This document provides a functional description of the Ethernet Media Access Controller (EMAC) and
physical layer (PHY) device Management Data Input/Output (MDIO) module integrated in the
TMS320DM36x Digital Media System-on-Chip (DMSoC). Included are the features of the EMAC and
MDIO modules, a discussion of their architecture and operation, how these modules connect to the
outside world, and the registers description for each module.
Notational Conventions
This document uses the following conventions.
•Hexadecimal numbers are shown with the suffix h. For example, the following number is 40
hexadecimal (decimal 64): 40h.
•Registers in this document are shown in figures and described in tables.
– Each register figure shows a rectangle divided into fields that represent the fields of the register.
Each field is labeled with its bit name, its beginning and ending bit numbers above, and its
read/write properties below. A legend explains the notation used for the properties.
– Reserved bits in a register figure designate a bit that is used for future device expansion.
Preface
SPRUFI5B–March 2009–Revised December 2010
Read This First
Related Documentation From Texas Instruments
The following documents describe the TMS320DM36x Digital Media System-on-Chip (DMSoC). Copies of
these documents are available on the internet at www.ti.com.
SPRUFG5 — TMS320DM365 Digital Media System-on-Chip (DMSoC) ARM Subsystem Reference
Guide This document describes the ARM Subsystem in the TMS320DM36x Digital Media
System-on-Chip (DMSoC). The ARM subsystem is designed to give the ARM926EJ-S (ARM9)
master control of the device. In general, the ARM is responsible for configuration and control of the
device; including the components of the ARM Subsystem, the peripherals, and the external
memories.
SPRUFG8 — TMS320DM36x Digital Media System-on-Chip (DMSoC) Video Processing Front End
(VPFE) Users Guide This document describes the Video Processing Front End (VPFE) in the
TMS320DM36x Digital Media System-on-Chip (DMSoC).
SPRUFG9 — TMS320DM36x Digital Media System-on-Chip (DMSoC) Video Processing Back End
(VPBE) Users Guide This document describes the Video Processing Back End (VPBE) in the
TMS320DM36x Digital Media System-on-Chip (DMSoC).
SPRUFH0 — TMS320DM36x Digital Media System-on-Chip (DMSoC) 64-bit Timer Users Guide This
document describes the operation of the software-programmable 64-bit timers in the
TMS320DM36x Digital Media System-on-Chip (DMSoC).
SPRUFH1 — TMS320DM36x Digital Media System-on-Chip (DMSoC) Serial Peripheral Interface
(SPI) Users Guide This document describes the serial peripheral interface (SPI) in the
TMS320DM36x Digital Media System-on-Chip (DMSoC). The SPI is a high-speed synchronous
serial input/output port that allows a serial bit stream of programmed length (1 to 16 bits) to be
shifted into and out of the device at a programmed bit-transfer rate. The SPI is normally used for
communication between the DMSoC and external peripherals. Typical applications include an
interface to external I/O or peripheral expansion via devices such as shift registers, display drivers,
Ethernet Media Access Controller (EMAC)/Management
Data Input/Output (MDIO)
1Introduction
This document provides a functional description of the Ethernet Media Access Controller (EMAC) and
physical layer (PHY) device Management Data Input/Output (MDIO) module integrated in the device.
Included are the features of the EMAC and MDIO modules, a discussion of their architecture and
operation, how these modules connect to the outside world, and a description of the registers for each
module.
The EMAC controls the flow of packet data from the system to the PHY. The MDIO module controls PHY
configuration and status monitoring.
Both the EMAC and the MDIO modules interface to the system core through a custom interface that
allows efficient data transmission and reception. This custom interface is referred to as the EMAC control
module and is considered integral to the EMAC/MDIO peripheral.
1.1Purpose of the Peripheral
The EMAC module is used to move data between theDM36x DMSoC and another host connected to the
same network, in compliance with the Ethernet protocol. The EMAC is controlled by the ARM CPU of the
device; control by the DSP CPU is not supported.
1.2Features
The EMAC/MDIO has the following features:
•Synchronous 10/100 Mbps operation
•MII interface to the physical layer device (PHY)
•EMAC acts as DMA master to either internal or external device memory space
•Hardware error handling including CRC
•Eight receive channels with VLAN tag discrimination for receive quality-of-service (QOS) support
•Eight transmit channels with round-robin or fixed priority for transmit quality-of-service (QOS) support
•Ether-Stats and 802.3-Stats RMON statistics gathering
•Transmit CRC generation selectable on a per channel basis
•Broadcast frames selection for reception on a single channel
•Multicast frames selection for reception on a single channel
•Promiscuous receive mode frames selection for reception on a single channel (all frames, all good
frames, short frames, error frames)
•Hardware flow control
•8K-byte local EMAC descriptor memory that allows the peripheral to operate on descriptors without
affecting the CPU. The descriptor memory holds enough information to transfer up to 512 Ethernet
packets without CPU intervention.
•Programmable interrupt logic permits the software driver to restrict the generation of back-to-back
interrupts, which allows more work to be performed in a single call to the interrupt service routine.
•TI Adaptive Performance Optimization for improved half duplex performance
Figure 1 shows the three main functional modules of the EMAC/MDIO peripheral:
•EMAC control module
•EMAC module
•MDIO module
The EMAC control module is the main interface between the device core processor and the EMAC
module and MDIO module. The EMAC control module contains the necessary components to allow the
EMAC to make efficient use of device memory, plus it controls device interrupts. The EMAC control
module incorporates 8K-byte internal RAM to hold EMAC buffer descriptors.
