the MAC speed using CTRL.FRCSPD overrides all other mechanisms for
configuring the MAC speed and can yield non-functional links if the MAC and PHY are
not operating at the same speed/configuration.
When forcing the 82583V to a specific speed configuration, the software device driver
must also ensure the PHY is configured to a speed setting consistent with MAC speed
settings. This implies that software must access the PHY registers to either force the
PHY speed or to read the PHY status register bits that indicate link speed of the PHY.
Forcing speed settings by CTRL.SPEED can also be accomplished by setting the
CTRL_EXT.SPD_BYPS bit. This bit bypasses the MAC's internal clock switching logic and
enables the software device driver complete control of when the speed setting takes
place. The CTRL.FRCSPD bit uses the MAC's internal clock switching logic, which does
delay the affect of the speed change.
6.2.3.1.2.2 Using PHY Direct Link-Speed Indication
The 82583V PHY provides a direct internal indication of its speed to the MAC
(SPD_IND). The most direct method for determining the PHY link speed and either
manually or automatically configuring the MAC speed is based on these direct speed
indications.
For MAC speed to be set/determined from these direct internal indications from the
PHY, the MAC must be configured such that CTRL.ASDE and CTRL.FRCSPD are both 0b
(both auto-speed detection and forced-speed override are disabled). As a result, the
MAC speed is reconfigured automatically each time the PHY indicates a new link-up
event to the MAC.
99
82583V GbE Controller—Interconnects
When MAC speed is neither forced nor auto-sensed by the MAC, the current MAC speed
setting and the speed indicated by the PHY is reflected in the Device Status register bits
STATUS.SPEED.
6.2.3.1.3
MAC Full/Half Duplex Resolution
The duplex configuration of the link is also resolved by the PHY during the auto-
negotiation process. The 82583V PHY provides an internal indication to the MAC of the
resolved duplex configuration using an internal full-duplex indication (FDX).
This internal duplex indication is normally sampled by the MAC each time the PHY
indicates the establishment of a good link (LINK indication). The PHY's indicated duplex
configuration is applied in the MAC and reflected in the MAC Device Status register
(STATUS.FD).
Software can override the duplex setting of the MAC via the CTRL.FD bit when the
CTRL.FRCDPLX (force duplex) bit is set. If CTRL.FRCDPLX is 0b, the CTRL.FD bit is
ignored and the PHY's internal duplex indication applied.
6.2.3.1.4
Using PHY Registers
The software device driver might be required under some circumstances to read from
or write to the MII management registers in the PHY. These accesses are performed via
the MDIC registers (see section 9.2.2.7). The MII registers enable the software device
driver to have direct control over the PHY's operation, which might include:
• Resetting the PHY
• Setting preferred link configuration for advertisement during the auto-negotiation
process
• Restarting the auto-negotiation process
• Reading auto-negotiation status from the PHY
• Forcing the PHY to a specific link configuration
The set of PHY management registers required for all PHY devices can be found in the
IEEE P802.3ab draft standard. The registers for the 82583V PHY are described in
section 9.2.
6.2.3.1.5
Comments Regarding Forcing Link
Forcing link requires the software device driver to configure both the MAC and PHY in a
consistent manner with respect to each other. After initialization, the software device
driver configures the desired modes in the MAC, then accesses the PHY registers to set
the PHY to the same configuration.
Before enabling the link, the speed and duplex settings of the MAC can be forced by
software using the CTRL.FRCSPD, CTRL.FRCDPX, CTRL.SPEED, and CTRL.FD bits. After
the PHY and MAC have both been configured, the software device driver should write a
1b to the CTRL.SLU bit.
6.2.4
Loss of Signal/Link Status Indication
PHY LOS/LINK signal provides an indication of physical link status to the MAC. This
signal from the PHY indicates whether the link is up or down; typically indicated after
successful auto-negotiation. Assuming that the MAC is configured with CTRL.SLU = 1b,
the MAC status bit STATUS.LU when read, generally reflects whether the PHY has link
(except under forced-link setup where even the PHY link indication might have been
forced).
100
Interconnects—82583V GbE Controller
When the link indication from the PHY is de-asserted, the MAC considers this to be a
transition to a link-down situation (such as, cable unplugged, loss of link partner, etc.).
If the LSC (Link Status Change) interrupt is enabled, the MAC generates an interrupt to
be serviced by the software device driver. See section 7.4 and section 9.2.4 for more
details.
6.2.5
10/100 Mb/s Specific Performance Enhancements
Adaptive IFS
6.2.5.1
The 82583V supports back-to-back transmit Inter-Frame-Spacing (IFS) of 960 ns in
100 Mb/s operation and 9.6 μs in 10 Mb/s operation. Although back-to-back
transmission is normally desirable, sometimes it can actually hurt performance in half-
duplex environments due to excessive collisions. Excessive collisions are likely to occur
in environments where one station is attempting to send large frames back-to-back,
while another station is attempting to send acknowledge (ACK) packets.
The 82583V contains an Adaptive IFS register (see section 9.2.6.3) that enables the
implementation of a driver-based adaptive IFS algorithm for collision reduction, which
is similar to Intel's other Ethernet products (such as PRO/100 adapters). Adaptive IFS
throttles back-to-back transmissions in the transmit MAC and delays their transfer to
the CSMA/CD transmit function and then can be used to delay the transmission of
back-to-back packets on the wire. Normally, this register should be set to zero.
However, if additional delay is desired between back-to-back transmits, then this
register can be set with a value greater than zero. This can be helpful in high-collision
half-duplex environments.
The AIFS field provides a similar function to the IGPT field in the TIPG register (see
section 9.2.6.3). However, this Adaptive IFS throttle register counts in units of GTX/
MTX_CLK clocks, which are 800 ns, 80 ns, 8 ns for 10/100/1000 Mb/s mode
respectively, and is 16 bits wide, thus providing a greater maximum delay value.
Using values lower than a certain minimum (determined by the ratio of GTX/MTX_CLK
clock to link speed), has no effect on back-to-back transmission. This is because the
82583V does not start transmission until the minimum IEEE IFS (9.6 μs at 10 Mb/s,
960 ns at 100 Mb/s, and 96 ns at 1000 Mb/s) has been met regardless of the value of
Adaptive IFS. For example, if the 82583V is configured for 100 Mb/s operation, the
minimum IEEE IFS at 100 Mb/s is 960 ns. Setting AIFS to a value of 10 (decimal) would
not effect back-to-back transmission time on the wire because the 800 ns delay
introduced (10 * 80 ns = 800 ns) is less than the minimum IEEE IFS delay of 960 ns.
However, setting this register with a value of 20 (decimal), which corresponds to
1600 ns for the above example, would delay back-to-back transmits because the
ensuing 1600 ns delay is greater than the minimum IFS time of 960 ns.
It is important to note that this register has no effect on transmissions that occur
immediately after receives or on transmissions that are not back-to-back (unlike the
IPGR1 and IPGR2 values in the TIPG register (see section 9.2.6.2). In addition,
Adaptive IFS also has no effect on re-transmission timing (re-transmissions occur after
collisions). Therefore, AIFS is only enabled in back-to-back transmission.
Note:
The AIFS value is not additive to the TIPG.IPGT value; instead, the actual IPG equals
the larger of the two, AIFS and TIPG.IPGT.
101
82583V GbE Controller—Interconnects
6.2.6
Flow Control
Flow control as defined in 802.3x, as well as the specific operation of asymmetrical flow
control defined by 802.3z, are supported in the MAC. The following seven registers are
defined for the implementation of flow control:
• Flow Control Address Low (FCAL) - 6-byte flow control multicast address
• Flow Control Address High (FCAH) - 6-byte flow control multicast address
• Flow Control Type (FCT) - 16-bit field that indicates flow control type
• Flow Control Receive Thresh Hi (FCRTH) - 13-bit high-water mark indicating receive
buffer fullness
• Flow Control Receive Thresh Lo (FCRTL) - 13-bit low-water mark indicating receive
buffer emptiness
• Flow Control Transmit Timer Value (FCTTV) - 16-bit timer value to include in
transmitted pause frames
• Flow Control Refresh Threshold Value (FCRTV) - 16-bit pause refresh threshold
value
Flow control allows for local controlling of network congestion levels. Flow control is
implemented as a means of reducing the possibility of receive buffer overflows. Receive
buffer overflows result in the dropping of received packets. Flow control is
accomplished by notifying the transmitting station that the receiving station receive
buffer is nearly full.
Implementing asymmetric flow control allows for one link partner to send flow control
packets while being allowed to ignore their reception. For example, not required to
respond to pause frames.
6.2.6.1
MAC Control Frames and Reception of Flow Control Packets
Three comparisons are used to determine the validity of a flow control frame. All three
must be true for a positive result.
1. A match on the six-byte multicast address for MAC control frames or to the station
address of the device (Receive Address Register 0).
2. A match on the Type field.
3. A comparison of the MAC Control Opcode field.
The 802.3x standard defines the MAC control frame multicast address as 01-80-C2-00-
00-01. This address must be loaded into the Flow Control Address Low/High registers
(FCAL/FCAH).
The Flow Control Type (FCT) register contains a 16-bit field that is compared against
the flow control packet's Type field to determine if it is a valid flow control packet: XON
or XOFF. 802.3x reserves this as 0x8808. This value must be loaded into the Flow
Control Type register.
The final check for a valid pause frame is the MAC control opcode. At this time, only the
pause control frame opcode is defined. It has a value of 0x0001.
Frame-based flow control differentiates XOFF from XON based on the value of the
Pause Timer field. Non-zero values constitute XOFF frames while a value of zero
constitutes an XON frame. Values in the timer field are in units of slot time. A slot time
is hard wired to 64-byte times or 512 ns.
Note:
An XON frame signals the cancellation of the pause from being initiated by an XOFF
frame (pause for zero slot times).
102
Interconnects—82583V GbE Controller
1 Byte
Up to 6 Bytes
1 Byte
S
Preamble...
SFD
Destination
Address
6 Bytes
Source
Address
6 Bytes
2 Bytes
2 Bytes
Type/Length
MAC Control
Opcode
MAC Control
Parameters
(min_FrameSize -160)/8
Bytes
4 Bytes
1 Byte
FCS
T
Figure 25.
802.3x MAC Control Frame Format
Where S is the start-of-packet delimiter and T is the first part of the end-of-packet
delimiters for 802.3z encapsulation.
The receiver is enabled to receive flow control frames if flow control is enabled via the
RFCE bit in the Device Control (CTRL) register.
Note:
Flow control capability must be negotiated between link partners via the auto-
negotiation process. The auto-negotiation process might modify the value of these bits
based on the resolved capability between the local device and the link partner.
Once the receiver validates receiving an XOFF or pause frame, the 82583V performs
the following:
• Increments the appropriate statistics register(s).
• Sets the TXOFF bit in the Device Status (STATUS) register.
• Initializes the pause timer based on the packet's Pause Timer field.
• Disables packet transmission or schedules the disabling of transmissions after the
current packet completes.
Resuming transmission can occur under the following conditions:
• An expired pause timer
• Receiving an XON frame (a frame with its pause timer set to zero)
Either condition clears the TXOFF status bit in the Device Status register and
transmission can resume. Note that hardware records the number of received XON
frames.
103
82583V GbE Controller—Interconnects
6.2.6.2
Discard Pause Frames and Pass MAC Control Frames
Two bits in the Receive Control register are implemented specifically for control over
receipt of pause and MAC control frames. These bits are Discard PAUSE Frames (DPF)
and Pass MAC Control Frames (PMCF). See section 9.2.6.2 for DPF and PMCF bit
definitions.
The DPF bit forces the discarding of any valid pause frame addressed to the 82583V's
station address. If the packet is a valid pause frame and is addressed to the station
address (receive address [0]), the 82583V does not pass the packet to host memory if
the DPF bit is set to logic high. However, if a flow control packet is sent to the station
address and is a valid flow control frame, it is then be transferred when DPF is set to
0b. This bit has no affect on pause operation, only the DMA function.
The PMCF bit enables for the passing of any valid MAC control frames to the system,
which does not have a valid pause opcode. In other words, the frame must have the
correct MAC control frame multicast address (or the MAC station address) as well as
the correct Type field match with the FCT register, but does not have the defined pause
opcode of 0x0001. Frames of this type are transferred to host memory when PMCF is a
logic high.
6.2.6.3
Transmitting PAUSE Frames
Transmitting pause frames is enabled by software by writing a 1b to the TFCE bit in the
Device Control register.
Note:
Similar to receiving flow control packets, XOFF packets can be transmitted only if this
configuration has been negotiated between the link partners via the auto-negotiation
process. In other words, setting this bit indicates the desired configuration. Resolving
the auto-negotiation process is described in section 6.2.3.
The content of the Flow Control Receive Threshold High register determines at what
point hardware transmits a pause frame. Hardware monitors the fullness of the receive
FIFO and compares it with the contents of FCRTH. When the threshold is reached,
hardware sends a pause frame with its pause time field equal to FCTTV.
At the time threshold is reached, the hardware starts counting an internal shadow
counter FCRTV (reflecting the pause time-out counter at the partner end) from zero.
When the counter reaches the value indicated in the FCRTV register, then, if the pause
condition is still valid (meaning that the buffer fullness is still above the low
watermark), an XOFF message is sent again and the shadow counter starts counting
again.
Once the receive buffer fullness reaches the low water mark, hardware sends an XON
message (a pause frame with a timer value of zero). Software enables this capability
with the XONE field of the FCRTL.
Hardware sends one more pause frame if it has previously sent one and the FIFO
overflows (so the threshold must not be set greater than the FIFO size). This is
intended to minimize the amount of packets dropped if the first pause frame does not
reach its target. Since the secure receive packets use the same data path, the behavior
is identical when secure packets are received.
Note:
Note:
Transmitting flow control frames should only be enabled in full-duplex mode per the
IEEE 802.3 standard. Software should ensure that transmitting flow control packets is
disabled when the 82583V is operating in half-duplex mode.
Regardless of the mechanism above, each time a receive packet is dropped due to lack
of space in the internal receive buffer, a pause frame is transmitted as well (if TFCE bit
in the Device Control register is enabled).
104
Interconnects—82583V GbE Controller
6.2.6.4
Software Initiated Pause Frame Transmission
The 82583V has the added capability to transmit an XOFF frame via software. This is
accomplished by software writing a 1b to the SWXOFF bit of the Transmit Control
register. Once this bit is set, hardware initiates transmitting a pause frame in a manner
similar to that automatically generated by hardware.
The SWXOFF bit is self-clearing after the pause frame has been transmitted.
The state of the CTRL.TFCE bit or the negotiated flow control configuration does not
affect software generated pause frame transmission.
Note:
Note:
Software sends an XON frame by programming a zero in the Pause Timer field of the
FCTTV register.
XOFF transmission is not supported in 802.3x for half-duplex links. Software should not
initiate an XOFF or XON transmission if the 82583V is configured for half-duplex
operation.
6.3
SPI Non-Volatile Memory Interface
General Overview
6.3.1
The 82583V requires non-volatile content for the 82583V configuration. The Non-
Volatile Memory (NVM) might contain the following main regions:
• LAN configuration space accessed by hardware - loaded by the 82583V after power
up, PCI reset de-assertion, D3->D0 transition, or a software commanded EEPROM
read (CTRL_EXT.EE_RST).
• LAN configuration space accessed by software - used by software only. The
meaning of these registers as listed here as a convention for the software only and
is ignored by the 82583V.
6.3.2
Supported NVM Devices
Some Intel LAN controllers require both an EEPROM and Flash device for storing LAN
data. However, the 82583 reduces the Bill of Material (BOM) cost by consolidating the
EEPROM and Flash into a single non-volatile memory device. The NVM is connected to a
single Serial Peripheral Interface (SPI).
The 82583 is compatible with many sizes of 4-wire SPI NVM devices. The required NVM
size is dependent upon system requirements.
105
82583V GbE Controller—Interconnects
Table 34.
NVM Configuration Size
Configuration
Minimum NVM Size
Memory Family
1
No iSCSI boot
1 Kb
SPI EEPROM
SPI Flash
SPI Flash
SPI Flash
PXE only
512 Kb
2 Mb
4 Mb
iSCSI boot only
Both iSCSI boot and PXE
Table 35.
Compatible EEPROM Parts
Vendor
1 Kb
2 Kb
32 Kb
Atmel*
STM*
AT25010N
95010W6
CAT25010S
AT25020N
95020W6
AT25010N-10SI-2.7
AT25320N
Catalyst*
CAT25020S
CAT25C320S
Table 36.
Compatible Flash Parts
Vendor
Winbond*
512 Kb
N/A
1 Mb
2 Mb
4 Mb
8 Mb
W25X10BV
N/A
W25X20BV
AT25DF021
25LV020
W25X40BV
AT25DF041A
N/A
N/A
Atmel
PMC*
SST*
AT25F512B
25LV512A
AT25DF081A
N/A
25LV010A
SST25VF010A
SST25VF512
SST25VF020B
SST25VF040B
SST25VF080B
6.3.3
NVM Device Detection
The 82583V detects the device connected on the SPI interface in two phases.
1. It first detects the device type by the state of the NVMT strapping pin.
2. It then looks at the NVM content depending on a valid signature in word 0x12 in the
NVM.
In reference to the EEPROM, the 82583V detects the length of the address bytes by
sensing the signature at word 0x12. It then sets the NVADDS field in the EEC register.
The exact size of the NVM is fetched by the 82583V from word 0x0F and is stored in the
NVSize field in the EEC register. When operating with an EEPROM that has an invalid
signature, software can force the address length via the NVADDS field in the EEC
register. Controlling the address length enables software to access the EEPROM via the
parallel EERD and EEWR registers in all cases including invalid signature.
6.3.3.1
CRC Field
CRC calculation is done by software.
106
Interconnects—82583V GbE Controller
6.3.4
6.3.5
Device Operation with an External EEPROM
When the 82583V is connected to an external EEPROM, it provides similar functionality
to its predecessors with the following enhancements:
• Enables a complete parallel interface for read/write to the EEPROM.
• Enables software to specify explicitly the address length, thus eliminating the need
for bit banging access even on an empty EEPROM.
Device Operation with Flash
As previously stated, the 82583V merges the legacy EEPROM and Flash content in a
single Flash device. The 82583V copies the lower section in the Flash device to an
internal shadow RAM. The interface to the shadow RAM is the same as the interface for
an external EEPROM device. This mechanism provides a seamless backward compatible
interface for software to the legacy EEPROM space as if an external EEPROM device is
connected.
The 82583V supports Flash devices with a block erase size of 4 KB. Note that many
Flash vendors are using the term sector differently. This document uses the term Flash
sector for a logic section of 4 KB.
6.3.5.1
LAN Configuration Sectors
Flash devices require a block erase instruction in case a cell is modified from 0b to 1b.
As a result, in order to update a single byte (or block of data) it is required to erase it
first. The first addresses of the Flash contain the device configuration and must always
be valid. The 82583V maintains two sectors of 4 KB: S0 and S1 for the configuration
content. At least one of these two sectors is valid at any given time or else the 82583V
is set by the hardware default. section 6.3.6 provides more details on the shadow RAM
and the first two sectors.
6.3.6
Shadow RAM
The 82583V includes an internal 4 KB shadow RAM of the first 4 KB Flash sector(s).
When the 82583V is connected to a Flash device the legacy configuration parameters
might reside in any of the first two 4 KB sectors (S0 or S1) in the Flash. The 82583V
copies that data to an internal shadow memory. The shadow RAM emulates a seamless
EEPROM interface to the rest of the 82583V and host CPU. This way the legacy
configuration content is accessible to software and firmware on the same EEPROM
registers as on previous GbE controllers.
Figure 26 shows the shadow RAM mapping and interface relative to the Flash and the
EEPROM. The external EEPROM and the shadow RAM share the same interface. The
82583V might access the EEPROM or shadow RAM according to the setting of the
SELSHAD bit in the EEC register. By hardware default, the SELSHAD bit is set by the
NVMT strapping pin so that the EEPROM is selected in case of external EEPROM and the
shadow RAM is selected in the case of external Flash.
Note:
Access to the shadow RAM uses the same interface as the external EEPROM with the
exception that bit banging is not supported for the shadow RAM.
107
82583V GbE Controller—Interconnects
LAN Flash
…
Address 8K
Address 4K
EEPROM
Sector 1
Sector 0
Shadow RAM
…
Address 00
EEC.SELSHAD
EEPROM Interface
Figure 26.
6.3.6.1
NVM Shadow RAM
Flash Mode
The 82583V is initialized from the NVM. As part of the initialization sequence, the
82583V copies the 4 KB content of S0 or S1 from the Flash to the shadow RAM. Any
access to the EEPROM interface is directed to the shadow RAM. Following any write
access to the shadow RAM by software or firmware, the data should also be updated in
the Flash. The 82583V maintains a watchdog timer defined by the FLASHT register to
minimize Flash updates. The timer is triggered by any write access to the shadow RAM.
The 82583V updates the Flash from the shadow RAM when the FLASHT timer expires or
when firmware or software request explicitly to update the Flash by setting the FLUPD
bit in the FLA register. The 82583V copies the content of the shadow RAM to the
inactive configuration sector and then makes it the active one. The Flash update
sequence is listed in the steps that follow:
1. Initiates block erase instruction(s) to the inactive sector (the inactive sector is
defined by the inverse value of the SEC1VAL bit in the EEC register).
2. Copy the shadow RAM to the inactive sector while the signature word is copied last.
3. Clear the signature word in the active sector to make it invalid.
4. Toggle the state of the SEC1VAL bit in the EEC register to indicate that the inactive
sector became the active one and visa versa.
Note:
Software should be aware of the fact that actual programming to the Flash might
require a long latency following the write access to the shadow RAM. Software might
poll the FLUDONE bit in the FLCTL register to complete the Flash programming, when
required.
6.3.6.2
EEPROM Mode
When the 82583V is attached to an external EEPROM, any access to the EEPROM
interface is directed to the external EEPROM.
108
Interconnects—82583V GbE Controller
6.3.7
NVM Clients and Interfaces
There are several clients that might access the NVM or shadow RAM listed in the
following table. Listed are the various clients and their access type to the NVM:
software device driver, BIOS, firmware and hardware.
Table 37.
Clients and Access Type to the NVM
Client + Interface
NVM port
EEPROM
NVM instructions
Host CPU on EEC CSR
Legacy bit banging
Host CPU on EERD and
EEWR
Parallel word read and write to EEPROM or shadow RAM
(controlled by the EEC.SELSHAD bit)
EEPROM
EE CSR
EEPROM
Flash
Parallel word read and write to EEPROM or shadow RAM
Legacy bit banging and Flash erase instructions
Host CPU on FLA CSR
Host CPU via BAR
1
Flash
Read byte word and Dword and byte programming
Host CPU via FLSWxxx
CSR registers
Host write access to the Flash no support for burst (multiple
byte) writes
Flash
Both
Direct HW accesses
Read EEPROM/shadow RAM at device initialization
1. Following a write instruction or erase instructions to the Flash, the 82583V initiates seamless write enable
before the write or erase instructions and polls the status at the end to check its completion.
6.3.7.1
Memory Mapped Host Interface via LAN Flash BAR
Software might read and write to the Flash via the LAN Flash BAR. The Flash BAR is
mapped to the physical Flash at offset 0x0. The 82583V supports read byte, word or
Dword and write byte through this interface. The host CPU waits (stalled) until the read
access to the Flash completes.
Note:
One of the first two sectors of 4 KB in the Flash are also reflected in the shadow RAM.
During normal operation, when software requires access to these sectors it should
access the shadow RAM. Direct write accesses to the Flash in this space via the Flash
BAR might cause non-coherency between the Flash and the shadow RAM.
Note:
Flash BAR access while FLA.FL_REQ is asserted (and granted) is forbidden.
6.3.7.2
CSR Mapped Host Interface
Software has bit banging and parallel accesses to the NVM or shadow RAM via the
registers in the CSR space. The 82583V supports the following cycles on the parallel
interface: posted write, posted read, block erase and device erase. Access to the
configuration space in the first two sectors is directed via the EEPROM registers
regardless of the external physical device. Access to the rest of the NVM space is done
according to the type of the physical device: Flash registers in reference to Flash and
EEPROM registers in reference to EEPROM. EEPROM CSR registers are as follows:
• EEC register for bit banging and device control
• EERD and EEWR registers for parallel read and write access
The Flash CSR registers are as follows:
• FLA register and EEC register for bit banging and device control
109
82583V GbE Controller—Interconnects
Note:
When software accesses the EEPROM or Flash spaces via the bit banging interface, it
should follow these steps:
1. Write a 1b to the Request bit in the FLA or EEC registers.
2. Poll the Grant bit in the FLA or EEC registers until its ready.
3. Access the NVM using the direct interface to its signaling via the EEC or FLA
registers.
4. When access completes, software should clear the Request bit.
Note:
Following a write or erase instruction, software should clear the Request bit only after it
checked that the cycles were completed by the NVM.
6.3.7.3
CSR Mapped Firmware Interface
Firmware might access the NVM or shadow RAM via the NVM Control registers in the
CSR space with the following capabilities:
• Word read and write accesses to the EEPROM or shadow RAM via the EECTL and
EEDATA registers.
• Read and write DMA and block erase to the Flash interface via the FLCTL and
FLDATA registers. Flash accesses are mapped to the physical NVM at offset 0x0.
Note that nominal accesses to the first two 4 KB sectors should be addressed to the
shadow RAM via the EEPROM interface.
6.3.8
NVM Write and Erase Sequence
6.3.8.1
Software Flow to the Bit Banging Interface
When software accesses the EEPROM or Flash CSR registers to the bit banging interface
it should follow these steps:
1. Write a 1b to the Request bit in the FLA or EEC registers.
2. Poll the Grant bit in the FLA or EEC registers until its ready.
3. Access the NVM using the direct interface to its signaling via the EEC or FLA
registers.
4. When access is achieved, software should clear the Request bit. Note that following
a write or erase instruction, software should clear the Request bit only after it
checked that the cycles were completed by the NVM.
6.3.8.2
Software Byte Program Flow to the EEPROM Interface
Software initiates a write cycle to the NVM on the parallel EEPROM as follows:
1. Poll the Done bit in the EEWR register until its set.
2. Write the data word, its address, and the Start bit to the EEWR register.
As a response, hardware executes the following steps:
Case 1 - The 82583V is connected to a physical EEPROM device:
1. Initiate an autonomous write enable instruction.
2. Initiate the program instruction right after the enable instruction.
3. Poll the EEPROM status until programming completes.
4. Set the Done bit in the EEWR register.
110
Interconnects—82583V GbE Controller
Case 2 - The 82583V is connected to a physical Flash device:
1. The 82583V writes the data to the shadow RAM and sets the Done bit in the EEWR
register.
2. Update of the shadow RAM to the Flash device as described in section 6.3.6.
6.3.8.3
Flash Byte Program Flow
Software initiates a byte write cycle via the Flash BAR as follows:
1. Write access to the Flash must be first enabled in the FLEW field in the EEC register.
2. Poll the FLBUSY flag in the FLA register until cleared.
3. Write the data byte to the Flash through the Flash BAR.
4. Repeat the steps 2 and 3 if multiple bytes should be programmed.
5. Clear the write enable in the FLEW field in the EEC register to protect the Flash
device.
As a response, hardware executes the following steps for each write access:
1. Initiate autonomous write enable instruction.
2. Initiate the program instruction right after the enable instruction.
3. Poll the Flash status until programming completes.
4. Clear the FLBUSY bit in the FLA register.
Note:
This section explains only the actual programming of a single byte or multiple bytes.
6.3.8.4
Flash Erase Flow
Device Erase Flow:
Erase instructions flow by software is almost identical to the program flow:
1. Erase access to the Flash must be first enabled in the FLEW field in the EEC
register.
2. Poll the FLBUSY flag in the FLA register until cleared.
3. Set the Flash Erase bit (FL_ER) in the FLA register.
4. Clear the Erase enable in the FLEW field in the EEC register to protect the Flash
device.
6.3.8.5
Flash Burst Program Flow
The 82583V provides a burst engine that can be useful for initial programming of the
entire Flash image according to the following flow:
1. Set the ADDR field with the byte resolution address in the FLSWCTL register.
2. Set the CMD field to 01b, which is the DMA write setting in the FLSWCTL register.
3. Write the first 32 bits of data to the FLSWGDATA register.
4. Set the RDCNT field to the byte count number in the FLSWCNT register.
5. Set the CMDV field in the FLSWCTL register to start a DMA write.
6. Hardware starts accessing the SPI bus and begins writing the first 32 bits from the
FLSWDATA register.
7. Once hardware writes the 32-bit data to the Flash, the DONE bit in the FLSWCTL
register is set indicating the next 32 bits are required.
111
82583V GbE Controller—Interconnects
8. Until new data is written to the FLSWDATA register, the Flash clock is paused.
9. Once data is written to the FLSWDATA by the software, the DONE bit in the
FLSWCTL register is cleared and is set after hardware writes it to the Flash.
10.After all bytes are written to the Flash, hardware completes the cycle on the SPI
bus and sets the WRDONE bit in the FLSWCTL register indicating that the entire
burst has completed.
6.3.8.6
Flash Programming Flow of S0 and S1
Other than initial programming of the Flash device, software and firmware should not
access the configuration sectors: S0 and S1. Any access to the configuration flow
should go to the Shadow RAM via the EEPROM interface registers.
112
Interconnects—82583V GbE Controller
Note:
This page intentionally left blank.
113
82583V GbE Controller—Inline Functions
7.0
Inline Functions
7.1
Packet Reception
Packet reception consists of recognizing the presence of a packet on the wire,
performing address filtering, storing the packet in the receive data FIFO, transferring
the data to the receive queue in host memory, and updating the state of a receive
descriptor.
7.1.1
Packet Address Filtering
Hardware stores incoming packets in host memory subject to the following filter
modes. If there is insufficient space in the receive FIFO, hardware drops them and
indicates the missed packet in the appropriate statistics registers.
The following filter modes are supported:
• Exact unicast/multicast
— The destination address must exactly match one of 16 stored addresses. These
addresses can be unicast or multicast.
Note:
The software device driver can only use 15 entries (entries 0-14).
• Promiscuous unicast
— Receive all unicasts
• Multicast
The upper bits of the incoming packet's destination address index is a bit vector that
indicates whether to accept the packet; if the bit in the vector is one, accept the
packet, otherwise, reject it. The 82583V provides a 4096-bit vector. Software provides
four choices of which bits are used for indexing. These are [47:36], [46:35], [45:34],
or [43:32] of the internally stored representation of the destination address (see
Figure 43)
• Promiscuous multicast
— Receive all multicast packets
• VLAN
Receive all VLAN packets that are for this station and have the appropriate bit set in the
VLAN filter table. A detailed discussion and explanation of VLAN packet filtering is
contained in section 7.5.3.
Normally, only good packets are received.
114
Inline Functions—82583V GbE Controller
Good packets are defined as those packets with no:
• CRC error
• Symbol error
• Sequence error
• Length error
• Alignment error
• Where carrier extension or RX_ERR errors are detected.
However, if the Store-Bad-Packet bit is set in the Device Control register (RCTL.SBP),
then bad packets that pass the filter function are stored in host memory. Packet errors
are indicated by error bits in the receive descriptor (RDESC.ERRORS). It is possible to
receive all packets, regardless of whether they are bad, by setting the promiscuous
enables and the Store-Bad-Packet bit.
Note:
CRC errors before the SFD are ignored. Every packet must have a valid SFD (RX_DV
with no RX_ER in the GMII/MII interface) in order to be recognized by the device (even
bad packets).
7.1.2
Receive Data Storage
Memory buffers pointed to by descriptors store packet data. Hardware supports the
following receive buffer sizes:
• 256B 512B 1024B 2048B 4096B 8192B 16384B
• FLXBUF x 1024B while FLXBUF=1,2,3,…15
Buffer size is selected by bit settings in the Receive Control register (RCTL.BSIZE,
RCTL.BSEX, RCTL.DTYP and RCTL. FLXBUF).
The 82583V (in legacy mode) places no alignment restrictions on receive memory
buffer addresses. This is desirable in situations where the receive buffer was allocated
by higher layers in the networking software stack, as these higher layers might have no
knowledge of a specific device's buffer alignment requirements.
Note:
Note:
Although alignment is completely unrestricted, it is highly recommended that software
allocate receive buffers on at least cache-line boundaries whenever possible.
The larger buffer sizes are only offered to provide compatibility with other silicon types.
Jumbo frames are not supported with the 82583V.
7.1.3
Legacy Receive Descriptor Format
A receive descriptor is a data structure that contains the receive data buffer address
and fields for hardware to store packet information. If the RFCTL.EXSTEN bit is clear
and the RCTL.DTYP equals 00b, the 82583V uses the Legacy Rx Descriptor as shown in
the following figure.
115
82583V GbE Controller—Inline Functions
63
48 47
40 39
Buffer Address [63:0]
Status Packet Checksum
32 31
16 15
0
0
8
1
VLAN Tag
Errors
Length
1. The checksum indicated here is the unadjusted 16-bit ones complement of the packet. A software assist might
be required to back out appropriate information prior to sending it up to upper software layers. The packet
checksum is always reported in the first descriptor (even in the case of multi-descriptor packets).
Figure 27.
7.1.3.1
82583V Legacy Rx Descriptor
Length Field (16-Bit, Offset 0)
Upon receipt of a packet for this device, hardware stores the packet data into the
indicated buffer and writes the length, Packet Checksum, Status, Errors, and Status
fields. Length covers the data written to a receive buffer including CRC bytes (if any).
Note:
Software must read multiple descriptors to determine the complete length for packets
that span multiple receive buffers.
7.1.3.2
Packet Checksum (16-Bit, Offset 16)
For standard 802.3 packets (non-VLAN) the packet checksum is by default computed
over the entire packet from the first byte of the DA through the last byte of the CRC,
including the Ethernet and IP headers. Software can modify the starting offset for the
packet checksum calculation via the Receive Checksum Control register (RXCSUM).
This register is described in section 9.2.5.15. To verify the TCP/UDP checksum using
the packet checksum, software must adjust the packet checksum value to back out the
bytes that are not part of the true TCP checksum. When operating with the legacy Rx
descriptor, the RXCSUM.IPPCSE and the RXCSUM.PCSD should be cleared (the default
value).
For packets with VLAN header the packet checksum includes the header if VLAN
striping is not enabled by the CTRL.VME. If VLAN header strip is enabled, the packet
checksum and the starting offset of the packet checksum exclude the VLAN header.
7.1.3.3
Status Field (8-Bit, Offset 32)
Status information indicates whether the descriptor has been used and whether the
referenced buffer is the last one for the packet.
7
6
5
4
3
2
1
0
Rsvd
IPCS
TCPCS
UDPCS
VP
Rsvd
EOP
DD
Figure 28.
Receive Status (RDESC.STATUS-0) Layout
Rsvd (bit 7) - Reserved
IPCS (bit 6) - IPv4 checksum calculated on packet
116
Inline Functions—82583V GbE Controller
TCPCS (bit 5) - TCP checksum calculated on packet
UDPCS (bit 4) - UDP checksum calculated on packet
VP (bit 3) - Packet is 802.1q (matched VET)
Reserved (bit 2) - Reserved
EOP (bit 1) - End of packet
DD (bit 0) - Descriptor done
EOP: Packets that exceed the receive buffer size spans multiple receive buffers. EOP
indicates whether this is the last buffer for an incoming packet. DD indicates whether
hardware is done with the descriptor. When the DD bit is set along with EOP, the
received packet is completely in main memory. Software can determine buffer usage by
setting the status byte to zero before making the descriptor available to hardware, and
checking it for non-zero content at a later time. For multi-descriptor packets, packet
status is provided in the final descriptor of the packet (EOP set). If EOP is not set for a
descriptor, only the Address, Length, and DD bits are valid.
VP: The VP field indicates whether the incoming packet's type matches VET (for
example, if the packet is a VLAN (802.1q) type). It is set if the packet type matches
VET and CTRL.VME is set. For a further description of 802.1q VLANs, see section 7.5.
IPCS TCPCS UDPCS: These bit descriptions are listed in the following table:
TCPCS
UDPCS
IPCS
Functionality
0b
0b
0b
Hardware does not provide checksum offload.
Hardware provides IPv4 checksum offload if IPCS active and TCP
checksum offload. Pass/fail indication is provided in the Error field
– IPE and TCPE.
1b
1b
0b
1b
1b/0b
1b/0b
Hardware provides IPv4 checksum offload if IPCS active and UDP
checksum offload. Pass/Fail indication is provided in the Error field
– IPE and TCPE.
IPv6 packets do not have the IPCS bit set, but might have the TCPCS bit set if the
82583V recognized the TCP or UDP packet.
7.1.3.4
Error Field (8-Bit, Offset 40)
Most error information appears only when the Store-Bad-Packet bit (RCTL.SBP) is set
and a bad packet is received. Figure 29 shows the definition of the possible errors and
their bit positions.
7
6
5
4
3
2
1
0
RXE
IPE
TCPE
CXE
Rsv
SEQ
SE
CE
Figure 29.
Receive Errors (RDESC.ERRORS) Layout
RXE (bit 7) - Rx data error
IPE (bit 6) - IPv4 checksum error
117
82583V GbE Controller—Inline Functions
TCPE (bit 5) - TCP/UDP checksum error
CXE (bit 4) - Carrier extension error
Rsv (bit 3) - Reserved
SEQ (bit 2) - Sequence error
SE (bit 1) - Symbol error
CE (bit 0) - CRC error or alignment error
The IP and TCP checksum error bits are valid only when the IPv4 or TCP/UDP
checksum(s) is performed on the received packet as indicated via IPCS and TCPCS
previously mentioned. These, along with the other error bits, are valid only when the
EOP and DD bits are set in the descriptor.
Note:
Receive checksum errors have no effect on packet filtering.
If receive checksum offloading is disabled (RXCSUM.IPOFL and RXCSUM.TUOFL), the
IPE and TCPE bits are 0b.
The RXE bit indicates that a data error occurred during the packet reception that has
been detected by the PHY. This generally corresponds to signal errors occurring during
the packet reception. This bit is valid only when the EOP and DD bits are set and are
not set in descriptors unless RCTL.SBP (Store-Bad-Packets) is set.
CRC errors and alignment errors are both indicated via the CE bit. Software can
distinguish between these errors by monitoring the respective statistics registers.
7.1.3.5
VLAN Tag Field (16-Bit, Offset 48)
Hardware stores additional information in the receive descriptor for 802.1q packets. If
the packet type is 802.1q (determined when a packet matches VET and RCTL.VME =
1b), then the VLAN Tag field records the VLAN information and the four-byte VLAN
information is stripped from the packet data storage. Otherwise, the VLAN Tag field
contains 0x0000.
15
13
PRI
12
11
0
CFI
VLAN
7.1.4
Extended Rx Descriptor
If the RFCTL.EXSTEN bit is set and RCTL.DTYP equals 00b, the 82583V uses the
extended Rx descriptor as follows:
Descriptor Read Format:
63
0
0
8
Buffer Address [63:0]
Reserved
0
118
Inline Functions—82583V GbE Controller
7.1.4.1
7.1.4.2
Buffer Address (64-Bit, Offset 0.0)
The field contains the physical address of the receive data buffer. The size of the buffer
is defined by the RCTL register (RCTL.BSIZE, RCTL.BSEX, RCTL.DTYP and RCTL.
FLXBUF fields).
DD (1-Bit, Offset 8.0)
This is the location of the DD bit in the Status field. The software device driver must
clear this bit before it handles the receive descriptor to the 82583V. The software
device driver can use this bit field later on as a completion indication of the hardware.
Descriptor Write-Back Format:
63
48
47
32
31
20
19
MRQ
0
0
8
Reserved
Packet Checksum
VLAN Tag
IP Identification
Length
Extended Error
Extended Status
Note:
Light-blue fields are mutually exclusive by RXCSUM.PCSD
7.1.4.3
MRQ Field (32-Bit, Offset 0.0)
Field
Bit(s)
Description
Reserved
Reserved
Reserved
Reserved
3:0
Reserved
Reserved
Reserved
Reserved
7:4
12:8
31:13
7.1.4.4
Packet Checksum (16-Bit, Offset 0.48)
For standard 802.3 packets (non-VLAN) the packet checksum is by default computed
over the entire packet from the first byte of the DA through the last byte of the CRC,
including the Ethernet and IP headers. Software can modify the starting offset for the
packet checksum calculation via the Receive Checksum Control register (RXCSUM).
This register is described in section 9.2.5.15. To verify the TCP/UDP checksum using
the packet checksum, software must adjust the packet checksum value to back out the
bytes that are not part of the true TCP checksum. Likewise, for fragmented UDP
packets, the Packet Checksum field can be used to accelerate UDP checksum
verification by the host processor. This operation is enabled by the RXCSUM.IPPCSE bit
as described in section 9.2.5.15.
For packets with VLAN header the packet checksum includes the header if VLAN
striping is not enabled by the CTRL.VME bit. If VLAN header strip is enabled, the packet
checksum and the starting offset of the packet checksum exclude the VLAN header.
119
82583V GbE Controller—Inline Functions
7.1.4.5
7.1.4.6
IP Identification (16-Bit, Offset 0.32)
This field stores the IP Identification field in the IP header of the incoming packet. The
software device driver should ignore this field when IPIDV is not set.
Extended Status (20-Bit, Offset 8.0)
9
8
7
6
5
4
3
2
1
0
IPIDV
TST
Rsvd
IPCS
TCPCS
UDPCS
VP
Rsvd
EOP
DD
19
18
17
16
15
14
13
12
11
10
PKTTYPE
ACK
Reserved
UDPV
PKTTYPE (bits 19:16) - Packet type
ACK (bit 15) - ACK packet indication
Reserved (bits 14:11) - Reserved
UDPV (bit 10) - Valid UDP XSUM
IPIDV (bit 9) - IP identification valid
TST (bit 8) - Time stamp taken
Rsvd (bit 7) - Reserved
IPCS (bit 6) IPv4 checksum calculated on packet - same as legacy descriptor.
TCPCS (bit 5) - TCP checksum calculated on packet - same as legacy descriptor.
UDPCS (bit 4) - UDP checksum calculated on packet.
VP (bit 3) - Packet is 802.1q (matched VET) - same as legacy descriptor.
Rsv (bit 2) - Reserved
EOP (bit 1) - End of packet - same as legacy descriptor.
DD (bit 0) - Descriptor done - same as legacy descriptor.
DD EOP IXSM VP UDPCS TCPCS IPCS: Same meaning as in the legacy receive
descriptor.
120
Inline Functions—82583V GbE Controller
IPCS TCPCS UDPCS: The meaning of these bits is shown in the following table:
TCPCS
UDPCS
IPCS
Functionality
0b
0b
1b/0b
Hardware provides IPv4 checksum offload if IPCS active.
Hardware provides IPv4 checksum offload if IPCS active and TCP
checksum offload. Pass/fail indication is provided in the Error field
– IPE and TCPE.
1b
0b
1b
0b
1b
1b
1b/0b
1b/0b
1b/0b
For IPv4 packets, hardware provides IP checksum offload if IPCS
active and fragmented UDP checksum offload. IP Pass/fail
indication is provided in the IPE field. Fragmented UDP checksum
is provided in the packet checksum field if the RXCSUM.PCSD bit is
cleared.
Hardware provides IPv4 checksum offload if IPCS active and UDP
checksum offload. Pass/fail indication is provided in the Error field
– IPE and TCPE.
Unsupported packet types do not have the IPCS or TCPCS bits set. IPv6 packets do not
have the IPCS bit set, but might have the TCPCS bit set if the 82583V recognized the
TCP or UDP packet.
IPIDV (bit 9): The IPIDV bit indicates that the incoming packet was identified as a
fragmented IPv4 packet. The IPID field contains a valid IP identification value if the
RXCSUM.PCSD is cleared.
UDPV (bit 10): The UDPV bit indicates that the incoming packet contains a valid (non-
zero value) checksum field in an incoming fragmented UDP IPv4 packet. It means that
the Packet Checksum field contains the UDP checksum as described in this section.
When this field is cleared in the first fragment that contains the UDP header, it means
that the packet does not contain a valid UDP checksum and the checksum field in the
Rx descriptor should be ignored. This field is always cleared in incoming fragments that
do not contain the UDP header.
ACK (bit 15): The ACK bit indicates that the received packet was an ACK packet with
or without TCP payload depending on the RFCTL.ACKD_DIS bit.
PKTTYPE (bit 19:16): The PKTTYPE field defines the type of the packet that was
detected by the 82583V. The 82583V tries to find the most complex match until the
most common one as shown in the following packet type table:
Packet Type
0x0
Description
MAC, (VLAN/SNAP) payload
0x1
0x2
0x3
0x4
0x5
0x6
0x7
0x8
0x9
0xA
0xB
MAC, (VLAN/SNAP) IPv4, payload
MAC, (VLAN/SNAP) IPv4, TCP/UDP, payload
MAC (VLAN/SNAP), IPv4, IPv6, payload
MAC (VLAN/SNAP), IPv4, IPv6, TCP/UDP, payload
MAC (VLAN/SNAP), IPv6, payload
MAC (VLAN/SNAP), IPv6,TCP/UDP, payload
MAC, (VLAN/SNAP), IPv4, TCP, ISCSI, payload
MAC, (VLAN/SNAP), IPv4, TCP/UDP, NFS, payload
MAC (VLAN/SNAP), IPv4, IPv6,TCP, ISCSI, payload
MAC (VLAN/SNAP), IPv4, IPv6,TCP/UDP,NFS, payload
MAC (VLAN/SNAP), IPv6,TCP, ISCSI, payload
121
82583V GbE Controller—Inline Functions
Packet Type
0xC
Description
MAC (VLAN/SNAP), IPv6,TCP/UDP, NFS, payload
0xD
0xE
Reserved
Reserved
— Payload does not mean raw data but can also be unsupported header.
— If there is an NFS/iSCSI header in the packets it can be seen in the packet type
field.
Note:
If the device is not configured to provide any offload that requires packet parsing, the
packet type field is set to 0b regardless of the actual packet type.
7.1.4.7
Extended Errors (12-Bit, Offset 8.20)
11
10
9
8
7
6
5
4
3
2
1
0
RXE
IPE
TCPE
CXE
Rsvd
SEQ
SE
CE
Rsvd
Rsvd
RXE (bit 11) - Rx data error - Same as legacy descriptor.
IPE (bit 10) - IPv4 checksum error - Same as legacy descriptor.
TCPE (bit 9) - TCP/UDP checksum error - Same as legacy descriptor.
CXE (bit 8) - Carrier extension error - Same as legacy descriptor.
SEQ (bit 6) - Sequence error - Same as legacy descriptor.
SE (bit 5) - Symbol error - Same as legacy descriptor.
CE (bit 4) - CRC error or alignment error - Same as legacy descriptor.
Reserved (bits 7, 3:0) - Reserved
RXE IPE TCPE CXE SEQ SE CE: Same as legacy descriptor.
Length (16-bit, offset 8.32): Same as the length field at offset 8.0 in the legacy
descriptor.
VLAN Tag (16-bit, offset 8.48): Same as legacy descriptor.
7.1.4.7.1
Receive UDP Fragmentation Checksum
The 82583V might provide receive fragmented UDP checksum offload. The following
setup should be made to enable this mode:
RXCSUM.PCSD bit should be cleared. When the PCSD bit is cleared, Packet Checksum
and IP Identification are active.
RXCSUM.IPPCSE bit should be set. This field enables the IP payload checksum enable
that is designed for the fragmented UDP checksum.
122
Inline Functions—82583V GbE Controller
RXCSUM.PCSS field must be zero. The packet checksum start should be zero to enable
auto start of the checksum calculation. See the following table for an exact description
of the checksum calculation.
The following table lists the outcome descriptor fields for the following incoming
packets types:
Incoming
Packet Type
IP
Packet Checksum
UDPV/IPIDV
UDPCS/TCPCS
Identification
Unadjusted 16-bit ones
complement checksum of the
entire packet (excluding VLAN
header)
None IPv4
Packet
Reserved
0b/0b
0b/0b
Fragment
IPv4 with TCP Same as above
header
Incoming IP
Identification
0b/1b
0b/0b
0b/0b
Non-
fragmented
IPv4 packet
Depend on transport
header and TUOFL
field
Same as above
Reserved
Fragmented
IPv4 without
transport
The unadjusted 1’s complement Incoming IP
0b/1b
1b/0b
1b/0b
checksum of the IP payload
Identification
header
1b if the UDP
header
checksum is
valid/1b
Fragmented
IPv4 with UDP Same as above
header
Incoming IP
Identification
Note:
When the software device driver computes the 16-bit ones complement sum on the
incoming packets of the UDP fragments, it should expect a value of 0xFFFF. See
section 7.1.10 for supported packet formats.
7.1.5
Packet Split Receive Descriptor
The 82583V uses the packet split feature when the RFCTL.EXSTEN bit is set and
RCTL.DTYP=01b. The software device driver must also program the buffer sizes in the
PSRCTL register.
Descriptor Read Format:
63
0
0
8
Buffer Address 0
Buffer Address 1
Buffer Address 2
Buffer Address 3
0
16
24
123
82583V GbE Controller—Inline Functions
7.1.5.1
Buffer Addresses [3:0] (4 x 64 bit)
The physical address of each buffer is written in the Buffer Addresses fields. The sizes
of these buffers are statically defined by BSIZE0-BSIZE3 in the PSRCTL register.
Note:
Software Notes:
• All buffers' addresses in a packet split descriptor must be word aligned.
• Packet header can't span across buffers, therefore, the size of the first buffer must
be larger than any expected header size. Otherwise the packet will not be split.
• If software sets a buffer size to zero, all buffers following that one should be set to
zero as well. Pointers in the packet split receive descriptors to buffers with a zero
size should be set to any address, but not to NULL pointers. Hardware does not
write to this address.
• When configured to packet split and a given packet spans across two or more
packet split descriptors, the first buffer of any descriptor (other than the first one)
is not used.
7.1.5.2
DD (1-Bit, Offset 8.0)
The software device driver might use the DD bit from the Status field to determine
when a descriptor has been used. Therefore, the software device driver must ensure
that the Least Significant B (LSB) of Buffer Address 1 is zero. This should not be an
issue, since the buffers should be page aligned for the packet split feature to be useful.
Note:
Any software device driver that cannot align buffers should not be using this descriptor
format.
124
Inline Functions—82583V GbE Controller
Descriptor Write-Back Format:
19
16
63
48
47
32
31
20
15
0
0
8
Reserved
MRQ
Packet Checksum
VLAN Tag
IP Identification
Length 0
Extended Error
Length 1
Extended Status
Header Status
1
6
Length 3
Length 2
2
4
Reserved
Note:
Light-blue fields are mutually exclusive by RXCSUM.PCSD
MRQ - Same as extended Rx descriptor.
Packet Checksum, IP Identification - Same as extended Rx descriptor.
Extended Status, Extended Errors, VLAN Tag - Same as extended Rx descriptor.
7.1.5.3
Length 0 (16-Bit, Offset 8.32), Length [3:1] (3- x 16-Bit, Offset 16.16)
Upon a packet reception, hardware stores the packet data in one or more of the
indicated buffers. Hardware writes in the Length field of each buffer the number of
bytes that were posted in the corresponding buffer. If no packet data is stored in a
given buffer, hardware writes 0b in the corresponding Length field. Length covers the
data written to receive buffer including CRC bytes (if any).
Software is responsible for checking the Length fields of all buffers for data that
hardware might have written to the corresponding buffers.
Note:
The larger buffer sizes are only offered to provide compatibility with other silicon types.
Jumbo frames are not supported with the 82583V.
7.1.5.4
Header Status (16-Bit, Offset 16.0):
15
14
10
9
0
HDRSP
Reserved
HLEN (Header Length)
HDRSP (bit 15) - Headers were split
Reserved (bits 14:10) - Reserved
Header Length (bits 9:0) - Packet header length
125
82583V GbE Controller—Inline Functions
HDRSP (bit 15): The HDRSP bit (when active) indicates that hardware split the
headers from the packet data for the packet contained in this descriptor. The following
table identifies the packets that are supported by header/data split functionality. In
addition, packets with a data portion smaller than 16 bytes are not guaranteed to be
split. If the device is not configured to provide any offload that requires packet parsing,
the HDRSP bit is set to 0b' even if packet split was enabled. Non-split packets are
stored linearly in the buffers of the receive descriptor.
HLEN (bit 9:0): The HLEN field indicates the header length in byte count that was
analyzed by the 82583V. The 82583V posts the first HLEN bytes of the incoming packet
to buffer zero of the Rx descriptor.
Packet types supported by the packet split: The 82583V provides header split for
the packet types listed in the following table. Other packet types are posted
sequentially in the buffers of the packet split receive buffers.
Packet
Type
Description
MAC, (VLAN/SNAP), payload
Header Split
0x0
No.
0x1
0x2
MAC, (VLAN/SNAP), IPv4, payload
Split header after L3 if fragmented packets.
Split header after L4 if not fragmented,
otherwise treat as packet type 1.
MAC, (VLAN/SNAP), IPv4, TCP/UDP, payload
Split header after L3 if either IPv4 or IPv6
indicates a fragmented packet.
0x3
0x4
MAC (VLAN/SNAP), IPv4, IPv6, payload
Split header after L4 if IPv4 not fragmented
and if IPv6 does not include fragment
extension header, otherwise treat as packet
type 3.
MAC (VLAN/SNAP), IPv4, IPv6,TCP/UDP, payload
0x5
0x6
MAC (VLAN/SNAP), IPv6, payload
Split header after L3 if fragmented packets.
Split header after L4 if IPv6 does not include
fragment extension header, otherwise treat
as packet type 5.
MAC (VLAN/SNAP), IPv6,TCP/UDP, payload
Split header after L5 if not fragmented,
otherwise treat as packet type 1.
0x7
0x8
MAC, (VLAN/SNAP) IPv4, TCP, ISCSI, payload
MAC, (VLAN/SNAP) IPv4, TCP/UDP, NFS, payload
Split header after L5 if not fragmented,
otherwise treat as packet type 1.
Split header after L5 if IPv4 not fragmented
and if IPv6 does not include fragment
extension header, otherwise treat as packet
type 3.
0x9
0xA
MAC (VLAN/SNAP), IPv4, IPv6, TCP, ISCSI, payload
Split header after L5 if IPv4 not fragmented
and if IPv6 does not include fragment
extension header, otherwise treat as packet
type 3.
MAC (VLAN/SNAP), IPv4, IPv6, TCP/UDP,NFS,
payload
Split header after L5 if IPv6 does not include
fragment extension header, otherwise treat
as packet type 5.
0xB
0xC
MAC (VLAN/SNAP), IPv6, TCP, ISCSI, payload
MAC (VLAN/SNAP), IPv6, TCP/UDP, NFS, payload
Split header after L5 if IPv6 does not include
fragment extension header, otherwise treat
as packet type 5.
0xD
0xE
Reserved
Reserved
Note:
A header of a fragmented IPv6 packet is defined until the fragmented extension header.
126
Inline Functions—82583V GbE Controller
Note:
If the device is not configured to provide any offload that requires packet parsing, the
packet type field is set to 0b regardless of the actual packet type. When packet split is
enabled, the packet type field is always valid.
7.1.6
Receive Descriptor Fetching
The fetching algorithm attempts to make the best use of PCIe bandwidth by fetching a
cache-line (or more) descriptor with each burst. The following paragraphs briefly
describe the descriptor fetch algorithm and the software control provided.
When the on-chip buffer is empty, a fetch happens as soon as any descriptors are made
available (host writes to the tail pointer). When the on-chip buffer is nearly empty
(RXDCTL.PTHRESH), a prefetch is performed each time enough valid descriptors
(RXDCTL.HTHRESH) are available in host memory and no other PCIe activity of greater
priority is pending (descriptor fetches and write backs or packet data transfers).
When the number of descriptors in host memory is greater than the available on-chip
descriptor storage, the chip might elect to perform a fetch that is not a multiple of
cache line size. The hardware performs this non-aligned fetch if doing so results in the
next descriptor fetch being aligned on a cache line boundary. This enables the
descriptor fetch mechanism to be most efficient in the cases where it has fallen behind
software.
Note:
The 82583V NEVER fetches descriptors beyond the descriptor tail pointer.
7.1.7
Receive Descriptor Write Back
Processors have cache line sizes that are larger than the receive descriptor size (16
bytes). Consequently, writing back descriptor information for each received packet can
cause expensive partial cache line updates. Two mechanisms minimize the occurrence
of partial line write backs:
• Receive descriptor packing
• Null descriptor padding
The following sections explain these mechanisms.
7.1.7.1
Receive Descriptor Packing
To maximize memory efficiency, receive descriptors are packed together and written as
a cache line whenever possible. Descriptors accumulate and are opportunistically
written out in cache line-oriented chunks. Used descriptors are also explicitly written
out under the following scenarios:
• RXDCTL.WTHRESH descriptors have been used (the specified maximum threshold
of unwritten used descriptors has been reached)
• The last descriptors of the allocated descriptor ring have been used (to enable
hardware to re-align to the descriptor ring start)
• A receive timer expires (RADV or RDTR)
• Explicit software flush (RDTR.FPD)
127
82583V GbE Controller—Inline Functions
When the number of descriptors specified by RXDCTL.WTHRESH have been used, they
are written back, regardless of cache line alignment. It is therefore recommended that
WTHRESH be a multiple of cache line size. When a receive timer (RADV or RDTR)
expires, all used descriptors are forced to be written back prior to initiating the
interrupt, for consistency. Software might explicitly flush accumulated descriptors by
writing the RDTR register with the high order bit (FPD) set.
7.1.7.2
Null Descriptor Padding
Hardware stores no data in descriptors with a null data address. Software can make
use of this property to cause the first condition under receive descriptor packing to
occur early. Hardware writes back null descriptors with the DD bit set in the status byte
and all other bits unchanged.
Note:
Null descriptor padding is not supported for packet split descriptors.
7.1.8
Receive Descriptor Queue Structure
Figure 30 shows the structure of the receive descriptor ring. Hardware maintains a
circular queue of descriptors and writes back used descriptors just prior to advancing
the head pointer. Head and tail pointers wrap back to base when size descriptors have
been processed.
Circular Buffer Queues
Base
Head
Receive
Queue
Tail
Base + Size
Figure 30.
Receive Descriptor Ring Structure
128
Inline Functions—82583V GbE Controller
Software adds receive descriptors by advancing the tail pointer to refer to the address
of the entry just beyond the last valid descriptor. This is accomplished by writing the
descriptor tail register with the offset of the entry beyond the last valid descriptor. The
hardware adjusts its internal tail pointer accordingly. As packets arrive, they are stored
in memory and the head pointer is incremented by hardware. When the head pointer is
equal to the tail pointer, the queue is empty. Hardware stops storing packets in system
memory until software advances the tail pointer, making more receive buffers available.
The receive descriptor head and tail pointers reference 16-byte blocks of memory.
Shaded boxes in the figure represent descriptors that have stored incoming packets but
have not yet been recognized by software. Software can determine if a receive buffer is
valid by reading descriptors in memory rather than by I/O reads. Any descriptor with a
non-zero status byte has been processed by the hardware, and is ready to be handled
by the software.
Note:
Note:
When configured to work as a packet split feature, the descriptor tail needs to be
increment by software by two for every descriptor ready in memory (as the packet split
descriptors are 32 bytes while regular descriptors are 16 bytes).
The head pointer points to the next descriptor that will be written back. At the
completion of the descriptor write-back operation, this pointer is incremented by the
number of descriptors written back. Hardware OWNS all descriptors between [head...
tail]. Any descriptor not in this range is owned by software.
The receive descriptor ring is described by the following registers:
• Receive Descriptor Base Address registers (RDBA0)
— This register indicates the start of the descriptor ring buffer; this 64-bit address
is aligned on a 16-byte boundary and is stored in two consecutive 32-bit
registers. Hardware ignores the lower 4 bits.
• Receive Descriptor Length registers (RDLEN0)
— This register determines the number of bytes allocated to the circular buffer.
This value must be a multiple of 128 (the maximum cache line size). Since each
descriptor is 16 bytes in length, the total number of receive descriptors is
always a multiple of 8.
• Receive Descriptor Head register (RDH0)
— This register holds a value that is an offset from the base, and indicates the in-
progress descriptor. There can be up to 64 KB descriptors in the circular buffer.
Hardware maintains a shadow copy that includes those descriptors completed
but not yet stored in memory.
• Receive Descriptor Tail register (RDT0)
— This register holds a value that is an offset from the base, and identifies the
location beyond the last descriptor hardware can process. This is the location
where software writes the first new descriptor.
If software statically allocates buffers, and uses memory read to check for completed
descriptors, it simply has to zero the status byte in the descriptor to make it ready for
reuse by hardware. This is not a hardware requirement (moving the hardware tail
pointer is), but is necessary for performing an in-memory scan.
129
82583V GbE Controller—Inline Functions
7.1.9
Receive Interrupts
The following indicates the presence of new packets:
• Receive Timer (ICR.RXT0) due to packet delay timer (RDTR)
A predetermined amount of time has elapsed since the last packet was received and
transferred to host memory. Every time a new packet is received and transferred to the
host memory, the timer is re-initialized to the predetermined value. The timer then
counts down and triggers an interrupt if no new packet is received and transferred to
host memory completely before the timer expires. Software can set the timer value to
zero if it needs to be notified immediately (no interval delay) whenever a new packet
has been stored in memory.
Writing the absolute timer with its high order bit set to 1b forces an explicit flush of any
partial cache lines worth of consumed descriptors. Hardware writes all used descriptors
to memory and updates the globally visible value of the RXDH head pointer.
This timer is re-initialized when an interrupt is generated and restarts when a new
packet is observed. It stays disabled until a new packet is received and transferred to
the host memory. The packet delay timer is also re-initialized when an interrupt occurs
due to an absolute timer expiration or small packet-detection interrupt.
• Receive Timer (ICR.RXT0) due to absolute timer (RADV)
A predetermined amount of time has elapsed since the first packet received after the
hardware timer was written (specifically, after the last packet data byte was written to
memory).
This timer is re-initialized when an interrupt is generated and restarts when a new
packet is observed. It stays disabled until a new packet is received and transferred to
the host memory. The absolute delay timer is also re-initialized when an interrupt
occurs due to a packet timer expiration or small packet-detection interrupt.
The absolute timer and the packet delay timer can be used together. The following
table lists the conditions when the absolute timer and the packet delay timer are
initialized, disabled and when they start counting. The timer is always disabled if the
value of the RDTR = 0b.
Interrupt
Timers
When Starts
Counting
When Re-initialized
When Disabled
Timer inactive and
receive packet
transferred to host
memory.
On expiration
Due to other receive
interrupt.
Absolute delay
timer
At start
Timer inactive and
receive packet
transferred to host
memory.
At start
New packet received and
transferred to host memory
On expiration
Due to other receive
interrupt.
Packet delay
timer
Figure 31 further clarifies the packet timer operation.
130
Inline Functions—82583V GbE Controller
packet received & xferred
Initial State
DISABLED
to host mem
Action: Re-initialized
other receive interrupts
packet received & xfer to
host memory
RUNNING
Action - Re-initialize
Timer expires
INT
GENERATED
Figure 31.
Packet Delay Timer Operation (With State Diagram)
Figure 32 shows how the packet timer and absolute timer can be used together:
131
82583V GbE Controller—Inline Functions
Case A: Using only an absolute timer
A bsolute Timer Value
Interrupt generated due to PKT #1
PKT #1
PKT #2
PKT #3
PKT #4
Case B: Using an absolute time in conjunction with the Packet timer
A bsolute Timer Value
A bsolute Timer Value
PKT #1
PKT #2
PKT #3
PKT #4
PKT #5
PKT #6
...
...
...
1) Packet timer expires
2) Interrupt generated
3) Absolute timer reset
Interrupt generalted (due to PKT #4)
as absolute timer expires.
Packet delay timer disabled untill
next packet is received and
transferred to host memory.
Case C: Packet timer expiring while a packet is transferred to host memory.
Illustrates that packet timer is re-started only after a packet is transferred to host memory.
A bsolute Timer Value
A bsolute Timer Value
PKT #1
PKT #2
PKT #3
PKT #4
PKT #5
PKT #6
...
...
...
1) Packet timer expires
2) Interrupt generated
3) Absolute timer reset
Interrupt generalted (due to PKT #4)
as absolute timer expires.
Packet delay timer disabled untill
next packet is received and
transferred to host memory.
Figure 32.
Packet and Absolute Timers
• Small Receive Packet Detect (ICR.SRPD)
— A receive interrupt is asserted when small-packet detection is enabled (RSRPD
is set with a non-zero value) and a packet of (size < RSRPD.SIZE) has been
transferred into the host memory. When comparing the size the headers and
CRC are included (if CRC stripping is not enabled). CRC and VLAN headers are
not included if they have been stripped. A receive timer interrupt cause
(ICR.RXT0) will also be noted when the small packet-detect interrupt occurs.
• Receive ACK frame interrupt is asserted when a frame is detected to be an ACK
frame. Detection of ACK frames are masked through the IMS register. When a
frame is detected as an ACK frame an interrupt is asserted after the
RAID.ACK_DELAY timer had expired and the ACK frames interrupts were not
masked in the IMS register.
Note:
The ACK frame detect feature is only active when configured to packet split
(RCTL.DTYP=01b) or the extended status feature is enabled (RFCTL.EXSTEN is set).
132
Inline Functions—82583V GbE Controller
Receive interrupts can also be generated for the following events:
• Receive Descriptor Minimum Threshold (ICR.RXDMT)
— The minimum descriptor threshold helps avoid descriptor under-run by
generating an interrupt when the number of free descriptors becomes equal to
the minimum. It is measured as a fraction of the receive descriptor ring size.
This interrupt would stop and re-initialize the entire active delayed receives
interrupt timers until a new packet is observed.
• Receiver FIFO Overrun (ICR.RXO)
— FIFO overrun occurs when hardware attempts to write a byte to a full FIFO. An
overrun could indicate that software has not updated the tail pointer to provide
enough descriptors/buffers, or that the PCIe bus is too slow draining the
receive FIFO. Incoming packets that overrun the FIFO are dropped and do not
affect future packet reception. This interrupt would stop and re-initialize the
entire active delayed receives interrupts.
7.1.10
Receive Packet Checksum Offloading
The 82583V supports the offloading of three receive checksum calculations: the packet
checksum, the IPv4 header checksum, and the TCP/UDP checksum.
The packet checksum is the one's complement over the receive packet, starting from
the byte indicated by RXCSUM.PCSS (zero corresponds to the first byte of the packet),
after stripping. For packets with VLAN header the packet checksum includes the header
if VLAN striping is not enabled by the CTRL.VME. If VLAN header strip is enabled, the
packet checksum and the starting offset of the packet checksum exclude the VLAN
header due to masking of VLAN header. For example, for an Ethernet II frame
encapsulated as an 802.3ac VLAN packet and CTRL.VME is set and with RXCSUM.PCSS
set to 14, the packet checksum would include the entire encapsulated frame, excluding
the 14-byte Ethernet header (DA, SA, Type/Length) and the 4-byte q-tag. The packet
checksum does not include the Ethernet CRC if the RCTL.SECRC bit is set.
