DMA With Bridge (XDMA)

DMA/Bridge Subsystem for
PCI Express v4.1
Product Guide
Vivado Design Suite
PG195 (v4.1) November 24, 2023
AMD Adaptive Computing is creating an environment where

employees, customers, and partners feel welcome and included. To
that end, we’re removing non-inclusive language from our products
and related collateral. We’ve launched an internal initiative to remove
language that could exclude people or reinforce historical biases,
including terms embedded in our software and IPs. You may still find
examples of non-inclusive language in our older products as we work
to make these changes and align with evolving industry standards.
Follow this link for more information.
Table of Contents
Chapter 1: Introduction.............................................................................................. 4
Features........................................................................................................................................ 4
IP Facts..........................................................................................................................................5
Chapter 2: Overview......................................................................................................6
Feature Summary........................................................................................................................ 8
Applications..................................................................................................................................8
Unsupported Features................................................................................................................9
Limitations....................................................................................................................................9
Licensing and Ordering............................................................................................................ 10
Chapter 3: Product Specification......................................................................... 12

Standards................................................................................................................................... 12
Performance and Resource Utilization...................................................................................12
Minimum Device Requirements.............................................................................................. 12
Configurable Components of the Subsystem........................................................................13
XDMA Operations......................................................................................................................19
Port Descriptions.......................................................................................................................30
Register Space........................................................................................................................... 43
Chapter 4: Designing with the Subsystem..................................................... 77

Clocking and Resets.................................................................................................................. 77
Tandem Configuration..............................................................................................................78
Chapter 5: Design Flow Steps.................................................................................83

Customizing and Generating the Subsystem........................................................................ 83
Constraining the Subsystem....................................................................................................96
Simulation.................................................................................................................................. 98
Synthesis and Implementation............................................................................................. 101
Chapter 6: Example Design................................................................................... 102

Available Example Designs.................................................................................................... 102
PG195 (v4.1) November 24, 2023

Send Feedback
DMA/Bridge Subsystem for PCIe 2
Customizing and Generating the Example Design............................................................. 112
Chapter 7: Test Bench...............................................................................................114

Root Port Model Test Bench for Endpoint............................................................................114
Appendix A: GT Locations.......................................................................................122
Appendix B: Application Software Development..................................... 123

Device Drivers.......................................................................................................................... 123
Linux Device Driver................................................................................................................. 123
Using the Driver...................................................................................................................... 124
Interrupt Processing............................................................................................................... 124
Example H2C Flow...................................................................................................................125
Example C2H Flow...................................................................................................................126
Appendix C: Upgrading............................................................................................127
New Parameters......................................................................................................................127
New Ports................................................................................................................................. 127
Appendix D: Debugging.......................................................................................... 130

Finding Help with AMD Adaptive Computing Solutions.....................................................130
Debug Tools............................................................................................................................. 131
Hardware Debug..................................................................................................................... 132
Appendix E: Using the Xilinx Virtual Cable to Debug............................. 135

Overview...................................................................................................................................135
Host PC XVC-Server Application............................................................................................ 136
Host PC XVC-over-PCIe Driver............................................................................................... 136
XVC-over-PCIe Enabled FPGA Design................................................................................... 137
Using the PCIe-XVC-VSEC Example Design.......................................................................... 143
Appendix F: Additional Resources and Legal Notices............................ 152

Finding Additional Documentation.......................................................................................152
Support Resources.................................................................................................................. 153
References................................................................................................................................153
Revision History....................................................................................................................... 154
Please Read: Important Legal Notices................................................................................. 158
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 1: Introduction
Chapter 1
Introduction
The AMD DMA/Bridge Subsystem for PCI Express® (PCIe®) implements a high performance,
configurable Scatter Gather DMA for use with the PCI Express® 2.1 and 3.x Integrated Block.
The IP provides a choice between an AXI4 Memory Mapped or AXI4-Stream user interface.
This IP optionally also supports a PCIe AXI Bridge mode which is enabled for only AMD
UltraScale+™ devices. For details about PCIe AXI Bridge mode operation, see AXI Bridge for PCI
Express Gen3 Subsystem Product Guide (PG194).
This document covers DMA mode operation only.
Note: For details about the AMD Versal™ Adaptive SoC subsystem, refer to the Versal Adaptive SoC DMA
and Bridge Subsystem for PCI Express Product Guide (PG344).
Features
• Supports AMD UltraScale+™, AMD UltraScale™, AMD Virtex™ 7 XT Gen3 (Endpoint), and 7
series 2.1 (Endpoint) Integrated Blocks for PCIe. 7A15T and 7A25T are not supported
• Support for 64, 128, 256, 512-bit datapath (64, and 128-bit datapath only for 7 series Gen2
IP)
• 64-bit source, destination, and descriptor addresses
• Up to four host-to-card (H2C/Read) data channels (up to two for 7 series Gen2 IP)
• Up to four card-to-host (C2H/Write) data channels (up to two for 7 series Gen2 IP)
• Selectable user interface
○ Single AXI4 memory mapped (MM) user interface
○ AXI4-Stream user interface (each channel has its own AXI4-Stream interface)
• AXI4 Master and AXI4-Lite Master optional interfaces allow for PCIe traffic to bypass the
DMA engine
• AXI4-Lite Slave to access DMA status registers
• Scatter Gather descriptor list supporting unlimited list size
• 256 MB max transfer size per descriptor
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 1: Introduction
• Legacy, MSI, and MSI-X interrupts

• Block fetches of contiguous descriptors
• Poll Mode
• Descriptor Bypass interface
• Arbitrary source and destination address
• Parity check or Propagate Parity on AXI bus (not available for 7 series Gen2 IP)
IP Facts
AMD LogiCORE™ IP Facts Table
Subsystem Specifics
Supported Device Family1 AMD UltraScale+™, AMD UltraScale™, 7 series Gen2 devices
Supported User Interfaces AXI4 MM, AXI4-Lite, AXI4-Stream
Resources See Resource Utilization web page.
Provided with Subsystem
Design Files Encrypted System Verilog
Example Design Verilog
Test Bench Verilog
Constraints File XDC
Simulation Model Verilog
Supported S/W Driver Linux and Windows Drivers2
Tested Design Flows3
Design Entry AMD Vivado™ Design Suite
Simulation For supported simulators, see the Vivado Design Suite User Guide: Release Notes,
Installation, and Licensing (UG973).
Synthesis Vivado synthesis
Support
Release Notes and Known Issues Master Answer Record: 65443
All Vivado IP Change Logs Master Vivado IP Change Logs: 72775
Support web page
Notes:
1. For a complete list of supported devices, see the AMD Vivado™ IP catalog.
2. For details, see Appendix B: Application Software Development and AR: 65444.
3. For the supported versions of the tools, see the Vivado Design Suite User Guide: Release Notes, Installation, and Licensing
(UG973).
4. For AMD Versal™ Adaptive SoC, refer to Versal Adaptive SoC DMA and Bridge Subsystem for PCI Express Product Guide
(PG344).
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 2: Overview
Chapter 2
Overview
The DMA/Bridge Subsystem for PCI Express® (PCIe®) can be configured to be either a high-
performance direct memory access (DMA) data mover or a bridge between the PCI Express and
AXI memory spaces.
• DMA Data Mover: As a DMA, the core can be configured with either an AXI (memory
mapped) interface or with an AXI streaming interface to allow for direct connection to RTL
logic. Either interface can be used for high performance block data movement between the
PCIe address space and the AXI address space using the provided character driver. In addition
to the basic DMA functionality, the DMA supports up to four upstream and downstream
channels, the ability for PCIe traffic to bypass the DMA engine (Host DMA Bypass), and an
optional descriptor bypass to manage descriptors from the FPGA fabric for applications that
demand the highest performance and lowest latency.
• Bridge Between PCIe and AXI Memory: When configured as a PCIe Bridge, received PCIe
packets are converted to AXI traffic and received AXI traffic is converted to PCIe traffic. The
bridge functionality is ideal for AXI peripherals needing a quick and easy way to access a PCI
Express subsystem. The bridge functionality can be used as either an Endpoint or as a Root
Port. PCIe Bridge functionality is only supported for AMD UltraScale+™ devices. For Bridge
only option and for 7 series non-XT device, you should use AXI Memory Mapped to PCI
Express (PCIe) Gen2. For details see, AXI Memory Mapped to PCI Express (PCIe) Gen2 LogiCORE
IP Product Guide (PG055). For 7 series XT and UltraScale device, you should use AXI Bridge for
PCI Express Gen3. For details, see AXI Bridge for PCI Express Gen3 Subsystem Product Guide
(PG194).
This document covers DMA mode operation only.
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 2: Overview
Figure 1: DMA/Bridge Subsystem for PCI Express® Overview
DMA Subsystem for PCIe
AXI Write
H2C
Interface (MM
Channels
or ST)
RQ/RC
Interface
AXI Read
C2H
Interface (MM
Channels
or ST)
Integrated Block PCIe RX

for PCIe IP (with
User wrapper as
IRQ Module
Logic needed)
Configured PCIe TX
as EndPoint
Cfg Master
(AXI4-Lite
Master)
Cfg Master CQ/CC

(AXI4-Lite Slave) Bridge Interface
Host DMA
Bypass (AXI MM
Master)
X14718-042121
This diagram refers to the Requester Request (RQ)/Requester Completion (RC) interfaces, and
the Completer Request (CQ)/Completer Completion (CC) interfaces. For more information about
these, see the UltraScale+ Devices Integrated Block for PCI Express LogiCORE IP Product Guide
(PG213).
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 2: Overview
Feature Summary
The DMA/Bridge Subsystem for PCI Express® allows for the movement of data between Host
memory and the DMA subsystem. It does this by operating on 'descriptors' that contain
information about the source, destination and amount of data to transfer. These direct memory
transfers can be both in the Host to Card (H2C) and Card to Host (C2H) transfers. The DMA can
be configured to have a single AXI4 Master interface shared by all channels or one AXI4-Stream
interface for each channel enabled. Memory transfers are specified on a per-channel basis in
descriptor linked lists, which the DMA fetches from host memory and processes. Events such as
descriptor completion and errors are signaled using interrupts. The core also provides up to 16
user interrupt wires that generate interrupts to the host.
The host is able to directly access the user logic through two interfaces:
• The AXI4-Lite Master Configuration port: This port is a fixed 32-bit port and is intended for
non-performance-critical access to user configuration and status registers.
• The AXI Memory Mapped Master CQ Bypass port: The width of this port is the same as the
DMA channel datapaths and is intended for high-bandwidth access to user memory that
might be required in applications such as peer-to-peer transfers.
The user logic is able to access the DMA/Bridge Subsystem for PCIe internal configuration and
status registers through an AXI4-Lite Slave Configuration interface. Requests that are mastered
on this interface are not forwarded to PCI Express.
Applications
The core architecture enables a broad range of computing and communications target
applications, emphasizing performance, cost, scalability, feature extensibility, and mission-critical
reliability. Typical applications include:
• Data communications networks

• Telecommunications networks
• Broadband wired and wireless applications
• Network interface cards
• Chip-to-chip and backplane interface cards
• Server add-in cards for various applications
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 2: Overview
Unsupported Features
The following features of the standard are not supported by this core:
• Tandem Configuration solutions (Tandem PROM, Tandem PCIe, Tandem with Field Updates,
PR over PCIe) are not supported for AMD Virtex™ 7 XT and 7 series Gen2 devices
• Tandem Configuration is not yet supported for Bridge mode in UltraScale+ devices.
• SR-IOV
• ECRC
• Example design not supported for all configurations
• Narrow burst (not supported on the master interface)
• BAR translation for DMA addresses to the AXI4 Memory Mapped interface
Limitations
PCIe Transaction Type
The PCIe® transactions that can be generated are those that are compatible with the AXI4
specification. The following table lists the supported PCIe transaction types.
Table 1: Supported PCIe Transaction Types
TX RX
MRd32 MRd32
MRd64 MRd64
MWr32 MWr32
MWr64 MWr64
Msg(INT/Error) Msg(SSPL,INT,Error)
Cpl Cpl
CplD CplD
PCIe Capability
For the DMA/Bridge Subsystem for PCI Express®, only the following PCIe capabilities are
supported due to the AXI4 specification:
• 1 PF
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 2: Overview
• MSI
• MSI-X
• PM
• AER (only PCIe 3.x core)
DMA/Bridge Subsystem for PCI Express® does not support expansion ROM.
Others
• Only supports the INCR burst type. Other types result in a Slave Illegal Burst (SIB) interrupt.
• No memory type support (AxCACHE)
• No protection type support (AxPROT)
• No lock type support (AxLOCK)
• No non-contiguous byte enable support (WSTRB)
• For 7 series Gen2 IP, PCIe access from the Host system must be limited to 1DW (4 Bytes)
transaction only.
Note: Both AXI Bypass and Register access are limited by this restriction.
PCIe to DMA Bypass Master

• Only issues the INCR burst type
• Only issues the data, non-secure, and unprivileged protection type
• For 7 series Gen2 IP, limited to 1DW (4 Bytes) transaction only
User Interrupt in MSI-X Mode