The MDIO module implements the 802.3 serial management interface to interrogate and control up to 32
Ethernet PHYs connected to the device, using a shared two-wire bus. Host software uses the MDIO
module to configure the autonegotiation parameters of each PHY attached to the EMAC, retrieve the
negotiation results, and configure required parameters in the EMAC module for correct operation. The
module is designed to allow almost transparent operation of the MDIO interface, with very little
maintenance from the core processor.
The EMAC module provides an efficient interface between the processor and the networked community.
The EMAC on this device supports 10 Mbits/second and 100 Mbits/second in either half-duplex or
full-duplex mode, with hardware flow control and quality-of-service (QOS) support.
www.ti.com
14
Figure 1. EMAC and MDIO Block Diagram
Figure 1 also shows the main interface between the EMAC control module and the CPU. The following
connections are made to the device core:
•The peripheral bus connection from the EMAC control module allows the EMAC module to read and
write both internal and external memory through the DMA memory transfer controller.
•The EMAC control module, EMAC, and MDIO all have control registers. These registers are
memory-mapped into device memory space via the device configuration bus. Along with these
registers, the control module’s internal RAM is mapped into this same range.
•The EMAC and MDIO interrupts are combined into a single interrupt within the control module. The
interrupt from the control module then goes to the ARM interrupt controller.
Ethernet Media Access Controller (EMAC)/Management Data Input/OutputSPRUFI5B–March 2009–Revised December 2010
(MDIO)
The EMAC and MDIO interrupts are combined within the control module, so only the control module
interrupt needs to be monitored by the application software or device driver. The EMAC control module
combines the EMAC and MDIO interrupts and generates 4 separate interrupts to the ARM through the
ARM interrupt controller. See Section 2.17.4 for details of interrupt multiplex logic of the EMAC control
module.
1.4Industry Standard(s) Compliance Statement
The EMAC peripheral conforms to the IEEE 802.3 standard, describing the Carrier Sense Multiple Access
with Collision Detection (CSMA/CD) Access Method and Physical Layer specifications. The IEEE 802.3
standard has also been adopted by ISO/IEC and re-designated as ISO/IEC 8802-3:2000(E).
In difference from this standard, the EMAC peripheral does not use the Transmit Coding Error signal
MTXER. Instead of driving the error pin when an underflow condition occurs on a transmitted frame, the
EMAC intentionally generates an incorrect checksum by inverting the frame CRC, so that the transmitted
frame is detected as an error by the network.
2Architecture
This section discusses the architecture and basic function of the EMAC/MDIO module.
2.1Clock Control
The frequencies for the transmit and receive clocks are fixed by the IEEE 802.3 specification as:
•2.5 MHz at 10 Mbps
•25 MHz at 100 Mbps
Architecture
All EMAC logic is clocked synchronously with the PLL peripheral clock. The MDIO clock can be controlled
through the application software, by programming the divide-down factor in the MDIO control register
(CONTROL).
2.1.1MII Clocking
The transmit and receive clock sources are provided from an external PHY via the EMAC_TX_CLK and
EMAC_RX_CLK pins. These clocks are inputs to the EMAC module and operate at 2.5 MHz in 10 Mbps
mode and at 25 MHz in 100 Mbps mode. For timing purposes, data is transmitted and received with
reference to EMAC_TX_CLK and EMAC_RX_CLK, respectively.
SPRUFI5B–March 2009–Revised December 2010Ethernet Media Access Controller (EMAC)/Management Data Input/Output
The EMAC peripheral includes internal memory that is used to hold information about the Ethernet
packets received and transmitted. This internal RAM is 2K × 32 bits in size. Data can be written to and
read from the EMAC internal memory by either the EMAC or the CPU. It is used to store buffer descriptors
that are 4-words (16-bytes) deep. This 8K local memory holds enough information to transfer up to 512
Ethernet packets without CPU intervention.
The packet buffer descriptors can also be placed in the internal processor memory (L2), or in EMIF
memory (DDR). There are some tradeoffs in terms of cache performance and throughput when
descriptors are placed in the system memory, versus when they are placed in the EMAC’s internal
memory. Cache performance is improved when the buffer descriptors are placed in internal memory.
However, the EMAC throughput is better when the descriptors are placed in the local EMAC RAM.
2.3Signal Descriptions
The DM36x DMSoC supports the MII interface (for 10/100 Mbps) operation.
Figure 2 shows a device with integrated EMAC and MDIO interfaced via a MII connection. The EMAC
module does not include a transmit error (MTXER) pin. In the case of transmit error, CRC inversion is
used to negate the validity of the transmitted frame.
The individual EMAC and MDIO signals for the MII interface are summarized in Table 1. For more
information, refer to either the IEEE 802.3 standard or ISO/IEC 8802-3:2000(E).
www.ti.com
Figure 2. Ethernet Configuration MII Connections
16
Ethernet Media Access Controller (EMAC)/Management Data Input/OutputSPRUFI5B–March 2009–Revised December 2010
(MDIO)
EMAC_TX_CLKITransmit clock (EMAC_TX_CLK). The transmit clock is a continuous clock that provides the timing
reference for transmit operations. The EMAC_TXD and EMAC_TX_EN signals are tied to this clock.