Software must make the required offsetting computation (to back out the bytes that
should not have been included and to include the pseudo-header) prior to comparing
the packet checksum against the TCP checksum stored in the packet.
For supported packet/frame types, the entire checksum calculation can be offloaded to
the 82583V. If RXCSUM.IPOFLD is set to 1b, the 82583V calculates the IPv4 checksum
and indicates a pass/fail indication to software via the IPv4 Checksum Error bit
(RDESC.IPE) in the Error field of the receive descriptor. Similarly, if RXCSUM.TUOFLD is
set to 1b, the 82583V calculates the TCP or UDP checksum and indicates a pass/fail
condition to software via the TCP/UDP Checksum Error bit (RDESC.TCPE). These error
bits are valid when the respective status bits indicate the checksum was calculated for
the packet (RDESC.IPCS and RDESC.TCPCS respectively). Similarly, if RFCTL.Ipv6_DIS
and RFCTL.IP6Xsum_DIS are cleared to 0b and RXCSUM.TUOFLD is set to 1b, the
82583V calculates the TCP or UDP checksum for IPv6 packets. It then indicates a pass/
fail condition in the TCP/UDP Checksum Error bit (RDESC.TCPE).
If neither RXCSUM.IPOFLD nor RXCSUM.TUOFLD are set, the Checksum Error bits (IPE
and TCPE) are 0b for all packets.
133
82583V GbE Controller—Inline Functions
Supported frame types:
• Ethernet II
• Ethernet SNAP
Table 38.
Supported Receive Checksum Capabilities
HW IP Checksum
Calculation
HW TCP/UDP Checksum
Calculation
Packet Type
IPv4 packets
IPv6 packets
Yes
Yes
Yes
No (n/a)
IPv6 packet with next header options:
Hop-by-Hop options
Destinations options
Routing (with len 0)
Routing (with len >0)
Fragment
No (n/a)
No (n/a)
No (n/a)
No (n/a)
No (n/a)
No (n/a)
Yes
Yes
Yes
No
No
No
Home option
IPv4 tunnels:
IPv4 packet in an IPv4 tunnel
IPv6 packet in an IPv4 tunnel
No
Yes (IPv4)
No
Yes
1
IPv6 tunnels:
IPv4 packet in an IPv6 tunnel
IPv6 packet in an IPv6 tunnel
No
No
No
No
Packet is an IPv4 fragment
Packet has 802.3ac tag
Yes
Yes
No
Yes
IPv4 Packet has IP options
(IP header is longer than 20 bytes)
Yes
Yes
Yes
Yes
Yes
No
Packet has TCP or UDP options
IP header’s protocol field contains a
protocol # other than TCP or UDP.
1. The IPv6 header portion can include supported extension headers as described in the IPv6 filter section.
The previous table lists the general details about what packets are processed. In more
detail, the packets are passed through a series of filters to determine if a receive
checksum is calculated:
7.1.10.1
7.1.10.2
MAC Address Filter
This filter checks the MAC destination address to be sure it is valid (such as, IA match,
broadcast, multicast, etc.). The receive configuration settings determine which MAC
addresses are accepted. See the various receive control configuration registers such as
RCTL (RTCL.UPE, RCTL.MPE, RCTL.BAM), MTA, RAL, and RAH.
SNAP/VLAN Filter
This filter checks the next headers looking for an IP header. It is capable of decoding
Ethernet II, Ethernet SNAP, and IEEE 802.3ac headers. It skips past any of these
intermediate headers and looks for the IP header. The receive configuration settings
determine which next headers are accepted. See the various receive control
configuration registers such as RCTL (RCTL.VFE), VET, and VFTA.
134
Inline Functions—82583V GbE Controller
7.1.10.3
IPv4 Filter
This filter checks for valid IPv4 headers. The version field is checked for a correct value
(4).
IPv4 headers are accepted if they are any size greater than or equal to 5 (Dwords). If
the IPv4 header is properly decoded, the IP checksum is checked for validity. The
RXCSUM.IPOFL bit must be set for this filter to pass.
7.1.10.4
IPv6 Filter
This filter checks for valid IPv6 headers, which are a fixed size and have no checksum.
The IPv6 extension headers accepted are: hop-by-hop, destination options, and
routing. The maximum size next header accepted is 16 Dwords (64 bytes).
All of the IPv6 extension headers supported by the 82583V have the same header
structure:
Byte0
Byte1
Byte2
Byte3
NEXT HEADER
HDR EXT LEN
NEXT HEADER is a value that identifies the header type. The supported IPv6 next
headers values are:
• Hop-by-hop = 0x00
• Destination options = 0x3C
• Routing = 0x2B
HDR EXT LEN is the 8-byte count of the header length, not including the first 8 bytes.
For example, a value of three means that the total header size including the NEXT
HEADER and HDR EXT LEN fields is 32 bytes (8 + 3*8).
The RFCTL.Ipv6_DIS bit must be cleared for this filter to pass.
7.1.10.5
UDP/TCP Filter
This filter checks for a valid UDP or TCP header. The prototype next header values are
0x11 and 0x06, respectively. The RXCSUM.TUOFL bit must be set for this filter to pass.
7.2
Packet Transmission
7.2.1
Transmit Functionality
The 82583V transmit flow is a descriptor-based transmit where the hardware gets the
per-packet details for the transmit tasks through descriptors created by software.
This section outlines the transmit structures and process along with features and
offloads supported by the 82583V.
135
82583V GbE Controller—Inline Functions
7.2.2
Transmission Flow Using Simplified Legacy Descriptors
Software defines a descriptor ring and configures the 82583V's transmit queue with the address
location, length, head, and tail pointers of the ring. See section 7.2.4 for more details on the
descriptor ring structure.
1
2
Software prepares the packet headers and data for the transmit within one or more data buffers.
Software prepares Tx descriptors according to the number of data buffers that are used. Each
descriptor points to a different data buffer and holds the required hardware processing. See
section 7.2.9 for more details on the descriptor format. The software places the descriptors in the
correct location in the Tx descriptor ring.
3
Software updates the transmit descriptor tail pointer (TDT) to indicate the hardware that Tx
descriptors are ready.
4
5
6
Hardware senses a change of the TDT and initiates a PCIe request to fetch the descriptors from host
memory.
The descriptors’ content is received in a PCIe read completion and is written to the appropriate
location in the descriptor queue.
According to the descriptors content the corresponding memory data buffers are then fetched from
the host to the hardware on-chip transmit FIFO.
While the packet is passing through the DMA and MAC units, relevant off load functions are
incorporated according to the commands in the descriptors.
7
10
11
After the entire packet is fetched by the hardware it is transmitted to the Ethernet link.
After a DMA of each buffer is complete, if the RS bit in the command byte is set, the DMA updates the
Status field of the appropriate descriptor and writes back the descriptor to the descriptor ring in host
memory.
The hardware moves the transmit descriptor head pointer (TDH) in the direction of the tail to point to
the next descriptor in the ring.
12
13
After the entire packet is fetched by the hardware an interrupt might be generated by the hardware to
notify the software device driver that it can release the relevant buffers to the operating system.
7.2.3
Transmission Process Flow Using Extended Descriptors
The 82583V supports extended Tx descriptors that provide more offload capabilities.
The extended offload capabilities are indicated to the hardware by two types of
descriptors: context descriptors and data descriptors. The context descriptors define a
set of offload capabilities applicable for multiple packets while the data descriptors
define the data buffers and specific off load capabilities per packet.
The software/hardware flow while using the extended descriptors is as follows:
• Software prepares the context descriptor that defines the offload capabilities for
the incoming packets.
• Software prepares the data packets in host memory within one or more data
buffers and their descriptors.
• All steps are the same as the legacy Tx descriptors as previously described
(starting at step number 4) while the data buffers belong to a single packet.
The software/hardware flow for TCP segmentation using the extended descriptors is as
follows:
• Software prepares the context descriptor that defines the upcoming TCP
segmentation, In this case, the data buffers belong to multiple packets.
• Software places a prototype header in host memory and indicates it to the
hardware by a data descriptor.
136
Inline Functions—82583V GbE Controller
• Software places the rest of the data to be transmitted in the host memory indicated
to the hardware by additional data descriptors.
• Hardware splits the data into multiple packets according to the Maximum Segment
Size (MSS) defined in the context descriptor. Hardware uses the prototype header
for each packet while it auto-updates some of the fields in the IP and TCP headers.
See more details in section 7.3.6.2.
• For each packet, the proceeding steps are the same as the legacy Tx descriptors as
previously described (starting at step number 4).
7.2.4
Transmit Descriptor Ring Structure
The transmit descriptor ring is described by the following registers:
• Transmit Descriptor Base Address register (TDBA)
— This register indicates the start address of the descriptor ring buffer in the host
memory; this 64-bit address is aligned on a 16-byte boundary and is stored in
two consecutive 32-bit registers. Hardware ignores the lower four bits.
• Transmit Descriptor Length register (TDLEN)
— This register determines the number of bytes allocated to the circular ring. This
value must be aligned to 128 bytes.
• Transmit Descriptor Head register (TDH)
— This register holds an index value that indicates the in-progress descriptor.
There can be up to 64 KB descriptors in the circular buffer. Reading this register
returns the value of head corresponding to descriptors already loaded in the
transmit FIFO.
• Transmit Descriptor Tail register (TDT)
— This register holds a value, which is an offset from the base (TDBA), and
indicates the location beyond the last descriptor hardware can process. This is
the location where software writes the next new descriptor.
Base +
TDLEN
Base
TDBA
Base+1
Head
TDH
Tail
TDT
Figure 33.
Transmit Descriptor Ring Structure
137
82583V GbE Controller—Inline Functions
Descriptors between the head and the tail pointers are descriptors that have been
prepared by software and are owned by hardware.
7.2.4.1
Transmit Descriptor Fetching
The descriptor processing strategy for transmit descriptors is essentially the same as
for receive descriptors.
When the on-chip descriptor queue is empty, a fetch occurs as soon as any descriptors
are made available (host writes to the tail pointer). Hardware might elect to perform a
fetch which is not a multiple of cache line size. The hardware performs this non-aligned
fetch if doing so results in the next descriptor fetch being aligned on a cache line
boundary. This enables the descriptor fetch mechanism to be most efficient in the cases
where it has fallen behind software.
After the initial fetch of descriptors, as the on-chip buffer empties, the hardware can
decide to pre-fetch more transmit descriptors if the number of on-chip descriptors drop
below TXDCTL.PTHRESH and enough valid descriptors TXDCT is performed.
Note:
The 82583V NEVER fetches descriptors beyond the descriptor tail pointer.
7.2.4.2
Transmit Descriptor Write Back
The descriptor write-back policy for transmit descriptors is similar to that for receive
descriptors with a few additional factors.
There are three factors: the Report Status (RS) bit in the transmit descriptor, the write
back threshold (TXDCTL.WTHRESH) and the Interrupt Delay Enable (IDE) bit in the
transmit descriptor.
Descriptors are written back in one of three cases:
• TXDCTL.WTHRESH = zero, IDE = zero and a descriptor with RS set to 1b is ready
to be written back, for this condition write backs are immediate. The device writes
back only the status byte of the descriptor (TDESCR.STA) and all other bytes of the
descriptor are left unchanged.
• IDE = 1b and the Transmit Interrupt Delay (TIDV) register timer expires, this timer
is used to force a timely write back of descriptors. Timer expiration flushes any
accumulated descriptors and sets an interrupt event.
• TXDCTL.WTHRESH > zero and the write back of the full descriptors are performed
only when TXDCTL.WTHRESH number of descriptors are ready for a write back.
138
Inline Functions—82583V GbE Controller
7.2.4.3
Determining Completed Frames as Done
Software can determine if a packet has been sent by the following method:
• Setting the RS bit in the transmit descriptor command field and checking the DD bit
of the relevant descriptors in host memory.
The process of checking for completed descriptors consists of the following:
• The software device driver scans the host memory for the value of the DD status
bit. When the DD bit =1b, indicates a completed packet, and also indicates that all
packets preceding that packet have been put in the output FIFO.
7.2.5
Overview of On-Chip Transmit Modes
Transmit mode is used to refer to a set of configurations that support some of the
transmit path offloads. These modes are updated and controlled with the transmit
descriptors.
There are three types of transmit modes:
• Legacy mode
• Extended mode
• Segmentation mode
The first mode (legacy) is an implied mode as it is not explicitly specified with a context
descriptor. This mode is constructed by the device from the first and last descriptors of
a legacy transmit and from some internal constants. The legacy mode enables insertion
of one checksum.
The other two modes are indicated explicitly by a transmit context descriptor. The
extended mode is used to control the checksum offloading feature for packet
transmission. The segmentation mode is used to control the packet segmentation
capabilities of the device. The TSE bit, in the context descriptor, selects which mode is
updated, that is, extended mode or segmentation mode. The extended and
segmentation modes enable insertion of two checksums. In addition, the segmentation
mode adds information specific to the segmentation capability.
139
82583V GbE Controller—Inline Functions
The device automatically selects the appropriate mode to use based on the current
packet transmission: legacy, extended, or segmentation.
Note:
While the architecture supports arbitrary ordering rules for the various descriptors,
there are restrictions including:
— Context descriptors should not occur in the middle of a packet or of a
segmentation.
— Data descriptors of different packet types (legacy, extended, or segmentation)
should not be intermingled except at the packet (or segmentation) level.
There are dedicated resources on-chip for both the extended and segmentation modes.
These modes remain constant until they are modified by another context descriptor.
This means that a set of configurations relevant to one mode can (and will) be used for
multiple packets unless a new mode is loaded prior to sending a new packet.
7.2.6
Pipelined Tx Data Read Requests
Transmit data request pipelining is the process by which a request for transmit data is
sent to the host memory before the read DMA request of the previously requested data
completes. Transmit pipeline requests is enabled using the MULR bit in the Transmit
Control (TCTL) register, Its initial value is loaded from the NVM.
The 82583V supports four pipelined requests from the Tx data DMA. In general, the
four requests can belong to the same packet or to consecutive packets. However, the
following restrictions apply:
• All requests for a packet are issued before a request is issued for a following
packet.
• If a request (for the following packet) requires context change, the request for the
following packet is not issued until the previous request is completed (such as, no
pipeline across contexts).
The PCIe specification does not ensure that completions for separate requests return in
order. The 82583V can handle completions that arrive in any order.
The 82583V incorporates a 2 KB buffer to support re-ordering of completions for the
four requests. Each request/completion can be up to 512 bytes long. The maximum
size of a read request is defined as follows:
• When the MULR bit is cleared, maximum request size in bytes is the min{2K,
Max_Read_Request_Size}
• When the MULR bit is set, maximum request size in bytes is the min{512,
Max_Read_Request_Size}
Note:
In addition to the four pipeline requests from the Tx data DMA, the 82583V can issue a
single read request from each of the 2 Tx descriptor and 2 Rx descriptor DMA engines.
The requests from the three sources (Tx data, Tx descriptor and Rx descriptor) are
independently issued. Each descriptor read request can fetch up to 16 descriptors
(equal to 256 bytes of data).
140
Inline Functions—82583V GbE Controller
7.2.7
Transmit Interrupts
Hardware supplies the transmit interrupts described below. These interrupts are
initiated via the following conditions:
• Transmit Descriptor Ring Empty (ICR.TXQE) - All descriptors have been processed.
The head pointer is equal to the tail pointer.
• Any write backs are performed; either with the RS bit set or when accumulated
descriptors are written back when TXDCTL.WTHRESH descriptors have been
completed and accumulated; Transmit Descriptor Write Back (ICR.TXDW).
• Transmit Delayed Interrupt (ICR.TXDW) - in conjunction with Interrupt Delay
Enable (IDE), the TXDW indication is delayed per the TIDV and/or TADV registers.
The interrupt is set when one of the transmit interrupt countdown timers expire. A
transmit delayed interrupt is scheduled for a transmit descriptor with its RS bit set
and the IDE bit set. When a transmit delayed interrupt occurs, the TXDW interrupt
bit is set (just as when a transmit descriptor write-back interrupt occurs). This
interrupt can be masked in the same manner as the TXDW interrupt. This interrupt
is used frequently by software that performs dynamic transmit chaining by adding
packets one at a time to the transmit chain.
Note:
The transmit delay interrupt is indicated with the same interrupt bit as the transmit
write-back interrupt, TXDW. The transmit delay interrupt is only delayed in time as
previously discussed.
• Transmit Descriptor Ring Low Threshold Hit (ICR.TXD_LOW) - Set when the total
number of transmit descriptors available hits the low threshold specified in the
TXDCTL.LWTHRESH field in the Transmit Descriptor Control register. For the
purposes of this interrupt, number of transmit descriptors available is the
difference between the transmit descriptor tail and transmit descriptor head values,
minus the number of transmit descriptors that have been pre-fetched. Up to eight
descriptors can be pre-fetched.
7.2.7.1
Delayed Transmit Interrupts
This mechanism allows software the flexibility of delaying transmit interrupts in order
to allow more time for new descriptors to be written to the memory ring and potentially
prevent an interrupt when the device's head pointer catches the tail pointer.
This feature is desirable, because a software device driver usually has no knowledge of
when it is going to be asked to send another frame. For performance reasons, it is best
to generate only one transmit interrupt after a burst of packets have been sent.
7.2.8
Transmit Data Storage
Data is stored in buffers pointed to by the descriptors. Alignment of data is on an
arbitrary byte boundary with the maximum size per descriptor limited only to the
maximum allowed length size. A packet typically consists of two (or more) descriptors,
one (or more) for the header and one (or more) for the actual data. Some software
implementations copy the header and packet data into one buffer and use only one
descriptor per transmitted packet.
141
82583V GbE Controller—Inline Functions
7.2.9
Transmit Descriptor Formats
The original descriptor is referred to as the legacy descriptor and is described in
section 7.2.9.1. The two new descriptor types are collectively referred to as extended
descriptors. One of the new descriptor types is quite similar to the legacy descriptor in
that it points to a block of packet data. This descriptor type is called the extended data
descriptor. The other new descriptor type is fundamentally different as it does not point
to packet data. This descriptor type is called the context descriptor. It only contains
control information, which is loaded into registers of the 82583V, and affects the
processing of future packets. The following paragraphs describe the three descriptor
formats.
The new descriptor types are specified by setting the TDESC.DEXT bit to 1b. If this bit
is set, the TDESC.DTYP field is examined to determine the descriptor type (extended
data or context). Figure 35 shows the context descriptor generic layout. Figure 37
shows the data descriptor generic layout.
7.2.9.1
Legacy Transmit Descriptor Format
63
48 47
40 39
36 35 32 31
24 23
16 15
0
0
8
Buffer Address [63:0]
VLAN
CSS ExtCMD STA CMD
CSO
Length
Checksum
Start
Checksum
Offset
VLAN
12
Command
4
Status
Length
015
13
15
3
2
1
0
7
6
5
3
2
1
0
11
0
8
0
3
1
0
8
PRI
CFI
VLAN
CSS
ExtCMD
CSO
LENGTH
TS
Rsv Res Res DD
VLE DEXT RSV RS IC IFCS EOP
IDE
Figure 34.
7.2.9.1.1
Legacy Transmit Descriptor Format
The legacy Tx descriptor is defined by setting the DEXT bit in the command field to 0b.
The legacy Tx descriptor format is shown in Figure 34.
Buffer Address
The buffer address (TDESC.Buffer Address) specifies the location (address) in main
memory of the data to be fetched.
142
Inline Functions—82583V GbE Controller
7.2.9.1.2
Length
Length (TDESC.LENGTH) specifies the length in bytes to be fetched from the buffer
address. The maximum length associated with any single legacy descriptor is 16288
bytes.
Note:
The maximum allowable packet size for transmits might change based on the value
configured for the transmit FIFO size written to the Packet Buffer Allocation (PBA)
register. For any individual packet, the sum of the individual descriptors' lengths must
be at least 80 bytes less than the allocated size of the transmit FIFO.
7.2.9.1.3
Checksum Offset and Checksum Start - CSO and CSS
The checksum start (TDESC.CSS) field indicates where to begin computing the
checksum. CSS must be set in the first descriptor of a packet. The checksum offset
(TDESC.CSO) field indicates where to insert the TCP checksum, relative to the start of
the packet. Both CSO and CSS are in units of bytes while they must be within the range
of data provided to the device in the descriptor. This means, for short packets that are
padded by software, CSS and CSO must be in the range of the unpadded data length,
not the eventual padded length (64 bytes).
Note:
Note:
CSO must be set in the last descriptor of the packet. Only when EOP is set does the
hardware interpret Insert Checksum (IC), and CSO bits.
In the case of 802.1Q header, the offset values depend on the VLAN insertion enable bit
- CTRL.VME and the VLE bit. When the CTRL.VME and the VLE bit are not set (VLAN
tagging included in the packet buffers), the offset values should include the VLAN
tagging. When these bits are set (VLAN tagging is taken from the packet descriptor),
the offset values should exclude the VLAN tagging.
Although the 82583V can be programmed to calculate and insert TCP checksum using
the legacy descriptor format as previously described, it is recommended that software
use the newer context descriptor format. This newer descriptor format enables
hardware to calculate both the IP and TCP checksums for outgoing packets. See
section 7.2.6 for more information about how the new descriptor format can be used to
accomplish this task.
Note:
Note:
UDP checksum calculation is not supported by the legacy descriptor.
As the CSO field is eight bits wide, it limits the location of the checksum to 255 bytes
from the beginning of the packet.
Software must compute an offsetting entry and store it in the position where the
hardware computed checksum is to be inserted. This offset is needed to back out the
bytes of the header that should not be included in the TCP checksum.
7.2.9.1.4
Table 39.
Command Byte - CMD
The CMD byte stores the applicable command and has the fields shown in Table 39.
Command Byte Fields
7
6
5
4
3
2
1
0
IDE
VLE
DEXT
RSV
RS
IC
IFCS
EOP
143
82583V GbE Controller—Inline Functions
IDE (bit 7) - Interrupt Delay Enable
VLE (bit 6) - VLAN Packet Enable
DEXT (bit 5) - Descriptor extension (0b for legacy mode)
RSV (bit 4) - Reserved
RS (bit 3) - Report status
IC (bit 2) - Insert checksum
IFCS (bit 1) - Insert FCS (CRC)
EOP (bit 0) - End of packet
IDE activates a transmit interrupt delay timer. Hardware loads a countdown register
when it writes back a transmit descriptor that has RS and IDE set. The value loaded
comes from the IDV field of the Interrupt Delay (TIDV) register. When the count
reaches zero, a transmit interrupt occurs if transmit descriptor write-back interrupts
(TXDW) are enabled. Hardware always loads the transmit interrupt counter whenever it
processes a descriptor with IDE set even if it is already counting down due to a
previous descriptor. If hardware encounters a descriptor that has RS set, but not IDE, it
generates an interrupt immediately after writing back the descriptor and clears the
interrupt delay timer. Setting the IDE bit has no meaning without setting the RS bit.
Note:
Although the transmit interrupt might be delayed, the descriptor write-back requested
by setting the RS bit is performed without delay unless descriptor write-back bursting is
enabled.
VLE indicates that the packet is a VLAN packet (for example, that the hardware should
add the VLAN Ether type and an 802.1q VLAN tag to the packet).
Note:
If the VLE bit is set, the CTRL.VME bit should also be set to enable VLAN tag insertion.
Table 40.
VLAN Tag Insertion Decision Table when VLAN Mode Enabled (CTRL.VME=1b)
VLE
Action
Send generic Ethernet packet. IFCS controls insertion of FCS in normal Ethernet
packets.
0
1
Send 802.1Q packet; the Ethernet Type field comes from the VET register and the
VLAN data comes from the special field of the TX descriptor; hardware appends the
FCS/CRC - command should reflect by setting IFCS to 1b.
The DEXT bit identifies this descriptor as either a legacy or an extended descriptor type
and must be set to 0b to indicate legacy descriptor.
When the RS bit is set, hardware writes back the DD bit once the DMA fetch completes.
Note:
Note:
Descriptors with the null address (0), or zero length, transfer no data. If they have the
RS bit in the command byte set, then the DD field in the status word is written when
hardware processes them. Hardware only sets the DD bit for descriptors with RS set.
The software can set the RS bit in each descriptor or, more likely, in specific descriptors
such as the last descriptor of each packet.
144
Inline Functions—82583V GbE Controller
When IC is set, hardware inserts a checksum value calculated from the CSS bit value to
the CSE bit value, or to the end of packet. The checksum value is inserted in the header
at the CSO bit value location. One or many descriptors can be used to form a packet.
Checksum calculations are for the entire packet starting at the byte indicated by the
CSS field. A value of zero for CSS corresponds to the first byte in the packet. CSS must
be set in the first descriptor for a packet. In addition, IC is ignored if CSO or CSS are
out of range. This occurs if (CSS ≥ Length) or (CSO ≥ Length – 1).
When IFCS is set, hardware appends the MAC FCS at the end of the packet. When
cleared, software should calculate the FCS for proper CRC check. The software must set
IFCS in the following instances:
• Transmission of short packets while padding is enabled by the TCTL.PSP bit
• Checksum offload is enabled by the IC bit in the TDESC.CMD
• VLAN header insertion enabled by the VLE bit in the TDESC.CMD
• Large send or TCP/IP checksum offload using context descriptor
EOP stands for end-of-packet and when set, indicates the last descriptor making up the
packet.
Note:
VLE, IFCS, CSO, and IC are qualified by EOP. In other words, hardware interprets these
bits ONLY when the EOP bit is set.
7.2.9.1.5
Extended Command - ExtCMD
3
2
1
0
Rsv
TS
RSV (bit 3:1) - Reserved
TS (bit 0) - Time stamp
The TS bit indicates to the 82583V to put a time stamp on the packet designated by the
descriptor.
7.2.9.1.6
Status - STA
3
2
1
0
Rsv
DD
RSV (bit 3:1) - Reserved
DD (bit 0) - Descriptor done status
DD indicates that the descriptor is done and is written back after the descriptor has
been processed (assuming the RS bit was set). The DD bit can be used as an indicator
to the software that all descriptors, in the memory descriptor ring, up to and including
the descriptor with the DD bit set are again available to the software.
145
82583V GbE Controller—Inline Functions
7.2.9.1.7
VLAN Field
The VLAN field is used to provide the 802.1Q/802.1ac tagging information. The VLAN
field is ignored if the VLE bit is 0b or if the EOP bit is 0b.
15
13
12
11
VLAN tag
0
PRI
CFI
7.2.9.2
Context Transmit Descriptor Format
TUCSE, TUCSS, TUCSO are
TCP/UDP Checksum Controls
IPCSE, IPCSS, IPCSO are IP
Checksum Controls
16
15
8
7
0
0
63
48 47
40 39
TUCSO TUCSS
HDRLEN
48 47 40 39 36 35 32 31
32 31
0
8
TUCSE
MSS
63
IPCSE
IPCSO
IPCSS
RSV STA TUCMD DTYPPAYLEN
24 23 20 19
Command
4
Status
1
3
2
0
7
6
5
3
2
1
0
3
2
1
0
SNAP DEXT RSV RS TSE
IP
TCP
RSV
DD
DTYP
IDE
DEXT must =1
for context
DTYP must =0000
for context
descriptor format
descriptor format
Figure 35.
Context Transmit Descriptor Format
The context descriptor provides access to the enhanced checksum off load and TCP
segmentation features available in the 82583V.
A context descriptor differs from a data descriptor as it does not point to packet data.
Instead, this descriptor provides access to on-chip contexts that support the transmit
checksum offloading or the segmentation feature of the 82583V. A context refers to a
set of parameters loaded or unloaded as a group to provide a particular function.
To select this descriptor format, the DEXT bit in the command field should be set to 1b
and TDESC.DTYP should be set to 0x0000. In this case, the descriptor format is defined
as shown in Figure 35.
7.2.9.2.1
IP and TCP/UDP Checksum Control
The first Qword of this descriptor type contains parameters used to calculate the two
checksums, which can be offloaded.
146
Inline Functions—82583V GbE Controller
IPCSS - IP Checksum Start - Specifies the byte offset from the start of the DMA'd data
to the first byte to be included in the checksum. Setting this value to 0b means the first
byte of the data would be included in the checksum. This field is limited to the first 256
bytes of the packet and must be less than or equal to the total length of a given packet.
If this is not the case, the results are unpredictable.
IPCSO - IP Checksum Offset - Specifies where the resulting checksum should be
placed. This field is limited to the first 256 bytes of the packet and must be less than or
equal to the total length of a given packet. If this is not the case, the checksum is not
inserted.
IPCSE - IP Checksum End - Specifies where the checksum should stop. A 16-bit value
supports checksum off loading of packets as large as 64 KB. Setting the IPCSE field to
all zeros means EOP. In this way, the length of the packet does not need to be
calculated.
Note:
Note:
When doing checksum or TCP segmentation with IPv6 headers IPCSE field should be
set to 0x0000, IPCSS should be valid as in IPv4 packet and the IXSM bit in the data
descriptor should be cleared.
For proper IP checksum calculation, the IP Header Checksum field should be set to zero
unless some adjustment is needed by the driver.
Similarly, TUCSS, TUCSO, TUCSE specify the same parameters for the TCP or UDP
checksum.
Note:
Note:
For proper TCP/UDP checksum calculation the TCP/UDP Checksum field should be set to
the partial pseudo-header checksum value.
In case of 802.1Q header, the offset values depend on the VLAN insertion enable bit -
CTRL.VME. When the CTRL.VME is not set (VLAN tagging included in the packet
buffers), the offset values should include the VLAN tagging. When the CTRL.VME is set
(VLAN tagging is taken from the packet descriptor), the offset values should exclude
the VLAN tagging.
When setting the TCP segmentation context, IPCSS and TUCSS are used to indicate the
start of the IP and TCP headers respectively, and must be set even if checksum
insertion is not desired.
In certain situations, software might need to calculate a partial checksum (the TCP
pseudo-header for instance) to include bytes that are not contained within the range of
start and end. If this is the case, this partial checksum should be placed in the packet
data buffer, at the appropriate offset for the checksum. If no partial checksum is
required, software must write a value of zero at this offset.
7.2.9.3
Max Segment Size - MSS
MSS controls the maximum segment size. This specifies the maximum TCP or UDP
payload segment sent per frame, not including any header. The total length of each
frame (or section) sent by the TCP segmentation mechanism (excluding 802.3ac
tagging and Ethernet CRC) is MSS bytes + HRDLEN. The one exception is the last
packet of a TCP segmentation that might be shorter. This field is ignored if TDESC.TSE
is not set.
147
82583V GbE Controller—Inline Functions
7.2.9.3.1
Header Length - HDRLEN
HDRLEN is used to specify the length (in bytes) of the header to be used for each frame
of a TCP segmentation operation. The first HDRLEN bytes fetched from data descriptor
are stored internally and are used as a prototype header. The prototype header is
updated for each packet and is prepended to the packet payload. For UDP packets this
will normally be equal to UDP checksum offset + 2. For TCP messages it will normally
be equal to TCP checksum offset + 4 + TCP header option bytes. This field is ignored if
TDESC.TSE is not set.
Maximum limits for the HDRLEN and MSS fields are dictated by the lengths variables.
However, there is a further restriction that for any TCP segmentation operation, the
hardware must be capable of storing a complete framed fragment (completely-built
frames) in the transmit FIFO prior to transmission. Therefore, the output TX FIFO
(packet buffer) should at least have (MSS + HDRLEN) space available. In addition MSS
must be set to a value more than 0x10 and HDRLEN must be smaller than 256 bytes.
7.2.9.4
Payload - PAYLEN
The Packet Length field (PAYLEN) is the total number of payload bytes for this TCP
segmentation offload (for example, the total number of payload bytes includes those
that are distributed across multiple frames after TCP segmentation is performed).
Following the fetch of the prototype header, PAYLEN specifies the length of data that is
fetched next from data descriptor. This field is also used to determine when last-frame
processing needs to be performed. The PAYLEN specification does not include any
header bytes. This field is ignored if TDESC.TSE is not set.
Note:
There is no restriction on the overall PAYLEN specification with respect to the transmit
FIFO size, once the MSS and HDRLEN specifications are legal.
7.2.9.5
Descriptor Type - DTYP
Setting the descriptor type (TDESC.DTYP) field to 0x0000 identifies this descriptor as a
context descriptor.
7.2.9.6
Command - TUCMD
The command field (TDESC.TUCMD) provides options that control the checksum
offloading and TCP segmentation features, along with some of the generic descriptor
processing functions. Table 41 lists the bit definitions for the TDESC.TUCMD field. The
IDE, DEXT, and RS bits are valid regardless of the state of TSE. All other bits are
ignored if TSE=0b.
Table 41.
Command TUCMD Fields
7
6
5
4
3
2
1
0
IDE
SNAP
DEXT
Rsv
RS
TSE
IP
TCP
148
Inline Functions—82583V GbE Controller
IDE (bit 7) - Interrupt Delay Enable
SNAP (bit 6) - SNAP
DEXT (bit 5) - Descriptor extension (must be 1b for this descriptor type)
Rsv (bit 4) - Reserved
RS (bit 3) - Report status
TSE (bit 2) - TCP segmentation enable
IP (bit 1) - IP Packet type (IPv4=1b, IPv6=0b)
TCP (bit 0) - Packet type (TCP=1b,UDP=0b)
IDE activates a transmit interrupt delay timer. Hardware loads a countdown register
when it writes back a transmit descriptor that has RS and IDE set. The value loaded
comes from the IDV field of the Interrupt Delay (TIDV) register. When the count
reaches zero, a transmit interrupt occurs if transmit descriptor write-back interrupts
(TXDW) are enabled. Hardware always loads the transmit interrupt counter whenever it
processes a descriptor with IDE set even if it is already counting down due to a
previous descriptor. If hardware encounters a descriptor that has RS set, but not IDE, it
generates an interrupt immediately after writing back the descriptor and clears the
interrupt delay timer. Setting the IDE bit has no meaning without setting the RS bit.
Note:
Although the transmit interrupt may be delayed, the descriptor write-back requested
by setting the RS bit is performed without delay unless descriptor write-back bursting is
enabled.
SNAP indicates that the TCP segmentation MAC header includes a SNAP header that
needs to be updated by hardware.
The DEXT bit identifies this descriptor as one of the extended descriptor types and
must be set to 1b.
When the RS bit is set, hardware writes back the DD bit once the DMA fetch completes.
Note:
Note:
Descriptors with the null address (0), or zero length, transfer no data. If they have the
RS bit in the command byte set, then the DD field in the status word is written when
hardware processes them. Hardware only sets the DD bit for descriptors with RS set.
Software can set the RS bit in each descriptor or, more likely, in specific descriptors
such as the last descriptor of each packet.
TSE indicates that this descriptor is setting the TCP segmentation context. If this bit is
zero, the descriptor defines a single packet TCP/UDP, IP checksum offload mode. When
a descriptor of this type is processed, the device immediately updates the mode in
question (TCP segmentation or checksum offloading) with values from the descriptor.
This means that if any normal packets or TCP segmentation packets are in progress (a
descriptor with EOP set has not been received for the given context) the results will
likely be undesirable.
The IP bit is used to indicate what type of IP (IPv4 or IPv6) packet is used in the
segmentation process. This is necessary for the 82583V to know where the IP Payload
Length field is located. This does not override the checksum insertion bit, IXSM. The IP
bit must only be set for IPv4 packets and cleared for IPv6 packets.
149
82583V GbE Controller—Inline Functions
The TCP bit identifies the packet as either TCP or UDP (non-TCP). This affects the
processing of the header information.
7.2.9.7
Status - STA
Four bits are reserved to provide transmit status, although only one is currently
assigned for this specific descriptor type.
The status word will only be written back to host memory in cases where the RS bit is
set in the command. DD indicates that the descriptor is done and is written back after
the descriptor has been processed only if the RS bit was set.
3
2
1
0
Reserved
DD
Figure 36.
Transmit Status Layout
Rsv (bits 3-1) - Reserved
DD (bit 0) - Descriptor Done
7.2.10
Extended Data Descriptor Format
63
48 47
40 39
36 35
32
31
24
23
20 19
0
0
8
Addresses
VLAN
POPTS
ExtCMD
STA
DCMD
DTYP
DTALEN
Command
Status
0
0
DD IDE
7
2
1
0
3
7
6
5
4
3
2
1
0
13 12 11
3
1
0
1
15
TXSM IXSM
TS
RSV
RSV
PRI
CFI VLAN ID
VLE DEXT RSV RS TSE IFCS EOP
RSV
Figure 37.
Extended Data Descriptor Format
The extended data descriptor is the companion to the context descriptor described in
the previous section. This descriptor type points to the location of the data in the host
memory.
To select this descriptor format, bit 29 (TDESC.DEXT) must be set to 1b and
TDESC.DTYP must be set to 0x0001. In this case, the descriptor format is defined as
shown in Figure 37.
The first Qword of this descriptor type contains the address of a data buffer in host
memory. This buffer contains all or a portion of a transmit packet.
The second Qword of this descriptor contains information about the data pointed to by
this descriptor as well as descriptor processing options.
150
Inline Functions—82583V GbE Controller
7.2.10.1
Data Length - DTALEN
The Data Length field (TDESC.DTALEN) is the total length of the data pointed to by this
descriptor (the entire send), in bytes. For data descriptors not associated with a TCP
segmentation operation (TDESC.TSE not set), the descriptor lengths are subject to the
same restrictions specified for legacy descriptors (the sum of the lengths of the data
descriptors comprising a single packet must be at least 80 bytes less than the allocated
size of the transmit FIFO).
7.2.10.2
Descriptor Type - DTYP
Setting the descriptor type (TDESC.DTYP) field to 0x0001 identifies this descriptor as
an extended data descriptor.
7.2.10.3
Table 42.
Command - DCMD
The command field (TDESC.DCMD) provides options that control the checksum
offloading TCP segmentation features, along with some of the generic descriptor
processing features. Table 42 lists the bit definitions for the DCMD field.
Command DCMD Fields
7
6
5
4
3
2
1
0
IDE
VLE
DEXT
RSV
RS
TSE
IFCS
EOP
IDE (bit 7) - Interrupt delay enable
VLE (bit 6) - VLAN enable
DEXT (bit 5) - Descriptor extension (must be 1b for this descriptor type)
RSV (bit 4) - Reserved
RS (bit 3) - Report status
TSE (bit 2) - TCP segmentation enable
IFCS (bit 1) - Insert FCS (also controls insertion of Ethernet CRC)
EOP (bit 0) - End of packet
IDE activates a transmit interrupt delay timer. Hardware loads a countdown register
when it writes back a transmit descriptor that has RS and IDE set. The value loaded
comes from the IDV field of the Interrupt Delay (TIDV) register. When the count
reaches zero, a transmit interrupt occurs if transmit descriptor write-back interrupts
(TXDW) are enabled. Hardware always loads the transmit interrupt counter whenever it
processes a descriptor with IDE set even if it is already counting down due to a
previous descriptor. If hardware encounters a descriptor that has RS set, but not IDE, it
generates an interrupt immediately after writing back the descriptor and clears the
interrupt delay timer. Setting the IDE bit has no meaning without setting the RS bit.
151
82583V GbE Controller—Inline Functions
Although the transmit interrupt might be delayed, the descriptor write-back requested
by setting the RS bit is performed without delay unless descriptor write-back bursting is
enabled.
VLE indicates that the packet is a VLAN packet (for example, that the hardware should
add the VLAN Ether type and an 802.1Q VLAN tag to the TCP message).
Table 43.
VLAN Tag Insertion Decision Table
VLE
Action
Send generic Ethernet packet. IFCS controls insertion of FCS in normal Ethernet
packets.
0
1
Send 802.1Q packet; the Ethernet Type field comes from the VET register and the
VLAN data comes from the special field of the TX descriptor; hardware always
appends the FCS/CRC.
Note:
Note:
If the VLE bit is set to enable VLAN tag insertion, the CTRL.VME bit should also be set.
The DEXT bit identifies this descriptor as one of the extended descriptor types and
must be set to 1b.
When the RS bit is set, the hardware writes back the DD bit once the DMA fetch
completes.
Descriptors with the null address (0), or zero length, transfer no data. If they have the
RS bit in the command byte set, then the DD field in the status word is written when
hardware processes them. Hardware only sets the DD bit for descriptors with RS set.
Software can set the RS bit in each descriptor or, more likely, in specific descriptors
such as the last descriptor of each packet.
TSE indicates that this descriptor is part of the current TCP segmentation context. If
this bit is not set, the descriptor is part of the normal non-segmentation context.
IFCS controls insertion of the Ethernet CRC. The packet FCS covers the TCP/IP headers.
Therefore, when using the TCP segmentation offload, software must also use the FCS
insertion.
Note:
The VLE, IFCS, and VLAN fields are only valid in certain descriptors. If TSE is enabled,
the VLE, IFCS, and VLAN fields are only valid in the first data descriptor of the TCP
segmentation context. If TSE is not enabled, then these fields are only valid in the last
descriptor of the given packet (qualified by the EOP bit).
EOP when set, indicates the last descriptor making up the packet.
152
Inline Functions—82583V GbE Controller
7.2.10.4
Status - STA
The status field is written back to host memory in cases where the RS bit is set in the
command field. The DD bit indicates that the descriptor is done after the descriptor has
been processed.
3
2
1
0
Rsv
DD
Rsv (bit 3:1) - Reserved
DD (bit 0) - Descriptor done
7.2.10.5
Packet Options - POPTS
The POPTS field provides a number of options, which control the handling of this
packet. This field is relevant only on the first data descriptor of a packet or
segmentation context.
7
2
1
0
Rsv
TXSM
IXSM
Rsv (bits 7:2) - Reserved
TXSM (bit 1) - Insert TCP/UDP checksum
IXSM (bit 0) - Insert IP checksum
IXSM and TXSM are used to control insertion of the IP and TCP/UDP checksums,
respectively. If the corresponding bit is not set, whatever value software has placed
into the checksum field of the packet data is placed on the wire.
Note:
Note:
For proper values of the IP and TCP checksum, software must set the IXSM and TXSM
when using the transmit segmentation.
Software should not set this field for IPv6 packets.
7.2.10.6
VLAN
The VLAN field is used to provide the 802.1Q tagging information. The special field is
ignored if the VLE bit in the DCMD command byte is 0b.
15
13
12
11
0
PRI
CFI
VLAN ID
153
82583V GbE Controller—Inline Functions
7.3
TCP Segmentation
TCP segmentation is an offloading option of the TCP/IP stack. This is often referred to
as Transmit Segmentation Offloading (TSO). This feature obligates the software device
driver and hardware to carve up TCP messages, larger than the Maximum Transmission
Unit (MTU) of the medium, into MSS sized frames that have appropriate layer 2, 3 (IP),
and 4 (TCP) headers. These headers must have the correct sequence number, IP
identification, checksum fields, options and flag values as required. This is done by
breaking up the data into segments smaller than or equal to the MSS.
Note:
Note that some of these values (such as the checksum values) are unique for each
packet of the TCP message, and other fields such as the source IP address are constant
for all frames associated with the TCP message.
The offloading of these mechanisms to the software device driver and the 82583V
saves significant CPU cycles. The software device driver shares the additional tasks to
support these options with the 82583V.
7.3.1
TCP Segmentation Performance Advantages
Performance advantages for a hardware implementation of TCP segmentation offload
include:
• The stack does not need to partition the block to fit the MTU size, saving CPU
cycles.
• The stack only computes one Ethernet, IP, and TCP header per segment (entire
packet), saving CPU cycles.
• The stack interfaces with the software device driver only once per block transfer,
instead of once per frame.
• Interrupts are easily reduced to once per TCP message instead of once per frame.
• Fewer I/O accesses are required to command the the 82583V.
Note:
TCP segmentation requires the transmit context descriptor format and the transmit
data descriptor format.
7.3.2
Ethernet Packet Format
A TCP message can be fragmented across multiple pages in host memory. The 82583V
partitions the data packet into standard Ethernet frames prior to transmission. The
82583V supports calculating the Ethernet, IP, TCP, and UDP headers, including
checksum, on a frame-by-frame basis.
L2
L3
L4
Ethernet
IP
TCP
DATA
FCS
Figure 38.
TCP/IP Packet Format
154
Inline Functions—82583V GbE Controller
Frame formats supported by the 82583V include:
• Ethernet 802.3
• IEEE 802.1q VLAN (Ethernet 802.3ac)
• Ethernet Type 2
• Ethernet SNAP
• IPv4 headers with options
• IPv6 headers with IP option next headers
• TCP with options
• UDP with options
VLAN tag insertion is handled by hardware.
Note:
IP tunneled packets are not supported for TSO operation.
Once the TCP segmentation context has been set, the next descriptor provides the
initial data to transfer. This first descriptor(s must point to a packet of the type
indicated. Furthermore, the data it points to might need to be modified by software as
it serves as the prototype (partial pseudo-header) header for all packets within the TCP
segmentation context. The following sections describe the supported packet types and
the various updates which are performed by hardware. This should be used as a guide
to determine what must be modified in the original packet header to make it a suitable
prototype (partial pseudo-header) header.
7.3.3
TCP Segmentation Data Descriptors
The TCP segmentation data descriptor is the companion to the TCP segmentation
context descriptor described in the previous section. For a complete description of the
descriptor please refer to section 7.2.10.
To select this descriptor format, bit 29 (TDESC.DEXT) must be set to 1b and
TDESC.DTYP must be set to 0x0001.
155
82583V GbE Controller—Inline Functions
7.3.4
TCP Segmentation Source Data
Once the TCP segmentation context has been set, the next descriptor (data descriptor)
provides the initial data to transfer. This first data descriptor must point to data
containing an Ethernet header of the type indicated. The 82583V fetches the prototype
(partial pseudo-header) header from the host data buffer into an internal buffer and
this header is prepended to every packet for this TSO operation. The prototype (partial
pseudo-header) header is modified accordingly for each MSS sized segment. The
following sections describe the supported packet types and the various updates that
are performed by hardware. This should be used as a guide to determine what must be
modified in the original packet header to make it a suitable prototype (partial pseudo-
header) header.
The following summarizes the fields considered by the driver for modification in
constructing the prototype (partial pseudo-header) header.
MAC Header (for SNAP)
• MAC Header LEN field should be set to 0b.
IPv4 Header
• Length should be set to zero.
• Identification field should be set as appropriate for first packet of send (if not
already).
• Header checksum should be zeroed out unless some adjustment is needed by the
software device driver.
IPv6 Header
• Length should be set to zero.
TCP Header
• Sequence number should be set as appropriate for first packet of send (if not
already).
• PSH, and FIN flags should be set as appropriate for LAST packet of send.
• TCP checksum should be set to the partial pseudo-header checksum.
UDP Header
• UDP checksum should be set to the partial pseudo-header checksum.
The 82583V's DMA function fetches the IP, and TCP/UDP prototype (partial pseudo-
header) header information from the initial descriptor and save them on-chip for
individual packet header generation.
7.3.5
Hardware Performed Updating for Each Frame
The following sections describe the updating process performed by the hardware for
each frame sent using the TCP segmentation capability.
156
Inline Functions—82583V GbE Controller
7.3.6
TCP Segmentation Use of Multiple Data Descriptors
TCP segmentation enables a series of data descriptors, each referencing a single
physical address page, to reference a large packet contained in a single virtual-address
buffer.
The only requirement on use of multiple data descriptors for TCP segmentation is as
follows:
• If multiple data descriptors are used to describe the IP/TCP/UDP header section,
each descriptor must describe one or more complete headers; descriptors
referencing only parts of headers are not supported.
Note:
It is recommended that the entire header section, as described by the TCP Context
Descriptor HDRLEN field, be coalesced into a single buffer and described using a single
data descriptor. If all the layer headers (L2-L4) are not coalesced into a single buffer,
each buffer must not cross a 4 KB boundary, or be bigger than MAX_READ_REQUEST.
7.3.6.1
Transmit Checksum Offloading with TCP Segmentation
The 82583V supports checksum offloading as a component of the TCP segmentation
offload feature and as a standalone capability.
The 82583V supports IP and TCP/UDP header options in the checksum computation for
packets that are derived from the TCP segmentation feature.
Note:
The 82583V is capable of computing one level of IP header checksum and one TCP/UDP
header and payload checksum. In case of multiple IP headers, the software device
driver has to compute all but one IP header checksum. The 82583V calculates
checksums on the fly on a frame-by-frame basis and inserts the result in the IP/TCP/
UDP headers of each frame. TCP and UDP checksum are a result of performing the
checksum on all bytes of the payload and the pseudo header.
Three specific types of checksum are supported by the hardware in the context of the
TCP Segmentation off load feature:
• IPv4 checksum (IPv6 does not have a checksum)
• TCP checksum
• UDP checksum
Each packet that is sent via the TCP segmentation offload feature optionally includes
the IPv4 checksum and either the TCP or UDP checksum.
All checksum calculations use a 16-bit wide ones complement checksum. The
checksum word is calculated on the outgoing data. The checksum field is written with
the 16-bit ones complement sum of all 16-bit words in the range of CSS to CSE,
including the checksum field itself.
157
82583V GbE Controller—Inline Functions
7.3.6.2
IP/TCP/UDP Header Updating
IP/TCP/UDP header is updated for each outgoing frame based on the IP/TCP header
prototype (partial pseudo-header) which the hardware gets from the first descriptor
and stores on chip. The IP/TCP/UDP headers are fetched from host memory into an on-
chip 240 byte header buffer once for each TCP segmentation context (for performance
reasons, this header is not fetched for each additional packet that will be derived from
the TCP segmentation process). The checksum fields and other header information are
updated on a frame-by-frame basis. The updating process is performed concurrently
with the packet data fetch.
7.3.6.2.1
TCP/IP/UDP Header for the First Frame
The hardware makes the following changes to the headers of the first packet that is
derived from each TCP segmentation context.
MAC Header (for SNAP)
• Type/Len field = MSS + HDRLEN - 14
IPv4 Header
• IP Total Length = MSS + HDRLEN - IPCSS
• IP Checksum
IPv6 Header
• Payload Length = MSS + HDRLEN - IPCSS - Ipv6Size (while Ipv6Size = 40Bytes)
TCP Header
• Sequence Number: The value is the Sequence Number of the first TCP byte in this
frame.
• If FIN flag = 1b, it is cleared in the first frame.
• If PSH flag =1b, it is cleared in the first frame.
• TCP Checksum
UDP Header
• UDP length: MSS + HDRLEN - TUCSS
• UDP Checksum
7.3.6.2.2
TCP/IP/UDP Header for the Subsequent Frames
The hardware makes the following changes to the headers of the subsequent packets
that is derived from each TCP segmentation context.
Note:
Number of bytes left for transmission = PAYLEN - (N * MSS). Where N is the number of
frames that have been transmitted.
MAC Header (for SNAP Packets)
• Type/Len field = MSS + HDRLEN - 14
158
Inline Functions—82583V GbE Controller
IPv4 Header
• IP Identification: incremented from last value (wrap around)
• IP Total Length = MSS + HDRLEN - IPCSS
• IP Checksum
IPv6 Header
• Payload Length = MSS + HDRLEN - IPCSS - Ipv6Size (while Ipv6Size = 40Bytes)
TCP Header
• Sequence Number update: Add previous TCP payload size to the previous sequence
number value. This is equivalent to adding the MSS to the previous sequence
number.
• If FIN flag = 1b, it is cleared in these frames.
• If PSH flag =1b, it is cleared in these frames.
• TCP Checksum
UDP Header
• UDP Length: MSS + HDRLEN - TUCSS
• UDP Checksum
7.3.6.2.3
TCP/IP/UDP Header for the Last Frame
The hardware makes the following changes to the headers of the last packet that is
derived from each TCP segmentation context.
Note:
Last frame payload bytes = PAYLEN - (N * MSS)
MAC Header (for SNAP Packets)
• Type/Len field = Last frame payload bytes + HDRLEN - 14
IPv4 Header
• IP Total Length = (last frame payload bytes + HDRLEN) - IPCSS
• IP Identification: incremented from last value (wrap around)
• IP Checksum
IPv6 Header
• Payload Length = last frame payload bytes + HDRLEN - IPCSS - Ipv6Size (while
Ipv6Size = 40Bytes)
TCP Header
• Sequence Number update: Add previous TCP payload size to the previous sequence
number value. This is equivalent to adding the MSS to the previous sequence
number.
• If FIN flag = 1b, set it in this last frame
• If PSH flag =1b, set it in this last frame
• TCP Checksum
159
82583V GbE Controller—Inline Functions
UDP Header
• UDP length: (last frame payload bytes + HDRLEN) - TUCSS
• UDP Checksum
7.4
Interrupts
The 82583V supports the following interrupt modes:
• PCI legacy interrupts
• PCI MSI - Message Signaled Interrupts
7.4.1
Legacy and MSI Interrupt Modes
In legacy and MSI modes, an interrupt cause is reflected by setting one of the bits in
the ICR register, where each bit reflects one or more causes. This description of ICR
register provides the mapping of interrupt causes (for example, a specific Rx queue
event or a LSC event) to bits in the ICR.
Mapping of causes relating to the Tx and Rx queues as well as non-queue causes in this
mode is not configurable. Each possible queue interrupt cause (such as the Rx queue,
Tx queue or any other interrupt source) has an entry in the ICR.
The following configuration and parameters are involved:
• The ICR[31:0] bits are allocated to specific interrupt causes
7.4.2
Registers
The interrupt logic consists of the registers listed in the following table, plus the
registers associated with MSI signaling.
Register
Acronym
Function
Records all interrupt causes - an interrupt is signaled when
unmasked bits in this register are set.
Interrupt Cause
ICR
Interrupt Cause Set
Interrupt Mask Set/Read
Interrupt Mask Clear
ICS
IMS
IMC
Enables software to set bits in the Interrupt Cause register.
Sets or reads bits in the interrupt mask.
Clears bits in the Interrupt mask.
Enables bits in the ICR and IMS without a read or write of the
ICR.
Interrupt Auto Clear
Interrupt Auto Mask
EIAC
IAM
Enables bits in the IMS to be set automatically.
Interrupt Cause Registers (ICR)
This register records the interrupts causes to provide to the software information on
the interrupt source.
160
Inline Functions—82583V GbE Controller
The interrupt causes include:
• The receive and transmit related interrupts.
• Other bits in this register are the legacy indication of interrupts as the MDIC
completes a link status change. There is a specific Other Cause bit that is set if one
of these bits are set.
Interrupt Cause Set Register (ICS)
This registers allows triggering an immediate interrupt by software, By writing 1b to
bits in ICS the corresponding bits in ICR is set Used usually to rearm interrupts the
software didn't have time to handle in the current interrupt routine.
Interrupt Mask Set and Read Register (IMS) and Interrupt Mask Clear
Register (IMC)
Interrupts appear on PCIe only if the interrupt cause bit is a one and the corresponding
interrupt mask bit is a one. Software blocks assertion of an interrupt by clearing the
corresponding bit in the mask register. The cause bit stores the interrupt event
regardless of the state of the mask bit. Clear and set make this register more thread
safe by avoiding a read-modify-write operation on the mask register. The mask bit is
set for each bit written to a one in the set register and cleared for each bit written in
the clear register. Reading the set register (IMS) returns the current mask register
value.
Interrupt Auto Clear Enable Register (EIAC)
Bits 24:20 in this register enables clearing of the corresponding bit in ICR following
interrupt generation. When a bit is set, the corresponding bit in ICR and in IMS is
automatically cleared following an interrupt.
Bits in the ICR that are not set in EIAC need to be cleared with ICR read or ICR write-
to-clear.
Interrupt Auto Mask Enable register (IAM)
Each bit in this register enables setting of the corresponding bit in IMS following write
to-clear to ICR.
7.4.3
Interrupt Moderation
The 82583V implements interrupt moderation to reduce the number of interrupts
software processes. The moderation scheme is based on a timer called ITR Interrupt
Throttle register). In general terms, the ITR defines an interrupt rate by defining the
time interval between consecutive interrupts.
The number of ITR registers is:
• A single ITR is used (ITR).
Software uses ITR to limit the rate of delivery of interrupts to the host CPU. It provides
a guaranteed inter-interrupt delay between interrupts asserted by the network
controller, regardless of network traffic conditions.
161
82583V GbE Controller—Inline Functions
The following algorithm converts the inter-interrupt interval value to the common
'interrupts/sec' performance metric:
Interrupts/sec = (256 * 10-9 sec x interval) -1
For example, if the interval is programmed to 500d, the 82583V guarantees the CPU is
not interrupted by it for at least 128 μs from the last interrupt.
Inversely, inter-interrupt interval value can be calculated as:
Inter-interrupt interval = (256 * 10-9 sec x interrupts/sec) -1
The optimal performance setting for this register is very system and configuration
specific.
ITR rules:
• The maximum observable interrupt rate from the adapter should not exceed 7813
interrupts/sec.
• The Extended Interrupt Throttle register should default to 0x0 upon initialization
and reset.
Each time an interrupt event happens, the corresponding bit in the ICR is activated.
The interrupt flow should follow the following diagram:
162
Inline Functions—82583V GbE Controller
Load counter
with interval
Start count
down
No
Counter = 0
?
Yes
Yes
No
Interrupt
active
?
Yes
v
Assert Interrupt
No
Intr ack
?
Yes
v
Clear Interrupt
Figure 39.
Interrupt Throttle Flow Diagram
For cases where the 82583V is connected to a small number of clients, it is desirable to
fire off the interrupt as soon as possible with minimum latency.
163
82583V GbE Controller—Inline Functions
Case A: Heavy load, interrupts moderated
Intr
Intr
Intr
ITR delay
Pkt
ITR delay
Pkt
Pkt
Pkt
Pkt
Pkt
Pkt
Pkt
Case B: Light load, interrupts immediately on packet receive
Intr
Intr
ITR delay
Pkt
Pkt
7.4.4
Clearing Interrupt Causes
The 82583V has two methods available for to clear ICR bits: clear-on-write and clear-
on-read.
164
Inline Functions—82583V GbE Controller
Write to Clear
The ICR register clears specific interrupt cause bits in the register after writing 1b to
those bits. Any bit that was written with a 0b remains unchanged.
Read to clear
All bits in the ICR register are cleared on a read to ICR.
7.5
802.1q VLAN Support
The 82583V provides several specific mechanisms to support 802.1q VLANs:
• Optional adding (for transmits) and ping (for receives) of IEEE 802.1q VLAN tags.
• Optional ability to filter packets belonging to certain 802.1q VLANs.
7.5.1
802.1q VLAN Packet Format
The following diagram compares an untagged 802.3 Ethernet packet with an 802.1q
VLAN tagged packet:
802.1q VLAN
802.3 Packet
#Octets
#Octets
Packet
DA
SA
6
DA
SA
6
6
6
Type/Length
Data
2
46-1500
4
802.1q Tag
Type/Length
Data
4
2
46-1500
4
CRC
CRC*
Note:
The CRC for the 802.1q tagged frame is re-computed, so that it covers the entire
tagged frame including the 802.1q tag header. Also, maximum frame size for an 802.1q
VLAN packet is 1522 octets as opposed to 1518 octets for a normal 802.3z Ethernet
packet.
7.5.1.1
802.1q Tagged Frames
For 802.1q, the Tag Header field consists of four octets comprised of the Tag Protocol
Identifier (TPID) and Tag Control Information (TCI); each taking two octets. The first
16 bits of the tag header makes up the TPID. It contains the protocol type, which
identifies the packet as a valid 802.1q tagged packet.
The two octets making up the TCI contain three fields:
• User Priority (UP)
• Canonical Form Indicator (CFI). Should be 0b for transmits. For receives, the
device has the capability to filter out packets that have this bit set. See the CFIEN
and CFI bits in the RCTL described in section 9.2.5.1.
• VLAN Identifier (VID)
165
82583V GbE Controller—Inline Functions
The bit ordering is as follows:
Octet 1
Octet 2
UP
CFI
VID
7.5.2
Transmitting and Receiving 802.1q Packets
Since the 802.1q tag is only four bytes, adding and stripping of tags could be done
completely in software. (In other words, for transmits, software inserts the tag into
packet data before it builds the transmit descriptor list, and for receives, software strips
the 4-byte tag from the packet data before delivering the packet to upper layer
software.)
However, because adding and stripping of tags in software results in more overhead for
the host, the 82583V has additional capabilities to add and strip tags in hardware. See
section 7.5.2.1 and section 7.5.2.2.
7.5.2.1
Adding 802.1q Tags on Transmits
Software might command the 82583V to insert an 802.1q VLAN tag on a per packet
basis. If CTRL.VME is set to 1b, and the VLE bit in the transmit descriptor is set to 1b,
then the 82583V inserts a VLAN tag into the packet that it transmits over the wire.
The Tag Protocol Identifier (TPID) field of the 802.1q tag comes from the VET register,
and the Tag Control Information (TCI) of the 802.1q tag comes from the special field of
the transmit descriptor.
7.5.2.2
Stripping 802.1q Tags on Receives
Software might instruct the 82583V to strip 802.1q VLAN tags from received packets. If
the CTRL.VME bit is set to 1b, and the incoming packet is an 802.1q VLAN packet (for
example, it's Ethernet Type field matched the VET), then the 82583V strips the 4-byte
VLAN tag from the packet, and stores the TCI in the Special field of the receive
descriptor.
The 82583V also sets the VP bit in the receive descriptor to indicate that the packet had
a VLAN tag that was stripped. If the CTRL.VME bit is not set, the 802.1q packets can
still be received if they pass the receive filter, but the VLAN tag is not stripped and the
VP bit is not set.
7.5.3
802.1q VLAN Packet Filtering
VLAN filtering is enabled by setting the RCTL.VFE bit to 1b. If enabled, hardware
compares the type field of the incoming packet to a 16-bit field in the VLAN Ether Type
(VET) register. If the VLAN type field in the incoming packet matches the VET register,
the packet is then compared against the VLAN filter table array for acceptance.
166
Inline Functions—82583V GbE Controller
The Virtual LAN ID field indexes a 4096 bit vector. If the indexed bit in the vector is
one; there is a virtual LAN match. Software might set the entire bit vector to ones if the
node does not implement 802.1q filtering. The register description of the VLAN filter
table array is described in detail in section 9.2.5.20.
In summary, the 4096-bit vector is comprised of 128, 32-bit registers. Matching to this
bit vector follows the same algorithm as indicated in section 7.1.1 for multicast address
filtering. The VLAN Identifier (VID) field consists of 12 bits. The upper 7 bits of this field
are decoded to determine the 32-bit register in the VLAN filter table array to address
and the lower 5 bits determine which of the 32 bits in the register to evaluate for
matching.
Two other bits in the Receive Control register (see section 9.2.5.1), CFIEN and CFI, are
also used in conjunction with 802.1q VLAN filtering operations. CFIEN enables the
comparison of the value of the CFI bit in the 802.1q packet to the Receive Control
register CFI bit as acceptance criteria for the packet.
Note:
The VFE bit does not effect whether the VLAN tag is stripped. It only affects whether
the VLAN packet passes the receive filter.
Table 44 lists reception actions per control bit settings.
Table 44.
Packet Reception Decision Table
Is
packet
802.1q?
CTRL.
VME
RCTL.
VFE
Action
No
X
X
Normal packet reception.
Receive a VLAN packet if it passes the standard filters (only).
Leave the packet as received in the data buffer. VP bit in receive
descriptor is cleared.
Yes
0b
0b
1b
1b
0b
1b
0b
1b
Receive a VLAN packet if it passes the standard filters and the
VLAN filter table. Leave the packet as received in the data buffer
(for example, the VLAN tag would not be stripped). VP bit in
receive descriptor is cleared.
Yes
Yes
Yes
Receive a VLAN packet if it passes the standard filters (only). Strip
off the VLAN information (four bytes) from the incoming packet
and store in the descriptor. Sets the VP bit in receive descriptor.
Receive a VLAN packet if it passes the standard filters and the
VLAN filter table. Strip off the VLAN information (four bytes) from
the incoming packet and store in the descriptor. Sets the VP bit in
receive descriptor.
Note:
A packet is defined as a VLAN/802.1q packet if its type field matches the VET.
7.6
LEDs
The 82583V implements three output drivers intended for driving external LED circuits
per port. Each of the three LED outputs can be individually configured to select the
particular event, state, or activity, which is indicated on that output. In addition, each
LED can be individually configured for output polarity as well as for blinking versus non-
blinking (steady-state) indication.
The configuration for LED outputs is specified via the LEDCTL register. Furthermore, the
hardware-default configuration for all the LED outputs, can be specified via NVM fields,
thereby supporting LED displays configurable to a particular OEM preference.
167
82583V GbE Controller—Inline Functions
Each of the three LED's might be configured to use one of a variety of sources for
output indication. The Mode bits control the LED source:
• LINK_100/1000 is asserted when link is established at either 100 or 1000 Mb/s.
• LINK_10/1000 is asserted when link is established at either 10 or 1000 Mb/s.
• LINK_UP is asserted when any speed link is established and maintained.
• ACTIVITY is asserted when link is established and packets are being transmitted or
received.
• LINK/ACTIVITY is asserted when link is established AND there is NO transmit or
receive activity
• LINK_10 is asserted when a 10 Mb/s link is established and maintained.
• LINK_100 is asserted when a 100 Mb/s link is established and maintained.
• LINK_1000 is asserted when a 1000 Mb/s link is established and maintained.
• FULL_DUPLEX is asserted when the link is configured for full duplex operation.
• COLLISION is asserted when a collision is observed.
• PAUSED is asserted when the device's transmitter is flow controlled.
• LED_ON is always asserted; LED_OFF is always de-asserted.
The IVRT bits enable the LED source to be inverted before being output or observed by
the blink-control logic. LED outputs are assumed to normally be connected to the
negative side (cathode) of an external LED.
The BLINK bits control whether the LED should be blinked while the LED source is
asserted, and the blinking frequency (either 200 ms on and 200 ms off or 83 ms on and
83 ms off)1. The blink control can be especially useful for ensuring that certain events,
such as ACTIVITY indication, cause LED transitions, which are sufficiently visible to a
human eye. The same blinking rate is shared by all LEDs.
Note:
Note that the LINK/ACTIVITY source functions slightly different from the others when
BLINK is enabled. The LED is off if there is no LINK, on if there is LINK and no
ACTIVITY, and blinking if there is LINK and ACTIVITY.
1. While in Smart Power Down mode, the blinking durations are increased by 5x to 1 second and
415 ms, respectively.
168
Inline Functions—82583V GbE Controller
Note:
This page intentionally left blank.
169
82583V GbE Controller—Power Management and Delivery
8.0
Power Management and Delivery
The 82583V supports the Advanced Configuration and Power Interface (ACPI 2.0)
specification as well as Advanced Power Management (APM). This section describes
how power management is implemented in the 82583V.
Implementation requirements were obtained from the following documents:
• PCI Bus Power Management Interface Specification .................................Rev 1.1
• PCI Express Base Specification .............................................................Rev.1.1
• ACPI Specification ...............................................................................Rev 2.0
• PCI Express Card Electromechanical Specification....................................Rev 1.1
8.1
Assumptions
The following assumptions apply to the implementation of power management for the
82583V.