Users need to program a different vector number for each user interrupts in the IRQ Block User
Vector Number register to generate acks for all user interrupts. This generates acks for all user
interrupts when there are simultaneous interrupts. When all vector numbers are pointing to the
same MSI-X entry, there is only one ack.
Licensing and Ordering

This AMD IP module is provided at no additional cost with the AMD Vivado™ Design Suite under
the terms of the End User License. Information about this and other AMD IP modules is available
at the AMD Intellectual Property page. For information about pricing and availability of other
AMD IP modules and tools, contact your local AMD sales representative.
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 2: Overview
For more information, visit the DMA Subsystem for PCI Express product page.
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 3: Product Specification
Chapter 3
Product Specification
The DMA/Bridge Subsystem for PCI Express® in conjunction with the Integrated Block for PCI
Express IP, provides a highly configurable DMA Subsystem for PCIe, and a high performance
DMA solution.
Standards
The DMA/Bridge Subsystem for PCIe is compliant with the following specifications:
• AMBA AXI4-Stream Protocol Specification (ARM IHI 0051A)

• PCI Express Base Specification v2.1 and v3.0
Performance and Resource Utilization

For DMA Perfomance data, see AR 68049.
For DMA Resource Utilization, see Resource Utilization web page.
Minimum Device Requirements

The following table lists the link widths and supported speed for a given speed grade.
Table 2: Minimum Device Requirements
Capability Link Speed Capability Link Width Supported Speed Grades

AMD UltraScale+™ Architecture (PCIE4)
Gen1/Gen2 x1, x2, x4, x8, x16 -1, -1L, -1LV, -2, -2L, -2LV, -31
x1, x2, x4 -1, -1L, -1LV, -2, -2L, -2LV, -31
Gen3 x8 -1, -2, -2L, -2LV, -31
x16 -1, -2, -2L, -31
PG195 (v4.1) November 24, 2023

Send Feedback
Table 2: Minimum Device Requirements (cont'd)
Capability Link Speed Capability Link Width Supported Speed Grades

AMD Virtex™ UltraScale+™ Devices with HBM (PCIE4C)2
Gen1/Gen2 x1, x2, x4, x8, x16 -1, -2, -2L, -2LV, -3
x1, x2, x4 -1, -2, -2L, -2LV, -3
Gen3 x8 -1, -2, -2L, -2LV, -3
x16 -1, -2, -2L, -2LV, -3
Gen46 x1, x2, x4, x8 -2, -2L, -3
AMD UltraScale™ Devices
Gen1 x1, x2, x4, x8 -1, -1L, -1LV, -1H, -1HV, -2, -33
Gen2 x1, x2, x4, x8 -1, -1L, -1LV, -1H, -1HV, -2, -33
Gen3 x1, x2, x4 -1, -1L, -1LV, -1H, -1HV, -2, -33, 4
Gen3 x8 -2, -3
7 series Gen3 Devices
Gen1 x1, x2, x4, x8 -1, -1M, -1I, -2, -2L, -2G, -2I, -3
Gen2 x1, x2, x4, x8 -1, -1M, -1I, -2, -2L, -2G, -2I, -3
Gen3 x1, x2, x4, x8 -2, -2L, -2G, -2I, -3
7 series Gen2 Devices
Gen1 x1, x2, x4, x8 -15, -25, -3
x1, x2, x4 -15, -25, -3
Gen2
x8 -25, -3
Notes:
1. -1L(0.95V), -1LV(0.90V), -2L(0.85V), -2LV(0.72V).
2. AMD Virtex™ UltraScale+™ devices with high bandwidth memory (HBM) contain both PCIE4 and PCIE4C blocks. Only
the PCIE4C blocks support Gen3 x16 in the -2LV speed grade.
3. -1L(0.95V), -1LV(0.90V), -1H(1.0V), -1HV(0.95V).
4. The Core Clock Frequency option must be set to 250 MHz for -1, -1LV, -1L, -1H and -1HV speed grades.
5. Available -1 speed grades are -1M, -1I, -1Q depending on family selected. Available -2 speed grades are -2, -2G, -2I,
-2IL, -2L depending on the family selected.
6. For Gen4 mode restrictions, see UltraScale+ Devices Integrated Block for PCI Express LogiCORE IP Product Guide (PG213).
Configurable Components of the Subsystem

Internally, the subsystem can be configured to implement up to eight independent physical DMA
engines (up to four H2C and four C2H). These DMA engines can be mapped to individual AXI4-
Stream interfaces or a shared AXI4 memory mapped (MM) interface to the user application. On
the AXI4 MM interface, the DMA/Bridge Subsystem for PCI Express® generates requests and
expected completions. The AXI4-Stream interface is data-only.
PG195 (v4.1) November 24, 2023

Send Feedback
The type of channel configured determines the transactions on which bus:
• A Host-to-Card (H2C) channel generates read requests to PCIe and provides the data or
generates a write request to the user application.
• A Card-to-Host (C2H) channel either waits for data on the user side or generates a read
request on the user side and then generates a write request containing the data received to
PCIe.
The DMA/Bridge Subsystem for PCIe also enables the host to access the user logic. Write
requests that reach PCIe to DMA bypass Base Address Register (BAR) are processed by the
DMA. The data from the write request is forwarded to the user application through the NoC
interface to the PL logic.
The host access to the configuration and status registers in the user logic is provided through an
AXI master port. These requests are 32-bit reads or writes. The user application also has access
to internal DMA configuration and status registers through an AXI slave port.
When multiple channels for H2C and C2H are enabled, transactions on the AXI4 Master
interface are interleaved between all selected channels. Simple round robin protocol is used to
service all channels. Transactions granularity depends on host Max Payload Size (MPS), page size,
and other host settings.
Target Bridge
The target bridge receives requests from the host. Based on BARs, the requests are directed to
the internal target user through the AXI4-Lite master, or the CQ bypass port. After the
downstream user logic has returned data for a non-posted request, the target bridge generates a
read completion TLP and sends it to the PCIe IP over the CC bus.
In the following tables, the PCIe BARs selection corresponds to the options set in the PCIe BARs
tab in the IP Configuration GUI.
Table 3: 32-Bit BARs
PCIe BARs Selection

During IP BAR0 (32-bit) BAR1 (32-bit) BAR2 (32-bit)
Customization
Default DMA
PCIe to AXI Lite Master PCIe to AXI4-Lite Master DMA
enabled
PCIe to AXI Lite Master and PCIe to AXI4-Lite Master DMA PCIe to DMA Bypass
PCIe to DMA Bypass
enabled
PCIe to DMA Bypass DMA PCIe to DMA Bypass
enabled
PG195 (v4.1) November 24, 2023

Send Feedback
Table 4: 64-Bit BARs
PCIe BARs Selection

During IP BAR0 (64-bit) BAR2 (64-bit) BAR4 (64-bit)
Customization
Default DMA
PCIe to AXI Lite Master PCIe to AXI4-Lite Master DMA
enabled
PCIe to AXI Lite Master and PCIe to AXI4-Lite Master DMA PCIe to DMA Bypass
PCIe to DMA Bypass
enabled
PCIe to DMA Bypass DMA PCIe to DMA Bypass
enabled
Different combinations of BARs can be selected. The tables above list only 32-bit selections and
64-bit selections for all BARs as an example. You can select different combinations of BARs
based on your requirements.
Related Information
PCIe BARs Tab
H2C Channel
The previous tables represents PCIe to AXI4-Lite Master, DMA, and PCIe to DMA Bypass for 32-
bit and 64-bit BAR selections. Each space can be individually selected for 32-bits or 64-bits BAR.
The number of H2C channels is configured in the AMD Vivado™ Integrated Design Environment
(IDE). The H2C channel handles DMA transfers from the host to the card. It is responsible for
splitting read requests based on maximum read request size, and available internal resources. The
DMA channel maintains a maximum number of outstanding requests based on the RNUM_RIDS,
which is the number of outstanding H2C channel request ID parameter. Each split, if any, of a
read request consumes an additional read request entry. A request is outstanding after the DMA
channel has issued the read to the PCIe RQ block to when it receives confirmation that the write
has completed on the user interface in-order. After a transfer is complete, the DMA channel
issues a writeback or interrupt to inform the host.
The H2C channel also splits transaction on both its read and write interfaces. On the read
interface to the host, transactions are split to meet the maximum read request size configured,
and based on available Data FIFO space. Data FIFO space is allocated at the time of the read
request to ensure space for the read completion. The PCIe RC block returns completion data to
the allocated Data Buffer locations. To minimize latency, upon receipt of any completion data, the
H2C channel begins issuing write requests to the user interface. It also breaks the write requests
into maximum payload size. On an AXI4-Stream user interface, this splitting is transparent.
PG195 (v4.1) November 24, 2023

Send Feedback
When multiple channels are enabled, transactions on the AXI4 Master interface are interleaved
between all selected channels. Simple round robin protocol is used to service all channels.
Transactions granularity depends on host Max Payload Size (MPS), page size, and other host
settings.
C2H Channel
The C2H channel handles DMA transfers from the card to the host. The instantiated number of
C2H channels is controlled in the AMD Vivado™ IDE. Similarly the number of outstanding
transfers is configured through the WNUM_RIDS, which is the number of C2H channel request
IDs. In an AXI4-Stream configuration, the details of the DMA transfer are set up in advance of
receiving data on the AXI4-Stream interface. This is normally accomplished through receiving a
DMA descriptor. After the request ID has been prepared and the channel is enabled, the AXI4-
Stream interface of the channel can receive data and perform the DMA to the host. In an AXI4
MM interface configuration, the request IDs are allocated as the read requests to the AXI4 MM
interface are issued. Similar to the H2C channel, a given request ID is outstanding until the write
request has been completed. In the case of the C2H channel, write request completion is when
the write request has been issued as indicated by the PCIe IP.
When multiple channels are enabled, transactions on the AXI4 Master interface are interleaved
between all selected channels. Simple round robin protocol is used to service all channels.
Transactions granularity depends on host MaxPayload Size (MPS), page size, and other host
settings.
AXI4-Lite Master
This module implements the AXI4-Lite master bus protocol. The host can use this interface to
generate 32-bit read and 32-bit write requests to the user logic. The read or write request is
received over the PCIe to AXI4-Lite master BAR. Read completion data is returned back to the
host through the target bridge over the PCIe IP CC bus.
AXI4-Lite Slave
This module implements the AXI4-Lite slave bus protocol. The user logic can master 32-bit reads
or writes on this interface to DMA internal registers only. You cannot access the PCIe integrated
block register through this interface. This interface does not generate requests to the host.
Host-to-Card Bypass Master

Host requests that reach the PCIe to DMA bypass BAR are sent to this module. The bypass
master port is an AXI4 MM interface and supports read and write accesses.
PG195 (v4.1) November 24, 2023

Send Feedback
IRQ Module
The IRQ module receives a configurable number of interrupt wires from the user logic and one
interrupt wire from each DMA channel. This module is responsible for generating an interrupt
over PCIe. Support for MSI-X, MSI, and legacy interrupts can be specified during IP configuration.
Note: The Host can enable one or more interrupt types from the specified list of supported interrupts
during IP configuration. The IP only generates one interrupt type at a given time even when there are more
than one enabled. MSI-X interrupt takes precedence over MSI interrupt, and MSI interrupt take
precedence over Legacy interrupt. The Host software must not switch (either enable or disable) an
interrupt type while there is an interrupt asserted or pending.
Legacy Interrupts
Asserting one or more bits of usr_irq_req when legacy interrupts are enabled causes the
DMA to issue a legacy interrupt over PCIe. Multiple bits may be asserted simultaneously but
each bit must remain asserted until the corresponding usr_irq_ack bit has been asserted.
After a usr_irq_req bit is asserted, it must remain asserted until both corresponding
usr_irq_ack bit is asserted and the interrupt is serviced and cleared by the Host. This ensures
interrupt pending register within the IP remains asserted when queried by the Host's Interrupt
Service Routine (ISR) to determine the source of interrupts. The usr_irq_ack assertion
indicates the requested interrupt has been sent to the PCIe block. You must implement a
mechanism in the user application to know when the interrupt routine has been serviced. This
detection can be done in many different ways depending on your application and your use of this
interrupt pin. This typically involves a register (or array of registers) implemented in the user
application that is cleared, read, or modified by the Host software when an interrupt is serviced.
After the usr_irq_req bit is deasserted, it cannot be reasserted until the corresponding
usr_irq_ack bit has been asserted for a second time. This indicates the deassertion message
for the legacy interrupt has been sent over PCIe. After a second usr_irq_ack has occurred, the
xdma0_usr_irq_req wire can be reasserted to generate another legacy interrupt.
The xdma0_usr_irq_req bit and DMA interrupts can be mapped to legacy interrupt INTA,
INTB, INTC, and INTD through the configuration registers. The following figure shows the
legacy interrupts.
Note: This figure shows only the handshake between xdma0_usr_irq_req and usr_irq_ack. Your
application might not clear or service the interrupt immediately, in which case, you must keep
usr_irq_req asserted past usr_irq_ack. The figure below shows one possible scenario where
usr_irq_ackis deasserted at the same cycle for both requests[1:0], which might not be the case in other
situations.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 2: Legacy Interrupts
MSI and MSI-X Interrupts