The clock is generated by the PHY and is 2.5 MHz at 10 Mbps operation and 25 MHz at 100 Mbps
operation.
EMAC_TXD[3-0]OTransmit data (EMAC_TXD). The transmit data pins are a collection of 4 data signals comprising
4 bits of data. MTDX0 is the least-significant bit (LSB). The signals are synchronized by
EMAC_TX_CLK and valid only when EMAC_TX_EN is asserted.
EMAC_TX_ENOTransmit enable (EMAC_TX_EN). The transmit enable signal indicates that the EMAC_TXD pins are
generating nibble data for use by the PHY. It is driven synchronously to EMAC_TX_CLK.
EMAC_COLICollision detected (EMAC_COL). The EMAC_COL pin is asserted by the PHY when it detects a
collision on the network. It remains asserted while the collision condition persists. This signal is not
necessarily synchronous to EMAC_TX_CLK nor EMAC_RX_CLK. This pin is used in half-duplex
operation only.
EMAC_CRSICarrier sense (EMAC_CRS). The EMAC_CRS pin is asserted by the PHY when the network is not
idle in either transmit or receive. The pin is deasserted when both transmit and receive are idle. This
signal is not necessarily synchronous to EMAC_TX_CLK nor EMAC_RX_CLK. This pin is used in
half-duplex operation only.
EMAC_RX_CLKIReceive clock (EMAC_RX_CLK). The receive clock is a continuous clock that provides the timing
reference for receive operations. The EMAC_RXD, EMAC_RX_DV, and MRXER signals are tied to
this clock. The clock is generated by the PHY and is 2.5 MHz at 10 Mbps operation and 25 MHz at
100 Mbps operation.
EMAC_RXD[3-0]IReceive data (EMAC_RXD). The receive data pins are a collection of 4 data signals comprising
4 bits of data. MRDX0 is the least-significant bit (LSB). The signals are synchronized by
EMAC_RX_CLK and valid only when EMAC_RX_DV is asserted.
EMAC_RX_DVIReceive data valid (EMAC_RX_DV). The receive data valid signal indicates that the EMAC_RXD
pins are generating nibble data for use by the EMAC. It is driven synchronously to EMAC_RX_CLK.
MRXERIReceive error (MRXER). The receive error signal is asserted for one or more EMAC_RX_CLK
periods to indicate that an error was detected in the received frame. This is meaningful only during
data reception when EMAC_RX_DV is active.
MDCLKOManagement data clock (MDCLK). The MDIO data clock is sourced by the MDIO module on the
system. It is used to synchronize MDIO data access operations done on the MDIO pin. The
frequency of this clock is controlled by the CLKDIV bits in the MDIO control register (CONTROL).
MDIOI/OManagement data input output (MDIO). The MDIO pin drives PHY management data into and out of
the PHY by way of an access frame consisting of start of frame, read/write indication, PHY address,
register address, and data bit cycles. The MDIO pin acts as an output for all but the data bit cycles
at which time it is an input for read operations.
2.4Pin Multiplexing
On the DM36x processor, the EMAC pins are multiplexed with other pin functions. For these pins to be
used as EMAC functions, the pin multiplexing registers must be configured appropriately. For specific
information on pin multiplexing, refer to the device-specific data manual and the TMS320DM365 DigitalMedia System-on-Chip (DMSoiC) ARM Subsystem Users Guide (SPRUFG5).
SPRUFI5B–March 2009–Revised December 2010Ethernet Media Access Controller (EMAC)/Management Data Input/Output
Ethernet provides an unreliable, connection-less service to a networking application. A brief overview of
the Ethernet protocol is given in the following subsections. For in-depth information on the Carrier Sense
Multiple Access with Collision Detection (CSMA/CD) Access Method, which is the Ethernet’s multiple
access protocol, see the IEEE 802.3 standard document.
2.5.1Ethernet Frame Format
All the Ethernet technologies use the same frame structure. The format of an Ethernet frame is shown in
Figure 3 and described in Table 2. The Ethernet packet, which is the collection of bytes representing the
data portion of a single Ethernet frame on the wire, is shown outlined in bold. The Ethernet frames are of
variable lengths, with no frame smaller than 64 bytes or larger than RXMAXLEN bytes (header, data, and
CRC).
www.ti.com
Figure 3. Ethernet Frame Format
Table 2. Ethernet Frame Description
FieldBytesDescription
Preamble7Preamble. These 7 bytes have a fixed value of 55h and serve to wake up the receiving
SFD1Start of Frame Delimiter. This field with a value of 5Dh immediately follows the preamble
Destination6Destination address. This field contains the Ethernet MAC address of the EMAC port for
Source6Source address. This field contains the MAC address of the Ethernet port that transmits the
Len2Length/Type field. The length field indicates the number of EMAC client data bytes
Data46 toData field. This field carries the datagram containing the upper layer protocol frame, that is,
(RXMAXLEN - 18)IP layer datagram. The maximum transfer unit (MTU) of Ethernet is (RXMAXLEN - 18)
FCS4Frame Check Sequence. A cyclic redundancy check (CRC) is used by the transmit and
EMAC ports and to synchronize their clocks to that of the sender’s clock.
pattern and indicates the start of important data.
which the frame is intended. It may be an individual or multicast (including broadcast)
address. When the destination EMAC port receives an Ethernet frame with a destination
address that does not match any of its MAC physical addresses, and no promiscuous,
multicast or broadcast channel is enabled, it discards the frame.
frame to the Local Area Network.
contained in the subsequent data field of the frame. This field can also be used to identify
the type of data the frame is carrying.
bytes. This means that if the upper layer protocol datagram exceeds (RXMAXLEN - 18)
bytes, then the host has to fragment the datagram and send it in multiple Ethernet packets.