• The software device driver sets up the filters prior to the system transition of the
82583V to a D3 state.
• Prior to transition from D0 to the D3 state, the operating system ensures that the
software device driver has been disabled. See Section 8.4.4.2.3 for the 82583V
behavior on D3 entry.
• No wake up capability, except APM wake up if enabled in the NVM, is required after
the system puts the 82583V in D3 state and then returns the 82583V to D0.
• If the APMPME bit in the Wake Up Control (WUC) register is 1b, it is permissible to
assert PE_WAKE_N even when PME_En is 0b.
8.2
Power Consumption
Table 16 and Table 17 list power consumption in various modes (see Section 3.5). The
following sections describe the requirements in specific power states.
170
Power Management and Delivery—82583V GbE Controller
8.3
Power Delivery
82583V operates from the following power rails:
• A 3.3 V dc power rail for internal power regulation and for periphery. The 3.3 V dc
should be supplied by an external power source.
• A 1.9 V dc power rail.
• A 1.05 V dc power rail.
8.3.1
The 1.9 V dc Rail
The 1.9 V dc rail is used for core and I/O functions. It also feeds internal regulators to a
lower 1.05 V dc core voltage. The 1.9 V dc rail can be generated in one of two ways:
• An external power supply not dependent on support from the 82583V. For
example, the platform designer might choose to route a platform-available 1.9 V dc
supply to the 82583V.
• Internal voltage regulator solution, where the control logic for the power transistor
is embedded in the 82583V, while the power transistor is placed externally. Control
is done using the CTRL19 pin.
8.3.2
The 1.05 V dc Rail
The 1.05 V dc rail is used for core functions and can be generated in one of the
following ways:
• An external power supply not dependent on support from the 82583V.
• Internal voltage regulator solution, where the control logic for the power transistor
is embedded in the 82583V, while the power transistor is placed externally. Control
is done using the CTRL10 pin.
• A complete internal voltage regulator solution. The internal voltage regulator can
be disabled by the DIS_REG10 pin.
8.4
Power Management
8.4.1
82583V Power States
The 82583V supports D0 and D3 power states defined in the PCI Power Management
and PCIe specifications. D0 is divided into two sub-states: D0u (D0 Un-initialized), and
D0a (D0 active). In addition, the 82583V supports a Dr state that is entered when
PE_RST_N is asserted (including the D3cold state).
Figure 40 shows the power states and transitions between them.
171
82583V GbE Controller—Power Management and Delivery
Hot (in-band)
Reset
Internal Power
On Reset
PE_RST_N de-
assertion
assertion and
EEPROM read
done
Dr
D0u
PE_RST_N
assertion
Enable
master or
slave access
PE_RST_N
assertion
Write 00b
to power
state
PE_RST_N
assertion
Write 11b
to power
state
D3
D0a
Figure 40.
Power Management State Diagram
8.4.2
Auxiliary Power Usage
If ADVD3WUC=1b, the 82583V uses the AUX_PWR indication that auxiliary power is
available to the controller, and therefore advertises D3cold wake up support. The
amount of power required for the function (which includes the entire NIC) is advertised
in the Power Management Data register, which is loaded from the NVM.
If D3cold is supported, the PME_En and PME_Status bits of the Power Management
Control/Status Register (PMCSR), as well as their shadow bits in the Wake Up Control
(WUC) register is reset only by the power up reset (detection of power rising).
The only effect of setting AUX_PWR to 1b is advertising D3cold wake up support and
changing the reset function of PME_En and PME_Status. AUX_PWR is a strapping option
in the 82583V.
The 82583V tracks the PME_En bit of the Power Management Control / Status Register
(PMCSR) and the Auxiliary (AUX) Power PM Enable bit of the PCIe Device Control
register to determine the power it might consume (and therefore its power state) in the
D3cold state (internal Dr state).
172
Power Management and Delivery—82583V GbE Controller
The AUX Power PM Enable bit in the PCIe Device Control register determines if the
82583V complies with the auxiliary power regime defined in the PCIe specification. If
set, the 82583V might consume higher power for any purpose (such as, even if PME_En
is not set).
If the AUX Power PM Enable bit of the PCIe Device Control register is cleared, higher
power consumption is determined by the PCI-PM legacy PME_En bit in the Power
Management Control / Status Register (PMCSR).
Note:
In the current implementation, the AUX Power PM Enable bit is hardwired to 0b.
8.4.3
Power Limits by Certain Form Factors
Table 45 lists the power limitations introduced by different form factors.
Power Limits by Form Factor
Table 45.
Form Factor
LOM
PCIe NIC (x1 connector)
3 A @ 3.3 V dc
Main
3 A @ 3.3 V dc
375 mA @ 3.3 V dc
20 mA @ 3.3 V dc
Auxiliary (aux enabled)
Auxiliary (aux disabled)
375 mA @ 3.3 V dc
1. This auxiliary current limit only applies when the primary 3.3 V dc voltage source is
not available (such as, the NIC is in a low power D3 state.
2. The 82583V exceeds the allowed power consumption in GbE speed. It therefore
cannot run from aux power, restricting the 82583V speed in Dr state.
The 82583V therefore implements two NVM bits to disable GbE operation in certain
cases:
1. The Disable 1000 NVM bit disables 1000 Mb/s operation under all conditions.
2. The Disable 1000 in non-D0a CSR bit disables 1000 Mb/s operation in non-D0a
states. If Disable 1000 in non-D0a is set, and the 82583V is at GbE speed on entry
to a non-D0a state, then the device removes advertisement for 1000 Mb/s and
auto-negotiates. The Disable 1000 in non-D0a bit is loaded from the NVM.
Note:
The 82583V restarts link auto-negotiation each time it transitions from a state where
GbE speed is enabled to a state where GbE speed is disabled, or vice versa. For
example, if Disable 1000 in non-D0a is set but Disable 1000 is clear, the 82583V
restarts link auto-negotiation on transition from D0 state to D3 or Dr states.
8.4.4
Power States
8.4.4.1
D0 Uninitialized State
The D0u state is a low-power state used after PE_RST_N is de-asserted following a
power up (cold or warm), on hot reset (in-band reset through a PCIe physical layer
message), or on D3 exit.
173
82583V GbE Controller—Power Management and Delivery
When entering the D0u state, the 82583V disables all wake ups and asserts a reset to
the PHY while the NVM is being read. If the APM Mode bit in the NVM's Initialization
Control Word 2 is set, then APM wake up is enabled.
8.4.4.1.1
Entry into D0u state
D0u is reached from either the Dr state (on assertion of Internal PwrGd) or the D3hot
state (by configuration software writing a value of 00b to the Power State field of the
PCI-PM registers).
Asserting Internal PwrGd means that the entire state of the device is cleared, other
than sticky bits. The state is loaded from the NVM, followed by establishment of the
PCIe link. Once this is done, configuration software can access the device.
On a transition from the D3 to D0u state, the 82583V’s PCI configuration space is not
reset. Per the PCI Power Management Specification (revision 1.1, Section 5.4),
software “will need to perform a full re-initialization of the function including its PCI
Configuration Space.”
8.4.4.2
D0active State
Once memory space is enabled, all internal clocks are activated and the 82583V enters
an active state. It can transmit and receive packets if properly configured by the
software device driver. The PHY is enabled or re-enabled by the software device driver
to operate / auto-negotiate to full-line speed/power if not already operating at full
capability. Any APM Wakeup previously active remains active. The software device
driver can deactivate APM Wakeup by writing to the WUC register, or activate other
wake-up filters by writing to the Wake Up Filter Control (WUFC) register.
Note:
Fields that are auto-loaded from the NVM, like WUC.APME, should be configured
through an NVM setting, because D3 to D0 power state transition causes NVM auto-
read to reload those bits from the NVM.
8.4.4.2.1
Entry to D0a State
D0a is entered from the D0u state by writing a 1b to the Memory Access Enable or the
I/O Access Enable bit in the PCI Command register. The DMA, MAC, and PHY are
enabled.
8.4.4.2.2
D3 State (=PCI-PM D3hot)
When the system writes a 11b to the Power State field in the PMCSR, the 82583V
transitions to D3. Any wake-up filter settings that were enabled before entering this
reset state are maintained. Upon transition to D3 state, the 82583V clears the Memory
Access Enable and I/O Access Enable bits of the PCI Command register, which disables
memory access decode. In D3, the 82583V only responds to PCI configuration accesses
and does not generate master cycles.
A D3 state is followed by either a D0u state (in preparation for a D0a state) or by a
transition to Dr state (PCI-PM D3cold state). To transition back to D0u, the system
writes a 00b to the Power State field of the PMCSR. Transition to Dr state is through
PE_RST_N assertion.
174
Power Management and Delivery—82583V GbE Controller
8.4.4.2.3
Entry to D3 State
Transition to the D3 state is through a configuration write to the Power State field of the
PCI-PM registers.
Prior to transition from D0 to the D3 state, the software device driver disables
scheduling of further tasks to the 82583V, as follows:
• It masks all interrupts
• It does not write to the Transmit Descriptor Tail (TDT) register
• It does not write to the Receive Descriptor Tail (RDT) register
• Operates the master disable algorithm as defined in Section 6.1.3.10.
If wake-up capability is needed, the software device driver should set up the
appropriate wake-up registers and the system should write a 1b to the PME_En bit in
the PMCSR or to the AUX Power PM Enable bit of the PCIe Device Control register prior
to the transition to D3.
As a response to being programmed into the D3 state, the 82583V brings its PCIe link
into the L1 link state. As part of the transition into L1 state, the 82583V suspends
scheduling of new Transaction Layer Protocols (TLPs) and waits for the completion of all
previous TLPs it has sent. The 82583V clears the Memory Access Enable and I/O Access
Enable bits of the PCI Command register, which disables memory access decode. Any
receive packets that have not been transferred into system memory are kept in the
device (and discarded later on D3 exit). Any transmit packets that were not sent, can
still be transmitted (assuming the Ethernet link is up).
To reduce power consumption, if APM wake and PCI-PM PME are enabled, the PHY auto-
negotiates to a lower link speed on D3 entry (see Section 8.4.4.2.3).
175
82583V GbE Controller—Power Management and Delivery
8.4.4.3
Dr State
Transition to Dr state is initiated on three occasions:
• At system power up - Dr state begins with the assertion of the internal power
detection circuit (Internal Power On Reset) and ends with the assertion of the
Internal Pwrgd signal (indicating that the system de-asserted its PCIe PE_RST_N
signal).
• At transition from a D0a state - During operation, the system might assert PCIe
PE_RST_N at any time. In an ACPI system, a system transition to the G2/S5 state
causes a transition from D0a to Dr state.
• At transition from a D3 state - The system transitions the device into the Dr state
by asserting PCIe PE_RST_N.
The 82583V meets the restrictions on using auxiliary power, defined in the PCI-PM
specification:
1. If wake is enabled (either APM wake or ACPI wake), then the 82583V might
consume up to 375 mA @ 3.3 V dc.
2. If wake is disabled, then the 82583V might consume up to 20 mA @ 3.3 V dc.
The restrictions apply to all cases of Dr state (power up, D3 entry, Dr entry from D0).
Note:
When the wake configuration is unknown (for example, during power up before an NVM
read), the 82583V must meet the 20 mA limit.
The system might maintain PE_RST_N asserted for an arbitrary time. The de-assertion
(rising edge) of PE_RST_N causes a transition to D0u state.
Any Wake-up filter settings that were enabled before entering this reset state are
maintained.
8.4.4.3.1
Entry to Dr State
Dr entry on platform power up begins by asserting the internal power detection circuit
(Internal Power On Reset). The NVM is read and determines device configuration. If the
APM Enable bit in the NVM's Initialization Control Word 2 is set, then APM wake up is
enabled. The PHY and MAC states are determined by the state of APM wake. To reduce
power consumption, if APM wake is enabled, the PHY auto-negotiates to a lower link
speed on Dr entry (see Section 8.4.4.3.1). The PCIe link is not enabled in Dr state
following system power up (since PERS# is asserted).
Entry to Dr state from D0a state is by asserting the PE_RST_N signal. An ACPI
transition to the G2/S5 state is reflected in a device transition from D0a to Dr state.
The transition might be orderly (for example, the designer selected the shut down
option), in which case the software device driver might have a chance to intervene. Or,
it might be an emergency transition (such as, power button override), in which case,
the software device driver is not notified.
To reduce power consumption, if APM wake or PCI-PM PME is enabled, the PHY auto-
negotiates to a lower link speed on D0a to Dr transition (see Section 8.4.4.3.1).
Transition from D3 state to Dr state is done by asserting the PE_RST_N signal. Prior to
that, the system initiates a transition of the PCIe link from the L1 state to either the L2
or L3 state. The link enters L2 state if PCI-PM PME is enabled.
176
Power Management and Delivery—82583V GbE Controller
8.4.4.4
Device Disable
For a LOM design, it might be desirable for the system to provide BIOS-setup capability
for selectively enabling or disabling LOM devices. This might allow the designers more
control over system resource-management, avoid conflicts with add-in NIC solutions,
etc. The 82583V provides support for selectively enabling or disabling it.
• Device Disable - the device is in a global power down state.
Device disable is initiated by asserting the asynchronous DEV_OFF_N pin. The
DEV_OFF_N pin has an internal pull-up resistor, so that it can be left not connected to
enable device operation.
While in device disable mode, the PCIe link is in L3 state. The PHY is in power-down
mode. All internal clocks are gated. Output buffers are tri-stated.
Asserting or de-asserting PCIe PE_RST_N does not have any effect while the device is
in device disable mode (for example, the device stays in the respective mode as long as
DEV_OFF_N is asserted). However, the device might momentarily exit the device
disable mode from the time PCIe PE_RST_N is de-asserted again and until the NVM is
read.
Note:
Note to system designers: The DEV_OFF_N pin should maintain its state during system
reset and system sleep states. It should also insure the proper default value on system
power up. For example, a system designer could use a GPIO pin that defaults to 1b
(enable) and is on system suspend power (for example, it maintains state in S0-S5
ACPI states).
8.4.4.5
Link-Disconnect
In any of D0u, D0a, D3, or Dr states, the 82583V enters a link-disconnect state if it
detects a link-disconnect condition on the Ethernet link. Note that the link-disconnect
state is invisible to software (other than the Link Energy Detect bit state). In particular,
while in D0 state, software might be able to access any of the device registers as in a
link-connect state.
During link disconnect mode, the CCM PLL might be shut down. See Section 8.4.4.5.
8.4.5
Timing of Power-State Transitions
The following sections give detailed timing for the state transitions. In the diagrams the
dotted connecting lines represent the 82583V requirements, while the solid connecting
lines represent the 82583V guarantees.
The timing diagrams are not to scale. The clocks edges are shown to indicate running
clocks only, they are not used to indicate the actual number of cycles for any operation.
8.4.5.1
Transition From D0a to D3 and Back Without PE_RST_N
Figure 41 shows the 82583V’s reaction to a D3 transition.
177
82583V GbE Controller—Power Management and Delivery
PCIe Reference
Clock
PCIe PwrGd
D0 Write
td0me
2
3
tee
m
Memory Access Enable
6
Auto
Read
Ext.
Conf.
Reading EEPROM
PHY Reset
5
7
D3 write
1
PCIe Link
L0
L0
L1
4
Wake Up Enabled
PHY Power State
DState
Any mode
APM
power-
managed
full
power-managed
full
D0
D0a
D3
D0u
Figure 41.
Table 46.
D3hot Transition Timing Diagram
Notes to D3hot Timing Diagram
Note
Description
1
2
3
4
Writing 11b to the Power State field of the PMCSR transitions the 82583V to D3.
The system keeps the 82583V in D3 state for an arbitrary amount of time.
To exit D3 state the system writes 00b to the Power State field of the PMCSR.
APM wake up mode can be enabled based on what is read in the NVM.
After reading the NVM, reset to the PHY is de-asserted. The PHY operates at reduced-speed if APM
wake up is enabled, else powered-down.
5
6
The system can delay an arbitrary time before enabling memory access.
Writing a 1b to the Memory Access Enable bit or to the I/O Access Enable bit in the PCI Command
register transitions the 82583V from D0u to D0 state and returns the PHY to full-power/speed
operation.
7
8.4.5.2
Transition From D0a to D3 and Back with PE_RST_N
Figure 42 shows the 82583V’s reaction to a D3 transition.
178
Power Management and Delivery—82583V GbE Controller
4a
PCIe Reference
Clock
4b
tl2clk
tclkp
g
tPWRGD-CLK
6
PCIe PwrGd
3
tpgdl
tppg-
clkint
tl2pg
7
Internal PCIe clock
(2.5 GHz)
tclkp
r
8
Internal PwrGd
(PLL)
9
tee
5
Auto
Read
Ext.
Conf.
Reading EEPROM
11
Reset to PHY
(active low)
tpgcfg
tpgres
D3 write
tpgtrn
12
15
13
2
14
1
PCIe Link
Wake Up Enabled
PHY Power State
L0
L1
L2/L3
L0
L0
10
Any mode
APM
full
power-managed
Dr
full
DState D0a
D3
D0u
D0a
Figure 42.
Table 47.
D3cold Transition Timing Diagram
Notes to D3cold Timing Diagram
Note
Description
Writing 11b to the Power State field of the PMCSR transitions the 82583V to D3. PCIe link transitions
to L1 state.
1
The system can delay an arbitrary amount of time between setting D3 mode and transition the link to
an L2 or L3 state.
2
3
4
Following link transition, PE_RST_N is asserted.
The system must assert PE_RST_N before stopping the PCIe reference clock. It must also wait tl2clk
after link transition to L2/L3 before stopping the reference clock.
5
6
7
8
On assertion of PE_RST_N, the 82583V transitions to Dr state.
The system starts the PCIe reference clock t
before de-asserting PE_RST_N.
PWRGD-CLK
The Internal PCIe clock is valid and stable t
from PE_RST_N de-assertion.
ppg-clkint
The PCIe Internal PWRGD signal is asserted tclkpr after the external PE_RST_N signal.
Asserting Internal PCIe PWRGD causes the NVM to be re-read, asserts PHY reset, and disables wake
up.
9
10
11
12
13
14
APM wake-up mode can be enabled based on what is read from the NVM.
After reading the NVM, PHY reset is de-asserted.
Link training starts after tpgtrn from PE_RST_N de-assertion.
A first PCIe configuration access might arrive after t
from PE_RST_N de-assertion.
pgcfg
A first PCI configuration response can be sent after tpgres from PE_RST_N de-assertion
Writing a 1b to the Memory Access Enable bit in the PCI Command register transitions the device
from the D0u to D0 state.
15
179
82583V GbE Controller—Power Management and Delivery
8.5
Wake Up
The 82583V supports two types of wake-up mechanisms:
• Advanced Power Management (APM) wake up
• PCIe power management wake up
The PCIe power management wake up uses the PE_WAKE_N pin to wake the system
up. The advanced power management wake up can be configured to use the
PE_WAKE_N pin as well.
8.5.1
Advanced Power Management Wake Up
Advanced power management wake up, or APM wake up, was previously known as
wake on LAN. It is a feature that has existed in the 10/100 Mb/s NICs for several
generations. The basic premise is to receive a broadcast or unicast packet with an
explicit data pattern, and then to assert a signal to wake up the system. In the earlier
generations, this was accomplished by using special signal that ran across a cable to a
defined connector on the motherboard. The NIC would assert the signal for
approximately 50 ms to signal a wake up. The 82583V uses (if configured to) an in-
band PM_PME message for this.
At power up, the 82583V reads the APM Enable bits from the NVM Initialization Control
Word 2 into the APM Enable (APME) bits of the WUC. These bits control enabling of APM
wake up.
When APM wake up is enabled, the 82583V checks all incoming packets for Magic
Packets. See Section 8.5.3.1.4 for a definition of Magic Packets.
Once the 82583V receives a matching wake-up packet, it:
• If the Assert PME On APM Wakeup (APMPME) bit is set in the WUC:
— Sets the PME_Status bit in the PMCSR and issues a PM_PME message (in some
cases, this might require asserting the PE_WAKE_N signal first to resume
power and clock to the PCIe interface).
• Stores the first 128 bytes of the packet in the WUPM.
• Sets the relevant <wake up packet type> received bit in the WUS.
The 82583V maintains the first wake-up packet received in the WUPM until the
software device driver writes a 1b to the Magic Packet Received MAG bit in the WUS.
Note:
Note:
The WUPM latches on the first wake-up packet. Subsequent wake-up packets are not
saved until the programmer writes 1b to the relevant bit in the WUS. The best course of
action is to write a 1b to ALL of the WUC's bits, for example, set WUC = 0xFFFFFFFF.
Full power-on reset also clears the WUC.
APM wake up is supported in all power states and only disabled if a subsequent NVM
read results in the APM Wake Up bit being cleared or software explicitly writes a 0b to
the APM Wake Up (APM) bit of the WUC register.
180
Power Management and Delivery—82583V GbE Controller
8.5.2
PCIe Power Management Wake Up
The 82583V supports PCIe power management based wake ups. It can generate
system wake-up events from three sources:
• Reception of a Magic Packet*.
• Reception of a network wake-up packet.
• Detection of a link change of state.
Activating PCIe power management wake up requires the following steps:
• The software device driver programs the WUFC to indicate the packets it needs to
use to indicate wake up and supplies the necessary data to the Ipv4/v6 Address
Table (IP4AT, IP6AT) and the Flexible Filter Mask Table (FFMT), Flexible Filter
Length Table (FFLT), and the Flexible Filter Value Table (FFVT). It can also set the
Link Status Change Wake Up Enable (LNKC) bit in the WUFC to cause a wake up
when the link changes state.
• The operating system (at configuration time) writes a 1b to the PME_EN bit of the
PMCSR (bit 8).
Normally, after enabling wake up, the operating system writes a 11b to the lower two
bits of the PMCSR to put the 82583V into a low-power mode.
Once wake up is enabled, the 82583V monitors incoming packets, first filtering them
according to its standard address filtering method, then filtering them with all of the
enabled wake-up filters. If a packet passes both the standard address filtering and at
least one of the enabled wake-up filters, the 82583V:
• Sets the PME_Status bit in the PMCSR.
• If the PME_En bit in the PMCSR is set, asserts PE_WAKE_N.
• Stores the first 128 bytes of the packet in the WPM.
• Sets one or more of the Received bits in the WUS. (the 82583V set more than one
bit if a packet matches more than one filter.)
If enabled, a link state change wake up causes similar results, setting PME_Status,
asserting PE_WAKE_N and setting the Link Status Changed (LNKC) bit in the WUS
when the link goes up or down.
PE_WAKE_N remains asserted until the operating system either writes a b1 to the
PME_Status bit of the PMCSR or writes a 0b to the PME_EN bit.
After receiving a wake-up packet, the 82583V ignores any subsequent wake-up packets
until the software device driver clears all of the Received bits in the WUS. It also
ignores link change events until the software device driver clears the Link Status
Changed (LNKC) bit in the WUS.
8.5.3
Wake-Up Packets
The 82583V supports various wake-up packets using two types of filters:
• Pre-defined filters
• Flexible filters
Each of these filters are enabled if the corresponding bit in the WUFC is set to 1b.
181
82583V GbE Controller—Power Management and Delivery
8.5.3.1
Pre-Defined Filters
The following packets are supported by the 82583V's pre-defined filters:
• Directed packet (including exact, multicast indexed, and broadcast)
• Magic Packet*
• ARP/Ipv4 request packet
• Directed IPv4 packet
• Directed IPv6 packet
Each of these filters are enabled if the corresponding bit in the WUFC is set to 1b.
The explanation of each filter includes a table showing which bytes at which offsets are
compared to determine if the packet passes the filter. Both VLAN frames and LLC/SNAP
can increase the given offsets if they are present.
8.5.3.1.1
Directed Exact Packet
The 82583V generates a wake-up event upon receipt of any packet whose destination
address matches one of the 16 valid programmed receive addresses if the Directed
Exact Wake Up Enable bit is set in the Wake Up Filter Control Register (WUFC.EX).
.
# of
bytes
Offset
Field
Value
Action
Comment
Match any pre-
programmed address
0
6
Destination Address
Compare
8.5.3.1.2
Directed Multicast Packet
For multicast packets, the upper bits of the incoming packet's destination address index
a bit vector, the Multicast Table Array that indicates whether to accept the packet. If the
Directed Multicast Wake Up Enable bit set in the Wake Up Filter Control Register
(WUFC.MC) and the indexed bit in the vector is one then the 82583V generates a wake-
up event. The exact bits used in the comparison are programmed by software in the
Multicast Offset field of the Receive Control Register (RCTL.MO).
# of
bytes
Offset
Field
Value
Action
Compare
Comment
0
6
Destination Address
See above paragraph.
8.5.3.1.3
Broadcast
If the Broadcast Wake Up Enable bit in the Wake Up Filter Control Register (WUFC.BC)
is set, the 82583V generates a wake-up event when it receives a broadcast packet.
# of
bytes
Offset
Field
Value
Action
Compare
Comment
0
6
Destination Address
0xFF*6
182
Power Management and Delivery—82583V GbE Controller
8.5.3.1.4
Magic Packet*
Once the 82583V has been put into the Magic Packet* mode, it scans all incoming
frames addressed to the node for a specific data sequence, which indicates to the
controller that 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,
DESTINATION ADDRESS (which may be the receiving station's IEEE address or a
MULTICAST address which includes the BROADCAST address), and CRC. The specific
data sequence consists of 16 duplications of the IEEE 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 enables
the scanning state machine to be much simpler. The synchronization stream is defined
as 6 bytes of 0xFF. The 82583V also accepts a broadcast frame, as long as the 16
duplications of the IEEE address match the address of the machine to be awakened.
The 82583V expects the destination address to either:
1. Be the broadcast address (0xFF.FF.FF.FF.FF.FF)
2. Match the value in Receive Address Register 0 (RAH0, RAL0). This is initially loaded
from the NVM but might be changed by the software device driver.
3. Match any other address filtering enabled by the software device driver.
The 82583V searches for the contents of Receive Address Register 0 (RAH0, RAL0) as
the embedded IEEE address. It considers any non-0xFF byte after a series of at least 6
0xFFs to be the start of the IEEE address for comparison purposes. (that is it catches
the case of 7 0xFFs followed by the IEEE address). As soon as one of the first 96 bytes
after a string of 0xFFs doesn't match, it continues to search for anther set of at least 6
0xFFs followed by the 16 copies of the IEEE address later in the packet.
Note:
This definition precludes the first byte of the destination address from being 0xFF.
A Magic Packet's* destination address must match the address filtering enabled in the
configuration registers with the exception that broadcast packets are considered to
match even if the Broadcast Accept bit of the Receive Control Register (RCTL.BAM) is
0b. If APM Wakeup is enabled in the NVM, the 82583V starts up with the Receive
Address Register 0 (RAH0, RAL0) loaded from the NVM. This enables the 82583V to
accept packets with the matching IEEE address before the software device driver
comes up.
# of
Bytes
Offset
Field
Value
Action
Comment
MAC Header –
processed by main
address filter
0
6
6
Destination Address
Compare
6
8
4
4
6
Source Address
Skip
12
Possible LLC/SNAP Header
Possible VLAN Tag
Type
Skip
12
Skip
12
Skip
Any
Synchronizing Stream
0xFF*6+
A*16
Compare
Compared to Receive
Address Register 0
(RAH0, RAL0)
any+6
96
16 copies of Node Address
Compare
183
82583V GbE Controller—Power Management and Delivery
Accepting broadcast Magic Packets* for wake up purposes when the Broadcast Accept
bit of the Receive Control Register (RCTL.BAM) is 0b is a change from previous devices,
which initialized RCTL.BAM to 1b if APM was enabled in the NVM, but then required that
bit to be 1b to accept broadcast Magic Packets*, unless broadcast packets passed
another perfect or multicast filter.
8.5.3.1.5
ARP/IPv4 Request Packet
The 82583V supports receiving ARP Request packets for wake up if the ARP bit is set in
the WUFC. Four IPv4 addresses are supported, which are programmed in the IPv4
Address Table (IP4AT). A successfully matched packet must contain a broadcast MAC
address, a protocol type of 0x0806, an ARP opcode of 0x01, and one of the four
programmed IPv4 addresses. The 82583V also handles ARP request packets that have
VLAN tagging on both Ethernet II and Ethernet SNAP types.
# of
Offset
0
Field
Value
Action
Comment
Bytes
MAC Header –
processed by main
address filter
6
Destination Address
Compare
6
6
8
4
2
2
2
1
1
2
6
4
6
Source Address
Skip
12
12
12
14
16
18
19
20
22
28
32
Possible LLC/SNAP Header
Possible VLAN Tag
Type
Skip
Skip
0x0806
0x0001
0x0800
0x06
0x04
0x0001
-
Compare
Compare
Compare
Compare
Compare
Compare
Ignore
Ignore
Ignore
ARP
Hardware Type
Protocol Type
Hardware Size
Protocol Address Length
Operation
Sender Hardware Address
Sender IP Address
Target Hardware Address
-
-
May match any of four
values in IP4AT
38
4
Target IP Address
IP4AT
Compare
8.5.3.1.6
Directed IPv4 Packet
The 82583V supports receiving directed IPv4 packets for wake up if the IPV4 bit is set
in the WUFC. Four IPv4 addresses are supported, which are programmed in the IPv4
Address Table (IP4AT). A successfully matched packet must contain the station's MAC
address, a protocol type of 0x0800, and one of the four programmed IPv4 addresses.
The 82583V also handles directed IPv4 packets that have VLAN tagging on both
Ethernet II and Ethernet SNAP types.
184
Power Management and Delivery—82583V GbE Controller
# of
Offset
0
Field
Value
Action
Comment
Bytes
MAC Header –
processed by main
address filter
6
Destination Address
Compare
6
6
8
4
2
1
1
2
2
2
1
1
2
4
Source Address
Possible LLC/SNAP Header
Possible VLAN Tag
Type
Skip
12
12
12
14
15
16
18
20
22
23
24
26
Skip
Skip
0x0800
Compare
Compare
Ignore
Ignore
Ignore
Ignore
Ignore
Ignore
Ignore
Ignore
IP
Version/ HDR Length
Type of Service
Packet Length
0x4X
Check IPv4
-
-
-
-
-
-
-
-
Identification
Fragment Information
Time to Live
Protocol
Header Checksum
Source IP Address
May match any of four
values in IP4AT
30
4
Destination IP Address
IP4AT
Compare
8.5.3.1.7
Directed IPv6 Packet
The 82583V supports receiving directed IPv6 packets for wake up if the IPV6 bit is set
in the WUFC. One IPv6 address is supported and is programmed in the IPv6 Address
Table (IP6AT). A successfully matched packet must contain the station's MAC address,
a protocol type of 0x0800, and the programmed IPv6 address. The 82583V also
handles directed IPv6 packets that have VLAN tagging on both Ethernet II and Ethernet
SNAP types.
# of
Offset
0
Field
Value
Action
Comment
Bytes
MAC Header –
processed by main
address filter
6
Destination Address
Compare
6
6
8
4
2
1
3
Source Address
Possible LLC/SNAP Header
Possible VLAN Tag
Type
Skip
12
12
12
14
15
Skip
Skip
0x0800
0x6X
-
Compare
Compare
Ignore
IP
Version/ Priority
Flow Label
Check IPv6
185
82583V GbE Controller—Power Management and Delivery
# of
Offset
Field
Payload Length
Value
Action
Ignore
Comment
Bytes
18
20
21
22
38
2
1
1
-
-
-
-
Next Header
Ignore
Ignore
Ignore
Compare
Hop Limit
16
16
Source IP Address
Destination IP Address
IP6AT
Match value in IP6AT
8.5.3.2
Flexible Filter
The 82583V supports four flexible filters for host wake up and two flexible filters for
TCO wake up. For more details refer to Section 9.2.8.2. Each filter can be configured to
recognize any arbitrary pattern within the first 128 bytes of the packet. To configure
the flexible filter, software programs:
• The mask values into the Flexible Filter Mask Table (FFMT)
• The required values into the Flexible Filter Value Table (FFVT)
• The minimum packet length into the Flexible Filter Length Table (FFLT).
These contain separate values for each filter. Software must also:
• Enable the filter in the WUFC.
• Enable the overall wake-up functionality by setting PME_En in the PMCSR or WUC.
Once enabled, the flexible filters scan incoming packets for a match. If the filter
encounters any byte in the packet where the mask bit is one and the byte doesn't
match the byte programmed in FFVT, then the filter failed that packet. If the filter
reaches the required length without failing the packet, it passes the packet and
generates a wake-up event. It ignores any mask bits set to one beyond the required
length.
The following packets are listed for reference purposes only. The flexible filter could be
used to filter these packets.
8.5.3.2.1
IPX Diagnostic Responder Request Packet
An IPX Diagnostic Responder Request Packet must contain a valid MAC address, a
Protocol Type of 0x8137, and an IPX Diagnostic Socket of 0x0456. It may include LLC/
SNAP Headers and VLAN Tags. Since filtering this packet relies on the flexible filters,
which use offsets specified by the operating system directly, the operating system must
account for the extra offset LLC/SNAP Headers and VLAN tags.
# of
bytes
Offset
Field
Value
Action
Compare
Comment
0
6
6
6
8
4
Destination Address
Source Address
Skip
Skip
Skip
12
12
Possible LLC/SNAP Header
Possible VLAN Tag
186
Power Management and Delivery—82583V GbE Controller
# of
Offset
Field
Value
Action
Compare
Comment
bytes
12
14
30
2
Type
0x8137
-
IPX
16
2
Typical IPX Information
IPX Diagnostic Socket
Ignore
0x0456
Compare
8.5.3.2.2
Directed IPX Packet
A valid directed IPX packet contains:
• The station's MAC address.
• A protocol type of 0x8137.
• an IPX node address that equals the station's MAC address.
It might also include LLC/SNAP Headers and VLAN Tags. Since filtering this packet
relies on the flexible filters, which use offsets specified by the operating system
directly, the operating system must account for the extra offset LLC/SNAP headers and
VLAN tags.
# of
bytes
Offset
0
Field
Value
Action
Comment
MAC Header –
processed by main
address filter
6
Destination Address
Compare
6
6
8
4
2
Source Address
Skip
12
12
12
14
Possible LLC/SNAP Header
Possible VLAN Tag
Type
Skip
Skip
0x8137
-
Compare
Ignore
IPX
10
6
Typical IPX Information
Receive
Address 0
Must match Receive
Address 0
24
IPX Node Address
Compare
8.5.3.2.3
8.5.3.3
IPv6 Neighbor Discovery Filter
In IPpv6, a neighbor discovery packet is used for address resolution. A flexible filter can
be used to check for a neighborhood discovery packet.
Wake-Up Packet Storage
The 82583V saves the first 128 bytes of the wake-up packet in its internal buffer, which
can be read through the WUPM after the system wakes up.
187
82583V GbE Controller—Driver Programing Interface
9.0
Driver Programing Interface
9.1
Introduction
This chapter details the programmer visible state inside the 82583V. In some cases, it
describes hardware structures invisible to software in order to clarify a concept.
The 82583V's address space is mapped into four regions. These regions are listed in
Table 48:
Table 48.
82583V Address Space
Addressable Content
Internal registers and memories
How Mapped
Size of Region
Direct memory mapped
Direct memory-mapped
Direct memory-mapped
I/O window mapped
128 KB
Flash (optional)
64 KB-16 MB
2 KB-256 KB
32 bytes
Expansion ROM (optional)
Internal registers and memories, Flash (optional)
Both the Flash and Expansion ROM Base Address Registers (BARs) map the same Flash
memory.
The internal registers, memories, and Flash can be accessed though I/O space
indirectly, as explained in the sections that follow.
9.1.1
Memory and I/O Address Decoding
9.1.1.1
Memory-Mapped Access to Internal Registers and Memories
The internal registers and memories can be accessed as direct memory-mapped offsets
from the Base Address Register 0 (BAR0). The appropriate offset for each specific
internal register is described in this section.
9.1.1.2
9.1.1.3
Memory-Mapped Access to Flash
The external Flash can be accessed using direct memory-mapped offsets from the Flash
Base Address Register 1 (BAR1). The Flash is only accessible if enabled through the
NVM Initialization Control Word, and if the Flash BAR1 contains a valid (non-zero) base
memory address. For accesses, the offset from the Flash BAR1 corresponds to the
offset into the Flash actual physical memory space.
Memory-Mapped Access to Expansion ROM
The external Flash can also be accessed as a memory-mapped expansion ROM.
Accesses to offsets starting from the Expansion ROM BAR reference the Flash, provided
that access is enabled through the NVM Initialization Control Word, and the Expansion
ROM BAR contains a valid (non-zero) base memory address.
188
Driver Programing Interface—82583V GbE Controller
9.1.1.4
I/O-Mapped Access to Internal Registers, Memories, and Flash
To support pre-boot operation (prior to the allocation of physical memory base
addresses), all internal registers, memories, and Flash can be accessed using I/O
operations. I/O accesses are supported only if:
• An I/O Base Address Register (BAR) is allocated and mapped (BAR2)
• The BAR contains a valid (non-zero) value
• I/O address decoding is enabled in the PCIe configuration
When an I/O BAR is mapped, the I/O address range allocated opens a 32-byte window
in the system I/O address map. Within this window, two I/O addressable registers are
implemented:
• IOADDR
• IODATA
The IOADDR register is used to specify a reference to an internal register, memory, or
Flash, and then the IODATA register is used as a window to the register, memory or
Flash address specified by IOADDR:
R/
W
Offset
Abbreviation
Name
Size
Internal register, internal memory, or Flash location
address.
0x00000-0x1FFFF – Internal registers and memories.
0x20000-0x7FFFF – Undefined.
0x00
IOADDR
R/W 4 bytes
0x80000-0xFFFFF – Flash.
Data field for reads or writes to the Internal Register,
Internal Memory, or Flash Location as identified by
the current value in IOADDR. All 32 bits of this
register are read/write-able.
0x04
IODATA
R/W 4 bytes
0x08 – 0x1F
Reserved
Reserved
RO
4 bytes
9.1.1.4.1
IOADDR (I/O Offset 0x00)
The IOADDR register must always be written as a Dword access. Writes that are less
than 32 bits are ignored. Reads of any size return a Dword of data. However, the
chipset or CPU might only return a subset of that Dword.
For software programmers, the IN and OUT instructions must be used to cause I/O
cycles to be used on the PCIe bus. Because writes must be to a 32-bit quantity, the
source register of the OUT instruction must be EAX (the only 32-bit register supported
by the OUT command). For reads, the IN instruction can have any size target register,
but it is recommended that the 32-bit EAX register be used.
Because only a particular range is addressable, the upper bits of this register are hard
coded to zero. Bits 31 through 20 cannot be written to and always read back as 0b.
At hardware reset (Internal Power On Reset) or PCI Reset, this register value resets to
0x00000000. Once written, the value is retained until the next write or reset.
9.1.1.4.2
IODATA (I/O Offset 0x04)
The IODATA register must always be written as a Dword access when the IOADDR
register contains a value for the internal register and memories (such as, 0x00000-
0x1FFFC). In this case, writes that are less than 32 bits are ignored.
189
82583V GbE Controller—Driver Programing Interface
The IODATA register may be written as a byte, word, or Dword access when the
IOADDR register contains a value for the Flash (such as, 0x80000-0xFFFFF). In this
case, the value in IOADDR must be properly aligned to the data value. The following
table lists the supported configurations:
82583V IOADDR Register Bits
[1:0]
Target IODATA Access BE[3:0]#
bits in Data Phase
Access Type
Byte (8 bit)
00b
01b
10b
11b
00b
10b
00b
1110b
1101b
1011b
0111b
1100b
0011b
0000b
Word (16 bit)
Dword (32 bit)
Note:
Software might have to implement non-obvious code to access the Flash, a byte, or
word at a time. Example code that reads a Flash byte is shown here to illustrate the
impact of the previous table:
char *IOADDR;
char *IODATA;
IOADDR = IOBASE + 0;
IODATA = IOBASE + 4;
*(IOADDR) = Flash_Byte_Address;
Read_Data = *(IODATA + (Flash_Byte_Address % 4));
Reads to IODATA of any size return a Dword of data. However, the chipset or CPU might
only return a subset of that Dword.
For software programmers, the IN and OUT instructions must be used to cause I/O
cycles to be used on the PCIe bus. Where 32-bit quantities are required on writes, the
source register of the OUT instruction must be EAX (the only 32-bit register supported
by the OUT command).
Writes and reads to IODATA when the IOADDR register value is in an undefined range
(0x20000-0x7FFFC) should not be performed. Results cannot be determined.
Note:
There are no special software timing requirements on accesses to IOADDR or IODATA.
All accesses are immediate except when data is not readily available or acceptable. In
this case, the 82583V delays the results through normal bus methods (for example,
split transaction or transaction retry).
Note:
Because a register/memory/Flash read or write takes two I/O cycles to complete,
software must provide a guarantee that the two I/O cycles occur as an atomic
operation. Otherwise, results can be non-deterministic from the software viewpoint.
9.1.1.4.3
Undefined I/O Offsets
I/O offsets 0x08 through 0x1F are considered to be reserved offsets with the I/O
window. Dword reads from these addresses return 0xFFFF; writes to these addresses
are discarded.
190
Driver Programing Interface—82583V GbE Controller
9.1.2
Registers Byte Ordering
This section defines the structure of registers that contain fields carried over the
network. Some examples are L2, L3, L4 fields.
The following example is used to describe byte ordering over the wire (hex notation):
Last
First
..., 06
05
04
03
02
01
00
where each byte is sent with the Least Significant Bit (LSB) first. That is, the bit order
over the wire for this example is
Last
First
....
0000 0011
0000 0010
0000 0001
0000 0000
The general rule for register ordering is to use host ordering. Using the previous
example, a 6-byte fields (such as, MAC address) is stored in a CSR in the following
manner:
Byte 3
Byte 2
Byte 1
Byte 0
DW address (N)
0x03
0x02
0x01
0x05
0x00
0x04
DW address (N+4)
The following exceptions use network ordering. Using the previous example, a 16-bit
field (such as, EtherType) is stored in a CSR in the following manner:
Byte 3
Byte 2
Byte 1
Byte 0
(DW aligned)
...
...
0x01
...
0x00
...
or (WORD aligned)
0x00
0x01
The following exception uses network ordering:
• All ETherType fields
Note:
The normal notation as it appears in text books, etc. is to use network ordering.
Example: Suppose a MAC address of 00-A0-C9-00-00-00. The order on the network is
00, then A0, then C9, etc. However, the host ordering presentation is:
Byte 3
Byte 2
Byte 1
Byte 0
Dword address (N)
00
...
C9
...
A0
00
00
00
Dword address (N+4)
9.1.3
Register Conventions
All registers in the 82583V are defined to be 32 bits. They should be accessed as 32-bit
double-words. There are some exceptions to this rule:
• Register pairs where two 32-bit registers make up a larger logical size.
• Accesses to Flash memory (via expansion ROM space, secondary BAR space, or the
I/O space) can be byte, word or double word accesses.
191
82583V GbE Controller—Driver Programing Interface
Reserved bit positions: Some registers contain certain bits that are marked as
reserved.
Reads from registers containing reserved bits might return indeterminate values in the
reserved bit-positions unless read values are explicitly stated. When read, these
reserved bits should be ignored by software.
Reserved and/or undefined addresses: any register address not explicitly declared
in this specification should be considered to be reserved, and should not be written to.
Note:
Writing to reserved or undefined register addresses can cause indeterminate behavior.
Reads from reserved or undefined configuration register addresses might return
indeterminate values unless read values are explicitly stated for specific addresses.
Initial values: most registers define the initial hardware values prior to being
programmed. In some cases, hardware initial values are undefined and are listed as
such via the text undefined, unknown, or X. Some of these configuration values should
be set via NVM configuration or via software in order to insure proper operation. This
need is dependent on the function of the bit. Other registers might cite a hardware
default which is overridden by a higher-precedence operation. Operations that might
supersede hardware defaults can include:
• A valid NVM load
• Completion of a hardware operation (such as hardware auto-negotiation)
• Writing of a different register whose value is then reflected in another bit
For registers that should be accessed as 32-bit double words, partial writes (less than a
32-bit double word) does not take effect (such as, the write is ignored). Partial reads
return all 32 bits of data regardless of the byte enables.
Note:
Note:
Partial reads to clear-by-read registers (such as, ICR) can have unexpected results
since all 32 bits are actually read regardless of the byte enables. Partial reads should
not be done.
All statistics registers are implemented as 32-bit registers. Though some logical
statistics registers represent counters in excess of 32-bits in width, registers must be
accessed using 32-bit operations (such as, independent access to each 32-bit field).
See special notes for VLAN Filter table and multicast table arrays in their specific
register definitions.
9.2
Configuration and Status Registers - CSR Space
Register Summary Table
9.2.1
All registers are listed in Section 49. These registers are ordered by grouping and are
not necessarily listed in the order that they appear in the address space.
192
Driver Programing Interface—82583V GbE Controller
Table 49.
82583V Register Summary
Alias
Offset
Link to
Page
Category
Offset
Abbreviation
Name
RW
0x00000 /
0x00004
General
General
General
N/A
CTRL
STATUS
EEC
Device Control Register
RW
R
page 197
page 200
page 201
0x00008
0x00010
N/A
N/A
Device Status Register
RW/
RO
EEPROM/FLASH Control Register
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
General
0x00014
0x00018
0x0001C
0x00020
0x00028
0x0002C
0x00030
0x00038
0x00170
0x05F40
0x00E00
0x00F00
0x00F08
0x01000
0x1010
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
EERD
EEPROM Read Register
Extended Device Control Register
Flash Access Register
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RO
page 203
page 203
page 205
page 206
page 208
page 208
page 209
page 209
page 209
page 210
page 210
page 212
page 212
page 213
page 214
page 203
page 214
page 214
page 214
page 214
page 215
page 215
page 216
page 216
page 216
page 216
CTRL_EXT
FLA
MDIC
MDI Control Register
FCAL
Flow Control Address Low
Flow Control Address High
Flow Control Type
FCAH
FCT
VET
VLAN Ether Type
FCTTV
Flow Control Transmit Timer Value
Flow Control Refresh Threshold Value
LED Control
FCRTV
LEDCTL
EXTCNF_CTRL
EXTCNF_SIZE
PBA
Extended Configuration Control
Extended Configuration Size
Packet Buffer Allocation
EEPROM Control Register
EEPROM Read/Write Data
Flash Control Register
EECTL
0x1014
EEDATA
FLCTL
RO
0x1018
RO
0x101C
0x1020
FLDATA
FLCNT
FLASH Read data
RO
FLASH Read Counter
RO
0x01028
0x0102C
0x1030
FLASHT
EEWR
FLASH Timer Register
RW
RW
RW
RW
RW
RW
RW
EEPROM Write Register
SW FLASH Burst Control Register
SW FLASH Burst Data Register
SW FLASH Burst Access Counter
FLASH Opcode Register
FLEEP Auto Load
FLSWCTL
FLSWDATA
FLSWCNT
FLOP
0x1034
0x1038
0x0103C
0x1050
FLOL
PCIe
PCIe
PCIe
PCIe
PCIe
PCIe
PCIe
PCIe
0x05B00
0x05B08
0x05B10
0x05B14
0x05B18
0x05B1C
0x05B20
0x05B24
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
GCR
3GIO Control Register
RW
RW
RW
RW
RW
RW
RW
RW
page 216
page 218
page 218
page 219
page 219
page 219
page 219
page 219
FUNCTAG
GSCL_1
GSCL_2
GSCL_3
GSCL_4
GSCN_0
GSCN_1
Function–Tag register
3GIO Statistic Control Register #1
3GIO Statistic Control Registers #2
3GIO Statistic Control Register #3
3GIO Statistic Control Register #4
3GIO Statistic Counter Registers #0
3GIO Statistic Counter Registers #1
193
82583V GbE Controller—Driver Programing Interface
Alias
Link to
Category
PCIe
Offset
Abbreviation
Name
RW
Offset
Page
0x05B28
0x05B2C
0x05B50
0x05B64
N/A
GSCN_2
GSCN_3
SWSM
3GIO Statistic Counter Registers #2
3GIO Statistic Counter Registers #3
Software Semaphore Register
3GIO Control Register 2
RW
RW
RW
RW
page 219
page 220
page 220
page 220
PCIe
PCIe
PCIe
N/A
N/A
N/A
GCR2
RC/
WC
Interrupt
0x000C0
N/A
ICR
Interrupt Cause Read Register
page 224
Interrupt
Interrupt
Interrupt
Interrupt
Interrupt
Interrupt
0x000C4
0x000C8
0x000D0
0x000D8
0x000DC
0x000E0
N/A
N/A
N/A
N/A
N/A
N/A
ITR
Interrupt Throttling Register
Interrupt Cause Set Register
Interrupt Mask Set/Read Register
Interrupt Mask Clear Register
Interrupt Auto Clear
R/W
W
page 225
page 227
page 228
page 229
page 230
page 230
ICS
IMS
IMC
EIAC
IAM
RW
W
RW
RW
Interrupt Acknowledge Auto–Mask
Receive
Receive
Receive
Receive
0x00100
0x02170
0x02160
0x02168
N/A
RCTL
Receive Control Register
RW
RW
RW
RW
page 230
page 233
page 234
page 234
N/A
PSRCTL
FCRTL
FCRTH
Packet Split Receive Control Register
Flow Control Receive Threshold Low
Flow Control Receive Threshold High
0x00168
0x00160
Receive Descriptor Base Address Low
queue 0
Receive
Receive
0x02800
0x02804
0x00110
0x00114
RDBAL0
RDBAH0
RW
RW
page 235
page 235
Receive Descriptor Base Address High
queue 0
Receive
Receive
Receive
Receive
Receive
Receive
Receive
Receive
Receive
Receive
0x02808
0x02810
0x02818
0x02820
0x02828
0x0282C
0x02C00
0x02C08
0x05000
0x05008
0x00118
0x00120
0x00128
0x00108
N/A
RDLEN0
RDH0
Receive Descriptor Length queue 0
Receive Descriptor Head queue 0
Receive Descriptor Tail queue 0
Rx Interrupt Delay Timer [Packet Timer]
Receive Descriptor Control
RW
RW
RW
RW
RW
RW
R/W
RW
RW
RW
page 235
page 236
page 236
page 236
page 237
page 238
page 239
page 239
page 239
page 241
RDT0
RDTR
RXDCTL
RADV
N/A
Receive Interrupt Absolute Delay Timer
Receive Small Packet Detect Interrupt
Receive ACK Interrupt Delay Register
Receive Checksum Control
N/A
RSRPD
RAID
N/A
N/A
RXCSUM
RFCTL
N/A
Receive Filter Control Register
0x05200-
0x053FC
Receive
MTA[127:0]
Multicast Table Array
RW
page 241
Receive
Receive
Receive
Receive
0x05400
0x05404
0x05408
0x0540C
0x00040
0x00044
0x00048
0x0004C
RAL(0)
RAH(0)
RAL(1)
RAH(1)
Receive Address Low (0)
Receive Address High (0)
Receive Address Low (1)
Receive Address High (1)
RW
RW
RW
RW
page 242
page 242
page 242
page 242
0x05600-
0x057FC
0x00600-
0x007FC
Receive
VFTA[127:0]
VLAN Filter Table Array
RW
page 244
0x05600-
0x057FC
0x00600-
0x006FC
Receive
Receive
VFTA[127:0]
RAL(15)
VLAN Filter Table Array (n)
Receive Address Low (15)
RW
RW
page 244
page 242
0x05478
0x000B8
194
Driver Programing Interface—82583V GbE Controller
Alias
Link to
Page
Category
Offset
Abbreviation
Name
RW
RW
Offset
Receive
0x0547C
x000BC
RAH(15)
Receive Address High (15)
page 242
Transmit
Transmit
Transmit
Transmit
Transmit
Transmit
Transmit
Transmit
Transmit
Transmit
Transmit
Transmit
0x00400
0x00410
0x00458
0x03800
0x03804
0x03808
0x03810
0x03818
0x03840
0x03820
0x03828
0x0382C
N/A
TCTL
TIPG
Transmit Control Register
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
page 245
page 246
page 247
page 247
page 248
page 248
page 248
page 249
page 249
page 249
page 250
page 252
N/A
Transmit IPG Register
N/A
AIT
Adaptive IFS Throttle
0x00420
0x00424
0x00428
0x00430
0x00438
N/A
TDBAL
TDBAH
TDLEN
TDH
Transmit Descriptor Base Address Low
Transmit Descriptor Base Address High
Transmit Descriptor Length
Transmit Descriptor Head
TDT
Transmit Descriptor Tail
TARC
TIDV
Transmit Arbitration Count
Transmit Interrupt Delay Value
Transmit Descriptor Control
Transmit Absolute Interrupt Delay Value
0x00440
N/A
TXDCTL
TADV
N/A
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
0x04000
0x04004
0x0400C
0x04010
0x04014
0x04018
0x0401C
0x04020
0x04028
0x04030
0x04034
0x0403C
0x04040
0x04048
0x0404C
0x04050
0x04054
0x04058
0x0405C
0x04060
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
CRCERRS
ALGNERRC
RXERRC
MPC
CRC Error Count
R
page 253
page 253
page 253
page 253
page 254
page 254
page 254
page 254
page 255
page 255
page 255
page 256
page 256
page 256
page 256
page 257
page 257
page 257
page 257
page 258
Alignment Error Count
RX Error Count
R
R
Missed Packets Count
Single Collision Count
Excessive Collisions Count
Multiple Collision Count
Late Collisions Count
R
SCC
R
ECOL
R
MCC
R
LATECOL
COLC
R
Collision Count
R
DC
Defer Count
R
TNCRS
CEXTERR
RLEC
Transmit with No CRS
Carrier Extension Error Count
Receive Length Error Count
XON Received Count
R
R
R
XONRXC
XONTXC
XOFFRXC
XOFFTXC
FCRUC
PRC64
PRC127
R
XON Transmitted Count
XOFF Received Count
R
R
XOFF Transmitted Count
FC Received Unsupported Count
Packets Received [64 Bytes] Count
Packets Received [65–127 Bytes] Count
R
RW
RW
RW
Packets Received [128–255 Bytes]
Count
Statistic
Statistic
Statistic
0x04064
0x04068
0x0406C
N/A
N/A
N/A
PRC255
PRC511
PRC1023
RW
RW
RW
page 258
page 258
page 258
Packets Received [256–511 Bytes]
Count
Packets Received [512–1023 Bytes]
Count
195
82583V GbE Controller—Driver Programing Interface
Alias
Offset
Link to
Page
Category
Offset
Abbreviation
Name
RW
Packets Received [1024 to Max Bytes]
Count
Statistic
0x04070
N/A
PRC1522
RW
page 259
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
Statistic
0x04074
0x04078
0x0407C
0x04080
0x04088
0x0408C
0x04090
0x04094
0x040A0
0x040A4
0x040A8
0x040AC
0x040B0
0x040C0
0x040C4
0x040C8
0x040D0
0x040D4
0x040D8
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
GPRC
BPRC
MPRC
GPTC
GORCL
GORCH
GOTCL
GOTCH
RNBC
RUC
Good Packets Received Count
Broadcast Packets Received Count
Multicast Packets Received Count
Good Packets Transmitted Count
Good Octets Received Count Low
Good Octets Received Count High
Good Octets Transmitted Count Low
Good Octets Transmitted Count High
Receive No Buffers Count
R
page 259
page 259
page 260
page 260
page 260
page 260
page 261
page 261
page 261
page 261
page 262
page 262
page 262
page 263
page 263
page 263
page 264
page 264
page 264
R
R
R
R
R
R
R
R
Receive Undersize Count
R
RFC
Receive Fragment Count
R
ROC
Receive Oversize Count
R
RJC
Receive Jabber Count
R
TORL
TORH
TOT
Total Octets Received
R
Total Octets Received
R
Total Octets Transmitted
RW
RW
RW
RW
TPR
Total Packets Received
TPT
Total Packets Transmitted
PTC64
Packets Transmitted [64 Bytes] Count
Packets Transmitted [65–127 Bytes]
Count
Statistic
Statistic
Statistic
Statistic
Statistic
0x040DC
0x040E0
0x040E4
0x040E8
0x040EC
N/A
N/A
N/A
N/A
N/A
PTC127
PTC255
PTC511
PTC1023
PTC1522
RW
RW
RW
RW
RW
page 264
page 265
page 265
page 265
page 265
Packets Transmitted [128–255 Bytes]
Count
Packets Transmitted [256–511 Bytes]
Count
Packets Transmitted [512–1023 Bytes]
Count
Packets Transmitted [Greater than 1024
Bytes] Count
Statistic
Statistic
0x040F0
0x040F4
N/A
N/A
MPTC
BPTC
Multicast Packets Transmitted Count
Broadcast Packets Transmitted Count
RW
RW
page 266
page 266
TCP Segmentation Context Transmitted
Count
Statistic
0x040F8
N/A
TSCTC
RW
page 266
TCP Segmentation Context Transmit Fail
Count
Statistic
Statistic
0x040FC
0x04100
N/A
N/A
TSCTFC
IAC
RW
R
page 266
page 267
Interrupt Assertion Count
Diagnostic
Diagnostic
Diagnostic
Diagnostic
Diagnostic
0x00F10
0x02410
0x02418
0x02420
0x02428
N/A
POEMB
RDFH
PHY OEM Bits Register
RW
RW
RW
RW
RW
page 296
page 296
page 297
page 297
page 297
0x08000
0x08008
N/A
Receive Data FIFO Head Register
Receive Data FIFO Tail Register
Receive Data FIFO Head Saved Register
Receive Data FIFO Tail Saved Register
RDFT
RDFHS
RDFTS
N/A
196
Driver Programing Interface—82583V GbE Controller
Alias
Offset
Link to
Page
Category
Offset
Abbreviation
Name
RW
Diagnostic
Diagnostic
Diagnostic
Diagnostic
Diagnostic
Diagnostic
0x02430
0x03410
0x03418
0x03420
0x03428
0x03430
N/A
RDFPC
TDFH
Receive Data FIFO Packet Count
Transmit Data FIFO Head Register
Transmit Data FIFO Tail Register
Transmit Data FIFO Head Saved Register
Transmit Data FIFO Tail Saved Register
Transmit Data FIFO Packet Count
RW
RW
RW
RW
RW
RW
page 298
page 298
page 298
page 299
page 299
page 299
0x08010
0x08018
N/A
TDFT
TDFHS
TDFTS
TDFPC
N/A
N/A
0x10000 -
0x17FFF
Diagnostic
Diagnostic
N/A
N/A
PBM
PBS
Packet Buffer Memory
Packet Buffer Size
RW
RW
page 299
page 300
0x01008
Note:
Certain registers maintain an alias address designed for backward compatibility with
software written for previous devices. For these registers, the alias address is shown in
Table 49. Those registers can be accessed by software at either the new offset or the
alias offset. It is recommended that software written solely for the 82583V, use the
new address offset.
9.2.2
General Register Descriptions
9.2.2.1
Device Control Register - CTRL (0x00000 / 0x00004; RW)
Initial
Value
Field
Bit(s)
Description
Full Duplex
0b = Half duplex
1
FD
0
1b
1b = Full duplex. Controls the MAC duplex setting when explicitly set
by software.
Reserved
Write as 0b for future compatibility.
Reserved
1
2
0b
0b
When set, the 82583V blocks new master requests using this
function. Once no master requests are pending by this function, the
GIO Master Enable Status bit is set.
GIO Master
Disable
Reserved
Set to 1b.
Reserved
Reserved
3
4
1b
0b
Reserved
Write as 0b for future compatibility.
Auto-Speed Detection Enable
When set to 1b, the MAC ignores the speed indicated by the PHY and
attempts to automatically detect the resolved speed of the link and
configure itself appropriately.
1
ASDE
5
0b
This bit must be set to 0b in the 82583V.
Set Link Up
The Set Link Up bit MUST be set to 1b to permit the MAC to recognize
the link signal from the PHY, which indicates the PHY has gotten the
link up, and to receive and transmit data.
See Section 6.2.3 for more information about auto-negotiation and
link configuration in the various modes.
1
SLU
6
7
0b
Set link up is normally initialized to 0b. However, if the APM Enable bit
is set in the NVM then it is initialized to 1b.
Reserved.
Must be set to 0b.
Reserved
0b
197
82583V GbE Controller—Driver Programing Interface
Initial
Value
Field
Bit(s)
Description
Speed selection
These bits can determine the speed configuration and are written by
software after reading the PHY configuration through the MDIO
interface. These signals are ignored when Auto-Speed Detection is
enabled. See Section 6.2.1 for details.
SPEED
9:8
10b
00b = 10 Mb/s
01b = 100 Mb/s
10b = 1000 Mb/s
11b =not used
Reserved
Write as 0b for future compatibility.
Reserved
FRCSPD
10
11
0b
0b
Force Speed
This bit is set when software wants to manually configure the MAC
speed settings according to the Speed bits. When using a PHY device,
note that the PHY device must resolve to the same speed
configuration, or software must manually set it to the same speed as
the MAC. Note that this bit is superseded by the CTRL_EXT.SPD_BYPS
bit which has a similar function.
1
Force Duplex
When set to 1b, software might override the duplex indication from
the PHY that is indicated in the FDX to the MAC. Otherwise, the
duplex setting is sampled from the PHY FDX indication into the MAC
on the asserting edge of the PHY LINK signal. When asserted, the
CTRL.FD bit sets duplex.
FRCDPLX
Reserved
12
0b
Reserved
Reads as 0b.
19:13
0x0
1b
D3Cold WakeUp Capability Advertisement Enable
When set, D3Cold wakeup capability is advertised based on whether
the AUX_PWR advertises presence of auxiliary power (yes if
AUX_PWR is indicated, no otherwise). When 0b, however, D3Cold
wakeup capability is not advertised even if AUX_PWR presence is
indicated.
ADVD3WUC
20
Note: This bit must be set to 1b.
Reserved
RST
25:21
26
0x0
0b
Reserved
Device Reset
This bit performs a reset of the MAC function of the device, as
described in Section 9.2.2.2. Normally 0b; writing 1b initiates the
reset. This bit is self-clearing.
Receive Flow Control Enable
Indicates that the device responds to the reception of flow control
packets. Reception of flow control packets requires the correct loading
of the FCAL/H and FCT registers. If auto-negotiation is enabled, this
bit is set to the negotiated duplex value. See Section 6.2.3 for more
information about auto-negotiation.
RFCE
27
0b
Transmit Flow Control Enable
Indicates that the device transmits flow control packets (XON and
XOFF frames) based on receiver fullness. If auto-negotiation is
enabled, this bit is set to the negotiated duplex value. See
Section 6.2.3 for more information about auto-negotiation.
TFCE
28
29
0b
0b
Reserved
Reads as 0b.
Reserved
VLAN Mode Enable
When set to 1b, all packets transmitted from the 82583V that have
VLE set is sent with an 802.1Q header added to the packet. The
contents of the header come from the transmit descriptor and from
the VLAN type register. On receive, VLAN information is stripped from
802.1Q packets. See Section 7.5.1 for more details.
VME
30
0b
198
Driver Programing Interface—82583V GbE Controller
Initial
Value
Field
Bit(s)
Description
PHY Reset
Controls a hardware-level reset to the internal PHY.
0b = Normal (operational).
PHY_RST
31
0b
1b = Reset to PHY asserted.
1. These bits are read from the NVM.
This register, as well as the Extended Device Control (CTRL_EXT) register, controls the
major operational modes for the device. While a software write to this register to
control device settings, several bits (such as FD and Speed) might be overridden
depending on other bit settings and the resultant link configuration determined by the
PHY's auto-negotiation resolution. See Section 6.2.3 for a detailed explanation on the
link configuration process.
Note:
In half-duplex mode, the 82583V transmits carrier extended packets and can receive
both carrier extended packets and packets transmitted with bursting.
When using an internal PHY, the FD (duplex) and Speed configuration of the device is
normally determined from the link configuration process. Software can specifically
override/set these MAC settings via these bits in a forced-link scenario; if so, the values
used to configure the MAC must be consistent with the PHY settings.
Manual link configuration is controlled through the PHY's MII management interface.
The ADVD3WUC bit (Advertise D3Cold Wakeup Capability Enable control) enables the
AUX_PWR pin to determine whether D3Cold support is advertised. If full 1 Gb/s
operation in D3 state is desired but the system's power requirements in this mode
would exceed the D3Cold Wakeup-Enabled specification limit (375 mA at 3.3 V dc), this
bit can be used to prevent the capability from being advertised to the system.
When using the internal PHY, by default the PHY re-negotiates the lowest functional link
speed in D3 and D0u states. The PHYREG 25.2 bit enables this capability to be disabled,
in case full 1 Gb/s speed is desired in these states.
Note:
The 82583V internal PHY automatically detects an unplugged LAN cable and reduce
operational power to the minimal amount required to maintain system operation.
Controller operations are not affected, except for the inability to transmit/receive due
to the lost link.
Device Reset (RST) might be used to globally reset the entire component. This register
is provided primarily as a last-ditch software mechanism to recover from an
indeterminate or suspected hung hardware state. Most registers (receive, transmit,
interrupt, statistics, etc.), and state machines are set to their power-on reset values,
approximating the state following a power-on or PCI reset. However, PCIe configuration
registers are not reset, thereby leaving the device mapped into system memory space
and accessible by a software device driver. One internal configuration register, the
Packet Buffer Allocation (PBA) register, also retains its value through a global reset.
Note:
To ensure that global device reset has fully completed and that the 82583V responds to
subsequent accesses, designers must wait approximately 1 μs after resetting before
attempting to check to see if the bit has cleared or attempting to access (read or write)
any other device register.
Before issuing this reset, software has to insure that Tx and Rx processes are stopped
by following the procedure described in Section 6.1.3.10.
199
82583V GbE Controller—Driver Programing Interface
9.2.2.2
Device Status Register - STATUS (0x00008; R)
Initial
Value
Field
Bit(s)
Description
Full Duplex
FD
LU
0
1
X
X
0b = half duplex
1b = Full duplex. Reflects duplex setting of the MAC and/or link.
Link Up
0b = No link established
1b = Link established. For this to be valid, the Set Link Up bit of the
Device Control (CTRL.SU) register must be set.
Reserved
TXOFF
3:2
4
00b
X
Reserved
Transmission Paused
Indication of pause state of the transmit function when symmetrical
flow control is enabled.
Reserved
5
0b
Reserved
Link speed setting. Reflects speed setting of the MAC and/or link
00b = 10 Mb/s
SPEED
ASDV
7:6
9:8
X
X
01b = 100 Mb/s
10b = 1000 Mb/s
11b = 1000 Mb/s
Auto-Speed Detection Value
Speed result sensed by the MAC auto-detection function.
PHY Reset Asserted
This bit is read/write. Hardware sets this bit following the assertion of
PHY reset. The bit is cleared on writing 0b to it. This bit is used by
firmware as an indication for required initialization of the PHY.
PHYRA
10
1b
Reserved
18:11
19
0x0
1b
Reserved
Cleared by the 82583V when the GIO Master Disable bit is set and no
master requests are pending by this function. Set otherwise.