Asserting one or more bits of usr_irq_req causes the generation of an MSI or MSI-X interrupt
if MSI or MSI-X is enabled. If both MSI and MSI-X capabilities are enabled, an MSI-X interrupt is
generated.
After a usr_irq_req bit is asserted, it must remain asserted until the corresponding
usr_irq_ack bit is asserted and the interrupt has been serviced and cleared by the Host. The
usr_irq_ack assertion indicates the requested interrupt has been sent to the PCIe block. This
will ensure the interrupt pending register within the IP remains asserted when queried by the
Host's Interrupt Service Routine (ISR) to determine the source of interrupts. You must implement
a mechanism in the user application to know when the interrupt routine has been serviced. This
detection can be done in many different ways depending on your application and your use of this
interrupt pin. This typically involves a register (or array of registers) implemented in the user
application that is cleared, read, or modified by the Host software when an Interrupt is serviced.
Configuration registers are available to map usr_irq_req and DMA interrupts to MSI or MSI-X
vectors. For MSI-X support, there is also a vector table and PBA table. The following figure shows
the MSI interrupt.
Note: This figure shows only the handshake between usr_irq_req and usr_irq_ack. Your application
might not clear or service the interrupt immediately, in which case, you must keep usr_irq_req asserted
past usr_irq_ack.
Figure 3: MSI Interrupts
The following figure shows the MSI-X interrupt.
Note: This figure shows only the handshake between usr_irq_req and usr_irq_ack. Your application
might not clear or service the interrupt immediately, in which case, you must keep usr_irq_req asserted
past usr_irq_ack.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 4: MSI-X Interrupts
For more details, see Interrupt Processing.
Config Block
The config module, the DMA register space which contains PCIe® solution IP configuration
information and DMA control registers, stores PCIe IP configuration information that is relevant
to the DMA/Bridge Subsystem for PCIe. This configuration information can be read through
register reads to the appropriate register offset within the config module.
XDMA Operations
Quick Start
At the most basic level, the PCIe® DMA engine typically moves data between host memory and
memory that resides in the FPGA which is often (but not always) on an add-in card. When data is
moved from host memory to the FPGA memory, it is called a Host to Card (H2C) transfer or
System to Card (S2C) transfer. Conversely, when data is moved from the FPGA memory to the
host memory, it is called a Card to Host (C2H) or Card to System (C2S) transfer.
These terms help delineate which way data is flowing (as opposed to using read and write which
can get confusing very quickly). The PCIe DMA engine is simply moving data to or from PCIe
address locations.
In a typical operation, an application in the host moves data between the FPGA and host
memory. To accomplish this transfer, the host sets up buffer space in system memory and creates
descriptors that the DMA engine use to move the data.
The contents of the descriptors will depend on a number of factors, including which user
interface is chosen for the DMA engine. If an AXI4-Stream interface is selected, C2H transfers do
not use the source address field and H2C fields do not use the destination address. This is
because the AXI4-Stream interface is a FIFO type interface that does not use addresses.
PG195 (v4.1) November 24, 2023

Send Feedback
If an AXI Memory Mapped interface is selected, then a C2H transfer has the source address as an
AXI address and the destination address is the PCIe address. For a H2C transfer, the source
address is a PCIe address and the destination address is an AXI address.
The following flow charts show typical transfers for both H2C and C2H transfers when the data
interface is selected during IP configuration for an AXI Memory Mapped interface.
Initial Setup For H2C and C2H Transfers

The following figure shows the initial setup for both H2C and C2H transfers.
Figure 5: Setup
Load driver
(setup)
Set ‘H2C Channel interrupt enable mask’ register 0x0090 to generate interrupts
for corresponding bits.
Set ‘C2H Channel interrupt enable mask’ register 0x1090 to generate interrupts
for corresponding bits.
Set ‘IRQ Block Channel Interrupt Enable Mask’ register 0x2010 and enable all
channels (both H2C and C2H) to generate interrupt.
X19438-061319
AXI-MM Transfer For H2C

The following figure shows a basic flow chart that explains the data transfer for H2C. The flow
chart color coding is as follows: Green is the application program; Orange is the driver; and Blue
is the hardware.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 6: DMA H2C Transfer Summary
Application program initiates H2C transfer, with transfer length, buffer location where data is stored.
Driver creates descriptors based on transfer length.
Driver writes first descriptor base address to Address 0x4080 and 0x4084.
Driver writes next adjacent descriptor count to 0x4088 if any.
Driver starts H2C transfer by writing to H2C engines control register,

address 0x0004.
DMA initiates Descriptor fetch request for one or more descriptors

(depending on adjacent descriptor count).
DMA receives one Descriptor or more descriptors (depending on

adjacent descriptor count).
No
Is this the last DMA sends read request to (Host) source address based
descriptor? on first available descriptor.
Yes
Stop fetching the descriptor from

DMA receives data from Host for that descriptor.
host.
Yes
Transmit data on (Card ) AXI-MM Master interface
Is there any more
descriptor left?
Yes
Is there more data
to transfer? No
No
Stop fetching data from Host.
Send interrupt to Host.
Interrupt process.
Read ‘IRQ Block Channel Interrupt Request’ 0x2044 to see which channels sent interrupt.
Mask corresponding channel interrupt writing to 0x2018.
Driver Reads corresponding ‘Status register’ 0x0044 which will also clear the status register.
Read channel ‘Completed descriptor count’ 0x0048 and compare with the number of
descriptor generated.
Write to channel ‘Control register’ 0x0004 to stop the DMA run.

Write to ‘Block channel interrupt Enable Mask’ 0x2014 to enable interrupt for next transfer. Exit application
Return control to the application program with the transfer size. program.
X19389-061319
PG195 (v4.1) November 24, 2023

Send Feedback
AXI-MM Transfer For C2H

The following figure shows a basic flow chart that explains the data transfer for C2H. The flow
chart color coding is as follows: Green is the application program; Orange is the driver; and Blue
is the hardware.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 7: DMA C2H Transfer Summary
Application program initiates C2H transfer, with transfer length,

receive buffer location.
Driver creates descriptors based on transfer length.
Driver writes first descriptor base address to Address 0x5080 and 0x5084.
Driver writes next adjacent descriptor count to 0x5088 if any.
Driver starts C2H transfer by writing to C2H engines

control register, address 0x1004.
DMA initiates Descriptor fetch request for one or more

descriptors (depending on adjacent descriptor count).
DMA receives one Descriptor or more descriptors

(depending on adjacent descriptor count).
No
Is this the DMA reads data from (Card) Source address for
last descriptor? a given descriptor.
Yes
Stop fetching descriptor Yes
from host.
Is there any more
descriptor left?
Transmit data to PCIe to (Host) Destination address. No

Stop fetching data from
Card.
Yes
Is there more
data to transfer?
No
Send interrupt to Host.
Interrupt process.
Read ‘IRQ Block Channel Interrupt Request’ 0x2044 to see which
channels sent interrupt.
Mask corresponding channel interrupt writing to 0x2018 .
Driver Reads corresponding ‘Status register’ 0x1044 which will also clear
status register.
Read channel ‘completed descriptor count’ 0x1048 and compare with
number of descriptor generated. Exit application
program.
Write to channel ‘Control register’ 0x1004 to stop DMA run.
Write to ‘Block channel interrupt Enable Mask’ 0x2014 to enable interrupt Application program reads transfer data from
for next transfer. assigned buffer and writes to a file.
Return control to application program with transfer size.
X19388-061319
PG195 (v4.1) November 24, 2023

Send Feedback
Descriptors
The DMA/Bridge Subsystem for PCI Express® uses a linked list of descriptors that specify the
source, destination, and length of the DMA transfers. Descriptor lists are created by the driver
and stored in host memory. The DMA channel is initialized by the driver with a few control
registers to begin fetching the descriptor lists and executing the DMA operations.
Descriptors describe the memory transfers that the DMA/Bridge Subsystem for PCIe should
perform. Each channel has its own descriptor list. The start address of each channel's descriptor
list is initialized in hardware registers by the driver. After the channel is enabled, the descriptor
channel begins to fetch descriptors from the initial address. Thereafter, it fetches from the
Nxt_adr[63:0] field of the last descriptor that was fetched. Descriptors must be aligned to a
32 byte boundary.
The size of the initial block of adjacent descriptors are specified with the Dsc_Adj register. After
the initial fetch, the descriptor channel uses the Nxt_adj field of the last fetched descriptor to
determine the number of descriptors at the next descriptor address. A block of adjacent
descriptors must not cross a 4K address boundary. The descriptor channel fetches as many
descriptors in a single request as it can, limited by MRRS, the number the adjacent descriptors,
and the available space in the channel's descriptor buffer.
Note: Because MRRS in most host systems is 512 bytes or 1024 bytes, having more than 32 adjacent
descriptors is not allowed on a single request. However, the design will allow a maximum 64 descriptors in
a single block of adjacent descriptors if needed.
Every descriptor in the descriptor list must accurately describe the descriptor or block of
descriptors that follows. In a block of adjacent descriptors, the Nxt_adj value decrements from
the first descriptor to the second to last descriptor which has a value of zero. Likewise, each
descriptor in the block points to the next descriptor in the block, except for the last descriptor
which might point to a new block or might terminate the list.
Termination of the descriptor list is indicated by the Stop control bit. After a descriptor with the
Stop control bit is observed, no further descriptor fetches are issued for that list. The Stop
control bit can only be set on the last descriptor of a block.
When using an AXI4 memory mapped interface, DMA addresses to the card are not translated. If
the Host does not know the card address map, the descriptor must be assembled in the user
logic and submitted to the DMA using the descriptor bypass interface.
Table 5: Descriptor Format
Offset Fields
0x0 Magic[15:0] Rsv[1:0] Nxt_adj[5:0] Control[7:0]
0x04 4’h0, Len[27:0]
0x08 Src_adr[31:0]
0x0C Src_adr[63:32]
PG195 (v4.1) November 24, 2023

Send Feedback
Table 5: Descriptor Format (cont'd)
Offset Fields
0x10 Dst_adr[31:0]
0x14 Dst_adr[63:32]
0x18 Nxt_adr[31:0]
0x1C Nxt_adr[63:32]
Table 6: Descriptor Fields
Offset Field Bit Index Sub Field Description

0x0 Magic 15:0 16'had4b. Code to
verify that the driver
generated descriptor
is valid.
0x0 1:0 Reserved set to 0's
0x0 Nxt_adj 5:0 The number of
additional adjacent
descriptors after the
descriptor located at
the next descriptor
address field.
A block of adjacent
descriptors cannot
cross a 4k boundary.
0x0 5, 6, 7 Reserved
0x0 4 EOP End of packet for
stream interface.
0x0 2, 3 Reserved
0x0 Set to 1 to interrupt
after the engine has
completed this
descriptor. This
requires global
1 Completed
IE_DESCRIPTOR_COMP
Control LETED control flag set
in the H2C/C2H
Channel control
register.
0x0 Set to 1 to stop
fetching descriptors
for this descriptor list.
The stop bit can only
0 Stop
be set on the last
descriptor of an
adjacent block of
descriptors.
0x04 Length 31:28 Reserved set to 0's
0x04 27:0 Length of the data in
bytes.
PG195 (v4.1) November 24, 2023

Send Feedback
Table 6: Descriptor Fields (cont'd)
Offset Field Bit Index Sub Field Description

0x0C-0x8 Src_adr 63:0 Source address for
H2C and memory
mapped transfers.
Metadata writeback
address for C2H
transfers.
0x14-0x10 Dst_adr 63:0 Destination address
for C2H and memory
mapped transfers. Not
used for H2C stream.
0x1C-0x18 Nxt_adr 63:0 Address of the next
descriptor in the list.
The DMA has Bit_width * 512 deep FIFO to hold all descriptors in the descriptor engine. This
descriptor FIFO is shared with all selected channels and it is used only in an internal mode (not in
descriptor bypass mode).
• For Gen3x8 with 2H2C and 2C2H design, AXI bit width is 256 bits. FIFO depth is
256 bit * 512 = 32 B * 512 = 16 KB (512 descriptors). This FIFO is shared by 4 DMA
engines.
Descriptor Bypass
The descriptor fetch engine can be bypassed on a per channel basis through AMD Vivado™ IDE
parameters. A channel with descriptor bypass enabled accepts descriptor from its respective
c2h_dsc_byp or h2c_dsc_byp bus. Before the channel accepts descriptors, the Control
register Run bit must be set. The NextDescriptorAddress and NextAdjacentCount, and Magic
descriptor fields are not used when descriptors are bypassed. The ie_descriptor_stopped
bit in Control register bit does not prevent the user logic from writing additional descriptors. All
descriptors written to the channel are processed, barring writing of new descriptors when the
channel buffer is full.
When XDMA is configured in descriptor bypass mode, there is an 8 deep descriptor FIFO which
is common for all descriptor channels from user.
Poll Mode
Each engine is capable of writing back completed descriptor counts to host memory. This allows
the driver to poll host memory to determine when the DMA is complete instead of waiting for an
interrupt.
PG195 (v4.1) November 24, 2023