The minimum size of the data field is 46 bytes. This means that if the upper layer datagram
is less then 46 bytes, the data field has to be extended to 46 bytes by appending extra bits
after the data field, but prior to calculating and appending the FCS.
receive algorithms to generate a CRC value for the FCS field. The frame check sequence
covers the 60 to (RXMAXLEN - 4) bytes of the packet data. Note that this 4-byte field may
or may not be included as part of the packet data, depending on how the EMAC is
configured.
18
Ethernet Media Access Controller (EMAC)/Management Data Input/OutputSPRUFI5B–March 2009–Revised December 2010
(MDIO)
Nodes in an Ethernet Local Area Network are interconnected by a broadcast channel, as a result, when
an EMAC port transmits a frame, all the adapters on the local network receive the frame. Carrier Sense
Multiple Access with Collision Detection (CSMA/CD) algorithms are used when the EMAC operates in
half-duplex mode. When operating in full-duplex mode, there is no contention for use of a shared medium,
since there are exactly two ports on the local network.
Each port runs the CSMA/CD protocol without explicit coordination with the other ports on the Ethernet
network. Within a specific port, the CSMA/CD protocol works as follows:
1. The port obtains data from upper layers protocols at its node, prepares an Ethernet frame, and puts
the frame in a buffer.
2. If the port senses that the medium is idle it starts to transmit the frame. If the port senses that the
transmission medium is busy, it waits until it senses no signal energy (plus an Inter-Packet Gap time)
and then starts to transmit the frame.
3. While transmitting, the port monitors for the presence of signal energy coming from other ports. If the
port transmits the entire frame without detecting signal energy from other Ethernet devices, the port is
done with the frame.
4. If the port detects signal energy from other ports while transmitting, it stops transmitting its frame and
instead transmits a 48-bit jam signal.
5. After transmitting the jam signal the port enters an exponential backoff phase. Specifically, when
transmitting a given frame, after experiencing a number of collisions in a row for the frame, the port
chooses a random value that is dependent on the number of collisions. The port then waits an amount
of time that is multiple of this random value, and returns to step 2.
Architecture
2.6Programming Interface
2.6.1Packet Buffer Descriptors
The buffer descriptor is a central part of the EMAC module and is how the application software describes
Ethernet packets to be sent and empty buffers to be filled with incoming packet data. The basic descriptor
format is shown in Figure 4 and described in Table 3.
For example, consider three packets to be transmitted, Packet A is a single fragment (60 bytes), Packet B
is fragmented over three buffers (1514 bytes total), and Packet C is a single fragment (1514 bytes). The
linked list of descriptors to describe these three packets is shown in Figure 5.
0Next DescriptorThe next descriptor pointer is used to create a single linked list of descriptors. Each descriptor
Pointerdescribes a packet or a packet fragment. When a descriptor points to a single buffer packet
or the first fragment of a packet, the start of packet (SOP) flag is set in the flags field. When a
descriptor points to a single buffer packet or the last fragment of a packet, the end of packet
(EOP) flag is set. When a packet is fragmented, each fragment must have its own descriptor
and appear sequentially in the descriptor linked list.
1Buffer PointerThe buffer pointer refers to the actual memory buffer that contains packet data during
transmit operations, or is an empty buffer ready to receive packet data during receive
operations.
2Buffer OffsetThe buffer offset is the offset from the start of the packet buffer to the first byte of valid data.
This field only has meaning when the buffer descriptor points to a buffer that actually contains
data.
Buffer LengthThe buffer length is the actual number of valid packet data bytes stored in the buffer. If the
buffer is empty and waiting to receive data, this field represents the size of the empty buffer.
3FlagsThe flags field contains more information about the buffer, such as, is it the first fragment in a
packet (SOP), the last fragment in a packet (EOP), or contains an entire contiguous Ethernet
packet (both SOP and EOP). The flags are described in Section 2.6.4 and Section 2.6.5.
Packet LengthThe packet length only has meaning for buffers that both contain data and are the start of a
new packet (SOP). In the case of SOP descriptors, the packet length field contains the length
of the entire Ethernet packet, regardless if it is contained in a single buffer or fragmented over
several buffers.
Figure 5. Typical Descriptor Linked List
20
Ethernet Media Access Controller (EMAC)/Management Data Input/OutputSPRUFI5B–March 2009–Revised December 2010
(MDIO)
The EMAC module processes descriptors in linked list chains as discussed in Section 2.6.1. The lists
controlled by the EMAC are maintained by the application software through the use of the head descriptor
pointer registers (HDP). Since the EMAC supports eight channels for both transmit and receive, there are
eight head descriptor pointer registers for both. They are:
•TXnHDP - Transmit Channel n DMA Head Descriptor Pointer Register
•RXnHDP - Receive Channel n DMA Head Descriptor Pointer Register
After an EMAC reset and before enabling the EMAC for send or receive, all 16 head descriptor pointer
registers must be initialized to 0.