Indicates that no master requests is issued by this function as long as
the GIO Master Disable bit is set.
GIO Master
Enable Status
Reserved
Reads as 0b.
Reserved
Reserved
30:20
31
0x0
0b
Reserved
FD reflects the actual MAC duplex configuration. This normally reflects the duplex
setting for the entire link, as it normally reflects the duplex configuration negotiated
between the PHY and link partner (copper link) or MAC and link partner (fiber link).
Link up provides a useful indication of whether something is attached to the port.
Successful negotiation of features/link parameters results in link activity. The link start-
up process (and consequently the duration for this activity after reset) can be several
100's of μs. It reflects whether the PHY's LINK indication is present. Refer to
Section 6.2.3 for more details.
TXOFF indicates the state of the transmit function when symmetrical flow control has
been enabled and negotiated with the link partner. This bit is set to 1b when
transmission is paused due to the reception of an XOFF frame. It is cleared upon
expiration of the pause timer or the receipt of an XON frame.
200
Driver Programing Interface—82583V GbE Controller
Speed indicates the actual MAC speed configuration. These bits normally reflect the
speed of the actual link, negotiated by the PHY and link partner, and reflected internally
from the PHY to the MAC (SPD_IND). These bits might represent the speed
configuration of the MAC only, if the MAC speed setting has been forced via software
(CTRL.SPEED) or MAC auto-speed detection used. Speed indications are mapped as
follows:
00b = 10 Mb/s
01b = 100 Mb/s
10b = 1000 Mb/s
11b = 1000 Mb/s
If Auto-Speed Detection is enabled, the device's speed is configured only once after the
link signal is asserted by the PHY.
The ASDV bits are provided for diagnostics purposes only. Even if the MAC speed
configuration is not set using this function (ASDE=0b), the ASD calculation can be
initiated by software writing a logic one to the CTRL_EXT.ASDCHK bit. The resultant
speed detection is reflected in these bits.
9.2.2.3
EEPROM/FLASH Control Register - EEC (0x00010; RW/RO)
Initial
Value
Field
Bit(s)
Description
Clock input to the NVM
When EE_GNT is 1b, the EE_SK output signal is mapped to this bit
and provides the serial clock input to the NVM. Software clocks the
NVM via toggling this bit with successive writes.
EE_SK
0
0b
Chip select input to the NVM
When EE_GNT is 1b, the EE_CS output signal is mapped to the chip
select of the NVM device. Software enables the NVM by writing a 1b to
this bit.
EE_CS
EE_DI
EE_DO
1
2
3
0b
0b
X
Data input to the NVM
When EE_GNT is 1b, the EE_DI output signal is mapped directly to
this bit. Software provides data input to the NVM via writes to this bit.
Data output bit from the NVM
The EE_DO input signal is mapped directly to this bit in the register
and contains the NVM data output. This bit is read-only from the
software perspective – writes to this bit have no effect.
Flash Write Enable Control
These two bits control whether writes to the Flash are allowed.
00b = Enable Flash erase and block erase.
01b = Flash writes and Flash erase disabled.
10b = Flash writes enabled.
FWE
5:4
01b
11b = Not allowed.
This field enables write and erase instructions from software to the
Flash via the Flash BAR and the software DMA registers (FLSW).
Request NVM Access
Software must write a 1b to this bit to get direct NVM access. It has
access when EE_GNT is 1b. When software completes the access it
must write a 0b.
EE_REQ
EE_GNT
6
7
0b
0b
Grant NVM Access
When this bit is set to 1b, software can access the NVM using the SK,
CS, DI, and DO bits.
201
82583V GbE Controller—Driver Programing Interface
Initial
Value
Field
Bit(s)
Description
NVM Present
EE_PRES
8
9
X
Setting this bit to 1b indicates that an NVM (either Flash or EEPROM)
is present and has the correct signature field. This bit is read only.
NVM Auto Read Done
When set to 1b, this bit indicates that the auto read by hardware from
the NVM is done. This bit is set also when the NVM is not present or
when its signature is not valid.
Auto_RD
Reserved
NVSize
0b
This field is read only.
10
0b
Reserved
NVM Size
This field defines the size of the NVM:
This field defines the size of the NVM in bytes which equal 128 * 2 **
NVSize. This field is loaded from word 0x0F in the NVM.
1
14:11
0010b
This field is read only.
NVM Address Size
This field defines the address size of the NVM:
00b = Reserved.
01b = EEPROM with 1 address byte.
10b = EEPROM with 2 address bytes.
11b = Flash with 3 address bytes.
NVADDS
16:15
00b
This field is set at power up by the NVMT strapping pin. With the
EEPROM, the address length is set following a detection of the
signature bits in word 0x12. If an EEPROM is attached to the 82583V
and a valid signature is not found, software can modify this field
enabling parallel access to empty device. In all other cases writes to
this field do not affect the device operation
Reserved
Reserved
Reserved
17
18
19
0b
0b
0b
Reserved
Reserved
Reserved
Enable Autonomous Flash Update
1b = Enables the 82583V to update the Flash autonomously. The
autonomous update is triggered by write cycles and expiration of the
FLASHT timer.
AUPDEN
20
0b
0b = Disables the auto-update logic.
Reserved
SEC1VAL
21
22
0b
0b
Reserved
Sector 1 Valid
In case EE_PRES is set, a 0b indicates that S0 in the Flash contains
valid signatures. 1b indicates that S1 contains valid signatures. In
EEPROM setup or if EE_PRES is not set, the SEC1VAL is 0b.
This is a read-only field indicating the NVM type:
0b = EEPROM.
1b = Flash.
1
NVMTYPE
23
0b
This bit is loaded from NVM word 0x0F and is informational only (the
design uses strapping to determine the actual NVM type).
Reserved
Reserved
24
25
0b
0b
Reserved
Reserved
Reserved
Reads as 0b.
Reserved
31:26
0x0
1. These bits are read from the NVM.
This register provides software direct access to the NVM. Software can control the NVM
by successive writes to this register. Data and address information is clocked into the
EEPROM by software toggling the EE_SK bit of this register with EE_CS set to 1. Data
202
Driver Programing Interface—82583V GbE Controller
output from the NVM is latched into bit 3 of this register via the internal 62.5 MHz clock
and may be accessed by software via reads of this register. See Section 6.3.8 for
details.
Note:
Attempts to write to the Flash device when writes are disabled (FWE=01) should not be
attempted. Behavior after such an operation is undefined, and can result in component
and/or system hangs.
9.2.2.4
EEPROM Read Register - EERD (0x00014; RW)
Initial
Value
Field
Bit(s)
Description
Start Read
Writing a 1b to this bit causes the 82583V to read a 16-bit word at the
address stored in the ADDR field from the NVM. The result is stored in
the DATA field. This bit is self-clearing
START
0
1
0b
Read Done
DONE
1b
Set to 1b when the word read completes. Set to 0b when the read is
in progress. Writes by software are ignored.
Read Address
ADDR
DATA
15:2
0x0
0x0
This field is written by software along with Start Read to indicate the
word address of the word to read.
Read Data
Data returned from the NVM.
31:16
This register is used by software to cause the 82583V to read individual words in the
EEPROM. To read a word, software writes the address to the Read Address field and
simultaneously writes a 1b to the Start Read field. The 82583V reads the word from the
EEPROM and places it in the Read Data field, setting the Read Done field to 1b.
Software can poll this register, looking for a 1b in the Read Done field, and then using
the value in the Read Data field.
Note:
When this register is used to read a word from the EEPROM, that word is not written to
any of the 82583V's internal registers even if it is normally a hardware accessed word.
9.2.2.5
Extended Device Control Register - CTRL_EXT (0x00018; RW)
Initial
Field
Bit(s)
11:0
Description
Value
Reserved
0x0
Reserved.
ASD (Auto Speed Detection) Check
Initiate an ASD sequence to sense the frequency of the RX_CLK signal
from the PHY. The results are reflected in STATUS.ASDV. This bit is
self-clearing.
ASDCHK
12
0b
EEPROM Reset
Initiates a reset-like event to the EEPROM function. This causes the
EEPROM to be read as if a PCI_RST_N assertion had occurred.
Note: All device functions should be disabled prior to setting this bit.
This bit is self-clearing.
EE_RST
13
14
0b
0b
Reserved
Should be set to 0b.
1
Reserved
203
82583V GbE Controller—Driver Programing Interface
Initial
Value
Field
SPD_BYPS
Reserved
Bit(s)
Description
Speed Select Bypass
When set to 1b, all speed detection mechanisms are bypassed and
the device is immediately set to the speed indicated by CTRL.SPEED.
This provides a method for software to have full control of the speed
settings of the device as well as when the change takes place by
overriding the hardware clock switching circuitry.
15
0b
Reserved
Should be set to 0b.
1
16
17
0b
0b
Relaxed Ordering Disable
When set to 1b, the device does not request any relaxed ordering
transactions regardless of the state of bit 4 (Enable Relaxed Ordering)
in the PCIe Device Control register. When this bit is cleared and bit 4
of the PCIe Device Control register is set, the device requests relaxed
ordering transactions as described in Section 6.1.3.8.2.
RO_DIS
Reserved
18
19
0b
0b
Reserved
DMA Dynamic
Gating Enable
When set, this bit enables dynamic clock gating of the DMA and MAC
units.
1
1
PHY Power
Down Enable
20
1b
When set, this bit enables the PHY to enter a low-power state.
1
1
1
Reserved
Tx LS Flow
Tx LS
21
22
23
0b
0b
0b
Reserved
Should be set for correct TSO functionality. Refer to Section 7.3.
Should be cleared for correct TSO functionality. Refer to Section 7.3.
Extended Interrupt Auto Mask Enable
EIAM is used only upon a read of the EICR register.
EIAME
24
0b
Reserved
26:25
00b
0b
Reserved
When the IAME (interrupt acknowledge auto-mask enable) bit is set,
a read or write to the ICR register has the side effect of writing the
value in the IAM register to the IMC register. When this bit is 0b, the
feature is disabled.
IAME
27
28
Driver Loaded
This bit should be set by the software device driver after it was
loaded, Cleared when the software device driver unloads or PCIe soft
reset.
DRV_LOAD
0b
0b
When set, this bit enables the clearing of the interrupt timers
following an IMS clear. In this state, successive interrupts occur only
after the timers expire again. When cleared, successive interrupts
following IMS clear might happen immediately.
INT_TIMERS_
CLEAR_ENA
29
30
Reserved
Reads as 0b.
Reserved
0b
0b
PBA Support
PBA_Supportr 31
The 82583V behaves in a way supporting legacy INT-x interrupts.
Should be cleared when working in INT-x or MSI mode.
1. These bits are read from the NVM.
204
Driver Programing Interface—82583V GbE Controller
This register provides extended control of device functionality beyond that provided by
the Device Control (CTRL) register.
Note:
Note:
Note:
Device Control register values are changed by a read of the EEPROM which occurs upon
assertion of the EE_RST bit. Therefore, if software uses the EE_RST function and
desires to retain current configuration information, the contents of the control registers
should be read and stored by software.
The EEPROM reset function might read configuration information out of the EEPROM
which affects the configuration of PCIe configuration space BAR settings. The changes
to the BARs are not visible unless the system is rebooted and the BIOS is allowed to re-
map them.
The SPD_BYPS bit performs a similar function to the CTRL.FRCSPD bit in that the
device's speed settings are determined by the value software writes to the CRTL.SPEED
bits. However, with the SPD_BYPS bit asserted, the settings in CTRL.SPEED take effect
rather than waiting until after the device's clock switching circuitry performs the
change.
9.2.2.6
Flash Access Register - FLA (0x0001C; RW)
Initial
Value
Field
Bit(s)
Description
Clock input to the FLASH
When FL_GNT is 1, the FL_NVM_SK output signal is mapped to
this bit and provides the serial clock input to the Flash. Software
clocks the Flash via toggling this bit with successive writes.
FL_NVM_SK
0
1
2
0b
Chip select input to the FLASH
When FL_GNT is 1, the FL_CE output signal is mapped to the chip
select of the FLASH device. Software enables the FLASH by
writing a 0 to this bit.
FL_CE
FL_SI
0b
0b
Data input to the FLASH
When FL_GNT is 1, the FL_SI output signal is mapped directly to
this bit. Software provides data input to the FLASH via writes to
this bit.
Data output bit from the FLASH
The FL_SO input signal is mapped directly to this bit in the
register and contains the Flash serial data output. This bit is read-
only from the software perspective – writes to this bit have no
effect.
1
FL_SO
3
X
Request FLASH Access
The software must write a 1 to this bit to get direct Flash access.
It has access when FL_GNT is 1. When the software completes
the access it must write a 0.
FL_REQ
FL_GNT
4
5
0b
0b
Grant FLASH Access
When this bit is set to 1b, the software can access the Flash using
the SK, CS, DI, and DO bits.
Reserved
8:6
9
000b
1b
Reserved
Status Bit
SW_WR_DONE
Reserved
Indicates that last LAN_BAR or LAN_EXP write was done.
10
1b
Reserved
Reserved
Reads as 0b.
Reserved
29:11
0x0
205
82583V GbE Controller—Driver Programing Interface
Initial
Value
Field
Bit(s)
Description
Flash Busy
This bit is set to 1b while a transaction to the Flash is in progress.
While this bit is clear (read as 0b), software can access the Flash.
FL_BUSY
30
0b
This field is read only.
Flash Erase Command
The command is sent to the Flash only if bits 5:4 in the EEC
register are set to 00b. This bit is auto-cleared and read as 0b.
FL_ER
31
0b
Certain Flash vendors do not support this operation.
Note:
This register provides the software with direct access to the Flash. Software can control
the Flash by successive writes to this register. Data and address information is clocked
into the Flash by software toggling the FL_NVM_SK bit (0) of this register with FL_CE
set to 1. Data output from the Flash is latched into bit 3 of this register via the internal
125 MHz clock and may be accessed by software via reads of this register.
Note:
In the 82583V, the FLA register is only reset at Internal Power On Reset and not as
legacy devices at a software reset.
9.2.2.7
MDI Control Register - MDIC (0x00020; RW)
Initial
Value
Field
Bit(s)
Description
Data
In a Write command, software places the data bits and the MAC shifts
them out to the PHY. In a Read command, the MAC reads these bits
serially from the PHY and software can read them from this location.
DATA
15:0
X
REGADD
PHYADD
20:16
25:21
0x0
0x0
PHY register address; i.e., Reg 0, 1, 2, … 31.
PHY Address
1 = Gigabit PHY.
2 = PCIe PHY.
Op-Code
01b = MDI write.
10b = MDI read.
Other values are reserved.
OP
27:26
0x0
Ready Bit
Set to 1b by the 82583V at the end of the MDI transaction (for
example, indicates a read or write has been completed). It should be
reset to 0b by software at the same time the command is written.
R
I
28
29
1b
0b
Interrupt Enable
When set to 1b by software, it causes an Interrupt to be asserted to
indicate the end of an MDI cycle.
Error
This bit set is to 1b by hardware when it fails to complete an MDI
read. Software should make sure this bit is clear (0b) before making
an MDI Read or Write command.
E
30
31
0b
0b
Reserved
Reserved. Write as 0b for future compatibility.
This register is used by software to read or write Management Data Interface (MDI)
registers in a GMII/MII PHY.
For an MDI read cycle the sequence of events is as follows:
1. The CPU performs a PCIe write cycle to the MII register with:
206
Driver Programing Interface—82583V GbE Controller
a. Ready = 0b.
b. Interrupt Enable bit set to 1b or 0b.
c. Op-Code = 10b (read).
d. PHYADD = PHY address from the MDI register.
e. REGADD = Register address of the specific register to be accessed (0 through
31).
2. The MAC applies the following sequence on the MDIO signal to the PHY:
<PREAMBLE><01><10><PHYADD><REGADD><Z> where the Z stands for the
MAC tri-stating the MDIO signal.
3. The PHY returns the following sequence on the MDIO signal: <0><DATA><IDLE>.
4. The MAC discards the leading bit and places the following 16 data bits in the MII
register.
5. The 82583V asserts an interrupt indicating MDI done if the Interrupt Enable bit was
set.
6. The 82583V sets the Ready bit in the MII register indicating the read is complete.
7. The CPU might read the data from the MII register and issue a new MDI command.
For an MDI write cycle, the sequence of events is as follows:
1. The CPU performs a PCIe write cycle to the MII register with:
a. Ready = 0b.
b. Interrupt Enable bit set to 1b or 0b.
c. Op-Code = 01b (write).
d. PHYADD = PHY address from the MDI register.
e. REGADD = Register address of the specific register to be accessed (0 through
31).
f. Data = Specific data for desired control of the PHY.
2. The MAC applies the following sequence on the MDIO signal to the PHY:
<PREAMBLE><01><01><PHYADD><REGADD><10><DATA><IDLE>.
3. The 82583V asserts an interrupt indicating MDI done if the Interrupt Enable bit was
set.
4. The 82583V sets the Ready bit in the MII register to indicate step 2 has been
completed.
5. The CPU might issue a new MDI command.
Note:
An MDI read or write might take as long as 64 μs from the CPU write to the Ready bit
assertion.
If an invalid op-code is written by software, the MAC does not execute any accesses to
the PHY registers.
If the PHY does not generate a zero as the second bit of the turn-around cycle for
reads, the MAC aborts the access, sets the E (error) bit, writes 0xFFFF to the data field
to indicate an error condition, and sets the Ready bit.
207
82583V GbE Controller—Driver Programing Interface
9.2.2.8
Flow Control Address Low - FCAL (0x00028; RW)
Initial
Value
Field
FCAL
Bit(s)
31:0
Description
X
Flow Control Address Low
Flow control packets are defined by 802.3X to be either a unique multicast address or
the station address with the EtherType field indicating pause. Hardware compares
incoming packets against the FCA register value to determine if it should pause its
output.
This register contains the lower bits of the internal 48-bit flow control Ethernet address.
All 32 bits are valid. Software can access the High and Low registers as a register pair if
it can perform a 64-bit access to the PCIe bus. This register should be programmed
with 0x00_C2_80_01. The complete flow control multicast address is:
0x01_80_C2_00_00_01; where 01 is the first byte on the wire, 80 is the second, etc.
Note:
Any packet matching the contents of {FCAH, FCAL, FCT} when CTRL.RFCE is set is
acted on by the 82583V. Whether flow control packets are passed to the host
(software) depends on the state of the RCTL.DPF bit and whether the packet matches
any of the normal filters.
9.2.2.9
Flow Control Address High - FCAH (0x0002C; RW)
Initial
Field
FCAH
Reserved
Bit(s)
15:0
31:16
Description
Value
X
Flow Control Address High
Reserved
Reads as 0x0.
0x0
This register contains the upper bits of the 48-bit flow control Ethernet address. Only
the lower 16 bits of this register have meaning. The complete flow control address is
{FCAH, FCAL}. This register should be programmed with 0x01_00. The complete flow
control multicast address is: 0x01_80_C2_00_00_01; where 01 is the first byte on the
wire, 80 is the second, etc.
Note:
At the time of the original implementation, the flow control multicast address was not
defined and thus hardware provided programmability. Since then, the final release of
the 802.3x standard has reserved the following multicast address for MAC control
frames: 0x01-80-C2-00-00-01.
208
Driver Programing Interface—82583V GbE Controller
9.2.2.10
Flow Control Type - FCT (0x00030; RW)
Initial
Value
Field
Bit(s)
15:0
31:16
Description
FCT
Reserved
X
Flow Control Type
Reserved
Reads as 0x0
0x0
This register contains the type field hardware uses to recognize a flow control packet.
Only the lower 16 bits of this register have meaning. This register should be
programmed with 0x88_08. The upper byte is first on the wire FCT[15:8].
Note:
At the time of the original implementation, the flow control type field was not defined
and thus hardware provided programmability. Since then, the final release of the
802.3x standard has specified the type/length value for MAC control frames as 88-08.
9.2.2.11
VLAN Ether Type - VET (0x00038; RW)
Initial
Field
Bit(s)
15:0
31:16
Description
Value
VET
Reserved
0x8100
VLAN Ether Type
Reserved
Reads as 0x0.
0x0
This register contains the type field hardware uses to recognize an 802.1Q (VLAN)
Ethernet packet. To be compliant with the 802.3ac standard, this register should be
programmed with the value 0x8100. For VLAN transmission the upper byte is first on
the wire (VET[15:8]).
9.2.2.12
Flow Control Transmit Timer Value - FCTTV (0x00170; RW)
Initial
Value
Field
Bit(s)
Description
Transmit Timer Value
Included in XOFF frame.
TTV
15:0
X
Reads as 0x0.
Should be written to 0x0 for future compatibility.
Reserved
31:16
0x0
The 16-bit value in the TTV field is inserted into a transmitted frame (either XOFF
frames or any pause frame value in any software transmitted packets). It counts in
units of slot time. If software needs to send an XON frame, it must set TTV to 0b prior
to initiating the pause frame.
Note:
The 82583V uses a fixed slot time value of 64-byte times.
209
82583V GbE Controller—Driver Programing Interface
9.2.2.13
Flow Control Refresh Threshold Value - FCRTV (0x05F40; RW)
Bit
Type
Reset
Description
Flow Control Refresh Threshold (FCRT)
This value indicates the threshold value of the flow control shadow
counter. When the counter reaches this value, and the conditions for a
pause state are still valid (buffer fullness above low threshold value), a
pause (XOFF) frame is sent to the link partner.
The FCRTV timer count interval is the same as other flow control timers
and counts at slot times of 64-byte times.
15:0
RW
RO
X
If this field contains a zero value, the Flow Control Refresh is disabled.
Reserved
Reads as 0x0.
31:16
0x0
Should be written to 0x0 for future compatibility.
9.2.2.14
LED Control - LEDCTL (0x00E00; RW)
Initial
Field
LED0_MODE
Reserved
Bit(s)
Description
Value
LED0 (LINK_UP_N) Mode
1
3:0
0010b
This field specifies the control source for the LED0 output. An initial
value of 0010b selects LINK_UP indication.
Reserved
4
5
0b
0b
Read-only as 0b. Write as 0b for future compatibility.
Global Blink Mode
This field specifies the blink mode of all LEDs.
0b = Blink at 200 ms on and 200 ms off.
1b = Blink at 83 ms on and 83 ms off.
GLOBAL_
BLINK_MODE
1
1
LED0 (LINK_UP_N) Invert
This field specifies the polarity/ inversion of the LED source prior to
output or blink control.
0b = Do not invert LED source.
1b = Invert LED source.
LED0_IVRT
6
7
0b
0b
LED0 (LINK_UP_N) Blink
This field specifies whether to apply blink logic to the (inverted) LED
control source prior to the LED output.
1
LED0_BLINK
0b = do not blink asserted LED output.
1b = blink asserted LED output.
LED1 (ACTIVITY_N) Mode
This field specifies the control source for the LED1 output. An initial
value of 0011b selects ACTIVITY indication.
1
LED1_MODE
Reserved
11:8
12
0011b
Reserved
0b
Read-only as 0b. Write as 0 for future compatibility.
LED1 (ACTIVITY_N) Blink Mode
This field needs to be configured with the same value as
GLOBAL_BLINK_MODE, it specifies the blink mode of the LED.
0b = Blink at 200 ms on and 200 ms off.
1b = Blink at 83 ms on and 83 ms off.
LED1_BLINK_
MODE
1
13
0b
1
LED1_IVRT
14
15
0b
LED1 (ACTIVITY_N) Invert.
LED1 (ACTIVITY_N) Blink
1
LED1_BLINK
1b
210
Driver Programing Interface—82583V GbE Controller
Initial
Field
LED2_MODE
Reserved
Bit(s)
Description
Value
LED2 (LINK_100_N) Mode
1
19:16
0110b
This field specifies the control source for the LED2 output. An initial
value of 0110b selects LINK_100 indication.
Reserved
20
21
0b
0b
Read-only as 0b. Write as 0b for future compatibility.
LED2 (LINK_100_N) Blink Mode
This field needs to be configured with the same value as
GLOBAL_BLINK_MODE, it specifies the blink mode of the LED.
0b = Blink at 200 ms on and 200 ms off.
1b = Blink at 83 ms on and 83 ms off.
LED2_BLINK_
MODE
1
1
1
LED2_IVRT
LED2_BLINK
Reserved
22
0b
0b
LED2 (LINK_100_N) Invert.
LED2 (LINK_100_N) Blink
Reserved
23
31:24
0x0
1. These bits are read from the NVM.
The following mapping is used to specify the LED control source (MODE) for each LED
output:
MODE
Selected Mode
LINK_10/1000
Source Indication
Asserted when either 10 or 1000 Mb/s link is established
and maintained.
0000
0001
0010
0011
0100
0101
0110
Asserted when either 100 or 1000 Mb/s link is
established and maintained.
LINK_100/1000
LINK_UP
Asserted when any speed link is established and
maintained.
Asserted when link is established and packets are being
transmitted or received that passed MAC filtering.
FILTER_ACTIVITY
LINK/ACTIVITY
LINK_10
Asserted when link is established AND when there is NO
transmit or receive activity.
Asserted when a 10 Mb/s link is established and
maintained.
Asserted when a 100 Mb/s link is established and
maintained.
LINK_100
Asserted when a 1000 Mb/s link is established and
maintained.
0111
1000
1001
1010
1011
LINK_1000
Reserved
Reserved
Asserted when the link is configured for full-duplex
operation.
FULL_DUPLEX
COLLISION
ACTIVITY
Asserted when a collision is observed.
Asserted when link is established and packets are being
transmitted or received.
Asserted when the device detects a 1-lane PCIe
connection.
1100
BUS_SIZE
1101
1110
1111
PAUSED
LED_ON
LED_OFF
Asserted when the device’s transmitter is flow controlled.
Always asserted.
Always de-asserted.
211
82583V GbE Controller—Driver Programing Interface
Notes:
1. When LED blink mode is enabled the appropriate LED Invert bit should be set to
zero.
2. The dynamic Leds modes (FILTER_ACTIVITY, LINK/ACTIVITY, COLLISION,
ACTIVITY, PAUSED) should be used with LED blink mode enabled.
3. When LED blink mode is enabled and CCM PLL is shut, the blinking frequencies are
1/5 of the rates stated in the previous table.
9.2.2.15
Extended Configuration Control - EXTCNF_CTRL (0x00F00; RW)
Initial
Value
Field
Bit(s)
31:28
Description
Reserved
Reserved
Reserved
0b
Reserved
Reserved
Reserved
27:16
15:8
0x0
0x0
MDIO MNG Ownership: Management request for access to MDIO. This
is part of the semaphore scheme for MDIO access (Section 4.6.2).
This is a RO bit.
Reserved
Reserved
Reserved
7
6
5
0b
0b
0b
Note: Use of this register is optional for the 82583V.
MDIO/NVM HW Ownership: Hardware request fo raccess to MDIO/
EEPROM. This is part of the semaphore scheme for MDIO access
(Section 4.6.2). This is a RO bit.
Note: Use of this register is optional for the 82583V.
MDIO/NVM SW Ownership: Software request fo raccess to MDIO/
EEPROM. This is part of the semaphore scheme for MDIO access
(Section 4.6.2). This is a RO bit.
Note: Use of this register is optional for the 82583V.
Reserved
Reserved
Reserved
Reserved
Reserved
4
3
2
1
0
0b
1b
0b
0b
0b
Reserved
Reserved
Reserved
Reserved
Should be set to 0b.
9.2.2.16
Extended Configuration Size - EXTCNF_SIZE (0x00F08; RW)
Initial
Value
Field
Bit(s)
Description
Reserved
Reserved
31:8
7:0
0x0
0x0
Reserved
Reserved
212
Driver Programing Interface—82583V GbE Controller
9.2.2.17
Packet Buffer Allocation - PBA (0x01000; RW)
Initial
Value
Field
Bit(s)
Description
Receive packet buffer allocation in KB. Upper 10 bits are read only as
0x0. Default is 20 KB.
RXA
TXA
15:0
31:16
0x0014
0x0014
Transmit packet buffer allocation in KB. These bits are read only.
Default is 20 KB.
This register sets the on-chip receive and transmit storage allocation ratio. The receive
allocation value is read/write for the lower 6 bits. The transmit allocation is read only
and is calculated based on RXA. The partitioning size is 1 KB.
Note:
Note:
Programming this register does not automatically re-load or initialize internal packet-
buffer RAM pointers. Software must reset both transmit and receive operation (using
the global device reset CTRL.RST bit) after changing this register in order for it to take
effect. The PBA register itself is not reset by asserting the global reset, but is only reset
upon initial hardware power on.
For best performance the transmit buffer allocation should be set to accept two full
sized packets.
Note:
Note:
Transmit packet buffer size should be configured to be more than 4 KB.
The 82583V supports a maximum frame size of 1514 bytes.
9.2.2.18
EEPROM Control Register - EECTL (0x1010; RO)
Initial
Value
Field
Bit(s)
Description
Address
ADDR
14:0
0x0
This field is written by Start Read or Start write to indicate the
EEPROM word address to read or write.
Start
START
WRITE
15
16
0b
0b
Writing a 1b to this bit causes the EEPROM to start the read or write
operation according to the write bit.
Write
This bit tells the EEPROM if the current operation is read or write.
0b = Read.
1b = Write.
EPROM Busy
EEBUSY
17
0b
This bit indicates that the EEPROM is busy doing an auto read.
Reserved
EE_TRANS_E
Reserved
18
0b
Reserved
Transaction
19
0b
This bit indicates that the register is in the middle of a transaction.
31:20
0x0
Reserved
Note:
This register is read/write by firmware and read only by software.
213
82583V GbE Controller—Driver Programing Interface
9.2.2.19
EEPROM Read/Write Data - EEDATA (0x1014; RO)
Initial
Value
Field
Bit(s)
Description
Write Data
Data to be written to the EEPROM.
WRDATA
15:0
0x0
Read Data
Data returned from the EEPROM read.
RDDATA
31:16
X
Note:
This register is read/write by firmware and read only by software.
9.2.2.20
Flash Control Register - FLCTL (0x1018; RO)
Note:
This register is Read-Write by FW and Read-Only by SW.
9.2.2.21
Flash Read Data - FLDATA (0x101C; RO)
Note:
This register is Read-Write by FW and Read-Only by SW.
9.2.2.22
Flash Read Counter - FLCNT (0x1020; RO)
Note:
This register is Read-Write by FW and Read-Only by SW.
9.2.2.23
Flash Timer Register- FLASHT (0x01028; RW)
Field
Bit(s)
Default
Description
Auto Flash Update Timer
Defines the idle time from the last write until the 82583V
autonomously updates the Flash. The time is measured in FLASHT.FLT
x 1024 cycles at 62.5 MHz (or 12.5 MHz when the 125 MHz clock is
gated). A value of 0x00 means that the update is not delayed.
FLT
15:0
0x2
The update timer is enabled by the Aupden bit in the EEC register.
Reserve
31:16
0x00
Reserved
214
Driver Programing Interface—82583V GbE Controller
9.2.2.24
EEPROM Write Register - EEWR (0x0102C; RW)
Field
Bit(s)
Default
Description
Start Write
Writing a 1b to this bit causes the 82583V to write a 16-bit word at
the address stored in the ADDR field in the external NVM. The data is
fetched from the DATA field. This bit is self-clearing.
START
0
1
0b
Write Done
DONE
1b
Set to 1b when the write completes. Set to 0b when the write is in
progress. Writes by software are ignored.
Write Address
ADDR
DATA
15:2
0x0
0x0
This field is written by software along with Start Write to indicate the
word address of the word to read.
Write Data
Data written to the NVM.
31:16
Note:
EEWR has direct access regardless of a valid signature in the NVM.
9.2.2.25
SW FLASH Burst Control Register - FLSWCTL (0x1030; RW)
Field
Bit(s)
Default
Description
Address
ADDR
23:0
0x0
This field is written by software along with Start Read or Start write to
indicate the Flash address to read or write.
Command
Indicates which command should be executed. Valid only when the
CMDV bit is set.
00b = Reserved.
01b = DMA Write command (write up to 256 bytes).
10b = Reserved.
CMD
25:24
00b
0b
11b = Reserved.
Command Valid
When set, indicates that software issues a new command.
Cleared by hardware at the end of the command.
CMDV
26
Flash Busy
FLBUSY
27
28
29
0b
0b
0b
This bit indicates that the Flash is busy processing a Flash transaction
and should not be accessed.
Reserved
FLUDONE
Reserved
Flash Update Done
This bit is set by the 82583V when it completes updating the Flash.
Software should clear it to zero before it updates the Flash.
Write Done
This bit clears after CMDV is set by software and is set back again
DONE
30
31
1b
1b
when the Flash write transaction is done.
When writing a burst transaction the bit is cleared every time
software writes FLSWDATA.
Global Done
This bit clears after the CMDV bit is set by software and is set back
again when the all Flash read/write transactions complete. For
example, the Flash unit finished to read/write all the requested read/
writes.
WRDONE
215
82583V GbE Controller—Driver Programing Interface
9.2.2.26
9.2.2.27
Software Flash Burst Data Register - FLSWDATA (0x1034; RW)
Field
NVDATA
Bit(s)
31:0
Default
0x0
Description
Write NVM Data
Data written to the NVM.
Software Flash Burst Access Counter - FLSWCNT (0x1038; RW)
Field
Bit(s)
Default
Description
Abort
Writing a 1b to this bit aborts the current burst operation. It is self-
cleared by the Flash interface block when the Abort command has
been executed. Abort request is not permitted after writing the last
Dword.
Abort
31
0b
Reserved
NVCNT
30:25
24:0
0x0
0x0
Reserved
NVM Counter
This counter holds the size of the Flash burst read or write in Dwords
and is also used as the write byte count but in this case it is byte
count.
9.2.2.28
9.2.2.29
Flash Opcode Register - FLOP (0x0103C; RW)
This register is used by the 82583V to initiate the appropriate instructions to the NVM
device.
FEEP Auto Load - FLOL (0x01050; RW)
Field
Bit(s)
Default
1b
Description
RAM_PWR_
SAVE_EN
When set to 1b, enables reduced power consumption by clock gating
the 82583V RAMs.
0
Reserved
Reserve
7:1
0x0
0x0
Auto loaded from NVM 0x11 bits 7:1.
Reserved
31:8
9.2.3
PCIe Register Descriptions
9.2.3.1
3GIO Control Register - GCR (0x05B00; RW)
Initial
Value
Field
Bit(s)
Description
Disable_
timeout_
31
0b
If set, the PCIe time-out mechanism is disabled.
mechanism
Self_test_
result
30
0b
0b
0b
If set, a self-test result finished successfully.
Force good PCIe L0s training.
Gio_good_l0s 29
Gio_dis_rd_
err
28
Disable running disparity error of PCIe 108b decoders.
216
Driver Programing Interface—82583V GbE Controller
Initial
Field
L1_act_
without_L0s_
rx
Bit(s)
Description
Value
If set, enables the device to enter ASPM L1 active without any
correlation to L0s_rx.
27
0b
Determines the idle time of the PCIe link in L0s state before initiating
a transition to L1 state. The initial value is loaded from NVM.
00b = 64 μs
01b = 256 μs
10b = 1 ms
11b = 4 ms
L1_Entry_
Latency (LSB)
26:25
24
11b
0b
(Read Only)
L0s Entry Latency
Set to 0b to indicate L0s entry latency is the same as L0s exit latency.
Set to 1b to indicate L0s entry latency is (L0s exit latency/4).
L0S_ENTRY_
LAT
Latency
000b = 2 μs.
001b = 8 μs.
010b = 1 6μs.
011b = 32 μs.
100b = 64 μs.
101b = 25 6μs.
110b = 1 ms.
111b = 4 ms (default).
L1_Entry_
Latency (MSB)
(Read Only)
23
1b
Reserved
Reserved
22
21
0b
0b
For proper operation, must be set to 1b by software during
initialization.
Header_log_
order
When set, indicates a need to change the order of the header log in
the error reporting registers.
Reserved
Reserved
20
0b
Reserved
Reserved
19:10
0x0
When set to 1b the reply-timer always adds the required L0s
adjustment. When cleared to 0b the adjustment is added only when
Tx L0s is active.
Rx_L0s_
Adjustment
9
1b
Reserved
8:6
5
0b
0b
Reserved
Transmit Descriptor Read – No Snoop Indication.
Read directly by transaction layer.
TXDSCR_
NOSNOOP
Transmit Descriptor Write – No Snoop Indication.
Read directly by transaction layer.
TXDSCW_
NOSNOOP
4
3
2
1
0
0b
0b
0
Transmit Data Read – No Snoop Indication.
Read directly by transaction layer.
TXD_
NOSNOOP
Receive Descriptor Read – No snoop indication.
Read directly by transaction layer.
RXDSCR_
NOSNOOP
Receive Descriptor Write – No Snoop Indication
Read directly by transaction layer.
RXDSCW_
NOSNOOP
0b
0b
Receive Data Write – No Snoop Indication
Read directly by transaction layer.
RXD_
NOSNOOP
217
82583V GbE Controller—Driver Programing Interface
9.2.3.2
Function–Tag Register - FUNCTAG (0x05B08; RW)
Initial
Value
Field
Bit(s)
31:29
Description
cnt_3_tag
cnt_3_func
cnt_2_tag
cnt_2_func
cnt_1_tag
cnt_1_func
cnt_0_tag
cnt_0_func
0x0
Tag number for event 6/1D, if located in counter 3.
Function number for event 6/1D, if located in counter 3.
Tag number for event 6/1D, if located in counter 2.
Function number for event 6/1D, if located in counter 2.
Tag number for event 6/1D, if located in counter 1.
Function number for event 6/1D, if located in counter 1.
Tag number for event 6/1D, if located in counter 0.
Function number for event 6/1D, if located in counter 0.
28:24
23:29
20:16
15:13
12:8
7:5
0x0
0x0
0x0
0x0
0x0
0x0
0x0
4:0
9.2.3.3
3GIO Statistic Control Register #1 - GSCL_1 (0x05B10; RW)
Initial
Value
Field
Bit(s)
Description
GIO_COUNT_
START
31
0b
Start indication of 3GIO statistic counters.
Stop indication of 3GIO statistic counters.
GIO_COUNT_
STOP
30
29
28
0b
0b
0b
GIO_COUNT_
RESET
Reset indication of 3GIO statistic counters.
Enable two 64-bit counters instead of four 32-bit counters.
GIO_64_BIT_
EN
Test Bit
GIO_COUNT_
TEST
27
0b
Forward counters for testability.
RESERVED
26:4
3
0x0
0b
Reserved
GIO_COUNT_
EN_3
Enable 3GIO statistic counter number 3.
GIO_COUNT_
EN_2
2
1
0
0b
0b
0b
Enable 3GIO statistic counter number 2.
Enable 3GIO statistic counter number 1.
Enable 3GIO statistic counter number 0.
GIO_COUNT_
EN_1
GIO_COUNT_
EN_0
218
Driver Programing Interface—82583V GbE Controller
9.2.3.4
3GIO Statistic Control Registers #2- GSCL_2 (0x05B14; RW)
Initial
Value
Field
Bit(s)
Description
GIO_EVENT_
NUM_3
31:24
0x0
The event number that counter 3 counts
The event number that counter counts
The event number that counter counts
The event number that counter counts
GIO_EVENT_
NUM_2
23:16
15:8
7:0
0x0
0x0
0x0
GIO_EVENT_
NUM_1
GIO_EVENT_
NUM_0
This counter contains the mapping of the event (which counter counts what event).
9.2.3.5
3GIO Statistic Control Register #3 - GSCL_3 (0x05B18; RW)
Initial
Field
Bit(s)
Description
Threshold of flow control credits.
Value
GIO_FC_TH_0 11:0
RESERVED 15:12
GIO_FC_TH_1 27:16
RESERVED 31:28
0x0
Optional values: 0 = (256-1).
0x0
0x0
0x0
Reserved
Threshold of flow control credits.
Optional values: 0 = (256-1).
Reserved
This counter holds the threshold values needed for some of the event counting. Note
that the event increases only after the value passes the threshold boundary.
9.2.3.6
3GIO Statistic Control Register #4 - GSCL_4 (0x05B1C; RW)
Initial
Field
Bit(s)
31:16
Description
Value
RESERVED
0x0
Reserved
GIO_RB_TH
15:10
9:0
0x0
0x0
Retry buffer threshold.
HOST_COML_
TH
Completions latency threshold.
This counter holds the threshold values needed for some of the event counting. Note
that the event increases only after the value passes the threshold boundary.
9.2.3.7
9.2.3.8
9.2.3.9
3GIO Statistic Counter Registers #0 - GSCN_0 (0x05B20; RW)
3GIO Statistic Counter Registers #1- GSCN_1 (0x05B24; RW)
3GIO Statistic Counter Registers #2- GSCN_2 (0x05B28; RW)
219
82583V GbE Controller—Driver Programing Interface
9.2.3.10
9.2.3.11
3GIO Statistic Counter Registers #3- GSCN_3 (0x05B2C; RW)
Software Semaphore Register - SWSM (0x05B50; RW)
Initial
Value
Field
Bit(s)
Description
This bit is set by hardware when this register is read by the software
device driver and cleared when the host driver writes 0b to it.
The first time this register is read, the value is 0b. In the next read,
the value is 1b (hardware mechanism). The value remains 1b until the
software device driver clears it.
Note: This bit is typically used by multiple software threads when a
device has multiple LAN functions. Since the 82583V is a single port
device it is typically not needed. However, if there are multiple
software threads of execution accessing the 82583V it could be used
as part of a software thread semaphore mechanism.
SMBI
0
1b
Software EEPROM Semaphore Bit
This bit should be set only by the software device driver (read only to
firmware).
Note: This bit should not be used in the 82583V. Instead, the
EXTCNF_CTRL.MDIO/NVW SW Ownership bit should be used when the
software device driver needs to access the PHY or NVM (see
Section 4.6.2).
SWESMBI
1
0b
Hardware clears this bit on GIO soft reset.
Reserved
Reserved
Reserved
2
3
0b
Reserved
Reserved
Reserved
0b
31:4
0x0
9.2.3.12
3GPIO Control Register 2 - GCR2 (0x05B64; RW)
Initial
Field
Bit(s)
31:1
Description
Value
Reserved
Reserved
0x0
0b
Reserved
Reserved. Must be set to 1b by software during initialization.
0
220
Driver Programing Interface—82583V GbE Controller
9.2.3.13
Statistic Event Mapping
Event
Mapping
(Hex)
Transaction layer Events
Description
Each 125 MHz cycle the counter increases by 1 (1
Dword) or 2 (2 Dwords).
Counted: completion, memory, message (not
replied).
Dwords of Transaction Layer Packet
(TLP) transmitted (transferred to the
physical layer), include payload and
header.
0
1
2
3
Only TLP packets. Each cycle, the counter increase by
1 if TLP packet was transmitted to the link.
Counted: completion, memory, message (not
replied).
All types of transmitted packets.
Transmit TLP packets of function #0
Transmit TLP packets of function #1
Each cycle, the counter increases by 1, if the packet
was transmitted.
Counted: memory, message of function 0 (not
replied).
Each cycle, the counter increases by 1, if the packet
was transmitted.
Counted: memory, message of function 1 (not
replied).
Each cycle, the counter increases by 1, if the packet
was transmitted.
Counted: memory (np) of function 0 (not replied).
Non posted transmit TLP packets of
function #0
4
5
Each cycle, the counter increases by 1, if the packet
was transmitted.
Counted: memory (np) of function 1 (not replied).
Non posted transmit TLP packets of
function #1
Each cycle, the counter increases by 1, if the packet
was transmitted.
Counted: memory, message for a given func# and
tag# (not replied).
Transmit TLP packets of function X and
tag Y, according to FUNC_TAG register
6
Each cycle, the counter increases by 1, if the packet
was received.
Counted: completion (only good), memory, I/O,
config.
All types of received packets (TLP only) 1A
Each cycle, the counter increases by 1, if the packet
was received.
Counted: good completions of func#0.
Receive TLP packets of function #0
Reserved
1B
1C
1D
Reserved
Each cycle, the counter increases by 1, if the packet
was received.
Counted: good completions for a given func# and
tag#.
Receive completion packets
Clock counter
20
Counts gio cycles.
Each cycle, the counter increases by 1, if a bad TLP is
received (bad CRC, error reported by AL, misplaced
special char, reset in thI of received TLP).
Bad TLP from LL
21
Only TLP, each 125 MHz cycle the counter increases
by 1 (1 Dword of header) or 2 (2 Dwords of the
header).
Counted: completion, memory, message (not
replied).
Header Dwords of transaction layer
packet transmitted.
25
Only TLP, each 125 MHz cycle the counter increases
by 1 (1 Dword of header) or 2 (2 Dwords of the
header).
Header Dwords of Transaction layer
packet received.
26
Counted: completion, memory, message.
221
82583V GbE Controller—Driver Programing Interface
Event
Mapping
(Hex)
Transaction layer Events
Description
The counter counts the number of times the
transaction layer stops transmitting because of this
(per packet).
Transaction layer stalls transmitting due
to lack of flow control credits of the next 27
part.
Counted: completion, memory, message.
The counter increases for each re-transmitted
packet.
Counted: completion, memory, message.
Retransmitted packets.
28
29
The counter counts the number of times transaction
layer stops transmitting because the retry buffer is
full (per packet).
Stall due to retry buffer full
Retry buffer is under threshold
Counted: completion, memory, message.
Threshold specified by software, Retry buffer is under
threshold per packet.
2A
2B
2C
2D
2E
2F
30
Counted: completion, memory, message.
Posted Request Header (PRH) flow
control credits (of the next part) below
threshold
Posted Request Data (PRD) flow control
credits (of the next part) below
threshold
Threshold specified by software.
Non-Posted Request Header (NPRH)
flow control credits (of the next part)
below threshold
The counter increases each time the number of the
specific flow control credits is lower than the
threshold.
Counted: According to credit type.
Completion Header (CPLH) flow control
credits (of the next part) below
threshold
Completion Data (CPLD) flow control
credits (of the next part) below
threshold
Posted Request Header (PRH) flow
control credits (of local part) get to
zero.
Non-Posted Request Header (NPRH)
flow control credits (of local part) get to 31
zero.
Threshold specified by software.
The counter increases each time the number of the
specific flow control credits reaches the value of zero.
(The period that the credit is zero is not counted).
Posted Request Data (PRD) flow control
32
Counted: According to credit type.
credits (of local part) get to zero.
Non-Posted Request Data (NPRD) flow
control credits (of local part) get to
zero.
33
34
Each 125 MHz cycle the counter increases by 1 (1
Dword) or 2 (2 Dwords).
Counted: completion, memory, message, I/O, config.
Dwords of TLP received, include payload
and header.
Each 125 MHz cycle the counter increases by 1.
Counted: messages (only good).
Messages packets received
35
36
Each 125 MHz cycle the counter increases by 1.
Counted: memory, I/O, config (only good).
Received packets to func_logic.
222
Driver Programing Interface—82583V GbE Controller
Event
Mapping
Host Arbiter Events
Description
Software selects the client that needs to be tested.
The statistic counter counts the number of read
requests of the required client.
In addition, the accumulated time of all requests are
saved in a time accumulator.
The average time for read request is:
[Accumulated time/number of read requests].
(Event 41 is for the counter).
Average latency of read request – from
initialization until end of completions.
Estimated latency is ~5 μs
40 + 41
Software selects the client that needs to be tested.
The statistic counter counts the number of read
requests of the required client.
In addition, the accumulated time of all RTT are
saved in a time accumulator.
Average latency of read request RTT–
from initialization until the first
completion is arrived (round trip time).
42 + 43
Estimated latency is 1 μs
The average time for read request is:
[Accumulated time/number of read requests].
(Event 43 is for the counter).
Requests that reached time out.
44
Number of requests that reached time out.
Software selects the client that needs to be tested.
Software programs the required threshold (in
GSCL_4 – units of 96 ns).
One statistic counter counts the time from the
beginning of the request until end of completions.
Completion latency above threshold
45 + 46
The other counter counts the number of events.
If the time is above threshold – add 1 to the event
counter.
(Event 46 is for the counter).
Software selects the client that needs to be tested.
Software programs the required threshold (in
GSCL_4 – units of 96 ns).
One statistic counter counts the time from the
beginning of the request until first completion arrival.
The other counter counts the number of events.
Completion Latency above Threshold –
for RTT
47 + 48
If the time is above threshold – add 1 to the event
counter.
(Event 48 is for the counter).
Event
Mapping
Link Layer Events
Description
Include DLLP (Link layer packets) and TLP
(transaction layer packets transmitted.
Each 125 MHz cycle the counter increases by 1 (1
Dword) or 2 (2 Dwords).
Dwords of the packet transmitted
(transferred to the physical layer),
include payload and header.
50
51
Include DLLP (Link layer packets) and TLP
(transaction layer packets transmitted.
Each 125 MHz cycle the counter increases by 1 (1
Dword) or 2 (2 Dwords).
Dwords of packet received (transferred
to the physical layer), include payload
and header.
All types of DLLP packets transmitted
from link layer.
Each cycle, the counter increases by 1, if DLLP packet
was transmitted.
52
53
54
55
Flow control DLLP transmitted from link
layer.
Each cycle, the counter increases by 1, if message
was transmitted
Each cycle, the counter increases by 1, if message
was transmitted.
Ack DLLP transmitted.
Each cycle, the counter increases by 1, if DLLP was
received.
All types of DLLP packets received.
223
82583V GbE Controller—Driver Programing Interface
Event
Mapping
Link Layer Events
Description
Each cycle, the counter increases by 1, if message
was received.
Flow control DLLP received in link layer. 56
Each cycle, the counter increases by 1, if message
was received.
Ack DLLP received.
Nack DLLP received.
57
58
Each cycle, the counter increases by 1, if message
was transmitted.
9.2.4
Interrupt Register Descriptions
9.2.4.1
Interrupt Cause Read Register - ICR (0x000C0; RC/WC)
Initial
Value
Field
Bit(s)
Description
Transmit Descriptor Written Back
Set when hardware processes a descriptor with RS set. If using
delayed interrupts (IDE set), the interrupt is delayed until after one of
the delayed-timers (TIDV or TADV) expires.
TXDW
0
1
0b
Transmit Queue Empty
Set when the last descriptor block for a transmit queue has been
used.
TXQE
0b
Link Status Change
This bit is set whenever the link status changes (either from up to
down, or from down to up). This bit is affected by the link indication
from the PHY.
LSC
2
3
4
5
6
0b
0b
0b
0b
0b
Reserved
RXDMT0
Reserved
RXO
Reserved
Receive Descriptor Minimum Threshold Hit.
This bit indicates that the number of receive descriptors has reached
the minimum threshold as set in RCTL.RDMTS. This indicates to the
software to load more receive descriptors.
Reserved
Receiver Overrun
Set on receive data FIFO overrun. Could be caused either because
there are no available buffers or because PCIe receive bandwidth is
inadequate.
Receiver Timer Interrupt
Set when the timer expires.
RXT0
7
8
9
0b
Reserved
MDAC
0b
Reserved
MDIO Access Complete
Set when MDIO access completes. See Section 9.2.7.36 for details.
0b
Reserved
14:10
15
0x0
Reserved
Transmit Descriptor Low Threshold Hit
Indicates that the number of descriptors in the transmit descriptor
ring has reached the level specified in the Transmit Descriptor Control
register (TXDCTL.LWTHRESH).
TXD_LOW
0b
Small Receive Packet Detected
Indicates that a packet of size < RSRPD.SIZE has been detected and
transferred to host memory. The interrupt is only asserted if
RSRPD.SIZE register has a non-zero value.
SRPD
ACK
16
17
0b
0b
Receive ACK Frame Detected
Indicates that an ACK frame has been received and the timer in
RAID.ACK_DELAY has expired.
224
Driver Programing Interface—82583V GbE Controller
Initial
Field
Bit(s)
19:18
Description
Value
Reserved
00b
Reserved
Receive Queue 0 Interrupt
RxQ0
20
0b
Indicates Receive queue 0 write back or receive queue 0 descriptor
minimum threshold hit.
Reserved
TxQ0
21
22
23
0b
0b
0b
Reserved
Transmit Queue 0 Interrupt
Indicates transmit queue 0 write back.
Reserved
Reserved
Other Interrupt. Indicates one of the following interrupts was set:
•
•
•
•
•
Link Status Change.
Receiver Overrun.
MDIO Access Complete.
Small Receive Packet Detected.
Receive ACK Frame Detected.
Other
24
0b
Reserved
Reads as 0x0.
Reserved
30:25
0x0
0b
Interrupt Asserted
INT_
ASSERTED
This bit is set when the LAN port has a pending interrupt. If the
interrupt is enabled in the PCI configuration space, an interrupt is
asserted.
31
This register contains all interrupt conditions for the 82583V. Whenever an interrupt
causing event occurs, the corresponding interrupt bit is set in this register. A PCIe
interrupt is generated whenever one of the bits in this register is set, and the
corresponding interrupt is enabled via the Interrupt Mask Set/Read register.
Whenever an interrupt causing event occurs, all timers of delayed interrupts are
cleared and their cause event is set in the ICR.
Reading from the ICR register has different effects according to the following three
cases:
• Case 1 - Interrupt Mask register equals 0x0000 (mask all): ICR content is cleared.
• Case 2 - Interrupt was asserted (ICR.INT_ASSERT=1) and auto mask is active: ICR
content is cleared, and the IAM register is written to the IMC register.
• Case 3 - Interrupt was not asserted (ICR.INT_ASSERT=0): Read has no side affect.
Writing a 1b to any bit in the register also clears that bit. Writing a 0b to any bit has no
effect on that bit.
Note:
The INT_ASSERTED bit is a special case. Writing a 1b or 0b to this bit has no affect. It
is cleared only when all interrupt sources are cleared.
9.2.4.2
Interrupt Throttling Register - ITR (0x000C4; R/W)
Initial
Value
Field
Bit(s)
Description
Minimum Inter-Interrupt Intervall
INTERVAL
Reserved
15:0
0x0
The interval is specified in 256 ns increments. Zero disables interrupt
throttling logic.
Reserved
31:16
0x0
Should be written with 0x0 to ensure future compatibility.
225
82583V GbE Controller—Driver Programing Interface
Software can use this register to prevent the condition of repeated, closely spaced,
interrupts to the host CPU, asserted by the 82583V, by guaranteeing a minimum delay
between successive interrupts.
To independently validate configuration settings, software can use the following
algorithm to convert the inter-interrupt interval value to the common interrupts/sec
performance metric:
interrupts/sec = (256 x 10-9sec x interval)-1
For example, if the interval is programmed to 500 (decimal), the 82583V guarantees
the CPU is not interrupted by it for 128 μs from the last interrupt. The maximum
observable interrupt rate from the 82583V should never exceed 7813 interrupts/sec.
Inversely, inter-interrupt interval value can be calculated as:
inter-interrupt interval = (256 x 10-9sec x interrupts/sec) -1
The optimal performance setting for this register is very system and configuration
specific. An initial suggested range is 651- 5580 decimal (or 0x28B - 0x15CC).
226
Driver Programing Interface—82583V GbE Controller
9.2.4.3
Interrupt Cause Set Register - ICS (0x000C8; W)
Initial
Value
Field
TXDW
Bit(s)
Description
Sets Transmit Descriptor Written Back
0
1
2
3
4
5
X
X
X
X
X
X
TXQE
Sets Transmit Queue Empty
Sets Link Status Change.
Reserved
LSC
Reserved
RXDMT0
Reserved
Sets Receive Descriptor Minimum Threshold Hit
Reserved
Sets Receiver Overrun
Set on receive data FIFO overrun.
RXO
6
X
RXT0
7
8
9
X
X
X
X
X
X
X
X
X
X
X
0
0
0
0
0
Sets Receiver Timer Interrupt
Reserved
reserved
MDAC
Sets MDIO Access Complete Interrupt
Reserved
Reserved
Reserved
Reserved
Reserved
TXD_LOW
SRPD
10
11
Reserved
12
Reserved
14:13
15
Reserved
Transmit Descriptor Low Threshold Hit
Small Receive Packet Detected and Transferred
Sets Receive ACK Frame Detected
Reserved
16
ACK
17
Reserved
RxQ0
19:18
20
Sets Receive Queue 0 Interrupt
Reserved
Reserved
TxQ0
21
22
Sets Transmit Queue 0 Interrupt
Reserved
Reserved
Other
23
24
Sets Other Interrupt
Reserved
Reserved
31:25
X
Should be written with 0x0 to ensure future compatibility
Software uses this register to set an interrupt condition. Any bit written with a 1b sets
the corresponding interrupt. This results in the corresponding bit being set in the
Interrupt Cause Read register (see Section 9.2.4.1). A PCIe interrupt is also generated
if one of the bits in this register is set and the corresponding interrupt is enabled via
the Interrupt Mask Set/Read register (see Section 9.2.4.4).
Bits written with 0b are unchanged.
227
82583V GbE Controller—Driver Programing Interface
9.2.4.4
Interrupt Mask Set/Read Register - IMS (0x000D0; RW)
Initial
Value
Field
TXDW
Bit(s)
Description
0
1
2
3
4
5
6
7
8
9
0b
Sets the mask for transmit descriptor written back.
TXQE
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
0x0
0b
0b
0b
X
Sets the mask for transmit queue empty.
LSC
Sets the mask for link status change.
Reserved
RXDMT0
Reserved
RXO
Reserved
Sets the mask for receive descriptor minimum threshold hit.
Reserved.
Sets mask for receiver overrun. Set on receive data FIFO overrun.
RXT0
Sets mask for receiver timer interrupt.
reserved
MDAC
Reserved
Sets mask for MDIO access complete interrupt.
Reserved
Reserved
Reserved
Reserved
TXD_LOW
SRPD
10
Reserved
11
Reserved
12
Reserved
14:13
15
Reserved
Sets the mask for transmit descriptor low threshold hit.
Sets the mask for small receive packet detection.
Sets the mask forreceive ACK frame detection.
Reserved
16
ACK
17
Reserved
RxQ0
19:18
20
0b
0b
0b
0b
0b
Sets the mask for receive queue 0 interrupt.
Reserved
Reserved
TxQ0
21
22
Sets the mask for transmit queue 0 interrupt.
Reserved
Reserved
Other
23
24
Sets the mask for other interrupt.
Reserved
Reserved
31:25
x0
Should be written with 0x0 to ensure future compatibility.
Reading this register returns which bits have an interrupt mask set. An interrupt is
enabled if its corresponding mask bit is set to 1b, and disabled if its corresponding
mask bit is set to 0b. A PCIe interrupt is generated whenever one of the bits in this
register is set, and the corresponding interrupt condition occurs. The occurrence of an
interrupt condition is reflected by having a bit set in the Interrupt Cause Read register
(see Section 9.2.4.1).
A particular interrupt can be enabled by writing a 1b to the corresponding mask bit in
this register. Any bits written with a 0b, are unchanged. Thus, if software desires to
disable a particular interrupt condition that had been previously enabled, it must write
to the Interrupt Mask Clear register (see Section 9.2.4.5), rather than writing a 0b to a
bit in this register.
When the CTRL_EXT.INT_TIMERS_CLEAR_ENA bit is set, then following writing all 1b's
to the IMS register (enable all interrupts) all interrupt timers are cleared to their initial
value. This auto clear provides the required latency before the next INT event.
228
Driver Programing Interface—82583V GbE Controller
9.2.4.5
Interrupt Mask Clear Register - IMC (0x000D8; W)
Initial
Value
Field
TXDW
Bit(s)
Description
0
1
2
3
4
0b
Clears the mask for transmit descriptor written back.
Clears the mask for transmit queue empty.
Clears the mask for link status change.
Reserved
TXQE
0b
0b
0b
0b
LSC
Reserved
RXDMT0
Clears the mask for receive descriptor minimum threshold hit.
Reserved
Reads as 0b.
Reserved
RXO
5
6
0b
0b
Clears the mask for receiver overrun. Set on receive data FIFO
overrun.
RXT0
7
8
9
0b
0b
0b
0b
Clears the mask for receiver timer interrupt.
reserved
MDAC
Reserved
Clears the mask for MDIO access complete interrupt.
Reserved
Reserved
10
11
Reserved
Reads as 0b.
Reserved
0b
Reserved
Reserved
TXD_LOW
SRPD
12
0b
00b
0b
0b
0
Reserved
14:13
15
Reserved
Clears the mask for transmit descriptor low threshold hit.
Clears the mask for small receive packet detect interrupt.
Clears the mask for receive ACK frame detect interrupt.
Reserved
16
ACK
17
Reserved
RxQ0
19:18
20
X
0
Clears the mask for receive queue 0 interrupt.
Reserved
Reserved
TxQ0
21
0
22
0
Clears the mask for transmit queue 0 interrupt.
Reserved
Reserved
Other
23
0
24
0
Clears the mask for other interrupt.
Reserved
Reserved
31:25
0
Should be written with 0x0 to ensure future compatibility.
Software uses this register to disable an interrupt. Interrupts are presented to the bus
interface only when the mask bit is 1b and the cause bit is 1b. The status of the mask
bit is reflected in the Interrupt Mask Set/Read register (see Section 9.2.4.4), and the
status of the cause bit is reflected in the Interrupt Cause Read register (see
Section 9.2.4.3).
Software blocks interrupts by clearing the corresponding mask bit. This is accomplished
by writing a 1b to the corresponding bit in this register. Bits written with 0b are
unchanged (for example, their mask status does not change).
In summary, the sole purpose of this register is to enable software a way to disable
certain, or all, interrupts. Software disables a given interrupt by writing a 1b to the
corresponding bit in this register.
229
82583V GbE Controller—Driver Programing Interface
9.2.4.6
Interrupt Auto Clear- EIAC (0x000DC; RW)
Initial
Value
Field
Bit(s)
19:0
Description
Reserved
Reserved
Reserved
0x0
Reserved
Reserved
Reserved
24:20
31:25
0x0
0x0
9.2.4.7
Interrupt Acknowledge Auto–Mask - IAM (0x000E0; RW)
Initial
Value
Field
Bit(s)
Description
When the CTRL_EXT.IAME bit is set and the ICR.INT_ASSERT=1b, an
ICR read or write has the side effect of writing the contents of this
register to the IMC register.
IAM_VALUE
31:0
0x0
9.2.5
Receive Register Descriptions
9.2.5.1
Receive Control Register - RCTL (0x00100; RW)
Initial
Value
Field
Bit(s)
Description
Reserved
This bit represented as a hardware reset of the receive-related
portion of the device in previous controllers, but is no longer
applicable. Only a full device reset CTRL.RST is supported. Write as 0b
for future compatibility.
Reserved
0
1
0b
Enable
The receiver is enabled when this bit is set to 1b. Writing this bit to
0b, stops reception after receipt of any in progress packet. All
subsequent packets are then immediately dropped until this bit is set
to 1b.
EN
0b
0b
Store Bad Packets
0b = Do not store
1b = Store.
Note that CRC errors before the SFD are ignored. Any packet must
have a valid SFD (RX_DV with no RX_ER in the GMII/MII interface) in
order to be recognized by the device (even bad packets)
SBP
2
Unicast Promiscuous Enable
0b = Disabled.
1b = Enabled.
UPE
MPE
LPE
3
4
5
0b
0b
0b
Multicast Promiscuous Enable
0b = Disabled.
1b = Enabled.
Long Packet Enable.
0b = Disabled (always set to 0b).
1b = Enabled (not supported).
230
Driver Programing Interface—82583V GbE Controller
Initial
Value
Field
Bit(s)
Description
Loopback mode
Should always be set to 00b.
00b = Normal operation (or PHY loopback in GMII/MII mode).
01b = MAC Loopback (test mode).
10b = Undefined.
LBM
7:6
00b
11b = Undefined.
Receive Descriptor Minimum Threshold Size
The corresponding interrupt is set whenever the fractional number of
free descriptors becomes equal to RDMTS. Table 50 lists which
fractional values correspond to RDMTS values. See Section 9.2.5.7 for
details regarding RDLEN.
RDMTS
DTYP
9:8
00b
00b
Descriptor Type
00b = Legacy descriptor type.
01b = Packet split descriptor type.
10b = Reserved.
11:10
11b = Reserved.
Multicast Offset
This determines which bits of the incoming multicast address are used
in looking up the bit vector.
00b = [47:36].
01b = [46:35].
10b = [45:34].
11b = [43:32].
MO
13:12
00b
Reserved
BAM
14
15
0b
0b
Reserved
Broadcast Accept Mode
0b = Ignore broadcast packets (unless they pass through exact or
imperfect filters).
1b = Accept broadcast packets.
Receive Buffer Size
If RCTL.BSEX = 0b:
00b = 2048 bytes.
01b = 1024 bytes.
10b = 512 bytes.
11b = 256 bytes.
If RCTL.BSEX = 1b:
BSIZE
17:16
0b
00b = reserved; software should not set to this value.
01b = 16384 bytes.
10b = 8192 bytes.
11b = 4096 bytes.
BSIZE is only used when DTYP = 00b. When DTYP = 01b, the buffer
sizes for the descriptor are controlled by fields in the PSRCTL register.
BSIZE is not relevant when FLXBUF is different from 0x0, in that case,
FLXBUF determines the buffer size.
VLAN Filter Enable.
0b = Disabled (filter table does not decide packet acceptance).
1b = Enabled (filter table decides packet acceptance for 802.1Q
packets).
VFE
18
0b
Canonical Form Indicator Enable
0b = Disabled (CFI bit not compared to decide packet acceptance).
1b = Enabled (CFI from packet must match next field to accept
802.1Q packets).
CFIEN
CFI
19
20
0b
0b
Canonical Form Indicator Bit Value
If CFI is set, then 802.1Q packets with CFI equal to this field are
accepted; otherwise, the 802.1Q packet is discarded.
231
82583V GbE Controller—Driver Programing Interface
Initial
Value
Field
Bit(s)
Description
Reserved
Reserved
21
0b
Should be written with 0b to ensure future compatibility.
Discard Pause Frames
Any valid pause frame is discarded regardless of whether it matches
any of the filter registers.
0b = Incoming frames subject to filter comparison.
1b = Incoming pause frames ignored even if they match filter
registers.
DPF
22
23
0b
0b
Pass MAC Control Frames
0b = Do not (specially) pass MAC control frames.
1b = Pass any MAC control frame (type field value of 0x8808) that
does not contain the pause opcode of 0x0001.
PMCF
Reserved
Reserved
BSEX
24
25
26
0b
0b
0b
Should be written with 0b to ensure future compatibility.
Buffer Size Extension
Modifies the buffer size indication (BSIZE). When set to 1b, the
original BSIZE values are multiplied by 16.
Strip Ethernet CRC from incoming packet.
Do not DMA to host memory.
SECRC
Determines a flexible buffer size. When this field is 0x0000, the buffer
size is determined by BSIZE. If this field is different from 0x0000, the
receive buffer size is the number represented in KB. For example,
0x0001 = 1 KB (1024 bytes).
FLXBUF
30:27
31
0x0
0b
Reserved
Reserved
Should be written with 0b to ensure future compatibility.
Hardware discards long packets if LPE is 0b. A long packet is one longer than 1522
bytes. Note that the 82583V does not support setting LPE to 1b.
RDMTS[1,0] determines the threshold value for free receive descriptors according to
the following table:
Table 50.