Send Feedback
For a given DMA engine, the completed descriptor count writeback occurs when the DMA
completes a transfer for a descriptor, and ie_descriptor_completed and
Pollmode_wb_enable are set. The completed descriptor count reported is the total number of
completed descriptors since the DMA was initiated (not just those descriptors with the
Completed flag set). The writeback address is defined by the Pollmode_hi_wb_addr and
Pollmode_lo_wb_addr registers.
Table 7: Completed Descriptor Count Writeback Format
Offset Fields
0x0 Sts_err 7’h0 Compl_descriptor_count[23:0]
Table 8: Completed Descriptor Count Writeback Fields
Field Description
Sts_err The bitwise OR of any error status bits in the channel Status register.
Compl_descriptor_count[23:0] The lower 24 bits of the Complete Descriptor Count register.
DMA H2C Stream

For host-to-card transfers, data is read from the host at the source address, but the destination
address in the descriptor is unused. Packets can span multiple descriptors. The termination of a
packet is indicated by the EOP control bit. A descriptor with an EOP bit asserts tlast on the
AXI4-Stream user interface on the last beat of data.
Data delivered to the AXI4-Stream interface will be packed for each descriptor. tkeep is all 1s
except for the last cycle of a data transfer of the descriptor if it is not a multiple of the datapath
width. The DMA does not pack data across multiple descriptors.
DMA C2H Stream

For card-to-host transfers, the data is received from the AXI4-Stream interface and written to the
destination address. Packets can span multiple descriptors. The C2H channel accepts data when
it is enabled, and has valid descriptors. As data is received, it fills descriptors in order. When a
descriptor is filled completely or closed due to an end of packet on the interface, the C2H
channel writes back information to the writeback address on the host with pre-defined WB
Magic value 16'h52b4 (Table 10: C2H Stream Writeback Fields), and updated EOP and Length
as appropriate. For valid data cycles on the C2H AXI4-Stream interface, all data associated with a
given packet must be contiguous.
Note: C2H Channel Writeback information is different then Poll mode updates. C2H Channel Writeback
information provides the driver current length status of a particular descriptor. This is different from
Pollmode_*, as is described in Poll Mode.
PG195 (v4.1) November 24, 2023

Send Feedback
The tkeep bits must be all 1s except for the last data transfer of a packet. On the last transfer of
a packet, when tlast is asserted, you can specify a tkeep that is not all 1s to specify a data
cycle that is not the full datapath width. The asserted tkeep bits need to be packed to the lsb,
indicating contiguous data. If tlast is asserted and tkeep has all zero's, this is not a valid
combination for DMA to function properly.
The length of a C2H Stream descriptor (the size of the destination buffer) must always be a
multiple of 64 bytes.
Table 9: C2H Stream Writeback Format
Offset Fields
0x0 WB Magic[15:0] Reserved [14:0] Status[0]
0x04 Length[31:0]
Table 10: C2H Stream Writeback Fields
Field Bit Index Sub Field Description

Status 0 EOP End of packet
Reserved 14:0 Reserved
WB Magic 15:0 16’h52b4. Code to verify the C2H
writeback is valid.
Length 31:0 Length of the data in bytes.
Note: C2H Streaming writeback address cannot cross 4K boundary.
Address Alignment
Table 11: Address Alignment
Datapath
Interface Type Address Restriction
Width
AXI4 MM 64, 128, 256, None
512
AXI4-Stream 64, 128, 256, None
512
AXI4 MM fixed address1 64 Source_addr[2:0] == Destination_addr[2:0] == 3’h0
AXI4 MM fixed address1 512 Source_addr[5:0] == Destination_addr[5:0]==6'h0
Notes:
1. For fixed address mode, you must set bit [25] in the control registers.
PG195 (v4.1) November 24, 2023

Send Feedback
Related Information
H2C Channel Control (0x04)

C2H Channel Control (0x04)
Length Granularity
Table 12: Length Granularity
Datapath
Interface Type Length Granularity Restriction
Width
AXI4 MM 64, 128, 256, None
512
AXI4-Stream 64, 128, 256, None1
512
AXI4 MM fixed address 64 Length[2:0] == 3’h0
AXI4 MM fixed address 512 Length[5:0] == 6'h0
Notes:
1. Each C2H descriptor must be sized as a multiple of 64 Bytes. However, there are no restrictions to the total number of
Bytes in the actual C2H transfer.
Parity
Parity checking occurs one of two ways. Set the Parity Checking option in the PCIe DMA Tab in
the AMD Vivado™ IDE during core customization:
When Check Parity is enabled, the DMA/Bridge Subsystem for PCIe checks for parity on read
data from PCIe, and generates parity for write data to the PCIe.
When Propagate Parity is enabled, the DMA/Bridge Subsystem for PCIe propagates parity to the
user AXI interface. You are responsible for checking and generating parity in the AXI Interface.
Parity is valid every clock cycle when a data valid signal is asserted, and parity bits are valid only
for valid data bytes. Parity is calculated for every byte; total parity bits are DATA_WIDTH/8.
• Parity information is sent and received on *_tuser ports in AXI4-Stream (AXI_ST) mode.
• Parity information is sent and received on *_ruser and *_wuser ports in AXI4 Memory
Mapped (AXI-MM) mode.
Odd parity is used for parity checking. By default, parity checking is not enabled.
Related Information
PCIe DMA Tab
PG195 (v4.1) November 24, 2023

Send Feedback
Port Descriptions
IMPORTANT! This document covers only DMA mode port descriptions. For AXI Bridge mode, see the AXI
Bridge for PCI Express Gen3 Subsystem Product Guide (PG194).
The AMD DMA/Bridge Subsystem for PCI Express® connects directly to the integrated block for
PCIe. The datapath interfaces to the PCIe integrated block IP are 64, 128, 256 or 512-bits wide,
and runs at up to 250 MHz depending on the configuration of the IP. The datapath width applies
to all data interfaces except for the AXI4-Lite interfaces. AXI4-Lite interfaces are fixed at 32-bits
wide.
Ports associated with this subsystem are described in the following tables.
XDMA Global Ports

Table 13: Top-Level Interface Signals
Signal Name Direction Description

sys_clk I 7 series Gen2 and Virtex 7 Gen3: PCIe reference clock. Should be
driven from the O port of reference clock IBUFDS_GTE2.
UltraScale: DRP clock and internal system clock (Half the frequency of
sys_clk_gt if PCIe Reference Clock is 250 MHz, otherwise same
frequency as sys_clk_gt frequency). Should be driven by the ODIV2
port of reference clock IBUFDS_GTE3.
sys_clk_gt I UltraScale only: PCIe reference clock. Should be driven from the O
port of reference clock IBUFDS_GTE3. See the UltraScale Devices Gen3
Integrated Block for PCI Express LogiCORE IP Product Guide (PG156), or
UltraScale+ Devices Integrated Block for PCI Express LogiCORE IP Product
Guide (PG213).
sys_rst_n I Reset from the PCIe edge connector reset signal
axi_aclk O PCIe derived clock output for m_axi* and s_axi* interfaces. axi_aclk is
a derived clock from the TXOUTCLK pin from the GT block; it is not
expected to run continuously while axi_aresetn is asserted.
axi_aresetn O AXI reset signal synchronous with the clock provided on the axi_aclk
output. This reset should drive all corresponding AXI Interconnect
aresetn signals.
dma_bridge_resetn I Optional pin and available only when SOFT_RESET_EN parameter is
set to TRUE. This pin is intended to be user driven reset when link
down, Function Level Reset, Dynamic Function eXchange, or another
error condition defined by user occurs. It is not required to be
toggled during initial link up operation.
When used, all PCIe traffic must be in quiesce state. The signal must
be asserted for longer than the Completion Timeout value (typically
50 ms).
• 0: Resets all internal Bridge engines and registers as well as
asserts the axi_aresetn signal while maintaining PCIe link up.
• 1: Normal operation.
See Clocking and Resets for further instruction on using this signal.
user_lnk_up O Output Active-High Identifies that the PCI Express core is linked up
with a host device.
PG195 (v4.1) November 24, 2023

Send Feedback
Table 13: Top-Level Interface Signals (cont'd)

msi_enable O Indicates when MSI is enabled.
msi_vector_width[2:0] O Indicates the size of the MSI field (the number of MSI vectors
allocated to the device).
msix_enable O Indicates when MSI-X is enabled.
Related Information
Clocking and Resets
PCIe Interface Signals

Table 14: PCIe Interface Signals

pci_exp_rxp[PL_LINK_CAP_MAX_LINK_WIDTH-1:0] I PCIe RX serial interface
pci_exp_rxn[PL_LINK_CAP_MAX_LINK_WIDTH-1:0] I PCIe RX serial interface
pci_exp_txp[PL_LINK_CAP_MAX_LINK_WIDTH-1:0] O PCIe TX serial interface
pci_exp_txn[PL_LINK_CAP_MAX_LINK_WIDTH-1:0] O PCIe TX serial interface
H2C Channel 0-3 AXI4-Stream Interface Signals

Table 15: H2C Channel 0-3 AXI4-Stream Interface Signals
Signal Name1 Direction Description

Assertion of this signal by the user logic indicates that it is
ready to accept data. Data is transferred across the interface
when m_axis_h2c_tready and m_axis_h2c_tvalid are asserted
m_axis_h2c_tready_x I
in the same cycle. If the user logic deasserts the signal when
the valid signal is High, the DMA keeps the valid signal
asserted until the ready signal is asserted.
The DMA asserts this signal in the last beat of the DMA
m_axis_h2c_tlast_x O
packet to indicate the end of the packet.
m_axis_h2c_tdata_x
O Transmit data from the DMA to the user logic.
[DATA_WIDTH-1:0]
The DMA asserts this whenever it is driving valid data on
m_axis_h2c_tvalid_x O
m_axis_h2c_tdata.
m_axis_h2c_tuser_x Parity bits. This port is enabled only in Propagate Parity
O
[DATA_WIDTH/8-1:0] mode.
m_axis_h2c_tkeep_x O The tkeep signal specifies how many bytes are valid when
[DATA_WIDTH/8-1:0] tlast is asserted.
Notes:
1. _x in the signal name changes based on the channel number 0, 1, 2, and 3. For example, for channel 0 use the
m_axis_h2c_tready_0 port, and for channel 1 use the m_axis_h2c_tready_1 port.
PG195 (v4.1) November 24, 2023

Send Feedback
C2H Channel 0-3 AXI4-Stream Interface Signals

Table 16: C2H Channel 0-3 AXI4-Stream Interface Signals
Signal Name1 Direction Description

Assertion of this signal indicates that the DMA is ready to
accept data. Data is transferred across the interface when
s_axis_c2h_tready and s_axis_c2h_tvalid are asserted in the
s_axis_c2h_tready_x O
same cycle. If the DMA deasserts the signal when the valid
signal is High, the user logic must keep the valid signal
asserted until the ready signal is asserted.
The user logic asserts this signal to indicate the end of the
s_axis_c2h_tlast_x I
DMA packet.
s_axis_c2h_tdata_x
I Transmits data from the user logic to the DMA.
[DATA_WIDTH-1:0]
The user logic asserts this whenever it is driving valid data
s_axis_c2h_tvalid_x I
on s_axis_c2h_tdata.
m_axis_c2h_tuser_x Parity bits. This port is enabled only in Propagate Parity
I
[DATA_WIDTH/8-1:0] mode.
The tkeep signal tells how many bytes are valid for each
m_axis_c2h_tkeep_x beat. It must be asserted for all beets except when tlast is
I
[DATA_WIDTH/8-1:0] asserted. You must specify how many bytes are valid when
tlast is asserted.
Notes:
1. _x in the signal name changes based on the channel number 0, 1, 2, and 3. For example, for channel 0 use the
m_axis_c2h_tready_0 port, and for channel 1 use the m_axis_c2h_tready_1 port.
AXI4 Memory Mapped Read Address Interface

Signals
Table 17: AXI4 Memory Mapped Read Address Interface Signals

m_axi_araddr O This signal is the address for a memory mapped read to the
[AXI_ADR_WIDTH-1:0] user logic from the DMA.
m_axi_arid [ID_WIDTH-1:0] O Standard AXI4 description, which is found in the AXI4

Protocol Specification.
m_axi_arlen[7:0] O Master read burst length.
m_axi_arsize[2:0] O Master read burst size.
m_axi_arprot[2:0] O 3’h0
m_axi_arvalid O The assertion of this signal means there is a valid read
request to the address on m_axi_araddr.
m_axi_arready I Master read address ready.
m_axi_arlock O 1’b0
m_axi_arcache[3:0] O 4’h0
m_axi_arburst O Master read burst type.
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4 Memory Mapped Read Interface Signals

Table 18: AXI4 Memory Mapped Read Interface Signals

m_axi_rdata [DATA_WIDTH-1:0] I Master read data.
m_axi_rid [ID_WIDTH-1:0] I Master read ID.
m_axi_rresp[1:0] I Master read response.
m_axi_rlast I Master read last.
m_axi_rvalid I Master read valid.
m_axi_rready O Master read ready.
m_axi_ruser Parity ports for read interface. This port is enabled
I
[DATA_WIDTH/8-1:0] only in Propagate Parity mode.
AXI4 Memory Mapped Write Address Interface

Signals
Table 19: AXI4 Memory Mapped Write Address Interface Signals

m_axi_awaddr This signal is the address for a memory mapped
O
[AXI_ADR_WIDTH-1:0] write to the user logic from the DMA.
m_axi_awid
O Master write address ID.
[ID_WIDTH-1:0]
m_axi_awlen[7:0] O Master write address length.
m_axi_awsize[2:0] O Master write address size.
m_axi_awburst[1:0] O Master write address burst type.
m_axi_awprot[2:0] O 3’h0
The assertion of this signal means there is a valid
m_axi_awvalid O
write request to the address on m_axi_araddr.
m_axi_awready I Master write address ready.
m_axi_awlock O 1’b0
m_axi_awcache[3:0] O 4’h0
AXI4 Memory Mapped Write Interface Signals

Table 20: AXI4 Memory Mapped Write Interface Signals

m_axi_wdata
O Master write data.
[DATA_WIDTH-1:0]
m_axi_wlast O Master write last.
m_axi_wstrb O Master write strobe.
PG195 (v4.1) November 24, 2023

Send Feedback
Table 20: AXI4 Memory Mapped Write Interface Signals (cont'd)

m_axi_wvalid O Master write valid.
m_axi_wready I Master write ready.
m_axi_wuser Parity ports for read interface. This port is enabled
O
AXI4 Memory Mapped Write Response Interface

Signals
Table 21: AXI4 Memory Mapped Write Response Interface Signals

m_axi_bvalid I Master write response valid.
m_axi_bresp[1:0] I Master write response.
m_axi_bid I Master response ID.
[ID_WIDTH-1:0]
m_axi_bready O Master response ready.
AXI4 Memory Mapped Master Bypass Read Address

Interface Signals
Table 22: AXI4 Memory Mapped Master Bypass Read Address Interface Signals

m_axib_araddr O This signal is the address for a memory mapped
[AXI_ADR_WIDTH-1:0] read to the user logic from the host.
m_axib_arid O Master read address ID.