The EMAC uses a simple system to determine if a descriptor is currently owned by the EMAC or by the
application software. There is a flag in the buffer descriptor flags called OWNER. When this flag is set, the
packet that is referenced by the descriptor is considered to be owned by the EMAC. Note that ownership
is done on a packet based granularity, not on descriptor granularity, so only SOP descriptors make use of
the OWNER flag. As packets are processed, the EMAC patches the SOP descriptor of the corresponding
packet and clears the OWNER flag. This is an indication that the EMAC has finished processing all
descriptors up to and including the first with the EOP flag set, indicating the end of the packet (note this
may only be one descriptor with both the SOP and EOP flags set).
To add a descriptor or a linked list of descriptors to an EMAC descriptor queue for the first time, the
software application simply writes the pointer to the descriptor or first descriptor of a list to the
corresponding HDP register. Note that the last descriptor in the list must have its “next” pointer cleared to
0. This is the only way the EMAC has of detecting the end of the list. So in the case where only a single
descriptor is added, its “next descriptor” pointer must be initialized to 0.
The HDP must never be written to a second time while a previous list is active. To add additional
descriptors to a descriptor list already owned by the EMAC, the NULL “next” pointer of the last descriptor
of the previous list is patched with a pointer to the first descriptor in the new list. The list of new
descriptors to be appended to the existing list must itself be NULL terminated before the pointer patch is
performed.
There is a potential race condition where the EMAC may read the “next” pointer of a descriptor as NULL in
the instant before an application appends additional descriptors to the list by patching the pointer. This
case is handled by the software application always examining the buffer descriptor flags of all EOP
packets, looking for a special flag called end of queue (EOQ). The EOQ flag is set by the EMAC on the
last descriptor of a packet when the descriptor’s “next” pointer is NULL. This is the way the EMAC
indicates to the software application that it believes it has reached the end of the list. When the software
application sees the EOQ flag set, and there are more descriptors to process, the application may at that
time submit the new list, or the portion of the appended list that was missed, by writing the new list pointer
to the same HDP that started the process.
This process applies when adding packets to a transmit list, and empty buffers to a receive list.
Architecture
SPRUFI5B–March 2009–Revised December 2010Ethernet Media Access Controller (EMAC)/Management Data Input/Output
The EMAC processes descriptors in linked list chains as discussed in Section 2.6.1, using the linked list
queue mechanism discussed in Section 2.6.2.
The EMAC synchronizes descriptor list processing through the use of interrupts to the software
application. The interrupts are controlled by the application using the interrupt masks, global interrupt
enable, and the completion pointer register (CP). The CP is also called the interrupt acknowledge register.
As the EMAC supports eight channels for both transmit and receive, there are eight completion pointer
registers for both. They are:
•TXnCP - Transmit Channel n Completion Pointer (Interrupt Acknowledge) Register
•RXnCP - Receive Channel n Completion Pointer (Interrupt Acknowledge) Register
These registers serve two purposes. When read, they return the pointer to the last descriptor that the
EMAC has processed. When written by the software application, the value represents the last descriptor
processed by the software application. When these two values do not match, the interrupt remains
asserted, after the respective End of interrupt bit is signaled in the EMAC control module.
The system configuration determines whether or not an active interrupt actually interrupts the CPU. In
general, the individual interrupts for different events from the EMAC and MDIO must be enabled in the
EMAC control module, and it also must be mapped in the ARM interrupt controller and enabled as a CPU
interrupt. If the system is configured properly, the interrupt for a specific receive or transmit channel
executes under the previously described conditions when the corresponding interrupt is enabled in the
EMAC using the receive interrupt mask set register (RXINTMASKSET) or the transmit interrupt mask set
register (TXINTMASKSET).
Whether or not the interrupt is enabled, the current state of the receive or transmit channel interrupt can
be examined directly by the software application reading the receive interrupt status (unmasked) register
(RXINTSTATRAW) and transmit interrupt status (unmasked) register (TXINTSTATRAW).
Interrupts are acknowledged when the application software updates the value of TXnCP or RXnCP with a
value that matches the internal value kept by the EMAC. This mechanism ensures that the application
software never misses an EMAC interrupt, since the interrupt and its acknowledgment are tied directly to
the actual buffer descriptors processing.
www.ti.com
22
Ethernet Media Access Controller (EMAC)/Management Data Input/OutputSPRUFI5B–March 2009–Revised December 2010
(MDIO)
A transmit (TX) buffer descriptor (Figure 6) is a contiguous block of four 32-bit data words aligned on a
32-bit boundary that describes a packet or a packet fragment. Example 1 shows the transmit buffer
descriptor described by a C structure.
Figure 6. Transmit Buffer Descriptor Format
Word 0
310
Next Descriptor Pointer
Word 1
310
Buffer Pointer
Word 2
3116 150
Buffer OffsetBuffer Length
Word 3
3130292827262516
SOPEOPOWNEREOQTDOWNCMPLT PASSCRCReserved
150
Packet Length
Example 1. Transmit Buffer Descriptor in C Structure Format
/*
// EMAC Descriptor
//
// The following is the format of a single buffer descriptor
// on the EMAC.