RDMTS Values
RDMTS
Free Buffer Threshold
00b
01b
10b
11b
1/2
1/4
1/8
Reserved
BSIZE controls the size of the receive buffers and permits software to trade-off
descriptor performance versus required storage space. Buffers that are 2048 bytes
require only one descriptor per receive packet maximizing descriptor efficiency. Buffers
that are 256 bytes maximize memory efficiency at a cost of multiple descriptors for
packets longer than 256 bytes.
Three bits control the VLAN filter table. The first determines whether the table
participates in the packet acceptance criteria. The next two are used to decide whether
the CFI bit found in the 802.1Q packet should be used as part of the acceptance
criteria.
232
Driver Programing Interface—82583V GbE Controller
DPF controls the DMA function of flow control packets addressed to the station address
(RAH/L[0]). If a packet is a valid flow control packet and is addressed to the station
address it is not DMA'd to host memory if DPF=1b.
PMCF controls the DMA function of MAC control frames (other than flow control). A MAC
control frame in this context must be addressed to either the MAC control frame
multicast address or the station address, match the type field and NOT match the
pause op-code of 0x0001. If PMCF=1b then frames meeting this criteria are DMA'd to
host memory.
The SECRC bit controls whether the hardware strips the Ethernet CRC from the
received packet. This stripping occurs prior to any checksum calculations. The stripped
CRC is not DMA'd to host memory and is not included in the length reported in the
descriptor.
9.2.5.2
Packet Split Receive Control Register - PSRCTL (0x02170; RW)
Initial
Value
Field
Bit(s)
Description
Receive Buffer Size for Buffer 0.
The value is in 128-byte resolution. Value can be from 128 bytes to
16256 bytes (15.875 KB). Default buffer size is 256 bytes. Software
should not program this field to a zero value.
BSIZE0
6:0
0x2
Reserved
Rsv
7
0b
Should be written with 0b to ensure future compatibility.
Receive Buffer Size for Buffer 1.
The value is in 1 KB resolution. Value can be from 1 KB to 63 KB.
Default buffer size is 4 KB. Software should not program this field to a
zero value.
BSIZE1
Rsv
13:8
15:14
21:16
23:22
29:24
31:30
0x4
00b
0x4
00b
0x0
00b
Reserved
Should be written with 00b to ensure future compatibility.
Receive Buffer Size for Buffer 2.
The value is in 1 KB resolution. Value can be from 1 KB to 63 KB.
Default buffer size is 4 KB. Software can program this field to any
value.
BSIZE2
Rsv
Reserved
Should be written with 00b to ensure future compatibility.
Receive Buffer Size for Buffer 3
The value is in 1 KB resolution. Value can be from 1 KB to 63 KB.
Default buffer size is 0 KB. Software can program this field to any
value.
BSIZE3
Rsv
Reserved
Should be written with 0b to ensure future compatibility.
Note:
If software sets a buffer size to zero, all buffers following that one must be set to zero
as well. Pointers in the receive descriptors to buffers with a zero size should be set to
null pointers.
233
82583V GbE Controller—Driver Programing Interface
9.2.5.3
Flow Control Receive Threshold Low - FCRTL (0x02160; RW)
Initial
Value
Field
Bit(s)
Description
Reserved
Reserved
2:0
0x0
The underlying bits might not be implemented in all versions of the
chip. Must be written with 0x0.
Receive Threshold Low
FIFO low water mark for flow control transmission.
RTL
15:3
0x0
0x0
Reserved
Reserved
30:16
Reads as 0x0. Should be written to 0x0 for future compatibility.
XON Enable
XONE
31
0b
0b = Disabled.
1b = Enabled.
This register contains the receive threshold used to determine when to send an XON
packet. It counts in units of bytes. The lower 3 bits must be programmed to zero (8-
byte granularity). Software must set XONE to enable the transmission of XON frames.
Whenever hardware crosses the receive high threshold (becoming more full), and then
crosses the receive low threshold and XONE is enabled (= 1b), hardware transmits an
XON frame.
Note:
Note:
Note that flow control reception/transmission are negotiated capabilities by the auto-
negotiation process. When the device is manually configured, flow control operation is
determined by the RFCE and TFCE bits of the Device Control register.
This register's address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00168.
9.2.5.4
Flow Control Receive Threshold High - FCRTH (0x02168; RW)
Initial
Value
Field
Bit(s)
Description
Reserved
Reserved
2:0
0x0
The underlying bits might not be implemented in all versions of the
chip. Must be written with 0x0.
Receive Threshold High
FIFO high water mark for flow control transmission.
RTH
15:3
0x0
0x0
Reserved
Reserved
31:16
Reads as 0b. Should be written to 0b for future compatibility.
This register contains the receive threshold used to determine when to send an XOFF
packet. It counts in units of bytes. This value must be at least 8 bytes less than the
maximum number of bytes allocated to the Receive Packet Buffer (PBA.RXA), and the
lower 3 bits must be programmed to zero (8-byte granularity). Whenever the receive
FIFO reaches the fullness indicated by RTH, hardware transmits a pause frame if the
transmission of flow control frames is enabled.
Note:
Note that flow control reception/transmission are negotiated capabilities by the auto-
negotiation process. When the device is manually configured, flow control operation is
determined by the RFCE and TFCE bits of the Device Control register.
234
Driver Programing Interface—82583V GbE Controller
Note:
This register's address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00160.
9.2.5.5
Receive Descriptor Base Address Low - RDBAL (0x02800; RW)
Initial
Value
Field
Bit(s)
Description
0
3:0
31:4
0x0
Ignored on writes. Returns 0b on reads.
Receive Descriptor Base Address Low
RDBAL
X
This register contains the lower bits of the 64-bit descriptor base address. The lower 4
bits are always ignored. The Receive Descriptor Base Address must point to a 16-byte
aligned block of data.
Note:
This register's address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00110.
9.2.5.6
Receive Descriptor Base Address High - RDBAH (0x02804; RW)
Initial
Value
Field
RDBAH
Bit(s)
31:0
Description
X
Receive Descriptor Base Address [63:32]
This register contains the upper 32 bits of the 64-bit descriptor base address.
Note:
This register's address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00114.
9.2.5.7
Receive Descriptor Length - RDLEN (0x02808; RW)
Initial
Field
Bit(s)
6:0
Description
Ignore on write. Reads back as 0x0.
Value
0
0x0
LEN
19:7
0x0
0x0
Descriptor Length
Reserved
31:20
Reads as 0x0. Should be written to 0x0 for future compatibility.
This register sets the number of bytes allocated for descriptors in the circular descriptor
buffer. It must be 128-byte aligned.
Note:
This register's address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00118.
235
82583V GbE Controller—Driver Programing Interface
9.2.5.8
Receive Descriptor Head - RDH (0x02810; RW)
Initial
Value
Field
Bit(s)
15:0
31:16
Description
RDH
Reserved
0x0
0x0
Receive Descriptor Head
Reserved
Should be written with 0x0
This register contains the head pointer for the receive descriptor buffer. The register
points to a 16-byte datum. Hardware controls the pointer. The only time that software
should write to this register is after a reset (hardware reset or CTRL.RST) and before
enabling the receive function (RCTL.EN). If software were to write to this register while
the receive function was enabled, the on-chip descriptor buffers might be invalidated
and the hardware could be become unstable.
Note:
This register's address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00120.
9.2.5.9
Receive Descriptor Tail - RDT (0x02818; RW)
Initial
Value
Field
Bit(s)
Description
RDT
Reserved
15:0
31:16
0x0
0x0
Receive Descriptor Tail
Reads as 0x0. Should be written to 0x0 for future compatibility.
This register contains the tail pointers for the receive descriptor buffer. The register
points to a 16-byte datum. Software writes the tail register to add receive descriptors
to the hardware free list for the ring.
Note:
This register's address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00128.
9.2.5.10
Rx Interrupt Delay Timer [Packet Timer] - RDTR (0x02820; RW)
Initial
Field
Delay
Bit(s)
15:0
Description
Value
0x0
Receive packet delay timer measured in increments of 1.024 μs.
Reserved
Reads as 0x0
Reserved
FPD
30:16
31
0x0
0x0
Flush Partial Descriptor Block
When set to 1b, flushes the partial descriptor block; ignored
otherwise. Reads 0b.
This register is used to delay interrupt notification for the receive descriptor ring by
coalescing interrupts for multiple received packets. Delaying interrupt notification helps
maximize the number of receive packets serviced by a single interrupt.
236
Driver Programing Interface—82583V GbE Controller
This feature operates by initiating a countdown timer upon successfully receiving each
packet to system memory. If a subsequent packet is received before the timer expires,
the timer is re-initialized to the programmed value and re-starts its countdown. If the
timer expires due to not having received a subsequent packet within the programmed
interval, pending receive descriptor write backs are flushed and a receive timer
interrupt is generated.
Setting the value to zero represents no delay from a receive packet to the interrupt
notification, and results in immediate interrupt notification for each received packet.
Writing this register with FPD set initiates an immediate expiration of the timer, causing
a write back of any consumed receive descriptors pending write back, and results in a
receive timer interrupt in the ICR.
Receive interrupts due to a Receive Absolute Timer (RADV) expiration cancels a
pending RDTR interrupt. The RDTR countdown timer is reloaded but halted, so as to
avoid generation of a spurious second interrupt after the RADV has been noted, but can
be restarted by a subsequent received packet.
Note:
Note:
FPD is self clearing.
This register's address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00108.
9.2.5.11
Receive Descriptor Control - RXDCTL (0x02828; RW)
Initial
Field
Bit(s)
5:0
Description
Value
PTHRESH
Rsv
0x00
Prefetch Threshold
Reserved
7:6
0x00
0x00
0b
HTHRESH
Reserved
Rsv
13:8
14
Host Threshold
Reserved
15
0b
Reserved
WTHRESH
Rsv
21:16
23:22
0x01
00b
Write-Back Threshold
Reserved
Granularity
Units for the thresholds in this register.
0b = Cache lines.
1b = Descriptors.
GRAN
Rsv
24
0b
31:25
0x0
Reserved
Note:
Any value written to RXDCTL0 is automatically written to RXDCTL1. Writes to RXDCTL1
affects RXDCTL1 only.
This register controls the fetching and write back of receive descriptors. The three
threshold values are used to determine when descriptors are read from and written to
host memory. The values can be in units of cache lines or descriptors (each descriptor
is 16 bytes) based on the GRAN flag. If GRAN=0b (specifications are in cache-line
granularity), the thresholds specified (based on the cache line size specified in the PCIe
header CLS field) must not represent greater than 31 descriptors.
237
82583V GbE Controller—Driver Programing Interface
PTHRESH is used to control when a prefetch of descriptors are considered. This
threshold refers to the number of valid, unprocessed receive descriptors the chip has in
its on-chip buffer. If this number drops below PTHRESH, the algorithm considers pre-
fetching descriptors from host memory. This fetch does not happen however, unless
there are at least HTHRESH valid descriptors in host memory to fetch.
Note:
Note:
HTHRESH should be given a non-zero value whenever PTHRESH is used.
WTHRESH controls the write back of processed receive descriptors. This threshold
refers to the number of receive descriptors in the on-chip buffer which are ready to be
written back to host memory. In the absence of external events (explicit flushes), the
write back occurs only after at least WTHRESH descriptors are available for write back.
Possible values:
GRAN = 1b (descriptor granularity):
PTHRESH = 0..47
WTHRESH = 0..63
HTHRESH = 0..63
GRAN = 0 (cacheline granularity):
PTHRESH = 0..3 (for 16 descriptors cacheline - 256 bytes)
WTHRESH = 0..3
HTHRESH = 0..4
Note:
Note:
For any WTHRESH value other than zero - packet and absolute timers must get a non-
zero value for WTHRESH feature to take affect.
Since the default value for write-back threshold is one, the descriptors are normally
written back as soon as one cache line is available. WTHRESH must contain a non-zero
value to take advantage of the write-back bursting capabilities of the 82583V.
9.2.5.12
Receive Interrupt Absolute Delay Timer- RADV (0x0282C; RW)
Initial
Value
Field
Bit(s)
Description
Receive absolute delay timer measured in increments of 1.024 μs (0=
disabled).
Delay
15:0
0x0
Reserved
Reads as 0x0.
Reserved
31:16
0x0
If the packet delay timer is used to coalesce receive interrupts, it ensures that when
receive traffic abates, an interrupt is generated within a specified interval of no
receives. During times when receive traffic is continuous, it might be necessary to
ensure that no receive remains unnoticed for too long an interval. This register can be
used to ensure that a receive interrupt occurs at some predefined interval after the first
packet is received.
When this timer is enabled, a separate absolute count-down timer is initiated upon
successfully receiving each packet to system memory. When this absolute timer
expires, pending receive descriptor write backs are flushed and a receive timer
interrupt is generated.
238
Driver Programing Interface—82583V GbE Controller
Setting this register to 0x0 disables the absolute timer mechanism (the RDTR register
should be used with a value of 0x0 to cause immediate interrupts for all receive
packets).
Receive interrupts due to a Receive Packet Timer (RDTR) expiration cancels a pending
RADV interrupt. If enabled, the RADV count-down timer is reloaded but halted, so as to
avoid generation of a serious second interrupt after the RDTR has been noted.
9.2.5.13
Receive Small Packet Detect Interrupt- RSRPD (0x02C00; R/W)
Initial
Value
Field
Bit(s)
Description
If the interrupt is enabled any received packet of size <= SIZE asserts
an interrupt. SIZE is specified in bytes and includes the headers and
the CRC. It does not include the VLAN header in size calculation if it is
stripped.
SIZE
11:0
0x0
Reserved
31:12
X
Reserved.
9.2.5.14
Receive ACK Interrupt Delay Register - RAID (0x02C08; RW)
Initial
Field
Bit(s)
16:31
Description
Value
RSV
0x0
Reserved
ACK delay timer measured in increments of 1.024 μs. When the
receive ACK frame detect interrupt is enabled in the IMS register, ACK
packets being received uses a unique delay timer to generate an
interrupt. When an ACK is received, an absolute timer loads to the
value of ACK_DELAY. The interrupt signal is set only when the timer
expires. If another ACK packet is received while the timer is counting
down, the timer is not reloaded to ACK_DELAY.
ACK_DELAY
15:0
0x0
If an immediate (non-scheduled) interrupt is desired for any received ACK frame, the
ACK_DELAY should be set to x00.
9.2.5.15
Receive Checksum Control - RXCSUM (0x05000; RW)
Initial
Field
PCSS
Bit(s)
7:0
Description
Value
0x0
Packet Checksum Start
IPOFLD
TUOFLD
Reserved
CRCOFL
IPPCSE
PCSD
8
1b
1b
0b
0b
0b
0b
0x0
IP Checksum Offload Enable
9
TCP/UDP Checksum Offload Enable
Reserved
10
11
12
13
31:14
CRC32 Offload Enable
IP Payload Checksum Enable
Packet Checksum Disable
Reserved
Reserved
The Receive Checksum Control register controls the receive checksum offloading
features of the 82583V. The 82583V supports the offloading of three receive checksum
calculations: the packet checksum, the IP header checksum, and the TCP/UDP
checksum.
239
82583V GbE Controller—Driver Programing Interface
PCSD: The Packet Checksum and IP Identification fields. Only one of the two options is
reported in the Rx descriptor. The RXCSUM.PCSD affect is listed as follows:
RXCSUM.PCSD
0b (Checksum Enable)
1b (Checksum Disable)
Legacy Rx Descriptor
(RCTL.DTYP = 00b)
Packet checksum is reported in the
Rx Descriptor
Unsupported configuration.
Extended or Header Split Rx
Descriptor
(RCTL.DTYP = 01b)
Packet checksum and IP
identification are reported in the Rx
Descriptor
Reserved.
PCSS IPPCSE: The PCSS and the IPPCSE control the packet checksum calculation. The
packet checksum is reported in the receive descriptor when the RXCSUM.PCSD bit is
cleared.
If RXCSUM.IPPCSE cleared (the default value), the checksum calculation that is
reported in the Rx Packet Checksum field is the unadjusted 16-bit ones complement of
the packet. The Packet Checksum starts from the byte indicated by RXCSUM.PCSS
(zero corresponds to the first byte of the packet), after VLAN stripping if enabled by the
CTRL.VME. For example, for an Ethernet II frame encapsulated as an 802.3ac VLAN
packet and with RXCSUM.PCSS set to 14, the packet checksum would include the entire
encapsulated frame, excluding the 14-byte Ethernet header (DA, SA, Type/Length) and
the 4-byte VLAN tag. The Packet Checksum does not include the Ethernet CRC if the
RCTL.SECRC bit is set. Software must make the required offsetting computation (to
back out the bytes that should not have been included and to include the pseudo-
header) prior to comparing the Packet Checksum against the TCP checksum stored in
the packet.
If RXCSUM.IPPCSE is set, the Packet Checksum is aimed to accelerate checksum
calculation of fragmented UDP packets.
Note:
The PCSS value should not exceed a pointer to IP header start or else it will erroneously
calculate IP header checksum or TCP/UDP checksum.
RXCSUM.IPOFLD is used to enable the IP Checksum offloading feature. If
RXCSUM.IPOFLD is set to one, the 82583V calculates the IP checksum and indicates a
pass/fail indication to software via the IP Checksum Error bit (IPE) in the Error field of
the receive descriptor. Similarly, if RXCSUM.TUOFLD is set to one, the 82583V
calculates the TCP or UDP checksum and indicates a pass/fail indication to software via
the TCP/UDP Checksum Error bit (TCPE). Similarly, if RFCTL.IPv6_DIS and
RFCTL.IP6Xsum_DIS are cleared to zero and RXCSUM.TUOFLD is set to one, the
82583V calculates the TCP or UDP checksum for IPv6 packets. It then indicates a pass/
fail condition in the TCP/UDP Checksum Error bit (RDESC.TCPE).
This applies to checksum offloading only. Supported frame types:
• Ethernet II
• Ethernet SNAP
RXCSUM.CRCOFL is used to enable the CRC32 checksum offloading feature. If
RXCSUM.CRCOFL is set to one, the 82583V calculates the CRC32 checksum and
indicates a pass/fail indication to software via the CRC32 Checksum Error bit (CRCE) in
the Error field of the receive descriptor.
This register should only be initialized (written) when the receiver is not enabled (for
example, only write this register when RCTL.EN = 0b).
240
Driver Programing Interface—82583V GbE Controller
9.2.5.16
Receive Filter Control Register - RFCTL (0x05008; RW)
Initial
Value
Field
Bit(s)
Description
iSCSI Disable
Disable the iSCSI filtering.
ISCSI_DIS
0
0b
iSCSI Dword Count
ISCSI_DWC
5:1
0x0
This field indicates the Dword count of the iSCSI header, which is used
for packet split mechanism.
NFS Write Disable
Disable filtering of NFS write request headers.
NFSW_DIS
NFSR_DIS
6
7
0b
0b
NFS Read Disable
Disable filtering of NFS read reply headers.
NFS Version
00b = NFS version 2.
01b = NFS version 3.
10b = NFS version 4.
11b = Reserved for future use.
NFS_VER
9:8
00b
IPv6 Disable.
Disable IPv6 packet filtering.
IPv6_dis
10
11
0 b
0b
IPv6 Xsum Disable
Disable XSUM on IPv6 packets.
IP6Xsum_dis
ACK Accelerate Disable
ACKDIS
12
13
0 b
0b
When this bit is set, the 82583V does not accelerate interrupt on TCP
ACK packets.
ACK data Disable
1b = The 82583V recognizes ACK packets according to the ACK bit in
the TCP header + No –CP data
0b = The 82583V recognizes ACK packets according to the ACK bit
only.
ACKD_DIS
This bit is relevant only if the ACKDIS bit is not set.
IP Fragment Split Disable
When this bit is set, the header of IP fragmented packets are not set.
IPFRSP_DIS
EXSTEN
14
15
0b
0b
Extended status Enable
When the EXSTEN bit is set or when the packet split receive
descriptor is used, the 82583V writes the extended status to the Rx
descriptor.
Reserved
Reserved
16
17
0b
0b
Reserved.
Reserved.
Reserved
Reserved
31:18
0x0
Should be written with 0x0 to ensure future compatibility.
9.2.5.17
Multicast Table Array - MTA[127:0] (0x05200-0x053FC; RW)
Initial
Value
Field
Bit(s)
Description
Word-wide bit vector specifying 32 bits in the multicast address
filter table.
Bit Vector
31:0
X
241
82583V GbE Controller—Driver Programing Interface
There is one register per 32 bits of the multicast address table for a total of 128
registers (thus the MTA[127:0] designation). The size of the word array depends on the
number of bits implemented in the multicast address table. Software must mask to the
desired bit on reads and supply a 32-bit word on writes.
Note:
Note:
All accesses to this table must be 32-bit.
These registers' addresses have been moved from where they were located in previous
devices. However, for backwards compatibility, these registers can also be accessed at
their alias offsets of 0x00200-0x003FC.
Figure 43 shows the multicast lookup algorithm. The destination address shown
represents the internally stored ordering of the received DA. Note that bit 0 indicated in
this diagram is the first on the wire.
Destination Address
47:40 39:32 31:24 23:16 15:8
7:0
bank[1:0]
Multicast Table Array
32 x 128
word
(4096 bit vector)
pointer[11:5]
?
...
...
bit
pointer[4:0]
Figure 43.
9.2.5.18
Multicast Table Array Algorithm
Receive Address Low - RAL (0x05400 + 8*n; RW)
While "n" is the exact unicast/multicast address entry and it is equals to 0,1,…15.
Initial
Value
Field
Bit(s)
Description
Receive Address Low
The lower 32 bits of the 48-bit Ethernet address.
RAL
31:0
X
242
Driver Programing Interface—82583V GbE Controller
These registers contain the lower bits of the 48-bit Ethernet address. All 32 bits are
valid.
If the NVM is present the first register (RAL0) is loaded from the NVM.
Note:
These registers' addresses have been moved from where they were located in previous
devices. However, for backwards compatibility, these registers can also be accessed at
their alias offsets of 0x0040-0x000BC.
9.2.5.19
Receive Address High - RAH (0x05404 + 8*n; RW)
While "n" is the exact unicast/Multicast address entry and it is equals to 0,1,…15
Initial
Value
Field
Bit(s)
Description
Receive Address High
The upper 16 bits of the 48-bit Ethernet address.
RAH
15:0
X
X
Address Select
Selects how the address is to be used. Decoded as follows:
00b = Destination address (must be set to this in normal mode).
01b = Source address.
ASEL
17:16
10b = Reserved.
11b = Reserved.
Reserved
Reads as 0x0. Ignored on write.
Reserved
AV
30:18
31
0x0
X
Address Valid
Cleared after master reset. If the NVM is present, the Address Valid
field of Receive Address Register 0 are set to 1b after a software or
PCI reset or NVM read.
In entries 0-14 this bit is cleared by master reset. The AV bit of entry
15 is cleared by Internal Power On Reset.
These registers contain the upper bits of the 48-bit Ethernet address. The complete
address is {RAH, RAL}. AV determines whether this address is compared against the
incoming packet. AV is cleared by a master reset in entries 0-14, and on Internal Power
On Reset in entry 15.
ASEL enables the device to perform special filtering on receive packets.
Note:
Note:
The first receive address register (RAR0) is also used for exact match pause frame
checking (DA matches the first register). Therefore RAR0 should always be used to
store the individual Ethernet MAC address of the 82583V.
These registers' addresses have been moved from where they were located in previous
devices. However, for backwards compatibility, these registers can also be accessed at
their alias offsets of 0x0040-0x000BC.
After reset, if the NVM is present, the first register (Receive Address Register 0) is
loaded from the IA field in the NVM, its Address Select field will be 00b, and its Address
Valid field will be 1b. If no NVM is present the Address Valid field for n=0b will be 0b.
The Address Valid field for all of the other registers is 0b.
Note:
The software device driver can use only entries 0-14.
243
82583V GbE Controller—Driver Programing Interface
9.2.5.20
VLAN Filter Table Array - VFTA[127:0] (0x05600-0x057FC; RW)
Initial
Value
Field
Bit(s)
31:0
Description
Bit Vector
X
Double word-wide bit vector specifying 32 bits in the VLAN filter table.
There is one register per 32 bits of the VLAN Filter table. The size of the word array
depends on the number of bits implemented in the VLAN filter table. Software must
mask to the desired bit on reads and supply a 32-bit word on writes.
Note:
Note:
All accesses to this table must be 32-bit.
The algorithm for VLAN filtering via the VFTA is identical to that used for the multicast
table array.
These registers' addresses have been moved from where they were located in previous
devices. However, for backwards compatibility, these registers can also be accessed at
their alias offsets of 0x00600-0x006FC
244
Driver Programing Interface—82583V GbE Controller
9.2.6
Transmit Register Descriptions
9.2.6.1
Transmit Control Register - TCTL (0x00400; RW)
Initial
Value
Field
Bit(s)
Description
Reserved
Write as 0b for future compatibility.
Reserved
0
1
2
0b
Enable
The transmitter is enabled when this bit is set to 1b. Writing this bit to
0b stops transmission after any in progress packets are sent. Data
remains in the transmit FIFO until the device is re-enabled. Software
should combine this with a reset if the packets in the FIFO need to be
flushed.
EN
0b
0b
Reserved
Reserved
Reads as 0b. Should be written to 0b for future compatibility.
Pad short packets (with valid data, NOT padding symbols).
0b = do not pad
1b = pad.
Padding makes the packet 64 bytes. This is not the same as the
minimum collision distance.
PSP
3
1b
If padding of short packet is allowed, the value in TX descriptor length
field should be not less than 17 bytes.
Collision Threshold
This determines the number of attempts at re-transmission prior to
giving up on the packet (not including the first transmission attempt).
While this can be varied, it should be set to a value of 15 in order to
comply with the IEEE specification requiring a total of 16 attempts.
The Ethernet back-off algorithm is implemented and clamps to the
maximum number of slot times after 10 retries. This field only has
meaning while in half-duplex operation.
CT
11:4
0x0
Collision Distance
Specifies the minimum number of byte times that must elapse for
proper CSMA/CD operation. Packets are padded with special symbols,
not valid data bytes. Hardware checks and pads to this value plus one
byte even in full-duplex operation.
COLD
21:12
22
0b
0b
Software XOFF Transmission
When set to 1b, the device schedules the transmission of an XOFF
(PAUSE) frame using the current value of the pause timer. This bit self
clears upon transmission of the XOFF frame.
SWXOFF
Packet Burst Enable
PBE
23
24
0b
0b
The 82583V does not support packet bursting for 1 Gb/s half-duplex
transmit operation. This bit must be set to 0b.
Re-Transmit on Late Collision
Enables the device to re-transmit on a late collision event. This bit is
ignored in full-duplex mode.
RTLC
UNORTX
25
Under run No Re-Transmit
Tx Descriptor Minimum Threshold
Multiple Request Support
TXDSCMT
27:26
This bit defines the number of read requests the 82583V issues for
transmit data. When set to 0b, the 82583V submits only one request
at a time, When set to 1b, the 82583V might submit up to four
concurrent requests. The software device driver must not modify this
register when the Tx head register is not equal to the tail register.
MULR
28
1b
This bit is loaded from the NVM word 0x24/0x14.
245
82583V GbE Controller—Driver Programing Interface
Initial
Value
Field
RRTHRESH
Reserved
Bit(s)
Description
Read Request Threshold
These bits define the threshold size for the intermediate buffer to
determine when to send the read command to the packet buffer.
Threshold is defined as follows:
RRTHRESH = 00b threshold = 2 lines of 16 bytes
RRTHRESH = 01b threshold = 4 lines of 16 bytes
RRTHRESH = 10b threshold = 8 lines of 16 bytes
30:29
01b
RRTHRESH = 11b threshold = No threshold (transfer data after all of
the request is in the RFIFO)
Reserved
31
0b
Reads as 0b. Should be written to 0b for future compatibility.
Two fields deserve special mention: CT and COLD. Software might choose to abort
packet transmission in less than the Ethernet mandated 16 collisions. For this reason,
hardware provides CT.
Wire speeds of 1000 Mb/s result in a very short collision radius with traditional
minimum packet sizes. COLD specifies the minimum number of bytes in the packet to
satisfy the desired collision distance. It is important to note that the resulting packet
has special characters appended to the end. These are NOT regular data characters.
Hardware strips special characters for packets that go from 1000 Mb/s environments to
100 Mb/s environments. Note that the hardware evaluates this field against the packet
size in full duplex as well.
Note:
While 802.3x flow control is only defined during full duplex operation, the sending of
pause frames via the SWXOFF bit is not gated by the duplex settings within the device.
Software should not write a 1b to this bit while the device is configured for half-duplex
operation.
RTLC configures the 82583V to perform retransmission of packets when a late collision
is detected. Note that the collision window is speed dependent: 64 bytes for 10/
100 Mb/s and 512 bytes for 1000 Mb/s operation. If a late collision is detected when
this bit is disabled, the transmit function assumes the packet is successfully
transmitted. This bit is ignored in full-duplex mode.
9.2.6.2
Transmit IPG Register - TIPG (0x00410; RW)
Initial
Value
Field
Bit(s)
Description
IPG Transmit Time
Measured in increments of the MAC clock:
8 ns @ 1 Gb/s
IPGT
9:0
0x8
80 ns @ 100 Mb/s
800 ns @ 10 Mb/s.
IPG Receive Time 1
Measured in increments of the MAC clock:
8 ns @ 1 Gb/s
80 ns @ 100Mb/s
IPGR1
IPGR2
19:10
29:20
0x8
0x6
800 ns @ 10 Mb/s.
IPG Receive Time 2
Measured in increments of the MAC clock:
8 ns @ 1 Gb/s
80 ns @ 100 Mb/s
800 ns @ 10 Mb/s.
246
Driver Programing Interface—82583V GbE Controller
Initial
Value
Field
Bit(s)
Description
Reserved
Reads as 0b. Should be written to 0b for future compatibility.
Reserved
31:30
0x0
This register controls the Inter Packet Gap (IPG) timer. IPGT specifies the IPG length for
back-to-back transmissions. IPGR1 contains the length of the first part of the IPG time
for non back-to-back transmissions. During this time, the IPG counter restarts if any
carrier sense event occurs. Once the time specified by IPGR1 has elapsed, carrier sense
does not affect the IPG counter. IPGR2 specifies the total IPG time for non back-to-back
transmissions. According to the IEEE 802.3 spec, IPGR1 should be 2/3 of IPGR2. IPGR1
and IPGR2 are significant only for half-duplex operation.
Note:
The actual time waited for IPGT and IPGR2 is 6 MAC clocks (48 ns @ 1 Gb/s) longer
than the value programmed in the register. This is due to the implementation of the
asynchronous interface between the internal DMA and MAC engines. Therefore, the
suggested value that software should program into this register is 0x00602006. This
corresponds to: IPGT = 6 (6+6 = total delay of 12); IPGR1 = 8; and IPGR2 = 6 (6+6 =
total delay of 12). Also, it should be noted that this six MAC clock delay is longer than
implementations. For previous implementations, the actual time waited for any of the
IPG timers was two MAC clocks (16 ns) longer than the value programmed in the
register. Thus, for previous implementations, the suggested value for software to
program this register was 0x00A00200A.
9.2.6.3
Adaptive IFS Throttle - AIT (0x00458; RW)
Initial
Value
Field
Bit(s)
Description
Adaptive IFS Value
This value is in units of 8 ns.
AIFS
Reserved
15:0
31:16
0x0000
0x0000
This field should be written with 0x0.
Adaptive IFS throttles back-to-back transmissions in the transmit packet buffer and
delays their transfer to the CSMA/CD transmit function, and thus can be used to delay
the transmission of back-to-back packets on the wire. Normally, this register should be
set to zero. However, if additional delay is desired between back-to-back transmits,
then this register can be set with a value greater than zero.
The Adaptive IFS field provides a similar function to the IPGT field in the TIPG register
(see Section 9.2.6.2). However, it only affects the initial transmission timing, not re-
transmission timing.
Note:
If the value of the Adaptive IFS field is less than the IPG Transmit Time field in the
Transmit IPG registers then it has no effect, as the chip selects the maximum of the two
values.
9.2.6.4
Transmit Descriptor Base Address Low - TDBAL (0x03800; RW)
Initial
Value
Field
Bit(s)
Description
0
3:0
31:4
0x0
Ignored on writes. Returns 0x0 on reads.
Transmit Descriptor Base Address Low
TDBAL
X
247
82583V GbE Controller—Driver Programing Interface
This register contains the lower bits of the 64-bit descriptor base address. The lower
four bits are ignored. The transmit descriptor base address must point to a 16-byte
aligned block of data.
Note:
This register’s address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00420.
9.2.6.5
Transmit Descriptor Base Address High - TDBAH (0x03804; RW)
Initial
Value
Field
TDBAH
Bit(s)
31:0
Description
X
Transmit Descriptor Base Address [63:32]
This register contains the upper 32 bits of the 64-bit descriptor base address.
Note:
This register’s address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00424.
9.2.6.6
Transmit Descriptor Length - TDLEN (0x03808; RW)
Initial
Field
Bit(s)
6:0
Description
Ignore on write. Reads back as 0x0.
Value
0
0x0
LEN
19:7
0x0
0x0
Descriptor Length
Reserved
31:20
Reads as 0x0. Should be written to 0x0.
This register contains the descriptor length and must be 128-byte aligned.
Note:
This register’s address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00428.
9.2.6.7
Transmit Descriptor Head - TDH (0x03810; RW)
Initial
Field
Bit(s)
15:0
31:16
Description
Value
TDH
Reserved
0x0
0x0
Transmit Descriptor Head
Reserved
Should be written with 0x0.
This register contains the head pointer for the transmit descriptor ring. It points to a
16-byte datum. Hardware controls this pointer. The only time that software should
write to this register is after a reset (hardware reset or CTRL.RST) and before enabling
the transmit function (TCTL.EN).
Note:
If software were to write to this register while the transmit function was enabled, the
on-chip descriptor buffers might be invalidated and the hardware could be become
unstable.
248
Driver Programing Interface—82583V GbE Controller
Note:
This register’s address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00430.
9.2.6.8
Transmit Descriptor Tail - TDT (0x03818; RW)
Initial
Value
Field
Bit(s)
Description
TDT
Reserved
15:0
31:16
0x0
0x0
Transmit Descriptor Tail
Reads as 0. Should be written to 0 for future compatibility.
This register contains the tail pointer for the transmit descriptor ring. It points to a 16-
byte datum. Software writes the tail pointer to add more descriptors to the transmit
ready queue. Hardware attempts to transmit all packets referenced by descriptors
between head and tail.
Note:
This register’s address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00438.
9.2.6.9
Transmit Arbitration Count - TARC (0x03840; RW)
Initial
Value
Field
Bit(s)
Description
Reserved
Reserved
Reserved
ENABLE
6:0
0x3
Writing 0x0 to this register is not allowed.
9:7
10
0b
1b
Reserved
Descriptor Enable
The Enable bit of transmit queue 0 should always be set.
Reserved
Reserved
Reserved
26:11
30:27
31
0x0
Reserved, reads as 0. Should be written to 0 for future compatibility.
Reserved
0000b
0b
Reads as 0b. Should be written to 0b for future compatibility.
9.2.6.10
Transmit Interrupt Delay Value - TIDV (0x03820; RW)
Initial
Value
Field
Bit(s)
Description
Interrupt Delay Value
Counts in units of 1.024 microseconds. A value of 0 is not allowed.
IDV
15:0
0x0
Reserved
FPD
30:16
31
0x0
0b
Reads as 0x0. Should be written to 0x0 for future compatibility.
Flush Partial Descriptor Block
When set to 1b, ignored. Reads as 0b.
This register is used to delay interrupt notification for transmit operations by coalescing
interrupts for multiple transmitted buffers. Delaying interrupt notification helps
maximize the amount of transmit buffers reclaimed by a single interrupt. This feature
ONLY applies to transmit descriptor operations where:
1. Interrupt-based reporting is requested (RS set).
2. The use of the timer function is requested (IDE is set).
249
82583V GbE Controller—Driver Programing Interface
This feature operates by initiating a count-down timer upon successfully transmitting
the buffer. If a subsequent transmit delayed-interrupt is scheduled BEFORE the timer
expires, the timer is re-initialized to the programmed value and re-starts its count
down. When the timer expires, a transmit-complete interrupt (ICR.TXDW) is generated.
Setting the value to 0b is not allowed. If an immediate (non-scheduled) interrupt is
desired for any transmit descriptor, the descriptor IDE should be set to 0b.
The occurrence of either an immediate (non-scheduled) or absolute transmit timer
interrupt halts the TIDV timer and eliminate any spurious second interrupts.
Transmit interrupts due to a Transmit Absolute Timer (TADV) expiration or an
immediate interrupt (RS=1b, IDE=0b) cancels a pending TIDV interrupt. The TIDV
countdown timer is re-loaded but halted, though it can be re-started by processing a
subsequent transmit descriptor.
Note:
This register’s address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x00440.
Writing this register with FPD set initiates an immediate expiration of the timer, causing
a write back of any consumed transmit descriptors pending write back, and results in a
transmit timer interrupt in the ICR.
Note:
FPD is self clearing.
9.2.6.11
Transmit Descriptor Control - TXDCTL (0x03828; RW)
Initial
Field
Bit(s)
5:0
Description
Value
PTHRESH
Rsv
0x0
Prefetch Threshold
Reserved
7:6
0x0
0x0
0x0
0x0
0x0
HTHRESH
Rsv
13:8
Host Threshold
Reserved
15:14
21:16
23:22
WTHRESH
Rsv
Write-Back Threshold
Reserved
Granularity
Units for the thresholds in this register.
0b = Cache lines
GRAN
24
0b
1b = Descriptors
Transmit Descriptor Low Threshold
Interrupt asserted when the number of descriptors pending service in
the transmit descriptor queue (processing distance from the TDT)
drops below this threshold.
LWTHRESH
31:25
0x0
This register controls the fetching and write back of transmit descriptors. The three
threshold values are used to determine when descriptors are read from and written to
host memory. The values can be in units of cache lines or descriptors (each descriptor
is 16 bytes) based on the GRAN flag.
Note:
When GRAN=1b all descriptors are written back (even if not requested).
250
Driver Programing Interface—82583V GbE Controller
PTHRESH is used to control when a prefetch of descriptors are considered. This
threshold refers to the number of valid, unprocessed transmit descriptors the chip has
in its on-chip buffer. If this number drops below PTHRESH, the algorithm considers pre-
fetching descriptors from host memory. However, this fetch does not happen unless
there are at least HTHRESH valid descriptors in host memory to fetch.
Note:
HTHRESH should be given a non-zero value when ever PTHRESH is used.
WTHRESH controls the write-back of processed transmit descriptors. This threshold
refers to the number of transmit descriptors in the on-chip buffer that are ready to be
written back to host memory. In the absence of external events (explicit flushes), the
write back occurs only after at least WTHRESH descriptors are available for write back.
• Possible values:
— GRAN = 1b (descriptor granularity):
— PTHRESH = 0..47
— WTHRESH = 0..63
— HTHRESH = 0..63
— GRAN = 0 (cacheline granularity):
— PTHRESH = 0..3 (for 16 descriptors cacheline - 256 bytes)
— WTHRESH = 0..3
— HTHRESH = 0..4
Note:
Note:
For any WTHRESH value other than zero - packet and absolute timers must get a non-
zero value for the WTHRESH feature to take affect.
Since the default value for write-back threshold is zero, descriptors are normally
written back as soon as they are processed. WTHRESH must be a non-zero value to
take advantage of the write-back bursting capabilities of the 82583V.
Since write-back of transmit descriptors is optional (under the control of RS bit in the
descriptor), not all processed descriptors are counted with respect to WTHRESH.
Descriptors start accumulating after a descriptor with RS is set. Furthermore, with
transmit descriptor bursting enabled, some descriptors are written back that did not
have RS set in their respective descriptors.
Note:
Leaving this value at its default causes descriptor processing to be similar to previous
devices.
As descriptors are transmitted the number of descriptors waiting in the transmit
descriptor queue decreases as noted by the transmit descriptor head and tail positions
in the circular queue. When the number of waiting descriptors drops to LWTHRESH (the
head and tail positions are sufficiently close to one another) an interrupt is asserted.
LWTHRESH controls the number of descriptors in transmit ring, at which a transmit
descriptor-low interrupt (ICR.TXD_LOW) is reported. This might enable software to
operate more efficiently by maintaining a continuous addition of transmit work,
interrupting only when the hardware nears completion of all submitted work.
LWTHRESH specifies a multiple of eight descriptors. An interrupt is asserted when the
number of descriptors available transitions from (threshold level=8*LWTHRESH)+1 ‡
(threshold level=8*LWTHRESH). Setting this value to zero disables this feature.
251
82583V GbE Controller—Driver Programing Interface
9.2.6.12
Transmit Absolute Interrupt Delay Value-TADV (0x0382C; RW)
Initial
Value
Field
Bit(s)
Description
Interrupt Delay Value
Counts in units of 1.024 μs. (0b = disabled).
IDV
Reserved
15:0
31:16
0x0
0x0
Reads as 0x0. Should be written to 0x0 for future compatibility.
The transmit interrupt delay timer (TIDV) can be used to coalesce transmit interrupts.
However, it might be necessary to ensure that no completed transmit remains
unnoticed for too long an interval in order to ensure timely release of transmit buffers.
This register can be used to ENSURE that a transmit interrupt occurs at some pre-
defined interval after a transmit completes. Like the delayed-transmit timer, the
absolute transmit timer ONLY applies to transmit descriptor operations where
1. Interrupt-based reporting is requested (RS set).
2. The use of the timer function is requested (IDE is set).
This feature operates by initiating a count-down timer upon successfully transmitting
the buffer. When the timer expires, a transmit-complete interrupt (ICR.TXDW) is
generated. The occurrence of either an immediate (non-scheduled) or delayed transmit
timer (TIDV) expiration interrupt halts the TADV timer and eliminates any spurious
second interrupts.
Setting the value to zero, disables the transmit absolute delay function. If an
immediate (non-scheduled) interrupt is desired for any transmit descriptor, the
descriptor IDE should be set to 0b.
9.2.7
Statistic Register Descriptions
Note:
All statistics registers reset when read. In addition, they stick at 0xFFFF_FFFF when the
maximum value is reached.
Note:
Note:
For the receive statistics it should be noted that a packet is indicated as received if it
passes the device’s filters and is placed into the packet buffer memory. A packet does
not have to be DMA’d to host memory in order to be counted as received.
Due to divergent paths between interrupt-generation and logging of relevant statistics
counts, it might be possible to generate an interrupt to the system for a noteworthy
event prior to the associated statistics count actually being incremented. This is
extremely unlikely due to expected delays associated with the system interrupt-
collection and ISR delay, but might be observed as an interrupt for which statistics
values do not quite make sense. Hardware guarantees that any event noteworthy of
inclusion in a statistics count is reflected in the appropriate count within 1 μs; a small
time-delay prior to read of statistics might be necessary to avoid the potential for
receiving an interrupt and observing an inconsistent statistics count as part of the ISR.
252
Driver Programing Interface—82583V GbE Controller
9.2.7.1
CRC Error Count - CRCERRS (0x04000; R)
Initial
Value
Field
Bit(s)
31:0
Description
CEC
0x0
CRC Error Count
Counts the number of receive packets with CRC errors. In order for a packet to be
counted in this register, it must pass address filtering and must be 64 bytes or greater
(from <Destination Address> through <CRC>, inclusively) in length. If receives are not
enabled, then this register does not increment.
9.2.7.2
Alignment Error Count - ALGNERRC (0x04004; R)
Initial
Value
Field
Bit(s)
31:0
Description
AEC
0x0
Alignment Error Count
Counts the number of receive packets with alignment errors (such as the packet is not
an integer number of bytes in length). In order for a packet to be counted in this
register, it must pass address filtering and must be 64 bytes or greater (from
<Destination Address> through <CRC>, inclusively) in length. If receives are not
enabled, then this register does not increment. This register is valid only in MII mode
during 10/100 Mb/s operation.
9.2.7.3
RX Error Count - RXERRC (0x0400C; R)
Initial
Value
Field
RXEC
Bit(s)
31:0
Description
0x0
RX Error Count
Counts the number of packets received in which RX_ER was asserted by the PHY. In
order for a packet to be counted in this register, it must pass address filtering and must
be 64 bytes or greater (from <Destination Address> through <CRC>, inclusively) in
length. If receives are not enabled, then this register does not increment.
9.2.7.4
Missed Packets Count - MPC (0x04010; R)
Initial
Value
Field
Bit(s)
31:0
Description
MPC
0x0
Missed Packets Count
Counts the number of missed packets. Packets are missed when the receive FIFO has
insufficient space to store the incoming packet. This could be caused because of too
few buffers allocated, or because there is insufficient bandwidth on the IO bus. Events
setting this counter cause RXO, the receiver overrun interrupt, to be set. This register
does not increment if receives are not enabled.
Note:
Note that these packets are also counted in the Total Packets Received register as well
as in the Total Octets Received register.
253
82583V GbE Controller—Driver Programing Interface
9.2.7.5
Single Collision Count - SCC (0x04014; R)
Initial
Value
Field
Bit(s)
31:0
Description
SCC
0x0
Number of times a transmit encountered a single collision.
This register counts the number of times that a successfully transmitted packet
encountered a single collision. This register only increments if transmits are enabled
and the device is in half-duplex mode.
9.2.7.6
Excessive Collisions Count - ECOL (0x04018; R)
Initial
Field
Bit(s)
31:0
Description
Value
ECC
0x0
Number of packets with more than 16 collisions.
When 16 or more collisions have occurred on a packet, this register increments,
regardless of the value of collision threshold. If collision threshold is set below 16, this
counter won’t increment. This register only increments if transmits are enabled and the
device is in half-duplex mode.
9.2.7.7
Multiple Collision Count - MCC (0x0401C; R)
Initial
Value
Field
Bit(s)
Description
Number of times a successful transmit encountered multiple
collisions.
MCC
31:0
0x0
This register counts the number of times that a transmit encountered more than one
collision but less than 16. This register only increments if transmits are enabled and the
device is in half-duplex mode.
9.2.7.8
Late Collisions Count - LATECOL (0x04020; R)
Initial
Field
Bit(s)
31:0
Description
Value
LCC
0x0
Number of packets with late collisions.
Late collisions are collisions that occur after one slot time. This register only increments
if transmits are enabled and the device is in half-duplex mode.
254
Driver Programing Interface—82583V GbE Controller
9.2.7.9
Collision Count - COLC (0x04028; R)
Initial
Value
Field
COLC
Bit(s)
31:0
Description
0x0
Total number of collisions experienced by the transmitter.
This register counts the total number of collisions seen by the transmitter. This register
only increments if transmits are enabled and the device is in half-duplex mode. This
register applies to clear as well as secure traffic.
9.2.7.10
Defer Count - DC (0x04030; R)
Initial
Field
Bit(s)
31:0
Description
Value
CDC
0x0
Number of defer events.
This register counts defer events. A defer event occurs when the transmitter cannot
immediately send a packet due to the medium being busy either because:
• Another device is transmitting
• The IPG timer has not expired
• Hhalf-duplex deferral events
• Reception of XOFF frames
• The link is not up
This register only increments if transmits are enabled. The behavior of this counter is
slightly different in the 82583V relative to previous devices. For the 82583V, this
counter does not increment for streaming transmits that are deferred due to TX IPG.
9.2.7.11
Transmit with No CRS - TNCRS (0x04034; R)
Initial
Value
Field
TNCRS
Bit(s)
31:0
Description
0x0
Number of transmissions without a CRS assertion from the PHY.
This register counts the number of successful packet transmissions in which the CRS
input from the PHY was not asserted within one slot time of start of transmission from
the MAC. Start of transmission is defined as the assertion of TX_EN to the PHY.
The PHY should assert CRS during every transmission. Failure to do so might indicate
that the link has failed, or the PHY has an incorrect link configuration. This register only
increments if transmits are enabled. This register is only valid when the 82583V is
operating at half duplex.
255
82583V GbE Controller—Driver Programing Interface
9.2.7.12
Carrier Extension Error Count - CEXTERR (0x0403C; R)
Initial
Value
Field
Bit(s)
31:0
Description
CEXTERR
0x0
Number of packets received with a carrier extension error.
This register counts the number of packets received in which the carrier extension error
was signaled across the GMII interface. The PHY propagates carrier extension errors to
the MAC when an error is detected during the carrier extended time of a packet
reception. An extension error is signaled by the PHY by the encoding of 0x1F on the
receive data inputs while RX_ER is asserted to the MAC. This register only increments if
receives are enabled and the device is operating at 1000 Mb/s.
9.2.7.13
Receive Length Error Count - RLEC (0x04040; R)
Initial
Field
RLEC
Bit(s)
31:0
Description
Value
0x0
Number of packets with receive length errors.
This register counts receive length error events. A length error occurs if an incoming
packet passes the filter criteria but is undersized or oversized. Packets less than 64
bytes are undersized. Packets over 1522 bytes are oversized if LongPacketEnable is 0b.
Note:
The 82583V does not support setting LPE to 1b.
If receives are not enabled, this register does not increment. These lengths are based
on bytes in the received packet from <Destination Address> through <CRC>,
inclusively.
9.2.7.14
XON Received Count - XONRXC (0x04048; R)
Initial
Field
XONRXC
Bit(s)
31:0
Description
Number of XON packets received.
Value
0x0
This register counts the number of XON packets received. XON packets can use the
global address, or the station address. This register only increments if receives are
enabled.
9.2.7.15
XON Transmitted Count - XONTXC (0x0404C; R)
Initial
Field
XONTXC
Bit(s)
31:0
Description
Number of XON packets transmitted.
Value
0x0
This register counts the number of XON packets transmitted. These can be either due
to queue fullness, or due to software initiated action (using SWXOFF). This register only
increments if transmits are enabled.
256
Driver Programing Interface—82583V GbE Controller
9.2.7.16
9.2.7.17
9.2.7.18
XOFF Received Count - XOFFRXC (0x04050; R)
Initial
Value
Field
Bit(s)
31:0
Description
Number of XOFF packets received.
XOFFRXC
0x0
This register counts the number of XOFF packets received. XOFF packets can use the
global address, or the station address. This register only increments if receives are
enabled.
XOFF Transmitted Count - XOFFTXC (0x04054; R)
Initial
Field
Bit(s)
31:0
Description
Number of XOFF packets transmitted.
Value
XOFFTXC
0x0
This register counts the number of XOFF packets transmitted. These can be either due
to queue fullness, or due to software initiated action (using SWXOFF). This register only
increments if transmits are enabled.
FC Received Unsupported Count - FCRUC (0x04058; RW)
Initial
Field
FCRUC
Bit(s)
31:0
Description
Value
0x0
Number of unsupported flow control frames received.
This register counts the number of unsupported flow control frames that are received.
The FCRUC counter is incremented when a flow control packet is received that matches
either the reserved flow control multicast address (in FCAH/L) or the MAC station
address, and has a matching flow control type field match (to the value in FCT), but has
an incorrect op-code field. This register only increments if receives are enabled.
9.2.7.19
Packets Received [64 Bytes] Count - PRC64 (0x0405C; RW)
Initial
Value
Field
PRC64
Bit(s)
31:0
Description
0
Number of packets received that are 64 bytes in length.
This register counts the number of good packets received that are exactly 64 bytes
(from <Destination Address> through <CRC>, inclusively) in length. Packets that are
counted in the Missed Packet Count register are not counted in this register. This
register does not include received flow control packets and increments only if receives
are enabled.
257
82583V GbE Controller—Driver Programing Interface
9.2.7.20
9.2.7.21
9.2.7.22
9.2.7.23
Packets Received [65–127 Bytes] Count - PRC127 (0x04060; RW)
Initial
Field
PRC127
Bit(s)
31:0
Description
Value
0x0
Number of packets received that are 65-127 bytes in length.
This register counts the number of good packets received that are 65-127 bytes (from
<Destination Address> through <CRC>, inclusively) in length. Packets that are
counted in the Missed Packet Count register are not counted in this register. This
register does not include received flow control packets and increments only if receives
are enabled.
Packets Received [128–255 Bytes] Count - PRC255 (0x04064; RW)
Initial
Field
PRC255
Bit(s)
31:0
Description
Value
0x0
Number of packets received that are 128-255 bytes in length.
This register counts the number of good packets received that are 128-255 bytes (from
<Destination Address> through <CRC>, inclusively) in length. Packets that are
counted in the Missed Packet Count register are not counted in this register. This
register does not include received flow control packets and increments only if receives
are enabled.
Packets Received [256–511 Bytes] Count - PRC511 (0x04068; RW)
Initial
Value
Field
PRC511
Bit(s)
31:0
Description
0x0
Number of packets received that are 256-511 bytes in length.
This register counts the number of good packets received that are 256-511 bytes (from
<Destination Address> through <CRC>, inclusively) in length. Packets that are
counted in the Missed Packet Count register are not counted in this register. This
register does not include received flow control packets and increments only if receives
are enabled.
Packets Received [512–1023 Bytes] Count - PRC1023 (0x0406C; RW)
Initial
Field
Bit(s)
31:0
Description
Value
PRC1023
0x0
Number of packets received that are 512-1023 bytes in length.
This register counts the number of good packets received that are 512-1023 bytes
(from <Destination Address> through <CRC>, inclusively) in length. Packets that are
counted in the Missed Packet Count register are not counted in this register. This
register does not include received flow control packets and increments only if receives
are enabled.
258
Driver Programing Interface—82583V GbE Controller
9.2.7.24
Packets Received [1024 to Max Bytes] Count - PRC1522 (0x04070;
RW)
Initial
Value
Field
PRC1522
Bit(s)
31:0
Description
0x0
Number of packets received that are 1024-maximum bytes in length.
This register counts the number of good packets received that are from 1024 bytes to
the maximum (from <Destination Address> through <CRC>, inclusively) in length. The
maximum is dependent on the current receiver configuration and the type of packet
being received. If a packet is counted in the Receive Oversized Count register, it is not
counted in this register (see Section 9.2.7.36). This register does not include received
flow control packets and only increments if the packet has passed address filtering and
receives are enabled.
Due to changes in the standard for maximum frame size for VLAN tagged frames in
802.3, this device accepts packets which have a maximum length of 1522 bytes. The
RMON statistics associated with this range has been extended to count 1522 byte long
packets.
9.2.7.25
Good Packets Received Count - GPRC (0x04074; R)
Initial
Value
Field
GPRC
Bit(s)
31:0
Description
0x0
Number of good packets received (of any length).
This register counts the number of good (non-erred) packets received of any legal
length. The legal length for the received packet is defined by the value of LPE (see
Section 9.2.7.13). This register does not include received flow control packets and only
counts packets that pass filtering. This register only increments if receives are enabled.
This register does not count packets counted by the Missed Packet Count (MPC)
register.
9.2.7.26
Broadcast Packets Received Count - BPRC (0x04078; R)
Initial
Field
BPRC
Bit(s)
31:0
Description
Value
0x0
Number of broadcast packets received.
This register counts the number of good (non-erred) broadcast packets received. This
register does not count broadcast packets received when the broadcast address filter is
disabled. This register only increments if receives are enabled.
259
82583V GbE Controller—Driver Programing Interface
9.2.7.27
Multicast Packets Received Count - MPRC (0x0407C; R)
Initial
Value
Field
MPRC
Bit(s)
31:0
Description
0x0
Number of multicast packets received.
This register counts the number of good (non-erred) multicast packets received. This
register does not count multicast packets received that fail to pass address filtering nor
does it count received flow control packets. This register only increments if receives are
enabled. This register does not count packets counted by the Missed Packet Count
(MPC) register.
9.2.7.28
Good Packets Transmitted Count - GPTC (0x04080; R)
Initial
Field
GPTC
Bit(s)
31:0
Description
Number of good packets transmitted.
Value
0x0
This register counts the number of good (non-erred) packets transmitted. A good
transmit packet is considered one that is 64 or more bytes in length (from <Destination
Address> through <CRC>, inclusively) in length. This does not include transmitted flow
control packets. This register only increments if transmits are enabled. This register
does not count packets counted by the Missed Packet Count (MPC) register. The
register counts clear as well as secure packets.
9.2.7.29
9.2.7.30
Good Octets Received Count - GORCL (0x04088; R)
Good Octets Received Count - GORCH (0x0408C; R)
Initial
Value
Field
Bit(s)
Description
GORCL
GORCH
31:0
31:0
0x0
0x0
Number of good octets received – lower 4 bytes.
Number of good octets received – upper 4 bytes.
These registers make up a logical 64-bit register that counts the number of good (non-
erred) octets received. This register includes bytes received in a packet from the
<Destination Address> field through the <CRC> field, inclusively. This register must be
accessed using two independent 32-bit accesses. This register resets whenever the
upper 32 bits are read (GORCH).
In addition, it sticks at 0xFFFF_FFFF_FFFF_FFFF when the maximum value is reached.
Only packets that pass address filtering are counted in this register. This register only
increments if receives are enabled.
These octets do not include octets in received flow control packets.
260
Driver Programing Interface—82583V GbE Controller
9.2.7.31
9.2.7.32
Good Octets Transmitted Count - GOTCL (0x04090; R)
Good Octets Transmitted Count - GOTCH (0x04094; R)
Initial
Value
Field
Bit(s)
Description
GOTCL
GOTCH
31:0
31:0
0x0
0x0
Number of good octets transmitted – lower 4 bytes.
Number of good octets transmitted – upper 4 bytes.
These registers make up a logical 64-bit register that counts the number of good (non-
erred) octets transmitted. This register must be accessed using two independent 32-bit
accesses. This register resets whenever the upper 32 bits are read (GOTCH).
In addition, it sticks at 0xFFFF_FFFF_FFFF_FFFF when the maximum value is reached.
This register includes bytes transmitted in a packet from the <Destination Address>
field through the <CRC> field, inclusively. This register counts octets in successfully
transmitted packets which are 64 or more bytes in length. This register only increments
if transmits are enabled. The register counts clear as well as secure octets.
These octets do not include octets in transmitted flow control packets.
9.2.7.33
Receive No Buffers Count - RNBC (0x040A0; R)
Initial
Field
RNBC
Bit(s)
31:0
Description
Value
0x0
Number of receive no buffer conditions.
This register counts the number of times that frames were received when there were
no available buffers in host memory to store those frames (receive descriptor head and
tail pointers were equal). The packet is still received if there is space in the FIFO. This
register only increments if receives are enabled.
This register does not increment when flow control packets are received.
9.2.7.34
Receive Undersize Count - RUC (0x040A4; R)
Initial
Value
Field
Bit(s)
31:0
Description
Number of receive undersize errors.
RUC
0x0
This register counts the number of received frames that passed address filtering, and
were less than minimum size (64 bytes from <Destination Address> through <CRC>,
inclusively), and had a valid CRC. This register only increments if receives are enabled.
261
82583V GbE Controller—Driver Programing Interface
9.2.7.35
Receive Fragment Count - RFC (0x040A8; R)
Initial
Value
Field
Bit(s)
31:0
Description
Number of receive fragment errors.
RFC
0x0
This register counts the number of received frames that passed address filtering, and
were less than minimum size (64 bytes from <Destination Address> through <CRC>,
inclusively), but had a bad CRC (this is slightly different from the Receive Undersize
Count register). This register only increments if receives are enabled.
9.2.7.36
Receive Oversize Count - ROC (0x040AC; R)
Initial
Value
Field
Bit(s)
31:0
Description
Number of receive oversize errors.
ROC
0x0
This register counts the number of received frames that passed address filtering, and
were greater than maximum size. Packets over 1522 bytes are oversized if LPE is 0b.
Note:
The 82583V does not support setting LPE to 1b.
If receives are not enabled, this register does not increment. These lengths are based
on bytes in the received packet from <Destination Address> through <CRC>,
inclusively.
9.2.7.37
Receive Jabber Count - RJC (0x040B0; R)
Initial
Field
Bit(s)
31:0
Description
Number of receive jabber errors.
Value
RJC
0x0
This register counts the number of received frames that passed address filtering, and
were greater than maximum size and had a bad CRC (this is slightly different from the
Receive Oversize Count register).
Packets over 1522 bytes are oversized if LPE is 0b.
Note:
The 82583V does not support setting LPE to 1b.
If receives are not enabled, this register does not increment. These lengths are based
on bytes in the received packet from <Destination Address> through <CRC>,
inclusively.
262
Driver Programing Interface—82583V GbE Controller
9.2.7.38
9.2.7.39
Total Octets Received - TORL (0x040C0; R)
Total Octets Received - TORH (0x040C4; R)
Initial
Value
Field
Bit(s)
Description
TORL
TORH
31:0
31:0
0x0
0x0
Number of total octets received – lower 4 bytes.
Number of total octets received – upper 4 bytes.
These registers make up a logical 64-bit register that counts the total number of octets
received. This register must be accessed using two independent 32-bit accesses. This
register resets whenever the upper 32 bits are read (TORH). In addition, it sticks at
0xFFFF_FFFF_FFFF_FFFF when the maximum value is reached.
All packets received have their octets summed into this register, regardless of their
length, whether they are erred, or whether they are flow control packets. This register
includes bytes received in a packet from the <Destination Address> field through the
<CRC> field, inclusively. This register only increments if receives are enabled.
Note:
Broadcast rejected packets are counted in this counter (in contradiction to all other
rejected packets that are not counted).
9.2.7.40
Total Octets Transmitted - TOT (0x040C8; RW)
Initial
Value
Field
Bit(s)
Description
TOTL
TOTH
31:0
31:0
0x0
0x0
Number of total octets transmitted – lower 4 bytes.
Number of total octets transmitted – upper 4 bytes.
These registers make up a logical 64-bit register that counts the total number of octets
transmitted. This register must be accessed using two independent 32-bit accesses.
This register resets whenever the upper 32 bits are read (TOTH). In addition, it sticks
at 0xFFFF_FFFF_FFFF_FFFF when the maximum value is reached.
All transmitted packets have their octets summed into this register, regardless of their
length or whether they are flow control packets. This register includes bytes
transmitted in a packet from the <Destination Address> field through the <CRC> field,
inclusively.
Octets transmitted as part of partial packet transmissions (for example, collisions in
half-duplex mode) are not included in this register. This register only increments if
transmits are enabled.
263
82583V GbE Controller—Driver Programing Interface
9.2.7.41
Total Packets Received - TPR (0x040D0; RW)
Initial
Value
Field
Bit(s)
31:0
Description
Number of all packets received.
TPR
0x0
This register counts the total number of all packets received. All packets received are
counted in this register, regardless of their length, whether they are erred, or whether
they are flow control packets. This register only increments if receives are enabled.
Note:
Broadcast rejected packets are counted in this counter (in contradiction to all other
rejected packets that are not counted).
9.2.7.42
Total Packets Transmitted - TPT (0x040D4; RW)
Initial
Value
Field
Bit(s)
31:0
Description
Number of all packets transmitted.
TPT
0x0
This register counts the total number of all packets transmitted. All packets transmitted
will be counted in this register, regardless of their length, or whether they are flow
control packets.
Partial packet transmissions (for example, collisions in half-duplex mode) are not
included in this register. This register only increments if transmits are enabled. This
register counts all packets, including standard packets and secure packets.
9.2.7.43
Packets Transmitted [64 Bytes] Count - PTC64 (0x040D8; RW)
Initial
Value
Field
PTC64
Bit(s)
31:0
Description
0x0
Number of packets transmitted that are 64 bytes in length.
This register counts the number of packets transmitted that are exactly 64 bytes (from
<Destination Address> through <CRC>, inclusively) in length. Partial packet
transmissions (for example, collisions in half-duplex mode) are not included in this
register. This register does not include transmitted flow control packets (which are 64
bytes in length). This register only increments if transmits are enabled. This register
counts all packets, including standard packets and secure packets.
9.2.7.44
Packets Transmitted [65–127 Bytes] Count- PTC127 (0x040DC; RW)
Initial
Field
PTC127
Bit(s)
31:0
Description
Value
0x0
Number of packets transmitted that are 65-127 bytes in length.
This register counts the number of packets transmitted that are 65-127 bytes (from
<Destination Address> through <CRC>, inclusively) in length. Partial packet
transmissions (for example, collisions in half-duplex mode) are not included in this
register. This register only increments if transmits are enabled. This register counts all
packets, including standard packets and secure packets.
264
Driver Programing Interface—82583V GbE Controller
9.2.7.45
9.2.7.46
9.2.7.47
Packets Transmitted [128–255 Bytes] Count - PTC255 (0x040E0; RW)
Initial
Value
Field
PTC255
Bit(s)
31:0
Description
0x0
Number of packets transmitted that are 128-255 bytes in length.
This register counts the number of packets transmitted that are 128-255 bytes (from
<Destination Address> through <CRC>, inclusively) in length. Partial packet
transmissions (for example, collisions in half-duplex mode) are not included in this
register. This register only increments if transmits are enabled. This register counts all
packets, including standard packets and secure packets.
Packets Transmitted [256–511 Bytes] Count - PTC511 (0x040E4; RW)
Initial
Field
PTC511
Bit(s)
31:0
Description
Value
0x0
Number of packets transmitted that are 256-511 bytes in length.
This register counts the number of packets transmitted that are 256-511 bytes (from
<Destination Address> through <CRC>, inclusively) in length. Partial packet
transmissions (for example, collisions in half-duplex mode) are not included in this
register. This register only increments if transmits are enabled. This register counts all
packets, including standard and secure packets.
Packets Transmitted [512–1023 Bytes] Count - PTC1023 (0x040E8;
RW)
Initial
Field
Bit(s)
31:0
Description
Value
PTC1023
0x0
Number of packets transmitted that are 512-1023 bytes in length.
This register counts the number of packets transmitted that are 512-1023 bytes (from
<Destination Address> through <CRC>, inclusively) in length. Partial packet
transmissions (for example, collisions in half-duplex mode) are not included in this
register. This register only increments if transmits are enabled. This register counts all
packets, including standard and secure packets.
9.2.7.48
Packets Transmitted [Greater than 1024 Bytes] Count - PTC1522
(0x040EC; RW)
Initial
Field
Bit(s)
31:0
Description
Value
PTC1522
0x0
Number of packets transmitted that are 1024 or more bytes in length.
This register counts the number of packets transmitted that are 1024 or more bytes
(from <Destination Address> through <CRC>, inclusively) in length. Partial packet
transmissions (for example, collisions in half-duplex mode) are not included in this
register. This register only increments if transmits are enabled.
265
82583V GbE Controller—Driver Programing Interface
Due to changes in the standard for maximum frame size for VLAN tagged frames in
802.3, this device transmits packets that have a maximum length of 1522 bytes. The
RMON statistics associated with this range has been extended to count 1522 byte long
packets. This register counts all packets, including standard and secure packets.
9.2.7.49
9.2.7.50
9.2.7.51
Multicast Packets Transmitted Count - MPTC (0x040F0; RW)
Initial
Value
Field
MPTC
Bit(s)
31:0
Description
0x0
Number of multicast packets transmitted.
This register counts the number of multicast packets transmitted. This register does
not include flow control packets and increments only if transmits are enabled. Counts
clear as well as secure traffic.