[ID_WIDTH-1:0]
m_axib_arlen[7:0] O Master read address length.
m_axib_arsize[2:0] O Master read address size.
m_axib_arprot[2:0] O 3’h0
m_axib_arvalid O The assertion of this signal means there is a valid
read request to the address on m_axib_araddr.
m_axib_arready I Master read address ready.
m_axib_arlock O 1’b0
m_axib_arcache[3:0] O 4’h0
m_axib_arburst O Master read address burst type.
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4 Memory Mapped Master Bypass Read Interface

Signals
Table 23: AXI4 Memory Mapped Master Bypass Read Interface Signals

m_axib_rdata I Master read data.
[DATA_WIDTH-1:0]
m_axib_rid I Master read ID.
[ID_WIDTH-1:0]
m_axib_rresp[1:0] I Master read response.
m_axib_rlast I Master read last.
m_axib_rvalid I Master read valid.
m_axib_rready O Master read ready.
m_axib_ruser I Parity ports for read interface. This port is enabled
AXI4 Memory Mapped Master Bypass Write Address

Interface Signals
Table 24: AXI4 Memory Mapped Master Bypass Write Address Interface Signals

m_axib_awaddr O This signal is the address for a memory mapped
[AXI_ADR_WIDTH-1:0] write to the user logic from the host.
m_axib_awid O Master write address ID.

[ID_WIDTH-1:0]
m_axib_awlen[7:0] O Master write address length.
m_axib_awsize[2:0] O Master write address size.
m_axib_awburst[1:0] O Master write address burst type.
m_axib_awprot[2:0] O 3’h0
m_axib_awvalid O The assertion of this signal means there is a valid
write request to the address on m_axib_araddr.
m_axib_awready I Master write address ready.
m_axib_awlock O 1’b0
m_axib_awcache[3:0] O 4’h0
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4 Memory Mapped Master Bypass Write Interface

Signals
Table 25: AXI4 Memory Mapped Master Bypass Write Interface Signals

m_axib_wdata O Master write data.
[DATA_WIDTH-1:0]
m_axib_wlast O Master write last.
m_axib_wstrb O Master write strobe.
m_axib_wvalid O Master write valid.
m_axib_wready I Master write ready.
m_axib_wuser O Parity ports for read interface. This port is enabled
AXI4 Memory Mapped Master Bypass Write Response

Interface Signals
Table 26: AXI4 Memory Mapped Master Bypass Write Response Interface Signals

m_axib_bvalid I Master write response valid.
m_axib_bresp[1:0] I Master write response.
m_axib_bid I Master write response ID.
[ID_WIDTH-1:0]
m_axib_bready O Master response ready.
Config AXI4-Lite Memory Mapped Write Master

Interface Signals
Table 27: Config AXI4-Lite Memory Mapped Write Master Interface Signals

m_axil_awaddr[31:0] O This signal is the address for a memory mapped
write to the user logic from the host.
m_axil_awprot[2:0] O 3’h0
m_axil_awvalid O The assertion of this signal means there is a valid
write request to the address on m_axil_awaddr.
m_axil_awready I Master write address ready.
m_axil_wdata[31:0] O Master write data.
m_axil_wstrb O Master write strobe.
m_axil_wvalid O Master write valid.
PG195 (v4.1) November 24, 2023

Send Feedback
Table 27: Config AXI4-Lite Memory Mapped Write Master Interface Signals (cont'd)

m_axil_wready I Master write ready.
m_axil_bvalid I Master response valid.
m_axil_bready O Master response ready.
Config AXI4-Lite Memory Mapped Read Master

Interface Signals
Table 28: Config AXI4-Lite Memory Mapped Read Master Interface Signals

m_axil_araddr[31:0] O This signal is the address for a memory mapped
read to the user logic from the host.
m_axil_arprot[2:0] O 3’h0
m_axil_arvalid O The assertion of this signal means there is a valid
read request to the address on m_axil_araddr.
m_axil_arready I Master read address ready.
m_axil_rdata[31:0] I Master read data.
m_axil_rresp I Master read response.
m_axil_rvalid I Master read valid.
m_axil_rready O Master read ready.
Config AXI4-Lite Memory Mapped Write Slave

Interface Signals
Table 29: Config AXI4-Lite Memory Mapped Write Slave Interface Signals

s_axil_awaddr[31:0] I This signal is the address for a memory mapped
write to the DMA from the user logic.
s_axil_awvalid I The assertion of this signal means there is a valid
write request to the address on s_axil_awaddr.
s_axil_awprot[2:0] I Unused
s_axil_awready O Slave write address ready.
s_axil_wdata[31:0] I Slave write data.
s_axil_wstrb I Slave write strobe.
s_axil_wvalid I Slave write valid.
s_axil_wready O Slave write ready.
s_axil_bvalid O Slave write response valid.
s_axil_bready I Save response ready.
PG195 (v4.1) November 24, 2023

Send Feedback
Config AXI4-Lite Memory Mapped Read Slave

Interface Signals
Table 30: Config AXI4-Lite Memory Mapped Read Slave Interface Signals

s_axil_araddr[31:0] I This signal is the address for a memory mapped
read to the DMA from the user logic.
s_axil_arprot[2:0] I Unused
s_axil_arvalid I The assertion of this signal means there is a valid
read request to the address on s_axil_araddr.
s_axil_arready O Slave read address ready.
s_axil_rdata[31:0] O Slave read data.
s_axil_rresp O Slave read response.
s_axil_rvalid O Slave read valid.
s_axil_rready I Slave read ready.
Interrupt Interface
Table 31: Interrupt Interface

usr_irq_req[NUM_USR_IRQ-1:0] I Assert to generate an interrupt. Maintain assertion
until interrupt is serviced.
usr_irq_ack[NUM_USR_IRQ-1:0] O Indicates that the interrupt has been sent on PCIe.
Two acks are generated for legacy interrupts. One
ack is generated for MSI interrupts.
Each bits in usr_irq_reqbus corresponds to the same bits in usr_irq_ack. For example,
usr_irq_ack[0] represents an ack for usr_irq_req[0].
PG195 (v4.1) November 24, 2023

Send Feedback
Channel 0-3 Status Ports

Table 32: Channel 0-3 Status Ports

h2c_sts [7:0] O Status bits for each channel. Bit:
7: Reserved
6: Control register 'Run' bit
5: IRQ event pending
4: Packet Done event (AXI4-Stream)
3: Descriptor Done event. Pulses for one cycle for
each descriptor that is completed, regardless of the
Descriptor.Completed field
2: Status register Descriptor_stop bit
1: Status register Descriptor_completed bit
0: Status register busy bit
c2h_sts [7:0] O Status bits for each channel. Bit:
7: Reserved
6: Control register 'Run' bit
5: IRQ event pending
4: Packet Done event (AXI4-Stream)
3: Descriptor Done event. Pulses for one cycle for
each descriptor that is completed, regardless of the
Descriptor.Completed field
2: Status register Descriptor_stop bit
1: Status register Descriptor_completed bit
0: Status register busy bit
Configuration Extend Interface Port Descriptions

The Configuration Extend interface allows the core to transfer configuration information with the
user application when externally implemented configuration registers are implemented. The
following table defines the ports in the Configuration Extend interface of the core.
Note: This interface is not available for 7 series Gen2 IP.
PG195 (v4.1) November 24, 2023

Send Feedback
Table 33: Configuration Extend Interface Port Descriptions
Port Direction Width Description

cfg_ext_read_received O 1 Configuration Extend Read Received
The core asserts this output when it has received a configuration
read request from the link. When neither user-implemented
legacy or extended configuration space is enabled, receipt of a
configuration read results in a one-cycle assertion of this signal,
together with valid cfg_ext_register_number and
cfg_ext_function_number.
When user-implemented legacy, extended configuration space,
or both are enabled, for the cfg_ext_register_number falling
below mentioned ranges, this signal is asserted and the user
logic must present the cfg_ext_read_data and
cfg_ext_read_data_valid.
Legacy Space:
0xB0-0xBF
Extended Configuration space:
0xE80 - 0xFFF (UltraScale+ HBM PCIe4C cores only)
0x480 - 0x4FF
cfg_ext_write_received O 1 Configuration Extend Write Received
The core generates a one-cycle pulse on this output when it has
received a configuration write request from the link.
cfg_ext_register_number O 10 Configuration Extend Register Number
The 10-bit address of the configuration register being read or
written. The data is valid when cfg_ext_read_received or
cfg_ext_write_received is High.
cfg_ext_function_number O 8 Configuration Extend Function Number
The 8-bit function number corresponding to the configuration
read or write request. The data is valid when
cfg_ext_read_received or cfg_ext_write_received is High.
cfg_ext_write_data O 32 Configuration Extend Write Data
Data being written into a configuration register. This output is
valid when cfg_ext_write_received is High.
cfg_ext_write_byte_enable O 4 Configuration Extend Write Byte Enable Byte enables for a
configuration write transaction.
cfg_ext_read_data I 32 Configuration Extend Read Data
You can provide data from an externally implemented
configuration register to the core through this bus. The core
samples this data on the next positive edge of the clock after it
sets cfg_ext_read_received High, if you have set
cfg_ext_read_data_valid.
cfg_ext_read_data_valid I 1 Configuration Extend Read Data Valid
The user application asserts this input to the core to supply data
from an externally implemented configuration register. The core
samples this input data on the next positive edge of the clock
after it sets cfg_ext_read_received High.
Configuration Management Interface Ports

The Configuration Management interface is used to read and write to the Configuration Space
Registers. The following table defines the ports in the Configuration Management interface of
the core.
PG195 (v4.1) November 24, 2023

Send Feedback
Table 34: Configuration Management Interface Ports
Port Direction Width Description

cfg_mgmt_addr I 19 Read/Write Address.
Configuration Space Dword-aligned address
cfg_mgmt_byte_enable I 4 Byte Enable
Byte Enable for write data, where cfg_mgmt_byte_enable[0]
corresponds to cfg_mgmt_write_data[7:0] and so on
cfg_mgmt_read_data O 32 Read Data Out
Read data provides the configuration of the Configuration and
Management registers
cfg_mgmt_read I 1 Read Enable
Asserted for a read operation. Active-High
cfg_mgmt_read_write_done O 1 Read/Write Operation Complete
Asserted for 1 cycle when operation is complete. Active-High
cfg_mgmt_write_data I 32 Write data
Write data is used to configure the Configuration and
Management registers
cfg_mgmt_write I 1 Write Enable
Asserted for a write operation. Active-High
Descriptor Bypass Mode

In the PCIe DMA tab of the Vivado IDE, if Descriptor Bypass for Read (H2C) or Descriptor
Bypass for Write (C2H) is selected, these ports are present. Each binary bit corresponds to a
channel (LSB correspond to Channel 0). Value 1 in bit positions means the corresponding channel
descriptor bypass is enabled.
Table 35: H2C 0-3 Descriptor Bypass Port
Port Direction Description

h2c_dsc_byp_ready O Channel is ready to accept new descriptors. After
h2c_dsc_byp_ready is deasserted, one additional descriptor
can be written. The Control register 'Run' bit must be
asserted before the channel accepts descriptors.
h2c_dsc_byp_load I Write the descriptor presented at h2c_dsc_byp_data into the
channel’s descriptor buffer.
h2c_dsc_byp_src_addr[63:0] I Descriptor source address to be loaded.
h2c_dsc_byp_dst_addr[63:0] I Descriptor destination address to be loaded.
h2c_dsc_byp_len[27:0] I Descriptor length to be loaded.
PG195 (v4.1) November 24, 2023