*/
typedef struct _EMAC_Desc {
struct _EMAC_Desc *pNext;/* Pointer to next descriptor in chain */
Uint8*pBuffer;/* Pointer to data buffer*/
Uint32BufOffLen; /* Buffer Offset(MSW) and Length(LSW)*/
Uint32PktFlgLen; /* Packet Flags(MSW) and Length(LSW)*/
The next descriptor pointer points to the 32-bit word aligned memory address of the next buffer descriptor
in the transmit queue. This pointer is used to create a linked list of buffer descriptors. If the value of this
pointer is zero, then the current buffer is the last buffer in the queue. The software application must set
this value prior to adding the descriptor to the active transmit list. This pointer is not altered by the EMAC.
The value of pNext should never be altered once the descriptor is in an active transmit queue, unless its
current value is NULL. If the pNext pointer is initially NULL, and more packets need to be queued for
transmit, the software application may alter this pointer to point to a newly appended descriptor. The
EMAC will use the new pointer value and proceed to the next descriptor unless the pNext value has
already been read. In this latter case, the transmitter will halt on the transmit channel in question, and the
software application may restart it at that time. The software can detect this case by checking for an end
of queue (EOQ) condition flag on the updated packet descriptor when it is returned by the EMAC.
2.6.4.2Buffer Pointer
The buffer pointer is the byte-aligned memory address of the memory buffer associated with the buffer
descriptor. The software application must set this value prior to adding the descriptor to the active transmit
list. This pointer is not altered by the EMAC.
2.6.4.3Buffer Offset
This 16-bit field indicates how many unused bytes are at the start of the buffer. For example, a value of
0000h indicates that no unused bytes are at the start of the buffer and that valid data begins on the first
byte of the buffer, while a value of 000Fh indicates that the first 15 bytes of the buffer are to be ignored by
the EMAC and that valid buffer data starts on byte 16 of the buffer. The software application must set this
value prior to adding the descriptor to the active transmit list. This field is not altered by the EMAC.
Note that this value is only checked on the first descriptor of a given packet (where the start of packet
(SOP) flag is set). It can not be used to specify the offset of subsequent packet fragments. Also, since the
buffer pointer may point to any byte–aligned address, this field may be entirely superfluous, depending on
the device driver architecture.
The range of legal values for this field is 0 to (Buffer Length – 1).
www.ti.com
2.6.4.4Buffer Length
This 16-bit field indicates how many valid data bytes are in the buffer. On single fragment packets, this
value is also the total length of the packet data to be transmitted. If the buffer offset field is used, the offset
bytes are not counted as part of this length. This length counts only valid data bytes. The software
application must set this value prior to adding the descriptor to the active transmit list. This field is not
altered by the EMAC.
2.6.4.5Packet Length
This 16-bit field specifies the number of data bytes in the entire packet. Any leading buffer offset bytes are
not included. The sum of the buffer length fields of each of the packet’s fragments (if more than one) must
be equal to the packet length. The software application must set this value prior to adding the descriptor to
the active transmit list. This field is not altered by the EMAC. This value is only checked on the first
descriptor of a given packet (where the start of packet (SOP) flag is set).
2.6.4.6Start of Packet (SOP) Flag
When set, this flag indicates that the descriptor points to a packet buffer that is the start of a new packet.
In the case of a single fragment packet, both the SOP and end of packet (EOP) flags are set. Otherwise,
the descriptor pointing to the last packet buffer for the packet sets the EOP flag. This bit is set by the
software application and is not altered by the EMAC.
24
Ethernet Media Access Controller (EMAC)/Management Data Input/OutputSPRUFI5B–March 2009–Revised December 2010
(MDIO)
When set, this flag indicates that the descriptor points to a packet buffer that is last for a given packet. In
the case of a single fragment packet, both the start of packet (SOP) and EOP flags are set. Otherwise, the
descriptor pointing to the last packet buffer for the packet sets the EOP flag. This bit is set by the software
application and is not altered by the EMAC.
2.6.4.8Ownership (OWNER) Flag
When set, this flag indicates that all the descriptors for the given packet (from SOP to EOP) are currently
owned by the EMAC. This flag is set by the software application on the SOP packet descriptor before
adding the descriptor to the transmit descriptor queue. For a single fragment packet, the SOP, EOP, and
OWNER flags are all set. The OWNER flag is cleared by the EMAC once it is finished with all the
descriptors for the given packet. Note that this flag is valid on SOP descriptors only.
2.6.4.9End of Queue (EOQ) Flag
When set, this flag indicates that the descriptor in question was the last descriptor in the transmit queue
for a given transmit channel, and that the transmitter has halted. This flag is initially cleared by the
software application prior to adding the descriptor to the transmit queue. This bit is set by the EMAC when
the EMAC identifies that a descriptor is the last for a given packet (the EOP flag is set), and there are no
more descriptors in the transmit list (next descriptor pointer is NULL).
The software application can use this bit to detect when the EMAC transmitter for the corresponding
channel has halted. This is useful when the application appends additional packet descriptors to a transmit
queue list that is already owned by the EMAC. Note that this flag is valid on EOP descriptors only.
Architecture
2.6.4.10Teardown Complete (TDOWNCMPLT) Flag
This flag is used when a transmit queue is being torn down, or aborted, instead of allowing it to be
transmitted. This would happen under device driver reset or shutdown conditions. The EMAC sets this bit
in the SOP descriptor of each packet as it is aborted from transmission.