Broadcast Packets Transmitted Count - BPTC (0x040F4; RW)
Initial
Value
Field
BPTC
Bit(s)
31:0
Description
0x0
Number of broadcast packets transmitted count.
This register counts the number of broadcast packets transmitted. This register only
increments if transmits are enabled. This register counts all packets, including standard
and secure packets.
TCP Segmentation Context Transmitted Count - TSCTC (0x040F8; RW)
Initial
Value
Field
TSCTC
Bit(s)
31:0
Description
0x0
Number of TCP Segmentation contexts transmitted count.
This register counts the number of TCP segmentation offload transmissions and
increments once the last portion of the TCP segmentation context payload is
segmented and loaded as a packet into the on-chip transmit buffer. Note that it is not a
measurement of the number of packets sent out (covered by other registers). This
register only increments if transmits and TCP segmentation offload are enabled.
9.2.7.52
TCP Segmentation Context Transmit Fail Count - TSCTFC (0x040FC;
RW)
Initial
Value
Field
Bit(s)
Description
Number of TCP segmentation contexts where the device failed to
transmit the entire data payload.
TSCTFC
31:0
0x0
This register counts the number of TCP segmentation offload requests to the hardware
that failed to transmit all data in the TCP segmentation context payload. There is no
indication by hardware of how much data was successfully transmitted. Only one failure
event is logged per TCP segmentation context. Failures could be due to Paylen errors.
This register will only increment if transmits are enabled.
266
Driver Programing Interface—82583V GbE Controller
9.2.7.53
Interrupt Assertion Count- IAC (0x04100; R)
Initial
Value
Field
Bit(s)
0-31
Description
IAC
0x0
This is a count of the Legacy interrupt assertions that have occurred.
This counter counts the total number of interrupts generated in the system.
9.2.8
PHY Registers
PHY registers can be accessed by using MDIC as described in Section 9.2.2.7
82583V PHY Register Summary
Table 51.
Alias
Offset
Link to
Page
Category
Offset
Abbreviation
Name
RW
Any Page,
Register 0
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
Control Register
Status Register
PHY Identifier 1
PHY Identifier 2
page 269
page 271
page 271
page 272
page 272
page 274
page 275
page 276
page 276
page 277
page 278
page 279
page 279
page 281
page 282
page 283
page 284
Any Page,
Register 1
Any Page,
Register 2
Any Page,
Register 3
Any Page,
Register 4
Auto-Negotiation Advertisement
Register
Any Page,
Register 5
Link Partner Ability Register - Base Page
Auto-Negotiation Expansion Register
Next Page Transmit Register
Any Page,
Register 6
Any Page,
Register 7
Any Page,
Register 8
Link Partner Next Page Register
1000BASE-T Control Register
1000BASE-T Status Register
Any Page,
Register 9
Any Page,
Register 10
Any Page,
Register 15
Extended Status Register
Page 0,
Register 16
Copper Specific Control Register 1
Copper Specific Status Register 1
Page 0,
Register 17
Page 0,
Register 18
Copper Specific Interrupt Enable
Register
Page 0,
Register 19
Copper Specific Status Register 2
Copper Specific Control Register 3
Page 0,
Register 20
267
82583V GbE Controller—Driver Programing Interface
Alias
Offset
Link to
Category
Offset
Page 0,
Abbreviation
Name
RW
Page
PHY
Receive Error Counter Register
Page Address
page 284
Register 21
Any Page,
Register 22
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
PHY
page 285
page 285
page 286
page 287
page 287
page 287
page 288
page 288
page 289
page 289
page 292
page 293
page 294
page 295
page 295
page 295
Page 0,
Register 25
OEM Bits
Page 0,
Register 26
Copper Specific Control Register 2
Bias Setting Register 1
Page 0,
Register 29
Page 0,
Register 30
Bias Setting Register 2
Page 2,
Register 16
MAC Specific Control Register 1
MAC Specific Interrupt Enable Register
MAC Specific Status Register
MAC Specific Control Register 2
LED[3:0] Function Control Register
LED[3:0] Polarity Control Register
LED Timer Control Register
Page 2,
Register 18
Page 2,
Register 19
Page 2,
Register 21
Page 3,
Register 16
Page 3,
Register 17
Page 3,
Register 18
Page 3,
Register 19
LED[5:4] Function Control and Polarity
Register
Page 5,
Register 20
1000 BASE-T Pair Skew Register
1000 BASE-T Pair Swap and Polarity
CRC Counters
Page 5,
Register 21
Page 6,
Register 17
268
Driver Programing Interface—82583V GbE Controller
9.2.8.1
Control Register (Any Page), PHY Address 01; Register 0
Bits
Field
Mode
HW Rst
SW Rst
Description
PHY Software Reset.
Writing a 1b to this bit causes the PHY state
machines to be reset. When the reset operation
completes, this bit is automatically cleared to 0b.
The reset occurs immediately.
15
Reset
R/W, SC
0x0
SC
1b = PHY reset.
0b = Normal operation.
When loopback is activated, the transmitter data
presented on TXD is looped back to RXD internally.
The link is broken when loopback is enabled.
Loopback speed is determined by registers
21_2.2:0.
14
Loopback
R/W
0x0
0x0
1b = Enable loopback.
0b = Disable loopback.
Changes to this bit are disruptive to the normal
operation; therefore, any changes to these
registers must be followed by a software reset to
take effect. A write to this register bit does not
take effect until any one of the following also
occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation (bit 6,
13).
Speed
Select (LSB)
13
R/W
0x0
Update
•
11b = Reserved.
10b = 1000 Mb/s.
01b = 100 Mb/s.
00b = 10 Mb/s.
Changes to this bit are disruptive to the normal
operation. A write to this register bit does not take
effect until any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
Auto-
Negotiation
Enable
If register 0.12 is set to 0b and speed is manually
forced to 1000 Mb/s in registers 0.13 and 0.6, then
auto- negotiation is still enabled and only
1000BASE-T full-duplex is advertised if register 0.8
is set to 1b, and 1000BASE-T half-duplex is
advertised if register 0.8 is set to 0b. Registers
4.8:5 and 9.9:8 are ignored. Auto-negotiation is
mandatory per IEEE for proper operation in
1000BASE-T.
12
R/W
0x1
Update
1b = Enable auto-negotiation process.
0b = Disable auto-negotiation process.
269
82583V GbE Controller—Driver Programing Interface
Bits
Field
Mode
HW Rst
SW Rst
Description
Power down is controlled via register 0.11 and
16_0.2. Both bits must be set to 0b before the PHY
transitions from power down to normal operation.
When the port is switched from power down to
normal operation, a software reset and restart
auto-negotiation are performed even when bits
Reset (0_15) and Restart Auto-Negotiation (0.9)
are not set by the user. IEEE power down shuts
down the 82583V except for the GMII interface if
16_2.3 is set to 1b. If 16_2.3 is set to 0b, then the
GMII interface also shuts down. After a hardware
reset, this bit takes on the value of pd_pwrdn_a.
See
Description
11
Power Down R/W
Retain
1b = Power down.
0b = Normal operation.
When pd_pwrdn_a transitions from 1b to 0b this
bit is set to 0b. When pd_pwrdn_a transitions from
0b to 1b this bit is set to 1b.
10
9
Isolate
RO
0x0
0x0
0x0
SC
This bit has no effect.
When pd_aneg_now_a transitions from 0b to 1b
this bit is set to 1b. Auto-negotiation automatically
restarts after hardware or software reset
regardless of whether or not the Restart bit (0.9) is
set.
1b = Restart auto-negotiation process.
0b = Normal operation.
Restart
Copper
Auto-
R/W,SC
Negotiation
Changes to this bit are disruptive to the normal
operation; therefore, any changes to these
registers must be followed by a software reset to
take effect. A write to this register bit does not
take effect until any one of the following also
occurs:
Copper
Duplex
Mode
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
8
7
R/W
0x1
Update
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
1b = Full-duplex.
0b = Half-duplex.
Collision
Test
RO
0x0
0x0
This bit has no effect.
Changes to this bit are disruptive to the normal
operation; therefore, any changes to these
registers must be followed by a software reset to
take effect. A write to this register bit does not
take effect until any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation (bit 6,
13).
Speed
Selection
(MSB)
6
R/W
0x1
Update
•
11b = Reserved.
10b = 1000 Mb/s.
01b = 100 Mb/s.
00b = 10 Mb/s.
Always
0x0
5:0
Reserved
RO
Always 0x0
Reserved, always 0x0.
270
Driver Programing Interface—82583V GbE Controller
9.2.8.2
Status Register (Any Page), PHY Address 01; Register 1
Bits
Field
Mode HW Rst
SW Rst
Description
100BASE-T4. This protocol is not available.
0b = PHY not able to perform 100BASE-T4.
Always
Always
0b
15
100BASE-T4
RO
0b
100BASE-X Full-
Duplex
Always
Always
1b
14
13
12
11
10
RO
1b
1b = PHY able to perform full-duplex 100BASE-X.
1b = PHY able to perform half-duplex 100BASE-X.
1b = PHY able to perform full-duplex 10BASE-T.
1b = PHY able to perform half-duplex 10BASE-T.
100BASE-X
Half-Duplex
Always
Always
1b
RO
1b
10 Mbps Full-
Duplex
Always
Always
1b
RO
1b
10 Mbps Half-
Duplex
Always
Always
1b
RO
1b
This protocol is not available.
0b = PHY not able to perform full-duplex.
100BASE-T2
Full-Duplex
Always
Always
0b
RO
0b
This protocol is not available.
0b = PHY not able to perform half-duplex.
100BASE-T2
Half-Duplex
Always
Always
0b
9
RO
0b
Always
Always
1b
8
7
6
Extended Status RO
1b = Extended status information in register 15.
Reserved, always 0b.
1b
Always
0b
Always
0b
Reserved
RO
RO
MF Preamble
Suppression
Always
1b
Always
1b
1b = PHY accepts management frames with
preamble suppressed.
Copper Auto-
Negotiation
Complete
1b = Auto-negotiation process complete.
0b = Auto-negotiation process not complete.
5
4
3
RO
0x0
0x0
0x0
0x0
1b = Remote fault condition detected.
0b = Remote fault condition not detected.
Copper Remote
Fault
RO,
LH
Auto-
Negotiation
Ability
Always
1b
Always
1b
RO
1b = PHY able to perform auto-negotiation.
This register bit indicates when link was LED[3] since
the last read. For the current link status, either read
this register back-to-back or read register 17_0.10
Link Real Time.
Copper Link
Status
RO,
LL
2
0x0
0x0
1b = Link is up.
0b = Link is down.
1b = Jabber condition detected.
0b = Jabber condition not detected.
RO,
LH
1
0
Jabber Detect
0x0
0x0
Extended
Capability
Always
1b
Always
1b
RO
1b = Extended register capabilities.
9.2.8.3
PHY Identifier 1 (Any Page), PHY Address 01; Register 2
Bits
Field
Mode HW Rst SW Rst
Description
0x005043 0000 0000 0101 0000 0100 0011
^ ^ bit 1....................................bit 24
Organizationally
Unique Identifier Bit
3:18
15:0
RO
0x0141
0x0141 register 2. [15:0] show bits 3 to 18 of the OUI.
0000000101000001 ^ ^ bit
3...................bit18
271
82583V GbE Controller—Driver Programing Interface
9.2.8.4
PHY Identifier 2 (Any Page), PHY Address 01; Register 3
Bits
Field
Mode
HW Rst
SW Rst
Description
Organizationally Unique Identifier bits
19:24 00 0011 ^.........^ bit 19...bit 24
15:10 OUI LSB
RO
RO
Always 000011b 0x00
Always 001011b 0x00
9:4
3:0
Model Number
Model Number 001011b.
Rev Number.
Contact FAEs for information on the
device revision number.
Revision
Number
See
RO
See Description
Description
9.2.8.5
Auto-Negotiation Advertisement Register (Any Page), PHY Address
01; Register 4
Bits
Field
Mode
HW Rst
SW Rst
Description
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
15
Next Page
R/W
0x0
Update
•
Copper link goes down.
If 1000BASE-T is advertised then the required next
pages are automatically transmitted. Register 4.15
should be set to 0b if no additional next pages are
needed.
1b = Advertise.
0b = Not advertised.
Always
0b
14
13
Ack
RO
Always 0b
Reserved, must be 0b.
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
Remote Fault R/W
0x0
Update
•
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
Copper link goes down.
1b = Set Remote Fault bit.
0b = Do not set Remote Fault bit.
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
12
Reserved
R/W
0x0
Update
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
•
Copper link goes down.
Reserved bit is R/W to allow for forward
compatibility with future IEEE standards.
272
Driver Programing Interface—82583V GbE Controller
Bits
Field
Mode
HW Rst
SW Rst
Description
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
•
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
Copper link goes down.
Asymmetric
Pause
See
Description
11
R/W
Update
After a hardware reset, this bit takes on the value of
pd_config_asm_pause_a.
1b = Asymmetric pause.
0b = No asymmetric pause.
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
•
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
Copper link goes down.
See
Description
10
Pause
R/W
R/W
Update
Retain
After a hardware reset, this bit takes on the value of
pd_config_pause_a.
1b = MAC pause implemented.
0b = MAC pause not implemented.
9
100BASE-T4
0x0
0b = Not capable of 100BASE-T4.
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
•
Copper link goes down.
100BASE-TX
Full-Duplex
If register 0.12 is set to 0b and speed is manually
forced to 1000 Mb/s in registers 0.13 and 0.6, then
auto-negotiation is still enabled and only 1000BASE-
T full-duplex is advertised if register 0.8 is set to 1b;
1000BASE-T half-duplex is advertised if 0.8 is set to
0b. Registers 4.8:5 and 9.9:8 are ignored.
8
R/W
0x1
Update
Auto-negotiation is mandatory per IEEE for proper
operation in 1000BASE-T.
1b = Advertise.
0b = Not advertised.
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15)
Restart auto-negotiation is asserted (register
0.9)
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation
•
Copper link goes down.
100BASE-TX
Half-Duplex
If register 0.12 is set to 0b and speed is manually
forced to 1000 Mb/s in registers 0.13 and 0.6, then
auto-negotiation is still enabled and only 1000BASE-
T full-duplex is advertised if register 0.8 is set to 1b;
1000BASE-T half-duplex is advertised if 0.8 is set to
0b. Registers 4.8:5 and 9.9:8 are ignored.
7
R/W
0x1
Update
Auto-negotiation is mandatory per IEEE for proper
operation in 1000BASE-T.
1b = Advertise.
0b = Not advertised.
273
82583V GbE Controller—Driver Programing Interface
Bits
Field
Mode
HW Rst
SW Rst
Description
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart Auto-Negotiation is asserted (register
0.9).
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
•
Copper link goes down.
10BASE-TX
Full-Duplex
If register 0.12 is set to 0b and speed is manually
forced to 1000 Mb/s in registers 0.13 and 0.6, then
auto-negotiation is still enabled and only 1000BASE-
T full-duplex is advertised if register 0.8 is set to 1;
1000BASE-T half-duplex is advertised if 0.8 is set to
0b. Registers 4.8:5 and 9.9:8 are ignored.
6
R/W
0x1
Update
Auto-negotiation is mandatory per IEEE for proper
operation in 1000BASE-T.
1b = Advertise.
0b = Not advertised.
A write to this register bit does not take effect until
any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted (register
0.9).
•
•
Power down (register 0.11, 16_0.2) transitions
from power down to normal operation.
Copper link goes down.
10BASE-TX
Half-Duplex
If register 0.12 is set to 0b and speed is manually
forced to 1000 Mb/s in registers 0.13 and 0.6, then
auto-negotiation is still enabled and only 1000BASE-
T full-duplex is advertised if register 0.8 is set to 1b;
1000BASE-T half-duplex is advertised if 0.8 is set to
0b. Registers 4.8:5 and 9.9:8 are ignored.
5
R/W
0x1
Update
Auto-negotiation is mandatory per IEEE for proper
operation in 1000BASE-T.
1b = Advertise.
0b = Not advertised.
Selector
Field
4:0
R/W
0x01
Retain
Selector Field mode 00001 = 802.3.
9.2.8.6
Link Partner Ability Register - Base Page (Any Page), PHY Address 01;
Register 5
Bits
Field
Mode HW Rst SW Rst
Description
Received Code Word Bit 15.
15
Next Page
RO
RO
0x0
0x0
0x0
0x0
1b = Link partner capable of next page.
0b = Link partner not capable of next page.
Acknowledge Received Code Word Bit 14.
1b = Link partner received link code word.
0b = Link partner does not have next page ability.
14
Acknowledge
Remote Fault Received Code Word Bit 13.
1b = Link partner detected remote fault.
0b = Link partner has not detected remote fault.
13
12
11
Remote Fault RO
0x0
0x0
0x0
0x0
0x0
0x0
Technology
RO
Received Code Word Bit 12.
Ability Field
Received Code Word Bit 11.
1b = Link partner requests asymmetric pause.
0b = Link partner does not request asymmetric pause.
Asymmetric
RO
Pause
274
Driver Programing Interface—82583V GbE Controller
Bits
Field
Mode HW Rst SW Rst
Description
Received Code Word Bit 10.
1b = Link partner is capable of pause operation.
0b = Link partner is not capable of pause operation.
Pause
Capable
10
RO
RO
RO
0x0
0x0
0x0
0x0
0x0
0x0
Received Code Word Bit 9.
1b = Link partner is 100BASE-T4 capable.
0b = Link partner is not 100BASE-T4 capable.
100BASE-T4
Capability
9
8
Received Code Word Bit 8.
1b = Link partner is 100BASE-TX full-duplex capable.
0b = Link partner is not 100BASE-TX full-duplex capable.
100BASE-TX
Full-Duplex
Capability
Received Code Word Bit 7.
1b = Link partner is 100BASE-TX half-duplex capable.
0b = Link partner is not 100BASE-TX half-duplex
capable.
100BASE-TX
Half-Duplex
Capability
7
6
RO
0x0
0x0
0x0
0x0
Received Code Word Bit 6.
1b = Link partner is 10BASE-T full-duplex capable.
0b = Link partner is not 10BASE-T full-duplex capable.
10BASE-T
Full-Duplex
Capability
RO
RO
Received Code Word Bit 5.
1b = Link partner is 10BASE-T half-duplex capable.
0b = Link partner is not 10BASE-T half-duplex capable.
10BASE-T
Half-Duplex
Capability
5
0x0
0x0
4:0
Selector Field RO
0x00
0x00
Selector Field Received Code Word Bit 4:0.
9.2.8.7
Auto-Negotiation Expansion Register (Any Page), PHY Address 01;
Register 6
Bits
Field
Mode HW Rst SW Rst
Description
15:5 Reserved
RO
0x000
0x000
Reserved. Must be 00000000000.
Register 6.4 is not valid until the auto-negotiation
complete bit (Reg 1.5) indicates completed.
Parallel
4
1b = A fault has been detected via the parallel
detection function.
RO,LH 0x0
0x0
Detection Fault
0b = A fault has not been detected via the parallel
detection function.
Register 6.3 is not valid until the auto-negotiation
complete bit (Reg 1.5) indicates completed.
1b = Link partner is next page able.
Link Partner
Next page Able
3
2
1
0
RO
RO
0x0
0x1
0x0
0x0
0x0
0x1
0x0
0x0
0b = Link partner is not next page able.
Register 6.2 is not valid until the auto-negotiation
complete bit (Reg 1.5) indicates completed.
1b = Local device is next page able.
Local Next
Page Able
0b = Local device is not next page able.
Register 6.1 is not valid until the auto-negotiation
complete bit (Reg 1.5) indicates completed.
1b = A new page has been received.
RO,
LH
Page Received
0b = A new page has not been received.
Register 6.0 is not valid until the auto-negotiation
complete bit (Reg 1.5) indicates completed.
1b = Link partner is auto-negotiation able.
0b = Link partner is not auto-negotiation able.
Link Partner
Auto-
Negotiation
Able
RO
275
82583V GbE Controller—Driver Programing Interface
9.2.8.8
Next Page Transmit Register (Any Page), PHY Address 01; Register 7
Bits
Field
Mode HW Rst SW Rst
Description
Transmit Code Word Bit 15.
A write to register 7 implicitly sets a variable in the
auto-negotiation state machine indicating that the next
page has been loaded. A link failure clears register 7.
15
Next Page
Reserved
R/W
0x0
0x0
14
13
RO
0x0
0x1
0x0
0x1
Transmit Code Word Bit 14.
Transmit Code Word Bit 13.
Message Page
Mode
R/W
12
11
Acknowledge2 R/W
0x0
0x0
0x0
0x0
Transmit Code Word Bit 12.
Transmit Code Word Bit 11.
Toggle
RO
Message/
10:0 Unformatted
Field
R/W
0x001
0x001
Transmit Code Word Bit 10:0.
9.2.8.9
Link Partner Next Page Register (Any Page), PHY Address 01; Register
8
Bits
Field
Mode HW Rst SW Rst
Description
15
14
13
12
11
Next Page
RO
RO
RO
RO
RO
RO
0x0
0x0
Received Code Word Bit 15.
Received Code Word Bit 14.
Received Code Word Bit 13.
Received Code Word Bit 12.
Received Code Word Bit 11.
Received Code Word Bit 10:0.
Acknowledge
Message Page
Acknowledge2
Toggle
0x0
0x0
0x0
0x0
0x0
0x0
0x0
0x0
10:0 Message Unformatted Field
0x000
0x000
276
Driver Programing Interface—82583V GbE Controller
9.2.8.10
1000BASE-T Control Register (Any Page), PHY Address 01; Register 9
Bits
Field
Mode
HW Rst
SW Rst
Description
TX_CLK comes from the RX_CLK pin for jitter
testing in test modes 2 and 3. After exiting the
test mode, a hardware reset or software reset
(register 0.15) should be issued to ensure
normal operation. A restart of auto-negotiation
clears these bits.
000b = Normal mode.
001b = Test mode 1 - transmit waveform test.
15:13 Test Mode
R/W
0x0
0x0
010b = Test mode 2 - transmit jitter test
(master mode).
011b = Test mode 3 - transmit jitter test (slave
mode).
100b = Test mode 4 - transmit distortion test.
101b, 110b, 111b = Reserved.
A write to this register bit does not take effect
until any of the following also occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted
(register 0.9).
Power down (register 0.11, 16_0.2)
transitions from power down to normal
operation.
Master/Slave
Manual
12
R/W
0x0
Update
•
Configuration
Enable
•
Copper link goes down.
1b = Manual master/slave configuration.
0b = Automatic master/slave configuration.
A write to this register bit does not take effect
until any of the following also occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted
(register 0.9).
•
Power down (register 0.11, 16_0.2)
transitions from power down to normal
operation.
Master/Slave
Configuration
Value
See
Description
11
R/W
Update
•
Copper link goes down.
After a hardware reset, this bit takes on the
value of pd_config_ms_a.
1b = Manual configure as master.
0b = Manual configure as slave.
A write to this register bit does not take effect
until any of the following also occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted
(register 0.9).
•
Power down (register 0.11, 16_0.2)
transitions from power down to normal
operation.
See
Description
10
Port Type
R/W
Update
•
Copper link goes down.
Register 9.10 is ignored if register 9.12 equals
1b. After a hardware reset, this bit takes on the
value of pd_config_ms_a.
1b = Prefer multi-port device (master).
0b = Prefer single port device (slave).
277
82583V GbE Controller—Driver Programing Interface
Bits
Field
Mode
HW Rst
SW Rst
Description
A write to this register bit does not take effect
until any of the following also occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted
(register 0.9).
1000BASE-T
Full-Duplex
9
R/W
0x1
Update
•
Power down (register 0.11, 16_0.2)
transitions from power down to normal
operation.
•
Copper link goes down.
1b = Advertise.
0b = Not advertised.
A write to this register bit does not take effect
until any of the following also occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted
(register 0.9).
•
Power down (register 0.11, 16_0.2)
transitions from power down to normal
operation.
1000BASE-T
Half-Duplex
See
Description
8
R/W
R/W
Update
Retain
•
Copper link goes down.
After a hardware reset, this bit takes on the
value of pd_config_1000hd_a.
1 = Advertise.
0 = Not advertised.
7:0
Reserved
0x00
Reserved, set to 0x00.
9.2.8.11
1000BASE-T Status Register (Any Page), PHY Address 01; Register 10
Bits
Field
Mode HW Rst SW Rst
Description
This register bit clears on reads.
1b = master/slave configuration fault detected.
0 = No master/slave configuration fault detected.
Master/Slave
Configuration
Fault
RO,
15
0x0
0x0
LH
RO
RO
RO
Master/Slave
Configuration
Resolution
1b = Local PHY configuration resolved to master.
0b = Local PHY configuration resolved to slave.
14
13
12
0x0
0x0
0x0
0x0
0x0
0x0
1b = Local receiver operational.
0b = Local receiver is not operational.
Local Receiver
Status
Remote
Receiver
Status
1b = Remote receiver operational.
0b = Remote receiver not operational.
1b = Link partner is capable of 1000BASE-T full-
duplex.
0b = Link partner is not capable of 1000BASE-T full
duplex.
Link Partner
1000BASE-T
Full-Duplex
Capability
11
RO
0x0
0x0
1b = Link partner is capable of 1000BASE-T half-
duplex.
0b = Link partner is not capable of 1000BASE-T half
duplex.
Link Partner
1000BASE-T
Half-Duplex
Capability
10
RO
RO
0x0
0x0
9:8
7:0
Reserved
0x0
0x0
Reserved.
MSB of Idle Error Counter.
Idle Error
Count
RO,
SC
These register bits report the idle error count since the
last time this register was read. The counter reaches
its maximum at 11111111b and does not roll over.
0x00
0x00
278
Driver Programing Interface—82583V GbE Controller
9.2.8.12
Extended Status Register (Any Page), PHY Address 01; Register 15
Bits
Field
Mode
HW Rst
SW Rst
Description
1000BASE-X
Full-Duplex
15
RO
Always 0b Always 0b 0b = Not 1000BASE-X full-duplex capable.
Always 0b Always 0b 0b = Not 1000BASE-X half-duplex capable.
Always 1b Always 1b 1b =1000BASE-T full-duplex capable.
Always 1b Always 1b 1b =1000BASE-T half-duplex capable.
1000BASE-X
Half-Duplex
14
13
12
RO
RO
1000BASE-T
Full-Duplex
1000BASE-T
Half-Duplex
RO
RO
11:0 Reserved
0x000
0x000
Reserved, set to 0x000.
9.2.8.13
Copper Specific Control Register 1 (Page 0), PHY Address 01; Register
16
Bits
15
Field
Mode
HW Rst
0x0
SW Rst
Description
1b = Disable link pulse.
Disable
Link Pulses
R/W
0x0
0b = Enable link pulse.
Changes to these bits are disruptive to the normal
operation; therefore, any changes to these registers
must be followed by software reset to take effect.
1x, 2x,...8x is the number of times the PHY attempts
to establish GbE link before the PHY downshifts to
the next highest speed.
000b = 1x.
100b = 5x.
001b = 2x.
101b = 6x.
010b = 3x.
110b = 7x.
011b = 4x.
111b = 8x.
Downshift
Counter
14:12
R/W
0x3
Update
Changes to these bits are disruptive to the normal
operation; therefore, any changes to these registers
must be followed by software reset to take effect.
1b = Enable downshift.
0 = Disable downshift.
Downshift
Enable
11
10
R/W
0x0
0x0
Update
Retain
If link is forced to be good, the link state machine is
bypassed and the link is always up. In 1000BASE-T
mode this has no effect.
Force
CopperLink R/W
Good
1b = Force link good.
0b = Normal operation.
After a hardware reset, both bits take on the value of
pd_config_edet_a.
0xb = Off.
Energy
R/W
See
9:8
Update
Retain
Detect
Description
10b = Sense only on Receive (energy detect).
11b = Sense and periodically transmit NLP (energy
detect+TM).
When using a cable exceeding 100 meters, the
10BASE-T receive threshold must be lowered in
order to detect incoming signals.
1b = Lower 10BASE-T receive threshold.
0b = Normal 10BASE-T receive threshold.
Enable
7
Extended
Distance
R/W
0x0
279
82583V GbE Controller—Driver Programing Interface
Bits
Field
Mode
HW Rst
SW Rst
Description
Changes to these bits are disruptive to the normal
operation; therefore, any changes to these registers
must be followed by a software reset to take effect.
00b = Manual MDI configuration.
01b = Manual MDIX configuration.
10b = Reserved.
MDI
Crossover
Mode
6:5
R/W
R/W
0x3
Update
11b = Enable automatic crossover for all modes.
4
3
Reserved
Copper
Transmitter R/W
Disable
0x0
0x0
Retain
Retain
Reserved, write as 0x0.
1b = Transmitter disable.
0b = Transmitter enable.
Power down is controlled via register 0.11 and
16_0.2.
Both bits must be set to 0b before the PHY
transitions from power down to normal operation.
When the port is switched from power down to
normal operation, a software reset and restart auto-
negotiation are done even when bits Reset (0_15)
and Restart Auto-Negotiation (0.9) are not set by
the user.
Power
R/W
2
0x0
Retain
Down
IEEE power down shuts down the 82583V except for
the GMII interface if 16_2.3 is set to 1b. If 16_2.3 is
set to 0b, then the GMII interface also shuts down.
1b = Power down.
0b = Normal operation.
If polarity is disabled, then the polarity is forced to
be normal in 10BASE-T.
1b = Polarity reversal disabled.
0b = Polarity reversal enabled.
The detected polarity status is shown in Register
17_0.1 or in 1000BASE-T mode, 21_5.3:0.
Polarity
Reversal
Disable
1
0
R/W
R/W
0x0
0x0
Retain
Retain
Jabber has affect only in 10BASE-T half-duplex
mode.
1b = Disable jabber function.
0b = Enable jabber function.
Disable
Jabber
280
Driver Programing Interface—82583V GbE Controller
9.2.8.14
Copper Specific Status Register 1 (Page 0), PHY Address 01; Register
17
Bits
Field
Mode HW Rst SW Rst
Description
These status bits are valid only after resolved bit
17_0.11 equals 1b. The resolved bit is set when
auto-negotiation completes or is disabled.
11b = Reserved.
10b = 1000 Mb/s.
01b = 100 Mb/s.
00b = 10 Mb/s.
15:14 Speed
RO
RO
0x2
0x0
Retain
Retain
This status bit is valid only after resolved bit 17_0.11
equals 1b. The resolved bit is set when auto-
negotiation completes or is disabled.
13
Duplex
1b = Full-duplex.
0b = Half-duplex.
1b = Page received.
0b = Page not received.
12
11
10
Page Received
RO, LH 0x0
0x0
0x0
0x0
When Auto-Negotiation is not enabled 17_0.11
equals 1b. 1b = Resolved.
0b = Not resolved.
Speed and
Duplex
Resolved
RO
RO
0x0
0x0
1b = Link up.
0b = Link down.
Copper Link
(real time)
This is a reflection of the MAC pause resolution. This
bit is for information purposes and is not used by the
82583V. This status bit is valid only after resolved
bit 17_0.11 = 1b. The resolved bit is set when auto-
negotiation completes or is disabled.
Transmit Pause
Enabled
9
RO
0x0
0x0
1b = Transmit pause enabled.
0b = Transmit pause disable.
This is a reflection of the MAC pause resolution. This
bit is for information purposes and is not used by the
82583V. This status bit is valid only after resolved
bit 17_0.11 equals 1b. The resolved bit is set when
auto-negotiation completes or is disabled.
1b = Receive pause enabled.
0b = Receive pause disabled.
Receive Pause
Enabled
8
7
RO
RO
0x0
0x0
0x0
0x0
Reserved
Reserved, set to 0x0.
This status bit is valid only after resolved bit 17_0.11
equals 1b. The resolved bit is set when auto-
negotiation completes or is disabled. This bit is 0b or
1b depending on what is written to 16.6:5 in manual
configuration mode. Register 16.6:5 are updated
with a software reset.
MDI Crossover
Status
6
RO
0x1
Retain
1b = MDI-X.
0b = MDI.
1b = Downshift.
0b = No downshift.
Downshift
Status
5
4
3
RO
RO
RO
0x0
0x0
0x0
0x0
0x0
0x0
1b = Sleep.
0b = Active.
Copper Energy
Detect Status
1b = Copper link is up.
0b = Copper link is down.
Global Link
Status
281
82583V GbE Controller—Driver Programing Interface
Bits
Field
Mode HW Rst SW Rst
Description
Reserved, set to 0x0.
2
Reserved
RO
0x0
0x0
Polarity reversal can be disabled by writing to
Register 16_0.1. In 1000BASE-T mode, polarity of all
pairs are shown in Register 21_5.3:0.
1b = Reversed.
0b = Normal.
Polarity (real
time)
1
0
RO
0x0
0x0
1b = Jabber.
0b = No jabber.
Jabber (real
time)
RO
0x0
0x0
9.2.8.15
Copper Specific Interrupt Enable Register (Page 0), PHY Address 01;
Register 18
Bits
Field
Mode HW Rst SW Rst
Description
1b = Interrupt enable.
0b = Interrupt disable.
Auto-Negotiation Error Interrupt
Enable
15
R/W
R/W
0x0
0x0
0x0
0x0
0x0
0x0
0x0
Retain
Retain
Retain
Retain
Retain
Retain
Retain
1b = Interrupt enable.
0b = Interrupt disable.
14
13
12
11
10
9
Speed Changed Interrupt Enable
1b = Interrupt enable.
0b = Interrupt disable.
Duplex Changed Interrupt Enable R/W
1b = Interrupt enable.
0b = Interrupt disable.
Page Received Interrupt Enable
R/W
R/W
R/W
R/W
1b = Interrupt enable.
0b = Interrupt disable.
Auto-Negotiation Completed
Interrupt Enable
1b = Interrupt enable.
0b = Interrupt disable.
Link Status Changed Interrupt
Enable
1b = Interrupt enable.
0b = Interrupt disable.
Symbol Error Interrupt Enable
1b = Interrupt enable.
0b = Interrupt disable.
8
7
6
False Carrier Interrupt Enable
Reserved
R/W
R/W
R/W
0x0
0x0
0x0
Retain
Retain
Retain
Reserved, set to 0x0.
1b = Interrupt enable.
0b = Interrupt disable.
MDI Crossover Changed Interrupt
Enable
1b = Interrupt enable.
0b = Interrupt disable.
5
4
Downshift Interrupt Enable
R/W
R/W
0x0
0x0
Retain
Retain
1b = Interrupt enable.
0b = Interrupt disable.
Energy Detect Interrupt Enable
1b = Interrupt enable.
0b = Interrupt disable.
FLP Exchange Complete But No
Link Interrupt Enable
3
2
1
R/W
R/W
R/W
0x0
0x0
0x0
Retain
Retain
Retain
Reserved
Reserved, set to 0x0.
1b = Interrupt enable.
0b = Interrupt disable.
Polarity Changed Interrupt
Enable
1b = Interrupt enable.
0b = Interrupt disable.
0
Jabber Interrupt Enable
R/W
0x0
Retain
282
Driver Programing Interface—82583V GbE Controller
9.2.8.16
Copper Specific Status Register 2 (Page 0), PHY Address 01; Register
19
Bits
Field
Mode HW Rst SW Rst
Description
An error occurs if the master/slave is
not resolved, parallel detect fault, no
common HCD, or the link does not come
up after negotiation completes.
Copper Auto-Negotiation
Error
15
RO,LH 0x0
0x0
1b = Auto-negotiation error.
0b = No auto-negotiation error.
1b = Speed changed.
0b = Speed not changed.
14
13
12
11
10
9
Copper Speed Changed
Copper Duplex Changed
Copper Page Received
RO,LH 0x0
RO,LH 0x0
RO,LH 0x0
RO,LH 0x0
0x0
0x0
0x0
0x0
0x0
0x0
0x0
1b = Duplex changed.
0b = Duplex not changed.
1b = Page received.
0b = Page not received.
1b = Auto-negotiation completed.
0b = Auto-negotiation not completed.
Copper Auto-Negotiation
Completed
1b = Link status changed.
0b = Link status not changed.
Copper Link Status Changed RO,LH 0x0
1b = Symbol error.
0b = No symbol error.
Copper Symbol Error
Copper False Carrier
Reserved
RO,LH 0x0
RO,LH 0x0
1b = False carrier.
0b = No false carrier.
8
Always
0b
Always
0b
7
RO
Reserved, always set to 0b.
1b = Crossover changed.
0b = Crossover not changed.
6
MDI Crossover Changed
Downshift Interrupt
RO,LH 0x0
RO,LH 0x0
0x0
0x0
1b = Downshift detected.
0b = No downshift.
5
1b = Energy detect state changed.
4
3
Energy Detect Changed
RO,LH 0x0
RO,LH 0x0
0x0
0x0
0b = No energy detect state change
detected.
1b = FLP exchange completed but link
not established.
0b = No event detected.
FLP Exchange Complete But
No Link
2
1
Reserved
RO
0x0
0x0
0x0
Reserved, set to 0x0.
1b = Polarity changed.
0b = Polarity not changed.
Polarity Changed
RO,LH 0x0
RO,LH 0x0
1b = Jabber.
0b = No jabber.
0
Jabber
0x0
283
82583V GbE Controller—Driver Programing Interface
9.2.8.17
Copper Specific Control Register 3 (Page 0), PHY Address 01; Register
20
Bits
Field
Mode HW Rst SW Rst
Description
Reserved, write as all zeros.
15:4 Reserved
Reverse
R/W
0x000
Retain
0b = Normal transmit polarity.
1b = Reverse transmit polarity.
MDI_PLUS/
MDI_MINUS[3]
Transmit Polarity
3
R/W
0x0
Retain
Reverse
MDI_PLUS/
MDI_MINUS[2]
Transmit Polarity
0b = Normal transmit polarity.
1b = Reverse transmit polarity.
2
1
0
R/W
R/W
R/W
0x0
0x0
0x0
Retain
Retain
Retain
Reverse
MDI_PLUS/
MDI_MINUS[1]
Transmit Polarity
0b = Normal transmit polarity.
1b = Reverse transmit polarity.
Reverse
MDI_PLUS/
MDI_MINUS[0]
Transmit Polarity
0b = Normal transmit polarity.
1b = Reverse transmit polarity.
9.2.8.18
Receive Error Counter Register (Page 0), PHY Address 01; Register 21
Bits
Field
Mode HW Rst SW Rst
Description
Counter reaches its maximum at 0xFFFF and does
not roll over.
Both false carrier and symbol errors are reported.
Receive Error
Count
15:0
RO, LH 0x0000
Retain
284
Driver Programing Interface—82583V GbE Controller
9.2.8.19
9.2.8.20
Page Address (Any Page), PHY Address 01; Register 22
Bits
Field
Mode
HW Rst
SW Rst
Description
Always
0x00
Always
0x00
15:8 Reserved
RO
Reserved, always set to 0x00.
Page number.
7:0
Page Select for Registers 0 to 28 R/W
0x00
Retain
OEM Bits (Page 0), PHY Address 01; Register 25
Bits
Field
Mode
HW Rst
0x0
SW Rst
Description
15:11 Reserved
R/W
0x0
Reserved, set to 0x0.
Restart auto-negotiation. Note that this bit is self
clearing.
10
Aneg_now
R/W
0b
0b
9:7
6
Reserved
a1000_dis
Reserved
rev_aneg
Reserved
R/W
R/W
R/W
R/W
R/W
0x0
0b
0x0
Reserved, set to 0x0.
GbE disable.
Retain
0x0
5:3
2
0x0
0b
Reserved, set to 0x0.
LPLU.
Retain
0x0
1:0
0x0
Reserved, set to 0x0.
285
82583V GbE Controller—Driver Programing Interface
9.2.8.21
Copper Specific Control Register 2 (Page 0), PHY Address 01; Register
26
Bits
15
Field
Mode
HW Rst
0x0
SW Rst
Description
0b = Class B.
1b = Class A.
1000 BASE-T
Transmitter Type
R/W
Retain
When set to disabled, 1000BASE-T is not
advertised even if registers 9.9 or 9.8 are set
to 1b.
A write to this register bit does not take effect
until any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted
(register 0.9).
•
Power down (register 0.11, 16_0.2)
transitions from power down to normal
operation.
Disable
1000BASE-T
See
Description
14
R/W
Retain
•
Copper link goes down.
After a hardware reset, this bit defaults as
follows:
•
•
ps_a1000_dis_s - bit 26_0.14 - 0, 0, 1, 1.
When ps_a1000_dis_s transitions from
one to zero, this bit is set to 0b.
•
When ps_a1000_dis_s transitions from
zero to one, this bit is set to 1b.
1b = Disable 1000BASE-T advertisement.
0b = Enable 1000BASE-T advertisement.
A write to this register bit does not take effect
until any one of the following occurs:
•
•
Software reset is asserted (register 0.15).
Restart auto-negotiation is asserted
(register 0.9).
•
Power down (register 0.11, 16_0.2)
transitions from power down to normal
operation.
•
Copper link goes down.
See
Description
13
Reverse Autoneg
R/W
Retain
After a hardware reset, this bit defaults as
follows:
•
•
pd_rev_aneg_a - bit 26_0.13 - 0, 0, 1, 1.
When pd_rev_aneg_a transitions from
one to zero this bit will be set to 0b.
•
When pd_rev_aneg_a transitions from
zero to one this bit will be set to 1b.
1b = Reverse auto-negotiation.
0b = Normal auto-negotiation.
0b = Class B.
1b = Class A.
100 BASE-T
12
R/W
R/W
0x0
Retain
Retain
Transmitter Type
11:4
Reserved
0x00
Reserved, write as 0x00.
0xb = Normal operation.
10b = Select 112 ns sequence.
11b = Select 16 ns sequence.
100 MB Test
Select
3:2
R/W
0x0
Retain
1b = Force negative polarity for receive only.
0b = Normal operation.
10 BT Polarity
Force
1
0
R/W
R/W
0x0
0x0
Retain
Retain
Reserved
Reserved, write as 0x0.
286
Driver Programing Interface—82583V GbE Controller
9.2.8.22
9.2.8.23
9.2.8.24
Bias Setting Register 1 (Page 0), PHY Address 01; Register 29
Bits
Field
Mode
HW Rst
SW Rst
Description
Used to optimize PHY performance in
1000Base-T mode. Set to 0x0003 when
initializing the 82583V to improve BER
performance.
15:0
Bias setting1
R/W
Retain
Bias Setting Register 2 (Page 0), PHY Address 01; Register 30
Bits
Field
Mode
HW Rst
SW Rst
Description
Used to optimize PHY performance in
1000Base-T mode. Set to 0x0000 when
initializing the 82583V to improve BER
performance.
15:0
Bias setting2
R/W
Retain
MAC Specific Control Register 1 (Page 2), PHY Address 01; Register 16
Bits
Field
Mode
HW Rst
SW Rst
Description
1000BASE-T:
00b = ± 16 bits.
01b = ± 24 bits.
10b = ± 32 bits.
11b = ± 40 bits.
Transmit
FIFO Depth
15:14
R/W
R/W
0x0
Retain
Retain
13:10 Reserved
0x00
Reserved, set to 0x00.
Changes to this bit are disruptive to the normal
operation; therefore, any changes to these registers
must be followed by a software reset to take effect.
After a hardware reset, this bit takes on the value of
pd_pwrdn_clk125_a. When pd_pwrdn_clk125_a
transitions from one to zero this bit is set to 0b. When
pd_pwrdn_clk125_a transitions from zero to one this
bit is set to 1b.
Disable
fi_125_clk
See
Description
9
R/W
Retain
1b = fi_125_clk low.
0b = fi_125_clk toggle
After a hardware reset, this bit takes on the value of
pd_pwrdn_clk50_a. When pd_pwrdn_clk50_a
transitions from one to zero this bit is set to 0b. When
pd_pwrdn_clk50_a transitions from zero to one this
bit is set to 1b.
Disable
fi_50_clk
See
Description
8
R/W
Retain
1b = fi_50_clk low.
0b = fi_50_clk toggle.
7
Reserved
Reserved
R/W
R/W
0x1
0x0
Update
Retain
Reserved, write as 0x1.
Reserved, write as 0x00.
6:4
Changes to this bit are disruptive to the normal
operation; therefore, any changes to these registers
must be followed by a software reset to take effect.
This bit determines whether the GMII RX_CLK powers
down when register 0.11, 16_0.2 are used to power
down the 82583V or when the PHY enters the energy
detect state.
GMII
Interface
Power
Down
3
R/W
R/W
0x1
0x0
Update
Retain
1b = Always power up.
0b = Can power down.
2:0
Reserved
Reserved, write as 0x00.
287
82583V GbE Controller—Driver Programing Interface
9.2.8.25
MAC Specific Interrupt Enable Register (Page 2), PHY Address 01;
Register 18
Bits
Field
Mode HW Rst SW Rst
Description
15:8 Reserved
R/W
R/W
R/W
R/W
0x00
0x0
0x0
0x0
Retain
Retain
Retain
Retain
Reserved, set to 0x00.
1b = Interrupt enable.
0b = Interrupt disable.
7
FIFO Over/ Underflow Interrupt Enable
Reserved
6:4
3
Reserved, set to 0x0.
1b = Interrupt enable.
0b = Interrupt disable.
FIFO Idle Inserted Interrupt Enable
1b = Interrupt enable.
0b = Interrupt disable.
2
FIFO Idle Deleted Interrupt Enable
Reserved
R/W
R/W
0x0
0x0
Retain
Retain
1:0
Reserved, set to 0x0.
9.2.8.26
MAC Specific Status Register (Page 2), PHY Address 01; Register 19
Bits
Field
Mode
HW Rst
SW Rst
Description
Always
0x00
Always
0x00
15:8 Reserved
RO
Reserved, always set to 0x00.
1b = Over/underflow error.
0b = No FIFO error.
7
FIFO Over/ Underflow RO,LH 0x0
0x0
Always
0x0
Always
0x0
6:4
3
Reserved
RO
Reserved, always set to 0x0.
1b = Idle inserted.
0b = No idle inserted.
FIFO Idle Inserted
FIFO Idle Deleted
Reserved
RO,LH 0x0
RO,LH 0x0
0x0
0x0
1b = Idle deleted.
0b = Idle not deleted.
2
Always
0x0
Always
0x0
1:0
RO
Reserved, always set to 0x0.
288
Driver Programing Interface—82583V GbE Controller
9.2.8.27
MAC Specific Control Register 2 (Page 2), PHY Address 01; Register 21
Bits
Field
Mode HW Rst SW Rst
Description
Reserved, set to 0x0.
15:14 Reserved
13:12 Reserved
R/W
R/W
R/W
R/W
R/W
0x0
0x1
0x00
0x1
0x0
0x0
Update
0x00
Reserved, set to 0x1.
Reserved, set to 0x00.
Reserved, set to 0x1.
Reserved, set to 0x0.
11:7
6
Reserved
Reserved
Reserved
Update
Retain
5:4
1b = Enable block carrier extension.
0b = Disable block carrier extension.
Block Carrier
Extension Bit
3
R/W
0x0
Retain
Changes to these bits are disruptive to the normal
operation; therefore, any changes to these registers
must be followed by software reset to take effect.
MAC interface speed during link down while auto-
negotiation is enabled and TX_CLK speed bit speed
link down 1000BASE-T.
Default MAC
Interface
Speed
000b = 10 Mb/s 2.5 MHz 0 MHz.
001b = 100 Mb/s 25 MHz 0 MHz.
01xb = 1000 Mb/s 0 MHz 0 MHz.
100b = 10 Mb/s 2.5 MHz 2.5 MHz.
101b = 100 Mb/s 25 MHz 25 MHz.
110b = 1000 Mb/s 2.5 MHz 2.5 MHz.
111b = 1000 Mb/s 25 MHz 25 MHz.
2:0
R/W
0x6
Update
9.2.8.28
LED[3:0] Function Control Register (Page 3), PHY Address 01;
Register 16
Bits
Field
Mode
HW Rst
SW Rst
Description
If 16_3.11:10 is set to 11b, then 16_3.15:12 has no
effect.
0000b = Reserved.
0001b = On - link, blink - activity, off - no link.
0010b = On - link, blink - receive, off - no link.
0011b = On - activity, off - no activity
0100b = Blink - activity, off - no activity.
0101b = On - transmit, off - no transmit.
0110b = On - 10 Mb/s or 1000 Mb/s master, off.
Else
LED[3]
Control
See
Description
15:12
R/W
Retain
0111b = On - full duplex, off - half-duplex.
1000b = Force off.
1001b = Force on.
1010b = Force hi-Z.
1011b = Force blink.
11xxb = Reserved.
After a hardware reset, this bit is a function of
pd_config_led_a[1:0].
00b = 0001b.
01b = 0001b.
10b = 0111b.
11b = 0001b.
289
82583V GbE Controller—Driver Programing Interface
Bits
Field
Mode
HW Rst
SW Rst
Description
0000b = On - link, off - no link.
0001b = On - link, blink - activity, off - no link.
0010b = Reserved.
0011b = On - activity, off - no activity.
0100b = Blink - activity, off - no activity.
0101b = On - transmit, off - no transmit.
0110b = On - 10/1000 Mb/s link, off.
Else
0111b = On - 10 Mb/s link, off.
Else
LED[2]
Control
See
Description
11:8
R/W
Retain
1000b = Force off.
1001b = Force on.
1010b = Force hi-Z.
1011b = Force blink.
1100b = Mode 1 (dual LED mode).
1101b = Mode 2 (dual LED mode).
1110b = Mode 3 (dual LED mode).
1111b = Mode 4 (dual LED mode).
After a hardware reset, this bit is a function of
pd_config_led_a[1:0].
00b = 0000b.
01b = 0111b.
10b = 0001b.
11b = 0111b.
If 16_3.3:2 is set to 11b, then 16_3.7:4 has no
effect.
0000b = Reserved.
0001b = On - link, blink - activity, off - no link.
0010b = On - link, blink - receive, off - no link.
0011b = On - activity, off - no activity.
0100b = Blink - activity, off - no activity.
0101b = Reserved.
0110b = On - 100/1000 Mb/s link, off.
Else
0111b = On - 100 Mb/s link, off.
Else
LED[1]
Control
See
Description
7:4
R/W
Retain
1000b = Force off.
1001b = Force on.
1010b = Force hi-Z.
1011b = Force blink.
11xxb = Reserved.
After a hardware reset, this bit is a function of
pd_config_led_a[1:0].
00b = 0001b.
01b = 0111b.
10b = 0111b.
11b = 0111b.
290
Driver Programing Interface—82583V GbE Controller
Bits
Field
Mode
HW Rst
SW Rst
Description
0000b = On - link, off - no link.
0001b = On - link, blink - activity, off - no link.
0010b = 3 blinks - 1000 Mb/s 2 blinks - 100 Mb/s 1
blink - 10 Mb/s 0 blink - no link.
0011b = On - activity, off - no activity.
0100b = Blink - activity, off - no activity.
0101b = On - transmit, off - no transmit.
0110b = On - copper link, off.
Else
0111b = On - 1000 Mb/s link, off.
Else
LED[0]
Control
See
Description
3:0
R/W
Retain
1000b = Force off.
1001b = Force on.
1010b = Force hi-Z.
1011b = Force blink.
1100b = Mode 1 (dual LED mode).
1101b = Mode 2 (dual LED mode).
1110b = Mode 3 (dual LED mode).
1111b = Mode 4 (dual LED mode).
After a hardware reset this bit is a function of
pd_config_led_a[1:0].
00b = 1110b.
01b = 0111b.
10b = 0111b.
11b = 0111b.
291
82583V GbE Controller—Driver Programing Interface
9.2.8.29
LED[3:0] Polarity Control Register (Page 3), PHY Address 01; Register
17
Bits
Field
Mode
HW Rst
SW Rst
Description
When using two-terminal bi-color LEDs, the mixing
percentage should not be set greater than 50%.
0000b = 0%.
0001b = 12.5%.
0111b = 87.5%.
1000b = 100%.
1001b - 1111b = Reserved.
After a hardware reset, this bit is a function of
pd_config_led_a[1:0].
LED[5],
LED[3],
LED[1] Mix
Percentage
See
Description
15:12
R/W
Retain
00b = 0100b.
01b = 0100b.
10b = 1000b.
11b = 1000b.
When using two-terminal bi-color LEDs, the mixing
percentage should not be set greater than 50%.
0000b = 0%.
0001b = 12.5%.
0111b = 87.5%.
1000b = 100%.
1001b - 1111b = Reserved.
After a hardware reset, this bit is a function of
pd_config_led_a[1:0].
LED[4],
LED[2],
See
Description
11:8
R/W
Retain
LED[0] Mix
Percentage
00b = 0100b.
01b = 0100b.
10b = 1000b.
11b = 1000b.
00b = On - drive LED[3] low, off - drive LED[3]
high.
01b = On - drive LED[3] high, off - drive LED[3]
low.
10b = On - drive LED[3] low, off - tristate LED[3]
11b = On - drive LED[3] high, off - tristate LED[3]
LED[3]
Polarity
7:6
5:4
3:2
1:0
R/W
R/W
R/W
R/W
0x0
0x0
0x0
0x0
Retain
Retain
Retain
Retain
00b = On - drive LED[2] low, off - drive LED[2]
high.
01b = On - drive LED[2] high, off - drive LED[2]
low.
10b = On - drive LED[2] low, off - tristate LED[2].
11b = On - drive LED[2] high, off - tristate LED[2].
LED[2]
Polarity
00b = On - drive LED[1] low, off - drive LED[1]
high.
01b = On - drive LED[1] high, off - drive LED[1]
low.
10b = On - drive LED[1] low, off - tristate LED[1].
11b = On - drive LED[1] high, off - tristate LED[1].
LED[1]
Polarity
00b = On - drive LED[0] low, off - drive LED[0]
high.
01b = On - drive LED[0] high, off - drive LED[0]
low.
LED[0]
Polarity
10b = On - drive LED[0] low, off - tristate LED[0].
11b = On - drive LED[0] high, off - tristate LED[0].
292
Driver Programing Interface—82583V GbE Controller
9.2.8.30
LED Timer Control Register (Page 3), PHY Address 01; Register 18
Bits
15
Field
Mode
HW Rst
0x0
SW Rst
Description
1b = Interrupt pin asserted is forced.
0b = Normal operation.
Force INT
R/W
Retain
000b = No pulse stretching.
001b = 21 ms to 42 ms.
010b = 42 ms to 84 ms.
011b = 84 ms to 170 ms.
100b = 170 ms to 340 ms.
101b = 340 ms to 670 ms.
110b = 670 ms to 1.3 s.
111b = 1.3 s to 2.7 s
Pulse
14:12 Stretch
Duration
R/W
R/W
0x4
Retain
Retain
After a hardware reset, this bit takes on the value of
pd_config_intpol_a.
0b = jt_int_s active high.
1b = jt_int_a active low
Interrupt
Polarity
See
Description
11
000b = 42 ms.
001b = 84 ms.
010b = 170 ms.
011b = 340 ms.
100b = 670 ms.
101b to 111b = Reserved.
After a hardware reset, this bit is a function of
pd_config_led_a[1:0].
00b = 001b.
01b = 000b.
10b = 001b.
11b = 001b.
See
10:8
Blink Rate
Reserved
R/W
Retain
Description
7:4
3:2
R/W
R/W
0x0
0x1
Retain
Retain
Reserved, set to 0x0.
00b = 84 ms.
01b = 170 ms.
10b = 340 ms.
11b = 670 ms.
Speed Off
Pulse Period
00b = 84 ms.
01b = 170 ms.
10b = 340 ms.
11b = 670 ms.
Speed On
Pulse Period
1:0
R/W
0x1
Retain
293
82583V GbE Controller—Driver Programing Interface
9.2.8.31
LED[5:4] Function Control and Polarity Register (Page 3), PHY
Address 01; Register 19
Bits
Field
Mode
HW Rst
0x0
SW Rst
Description
15:12 Reserved R/W
Retain
Reserved, set to 0x0.
00b = On - drive LED[5] low, off - drive LED[5] high.
01b = On - drive LED[5] high, off - drive LED[5] low.
10b = On - drive LED[5] low, off - tristate LED[5].
11b = On - drive LED[5] high, off - tristate LED[5].
LED[5]
Polarity
11:10
9:8
R/W
R/W
0x0
0x0
Retain
Retain
00b = On - drive LED[4] low, off - drive LED[4] high.
01b = On - drive LED[4] high, off - drive LED[4] low.
10b = On - drive LED[4] low, off - tristate LED[4].
11b = On - drive LED[4] high, off - tristate LED[4].
LED[4]
Polarity
If 19_3.3:2 is set to 11b, then 19_3.7:4 has no effect.
0000b = On - receive, off - no receive.
0001b = On - link, blink - activity, off - no link.
0010b = On - link, blink - receive, off - no link.
0011b = On - activity, off - no activity.
0100b = Blink - activity, off - no activity.
0101b = On - transmit, off - no transmit.
0110b = On - full-duplex, off - half-duplex.
0111b = On - full-duplex, blink - collision off - half
duplex.
LED[5]
Control
See
Description
7:4
R/W
Retain
1000b = Force off.
1001b = Force on.
1010b = Force hi-Z.
1011b = Force blink.
11xxb = Reserved.
After a hardware reset, this bit is a function of
pd_config_led_a[1:0].
00b = 0111b.
01b = 0100b.
10b = 0111b.
11b = 0111b.
0000b = On - receive, off - no receive.
0001b = On - link, blink - activity, off - no link.
0010b = On - link, blink - receive, off - no link.
0011b = On - activity, off - no activity.
0100b = Blink - activity, off - no activity.
0101b = On - transmit, off - no transmit.
0110b = On - full-duplex, off - half-duplex.
0111b = On - full-duplex, blink - collision off - half
duplex.
1000b = Force off.
1001b = Force on.
1010b = Force hi-Z.
1011b = Force blink.
LED[4]
Control
See
Description
3:0
R/W
Retain
1100b = Mode 1 (dual LED mode).
1101b = Mode 2 (dual LED mode).
1110b = Mode 3 (dual LED mode).
1111b = Mode 4 (dual LED mode).
After a hardware reset, this bit is a function of
pd_config_led_a[1:0].
00b = 0011b.
01b = 0110b.
10b = 0011b.
11b = 0011b.
294
Driver Programing Interface—82583V GbE Controller
9.2.8.32
1000 BASE-T Pair Skew Register (Page 5), PHY Address 01; Register
20
Bits
Field
Mode HW Rst SW Rst
Description
Skew = bit value times 8 ns. The value is correct to
within ± 8 ns. The contents of 20_5.15:0 are valid only if
register 21_5.6 = 1b.
Pair 7,8
(MDI[3]±)
15:12
RO
0x0
0x0
Pair 4,5
Skew = bit value times 8 ns. The value is correct to
within ± 8 ns.
11:8
7:4
RO
RO
RO
0x0
0x0
0x0
0x0
0x0
0x0
(MDI[2]±)
Pair 3,6
(MDI[1]±)
Skew = bit value times 8 ns. The value is correct to
within ± 8 ns.
Pair 1,2
(MDI[0]±)
Skew = bit value times 8 ns. The value is correct to
within ± 8 ns.
3:0
9.2.8.33
1000 BASE-T Pair Swap and Polarity (Page 5), PHY Address 01;
Register 21
Bits
Field
Mode HW Rst SW Rst
Description
15:7 Reserved
RO
0x000
0x000
The contents of 21_5.5:0 and 20_5.15:0 are valid
only if register 21_5.6 = 1b.
1b = Valid.
Register 20_5 And
21_5 Valid
6
5
4
RO
0x0
0x0
0b = Invalid.
1b = Channel C received on MDI[2]± Channel D
received on MDI[3]±.
0b = Channel D received on MDI[2]± Channel C
received on MDI[3]±.
C, D Crossover
RO
0x0
0x0
0x0
0x0
1b = Channel A received on MDI[0]± Channel B
received on MDI[1]±.
0b = Channel B received on MDI[0]± Channel A
received on MDI[1]±.
A, B Crossover
RO
RO
1b = Negative.
0b = Positive.
Pair 7,8 (MDI[3]±)
Polarity
3
2
1
0
0x0
0x0
0x0
0x0
0x0
0x0
0x0
0x0
1b = Negative.
0b = Positive.
Pair 4,5 (MDI[2]±)
Polarity
1b = Negative.
0b = Positive.
Pair 3,6 (MDI[1]±)
Polarity
RO
RO
1b = Negative.
0b = Positive.
Pair 1,2 (MDI[0]±)
Polarity
9.2.8.34
CRC Counters (Page 6), PHY Address 01; Register 17
Bits
Field
Mode HW Rst SW Rst
Description
0x00 = No packets received.
CRC Packet
Count
0xFF = 256 packets received (maximum count). Bit
16_6.4 must be set to 1b in order for the register to be
valid.
15:8
RO
RO
0x00
0x00
Retain
Retain
0x00 = no CRC errors detected in the packets received.
CRC Error
Count
0xFF = 256 CRC errors detected in the packets received
(maximum count). Bit 16_6.4 must be set to 1b in order
for the register to be valid.
7:0
295
82583V GbE Controller—Driver Programing Interface
9.2.9
Diagnostic Register Descriptions
The 82583V contains several diagnostic registers. These registers enable software to
directly access the contents of the 82583V’s internal Packet Buffer Memory (PBM), also
referred to as FIFO space. These registers also give software visibility into what
locations in the PBM the hardware currently considers to be the head and tail for both
transmit and receive operations.
9.2.9.1
PHY OEM Bits Register - POEMB (0x00F10; RW)
The bits in this register are connected to the PHY interface. They affect the auto-
negotiation speed resolution and enable GbE mode. Additionally, PHY class A or B
drivers are also controlled.
Initial
Value
Field
Bit(s)
Description
1
Reserved
d0lplu
0
1
1b
Reserved
PHY auto negotiation for slowest possible link (reverse auto-
negotiation) in all power states. This bit overrides the LPLU bit.
1
1
0b
1b
Enables PHY auto-negotiation for slowest possible link (reverse auto-
negotiation) in all power states except D0a (DR, D0u and D3).
lplu
2
an1000_dis_n
d0a
Prevents PHY from auto negotiating 1000 Mb/s link in all power states
except D0a (DR, D0u and D3).
1
1
1
1
3
4
5
6
1b
0b
0b
0b
class_ab
Class AB driver.
reautoneg_
now
This bit can be written by software to force link auto re-negotiation.
1000_dis
Prevents PHY auto-negotiating 1000 Mb/s link in all power states.
Auto-update CB
Disable auto update of the Flash from the shadow RAM when the
ER_RD register is written.
1
Auto_update
7
8
9
0b
1b
1b
Controls the pause advertisements by the PHY.
1b = MAC pause implemented.
0b = MAC pause not implemented.
Pause
Controls the metric pause advertisement by the PHY.
1b = Asymmetric pause supported.
0b = Semantics pause not supported.
Asymmetric
Pause
Reserved
31:10
0x0
Reserved
1. Bits 7:0 of this register are loaded from NVM word 0x1C[15:8].
Note:
When software changes LPLU, D0LPLU or an1000_dis_nd0a it must wait at least 80 ns
and then force the link to auto-negotiate in order to commit the changes to the PHY.
9.2.9.2
Receive Data FIFO Head Register - RDFH (0x02410; RW)
Initial
Value
Field
Bit(s)
Description
FIFO Head
Reserved
12:0
31:13
0x0
0x0
Receive FIFO Head Pointer
Reads as 0x0. Should be written to 0x0 for future compatibility.
296
Driver Programing Interface—82583V GbE Controller
This register stores the head pointer of the on–chip receive data FIFO. Since the
internal FIFO is organized in units of 64-bit words, this field contains the 64-bit offset of
the current receive FIFO head. So a value of 0x8 in this register corresponds to an
offset of eight Qwords or 64 bytes into the receive FIFO space. This register is available
for diagnostic purposes only, and should not be written during normal operation.
Note:
This register’s address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x08000. In addition, with the 82583V, the value in this register contains the
offset of the receive FIFO head relative to the beginning of the entire PBM space.
Alternatively, with previous devices, the value in this register contains the relative
offset to the beginning of the receive FIFO space (within the PBM space).
9.2.9.3
Receive Data FIFO Tail Register - RDFT (0x02418; RW)
Initial
Value
Field
Bit(s)
Description
FIFO Tail
Reserved
12:0
31:13
0x0
0x0
Receive FIFO Tail pointer.
Reads as 0x0. Should be written to 0x0 for future compatibility.
This register stores the tail pointer of the on–chip receive data FIFO. Since the internal
FIFO is organized in units of 64 bit words, this field contains the 64 bit offset of the
current Receive FIFO Tail. So a value of “0x8” in this register corresponds to an offset of
8 QWORDS or 64 bytes into the Receive FIFO space. This register is available for
diagnostic purposes only, and should not be written during normal operation.
Note:
This register’s address has been moved from where it was located in previous devices.
However, for backwards compatibility, this register can also be accessed at its alias
offset of 0x08008. In addition, with the 82583V, the value in this register contains the
offset of the receive FIFO tail relative to the beginning of the entire PBM space.
Alternatively, with previous devices, the value in this register contains the relative
offset to the beginning of the Receive FIFO space (within the PBM space).
9.2.9.4
Receive Data FIFO Head Saved Register - RDFHS (0x02420; RW)
Initial
Value
Field
Bit(s)
Description
FIFO Head
Reserved
12:0
31:13
0x0
0x0
A saved value of the receive FIFO head pointer.
Reads as 0x0. Should be written to 0x0 for future compatibility.
This register stores a copy of the Receive Data FIFO Head register if the internal
register needs to be restored. This register is available for diagnostic purposes only,
and should not be written during normal operation.
9.2.9.5
Receive Data FIFO Tail Saved Register - RDFTS (0x02428; RW)
Initial
Value
Field
Bit(s)
Description
FIFO Tail
Reserved
12:0
31:13
0x0
0x0
A saved value of the receive FIFO tail pointer.
Reads as 0x0. Should be written to 0x0 for future compatibility.
297
82583V GbE Controller—Driver Programing Interface
This register stores a copy of the Receive Data FIFO Tail register if the internal register
needs to be restored. This register is available for diagnostic purposes only, and should
not be written during normal operation.
9.2.9.6
Receive Data FIFO Packet Count - RDFPC (0x02430; RW)
Initial
Field
RX FIFO
Bit(s)
Description
Value
12:0
31:13
0x0
0x0
The number of received packets currently in the RX FIFO.
Packet Count
Reserved
Reads as 0x0. Should be written to 0x0 for future compatibility.
This register reflects the number of receive packets that are currently in the receive
FIFO. This register is available for diagnostic purposes only, and should not be written
during normal operation.
9.2.9.7
Transmit Data FIFO Head Register - TDFH (0x03410; RW)
Initial
Value
Field
Bit(s)
Description
1
FIFO Tail
Reserved
12:0
31:13
0x600
0x0
Transmit FIFO Head Pointer
Reads as 0x0. Should be written to 0x0 for future compatibility.