Send Feedback
Table 35: H2C 0-3 Descriptor Bypass Port (cont'd)

h2c_dsc_byp_ctl[15:0] I Descriptor control to be loaded.
[0]: Stop. Set to 1 to stop fetching next descriptor.
[1]: Completed. Set to 1 to interrupt after the engine has
completed this descriptor.
[3:2]: Reserved.
[4]: EOP. End of Packet for AXI-Stream interface.
[15:5]: Reserved.
All reserved bits can be forced to 0s.
control port "h2c_dsc_byp_ctl[4:0]" are same as in descriptor
control, refer to Table 6.
Table 36: C2H 0-3 Descriptor Bypass Ports

c2h_dsc_byp_ready O Channel is ready to accept new descriptors. After
c2h_dsc_byp_ready is deasserted, one additional descriptor
can be written. The Control register 'Run' bit must be
asserted before the channel accepts descriptors.
c2h_dsc_byp_load I Descriptor presented at c2h_dsc_byp_* is valid.
c2h_dsc_byp_src_addr[63:0] I Descriptor source address to be loaded.
c2h_dsc_byp_dst_addr[63:0] I Descriptor destination address to be loaded.
c2h_dsc_byp_len[27:0] I Descriptor length to be loaded.
c2h_dsc_byp_ctl[15:0] I Descriptor control to be loaded.
[0]: Stop. Set to 1 to stop fetching next descriptor.
[1]: Completed. Set to 1 to interrupt after the engine has
completed this descriptor.
[3:2]: Reserved.
[4]: EOP. End of Packet for AXI-Stream interface.
[15:5]: Reserved.
All reserved bits can be forced to 0s.
control port "h2c_dsc_byp_ctl[4:0]" are same as in descriptor
control, refer to Table 6.
The following timing diagram shows how to input the descriptor in descriptor bypass mode.
When dsc_byp_ready is asserted, a new descriptor can be pushed in with the
dsc_byp_load signal.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 8: Timing Diagram for Descriptor Bypass Mode
IMPORTANT! Immediately after dsc_byp_ready is deasserted, one more descriptor can be pushed in.
In the above timing diagram, a descriptor is pushed in when dsc_byp_ready is deasserted.
Related Information
PCIe DMA Tab
Register Space
Note: This document covers only DMA mode register space. For AXI Bridge mode, see the AXI Bridge for
PCI Express Gen3 Subsystem Product Guide (PG194).
Configuration and status registers internal to the DMA/Bridge Subsystem for PCI Express® and
those in the user logic can be accessed from the host through mapping the read or write request
to a Base Address Register (BAR). Based on the BAR hit, the request is routed to the appropriate
location. For PCIe BAR assignments, see Target Bridge.
PCIe to AXI Bridge Master Address Map

Transactions that hit the PCIe to AXI Bridge Master are routed to the AXI4 Memory Mapped user
interface.
PCIe to DMA Address Map

Transactions that hit the PCIe to DMA space are routed to the DMA Subsystem for the
PCIeDMA/Bridge Subsystem for PCI Express® internal configuration register bus. This bus
supports 32 bits of address space and 32-bit read and write requests.
DMA/Bridge Subsystem for PCIe registers can be accessed from the host or from the AXI Slave
interface. These registers should be used for programming the DMA and checking status.
PG195 (v4.1) November 24, 2023

Send Feedback
PCIe to DMA Address Format

Table 37: PCIe to DMA Address Format
31:16 15:12 11:8 7:0

Reserved Target Channel Byte Offset
Table 38: PCIe to DMA Address Field Descriptions
Bit Index Field Description

The destination submodule within the DMA
4’h0: H2C Channels
4’h1: C2H Channels
4’h2: IRQ Block
15:12 Target 4’h3: Config
4’h4: H2C SGDMA
4’h5: C2H SGDMA
4’h6: SGDMA Common
4'h8: MSI-X
This field is only applicable for H2C Channel, C2H Channel, H2C SGDMA,
11:8 Channel ID[3:0] and C2H SGDMA Targets. This field indicates which engine is being
addressed for these Targets. For all other Targets this field must be 0.
The byte address of the register to be accessed within the target.
7:0 Byte Offset
Bits[1:0] must be 0.
PCIe to DMA Configuration Registers

Table 39: Configuration Register Attribute Definitions
Attribute Description
RV Reserved
RW Read/Write
RC Clear on Read.
W1C Write 1 to Clear
W1S Write 1 to Set
RO Read Only
WO Write Only
Some registers can be accessed with different attributes. In such cases different register offsets
are provided for each attribute. Undefined bits and address space is reserved. In some registers,
individual bits in a vector might represent a specific DMA engine. In such cases the LSBs of the
vectors correspond to the H2C channel (if any). Channel ID 0 is in the LSB position. Bits
representing the C2H channels are packed just above them.
PG195 (v4.1) November 24, 2023

Send Feedback
H2C Channel Registers (0x0)

The H2C channel register space is described in this section.
Table 40: H2C Channel Register Space
Address (hex) Register Name

0x00 H2C Channel Identifier (0x00)
0x04 H2C Channel Control (0x04)
0x08 H2C Channel Control (0x08)
0x0C H2C Channel Control (0x0C)
0x40 H2C Channel Status (0x40)
0x44 H2C Channel Status (0x44)
0x48 H2C Channel Completed Descriptor Count (0x48)
0x4C H2C Channel Alignments (0x4C)
0x88 H2C Poll Mode Low Write Back Address (0x88)
0x8C H2C Poll Mode High Write Back Address (0x8C)
0x90 H2C Channel Interrupt Enable Mask (0x90)
0xC0 H2C Channel Performance Monitor Control (0xC0)
0xC4 H2C Channel Performance Cycle Count (0xC4)
0xC8 H2C Channel Performance Cycle Count (0xC8)
0xCC H2C Channel Performance Data Count (0xCC)
0xD0 H2C Channel Performance Data Count (0xD0)
H2C Channel Identifier (0x00)
Table 41: H2C Channel Identifier (0x00)
Bit Index Default Value Access Type Description
31:20 12’h1fc RO Subsystem identifier

19:16 4’h0 RO H2C Channel Target
15 1’b0 RO Stream
1: AXI4-Stream Interface
0: AXI4 Memory Mapped Interface
14:12 0 RO Reserved
11:8 Varies RO Channel ID Target [3:0]
PG195 (v4.1) November 24, 2023

Send Feedback
Table 41: H2C Channel Identifier (0x00) (cont'd)
Bit Index Default Value Access Type Description
7:0 8'h04 RO Version

8'h01: 2015.3 and 2015.4
8'h02: 2016.1
8'h03: 2016.2
8'h04: 2016.3
8'h05: 2016.4
8'h06: 2017.1 to current release
Table 42: H2C Channel Control (0x04)
Bit Index Default Access Type Description

31:28 Reserved
27 1’b0 RW When set write back information for C2H in AXI-Stream
mode is disabled, default write back is enabled.
26 0x0 RW pollmode_wb_enable
Poll mode writeback enable.
When this bit is set the DMA writes back the completed
descriptor count when a descriptor with the Completed bit
set, is completed.
25 1’b0 RW non_inc_mode
Non-incrementing address mode. Applies to m_axi_araddr
interface only.
24 Reserved
23:19 5’h0 RW ie_desc_error
Set to all 1s (0x1F) to enable logging of Status.Desc_error
and to stop the engine if the error is detected.
18:14 5’h0 RW ie_write_error
Set to all 1s (0x1F) to enable logging of Status.Write_error
13:9 5’h0 RW ie_read_error
Set to all 1s (0x1F) to enable logging of Status.Read_error
8:7 Reserved
6 1’b0 RW ie_idle_stopped
Set to 1 to enable logging of Status.Idle_stopped
5 1’b0 RW ie_invalid_length
Set to 1 to enable logging of Status.Invalid_length
4 1’b0 RW ie_magic_stopped
Set to 1 to enable logging of Status.Magic_stopped
3 1’b0 RW ie_align_mismatch
Set to 1 to enable logging of Status.Align_mismatch
2 1’b0 RW ie_descriptor_completed
Set to 1 to enable logging of Status.Descriptor_completed
PG195 (v4.1) November 24, 2023

Send Feedback
Table 42: H2C Channel Control (0x04) (cont'd)

1 1’b0 RW ie_descriptor_stopped
Set to 1 to enable logging of Status.Descriptor_stopped
0 1’b0 RW Run
Set to 1 to start the SGDMA engine. Reset to 0 to stop
transfer; if the engine is busy it completes the current
descriptor.
Notes:
1. The ie_* register bits are interrupt enabled. When these bits are set and the corresponding condition is met, status
will be logged in the H2C Channel Status (0x40). When the proper interrupt masks are set (per H2C Channel Interrupt
Enable Mask (0x90)), the interrupt will be generated.
Table 43: H2C Channel Control (0x08)
Bit Index Access Type Description

31:28 Reserved
27:0 W1S Control
Bit descriptions are the same as in H2C Channel Control (0x04).
H2C Channel Control (0x0C)
Table 44: H2C Channel Control (0x0C)
Bit Index Access Type Description

27:0 W1C Control
Bit descriptions are the same as in H2C Channel Control (0x04).
H2C Channel Status (0x40)
Table 45: H2C Channel Status (0x40)

31:24 Reserved
23:19 5’h0 RW1C descr_error[4:0]
Reset (0) on setting the Control register Run bit.
4: Unexpected completion
3: Header EP
2: Parity error
1: Completer abort
0: Unsupported request
PG195 (v4.1) November 24, 2023

Send Feedback
Table 45: H2C Channel Status (0x40) (cont'd)

18:14 5’h0 RW1C write_error[4:0]
Bit position:
4-2: Reserved
1: Slave error
0: Decode error
13:9 5’h0 RW1C read_error[4:0]
Bit position
4: Unexpected completion
3: Header EP
2: Parity error
1: Completer abort
8:7 Reserved
6 1’b0 RW1C idle_stopped
Reset (0) on setting the Control register Run bit. Set when
the engine is idle after resetting the Control register Run bit
if the Control register ie_idle_stopped bit is set.
5 1’b0 RW1C invalid_length
Reset on setting the Control register Run bit. Set when the
descriptor length is not a multiple of the data width of an
AXI4-Stream channel and the Control register
ie_invalid_length bit is set.
4 1’b0 RW1C magic_stopped
engine encounters a descriptor with invalid magic and
stopped if the Control register ie_magic_stopped bit is set.
3 1’b0 RW1C align_mismatch
Source and destination address on descriptor are not
properly aligned to each other.
2 1’b0 RW1C descriptor_completed
Reset on setting the Control register Run bit. Set after the
engine has completed a descriptor with the COMPLETE bit
set if the Control register ie_descriptor_stopped bit is set.
1 1’b0 RW1C descriptor_stopped
Reset on setting Control register Run bit. Set after the
engine completed a descriptor with the STOP bit set if the
Control register ie_descriptor_stopped bit is set.
0 1’b0 RO Busy
Set if the SGDMA engine is busy. Zero when it is idle.
PG195 (v4.1) November 24, 2023

Send Feedback
H2C Channel Status (0x44)
Table 46: H2C Channel Status (0x44)

23:1 RC Status
Clear on Read. Bit description is the same as in H2C Channel
Status (0x40).
Bit 0 cannot be cleared.
H2C Channel Completed Descriptor Count (0x48)
Table 47: H2C Channel Completed Descriptor Count (0x48)

31:0 32’h0 RO compl_descriptor_count
The number of competed descriptors update by the engine
after completing each descriptor in the list.
Reset to 0 on rising edge of Control register Run bit (H2C
Channel Control (0x04)).
H2C Channel Alignments (0x4C)
Table 48: H2C Channel Alignments (0x4C)

23:16 Configuration RO addr_alignment
based The byte alignment that the source and destination
addresses must align to. This value is dependent on
configuration parameters.
15:8 Configuration RO len_granularity
based The minimum granularity of DMA transfers in bytes.
7:0 Configuration RO address_bits
based The number of address bits configured.
H2C Poll Mode Low Write Back Address (0x88)
Table 49: H2C Poll Mode Low Write Back Address (0x88)

31:0 0x0 RW Pollmode_lo_wb_addr[31:0]
Lower 32 bits of the poll mode writeback address.
PG195 (v4.1) November 24, 2023

Send Feedback
H2C Poll Mode High Write Back Address (0x8C)
Table 50: H2C Poll Mode High Write Back Address (0x8C)

31:0 0x0 RW Pollmode_hi_wb_addr[63:32]
Upper 32 bits of the poll mode writeback address.
H2C Channel Interrupt Enable Mask (0x90)
Table 51: H2C Channel Interrupt Enable Mask (0x90)

23:19 5’h0 RW im_desc_error[4:0]
Set to 1 to interrupt when corresponding status register
read_error bit is logged.
18:14 5’h0 RW im_write_error[4:0]
set to 1 to interrupt when corresponding status register
write_error bit is logged.
13:9 5’h0 RW im_read_error[4:0]
set to 1 to interrupt when corresponding status register
read_error bit is logged.
8:7 Reserved
6 1’b0 RW im_idle_stopped
Set to 1 to interrupt when the status register idle_stopped
bit is logged.
5 1’b0 RW im_invalid_length
Set to 1 to interrupt when status register invalid_length bit is
logged.
4 1’b0 RW im_magic_stopped
set to 1 to interrupt when status register magic_stopped bit
is logged.
3 1’b0 RW im_align_mismatch
set to 1 to interrupt when status register align_mismatch bit
is logged.
2 1’b0 RW im_descriptor_completd
set to 1 to interrupt when status register
descriptor_completed bit is logged.
1 1’b0 RW im_descriptor_stopped
set to 1 to interrupt when status register descriptor_stopped
bit is logged.