Note that this flag is valid on SOP descriptors only. Also note that only the first packet in an unsent list has
the TDOWNCMPLT flag set. Subsequent descriptors are not even processed by the EMAC.
2.6.4.11Pass CRC (PASSCRC) Flag
This flag is set by the software application in the SOP packet descriptor before it adds the descriptor to the
transmit queue. Setting this bit indicates to the EMAC that the 4 byte Ethernet CRC is already present in
the packet data, and that the EMAC should not generate its own version of the CRC.
When the CRC flag is cleared, the EMAC generates and appends the 4-byte CRC. The buffer length and
packet length fields do not include the CRC bytes. When the CRC flag is set, the 4-byte CRC is supplied
by the software application and is already appended to the end of the packet data. The buffer length and
packet length fields include the CRC bytes, as they are part of the valid packet data. Note that this flag is
valid on SOP descriptors only.
SPRUFI5B–March 2009–Revised December 2010Ethernet Media Access Controller (EMAC)/Management Data Input/Output
A receive (RX) buffer descriptor (Figure 7) is a contiguous block of four 32-bit data words aligned on a
32-bit boundary that describes a packet or a packet fragment. Example 2 shows the receive buffer
descriptor described by a C structure.
2.6.5.1Next Descriptor Pointer
This pointer points to the 32–bit word aligned memory address of the next buffer descriptor in the receive
queue. This pointer is used to create a linked list of buffer descriptors. If the value of this pointer is zero,
then the current buffer is the last buffer in the queue. The software application must set this value prior to
adding the descriptor to the active receive list. This pointer is not altered by the EMAC.
The value of pNext should never be altered once the descriptor is in an active receive queue, unless its
current value is NULL. If the pNext pointer is initially NULL, and more empty buffers can be added to the
pool, the software application may alter this pointer to point to a newly appended descriptor. The EMAC
will use the new pointer value and proceed to the next descriptor unless the pNext value has already been
read. In this latter case, the receiver will halt the receive channel in question, and the software application
may restart it at that time. The software can detect this case by checking for an end of queue (EOQ)
condition flag on the updated packet descriptor when it is returned by the EMAC.
2.6.5.2Buffer Pointer
The buffer pointer is the byte-aligned memory address of the memory buffer associated with the buffer
descriptor. The software application must set this value prior to adding the descriptor to the active receive
list. This pointer is not altered by the EMAC.
Example 2. Receive Buffer Descriptor in C Structure Format
/*
// EMAC Descriptor
//
// The following is the format of a single buffer descriptor
// on the EMAC.
*/
typedef struct _EMAC_Desc {
struct _EMAC_Desc *pNext;/* Pointer to next descriptor in chain */
Uint8*pBuffer;/* Pointer to data buffer*/
Uint32BufOffLen; /* Buffer Offset(MSW) and Length(LSW)*/
Uint32PktFlgLen; /* Packet Flags(MSW) and Length(LSW)*/
This 16-bit field must be initialized to zero by the software application before adding the descriptor to a
receive queue.
Whether or not this field is updated depends on the setting of the RXBUFFEROFFSET register. When the
offset register is set to a non-zero value, the received packet is written to the packet buffer at an offset
given by the value of the register, and this value is also written to the buffer offset field of the descriptor.
When a packet is fragmented over multiple buffers because it does not fit in the first buffer supplied, the
buffer offset only applies to the first buffer in the list, which is where the start of packet (SOP) flag is set in
the corresponding buffer descriptor. In other words, the buffer offset field is only updated by the EMAC on
SOP descriptors.
The range of legal values for the BUFFEROFFSET register is 0 to (Buffer Length – 1) for the smallest
value of buffer length for all descriptors in the list.
SPRUFI5B–March 2009–Revised December 2010Ethernet Media Access Controller (EMAC)/Management Data Input/Output
•Before the descriptor is first placed on the receive queue by the application software, the buffer length
field is first initialized by the software to have the physical size of the empty data buffer pointed to by
the buffer pointer field.
•After the empty buffer has been processed by the EMAC and filled with received data bytes, the buffer
length field is updated by the EMAC to reflect the actual number of valid data bytes written to the
buffer.
2.6.5.5Packet Length
This 16-bit field specifies the number of data bytes in the entire packet. This value is initialized to zero by
the software application for empty packet buffers. The value is filled in by the EMAC on the first buffer
used for a given packet. This is signified by the EMAC setting a start of packet (SOP) flag. The packet
length is set by the EMAC on all SOP buffer descriptors.
2.6.5.6Start of Packet (SOP) Flag
When set, this flag indicates that the descriptor points to a packet buffer that is the start of a new packet.
In the case of a single fragment packet, both the SOP and end of packet (EOP) flags are set. Otherwise,
the descriptor pointing to the last packet buffer for the packet has the EOP flag set. This flag is initially
cleared by the software application before adding the descriptor to the receive queue. This bit is set by the
EMAC on SOP descriptors.
www.ti.com
2.6.5.7End of Packet (EOP) Flag
When set, this flag indicates that the descriptor points to a packet buffer that is last for a given packet. In
the case of a single fragment packet, both the start of packet (SOP) and EOP flags are set. Otherwise, the
descriptor pointing to the last packet buffer for the packet has the EOP flag set. This flag is initially cleared
by the software application before adding the descriptor to the receive queue. This bit is set by the EMAC
on EOP descriptors.