1. The initial value equals PBA.RXA times 128.
This register stores the head pointer of the on–chip transmit data FIFO. Since the
internal FIFO is organized in units of 64-bit words, this field contains the 64-bit offset of
the current Transmit FIFO Head. So a value of 0x8 in this register corresponds to an
offset of eight Qwords or 64 bytes into the transmit FIFO space. This register is
available for diagnostic purposes only, and should not be written during normal
operation.
Note:
This register’s address has been moved from where it was located in the previous
devices. However, for backwards compatibility, this register can also be accessed at its
alias offset of 0x08010. In addition, with the 82583V, the value in this register contains
the offset of the transmit FIFO head relative to the beginning of the entire PBM space.
Alternatively, with the previous devices, the value in this register contains the relative
offset to the beginning of the transmit FIFO space (within the PBM space).
9.2.9.8
Transmit Data FIFO Tail Register - TDFT (0x03418; RW)
Initial
Value
Field
Bit(s)
Description
1
FIFO Tail
Reserved
12:0
31:13
0x600
0x0
Transmit FIFO Tail Pointer
Reads as 0x0. Should be written to 0x0 for future compatibility.
1. The initial value equals PBA.RXA times 128.
This register stores the head pointer of the on–chip transmit data FIFO. Since the
internal FIFO is organized in units of 64 bit words, this field contains the 64 bit offset of
the current Transmit FIFO Tail. So a value of “0x8” in this register corresponds to an
offset of 8 QWORDS or 64 bytes into the Transmit FIFO space. This register is available
for diagnostic purposes only, and should not be written during normal operation.
298
Driver Programing Interface—82583V GbE Controller
This register’s address has been moved from where it was located in the previous
devices. However, for backwards compatibility, this register can also be accessed at its
alias offset of 0x08018. In addition, with the 82583V, the value in this register contains
the offset of the transmit FIFO head relative to the beginning of the entire PBM space.
Alternatively, with the previous devices, the value in this register contains the relative
offset to the beginning of the transmit FIFO space (within the PBM space).
9.2.9.9
Transmit Data FIFO Head Saved Register - TDFHS (0x03420; RW)
Initial
Value
Field
Bit(s)
Description
1
FIFO Head
Reserved
12:0
31:13
0x600
0x0
A saved value of the Transmit FIFO Head Pointer.
Reads as 0x0. Should be written to 0x0 for future compatibility.
1. The initial value equals PBA.RXA times 128.
This register stores a copy of the Transmit Data FIFO Head register if the internal
register needs to be restored. This register is available for diagnostic purposes only,
and should not be written during normal operation.
9.2.9.10
Transmit Data FIFO Tail Saved Register - TDFTS (0x03428; RW)
Initial
Value
Field
Bit(s)
Description
1
FIFO Tail
Reserved
12:0
31:13
0x600
0x0
A saved value of the Transmit FIFO Tail Pointer.
Reads as 0x0. Should be written to 0x0 for future compatibility.
1. The initial value equals PBA.RXA times 128.
This register stores a copy of the Receive Data FIFO Tail register if the internal register
needs to be restored. This register is available for diagnostic purposes only, and should
not be written during normal operation.
9.2.9.11
Transmit Data FIFO Packet Count - TDFPC (0x03430; RW)
Initial
Field
TX FIFO
Bit(s)
Description
Value
0x0
0x0
The number of packets to be transmitted that are currently in the TX
FIFO.
12:0
31:13
Packet Count
Reserved
Reads as 0x0. Should be written to 0x0 for future compatibility.
This register reflects the number of packets to be transmitted that are currently in the
transmit FIFO. This register is available for diagnostic purposes only, and should not be
written during normal operation.
9.2.9.12
Packet Buffer Memory - PBM (0x10000 - 0x17FFF; RW)
Initial
Value
Field
Bit(s)
31:0
Description
FIFO Data
X
Packet Buffer Data
299
82583V GbE Controller—Driver Programing Interface
All PBM (FIFO) data is available to diagnostics. Locations can be accessed as 32-bit or
64-bit words. The internal PBM is 40 KB in size. As mentioned in Section 9.2.7.36,
software can configure the amount of PBM space that is used as the transmit FIFO
versus the receive FIFO. The default is 16 KB of transmit FIFO space and 16 KB of
receive FIFO space. Regardless of the individual FIFO sizes that software configures,
the RX FIFO is located first in the memory mapped PBM space. So for the default FIFO
configuration, the RX FIFO occupies offsets 0x10000-0x13FFF of the memory mapped
space, while the TX FIFO occupies offsets 0x14000-0x17FFF of the memory mapped
space.
9.2.9.13
Packet Buffer Size -PBS (0x01008; RW)
Initial
Value
Field
Bit(s)
Description
Packet Buffer Size
Lower six bits declare the packet buffer size both for transmit and
receive in 1 KB granularity. The upper 10 bits are read as zero. The
default is 40 KB.
PBS
15:0
0x0028
0x0000
Rsvd
31:16
Reserved read as zero.
This register sets the on-chip receive and transmit storage allocation size, The
allocation value is read/write for the lower six bits. The division between transmit and
receive is done according to the PBA register.
Note:
Note:
Programming this register does not automatically re-load or initialize internal packet-
buffer RAM pointers. The software must reset both transmit and receive operation
(using the global device reset CTRL.RST bit) after changing this register in order for it
to take effect. The PBS register itself is not reset by asserting the global reset, but only
is reset at initial hardware power on.
Programming this register should be aligned with programming the PBA register. If PBA
and PBS are not coordinated, hardware operation is not determined.
300
Driver Programing Interface—82583V GbE Controller
Note:
This page intentionally left blank.
301
82583V GbE Controller—Programing Interface
10.0
Programing Interface
10.1
PCIe Configuration Space
PCIe Compatibility
10.1.1
PCIe is completely compatible with existing deployed PCI software. To achieve this,
PCIe hardware implementations conform to the following requirements:
• All devices required to be supported by the deployed PCI software must be
enumerable as part of a tree through PCI device enumeration mechanisms.
• Devices must not require any resources (such as address decode ranges and
interrupts) beyond those claimed by PCI resources for operation of software
compatible and software transparent features with respect to existing deployed PCI
software.
• Devices in their default operating state must conform to PCI ordering and cache
coherency rules from a software viewpoint.
• PCIe devices must conform to PCI power management specification. PCIe devices
must not require any register programming for PCI-compatible power
management, beyond those available through PCI power management capability
registers. Power management is expected to conform to standard PCI power
management using existing PCI bus drivers.
PCIe devices implement all registers required by the PCI specification as well as the
power management registers and capability pointers specified by the PCI power
management specification. In addition, PCIe defines a PCIe capability pointer to
indicate support for PCIe extensions and associated capabilities.
Note:
The 82583V is a single function device - the LAN function.
The 82583V contains the following regions of the PCI configuration space:
• Mandatory PCI configuration registers
• Power management capabilities
• MSI capabilities
• PCIe extended capabilities
302
Programing Interface—82583V GbE Controller
10.1.2
Mandatory PCI Configuration Registers
The PCI configuration registers map is depicted below. See a detailed description for
registers loaded from the NVM at initialization time. Initialization values of the
configuration registers are marked in parenthesis. Color Notation in Figure 44:
Light Blue
Dark Grey
Read-only fields
Not used. Hardwired to zero.
Configuration registers are assigned one of the attributes described in Table 52.
Table 52.
R/W Attribute Table
R/W
Description
Attribute
RO
Read-only register: Register bits are read-only and cannot be altered by software.
Read-write register: Register bits are read-write and can be either set or reset.
Read-only status, Write-1-to-clear status register, Writing a 0b to R/W1C bits has no effect.
RW
R/W1C
Read-only register with sticky bits: Register bits are read-only and cannot be altered by
software. Bits are not cleared by reset and can only be reset with the PWRGOOD signal.
Devices that consume AUX power are not allowed to reset sticky bits when AUX power
consumption (either via AUX power or PME Enable) is enabled.
ROS
RWS
Read-write register with sticky bits: Register bits are read-write and can be either set or reset
by software to the desired state. Bits are not cleared by reset and can only be reset with the
PWRGOOD signal. Devices that consume AUX power are not allowed to reset sticky bits when
AUX power consumption (either via AUX power or PME Enable) is enabled.
Read-only status, Write-1-to-clear status register with sticky bits: Register bits indicate status
when read, a set bit indicating a status event can be cleared by writing a 1b. Writing a 0b to R/
W1C bits has no effect. Bits are not cleared by reset and can only be reset with the PWRGOOD
signal. Devices that consume AUX power are not allowed to reset sticky bits when AUX power
consumption (either via AUX power or PME Enable) is enabled.
R/W1CS
HwInit
Hardware Initialized: Register bits are initialized by firmware or hardware mechanisms such as
pin strapping or serial NVM. Bits are read-only after initialization and can only be reset (for
write-once by firmware) with PWRGOOD signal.
Reserved and Preserved: Reserved for future R/W implementations; software must preserve
value read for writes to bits.
RsvdP
RsvdZ
Reserved and Zero: Reserved for future R/W1C implementations; software must use 0b for
writes to bits.
Byte Offset
Byte 3
Byte 2
Byte 1
Byte 0
0x0
0x4
0x8
Device ID
Status Register (0x0010)
Class Code (0x020000)
Vendor ID (0x8086)
Command Register (0x0000)
Revision ID (0x00)
Header Type (0x00 |
0x80)
Cache Line Size
(0x10)
0xC
BIST (0x00)
Latency Timer (0x00)
0x10
0x14
0x18
0x1C
0x20
0x24
Base Address 0
Base Address 1
Base Address 2
Base Address 3
Base Address 4
Base Address 5
303
82583V GbE Controller—Programing Interface
Byte Offset
Byte 3
Byte 2
Byte 1
Byte 0
0x28
0x2C
0x30
0x34
0x38
Cardbus CIS Pointer (0x00000000)
Subsystem ID (0x0000) Subsystem Vendor ID (0x8086)
Expansion ROM Base Address
Reserved (0x000000)
Reserved (0x00000000)
Cap_Ptr (0xC8)
Interrupt Line
(0x00)
Min_Grant
(0x00)
Interrupt Pin
(0x01)
0x3C
Max_Latency (0x00)
Figure 44.
10.1.2.1
PCI-Compatible Configuration Registers
Explanation of the various registers in the 82583V is as follows.
Vendor ID (Offset 0x0)
This is a read-only register that has the same value for all PCI functions. It uniquely
identifies Intel products. The field default value is 0x8086.
10.1.2.2
Device ID (Offset 0x2)
This is a read-only register. The value is loaded from NVM. Default value is 0x150C for
the 82583V.
PCI
Function
Default
Value
NVM Address
Meaning
10/100/1000 Mb/s Ethernet controller, x1
PCIe, copper
LAN
0x150C
0x0D
10.1.2.3
Command Reg (Offset 0x4)
Read-write register. Layout is as follows. Shaded bits are not used by this
implementation and are hardwired to 0b.
Bit(s)
Init Value
Description
0
1
2
3
4
5
6
7
8
9
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
I/O Access Enable.
Memory Access Enable.
Enable Mastering LAN R/W field.
Special Cycle Monitoring – Hardwired to 0b.
MWI Enable – Hardwired to 0b.
Palette Snoop Enable – Hardwired to 0b.
Parity Error Response.
Wait Cycle Enable – Hardwired to 0b.
SERR# Enable.
Fast Back-to-Back Enable – Hardwired to 0b.
Interrupt Disable
Controls the ability of a PCIe device to generate a legacy interrupt
message. When set, the device can’t generate legacy interrupt
messages.
10
0b
0b
15:11
Reserved
304
Programing Interface—82583V GbE Controller
10.1.2.4
Status Register (Offset 0x6)
Shaded fields are not used by this implementation and are hardwired to 0b.
Initial
Bits
2:0
R/W
Description
Value
000b
Reserved
1
3
0b
1b
RO
Interrupt Status
New Capabilities
Indicates that a device implements extended capabilities. The
82583V sets this bit, and implements a capabilities list, to
indicate that it supports PCI power management, message
signaled interrupts, and the PCIe extensions.
4
RO
5
0b
0b
0b
0b
00b
0
66MHz Capable – Hardwired to 0b.
Reserved.
6
7
Fast Back-to-Back Capable – Hardwired to 0b.
Data Parity Reported.
8
R/W1C
10:9
11
12
13
14
15
DEVSEL Timing – Hardwired to 0b.
Signaled Target Abort.
R/W1C
R/W1C
R/W1C
R/W1C
R/W1C
0bb
0b
0b
0b
Received Target Abort.
Received Master Abort.
Signaled System Error.
Detected Parity Error.
1. The Interrupt Status field is a read-only field that indicates that an interrupt message is
pending internally to the device.
10.1.2.5
10.1.2.6
Revision ID (Offset 0x8)
The default revision ID of this device is 0x0. The value of the rev ID is a logic XOR
between the default value and the value in the NVM word 0x1E.
Class Code (Offset 0x9)
The class code is a read-only, hard-coded value that identifies the device functionality.
LAN - 0x020000 - Ethernet Adapter
10.1.2.7
Cache Line Size (Offset 0xC)
This field is implemented by PCIe devices as a read-write field for legacy compatibility
purposes but has no impact on any PCIe device functionality. Loaded from NVM words
0x1A.
10.1.2.8
10.1.2.9
Latency Timer (Offset 0xD)
Not used. Hardwired to 0b.
Header Type (Offset 0xE)
This indicates if a device is single function or multifunction. For the 82583V this field
has a value of 0x00 to indicate a single function device.
305
82583V GbE Controller—Programing Interface
10.1.2.10 Base Address Registers (Offset 0x10 - 0x27)
The Base Address Registers (BARs) are used to map the 82583V register space. The
82583V BARs are defined as non-prefetchable, and therefore support 32-bit addressing
only.
BAR
Addr.
31
4
3
2
1
0
0
1
2
3
4
5
0x10
0x14
0x18
0x1C
0x20
0x24
Memory BAR (R/W - 31:17; 0b - 16:4)
Flash BAR (R/W - 31:23/16; 0b - 22/15:4)
IO BAR (R/W - 31:5; 0b - 4:1)
Reserved (read as all 0b’s)
0b
0b
00b
00b
0b
0b
0b
1b
Reserved (read as all 0b’s)
Reserved (read as all 0b’s)
Note:
Note:
Flash size is defined by the NVM.
The default setting of the Flash BAR enables software implement initial programming of
empty (non-valid) Flash via the (parallel) Flash BAR.
Note:
The 82583V requests I/O resources to support pre-boot operation (prior to allocating
physical memory base addresses).
All BARs have the following fields:
Initial
Value
Field
Bit(s)
R/W
Description
0b for
0b = Memory space
1b = I/O space.
memory
Mem
0
R
R
1b for I/O
Indicates the address space size.
00b = 32-bit
10b = 64-bit
00b (for
32-bit)
Mem Type
2:1
3
The 82583V BARs are 32-bit only.
0b = Non-prefetchable space.
1b = Prefetchable space.
The 82583V implements non-prefetchable space since it has
read side effects.
Prefetch
Mem
R
0b
Read/Write bits and hardwired to 0b depending on the
memory mapping window sizes:
LAN memory spaces are 128 KB.
LAN Flash spaces can be 64 KB and up to 4 MB in powers of
2.
Flash window size is set by the NVM. The Flash BAR can also
be disabled by the NVM.
Memory
Address
Space
31:4
31:2
R/W
R/W
0x0
0x0
Read/Write bits and hardwired to 0b depending on the I/O
mapping window sizes:
LAN I/O space is 32 bytes.
IO Address
Space
306
Programing Interface—82583V GbE Controller
Memory and I/O mapping:
Mapping
Window
Mapping Description
The internal registers and memories are accessed as direct memory
mapped offsets from the base address register. Software can access
byte, word or Dword.
Memory
BAR 0
The external Flash can be accessed using direct memory mapped
offsets from the Flash base address register. Software can access byte,
word or Dword.
Flash
BAR 1
The Flash BAR is enabled by the DISLFB field in NVM word 0x21.
All internal registers, memories, and Flash can be accessed using I/O
operations. There are two 4-byte registers in the I/O mapping window:
Addr Reg and Data Reg. Software can access byte, word or Dword.
I/O
BAR 2
Reserved
Reserved
10.1.2.11 CardBus CIS (Offset 0x28)
Not used. Hardwired to 0b.
10.1.2.12 Subsystem ID (Offset 0x2E)
This value can be loaded automatically from the NVM at power up with a default value
of 0x0000.
10.1.2.13 Subsystem Vendor ID (Offset 0x2C)
This value can be loaded automatically from the NVM address 0x0C at power up or
reset. The default value is 0x8086 at power up.
10.1.2.14 Expansion ROM Base Address (Offset 0x30)
This register is used to define the address and size information for boot-time access to
the optional Flash memory. The BAR size and enablement are set by the NVM.
Read/
Write
Initial
Value
Field
Bit(s)
Description
1b = Enables expansion ROM access.
0b = Disables expansion ROM access.
En
0
R/W
R
0b
Reserved
Address
10:1
0x0
0x0
Always read as 0b. Writes are ignored.
Read/Write bits and hardwired to 0b depending on the
memory mapping window size as defined in word 0x21 in
the NVM.
31:11
R/W
307
82583V GbE Controller—Programing Interface
10.1.2.15 Cap_Ptr (Offset 0x34)
The Capabilities Pointer field (Cap_Ptr) is an 8-bit field that provides an offset in the
device's PCI configuration space for the location of the first item in the capabilities
linked list. The 82583V sets this bit, and implements a capabilities list, to indicate that
it supports:
• PCI power management
• MSI
• PCIe extended capabilities
Its value, 0xC8, is the address of the first entry: PCI power management.
Address
0xC8-CF
Item
Next Pointer
PCI power management
MSI
0xD0
0xE0
0x00
0xA0
0xD0-DF
0xA0-AB
0xE0-F3
Reserved
PCIe Capabilities
10.1.2.16 Interrupt Line (Offset 0x3C)
Read/write register programmed by software to indicate which of the system interrupt
request lines this device's interrupt pin is bound to. See the PCI definition for more
details.
10.1.2.17 Interrupt Pin (Offset 0x3D)
Read-only register. The LAN implements legacy interrupt on INTA.
10.1.2.18 Max_Lat/Min_Gnt (Offset 0x3E)
Not used. Hardwired to 0b.
10.1.3
PCI Power Management Registers
All fields are reset on full power up. All of the fields except PME_En and PME_Status are
reset on exit from D3cold state.
See the detailed description for registers loaded from the NVM at initialization time.
Initialization values of the configuration registers are marked in parenthesis.
Some fields in this section depend on the Power Management Ena bits in the NVM word
0x0A.
Table 53 lists the organization of the PCI Power Management register block. Light-blue
fields are read only fields.
Table 53.
Power Management Register Block
Byte Offset
Byte 3
Byte 2
Byte 1
Byte 0
Next Pointer
(0xD0)
Capability ID
(0x01)
0xC8
Power Management Capabilities (PMC)
PMCSR_BSEBridge Power Management Control / Status
Support Extensions Register (PMCSR)
0xCC
Data
308
Programing Interface—82583V GbE Controller
The following section describes the register definitions, whether they are required or
optional for compliance, and how they are implemented in the 82583V.
10.1.3.1
10.1.3.2
10.1.3.3
Capability ID, Offset 0xC8, (RO)
This field equals 0x01 indicating the linked list item is the PCI Power Management
registers.
Next Pointer, Offset 0xC9, (RO)
This field provides an offset to the next capability item in the capability list. Its value of
0xD0 points to the MSI capability.
Power Management Capabilities (PMC), Offset 0xCA, (RO)
This field describes the device functionality at the power management states as
described in the following table.
Bits
Default
R/W
Description
PME_Support
This five-bit field indicates the power states in which the function might
assert PME# depending on NVM settings:
See value in
description
column
00000b = If PM is disabled in NVM (word 0x0A) than No PME support at
all states.
15:11
RO
01001b = If PM is enabled in NVM and no Aux_Pwr than PME is
supported at D0 and D3
hot.
11001b = If PM is Enabled in NVM and Aux_Pwr, then PME is supported
at D0, D3 and D3
hot
cold.
D2_Support
The 82583V does not support D2 state
10
9
0b
RO
RO
RO
D1_Support
The 82583V does not support D1 state
0b
AUX Current
Required current defined in the Data register
8:6
000b
DSI
5
1b
RO
The 82583V requires its software device driver to be executed following
transition to the D0 un-initialized state.
4
3
0b
0b
RO
RO
Reserved
PME_Clock
Disabled. Hardwired to 0b.
Version
2:0
010b
RO
The 82583V complies with PCI PM spec revision 1.1.
Figure 45.
Power Management Capabilities (PMC)
309
82583V GbE Controller—Programing Interface
10.1.3.4
Power Management Control/Status Register - (PMCSR), Offset 0xCC,
(RW)
This register is used to control and monitor power management events in the 82583V.
Bits
Default
Rd/Wr
Description
PME_Status
15
0b at power up R/W1C
This bit is set to 1b when the function detects a wake-up event
independent of the state of the PME_En bit. Writing a 1b clears this bit.
Data_Scale
This field indicates the scaling factor to be used when interpreting the
value of the Data register.
If the PM is enabled in the NVM, and the Data_Select field is set to 0, 3,
4 or7, than this field equals 01b (indicating 0.1 watt units). Else it
equals 00b.
see value in
RO
14:13
12:9
Data register
Data_Select
This four-bit field is used to select which data is to be reported through
the Data register and Data_Scale field. These bits are writeable only
when power management is enabled via the NVM.
0000b
R/W
PME_En
If power management is enabled in the NVM, writing a 1b to this
register enables wake up.
8
0b at power up R/W
If power management is disabled in the NVM, writing a 1b to this bit
has no affect, and does not set the bit to 1b.
Reserved
7:4
000000b
RO
The 82583V returns a value of 000000b for this field.
No_Soft_Reset
This bit is always set to 0b to indicate that the 82583V performs an
internal reset upon transitioning from D3hot to D0 via software control
of the PowerState bits. Configuration context is lost when performing
the soft reset. Upon transition from the D3hot to the D0 state, a full re-
initialization sequence is needed to return the 82583V to the D0
Initialized state.
3
2
0b
0b
RO
RO
Reserved
Power State
This field is used to set and report the power state of the 82583V as
follows:
00b = D0.
1:0
00b
R/W
01b = D1 (cycle ignored if written with this value).
10b = D2 (cycle ignored if written with this value).
11b = D3 (cycle ignored if PM is not enabled in the NVM).
Figure 46.
10.1.3.5
Power Management Control/Status - PMCSR
PMCSR_BSE Bridge Support Extensions, Offset 0xCE, (RO)
This register is not implemented in the 82583V, values set to 0x00.
310
Programing Interface—82583V GbE Controller
10.1.3.6
Data Register, Offset 0xCF, (RO)
This optional register is used to report power consumption and heat dissipation.
Reported register is controlled by the Data_Select field in the PMCSR and the power
scale is reported in the Data_Scale field in the PMCSR. The data of this field is loaded
from the NVM if power management is enabled in the NVM. Otherwise, it has a default
value of 0x00. The values for the 82583V are as follows:
Function
D0 (Consume/Dissipate)
D3 (Consume/Dissipate)
Data Select
Function 0
(0x0/0x4)
(0x3/0x7)
EEPROM address 0x22
EEPROM address 0x22
For other Data_Select values the Data register output is reserved (0b).
10.1.4
Message Signaled Interrupt (MSI) Configuration Registers
This structure is required for PCIe devices. Initialization values of the configuration
registers are marked in parenthesis. Light-blue fields represent read-only fields.
Note:
There are no changes to this structure from the PCI 2.2 specification.
Byte Offset
Byte 3
Byte 2
Byte 1
Byte 0
0xD0
0xD4
0xD8
0xDC
Message Control (0x0080)
Message Address
Next Pointer (0xE0)
Capability ID (0x05)
Message Upper Address
Reserved
Message Data
Figure 47.
10.1.4.1
MSI Configuration Registers
Capability ID, Offset 0xD0, (RO)
This field equals 0x05 indicating the linked list item as being the MS registers.
10.1.4.2
Next Pointer, Offset 0xD1, (RO)
This field provides an offset to the next capability item in the capability list. Its value of
0xE0 points to the PCIe capability.
311
82583V GbE Controller—Programing Interface
10.1.4.3
Message Control Offset 0xD2, (R/W)
The register fields are listed in the following table.
Bits
Default
R/W
Description
MSI Enable
0
0b
R/W
If set to 1b, MSI. In this case, the 82583V generates MSI for interrupt
assertion instead of INTx signaling.
Multiple Message Capable
The 82583V indicates a single requested message.
3:1
6:4
000b
000b
RO
RO
Multiple Message Enable
The 82583V returns 000b to indicate that it supports a single message.
64-bit capable. A value of 1b indicates that the 82583V is capable of
generating 64-bit message addresses.
7
1b
RO
RO
15:8
0x0
Reserved, reads as 0b.
10.1.4.4
Message Address Low Offset 0xD4, (R/W)
Written by the system to indicate the lower 32 bits of the address to use for the MSI
memory write transaction. The lower two bits always returns 0b regardless of the write
operation.
10.1.4.5
10.1.4.6
Message Address High, Offset 0xD8, (R/W)
Written by the system to indicate the upper 32 bits of the address to use for the MSI
memory write transaction.
Message Data, Offset 0xDC, (R/W)
Written by the system to indicate the lower 16 bits of the data written in the MSI
memory write Dword transaction. The upper 16 bits of the transaction are written as
0b.
10.1.5
PCIe Configuration Registers
PCIe provides two mechanisms to support native features:
• PCIe defines a PCIe capability pointer indicating support for PCIe.
• PCIe extends the configuration space beyond the 256 bytes available for PCI to
4096 bytes.
Initialization values of the configuration registers are marked in parenthesis.
312
Programing Interface—82583V GbE Controller
10.1.5.1
Table 54.
PCIe Capability Structure
The 82583V implements the PCIe capability structure for end-point devices as listed in
Table 54.
PCIe Configuration Registers
Byte Offset
Byte 3
Byte 2
Byte 1
Byte 0
0xE0
0xE4
0xE8
0xEC
0xF0
PCIe Capability Register
Next Pointer
Capability ID
Device Capability
Device Status
Link Status
Device Control
Link Capability
Link Control
10.1.5.1.1
Capability ID, Offset 0xE0, (RO)
This field equals 0x10 indicating the linked list item as being the PCIe Capabilities
registers.
10.1.5.1.2
10.1.5.1.3
Next Pointer, Offset 0xE1, (RO)
Offset to the next capability item in the capability list.
PCI Express CAP, Offset 0xE2, (RO)
The PCIe capabilities register identifies PCIe device type and associated capabilities.
This is a read-only register.
Bits
3:0
Default
R/W
RO
Description
Capability Version
0001b
Indicates the PCIe capability structure version number 1.
Device/Port Type
7:4
8
0000b
0b
RO
RO
Indicates the type of PCIe functions. LAN function in the 82583V is a native
PCIe functions with a value of 0000b.
Slot Implemented
The 82583V does not implement slot options therefore this field is hardwired
to 0b.
Interrupt Message Number
13:9
00000b
00b
RO
RO
The 82583V does not implement multiple MSI per function, therefore this field
is hardwired to 0x0.
15:14
Reserved
10.1.5.1.4
Device CAP, Offset 0xE4, (RO)
This register identifies the PCIe device specific capabilities. It is a read-only register.
Bits
R/W
Default
Description
Max Payload Size Supported
This field indicates the maximum payload that the device can support for
TLPs. It is loaded from the NVM PCIe Init Configuration 3 word 0x1A (bit 8)
with a default value of 256 bytes.
2:0
RO
001b
313
82583V GbE Controller—Programing Interface
Bits
4:3
R/W
RO
Default
00b
Description
Phantom Function Supported
Not supported by the 82583V.
Extended Tag Field Supported
Max supported size of the Tag field. The 82583V supports a 5-bit Tag field.
5
RO
RO
0b
End-Point L0s Acceptable Latency
This field indicates the acceptable latency that the 82583V can withstand due
to the transition from L0s state to the L0 state. The value is loaded from the
NVM PCIe Init Configuration 1 word 0x18.
8:6
011b
End-Point L1 Acceptable Latency
This field indicates the acceptable latency that the 82583V can withstand due
to the transition from L1 state to the L0 state. The value is loaded from the
NVM PCIe Init Configuration 1 word 0x18.
11:9
RO
110b
Attention Button Present
Hardwired in the 82583V to 0b.
12
13
14
RO
RO
RO
0b
0b
0b
Attention Indicator Present
Hardwired in the 82583V to 0b.
Power Indicator Present
Hardwired in the 82583V to 0b.
Role Based Error Reporting
Hardwired in the 82583V to 1b.
15
RO
RO
RO
1b
17:16
25:18
00b
0x0
Reserved, set to 00b
Slot Power Limit Value
Used in upstream ports only. Hardwired in the 82583V to 0x00.
Slot Power Limit Scale
Used in upstream ports only. Hardwired in the 82583V to 0b.
27:26
31:28
RO
RO
00b
0000b
Reserved
10.1.5.1.5
Device Control, Offset 0xE8, (RW)
This register controls PCIe specific parameters.
Bits
R/W
RW
Default
0b
Description
Correctable Error Reporting Enable
0
1
2
3
Enable error report.
Non-Fatal Error Reporting Enable
Enable error report.
RW
RW
RW
0b
0b
0b
Fatal Error Reporting Enable
Enable error report.
Unsupported Request Reporting Enable
Enable error report.
Enable Relaxed Ordering
If this bit is set, the device is permitted to set the Relaxed Ordering bit in the
attribute field of write transactions that do not need strong ordering. For
more details, also see register CTRL_EXT bit RO_DIS.
4
RW
RW
1b
Max Payload Size
This field sets maximum TLP payload size for the device functions. As a
000b (128 receiver, the device must handle TLPs as large as the set value. As a
7:5
Bytes)
transmitter, the device must not generate TLPs exceeding the set value.
The Maximum Payload Size supported in the Device Capabilities register
indicates permissible values that can be programmed.
314
Programing Interface—82583V GbE Controller
Bits
R/W
RW
Default
0b
Description
Extended Tag field Enable
8
9
Not implemented in the 82583V.
Phantom Functions Enable
Not implemented in the 82583V.
RW
RO
RW
0b
0b
1b
Auxiliary Power PM Enable
When set, enables the device to draw AUX power independent of PME AUX
power. In the 82583V, this bit is hardwired to 0b.
10
11
Enable No Snoop
Snoop is gated by NONSNOOP bits in the GCR register in the CSR space.
Max Read Request Size
This field sets maximum read request size for the device as a requester. The
default value is 010b (512 bytes).
This maximum read request configuration value should not be altered on the
fly.
14:12
15
RW
RO
010b
0b
Reserved.
10.1.5.1.6
PCIe Device Status, Offset 0xEA, (RO)
This register provides information about PCIe device specific parameters.
.
Bits
R/W
Default
0b
Description
Correctable Detected
0
1
2
3
RW1C
Indicates status of correctable error detection.
Non-Fatal Error Detected
Indicates status of non-fatal error detection.
RW1C
RW1C
RW1C
0b
0b
0b
Fatal Error Detected
Indicates status of fatal error detection.
Unsupported Request Detected
Indicates that the 82583V received an unsupported request.
Aux Power Detected
4
5
RO
0b
If Aux power is detected, this field is set to 1b. It is a strapping signal from
the periphery. Reset on Internal Power On Reset and PCIe Power Good only.
Transaction Pending
Indicates whether the 82583V has any transactions pending. (Transactions
include completions for any outstanding non-posted request for all used traffic
classes.).
RO
RO
0b
15:6
0x00
Reserved
315
82583V GbE Controller—Programing Interface
10.1.5.1.7
Link CAP, Offset 0xEC, (RO)
This register identifies PCIe link-specific capabilities. This is a read-only register.
Bits
3:0
R/W
RO
Default
Description
Max Link Speed
0001b
The 82583V indicates a maximum link speed of 2.5 Gb/s.
Max Link Width
Indicates the maximum link width. The 82583V supports x1 lane link.
9:4
RO
RO
0x01
Defined encoding:
000001b x1.
All other values - Reserved.
Active State Link PM Support
Indicates the level of active state power management supported in the
82583V. Defined encodings are:
00b = No ASPM support.
01b = L0s supported.
11:10
11b
10b = L1 supported.
11b = L0s and L1 supported.
This field is loaded from the NVM PCIe Init Configuration 3 word 0x1A.
L0s Exit Latency
Indicates the exit latency from L0s to L0 state. This field is loaded from the
NVM PCIe Init Configuration 1 word 0x18 (two values for common PCIe clock
or separate PCIe clock.
000b = Less than 64 ns.
001b = 64 ns – 128 ns.
010b = 128 ns – 256 ns.
011b = 256 ns - 512 ns.
100b = 512 ns - 1 μs.
101b = 1 μs – 2 μs.
001b
(64-
128 ns)
14:12
RO
110b = 2 μs – 4 μs.
111b = Reserved.
If the 82583V uses a common clock - PCIe Init Config 1 bits [2:0], if the
82583V uses a separate clock - PCIe Init Config 1 bits [5:3].
L1 Exit Latency
Indicates the exit latency from L1 to L0 state. This field is loaded from the
NVM PCIe Init Configuration 1 word 0x18.
000b = Less than 1 μs.
001b = 1 μs - 2 μs.
010b = 2 μs - 4 μs.
110b
(32-64 μs)
17:15
RO
011b = 4 μs - 8 μs.
100b = 8 μs - 16 μs.
101b = 16 μs - 32 μs.
110b = 32 μs - 64 μs.
111b = L1 transition not supported.
18
RO
RO
RO
RO
0b
Reserved.
19
0b
Surprise Down Error Reporting Capable.
Data Link Layer Link Active Reporting Capable.
Reserved.
20
0b
23:21
000b
Port Number
31:24
HwInit
0x0
The PCIe port number for the given PCIe link. Field is set in the link training
phase.
316
Programing Interface—82583V GbE Controller
10.1.5.1.8
Link Control, Offset 0xF0, (RO)
This register controls PCIe link specific parameters.
Bits
R/R
Default
Description
Active State Link PM Control
This field controls the active state PM supported on the link. Defined
encodings are:
00b = PM disabled.
1:0
RW
00b
01b = L0s entry supported.
10b = Reserved.
11b = L0s and L1 supported.
2
3
RO
0b
0b
Reserved.
RW
Read Completion Boundary.
Link Disable
4
5
RO
RO
0b
0b
Not applicable for end-point devices, hardwired to 0b.
Retrain Clock
Not applicable for end-point devices, hardwired to 0b.
Common Clock Configuration
When set, indicates that the 82583V and the component at the other end of
the link are operating with a common reference clock. A value of 0b indicates
that they operate with an asynchronous clock. This parameter affects the L0s
exit latencies.
6
RW
0b
Extended Sync
7
RW
RO
0b
This bit, when set, forces extended Tx of FTS ordered set in FTS and extra
TS1 at exit from L0s prior to enter L0.
15:8
0x0
Reserved.
10.1.5.1.9
Link Status, Offset 0xF2, (RO)
This register provides information about PCIe link-specific parameters. This is a read-
only register.
Bits
R/W
Default
Description
Link Speed
3:0
RO
0001b
Indicates the negotiated link speed. 0001b is the only defined speed, which is
2.5 Gb/s.
Negotiated Link Width
Indicates the negotiated width of the link.
Relevant encoding for the 82583V is:
000001b x1
9:4
RO
000001b
Link Training Error
Indicates that a link training error has occurred.
10
11
RO
RO
0b
0b
Link Training
Indicates that link training is in progress.
Slot Clock Configuration
When set, indicates that the 82583V uses the physical reference clock that
the platform provides on the connector. This bit must be cleared if the 82583V
uses an independent clock. Slot Clock Configuration bit is loaded from the
Slot_Clock_Cfg NVM bit.
12
HwInit
RO
1b
15:13
0000b
Reserved
317
82583V GbE Controller—Programing Interface
10.1.5.2
PCIe Extended Configuration Space
PCIe configuration space is located in a flat memory-mapped address space. PCIe
extends the configuration space beyond the 256 bytes available for PCI to 4096 bytes.
The 82583V decodes additional 4-bits (bits 27:24) to provide the additional
configuration space as shown. PCIe reserves the remaining 4 bits (bits 31:28) for
future expansion of the configuration space beyond 4096 bytes.
The configuration address for a PCIe device is computed using PCI-compatible bus,
device and function numbers as follows:
31
28
27
20
19
15
14 12
11
2
1
0
0000b
Bus #
Device #
Fun #
Register Address (offset) 00b
PCIe extended configuration space is allocated using a linked list of optional or required
PCIe extended capabilities following a format resembling PCI capability structures. The
first PCIe extended capability is located at offset 0x100 in the device configuration
space. The first Dword of the capability structure identifies the capability/version and
points to the next capability.
The 82583V supports the following PCIe extended capabilities:
• Advanced error reporting capability - offset 0x100
• Device serial number capability - offset 0x140
10.1.5.2.1
Advanced Error Reporting Capability
The PCIe advanced error reporting capability is an optional extended capability to
support advanced error reporting. The following table lists the PCIe advanced error
reporting extended capability structure for PCIe devices.
Register
Offset
Field
Description
0x00
PCIe CAP ID
PCIe Extended Capability ID.
UncorrectableError Reports error status of individual uncorrectable error sources on a PCIe
Status device.
0x04
0x08
0x0C
0x10
0x14
UncorrectableError Controls reporting of individual uncorrectable errors by device to the
Mask host bridge via a PCIe error message.
UncorrectableError Controls whether an individual uncorrectable error is reported as a fatal
Severity
error.
Correctable Error
Status
Reports error status of individual correctable error sources on a PCIe
device.
Correctable Error
Mask
Controls reporting of individual correctable errors by device to the host
bridge via a PCIe error message.
Identifies the bit position of the first uncorrectable error reported in the
Uncorrectable Error Status register.
0x18
First Error Pointer
Header Log
0x1C:0x28
Captures the header for the transaction that generated an error.
318
Programing Interface—82583V GbE Controller
10.1.5.2.1.1 PCI Express CAP ID, Offset 0x00
Bit Location
Attribute
Default Value
Description
Extended Capability ID
15;0
RO
0x0001
PCIe extended capability ID indicating advanced error
reporting capability.
Version Number
19:16
31:20
RO
RO
0x1
PCIe advanced error reporting extended capability version
number.
Next Capability Pointer - Next PCIe extended capability
pointer.
If serial number capability is enabled in NVM (PCIe init
configuration 2 word), the default value is 0x140.
Otherwise, it’s 0x000 indicating the end of capabilities list.
0x000/0x140
10.1.5.2.1.2 Uncorrectable Error Status, Offset 0x04
The Uncorrectable Error Status register reports error status of individual uncorrectable
error sources on a PCIe device. A value of 1b at a specific bit location indicates the
source of the error according to the following table. Software might clear an error
status by writing a 1b to the respective bit.
.
Bit Location
Attribute
RO
Default Value
0b
Description
3:0
4
Reserved.
R/W1CS
RO
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
Data Link Protocol Error Status.
Reserved.
11:5
12
R/W1CS
R/W1CS
R/W1CS
R/W1CS
R/W1CS
R/W1CS
R/W1CS
RO
Poisoned TLP Status.
13
Flow Control Protocol Error Status.
Completion Timeout Status.
Completion Abort Status.
Unexpected Completion Status.
Receiver Overflow Status.
Malformed TLP Status.
Reserved.
14
15
16
17
18
19
20
R/W1CS
RO
Unsupported Request Error Status.
Reserved.
31:21
319
82583V GbE Controller—Programing Interface
10.1.5.2.1.3 Uncorrectable Error Mask, Offset 0x08
The Uncorrectable Error Mask register controls reporting of individual uncorrectable
errors by device to the host bridge via a PCIe error message. A masked error
(respective bit set in mask register) is not reported to the host bridge by an individual
device. There is a mask bit per bit of the Uncorrectable Error Status register.
Bit Location
Attribute
RO
Default Value
0b
Description
3:0
4
Reserved.
RWS
RO
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
0b
Data Link Protocol Error Mask.
Reserved.
11:5
12
RWS
RWS
RWS
RWS
RWS
RWS
RWS
RO
Poisoned TLP Mask.
13
Flow Control Protocol Error Mask.
Completion Timeout Mask.
Completion Abort Mask.
Unexpected Completion Mask.
Receiver Overflow Mask.
Malformed TLP Mask.
Reserved.
14
15
16
17
18
19
20
RWS
RO
Unsupported Request Error Mask.
Reserved.
31:21
10.1.5.2.1.4 Uncorrectable Error Severity, Offset 0x0C
The Uncorrectable Error Severity register controls whether an individual uncorrectable
error is reported as a fatal error. An uncorrectable error is reported as fatal when the
corresponding error bit in the severity register is set. If the bit is cleared, the
corresponding error is considered non-fatal.
Bit Location
Attribute
RO
Default Value
0b
Description
3:0
4
Reserved.
RWS
RO
1b
0b
0b
1b
0b
0b
0b
1b
1b
0b
0b
0b
Data Link Protocol Error Severity.
Reserved.
11:5
12
RWS
RWS
RWS
RWS
RWS
RWS
RWS
RO
Poisoned TLP Severity.
13
Flow Control Protocol Error Severity.
Completion Timeout Severity.
Completion Abort Severity.
Unexpected Completion Severity.
Receiver Overflow Severity.
Malformed TLP Severity.
Reserved.
14
15
16
17
18
19
20
RWS
RO
Unsupported Request Error Severity.
Reserved.
31:21
320
Programing Interface—82583V GbE Controller
10.1.5.2.1.5 Correctable Error Status, Offset 0x10
The Correctable Error Status register reports error status of individual correctable error
sources on a PCIe device. When an individual error status bit is set to 1b it indicates
that a particular error occurred. Software might clear an error status by writing a 1b to
the respective bit.
Bit Location
Attribute
R/W1CS
Default Value
0b
Description
0
Receiver Error Status.
Reserved.
5:1
6
RO
0b
0b
0b
0b
0b
0b
0b
0b
R/W1CS
R/W1CS
R/W1CS
RO
Bad TLP Status.
Bad DLLP Status.
7
8
REPLAY_NUM Rollover Status.
Reserved.
11:9
12
13
15:14
R/W1CS
R/W1CS
RO
Replay Timer Timeout Status.
Advisory Non Fatal Error Status.
Reserved.
10.1.5.2.1.6 Correctable Error Mask, Offset 0x14
The Correctable Error Mask register controls reporting of individual correctable errors
by device to the host bridge via a PCIe error message. A masked error (respective bit
set in mask register) is not reported to the host bridge by an individual device. There is
a mask bit per bit in the Correctable Error Status register.
Bit Location
Attribute
RWS
Default Value
0b
Description
0
Receiver Error Mask.
Reserved.
5:1
6
RO
0b
0b
0b
0b
0b
0b
1b
0b
RWS
RWS
RWS
RO
Bad TLP Mask.
Bad DLLP Mask.
7
8
REPLAY_NUM Rollover Mask.
Reserved.
11:9
12
13
15:14
RWS
RWS
RO
Replay Timer Timeout Mask.
Advisory Non Fatal Error Mask.
Reserved.
10.1.5.2.1.7 First Error Pointer, Offset 0x18
The First Error Pointer is a read-only register that identifies the bit position of the first
uncorrectable error reported in the Uncorrectable Error Status register.
Bit Location
Attribute
RO
Default Value
0b
Description
Vector pointing to the first recorded error in the
Uncorrectable Error Status register.
3:0
321
82583V GbE Controller—Programing Interface
10.1.5.2.1.8 Header Log, Offset 0x1C
The header log register captures the header for the transaction that generated an error.
This register is 16 bytes.
Bit Location
Attribute
RO
Default Value
0x0
Description
127:0
Header of the defective packet (TLP or DLLP).
10.1.5.2.2
Device Serial Number Capability
The PCIe device serial number capability is an optional extended capability that can be
implemented by any PCIe device. The device serial number is a read-only 64-bit value
that is unique for a given PCIe device.
All multi-function devices that implement this capability must implement it for function
0; other functions that implement this capability must return the same device serial
number value as that reported by function 0. The 82583V is not a multi-function
device.
Table 55.
PCIe Device Serial Number Capability Structure
31
0
PCIe Enhanced Capability Header
Serial Number Register (Lower DW)
Serial Number Register (Upper DW)
10.1.5.2.2.1 Device Serial Number Enhanced Capability Header (Offset 0x00)
Figure 48 details the allocation of register fields in the device serial number enhanced
capability header. The Table below provides the respective bit definitions. The Extended
Capability ID for the Device Serial Number Capability is 0003h.
31
20 19
Capability Version
16 15
PCI Express Extended Capability ID
0
Next Capability Offset
Figure 48.
Allocation of Register Fields in the Device Serial Number Enhanced Capability
Header
Bit(s)
Location
Attributes
Description
PCIe Extended Capability ID
This field is a PCI-SIG defined ID number that indicates the nature and format of
the extended capability.
15:0
RO
Extended Capability ID for the Device Serial Number Capability is 0x0003.
Capability Version
This field is a PCI-SIG defined version number that indicates the version of the
capability structure present.
Must be 0x1 for this version of the specification.
19:16
31:20
RO
RO
Next Capability Offset
This field contains the offset to the next PCIe capability structure or 0x000 if no
other items exist in the linked list of capabilities.
For extended capabilities implemented in device configuration space, this offset is
relative to the beginning of PCI compatible configuration space and thus must
always be either 0x000 (for terminating list of capabilities) or greater than 0x0FF.
322
Programing Interface—82583V GbE Controller
10.1.5.2.2.2 Serial Number Register (Offset 0x04)
The Serial Number register is a 64-bit field that contains the IEEE defined 64-bit
extended unique identifier (EUI-64™). Figure 49 details the allocation of register fields
in the Serial Number register. The following table lists the respective bit definitions.
31
0
Serial Number Register (Lower DW)
Serial Number Register (Upper DW)
63
32
Figure 49.
Serial Number Register
Bit(s)
Attributes
Location
Description
PCIe Device Serial Number
This field contains the IEEE defined 64-bit extended unique identifier (EUI-64™).
This identifier includes a 24-bit company ID value assigned by IEEE registration
authority and a 40-bit extension identifier assigned by the manufacturer.
63:0
RO
10.1.5.2.2.3 Serial Number Definition in The 82583V
The serial number can be constructed from the 48-bit MAC address in the following
form:
Field
Order
Company ID
MAC Label
Addr+3 Addr+4
Extension identifier
Addr+0
Addr+1
Addr+2
Addr+5
Addr+6
Addr+7
Most significant bytes
Most significant bit
Least significant byte
Least significant bit
Figure 50.
Serial Number Definition in The 82583V 48-Bit MAC Address
The MAC label in the 82583V is 0xFFFF.
For example, assume that the company ID is (Intel) 00-A0-C9 and the extension
identifier is 23-45-67. In this case, the 64-bit serial number is:
Field
Order
Company ID
MAC Label
Extension identifier
Addr+0
00
Addr+1
A0
Addr+2
C9
Addr+3
FF
Addr+4
FF
Addr+5
23
Addr+6
45
Addr+7
67
Most significant byte
Most significant bit
Least significant byte
Least significant bit
The MAC address is the function 0 MAC address as loaded from NVM into the RAL and
RAH registers.
The official doc defining EUI-64 is: http://standards.ieee.org/regauth/oui/tutorials/
EUI64.html
323
82583V GbE Controller—Design Considerations
11.0
Design Considerations
This section provides general design considerations and recommendations when
selecting components and connecting special pins to the 82583V.
11.1
PCIe
11.1.1
Port Connection to the 82583V
PCIe is a dual simplex point-to-point serial differential low-voltage interconnect with a
signaling bit rate of 2.5 Gb/s per direction. The 82583V’s PCIe port consists of an
integral group of transmitters and receivers. The link between the PCIe ports of two
devices is a x1 lane that also consists of a transmitter and a receiver pair. Note that
each signal is 8b/10b encoded with an embedded clock.
The PCIe topology consists of a transmitter (Tx) located on one device connected
through a differential pair connected to the receiver (Rx) on a second device. The
82583V can be located on a motherboard or on an add-in card using a connector
specified by PCIe.
The lane is AC-coupled between its corresponding transmitter and receiver. The AC-
coupling capacitor is located on the board close to transmitter side. Each end of the link
is terminated on the die into nominal 100 Ω differential DC impedance. Board
termination is not required.
For more information on PCIe, refer to the PCI Express* Base Specification, Revision
1.1 and PCI Express* Card Electromechanical Specification, Revision 1.1RD.
For information about the 82583V’s PCIe power management capabilities, see
section 8.0.
11.1.2
11.1.3
PCIe Reference Clock
The 82583V uses a 100 MHz differential reference clock, denoted PECLKp and PECLKn.
This signal is typically generated on the system board and routed to the PCIe port. For
add-in cards, the clock is furnished at the PCIe connector.
The frequency tolerance for the PCIe reference clock is +/- 300 ppm.
Other PCIe Signals
The 82583V also implements other signals required by the PCIe specification. The
82583V signals power management events to the system using the PE_WAKE_N signal,
which operates very similarly to the familiar PCI PME# signal. Finally, there is a
PE_RST_N signal, which serves as the familiar reset function for the 82583V.
324
Design Considerations—82583V GbE Controller
11.1.4
11.2
PCIe Routing
Contact your Intel representative for information regarding the PCIe signal routing.
Clock Source
All designs require a 25 MHz clock source. The 82583V uses the 25 MHz source to
generate clocks up to 125 MHz and 1.25 GHz for the PHY circuits. For optimum results
with lowest cost, connect a 25 MHz parallel resonant crystal and appropriate load
capacitors at the XTAL1 and XTAL2 leads. The frequency tolerance of the timing device
should be 30 ppm or better. Refer to the Intel® Ethernet Controllers Timing Device
Selection Guide for more information on choosing crystals.
For further information regarding the clock for the 82583V, refer to the sections about
frequency control, crystals, and oscillators that follow.
11.2.1
Frequency Control Device Design Considerations
This section provides information regarding frequency control devices, including
crystals and oscillators, for use with all Intel Ethernet controllers. Several suitable
frequency control devices are available; none of which present any unusual challenges
in selection. The concepts documented herein are applicable to other data
communication circuits, including Platform LAN Connect devices (PHYs).
The Intel Ethernet controllers contain amplifiers, which when used with the specific
external components, form the basis for feedback oscillators. These oscillator circuits,
which are both economical and reliable, are described in more detail in section 11.3.1.
The Intel Ethernet controllers also have bus clock input functionality, however a
discussion of this feature is beyond the scope of this document, and will not be
addressed.
The chosen frequency control device vendor should be consulted early in the design
cycle. Crystal and oscillator manufacturers familiar with networking equipment clock
requirements may provide assistance in selecting an optimum, low-cost solution.
11.2.2
Frequency Control Component Types
Several types of third-party frequency reference components are currently marketed. A
discussion of each follows, listed in preferred order.
11.2.2.1
Quartz Crystal
Quartz crystals are generally considered to be the mainstay of frequency control
components due to their low cost and ease of implementation. They are available from
numerous vendors in many package types and with various specification options.
11.2.2.2
Fixed Crystal Oscillator
A packaged fixed crystal oscillator comprises an inverter, a quartz crystal, and passive
components conveniently packaged together. The device renders a strong, consistent
square wave output. Oscillators used with microprocessors are supplied in many
configurations and tolerances.
Crystal oscillators should be restricted to use in special situations, such as shared
clocking among devices or multiple controllers. As clock routing can be difficult to
accomplish, it is preferable to provide a separate crystal for each device.
325
82583V GbE Controller—Design Considerations
11.2.2.3
Programmable Crystal Oscillators
A programmable oscillator can be configured to operate at many frequencies. The
device contains a crystal frequency reference and a phase lock loop (PLL) clock
generator. The frequency multipliers and divisors are controlled by programmable
fuses.
A programmable oscillator’s accuracy depends heavily on the Ethernet device’s
differential transmit lines. The Physical Layer (PHY) uses the clock input from the
device to drive a differential Manchester (for 10 Mb/s operation), an MLT-3 (for 100
Mbps operation) or a PAM-5 (for 1000 Mbps operation) encoded analog signal across
the twisted pair cable. These signals are referred to as self-clocking, which means the
clock must be recovered at the receiving link partner. Clock recovery is performed with
another PLL that locks onto the signal at the other end.
PLLs are prone to exhibit frequency jitter. The transmitted signal can also have
considerable jitter even with the programmable oscillator working within its specified
frequency tolerance. PLLs must be designed carefully to lock onto signals over a
reasonable frequency range. If the transmitted signal has high jitter and the receiver’s
PLL loses its lock, then bit errors or link loss can occur.
PHY devices are deployed for many different communication applications. Some PHYs
contain PLLs with marginal lock range and cannot tolerate the jitter inherent in data
transmission clocked with a programmable oscillator. The American National Standards
Institute (ANSI) X3.263-1995 standard test method for transmit jitter is not stringent
enough to predict PLL-to-PLL lock failures, therefore, the use of programmable
oscillators is not recommended.
11.2.2.4
Ceramic Resonator
Similar to a quartz crystal, a ceramic resonator is a piezoelectric device. A ceramic
resonator typically carries a frequency tolerance of ±0.5%, – inadequate for use with
Intel Ethernet controllers, and therefore, should not be utilized.
326
Design Considerations—82583V GbE Controller
11.3
Crystal Support
11.3.1
Crystal Selection Parameters
All crystals used with Intel Ethernet controllers are described as AT-cut, which refers to
the angle at which the unit is sliced with respect to the long axis of the quartz stone.
Table 56 lists crystals which have been used successfully in other designs (however, no
particular product is recommended):
Table 56.
Crystal Manufacturers and Part Numbers
Manufacturer
KDS America
Part No.
DSX321G
NDK America Inc.
41CD25.0F1303018
7A25000165
9C25000008
TXC Corporation - USA
For information about crystal selection parameters, see section 11.7 and Table 18.
11.3.1.1
Vibrational Mode
Crystals in the above-referenced frequency range are available in both fundamental
and third overtone. Unless there is a special need for third overtone, use fundamental
mode crystals.
At any given operating frequency, third overtone crystals are thicker and more rugged
than fundamental mode crystals. Third overtone crystals are more suitable for use in
military or harsh industrial environments. Third overtone crystals require a trap circuit
(extra capacitor and inductor) in the load circuitry to suppress fundamental mode
oscillation as the circuit powers up. Selecting values for these components is beyond
the scope of this document.
11.3.1.2
11.3.1.3
11.3.1.4
Nominal Frequency
Intel Ethernet controllers use a crystal frequency of 25.000 MHz. The 25 MHz input is
used to generate a 125 MHz transmit clock for 100BASE-TX and 1000BASE-TX
operation – 10 MHz and 20 MHz transmit clocks, for 10BASE-T operation.
Frequency Tolerance
The frequency tolerance for an Ethernet Platform LAN Connect is dictated by the IEEE
802.3 specification as ±50 parts per million (ppm). This measurement is referenced to
a standard temperature of 25° C. Intel recommends a frequency tolerance of ±30 ppm.
Temperature Stability and Environmental Requirements
Temperature stability is a standard measure of how the oscillation frequency varies
over the full operational temperature range (and beyond). Some vendors separate
operating temperatures from temperature stability. Manufacturers may also list
temperature stability as 50 ppm in their data sheets.
Note:
Crystals also carry other specifications for storage temperature, shock resistance, and
reflow solder conditions. Crystal vendors should be consulted early in the design cycle
to discuss the application and its environmental requirements.
327
82583V GbE Controller—Design Considerations
11.3.1.5
Crystal Oscillation Mode
The terms series-resonant and parallel-resonant are often used to describe crystal
oscillator circuits. Specifying parallel mode is critical to determining how the crystal
frequency is calibrated at the factory.
A crystal specified and tested as series resonant oscillates without problem in a
parallel-resonant circuit, but the frequency is higher than nominal by several hundred
parts per million. The purpose of adding load capacitors to a crystal oscillator circuit is
to establish resonance at a frequency higher than the crystal’s inherent series resonant
frequency.
Figure 51 shows the recommended placement and layout of an internal oscillator
circuit. Note that pin X1 and X2 refers to XTAL1 and XTAL2 in the Ethernet device,
respectively. The crystal and the capacitors form a feedback element for the internal
inverting amplifier. This combination is called parallel-resonant, because it has positive
reactance at the selected frequency. In other words, the crystal behaves like an
inductor in a parallel LC circuit. Oscillators with piezoelectric feedback elements are
also known as “Pierce” oscillators.
11.3.1.6
Load Capacitance and Discrete Capacitors
The formula for crystal load capacitance is as follows:
(C1 ⋅ C2)
------------------
CL =
+ Cstray
(C1 + C2)
where:
CL is the rated Cload of the crystal component and C1 and C2 are discrete
crystal circuit capacitors.
C
stray allows for additional capacitance from solder pads, traces and the
82583V package. Individual stray capacitance components can be estimated
and added as parallel capacitances. Note that total Cstray is typically 3 pF to 7
pF.
Solve for the discrete capacitor values as follows:
C1 = C2 = 2 * [Cload - Cstray
]
For example:
If total Cstray = 4.0 pF and if the Cload rating is 18 pF, then the calculated C1
and C2 = 2 * [18 pF - 4.0 pF] = 28 pF.
Note:
Note:
Because 28 pF is not a standard value, use 27 pF capacitors for C1 and C2, which is the
closest standard value.
The oscillator frequency should be measured with a precision frequency counter where
possible. The values of C1 and C2 should be fine tuned for the design. As the actual
capacitive load increases, the oscillator frequency decreases.
Intel recommends COG or NPO capacitors with a tolerance of ±5% (approximately
±1 pF) or smaller.
328
Design Considerations—82583V GbE Controller
11.3.1.7
11.3.1.8
Shunt Capacitance
The shunt capacitance parameter is relatively unimportant compared to load
capacitance. Shunt capacitance represents the effect of the crystal’s mechanical holder
and contacts. The shunt capacitance should equal a maximum of 6 pF.
Equivalent Series Resistance
Equivalent Series Resistance (ESR) is the real component of the crystal’s impedance at
the calibration frequency, which the inverting amplifier’s loop gain must overcome. ESR
varies inversely with frequency for a given crystal family. The lower the ESR, the faster
the crystal starts up. Use crystals with an ESR value of 50 Ω or better.
11.3.1.9
Drive Level
Drive level refers to power dissipation in use. The allowable drive level for a Surface
Mounted Technology (SMT) crystal is less than its through-hole counterpart, because
surface mount crystals are typically made from narrow, rectangular AT strips, rather
than circular AT quartz blanks.
Some crystal data sheets list crystals with a maximum drive level of 1 mW. However,
Intel Ethernet controllers drive crystals to a level less than the suggested 0.3 mW
value. This parameter does not have much value for on-chip oscillator use.
11.3.1.10 Aging
Aging is a permanent change in frequency (and resistance) occurring over time. This
parameter is most important in its first year because new crystals age faster than old
crystals. Use crystals with a maximum of ±5 ppm per year aging.
11.3.1.11 Reference Crystal
The normal tolerances of the discrete crystal components can contribute to small
frequency offsets with respect to the target center frequency. To minimize the risk of
tolerance-caused frequency offsets causing a small percentage of production line units
to be outside of the acceptable frequency range, it is important to account for those
shifts while empirically determining the proper values for the discrete loading
capacitors, C1 and C2.
Even with a perfect support circuit, most crystals will oscillate slightly higher or slightly
lower than the exact center of the target frequency. Therefore, frequency
measurements (which determine the correct value for C1 and C2) should be performed
with an ideal reference crystal. When the capacitive load is exactly equal to the
crystal’s load rating, an ideal reference crystal will be perfectly centered at the desired
target frequency.
11.3.1.11.1 Reference Crystal Selection
There are several methods available for choosing the appropriate reference crystal:
• If a Saunders and Associates (S&A) crystal network analyzer is available, then
discrete crystal components can be tested until one is found with zero or nearly
zero ppm deviation (with the appropriate capacitive load). A crystal with zero or
near zero ppm deviation will be a good reference crystal to use in subsequent
frequency tests to determine the best values for C1 and C2.
329
82583V GbE Controller—Design Considerations
• If a crystal analyzer is not available, then the selection of a reference crystal can be
done by measuring a statistically valid sample population of crystals, which has
units from multiple lots and approved vendors. The crystal, which has an oscillation
frequency closest to the center of the distribution, should be the reference crystal
used during testing to determine the best values for C1 and C2.
• It may also be possible to ask the approved crystal vendors or manufacturers to
provide a reference crystal with zero or nearly zero deviation from the specified
frequency when it has the specified CLoad capacitance.
When choosing a crystal, customers must keep in mind that to comply with IEEE
specifications for 10/100 and 10/100/1000Base-T Ethernet LAN, the transmitter
reference frequency must be precise within 50 ppm. Intel® recommends customers to
use a transmitter reference frequency that is accurate to within 30 ppm to account for
variations in crystal accuracy due to crystal manufacturing tolerance.
11.3.1.11.2 Circuit Board
Since the dielectric layers of the circuit board are allowed some reasonable variation in
thickness, the stray capacitance from the printed board (to the crystal circuit) will also
vary. If the thickness tolerance for the outer layers of dielectric are controlled within
±17 percent of nominal, then the circuit board should not cause more than ±2 pF
variation to the stray capacitance at the crystal. When tuning crystal frequency, it is
recommended that at least three circuit boards are tested for frequency. These boards
should be from different production lots of bare circuit boards.
Alternatively, a larger sample population of circuit boards can be used. A larger
population will increase the probability of obtaining the full range of possible variations
in dielectric thickness and the full range of variation in stray capacitance.
Next, the exact same crystal and discrete load capacitors (C1 and C2) must be soldered
onto each board, and the LAN reference frequency should be measured on each circuit
board.
The circuit board, which has a LAN reference frequency closest to the center of the
frequency distribution, should be used while performing the frequency measurements
to select the appropriate value for C1 and C2.
11.3.1.11.3 Temperature Changes
Temperature changes can cause the crystal frequency to shift. Therefore, frequency
measurements should be done in the final system chassis across the system’s rated
operating temperature range.
11.3.2
Crystal Placement and Layout Recommendations
Crystal clock sources should not be placed near I/O ports or board edges. Radiation
from these devices can be coupled into the I/O ports and radiate beyond the system
chassis. Crystals should also be kept away from the Ethernet magnetics module to
prevent interference.
Note:
Failure to follow these guidelines could result in the 25 MHz clock failing to start.
When designing the layout for the crystal circuit, the following rules must be used:
• Place load capacitors as close as possible (within design-for-manufacturability
rules) to the crystal solder pads. They should be no more than 90 mils away from
crystal pads.
• The two load capacitors, crystal component, the Ethernet controller device, and the
crystal circuit traces must all be located on the same side of the circuit board
(maximum of one via-to-ground load capacitor on each XTAL trace).
330
Design Considerations—82583V GbE Controller
• Use 27 pF (5% tolerance) 0402 load capacitors.
• Place load capacitor solder pad directly in line with circuit trace (see Figure 51,
point A).
• Use 50 Ω impedance single-ended microstrip traces for the crystal circuit.
• Route traces so that electro-magnetic fields from XTAL2 do not couple onto XTAL1.
No differential traces.
• Route XTAL1 and XTAL2 traces to nearest inside corners of crystal pad (see
Figure 51, point B).
• Ensure that the traces from XTAL1 and XTAL2 are symmetrically routed and that
their lengths are matched.
• The total trace length of XTAL1 or XTAL2 should be less than 750 mils.
Crystal
“B”
“B”
Crystal Pad
Crystal Pad
Capacitor
Capacitor
“A”
27pF
0402
27pF
0402
Less than 660 mils
Xtal2
Xtal1
Ethernet Controller
Figure 51.
Recommended Crystal Placement and Layout
11.4
Oscillator Support
The 82583V clock input circuit is optimized for use with an external crystal. However,
an oscillator can also be used in place of the crystal with the proper design
considerations (see Table 19 for detailed clock oscillator specifications):
• The clock oscillator has an internal voltage regulator of 1.9 V dc to isolate it from
the external noise of other circuits to minimize jitter. If an external clock is used,
this imposes a maximum input clock amplitude of 1.9 V dc. For example, if a
3.3 V dc oscillator is used, it's signal should be attenuated to a maximum of
1.9 V dc with a resistive divider circuit.
331
82583V GbE Controller—Design Considerations
• The input capacitance introduced by the 82583V (approximately 20 pF) is greater
than the capacitance specified by a typical oscillator (approximately 15 pF).
• The input clock jitter from the oscillator can impact the 82583V clock and its
performance.
Note:
The power consumption of additional circuitry equals about 1.5 mW.
Table 57 lists oscillators that can be used with the 82583V. Please note that no
particular oscillator is recommended):
Table 57.
Oscillator Manufacturers and Part Numbers
Manufacturer
NDK AMERICA INC
Part No.
2560TKA-25M
6N25000160 or
7W25000025
TXC CORPORATION - USA
CITIZEN AMERICA CORP
Raltron Electronics Corp
MtronPTI
CSX750FJB25.000M-UT
CO4305-25.000-T-TR
M214TCN
Kyocera Corporation
KC5032C-C3
3.3 V dc
CLK
Oscillator
82583V
1000 pF
1 K ohm
VDD3p3
Out
XTAL1
C1
1 K ohm
Figure 52.
Oscillator Solution
11.4.1
Oscillator Placement and Layout Recommendations
Oscillator clock sources should not be placed near I/O ports or board edges. Radiation
from these devices can be coupled into the I/O ports and radiate beyond the system
chassis. Oscillators should also be kept away from the Ethernet magnetics module to
prevent interference.
332
Design Considerations—82583V GbE Controller
11.5
Ethernet Interface
11.5.1
Magnetics for 1000 BASE-T
Magnetics for the 82583V can be either integrated or discrete.
The magnetics module has a critical effect on overall IEEE and emissions conformance.
The device should meet the performance required for a design with reasonable margin
to allow for manufacturing variation. Occasionally, components that meet basic
specifications can cause the system to fail IEEE testing because of interactions with
other components or the printed circuit board itself. Carefully qualifying new magnetics
modules prevents this problem.
When using discrete magnetics it is necessary to use Bob Smith termination: Use four
75 Ω resistors for cable-side center taps and unused pins. This method terminates pair-
to-pair common mode impedance of the CAT5 cable.
Use an EFT capacitor attached to the termination plane. Suggested values are 1500 pF/
2 KV or 1000 pF/3 KV. A minimum of 50-mil spacing from capacitor to traces and
components should be maintained.
11.5.2
Magnetics Module Qualification Steps
The steps involved in magnetics module qualification are similar to those for crystal
qualification:
1. Verify that the vendor’s published specifications in the component datasheet meet
or exceed the specifications in section 11.6.
2. Independently measure the component’s electrical parameters on the test bench,
checking samples from multiple lots. Check that the measured behavior is
consistent from sample to sample and that measurements meet the published
specifications.
3. Perform physical layer conformance testing and EMC (FCC and EN) testing in real
systems. Vary temperature and voltage while performing system level tests.
11.5.3
Third-Party Magnetics Manufacturers
The following magnetics modules have been used successfully in previous designs.