W1S Interrupt Enable Mask
PG195 (v4.1) November 24, 2023

Send Feedback

W1C Interrupt Enable Mask
H2C Channel Performance Monitor Control (0xC0)
Table 54: H2C Channel Performance Monitor Control (0xC0)

2 1’b0 RW Run
Set to 1 to arm performance counters. Counter starts after
the Control register Run bit is set.
Set to 0 to halt performance counters.
1 1’b0 WO Clear
Write 1 to clear performance counters.
0 1’b0 RW Auto
Automatically stop performance counters when a descriptor
with the stop bit is completed. Automatically clear
performance counters when the Control register Run bit is
set. Writing 1 to the Performance Monitor Control register
Run bit is still required to start the counters.
H2C Channel Performance Cycle Count (0xC4)
Table 55: H2C Channel Performance Cycle Count (0xC4)

31:0 32’h0 RO pmon_cyc_count[31:0]
Increments once per clock while running. See the
Performance Monitor Control register (0xC0) bits Clear and
Auto for clearing.
H2C Channel Performance Cycle Count (0xC8)
Table 56: H2C Channel Performance Cycle Count (0xC8)

16 1’b0 RO pmon_cyc_count_maxed
Cycle count maximum was hit.
9:0 10’h0 RO pmon_cyc_count [41:32]
Auto for clearing.
PG195 (v4.1) November 24, 2023

Send Feedback
H2C Channel Performance Data Count (0xCC)
Table 57: H2C Channel Performance Data Count (0xCC)

31:0 32’h0 RO pmon_dat_count[31:0]
Increments for each valid read data beat while running. See
the Performance Monitor Control register (0xC0) bits Clear
and Auto for clearing.
H2C Channel Performance Data Count (0xD0)
Table 58: H2C Channel Performance Data Count (0xD0)

16 1’b0 RO pmon_dat_count_maxed
Data count maximum was hit
15:10 Reserved
9:0 10’h0 RO pmon_dat_count [41:32]
C2H Channel Registers (0x1)

The C2H channel register space is described in this section.
Table 59: C2H Channel Register Space

0x00 C2H Channel Identifier (0x00)
0x04 C2H Channel Control (0x04)
0x08 C2H Channel Control (0x08)
0x0C C2H Channel Control (0x0C)
0x40 C2H Channel Status (0x40)
0x44 C2H Channel Status (0x44)
0x48 C2H Channel Completed Descriptor Count (0x48)
0x4C C2H Channel Alignments (0x4C)
0x88 C2H Poll Mode Low Write Back Address (0x88)
0x8C C2H Poll Mode High Write Back Address (0x8C)
0x90 C2H Channel Interrupt Enable Mask (0x90)
0xC0 C2H Channel Performance Monitor Control (0xC0)
0xC4 C2H Channel Performance Cycle Count (0xC4)
PG195 (v4.1) November 24, 2023

Send Feedback
Table 59: C2H Channel Register Space (cont'd)

0xC8 C2H Channel Performance Cycle Count (0xC8)
0xCC C2H Channel Performance Data Count (0xCC)
0xD0 C2H Channel Performance Data Count (0xD0)
C2H Channel Identifier (0x00)
Table 60: C2H Channel Identifier (0x00)

19:16 4’h1 RO C2H Channel Target
15 1’b0 RO Stream
14:12 0 RO Reserved
8'h01: 2015.3 and 2015.4
8'h02: 2016.1
8'h03: 2016.2
8'h04: 2016.3
8'h05: 2016.4
Table 61: C2H Channel Control (0x04)

31:28 Reserved
27 0x0 RW Disables the metadata writeback for C2H AXI4-Stream. No
effect if the channel is configured to use AXI Memory
Mapped.
26 0x0 RW pollmode_wb_enable
Poll mode writeback enable.
When this bit is set, the DMA writes back the completed
descriptor count when a descriptor with the Completed bit
set, is completed.
25 1’b0 RW non_inc_mode
Non-incrementing address mode. Applies to m_axi_araddr
interface only.
23:19 5’h0 RW ie_desc_error
Set to all 1s (0x1F) to enable logging of Status.Desc_error
PG195 (v4.1) November 24, 2023

Send Feedback
Table 61: C2H Channel Control (0x04) (cont'd)

13:9 5’h0 RW ie_read_error
Set to all 1s (0x1F) to enable logging of Status.Read_error
and to stop the engine if the error is detected
8:7 Reserved
6 1’b0 RW ie_idle_stopped
Set to 1 to enable logging of Status.Idle_stopped
5 1’b0 RW ie_invalid_length
Set to 1 to enable logging of Status.Invalid_length
4 1’b0 RW ie_magic_stopped
Set to 1 to enable logging of Status.Magic_stopped
3 1’b0 RW ie_align_mismatch
Set to 1 to enable logging of Status.Align_mismatch
2 1’b0 RW ie_descriptor_completed
Set to 1 to enable logging of Status.Descriptor_completed
1 1’b0 RW ie_descriptor_stopped
Set to 1 to enable logging of Status.Descriptor_stopped
0 1’b0 RW Run
Set to 1 to start the SGDMA engine. Reset to 0 to stop the
transfer, if the engine is busy it completes the current
descriptor.
Notes:
1. The ie_* register bits are interrupt enabled. When these bits are set and the corresponding condition is met, the
status will be logged in the C2H Channel Status (0x40). When proper interrupt masks are set (per C2H Channel
Interrupt Enable Mask (0x90) ), the interrupt will be generated.
Table 62: C2H Channel Control (0x08)

W1S Control
Bit descriptions are the same as in C2H Channel Control
(0x04).
C2H Channel Control (0x0C)
Table 63: C2H Channel Control (0x0C)

W1C Control
Bit descriptions are the same as in C2H Channel Control
(0x04).
PG195 (v4.1) November 24, 2023

Send Feedback
C2H Channel Status (0x40)
Table 64: C2H Channel Status (0x40)

23:19 5’h0 RW1C descr_error[4:0]
Bit position:
4:Unexpected completion
3: Header EP
2: Parity error
1: Completer abort
13:9 5’h0 RW1C read_error[4:0]
Bit position:
4-2: Reserved
1: Slave error
0: Decode error
8:7 Reserved
6 1’b0 RW1C idle_stopped
Reset (0) on setting the Control register Run bit. Set when
the engine is idle after resetting the Control register Run bit
if the Control register ie_idle_stopped bit is set.
5 1’b0 RW1C invalid_length
descriptor length is not a multiple of the data width of an
AXI4-Stream channel and the Control register
ie_invalid_length bit is set.
4 1’b0 RW1C magic_stopped
engine encounters a descriptor with invalid magic and
stopped if the Control register ie_magic_stopped bit is set.
3 13’b0 RW1C align_mismatch
Source and destination address on descriptor are not
properly aligned to each other.
2 1’b0 RW1C descriptor_completed
engine has completed a descriptor with the COMPLETE bit
set if the Control register ie_descriptor_completed bit is set.
1 1’b0 RW1C descriptor_stopped
engine completed a descriptor with the STOP bit set if the
Control register ie_magic_stopped bit is set.
0 1’b0 RO Busy
Set if the SGDMA engine is busy. Zero when it is idle.
PG195 (v4.1) November 24, 2023

Send Feedback
C2H Channel Status (0x44)
Table 65: C2H Channel Status (0x44)

23:1 RC Status
Bit descriptions are the same as in C2H Channel Status
(0x40).
C2H Channel Completed Descriptor Count (0x48)
Table 66: C2H Channel Completed Descriptor Count (0x48)

31:0 32’h0 RO compl_descriptor_count
The number of competed descriptors update by the engine
after completing each descriptor in the list.
Reset to 0 on rising edge of Control register, run bit (C2H
Channel Control (0x04)).
C2H Channel Alignments (0x4C)
Table 67: C2H Channel Alignments (0x4C)

23:16 varies RO addr_alignment
The byte alignment that the source and destination
addresses must align to. This value is dependent on
configuration parameters.
15:8 Varies RO len_granularity
The minimum granularity of DMA transfers in bytes.
7:0 ADDR_BITS RO address_bits
The number of address bits configured.
C2H Poll Mode Low Write Back Address (0x88)
Table 68: C2H Poll Mode Low Write Back Address (0x88)

31:0 0x0 RW Pollmode_lo_wb_addr[31:0]
Lower 32 bits of the poll mode writeback address.
PG195 (v4.1) November 24, 2023

Send Feedback
C2H Poll Mode High Write Back Address (0x8C)
Table 69: C2H Poll Mode High Write Back Address (0x8C)

31:0 0x0 RW Pollmode_hi_wb_addr[63:32]
Upper 32 bits of the poll mode writeback address.
C2H Channel Interrupt Enable Mask (0x90)
Table 70: C2H Channel Interrupt Enable Mask (0x90)

23:19 5’h0 RW im_desc_error[4:0]
set to 1 to interrupt when corresponding Status.Read_Error
is logged.
13:9 5’h0 RW im_read_error[4:0]
set to 1 to interrupt when corresponding Status.Read_Error
is logged.
8:7 Reserved
6 1’b0 RW im_idle_stopped
set to 1 to interrupt when the Status.Idle_stopped is logged.
4 1’b0 RW im_magic_stopped
set to 1 to interrupt when Status.Magic_stopped is logged.
2 1’b0 RW im_descriptor_completd
set to 1 to interrupt when Status.Descriptor_completed is
logged.
1 1’b0 RW im_descriptor_stopped
set to 1 to interrupt when Status.Descriptor_stopped is
logged.
0 Reserved

W1S Interrupt Enable Mask
Bit descriptions are the same as in C2H Channel Interrupt
Enable Mask (0x90).
PG195 (v4.1) November 24, 2023

Send Feedback

W1C Interrupt Enable Mask
Bit Descriptions are the same as in C2H Channel Interrupt
Enable Mask (0x90).
C2H Channel Performance Monitor Control (0xC0)
Table 73: C2H Channel Performance Monitor Control (0xC0)

2 1’b0 RW Run
Set to 1 to arm performance counters. Counter starts after
the Control register Run bit is set.
Set to 0 to halt performance counters.
1 1’b0 WO Clear
Write 1 to clear performance counters.
0 1’b0 RW Auto
Automatically stop performance counters when a descriptor
with the stop bit is completed. Automatically clear
performance counters when the Control register Run bit is
set. Writing 1 to the Performance Monitor Control register
Run bit is still required to start the counters.
C2H Channel Performance Cycle Count (0xC4)
Table 74: C2H Channel Performance Cycle Count (0xC4)

31:0 32’h0 RO pmon_cyc_count[31:0]
Auto for clearing.
C2H Channel Performance Cycle Count (0xC8)
Table 75: C2H Channel Performance Cycle Count (0xC8)

16 1’b0 RO pmon_cyc_count_maxed
Cycle count maximum was hit.
15:10 Reserved
9:0 10’h0 RO pmon_cyc_count [41:32]
Auto for clearing.
PG195 (v4.1) November 24, 2023

Send Feedback
C2H Channel Performance Data Count (0xCC)
Table 76: C2H Channel Performance Data Count (0xCC)

31:0 32’h0 RO pmon_dat_count[31:0]
C2H Channel Performance Data Count (0xD0)
Table 77: C2H Channel Performance Data Count (0xD0)

16 1’b0 RO pmon_dat_count_maxed
Data count maximum was hit
15:10 Reserved
9:0 10’h0 RO pmon_dat_count [41:32]
IRQ Block Registers (0x2)

The IRQ Block registers are described in this section.
Table 78: IRQ Block Register Space

0x00 IRQ Block Identifier (0x00)
0x04 IRQ Block User Interrupt Enable Mask (0x04)
0x08 IRQ Block User Interrupt Enable Mask (0x08)
0x0C IRQ Block User Interrupt Enable Mask (0x0C)
0x10 IRQ Block Channel Interrupt Enable Mask (0x10)
0x40 IRQ Block User Interrupt Request (0x40)
0x44 IRQ Block Channel Interrupt Request (0x44)
0x48 IRQ Block User Interrupt Pending (0x48)
0x4C IRQ Block Channel Interrupt Pending (0x4C)
0x80 IRQ Block User Vector Number (0x80)
0x8C IRQ Block User Vector Number (0x8C)
PG195 (v4.1) November 24, 2023

Send Feedback
Table 78: IRQ Block Register Space (cont'd)

0xA0 IRQ Block Channel Vector Number (0xA0)
0xA4 IRQ Block Channel Vector Number (0xA4)
Interrupt processing registers are shared between AXI Bridge and AXI DMA. In AXI Bridge mode
when MSI-X Capabilities is selected, 64 KB address space from the BAR0 is reserved for the
MSI-X table. By default, register space is allocated in BAR0. You can select register space in a
different BAR, from BAR1 to BAR5, by using the CONFIG.bar_indicator {BAR0} Tcl
command. This option is valid only when MSI-X Capabilities option is selected. There is no
allocated space for other interrupt options.
IRQ Block Identifier (0x00)
Table 79: IRQ Block Identifier (0x00)