2.6.5.8Ownership (OWNER) Flag
When set, this flag indicates that the descriptor is currently owned by the EMAC. This flag is set by the
software application before adding the descriptor to the receive descriptor queue. This flag is cleared by
the EMAC once it is finished with a given set of descriptors, associated with a received packet. The flag is
updated by the EMAC on SOP descriptor only. So when the application identifies that the OWNER flag is
cleared on an SOP descriptor, it may assume that all descriptors up to and including the first with the EOP
flag set have been released by the EMAC. (Note that in the case of single buffer packets, the same
descriptor will have both the SOP and EOP flags set.)
2.6.5.9End of Queue (EOQ) Flag
When set, this flag indicates that the descriptor in question was the last descriptor in the receive queue for
a given receive channel, and that the corresponding receiver channel has halted. This flag is initially
cleared by the software application prior to adding the descriptor to the receive queue. This bit is set by
the EMAC when the EMAC identifies that a descriptor is the last for a given packet received (also sets the
EOP flag), and there are no more descriptors in the receive list (next descriptor pointer is NULL).
The software application can use this bit to detect when the EMAC receiver for the corresponding channel
has halted. This is useful when the application appends additional free buffer descriptors to an active
receive queue. Note that this flag is valid on EOP descriptors only.
2.6.5.10Teardown Complete (TDOWNCMPLT) Flag
This flag is used when a receive queue is being torn down, or aborted, instead of being filled with received
data. This would happen under device driver reset or shutdown conditions. The EMAC sets this bit in the
descriptor of the first free buffer when the tear down occurs. No additional queue processing is performed.
28
Ethernet Media Access Controller (EMAC)/Management Data Input/OutputSPRUFI5B–March 2009–Revised December 2010
(MDIO)
This flag is set by the EMAC in the SOP buffer descriptor if the received packet includes the 4-byte CRC.
This flag should be cleared by the software application before submitting the descriptor to the receive
queue.
2.6.5.12Jabber Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet is a jabber frame and was
not discarded because the RXCEFEN bit was set in the RXMBPENABLE. Jabber frames are frames that
exceed the RXMAXLEN in length, and have CRC, code, or alignment errors.
2.6.5.13Oversize Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet is an oversized frame and
was not discarded because the RXCEFEN bit was set in the RXMBPENABLE.
2.6.5.14Fragment Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet is only a packet fragment
and was not discarded because the RXCEFEN bit was set in the RXMBPENABLE.
2.6.5.15Undersized Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet is undersized and was
not discarded because the RXCSFEN bit was set in the RXMBPENABLE.
Architecture
2.6.5.16Control Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet is an EMAC control frame
and was not discarded because the RXCMFEN bit was set in the RXMBPENABLE.
2.6.5.17Overrun Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet was aborted due to a
receive overrun.
2.6.5.18Code Error (CODEERROR) Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet contained a code error
and was not discarded because the RXCEFEN bit was set in the RXMBPENABLE.
2.6.5.19Alignment Error (ALIGNERROR) Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet contained an alignment
error and was not discarded because the RXCEFEN bit was set in the RXMBPENABLE.
2.6.5.20CRC Error (CRCERROR) Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet contained a CRC error
and was not discarded because the RXCEFEN bit was set in the RXMBPENABLE.
2.6.5.21No Match (NOMATCH) Flag
This flag is set by the EMAC in the SOP buffer descriptor, if the received packet did not pass any of the
EMAC’s address match criteria and was not discarded because the RXCAFEN bit was set in the
RXMBPENABLE. Although the packet is a valid Ethernet data packet, it was only received because the
EMAC is in promiscuous mode.
SPRUFI5B–March 2009–Revised December 2010Ethernet Media Access Controller (EMAC)/Management Data Input/Output
The basic functions of the EMAC control module (Figure 8) are to interface the EMAC and MDIO modules
to the rest of the system, and to provide for a local memory space to hold EMAC packet buffer descriptors.
Local memory is used to help avoid contention to device memory spaces. Other functions include the bus
arbiter, and interrupt control and pacing logic control.
www.ti.com
Figure 8. EMAC Control Module Block Diagram
2.7.1Internal Memory
2.7.2Bus Arbiter
The EMAC control module includes 8K bytes of internal memory. The internal memory block is essential
for allowing the EMAC to operate more independently of the CPU. It also prevents memory underflow
conditions when the EMAC issues read or write requests to descriptor memory. (Memory accesses to
read or write the actual Ethernet packet data are protected by the EMAC's internal FIFOs).
A descriptor is a 16-byte memory structure that holds information about a single Ethernet packet buffer,
which may contain a full or partial Ethernet packet. Thus with the 8K memory block provided for descriptor
storage, the EMAC module can send and received up to a combined 512 packets before it needs to be
serviced by application or driver software.
The EMAC control module bus arbiter operates transparently to the rest of the system. It is used:
•To arbitrate between the CPU and EMAC buses for access to internal descriptor memory.
•To arbitrate between internal EMAC buses for access to system memory.
30
Ethernet Media Access Controller (EMAC)/Management Data Input/OutputSPRUFI5B–March 2009–Revised December 2010
(MDIO)