Manufacturer
Part Number
Low Profile Discrete:
Midcom Inc.
000-7412-35R-LF1
Standard Discrete:
BelFuse
Pulse Eng.
S558-5999-P3 (12-core)
H5007NL (12-core)
Integrated:
FOXCONN
Pulse Eng.
Amphenol
BelFuse
JFM38U1C-L1U1W
JW0-0013NL
RJMG2310 22830ER C03-002
0862-1J1T-Z4-F
6368472-1
Tyco
333
82583V GbE Controller—Design Considerations
11.5.4
Designing the 82583V as a 10/100 Mb/s Only Device
To connect the 82583V as a 10/100 Mb/s only device:
1. Set bit 14 of the LED 1 Configuration Defaults/PHY Configuration (Word 0x1C) to
1b. Setting bit 14 to 1b disables 1000 Mb/s operation in all power modes (see
section 6.1.1.18).
2. Connect MDI pair 0 (pins 57 and 58) and MDI pair 1 (pins 54 and 55) to your
magnetics. MDI pair 2 (pins 52 and 53) and MDI pair 3 (pins 49 and 50) can each
be pulled up to 1.9 V dc through a 50 Ω resistor or connected to the magnetics as a
gigabit device if it is desired at a later time to enable gigabit by altering bit 14 of
the LED 1 Configuration Defaults/PHY Configuration; word 0x1C (Giga Disable).
See Figure 53 and Figure 54 for details.
3. Use one of the approved discrete gigabit magnetics that were tested with the
82583V (refer to section 11.5.3). Note that Intel has not tested any 10/100
magnetics with the 82583V.
Note:
If you must use 10/100 magnetics in your design with the 82583V, the magnetics must
have a Common Mode Choke (CMC) in the receive path.
The 82583V supports auto-MDIX; this feature CANNOT be disabled. If there is no CMC
in the receive path, the system most likely will not pass regulatory radiated emission
tests when MDI-X mode is used.
M DI_PLUS (0)
MDI_M INUS (0)
M DI_PLUS (1)
MDI_M INUS (1)
58
57
55
54
Gigabit Magnetics
Module
82583V
M DI_PLUS (2)
MDI_M INUS (2)
M DI_PLUS (3)
MDI_M INUS (3)
53
52
50
49
Figure 53.
82583V Gigabit Magnetics Module Connections
334
Design Considerations—82583V GbE Controller
MDI_PLUS (0)
M DI_MINUS (0)
MDI_PLUS (1)
M DI_MINUS (1)
58
57
55
54
LAN_1.9V
10/100 Magnetics
Module with Common Mode
Choke in Both Rx and Tx
Paths
82583V
50 Ohm
Resistors
MDI_PLUS (2)
M DI_MINUS (2)
MDI_PLUS (3)
M DI_MINUS (3)
53
52
50
49
Figure 54.
82583V 10/100 Mb/s Magnetics Module Connections (With CMC)
11.5.5
Layout Considerations for the Ethernet Interface
These sections provide recommendations for performing printed circuit board layouts.
Good layout practices are essential to meet IEEE PHY conformance specifications and
EMI regulatory requirements.
Critical signal traces should be kept as short as possible to decrease the likelihood of
being affected by high frequency noise from other signals, including noise carried on
power and ground planes. Keeping the traces as short as possible can also reduce
capacitive loading.
Since the transmission line medium extends onto the printed circuit board, special
attention must be paid to layout and routing of the differential signal pairs.
Designing for 1000 BASE-T Gigabit operation is very similar to designing for 10 and 100
Mb/s. For the 82583V, system level tests should be performed at all three speeds.
11.5.5.1
Guidelines for Component Placement
Component placement can affect signal quality, emissions, and component operating
temperature This section provides guidelines for component placement.
Careful component placement can:
• Decrease potential problems directly related to electromagnetic interference (EMI),
which could cause failure to meet applicable government test specifications.
• Simplify the task of routing traces. To some extent, component orientation will
affect the complexity of trace routing. The overall objective is to minimize turns and
crossovers between traces.
335
82583V GbE Controller—Design Considerations
Minimizing the amount of space needed for the Ethernet LAN interface is important
because other interfaces compete for physical space on a motherboard near the
connector. The Ethernet LAN circuits need to be as close as possible to the connector.
Integrated
RJ-45
Keep silicon traces at least 1" from edge of
PB (2" is preferred).
w/LAN
Magnetics
Keep LAN silicon 1" - 4" from LAN connector.
Keep minimum distance between differential pairs
more than seven times the dielectric thickness away
from each other and other traces, including NVM
traces and parallel digital traces.
LAN
Silicon
Note: Figure 55 represents a 10/100 diagram. Use the same design considerations for the two
differential pairs not shown for gigabit implementations.
Figure 55.
General Placement Distances for 1000 BASE-T Designs
Figure 55 shows some basic placement distance guidelines. Figure 55 shows two
differential pairs, but can be generalized for a Gigabit system with four analog pairs.
The ideal placement for the Ethernet silicon would be approximately one inch behind
the magnetics module.
While it is generally a good idea to minimize lengths and distances, Figure 55 also
illustrates the need to keep the LAN silicon away from the edge of the board and the
magnetics module for best EMI performance.
11.5.5.2
Layout Guidelines for Use with Integrated and Discrete Magnetics
Layout requirements are slightly different when using discrete magnetics.
These include:
• Ground cut for HV installation (not required for integrated magnetics)
• A maximum of two (2) vias
• Turns less than 45°
• Discrete terminators
Figure 56 shows a reference layout for discrete magnetics.
336
Design Considerations—82583V GbE Controller
Magnetics Module
82583V
RJ-45
Figure 56.
11.5.5.3
Layout for Discrete Magnetics
Board Stack-Up Recommendations
Printed circuit boards for these designs typically have four, six, eight, or more layers.
Although, the 82583V does not dictate the stack up, here is an example of a typical six-
layer board stack up:
• Layer 1 is a signal layer. It can contain the differential analog pairs from the
Ethernet device to the magnetics module.
• Layer 2 is a signal ground layer. Chassis ground may also be fabricated in Layer 2
under the connector side of the magnetics module.
• Layer 3 is used for power planes.
• Layer 4 is a signal layer.
• Layer 5 is an additional ground layer.
• Layer 6 is a signal layer. For 1000 BASE-T (copper) Gigabit designs, it is common to
route two of the differential pairs (per port) on this layer.
This board stack up configuration can be adjusted to conform to specific OEM design
rules.
337
82583V GbE Controller—Design Considerations
11.5.5.4
Differential Pair Trace Routing for 10/100/1000 Designs
Trace routing considerations are important to minimize the effects of crosstalk and
propagation delays on sections of the board where high-speed signals exist. Signal
traces should be kept as short as possible to decrease interference from other signals,
including those propagated through power and ground planes. Observe the following
suggestions to help optimize board performance:
• Maintain constant symmetry and spacing between the traces within a differential
pair.
• Minimize the difference in signal trace lengths of a differential pair.
• Keep the total length of each differential pair under 4 inches. Although possible,
designs with differential traces longer than 5 inches are much more likely to have
degraded receive BER (Bit Error Rate) performance, IEEE PHY conformance
failures, and/or excessive EMI (Electromagnetic Interference) radiation.
• Keep differential pairs more than seven times the dielectric thickness away from
each other and other traces, including NVM traces and parallel digital traces.
• Keep maximum separation within differential pairs to 7 mils.
• For high-speed signals, the number of corners and vias should be kept to a
minimum. If a 90° bend is required, it is recommended to use two 45° bends
instead. Refer to Figure 57.
Note:
In manufacturing, vias are required for testing and troubleshooting purposes. The via
size should be a 17-mil (±2 mils for manufacturing variance) finished hole size (FHS).
• Traces should be routed away from board edges by a distance greater than the
trace height above the reference plane. This allows the field around the trace to
couple more easily to the ground plane rather than to adjacent wires or boards.
• Do not route traces and vias under crystals or oscillators. This will prevent coupling
to or from the clock. And as a general rule, place traces from clocks and drives at a
minimum distance from apertures by a distance that is greater than the largest
aperture dimension
.
45°
45°
Figure 57.
Trace Routing
• The reference plane for the differential pairs should be continuous and low
impedance. It is recommended that the reference plane be either ground or
1.9 V dc (the voltage used by the PHY). This provides an adequate return path for
and high frequency noise currents.
• Do not route differential pairs over splits in the associated reference plane as it
may cause discontinuity in impedances.
338
Design Considerations—82583V GbE Controller
11.5.5.5
11.5.5.6
Signal Termination and Coupling
The 82547L has internal termination on the MDI signals. External resistors are not
needed. Adding pads for external resistors can degrade signal integrity.
Signal Trace Geometry for 1000 BASE-T Designs
The key factors in controlling trace EMI radiation are the trace length and the ratio of
trace-width to trace-height above the reference plane. To minimize trace inductance,
high-speed signals and signal layers that are close to a reference or power plane should
be as short and wide as practical. Ideally, this trace width to height above the ground
plane ratio is between 1:1 and 3:1. To maintain trace impedance, the width of the trace
should be modified when changing from one board layer to another if the two layers are
not equidistant from the neighboring planes.
Each pair of signal should have a differential impedance of 100 Ω. +/- 15%. If a
particular tool cannot design differential traces, it is permissible to specify 55-65 Ω
single-ended traces as long as the spacing between the two traces is minimized. As an
example, consider a differential trace pair on Layer 1 that is 8 mils (0.2 mm) wide and
2 mils (0.05 mm) thick, with a spacing of 8 mils (0.2 mm). If the fiberglass layer is 8
mils (0.2 mm) thick with a dielectric constant, ER, of 4.7, the calculated single-ended
impedance would be approximately 61 Ω and the calculated differential impedance
would be approximately 100 Ω.
When performing a board layout, do not allow the CAD tool auto-router to route the
differential pairs without intervention. In most cases, the differential pairs will have to
be routed manually.
Note:
Measuring trace impedance for layout designs targeting 100 Ω often results in lower
actual impedance. Designers should verify actual trace impedance and adjust the
layout accordingly. If the actual impedance is consistently low, a target of 105 – 110 Ω
should compensate for second order effects.
It is necessary to compensate for trace-to-trace edge coupling, which can lower the
differential impedance by up to 10 Ω, when the traces within a pair are closer than 30
mils (edge to edge).
11.5.5.7
Trace Length and Symmetry for 1000 BASE-T Designs
As indicated earlier, the overall length of differential pairs should be less than four
inches measured from the Ethernet device to the magnetics.
The differential traces (within each pair) should be equal in total length to within 50
mils (1.25 mm) and as symmetrical as possible. Asymmetrical and unequal length
traces in the differential pairs contribute to common mode noise. If a choice has to be
made between matching lengths and fixing symmetry, more emphasis should be placed
on fixing symmetry. Common mode noise can degrade the receive circuit’s performance
and contribute to radiated emissions.
11.5.5.8
Routing 1.9 V dc to the Magnetics Center Tap
The central-tap 1.9 V dc should be delivered as a solid supply plane (1.9 V dc) directly
to the magnetic module or, if this is not possible, by a short and thick trace (lower than
0.2 Ω DC resistance). The decoupling capacitors for the central tap pins should be
placed as close as possible to the magnetic component. This improves both EMI and
IEEE compliance.
339
82583V GbE Controller—Design Considerations
11.5.5.9
Impedance Discontinuities
Impedance discontinuities cause unwanted signal reflections. Minimize vias (signal
through holes) and other transmission line irregularities. If vias must be used, a
reasonable budget is two per differential trace. Unused pads and stub traces should
also be avoided.
11.5.5.10 Reducing Circuit Inductance
Traces should be routed over a continuous reference plane with no interruptions. If
there are vacant areas on a reference or power plane, the signal conductors should not
cross the vacant area. This causes impedance mismatches and associated radiated
noise levels. Noisy logic grounds should be separated from analog signal grounds to
reduce coupling. Noisy logic grounds can sometimes affect sensitive DC subsystems
such as analog to digital conversion, operational amplifiers, etc. All ground vias should
be connected to every ground plane; and similarly, every power via, to all power planes
at equal potential. This helps reduce circuit inductance. Another recommendation is to
physically locate grounds to minimize the loop area between a signal path and its
return path. Rise and fall times should be as slow as possible. Because signals with fast
rise and fall times contain many high frequency harmonics, which can radiate
significantly. The most sensitive signal returns closest to the chassis ground should be
connected together. This will result in a smaller loop area and reduce the likelihood of
crosstalk. The effect of different configurations on the amount of crosstalk can be
studied using electronics modeling software.
11.5.5.11 Signal Isolation
To maintain best signal integrity, keep digital signals far away from the analog traces. A
good rule of thumb is no digital signal should be within 300 mils (7.5 mm) of the
differential pairs. If digital signals on other board layers cannot be separated by a
ground plane, they should be routed perpendicular to the differential pairs. If there is
another LAN controller on the board, take care to keep the differential pairs from that
circuit away.
Some rules to follow for signal isolation:
• Separate and group signals by function on separate layers if possible. Keep a
minimum distance between differential pairs more than seven times the dielectric
thickness away from each other and other traces, including NVM traces and parallel
digital traces.
• Physically group together all components associated with one clock trace to reduce
trace length and radiation.
• Isolate I/O signals from high-speed signals to minimize crosstalk, which can
increase EMI emission and susceptibility to EMI from other signals.
• Avoid routing high-speed LAN traces near other high-frequency signals associated
with a video controller, cache controller, processor, or other similar devices.
11.5.5.12 Traces for Decoupling Capacitors
Traces between decoupling and I/O filter capacitors should be as short and wide as
practical. Long and thin traces are more inductive and would reduce the intended effect
of decoupling capacitors. Also for similar reasons, traces to I/O signals and signal
terminations should be as short as possible. Vias to the decoupling capacitors should be
sufficiently large in diameter to decrease series inductance.
340
Design Considerations—82583V GbE Controller
11.5.5.13 Light Emitting Diodes for Designs Based on the 82583V
The 82583V provides three programmable high-current push-pull (active high) outputs
to directly drive LEDs for link activity and speed indication. Each LAN device provides
an independent set of LED outputs; these pins and their function are bound to a specific
LAN device. Each of the four LED outputs can be individually configured to select the
particular event, state, or activity, which is indicated on that output. In addition, each
LED can be individually configured for output polarity, as well as for blinking versus
non-blinking (steady-state) indication.
Since the LEDs are likely to be integral to a magnetics module, take care to route the
LED traces away from potential sources of EMI noise. In some cases, it may be
desirable to attach filter capacitors.
The LED ports are fully programmable through the NVM interface.
11.5.6
Physical Layer Conformance Testing
Physical layer conformance testing (also known as IEEE testing) is a fundamental
capability for all companies with Ethernet LAN products. PHY testing is the final
determination that a layout has been performed successfully. If your company does not
have the resources and equipment to perform these tests, consider contracting the
tests to an outside facility.
11.5.6.1
Conformance Tests for 10/100/1000 Mb/s Designs
Crucial tests are as follows, listed in priority order:
• Bit Error Rate (BER). Good indicator of real world network performance. Perform bit
error rate testing with long and short cables and many link partners. The test limit
is 10-11 errors.
• Output Amplitude, Rise and Fall Time (10/100 Mb/s), Symmetry and Droop
(1000Mbps). For the 82575 controller, use the appropriate PHY test waveform.
• Return Loss. Indicator of proper impedance matching, measured through the RJ-45
connector back toward the magnetics module.
• Jitter Test (10/100 Mb/s) or Unfiltered Jitter Test (1000 Mb/s). Indicator of clock
recovery ability (master and slave for Gigabit controller).
11.5.7
Troubleshooting Common Physical Layout Issues
The following is a list of common physical layer design and layout mistakes in LAN On
Motherboard Designs.
1. Lack of symmetry between the two traces within a differential pair. Asymmetry can
create common-mode noise and distort the waveforms. For each component and/or
via that one trace encounters, the other trace should encounter the same
component or a via at the same distance from the Ethernet silicon.
2. Unequal length of the two traces within a differential pair. Inequalities create
common-mode noise and will distort the transmit or receive waveforms.
3. Excessive distance between the Ethernet silicon and the magnetics. Long traces on
FR4 fiberglass epoxy substrate will attenuate the analog signals. In addition, any
impedance mismatch in the traces will be aggravated if they are longer than the
four inch guideline.
341
82583V GbE Controller—Design Considerations
4. Routing any other trace parallel to and close to one of the differential traces.
Crosstalk getting onto the receive channel will cause degraded long cable BER.
Crosstalk getting onto the transmit channel can cause excessive EMI emissions and
can cause poor transmit BER on long cables. At a minimum, other signals should be
kept 0.3 inches from the differential traces.
5. Routing one pair of differential traces too close to another pair of differential traces.
After exiting the Ethernet silicon, the trace pairs should be kept 0.3 inches or more
away from the other trace pairs. The only possible exceptions are in the vicinities
where the traces enter or exit the magnetics, the RJ-45 connector, and the
Ethernet silicon.
6. Use of a low-quality magnetics module.
7. Re-use of an out-of-date physical layer schematic in a Ethernet silicon design. The
terminations and decoupling can be different from one PHY to another.
8. Incorrect differential trace impedances. It is important to have ~100 Ω impedance
between the two traces within a differential pair. This becomes even more
important as the differential traces become longer. To calculate differential
impedance, many impedance calculators only multiply the single-ended impedance
by two. This does not take into account edge-to-edge capacitive coupling between
the two traces. When the two traces within a differential pair are kept close to each
other, the edge coupling can lower the effective differential impedance by 5 Ω to
20 Ω. Short traces have fewer problems if the differential impedance is slightly off
target.
11.6
82583V Power Supplies
The 82583V requires three power rails: 3.3 V dc, 1.9 V dc, and 1.05 V dc (see
section 8.4). A central power supply can provide all the required voltage sources or the
power can be derived from the 3.3 V dc supply and regulated locally using external
regulators. If the LAN wake capability is used, all voltages must remain present during
system power down. Local regulation of the LAN voltages from system 3.3 Vmain and
3.3 Vaux voltages is recommended. Refer to section 11.3 and section 11.5 for detailed
information about power supply sequencing rules and intended design options for
power solutions.
External voltage regulators need to generate the proper voltage, supply current
requirements (with adequate margin), and provide the proper power sequencing.
11.6.1
82583V GbE Controller Power Sequencing
Designs must comply with power sequencing requirements to avoid latch-up and
forward-biased internal diodes (see Figure 58).
The general guideline for sequencing is:
1. Power up the 3.3 V dc rail.
2. Power up the 1.9 V dc next.
3. Power up the 1.05 V dc rail last.
For power down, there is no requirement (only charge that remains is stored in the
decoupling capacitors).
342
Design Considerations—82583V GbE Controller
VDD3p3
AVDD1p9
VDD1p0
Figure 58.
11.6.1.1
Power Sequencing Guideline
Power Up Sequence (External LVR)
The board designer controls the power up sequence with the following stipulations (see
Figure 59):
• 1.9 V dc must not exceed 3.3 V dc by more than 0.3 V dc.
• 1.05 V dc must not exceed 1.9 V dc by more than 0.3 V dc.
• 1.05 V dc must not exceed 3.3 V dc by more than 0.3 V dc.
VDD3p3
AVDD1p9
VDD1p0
Figure 59.
External LVR Power-up Sequence
343
82583V GbE Controller—Design Considerations
11.6.1.2
Power Up-Sequence (Internal LVR)
The 82583V controls the power-up sequence internally and automatically with the
following conditions (see Figure 60):
• 3.3 V dc must be the source for the internal LVR.
• 1.9 V dc never exceeds 3.3 V dc.
• 1.05 V dc never exceeds 3.3 V dc or 1.9 V dc.
The ramp is delayed internally, with Tdelay depending on the rising slope of the 3.3 V dc
ramp.
VDD3p3
AVDD1p9
VDD1p0
Figure 60.
Internal LVR Power-Up Sequence
11.6.2
Power and Ground Planes
Good grounding requires minimizing inductance levels in the interconnections and
keeping ground returns short, signal loop areas small, and power inputs bypassed to
signal return, will significantly reduce EMI radiation.
The following guidelines help reduce circuit inductance in both backplanes and
motherboards:
• Route traces over a continuous plane with no interruptions. Do not route over a
split power or ground plane. If there are vacant areas on a ground or power plane,
avoid routing signals over the vacant area. This will increase inductance and EMI
radiation levels.
• Separate noisy digital grounds from analog grounds to reduce coupling. Noisy
digital grounds may affect sensitive DC subsystems.
• All ground vias should be connected to every ground plane; and every power via
should be connected to all power planes at equal potential. This helps reduce circuit
inductance.
• Physically locate grounds between a signal path and its return. This will minimize
the loop area.
• Avoid fast rise/fall times as much as possible. Signals with fast rise and fall times
contain many high frequency harmonics, which can radiate EMI.
344
Design Considerations—82583V GbE Controller
• The ground plane beneath a magnetics module should be split. The RJ45 connector
side of the transformer module should have chassis ground beneath it.
• Power delivery traces should be a minimum of 100 mils wide at all places from the
source to the destination. As power flows through pass transistors or regulators,
the traces must be kept wide as well. The distribution of power is better done with
a copper-pore under the PHY. This provides low inductance connectivity to
decoupling capacitors. Decoupling capacitors should be placed as close as possible
to the point of use and should avoid sharing vias with other decoupling capacitors.
Decoupling capacitor placement control should be done for the PHY as well as pass
transistors or regulators.
11.7
Device Disable
For a LOM design, it might be desirable for the system to provide BIOS-setup capability
for selectively enabling or disabling LOM devices. This enables designers more control
over system resource-management, avoid conflicts with add-in NIC solutions, etc. The
82583V provides support for selectively enabling or disabling it.
Device disable is initiated by asserting the asynchronous DEV_OFF_N pin. The
DEV_OFF_N pin has an internal pull-up resistor, so that it can be left not connected to
enable device operation.
The NVM’s Device Disable Power Down En bit enables device disable mode (hardware
default is that the mode is disabled).
While in device disable mode, the PCIe link is in L3 state. The PHY is in power down
mode. Output buffers are tri-stated.
Assertion or deassertion of PCIe PE_RST_N does not have any effect while the 82583V
is in device disable mode (that is, the 82583V stays in the respective mode as long as
DEV_OFF_N is asserted). However, the 82583V might momentarily exit the device
disable mode from the time PCIe PE_RST_N is de-asserted again and until the NVM is
read.
During power-up, the DEV_OFF_N pin is ignored until the NVM is read. From that point,
the 82583V might enter device disable if DEV_OFF_N is asserted.
Note:
The DEV_OFF_N pin should maintain its state during system reset and system sleep
states. It should also insure the proper default value on system power up. For example,
a designer could use a GPIO pin that defaults to 1b (enable) and is on system suspend
power. For example, it maintains the state in S0-S5 ACPI states).
11.7.1
BIOS Handling of Device Disable
Assume that in the following power-up sequence the DEV_OFF_N signal is driven high
(or it is already disabled)
1. The PCIe is established following the GIO_PWR_GOOD.
2. BIOS recognizes that the entire 82583V should be disabled.
3. The BIOS drives the DEV_OFF_N signal to the low level.
4. As a result, the 82583V samples the DEV_OFF_N signals and enters either the
device disable mode.
5. The BIOS could put the link in the Electrical IDLE state (at the other end of the PCIe
link) by clearing the Link Disable bit in the Link Control register.
6. BIOS might start with the device enumeration procedure (the entire 82583V
functions are invisible).
345
82583V GbE Controller—Design Considerations
7. Proceed with normal operation
8. Re-enable could be done by driving high the DEV_OFF_N signal, followed later by
bus enumeration.
11.8
82583V Exposed Pad*
Introduction
11.8.1
The 82583V is a 64-pin, 9 x 9 QFN package with an Exposed-Pad*. The Exposed-Pad*
is a central pad on the bottom of the package that provides the primary heat removal
path as well as electrical grounding for a Printed Circuit Board (PCB).
In order to maximize both the removal of heat from the package and the electrical
performance, a landing pattern must be incorporated on the PCB within the footprint of
the package corresponding to the exposed metal pad or exposed heat slug on the
package. The size of the landing pattern can be larger, smaller, or even take on a
different shape than the Exposed-Pad* on the package. However, the solderable area,
as defined by the solder mask, should be at least the same size/shape as the Exposed-
Pad* on the package to maximize the thermal/electrical performance.
While the landing pattern on the PCB provides a means of heat transfer/electrical
grounding from the package to the board through a solder joint, thermal vias are
necessary to effectively conduct from the surface of the PCB to the ground plane(s).
The number of vias are application specific and dependent upon the package power
dissipation as well as electrical conductivity requirements. As a result, thermal and
electrical analysis and/or testing are recommended to determine the minimum number
needed.
Warning:
Make sure that the 82583V has a good connection to ground. Check for solder voids on
the Exposed Pad,* solder wicking, or a complete lack of solder. Failure to ensure a good
connection to ground can result in functional failure.
The remainder of this section describes the silkscreen/component pads, solder mask,
solder paste, and two potential landing patterns that can be used for the 82583V
package. Note that these potential landing patterns have been used successfully in past
designs, however no particular landing pattern is recommended. Please work with your
manufacturer and assembler to ensure a process that is reliable.
11.8.2
Component Pad, Solder Mask and Solder Paste
Figure 61, Figure 62, and Figure 63 show the silkscreen/components pad, solder mask
and solder paste area for the 82583V package.
346
Design Considerations—82583V GbE Controller
Figure 61.
82583V Silkscreen and Components Pad (Top View)
Figure 62.
82583V Solder Mask
347
82583V GbE Controller—Design Considerations
0.12 in.
0.30 mm
0.12 in.
0.30 mm
0.054 in. (1.38 mm) Square x 9
Figure 63.
82583V Solder Paste
The stencil for the solder paste should be 5 mils thick. Also, use a solder paste alloy
consisting of 96.5Sn/3Ag/0.5Cu for a lead free process.
11.8.3
Landing Pattern A (No Via In Pad)
This landing pattern (vias outside Exposed Pad*) provides an extended ground
connection, adequate solder coverage and less solder voiding; however, it does not
provide thermal relief. This landing pattern also meets Intel’s recommendation for
coverage >= 80%.
Extended Ground Connection
Without Thermal Relief
Solder Mask Opening
Metal Pattern
Figure 64.
82583V Landing Pattern A (Top View - Vias on the Outside of the Exposed
Pad*)
Use 12 vias distributed on four sides (three per side, as shown in Figure 64) or three
sides (four per side). Additional vias can be added to improve conductivity. If larger
vias can be used (14 to 20 mil finished hole size), then a minimum of 9 vias can be
evenly placed around the extended ground connection.
348
Design Considerations—82583V GbE Controller
11.8.4
Landing Pattern B (Thermal Relief; No Via In Pad)
This landing pattern (vias outside Exposed Pad*) provides thermal relief, adequate
solder coverage, and less solder voiding; however, it does not provide an extended
ground connection. This landing pattern also meets Intel’s recommendation for
coverage >= 80%.
thermal relief
8-spoke pattern
40 mil mimimum
44 mil anti-pad
32 mil via pad
10 mil finished hole (small via)
14 to 20 mil finished hole (large thermal via)
Figure 65.
82583V Landing Pattern B (Top View - Vias on the Outside of the Exposed
Pad*)
Intel recommends using 16 vias evenly placed (as shown in Figure 65) around the
extended ground connection. Additional vias can be added to improve conductivity. A
minimum of 12 larger vias (14 to 20 mil finished hole size) can also be used.
349
82583V GbE Controller—Design Considerations
11.9
Assembly Process Flow
Figure 66 shows the typical process flow for mounting packages to the PCB.
Figure 66.
Assembly Flow
11.10
Reflow Guidelines
The typical reflow profile consists of four sections. In the preheat section, the PCB
assembly should be preheated at the rate of 1 to 2 °C/sec to start the solvent
evaporation and to avoid thermal shock. The assembly should then be thermally
soaked for 60 to 120 seconds to remove solder paste volatiles and for activation of flux.
The reflow section of the profile, the time above liquidus should be between 45 to 60
seconds with a peak temperature in the range of 245 to 250 °C, and the duration at the
peak should not exceed 30 seconds. Finally, the assembly should undergo cool down in
the fourth section of the profile. A typical profile band is provided in Figure 67, in which
220 °C is referred to as an approximation of the liquidus point. The actual profile
parameters depend upon the solder paste used and specific recommendations from the
solder paste manufacturers should be followed.
350
Design Considerations—82583V GbE Controller
Figure 67.
Typical Profile Band
Note:
1. Preheat: 125 °C -220 °C, 150 - 210 s at 0.4 k/s to 1.0 k/s
2. Time at T > 220 °C: 60 - 90 s
3. Peak Temperature: 245-250 °C
4. Peak time: 10 - 30 s
5. Cooling rate: <= 6 k/s
6. Time from 25 °C to Peak: 240 – 360 s
351
82583V GbE Controller—Design Considerations
11.11
XOR Testing
Note:
BSDL files are not available for the 82583V.
A common board or system-level manufacturing test for proper electrical continuity
between the 82583V and the board is some type of cascaded-XOR or NAND tree test.
The 82583V implements an XOR tree spanning most I/O signals. The component XOR
tree consists of a series of cascaded XOR logic gates, each stage feeding in the
electrical value from a unique pin. The output of the final stage of the tree is visible on
an output pin from the component.
Figure 68.
XOR Tree Concept
By connecting to a set of test-points or bed-of-nails fixture, a manufacturing test
fixture can test connectivity to each of the component pins included in the tree by
sequentially testing each pin, testing each pin when driven both high and low, and
observing the output of the tree for the expected signal value and/or change.
Note:
Some of the pins that are inputs for the XOR test are listed as “may be left
disconnected” in the pin descriptions. If XOR test is used, all inputs to the XOR tree
must be connected.
When the XOR tree test is selected, the following behaviors occur:
• Output drivers for the pins listed as “tested” are all placed in high-impedance (tri-
state) state to ensure that board/system test fixture can drive the tested inputs
without contention.
• Internal pull-up and pull-down devices for pins listed as “tested” are also disabled
to further ensure no contention with the board/system test fixture.
• The XOR tree is output on the LED1 pin.
To enter the XOR tree mode, a specific JTAG pattern must be sent to the test interface.
This pattern is described by the following TDF pattern: (dh = Drive High, dl = Drive
Low)
dh (TEST_EN, JTAG_TDI) dl(JTAG_TCK,JTAG_TMS);
dh(JTAG_TCK);
dl(JTAG_TCK);
dh(JTAG_TMS);
loop 2
dh(JTAG_TCK);
dl(JTAG_TCK);
end loop
dl(JTAG_TMS);
loop 2
dh(JTAG_TCK);
dl(JTAG_TCK);
end loop
352
Design Considerations—82583V GbE Controller
dl(JTAG_TDI);
dh(JTAG_TCK);
dl(JTAG_TCK);
dh(JTAG_TDI);
dh(JTAG_TCK);
dl(JTAG_TCK);
dl(JTAG_TDI);
dh(JTAG_TCK);
dl(JTAG_TCK);
dh(JTAG_TDI);
dh(JTAG_TCK);
dl(JTAG_TCK);
dl(JTAG_TDI);
dh(JTAG_TCK);
dl(JTAG_TCK);
dh(JTAG_TDI)
dh(JTAG_TMS);
dh(JTAG_TCK);
dl(JTAG_TCK);
dl(JTAG_TMS);
dh(JTAG_TCK);
dl(JTAG_TCK);
dh(JTAG_TMS);
dh(JTAG_TCK);
dl(JTAG_TCK);
dh(JTAG_TCK);
dl(JTAG_TCK);
dl(JTAG_TMS);
dh(JTAG_TCK);
dl(JTAG_TCK);
hold(JTAG_TMS,TEST_EN,JTAG_TCK,JTAG_TDI);
Note:
XOR tree reads left-to-right top-to-bottom.
Table 58.
Tested Pins Included in XOR Tree (7 pins)
Pin Name
Pin Name
Pin Name
LED2
NVM_SK
NVM_SO
LED0
NVM_SI
NVM_CS_N
LED1 (output of the XOR tree)
353
82583V GbE Controller—Thermal Design Considerations
12.0
Thermal Design Considerations
12.1
Introduction
This section describes the 82583V thermal characteristics and suggested thermal
solutions. Use this section to properly design a thermal solution for systems
implementing the 82583V.
Properly designed solutions provide adequate cooling to maintain the 82583V case
temperature (Tcase) at or below those listed in Table 60. Ideally, this is accomplished
by providing a low, local ambient temperature and creating a minimal thermal
resistance to that local ambient temperature. By maintaining the 82583V case
temperature at or below those recommended in this section, the 82583V will function
properly and reliably.
12.2
12.3
Intended Audience
The intended audience for this section is system design engineers using the 82583V.
System designers are required to address component and system-level thermal
challenges as the market continues to adopt products with higher-speeds and port
densities. New designs might be required to provide better cooling solutions for silicon
devices depending on the type of system and target operating environment.
Measuring the Thermal Conditions
This section provides a method for determining the operating temperature of the
82583V in a specific system based on case temperature. Case temperature is a function
of the local ambient and internal temperatures of the component. This section specifies
a maximum allowable Tcase for the 82583V.
Note:
Removal of the shield lid is required to measure the case temperature.
12.4
Thermal Considerations
Component temperature in a system environment is a function of the component,
board, and system thermal characteristics. The board/system-level thermal constraints
consist of the following:
• Local ambient temperature near the component
• Airflow over the component and surrounding board
• Physical constraints at, above, and surrounding the component that might limit the
size of a thermal enhancement
354
Thermal Design Considerations—82583V GbE Controller
• The component die temperature depends on the following:
— Component power dissipation
— Size
— Packaging materials (effective thermal conductivity)
— Type of interconnection to the substrate and motherboard
— Presence of a thermal cooling solution
— Thermal conductivity
— Power density of the substrate/package, nearby components, and circuit board
that is attached to it
Technology trends continue to push these parameters toward increased performance
levels (higher operating speeds), I/O density (smaller packages), and silicon density
(more transistors). Power density increases and thermal cooling solution space and
airflow become more constrained as operating frequencies increase and packaging
sizes decrease. These issues result in an increased emphasis on the following:
• Package and thermal enhancement technology to remove heat from the device.
• System design to reduce local ambient temperatures and ensure that thermal
design requirements are met for each component in the system.
12.5
Packaging Terminology
The following is a list of packaging terminology used in this section:
• Quad Flat No Leads - Plastic encapsulated package with a copper leadframe
substrate. Package uses perimeter lands on the bottom of the package to provide
electrical contact to the PCB. This package is also known as QFN.
• Junction - Refers to a P-N junction on the silicon. In this section, it is used as a
temperature reference point (for example, Theta JA refers to the junction to
ambient temperature).
• Ambient - Refers to local ambient temperature of the bulk air approaching the
component. It can be measured by placing a thermocouple approximately one inch
upstream from the component edge.
• Lands - The pads on the PCB that the BGA balls are soldered to.
• PCB - Printed Circuit Board.
• Printed Circuit Assembly (PCA) - An assembled PCB.
• Thermal Design Power (TDP) - The estimated maximum possible/expected power
generated in a component by a realistic application. Use the maximum power
requirement numbers from Table 59.
• LFM - Linear Feet per Minute (airflow)
12.6
Product Package Thermal Specification
Table 59.
Package Thermal Characteristics in Standard JEDEC Environment
Est. Power
(TDP)
Package Type
TJ Max
ΘJA
ΨJT
°
°
°
9 mm-64 QFN
473 mW
39.5 C/W
0.7 C/W
120 C
355
82583V GbE Controller—Thermal Design Considerations
Note:
Θ
JC and ΘJB are not included as a part of the thermal parameters for the 82583. ΘJC
and ΘJB each assume that all heat flows to either the case or board. However, since the
heat actually flows in multiple directions, ΨJT is more applicable to making thermal
calculations with the 82583.
The thermal parameters listed in Table 59 are based on simulated results of packages
assembled on a 4-layer 30 x 56 mm mini PCIe board connected to a system board in a
natural convection environment. The maximum case temperature is based on the
maximum junction temperature and defined by the relationship, Tcase-max = Tjmax -
(ΨJT x Power) where ΨJT is the junction-to-package top thermal characterization
parameter. ΘJA is the package junction-to-air thermal resistance.
12.7
Thermal Specifications
To ensure proper operation and reliability of the 82583V, the thermal solution must
maintain a case temperature at or below the values specified in Table 60. System-level
or component-level thermal enhancements are required to dissipate the generated heat
if the case temperature exceeds the maximum temperatures listed in Table 60.
Good system airflow is critical to dissipate the highest possible thermal power. The size
and number of fans, vents, and/or ducts, and, their placement in relation to
components and airflow channels within the system determine airflow. Acoustic noise
constraints might limit the size and types of fans, vents and ducts that can be used in a
particular design.
To develop a reliable, cost-effective thermal solution, all of the system variables must
be considered. Use system-level thermal characteristics and simulations to account for
individual component thermal requirements.
Table 60.
82583V Preliminary Thermal Absolute Maximum Rating
Parameter
Maximum
1
Tcase
109 °C
1. Tcase is defined as the maximum case temperature without any thermal enhancement to the package.
12.7.1
12.7.2
Case Temperature
The 82583V is designed to operate properly as long as the Tcase is not exceeded.
Section 12.12 describes the proper guidelines for measuring case temperature.
Designing for Thermal Performance
Section 12.14 describes the PCB and system design recommendations required to
achieve the required 82583V thermal performance.
356
Thermal Design Considerations—82583V GbE Controller
12.8
Thermal Attributes
12.8.1
Typical System Definitions
The following system example is used to generate thermal characteristics data. Note
that the evaluation board is a four-layer 30 x 56 mm mPCIe board.
• All data is preliminary and is not validated against physical samples. Specific
system designs might be significantly different.
• A larger board size with more than four copper layers might increase the 82583V
thermal performance.
Figure 69.
82583V Test Setup
Note:
The mPCIe board is connected to the bottom side of the system board.
357
82583V GbE Controller—Thermal Design Considerations
12.9
82583V Package Thermal Characteristics
Table 61.
Expected Tcase (°C) at TDP
Airflow (LFM)
0
100
101
91
200
99
89
84
79
69
59
49
14
300
400
97
87
82
77
67
57
47
12
85
103
93
88
83
73
63
53
18
98
88
83
78
68
58
48
13
75
70
65
55
45
35
0
86
Ambient
Temperature
(°C)
81
71
61
51
16
Figure 70.
Maximum Allowable Ambient Temperature vs. Air Flow
12.10
Reliability
Each PCAand system combination varies in attach strength and long-term adhesive
performance. Carefully evaluate the reliability of the completed assembly prior to high-
volume use. Some reliability recommendations are listed in Table 62.
358
Thermal Design Considerations—82583V GbE Controller
Table 62.
Reliability Validation
1
2
Test
Requirement
Pass/Fail Criteria
Mechanical shock
Random Vibration
50 G, board level 11 ms, 2 shocks/axis Visual and electrical check
7.3 G, board level 45 minutes/axis, 50
Visual and electrical check
to 2000 Hz
+85 °C 2000 hours total
Checkpoints occur at 168, 500, 1000,
and 2000 hours
High-temperature
life
Visual and mechanical check
Per-target environment (for example,
0 °C to +85 °C) 500 cycles
Thermal cycling
Humidity
Visual and mechanical check
Visual and mechanical check
85% relative humidity 85 °C, 1000
hours
1. Performed the above tests on a sample size of at least 12 assemblies from three lots of
material (total = 36 assemblies).
2. Additional pass/fail criteria can be added as necessary.
12.11
12.12
Measurements for Thermal Specifications
Determining the thermal properties of the system requires careful case temperature
measurements. Guidelines for measuring 82583V case temperature are provided in
Section 12.12.
Case Temperature Measurements
Maintain 82583V Tcase at or below the maximum case temperatures listed in Table 60
to ensure functionality and reliability. Special care is required when measuring the case
temperature to ensure an accurate temperature measurement. Use the following
guidelines when making case measurements:
• Measure the surface temperature of the case in the geometric center of the case
top.
• Calibrate the thermocouples used to measure Tcase before making temperature
measurements.
• Use 36-gauge (maximum) K-type thermocouples.
Care must be taken to avoid introducing errors into the measurements when
measuring a surface temperature that is a different temperature from the surrounding
local ambient air. Measurement errors might be due to a poor thermal contact between
the thermocouple junction and the surface of the package, heat loss by radiation,
convection, conduction through thermocouple leads, and/or contact between the
thermocouple cement and the heat-sink base (if used).
359
82583V GbE Controller—Thermal Design Considerations
12.12.1
Attaching the Thermocouple
The following approach is recommended to minimize measurement errors for attaching
the thermocouple to the case.
• Use 36 gauge or smaller diameter K type thermocouples.
• Ensure that the thermocouple has been properly calibrated.
• Attach the thermocouple bead or junction to the top surface of the package (case)
in the center of the package using high thermal conductivity cements.
Note:
It is critical that the entire thermocouple lead be butted tightly to the top of the
package.
• Attach the thermocouple at a 0° angle if there is no interference with the
thermocouple attach location or leads (Figure 71). This is the preferred method and
is recommended for use with non-enhanced packages.
Figure 71.
Technique for Measuring Tcase with a 0° Angle Attachment
12.13
Conclusion
Increasingly complex systems require better power dissipation. Care must be taken to
ensure that the additional power is properly dissipated. Heat can be dissipated using
improved system cooling, selective use of ducting or any combination.
The simplest and most cost effective method is to improve the inherent system cooling
characteristics through careful design and placement of fans, vents, and ducts. When
additional cooling is required, thermal enhancements may be implemented in
conjunction with enhanced system cooling. The size of the fan can be varied to balance
size and space constraints with acoustic noise.
This section has presented the conditions and requirements to properly design a
cooling solution for systems implementing the 82583V. Properly designed solutions
provide adequate cooling to maintain the 82583V case temperature at or below those
listed in Table 60. Ideally, this is accomplished by providing a low local ambient
temperature and creating a minimal thermal resistance to that local ambient
temperature.
By maintaining the 82583V case temperature at or below those recommended in this
section, the 82583V will function properly and reliably.
Use this section to understand the 82583V thermal characteristics and compare them
to your system environment. Measure the 82583V case temperatures to determine the
best thermal solution for your design.
360
Thermal Design Considerations—82583V GbE Controller
12.14
PCB Guidelines
The following general PCB design guidelines are recommended to maximize the thermal
performance of QFN packages:
1. When connecting ground (thermal) vias-to the ground planes, do not use thermal-
relief patterns.
2. Thermal-relief patterns are designed to limit heat transfer between the vias and the
copper planes, thus constricting the heat flow path from the component to the
ground planes in the PCB.
3. As board temperature also has an effect on the thermal performance of the
package, avoid placing 82583V adjacent to high power dissipation devices.
4. If airflow exists, locate the components in the mainstream of the airflow path for
maximum thermal performance. Avoid placing the components downstream,
behind larger devices or devices that obstruct the air flow or supply excessively
heated air.
Note:
The previously mentioned guidelines are not all inclusive and are defined to give
known, good design practices to maximize the thermal performance of the
components.
361
82583V GbE Controller—Diagnostics
13.0 Diagnostics
To assist in test and debug of the software device driver, a set of software-usable
features have been provided in the component. These features include controls for
specific test-mode usage, as well as some registers for verifying the 82583V’s internal
state against what the software device driver is expecting.
13.1
Introduction
The 82583V provides software visibility (and controllability) into certain major internal
data structures, including all of the transmit and receive FIFO space. However,
interlocks are not provided for any operations, so diagnostic accesses can only be
performed under very controlled circumstances.
The 82583V also provides software-controllable support for certain loopback modes, to
enable a software device driver to test transmit and receive flows to itself. Loopback
modes can also be used to diagnose communication problems and attempt to isolate
the location of a break in the communications path.
13.2
13.3
FIFO Pointer Accessibility
The 82583V’s internal pointers into its transmit and receive data FIFOs are visible
through the head and tail diagnostic data FIFO registers. See section 9.2.9. Diagnostics
software can read these FIFO pointers to confirm an expected hardware state following
a sequence of operation(s). Diagnostic software can further write to these pointers as a
partial-step to verify expected FIFO contents following a specific operation, or to
subsequently write data directly to the data FIFOs.
FIFO Data Accessibility
The 82583V’s internal transmit and receive data FIFOs contents are directly readable
and writeable through the PBM register. The specific locations read or written are
determined by the values of the FIFO pointers, which can be read and written. When
accessing the actual FIFO data structures, locations must be accessed as 32-bit words.
See section 9.2.9.
362
Diagnostics—82583V GbE Controller
13.4
Loopback Operations
Loopback operations are supported by the 82583V to assist with system and device
debug. Loopback operation can be used to test transmit and receive aspects of
software device drivers, as well as to verify electrical integrity of the connections
between the 82583V and the system (such as, PCIe bus connections, etc.). Loopback
operation is supported as follows:
Note:
Configuration for loopback operation varies depending on the link configuration being
used.
• MAC Loopback while operating with the internal PHY
• Loopback – To configure for loopback operation, the RCTL.LBM should remain
configured as for normal operation (set=00b). The PHY must be programmed,
using MDIO accesses to its MII management registers, to perform loopback within
the PHY.
Note:
Note:
All loopback modes are only allowed when the 82583V is configured for full-duplex
operation.
MAC loopback is not functional when the MAC is configured to work at 10 Mb/s.
363
82583V GbE Controller—Board Layout and Schematic Checklists
14.0
Board Layout and Schematic Checklists
Table 63.
Board Layout Checklist
Section
Check Item
Remarks
Obtain the most recent documentation
and specification updates.
Documents are subject to frequent change.
General
Route the transmit and receive differential
traces before routing the digital traces.
Layout of differential traces is critical.
With closer spacing, fields can follow the surface of the
magnetics module or wrap past edge of the board. As a result,
EMI might increase. The optimum location is approximately
one inch behind the magnetics module.
Place the 82583V at least one inch from
the edge of the board.
Placement of
the 82583V
Keep trace length under four inches from the 82583V through
the magnetics to the RJ-45 connector. Signal attenuation can
cause problems for traces longer than four inches. However,
due to near field EMI, the 82583V should be placed at least one
inch away from the magnetics module.
Place the 82583V at least one inch from
the integrated magnetics module but less
than four inches.
Place the AC coupling capacitors on the
PCI Express* (PCIe*) Tx traces as close as
possible to the 82583V but not further
than 250 mils.
Size 0402, X7R is recommended. The AC coupling capacitors
should be placed near the transmitter for PCIe.
Place the AC coupling capacitors on the
PCIe Rx traces as close as possible to the
upstream PCIe device but not further than
250 mils.
Size 0402, X7R is recommended. The AC coupling capacitors
should be placed near the transmitter for PCIe.
PCIe
Interface
Make sure the trace impedance for the
PCIe differential pairs is 100 Ω +/- 20%.
These traces should be routed differentially.
Match trace lengths within each PCIe pair
on a segment-by-segment basis. Match
trace lengths within a pair to five mils.
Place crystal within 0.75 inches of the
82583V.
This reduces EMI.
This reduces EMI.
Place the crystal load capacitors within
0.09 inches of the crystal.
Clock Source
(Crystal
Option)
Keep clock lines away from other digital
traces (especially reset signals), I/O ports,
board edge, transformers and differential
pairs.
364
Board Layout and Schematic Checklists—82583V GbE Controller
Section
Check Item
Remarks
Ensure the oscillator has a it's own local
power supply decoupling capacitor.
If the oscillator is shared or is more than
two inches away from the 82583V, a back-
termination resistor should be placed near
the oscillator for each 82583V.
This enables tuning to ensure that reflections do not distort the
clock waveform.
Clock Source
(Oscillator
Option)
Keep clock lines away from other digital
traces (especially reset signals), I/O ports,
board edge, transformers and differential
pairs.
This reduces EMI.
EEPROM or
Flash
Memory
The NVM can be placed a few inches away
from the 82583V to provide better spacing
of critical components.
Primary requirement for 10/100/1000 Mb/s Ethernet. Paired
50 Ω traces do not make 100 Ω differential. An impedance
calculator can be used to verify this.
Design traces for 100 Ω differential
impedance (± 20%).
Avoid highly resistive traces (for example,
avoid four mil traces longer than four
inches).
If trace length is a problem, use thicker board dielectrics to
allow wider traces. Thicker copper is even better than wider
traces.
If a LAN switch is used or the trace length
from the 82583V is greater than four
inches. It might be necessary to boost the
voltage at the center tap with a separate
power supply to optimize MDI
Consider using a second 82583V instead of a LAN switch and
long MDI traces. It is difficult to achieve excellent performance
with long traces and analog LAN switches. Additional
optimization effort is required to tune the system, the center
tap voltage, and magnetics modules.
performance.
Pairs should be matched at pads, vias and turns. Asymmetry
contributes to impedance mismatch.
Make traces symmetrical.
Do not make 90° bends.
Bevel corners with turns based on 45° angles
If vias are used, the budget is two per trace.
Avoid through holes (vias).
Keep traces close together inside a
differential pair.
Traces should be kept within 10 mils regardless of trace
geometry.
10/100/
1000Base-T
Interface
Traces
Keep trace-to-trace length difference
within each pair to less than 50 mils.
This minimizes signal skew and common mode noise.
Improves long cable performance.
Pair-to-pair trace length does not have to
be matched as differences are not critical.
The difference between the length of longest pair and the
length of the shortest pair should be kept below two inches.
Keep differential pairs more than seven
times the dielectric thickness away from
each other and other traces, including
NVM traces and parallel digital traces.
This minimizes crosstalk and noise injection. Tighter spacing is
allowed for the first 200 mils of trace near of the components.
Ensure that line side MDI traces and line
side termination are at least 80 mils from
all other traces.
This is to ensure the system can survive a high voltage on the
MDI cable. (Hi-POT)
Keep traces at least 0.1 inches away from
the board edge.
This reduces EMI.
Do not have stubs along the traces.
Stubs cause discontinuities that impact return loss.
Digital signals on adjacent layers must
cross at 90° angles. Splits in power and
ground planes must not cross.
Differential pairs should be run on different layers as needed to
improve routing.
365
82583V GbE Controller—Board Layout and Schematic Checklists
Section
Check Item
Remarks
Capacitors connected to center taps
should be placed very close (less than 0.1
inch recommended) to the integrated
magnetics module.
This improves Bit Error Rate (BER).
10/100/
1000Base-T
Interface
Magnetics
Module
The center tap voltage is critical to performance of MDI
interface. Any voltage drop can cause violations to the
specification. Some designs that have a resistive path to the
MDI transformer may require addition regulators to boost the
voltage to above 1.9 V dc at the transformer center tap.
The system side center tap on the
transformer should be connected to the
1.9 V dc power supply through a plane.
Provide a separate chassis ground “island”
to ground the shroud of the RJ-45
connector and if needed to terminate the
line side of the magnetics module. This
design improves EMI behavior.
The split in ground plane should be at least 50 mils. For
discrete magnetics modules, the split should run under center
of magnetics module. Differential pairs never cross the split.
10/100/
1000Base-T
Interface
Chassis
The Bob Smith termination and the MDI traces should be >=
80 mils away from all components and traces on the same
layer. Ensure there is at least 10 mils of single ply woven epoxy
(FR-4) between the chassis ground and any other nodes. Since
there can be small air pockets between woven fibers, it better
to use thicker, two ply, or three ply epoxy (FR-4) to provide
high voltage isolation.
Ensure there is a gap to provide high
voltage isolation to line side of the MDI
traces and the Bob Smith termination.
Ground
Place 4-6 pairs of pads for stitching
capacitors to bridge the gap from chassis
ground to signal ground.
Determine exact number and values empirically based on EMI
performance.
When using the internal regulator control
circuits of the 82583V with external PNP
transistors, keep the trace length from the
CTRL10 and CTRL19 output balls to the
transistors very short (less one inch) and
use 50 mil (minimum) wide traces.
A low inductive loop should be kept from the regulator control
pin, through the PNP transistor, and back to the chip from the
transistor's collector output. The power pins should connect to
the collector of the transistor through a power plane to reduce
the inductive path. This reduces oscillation and ripple in the
power supply.
Narrow finger-like planes and very wide traces are allowed. If
traces are used, 100 mils is the minimum.
Use planes if possible.
The 1.05 V dc and 1.9 V dc regulating
circuits require 1/2 inch x 1/2 inch thermal
relief pads for each PNP.
Power
The pads should be placed on the top layer, under the PNP.
Supply and
Signal
Ground
The 3.3 V dc rail should have at least 25
μF of capacitance.
The 1.05 V dc and 1.9 V dc rails should
have 20-40 μF of capacitance.
Place these to minimize the inductance
from each power pin to the nearest
decoupling capacitor.
Place decoupling and bulk capacitors close to 82583V, with
some along every side, using short, wide traces and large vias.
If power is distributed on traces, bulk capacitors should be
used at both ends. If power is distributed on cards, bulk
capacitors should be used at the connector.
If using decoupling capacitors on LED
lines, place them carefully.
Capacitors on LED lines should be placed near the LEDs.
Keep LED traces away from sources of
noise, for example, high speed digital
traces running in parallel.
LED traces can carry noise into integrated magnetics modules,
RJ-45 connectors, or out to the edge of the board, increasing
EMI.
LED Circuits
366
Board Layout and Schematic Checklists—82583V GbE Controller
Table 64.
Schematic Checklist
Section
Check Items
Remarks
Obtain the most recent documentation and
specification updates.
Documents are subject to frequent change.
General
Observe instructions for special pins needing
pull-up or pull-down resistors.
Connect PCIe interface pins to corresponding
pins on an upstream PCIe device.
Place AC coupling capacitors (0.1 μF) near the
PCIe transmitter.
Size 0402, X7R is recommended.
Connect PECLKn and PECLKp to 100 MHz PCIe
system clock.
PCIe Interface
This is required by the PCIe interface.
This is required for proper device initialization.
Connect PE_RST_N to PLTRST# on an
upstream PCIe device.
Connect PE_WAKE_N to PE_WAKE# on an
upstream PCIe device.
This is required to enable Wake on LAN functionality
required for advanced power management.
Connect to a super I/O pin that retains its value during
PCIe reset, is driven from the resume well and defaults
to one on power-up.
Connect pin 28 DEV_OFF_N to
SUPER_IO_GP_DISABLE# or a pull-up with a
1 KΩ resistor.
If device off functionality is not needed, then
DEV_OFF_N should be connected with an external pull-
up resistor. Ensure pull-ups are connected to aux
power.
Pull-down pin 48, RSET, with a 4.99 KΩ 1%
resistor.
This is required by the PCIe and MDI interfaces.
Support Pins
This pin impacts operation if the 82583V advertises D3
cold wakeup support on the PCIe bus.
Pull-up pin 39, AUX_PWR, with a 10 KΩ
resistor if the power supplies are derived from
always on auxiliary power rails.
Ensure pull-ups are connected to auxiliary power.
Required for normal operation.
Pull-down reserved pins RSVD2_PD,
RSVD3_PD, and RSVD7_PD with 10 KΩ
resistors.
Pull-up reserved pins RSVD5_PU, RSVD6_PU,
RSVD8_PU, RSVD9_PU, RSVD34_PU,
RSVD35_PU, and RSVD36_PU with 10 KΩ
resistors.
Required for normal operation.
Pull-down pin 29, TEST_EN, with a 1 KΩ
resistor.
This is required to prevent the device from going into
test mode during normal operation.
This pin must be driven high during the XOR test.
367
82583V GbE Controller—Board Layout and Schematic Checklists
Section
Check Items
Remarks
The oscillator needs to maintain 50 ppm under all
applicable temperature and voltage conditions. Avoid
PLL clock buffers. Clock buffers introduce additional
jitter. Broadband peak-to-peak jitter must be less than
200 ps.
Use 25 MHz 50 ppm oscillator.
Clock Source
(Oscillator
Option)
Use a local decoupling capacitor on the
oscillator power supply.
The signal from the oscillator must be AC
coupled into the 82583V.
The 82583V has internal circuitry to set the input
common mode voltage.
The clock signal going into the 82583V should
have an amplitude between 1.2 V dc and
1.9 V dc.
This can be achieved with a resistive divider network.
Parallel resonant crystals are required. The Cload
should be 18 pF. Specify Equivalent Series Resistance
(ESR) to be 50 Ω or less.
Use 25 MHz 30 ppm accuracy @ 25 °C crystal.
Avoid components that introduce jitter.
Capacitance affects accuracy of the frequency. Must be
matched to crystal specifications, including estimated
trace capacitance in calculation.
Clock Source
(Crystal Option)
Connect two load capacitors to crystal; one on
XTAL1 and one on XTAL2. Use 27 pF
capacitors as a starting point, but be prepared
to change the value based on testing.
Use capacitors with low ESR (types C0G or NPO, for
example). Refer to the design considerations section of
the datasheet and the Intel Ethernet Controllers Timing
Device Selection Guide for more information.
Use 0.1 μF decoupling capacitor.
Applies to EEPROM or Flash devices.
If SPI Flash is used, connect pin 38 (NVMT) to
ground through a 1 KΩ resistor. If an SPI
EEPROM is used, connect pin 38 (NVMT) to 3.3
V dc through a 1 KΩ resistor.
Ensure pull-ups are connected to auxiliary power.
NVM
The NVM must be powered from auxiliary
power.
The NVM is read when the system is powered on even
before main power is available.
Pins on the 82583V are connected to same named pins
on the NVM. (NVM_SI connects to SI on NVM.
NVM_SO connects to SO on NVM.)
Check connections to NVM_CS_N, NVM_SK,
NVM_SI, NVM_SO.
Primary requirement for 10/100/1000 Mb/s Ethernet.
Paired 50 Ω traces do not make 100 Ω differential. An
impedance calculator can be used to verify this.
Design traces for 100 Ω differential impedance
(± 20%)
If trace length is a problem, use thicker board
dielectrics to allow wider traces. Thicker copper is even
better than wider traces.
Avoid highly resistive traces (for example,
avoid four mil traces longer than four inches)
10/100/
1000Base-T
Interface
The boosted center tap voltage is between 1.9 V dc and
2.65 V dc and consume up to 200 mA.
If a LAN switch is used or the trace length
from the 82583V is greater than four inches.
It might be necessary to boost the voltage at
the center tap with a separate power supply to
optimize MDI performance.
Traces
Consider using a second 82583V instead of a LAN
switch and long MDI traces. It is difficult to achieve
excellent performance with long traces and analog LAN
switches. An optimization effort is required to tune the
system, the center tap voltage, and magnetics
modules.
368
Board Layout and Schematic Checklists—82583V GbE Controller
Section
Check Items
Remarks
Qualify magnetic modules carefully for return
loss, insertion loss, open circuit inductance,
common mode rejection, and crosstalk
isolation.
A magnetics module is critical to passing IEEE PHY
conformance tests and EMI test.
Supply 1.9 V dc to the transformer center taps
and use 0.01 μF bypass capacitors. If a LAN
switch is used or the trace length from the
82583V is greater than four inches, it might
be necessary to boost the voltage at the
center tap with a separate external power
supply to optimize MDI performance.
10/100/1000
Base-T Interface
Magnetic Module
(Integrated
Option)
1.9 V dc at the center tap biases the 82583V's output
buffers. Capacitors with low ESR should be used.
Ensure there are no termination resistors in
the path between the 82583V and the
magnetic module.
The 82583V has an internal termination network.
Bob Smith termination: use 4 x 75 Ω resistors
connected to each cable-side center tap.
Terminate pair-to-pair common mode impedance of the
CAT5 cable.
Bob Smith termination: use an EFT capacitor
attached to the chassis ground. Suggested
values are 1500 pF/2 KV or 1000 pF/3 KV.
These capacitors provide high voltage isolation.
Supply 1.9 V dc to the system side
transformer center taps and use 0.01 μF
bypass capacitors. If a LAN switch is used or
the trace length from the 82583V is greater
than four inches. It might be necessary to
boost the voltage at the center tap with a
separate power supply to optimize MDI
performance.
10/100/
1000Base-T
Interface
Magnetics Module
(Discrete Option
with
1.9 V dc at the center tap biases the 82583V's output
buffers. Capacitors with low ESR should be used.
The Bob Smith termination and the MDI traces should
be >= 80 mils away from all components and traces on
the same layer.
RJ-45 Connector)
Ensure there is high voltage isolation to line
side of the MDI traces and the Bob Smith
termination.
Do not use less than 10 mils of single ply woven epoxy
(FR-4). There can be small air pockets between woven
fibers. Use thicker, two ply, or three ply epoxy (FR-4).
Ensure there are no termination resistors in
the path between the 82583V and the
magnetics.
The 82583V has an internal termination network.
This design improves EMI behavior.
Provide a separate chassis ground to connect
the shroud of the RJ-45 connector and to
terminate the line side of the magnetic
module.
10/100/
1000Base-T
Interface
Typical values range from 0.1 μF to 4.7 μF. The correct
value should be determined experimentally to improve
EMI. Past experiments have shown they are not
required in some designs.
Place pads for approximately 4-6 stitching
capacitors to bridge the gap from chassis
ground to signal ground.
Chassis Ground
369
82583V GbE Controller—Board Layout and Schematic Checklists
Section
Check Items
Remarks
Provide a 3.3 V dc supply. Use an auxiliary
power supply.
Auxiliary power is necessary to support wake up from
power down states.
Connect external PNP transistor's base to
CTRL19 and the emitter to the 3.3 V dc
supply. The collector supplies 1.9 V dc. The
connections and transistor parameters are
critical.
Connect external PNP transistor's base to
CTRL10 and the emitter to the 3.3 V dc
supply. The collector supplies 1.05 V dc. The
connections and transistor parameters are
critical. For option B only.
Connect a 5 KΩ resistor from CTRL19 to the
3.3 V dc supply.
Integrated Power
Supply
Connect a 5 KΩ resistor from CTRL10 to the
3.3 V dc supply. For option B only.
(Option A and B)
For option A: Connect DIS_REG10 to ground.
Enable internal 1.05 V dc regulator if it is used.
For option B: Connect DIS_REG10 to the
3.3 V dc supply.
Ensure that there is at least 10 μF of
capacitance at the emitters of the PNPs.
Place decoupling and bulk capacitors close to 82583V,
with some along every side, using short, wide traces
and large vias. If power is distributed on traces, bulk
capacitors should be used at both ends. If power is
distributed on cards, bulk capacitors should be used at
the connector.
The 3.3 V dc rail should have at least 25 μF of
capacitance.
The 1.05 V dc and 1.9 V dc rails should have
20-40 μF of capacitance.
Place these to minimize the inductance from
each power pin to the nearest decoupling
capacitor.
370
Board Layout and Schematic Checklists—82583V GbE Controller
Section
Check Items
Remarks
Derive all three power supplies from auxiliary
power supplies.
Auxiliary power is necessary to support wake up from
power down states.
If the 1.05 V dc and 1.9 V dc rails are
externally supplied, ensure that CTRL10 and
CTRL19 are tied to ground through a 3.3 KΩ
resistor. Alternatively, they could be left
floating.
Pull-down resistors do not need to be exactly 3.3 KΩ;
however, they must be greater than 1 KΩ.
Connect DIS_REG10 to the 3.3 V dc supply
with a 1 KΩ resister.
Disable internal 1.05 V dc regulator.
It is recommended that the 1.9 V dc supply be
tunable with a resistor option.
Tuning the 1.9 V dc supply might be required to
optimize MDI performance.
External Power
supply
The 3.3 V dc rail should have at least 25 μF of
capacitance.
Place decoupling and bulk capacitors close to 82583V,
with some along every side, using short, wide traces
and large vias. If power is distributed on traces, bulk
capacitors should be used at both ends. If power is
distributed on cards, bulk capacitors should be used at
the connector.
(Option C)
The 1.05 V dc and 1.9 V dc rails should have
at least 20 μF of capacitance.
Place these to minimize the inductance from
each power pin to the nearest decoupling
capacitor.
The 82583V has a power on reset circuit that requires a
1-100 ms ramp time. The rise must be montonic to so
the power on reset triggers only once.
All voltages should ramp to within their control
bands in 100 ms or less. Voltages must ramp
in sequence (3.3 V dc ramps first, 1.9 V dc
ramps second, 1.05 V dc ramps last). The
voltage rise must be monotonic. The minimum
rise time on the 3.3 V dc power is 1 ms.
The sequence is required protect the ESD diodes
connected to the power supplies from being forward
biased
Provide a 3.3 V dc and 1.9 V dc supply. Derive
power supplies from auxiliary power supplies.
Auxiliary power is necessary to support wake up from
power down states.
Ensure that CTRL10 and CTRL19 are tied to
ground through a 3.3 KΩ resistor.
Alternatively, they could be left floating.
Pull-down resistors do not need to be exactly 3.3 KΩ;
however, they must be greater than 1 KΩ.
Connect DIS_REG10 to ground.
Enable internal 1.05 V dc regulator.
The 3.3 V dc rail should have at least 25 μF of
Place decoupling and bulk capacitors close to 82583V,
with some along every side, using short, wide traces
and large vias. If power is distributed on traces, bulk
capacitors should be used at both ends. If power is
distributed on cards, bulk capacitors should be used at
the connector.
Integrated Power
Supply
(Option D)
capacitance.
The 1.05 V dc and 1.9 V dc rails should have
20- 40 μF of capacitance.
Place these to minimize the inductance from
each power pin to the nearest decoupling
capacitor.
371
82583V GbE Controller—Board Layout and Schematic Checklists
Section
Check Items
Remarks
Two LED configurations are compatible with integrated
magnetic modules. For the Link/Activity LED, connect
the cathode to the LED1 pin and the anode to VCC. For
the bi-color speed LED pair, have the LED2 signal drive
one end. The other end should be connected to LED0.
When LED2 is low, the orange LED is lit. When LED0 is
low, the green LED is lit.
Basic recommendation is a single green LED
for activity and a dual (bi-color) LED for
speed. Many other configurations are possible.
LEDs are configurable through the NVM.
LED Circuits
Use 3.3 V dc AUX for designs supporting wake-up.
Consider adding one or two filtering capacitors per LED
for extremely noisy situations. Suggested starting
value is 470 pF.
Connect LEDs to 3.3 V dc as indicated in
reference schematics.
Typical current limiting resistors are 250 Ω to 330 Ω
when using a 3.3 V dc supply. Current limiting resistors
are sometimes included with integrated magnetic
modules.
Add current limiting resistors to LED paths.
Because of pin sharing the 82583V cannot be used in a
JTAG chain. The JTAG pins must be individually driven
and sampled.
The 82583V allows a JTAG Test Access Port to
enable an XOR tree test.
Mfg Test
372
Board Layout and Schematic Checklists—82583V GbE Controller
Note:
This page intentionally left blank.
373
82583V GbE Controller—Models
15.0
Models
Contact your Intel Representative for access to the 82583V IBIS and HSPICE models.
374
相关型号：
8259010U1000
RG-58A/U type, 20 AWG stranded (19x33) .035 tinned copper conductor, polyethylene insulation
ETC
8259010U500
RG-58A/U type, 20 AWG stranded (19x33) .035 tinned copper conductor, polyethylene insulation
ETC
©2020 ICPDF网 联系我们和版权申明