19:16 4’h2 RO IRQ Identifier
15:8 8’h0 RO Reserved
8'h01: 2015.3 and 2015.4
8'h02: 2016.1
8'h03: 2016.2
8'h04: 2016.3
8'h05: 2016.4
IRQ Block User Interrupt Enable Mask (0x04)
Table 80: IRQ Block User Interrupt Enable Mask (0x04)

[NUM_USR_INT-1:0] 'h0 RW user_int_enmask
User Interrupt Enable Mask
0: Prevents an interrupt from being generated when the
user interrupt source is asserted.
1: Generates an interrupt on the rising edge of the user
interrupt source. If the Enable Mask is set and the source
is already set, a user interrupt will be generated also.
PG195 (v4.1) November 24, 2023

Send Feedback
IRQ Block User Interrupt Enable Mask (0x08)
Table 81: IRQ Block User Interrupt Enable Mask (0x08)

W1S user_int_enmask
Bit descriptions are the same as in IRQ Block User Interrupt
Enable Mask (0x04).
IRQ Block User Interrupt Enable Mask (0x0C)
Table 82: IRQ Block User Interrupt Enable Mask (0x0C)

W1C user_int_enmask
Bit descriptions are the same as in IRQ Block User Interrupt
Enable Mask (0x04).
IRQ Block Channel Interrupt Enable Mask (0x10)
Table 83: IRQ Block Channel Interrupt Enable Mask (0x10)

[NUM_CHNL-1:0] ‘h0 RW channel_int_enmask
Engine Interrupt Enable Mask. One bit per read or write
engine.
0: Prevents an interrupt from being generated when
interrupt source is asserted. The position of the H2C bits
always starts at bit 0. The position of the C2H bits is the
index above the last H2C index, and therefore depends on
the NUM_H2C_CHNL parameter.
1: Generates an interrupt on the rising edge of the interrupt
source. If the enmask bit is set and the source is already set,
an interrupt is also be generated.

W1S channel_int_enmask
Bit descriptions are the same as in IRQ Block Channel
Interrupt Enable Mask (0x10).
PG195 (v4.1) November 24, 2023

Send Feedback

W1C channel_int_enmask
Bit descriptions are the same as in IRQ Block Channel
Interrupt Enable Mask (0x10).
The following figure shows the packing of H2C and C2H bits.
Figure 9: Packing H2C and C2H
Bits 7 6 5 4 3 2 1 0
4 H2C and 4 C2H enabled C2H_3 C2H_2 C2H_1 C2H_0 H2C_3 H2C_2 H2C_1 H2C_0
3 H2C and 3 C2H enabled X X C2H_2 C2H_1 C2H_0 H2C_2 H2C_1 H2C_0
1 H2C and 3 C2H enabled X X X X C2H_2 C2H_1 C2H_0 H2C_0
X15954-010115
IRQ Block User Interrupt Request (0x40)
Table 86: IRQ Block User Interrupt Request (0x40)

[NUM_USR_INT-1:0] ‘h0 RO user_int_req
User Interrupt Request
This register reflects the interrupt source AND’d with the
enable mask register.
IRQ Block Channel Interrupt Request (0x44)
Table 87: IRQ Block Channel Interrupt Request (0x44)

[NUM_CHNL-1:0] ‘h0 RO engine_int_req
Engine Interrupt Request. One bit per read or write engine.
This register reflects the interrupt source AND with the
enable mask register. The position of the H2C bits always
starts at bit 0. The position of the C2H bits is the index above
the last H2C index, and therefore depends on the
NUM_H2C_CHNL parameter. The previous figure shows the
packing of H2C and C2H bits.
PG195 (v4.1) November 24, 2023

Send Feedback
IRQ Block User Interrupt Pending (0x48)
Table 88: IRQ Block User Interrupt Pending (0x48)

[NUM_USR_INT-1:0] ‘h0 RO user_int_pend
User Interrupt Pending.
This register indicates pending events. The pending
events are cleared by removing the event cause condition
at the source component.
IRQ Block Channel Interrupt Pending (0x4C)
Table 89: IRQ Block Channel Interrupt Pending (0x4C)

[NUM_CHNL-1:0] ‘h0 RO engine_int_pend
Engine Interrupt Pending.
One bit per read or write engine. This register indicates
pending events. The pending events are cleared by
removing the event cause condition at the source
component. The position of the H2C bits always starts at bit
0. The position of the C2H bits is the index above the last
H2C index, and therefore depends on the NUM_H2C_CHNL
parameter.
The previous figure shows the packing of H2C and C2H bits.
IRQ Block User Vector Number (0x80)

If MSI is enabled, this register specifies the MSI or MSI-X vector number of the MSI. In legacy
interrupts, only the two LSBs of each field should be used to map to INTA, B, C, or D.
Table 90: IRQ Block User Vector Number (0x80)

28:24 5’h0 RW vector 3
The vector number that is used when an interrupt is
generated by the user IRQ usr_irq_req[3].
PG195 (v4.1) November 24, 2023

Send Feedback

interrupts, only the 2 LSB of each field should be used to map to INTA, B, C, or D.


interrupts only the 2 LSB of each field should be used to map to INTA, B, C, or D.

28:24 5’h0 RW vector 11
20:16 5’h0 RW vector 10
IRQ Block User Vector Number (0x8C)

interrupts only the 2 LSB of each field should be used to map to INTA, B, C, or D.
PG195 (v4.1) November 24, 2023

Send Feedback
Table 93: IRQ Block User Vector Number (0x8C)
28:24 5’h0 RW vector 15

20:16 5’h0 RW vector 14
IRQ Block Channel Vector Number (0xA0)

If MSI is enabled, this register specifies the MSI vector number of the MSI. In legacy interrupts,
only the 2 LSB of each field should be used to map to INTA, B, C, or D.
Similar to the other C2H/H2C bit packing clarification, see the previous figure. The first C2H
vector is after the last H2C vector. For example, if NUM_H2C_Channel = 1, then H2C0 vector is
at 0xA0, bits [4:0], and C2H Channel 0 vector is at 0xA0, bits [12:8]. If NUM_H2C_Channel = 4,
then H2C3 vector is at 0xA0 28:24, and C2H Channel 0 vector is at 0xA4, bits [4:0].
Table 94: IRQ Block Channel Vector Number (0xA0)

28:24 5’h0 RW vector3
generated by channel 3.
IRQ Block Channel Vector Number (0xA4)

If MSI is enabled, this register specifies the MSI vector number of the MSI. In legacy interrupts,
only the 2 LSB of each field should be used to map to INTA, B, C, or D.
PG195 (v4.1) November 24, 2023

Send Feedback
Similar to the other C2H/H2C bit packing clarification, see the previous figure. The first C2H
vector is after the last H2C vector. For example, if NUM_H2C_Channel = 1, then H2C0 vector is
at 0xA0, bits [4:0], and C2H Channel 0 vector is at 0xA0, bits [12:8].If NUM_H2C_Channel = 4,
then H2C3 vector is at 0xA0 28:24, and C2H Channel 0 vector is at 0xA4, bits [4:0].
Table 95: IRQ Block Channel Vector Number (0xA4)

Config Block Registers (0x3)

The Config Block registers are described in this section.
Table 96: Config Block Register Space

0x00 Config Block Identifier (0x00)
0x04 Config Block BusDev (0x04)
0x08 Config Block PCIE Max Payload Size (0x08)
0x0C Config Block PCIE Max Read Request Size (0x0C)
0x10 Config Block System ID (0x10)
0x14 Config Block MSI Enable (0x14)
0x18 Config Block PCIE Data Width (0x18)
0x1C Config PCIE Control (0x1C)
0x40 Config AXI User Max Payload Size (0x40)
0x44 Config AXI User Max Read Request Size (0x44)
0x60 Config Write Flush Timeout (0x60)
Config Block Identifier (0x00)
Table 97: Config Block Identifier (0x00)

PG195 (v4.1) November 24, 2023

Send Feedback
Table 97: Config Block Identifier (0x00) (cont'd)

19:16 4’h3 RO Config identifier
8'h01: 2015.3 and 2015.4
8'h02: 2016.1
8'h03: 2016.2
8'h04: 2016.3
8'h05: 2016.4
Config Block BusDev (0x04)
Table 98: Config Block BusDev (0x04)

[15:0] PCIe IP RO bus_dev
Bus, device, and function
Config Block PCIE Max Payload Size (0x08)
Table 99: Config Block PCIE Max Payload Size (0x08)

[2:0] PCIe IP RO pcie_max_payload
Maximum write payload size. This is the lesser of the PCIe IP
MPS and DMA/Bridge Subsystem for PCIe parameters.
3'b000: 128 bytes
3'b001: 256 bytes
3'b010: 512 bytes
3'b011: 1024 bytes
3'b100: 2048 bytes
3'b101: 4096 bytes
PG195 (v4.1) November 24, 2023

Send Feedback
Config Block PCIE Max Read Request Size (0x0C)
Table 100: Config Block PCIE Max Read Request Size (0x0C)

[2:0] PCIe IP RO pcie_max_read
Maximum read request size. This is the lesser of the PCIe IP
MRRS and DMA/Bridge Subsystem for PCIe parameters.
3'b000: 128 bytes
3'b001: 256 bytes
3'b010: 512 bytes
3'b011: 1024 bytes
3'b100: 2048 bytes
3'b101: 4096 bytes
Config Block System ID (0x10)
Table 101: Config Block System ID (0x10)

[15:0] 16’hff01 RO system_id
Core system ID
Config Block MSI Enable (0x14)
Table 102: Config Block MSI Enable (0x14)

[0] PCIe IP RO MSI_en
MSI Enable
[1] PCIe IP RO MSI-X Enable
Config Block PCIE Data Width (0x18)
Table 103: Config Block PCIE Data Width (0x18)

[2:0] C_DAT_WIDTH RO pcie_width
PCIe AXI4-Stream Width
0: 64 bits
1: 128 bits
2: 256 bits
3: 512 bits
PG195 (v4.1) November 24, 2023

Send Feedback
Config PCIE Control (0x1C)
Table 104: Config PCIE Control (0x1C)

[0] 1’b1 RW Relaxed Ordering
PCIe read request TLPs are generated with the relaxed
ordering bit set.
Config AXI User Max Payload Size (0x40)
Table 105: Config AXI User Max Payload Size (0x40)

6:4 3’h5 RO user_eff_payload
The actual maximum payload size issued to the user
application. This value might be lower than
user_prg_payload due to IP configuration or datapath width.
3'b000: 128 bytes
3'b001: 256 bytes
3'b010: 512 bytes
3'b011: 1024 bytes
3'b100: 2048 bytes
3'b101: 4096 bytes
3 Reserved
2:0 3’h5 RW user_prg_payload
The programmed maximum payload size issued to the user
application.
3'b000: 128 bytes
3'b001: 256 bytes
3'b010: 512 bytes
3'b011: 1024 bytes
3'b100: 2048 bytes
3'b101: 4096 bytes
Config AXI User Max Read Request Size (0x44)
Table 106: Config AXI User Max Read Request Size (0x44)

6:4 3’h5 RO user_eff_read
Maximum read request size issued to the user application.
This value may be lower than user_max_read due to PCIe
configuration or datapath width.
3'b000: 128 bytes
3'b001: 256 bytes
3'b010: 512 bytes
3'b011: 1024 bytes
3'b100: 2048 bytes
3'b101: 4096 bytes
PG195 (v4.1) November 24, 2023

Send Feedback
Table 106: Config AXI User Max Read Request Size (0x44) (cont'd)

3 Reserved
2:0 3’h5 RW user_prg_read
Maximum read request size issued to the user application.
3'b000: 128 bytes
3'b001: 256 bytes
3'b010: 512 bytes
3'b011: 1024 bytes
3'b100: 2048 bytes
3'b101: 4096 bytes
Config Write Flush Timeout (0x60)
Table 107: Config Write Flush Timeout (0x60)

4:0 5’h0 RW Write Flush Timeout
Applies to AXI4-Stream C2H channels. This register specifies
the number of clock cycles a channel waits for data before
flushing the write data it already received from PCIe. This
action closes the descriptor and generates a writeback. A
value of 0 disables the timeout. The timeout value in clocks
= 2value.
H2C SGDMA Registers (0x4)

Table 108: H2C SGDMA Registers (0x4)

0x00 H2C SGDMA Identifier (0x00)
0x80 H2C SGDMA Descriptor Low Address (0x80)
0x84 H2C SGDMA Descriptor High Address (0x84)
0x88 H2C SGDMA Descriptor Adjacent (0x88)
0x8C H2C SGDMA Descriptor Credits (0x8C)
H2C SGDMA Identifier (0x00)
Table 109: H2C SGDMA Identifier (0x00)

19:16 4’h4 RO H2C DMA Target
PG195 (v4.1) November 24, 2023

Send Feedback
Table 109: H2C SGDMA Identifier (0x00) (cont'd)
15 1’b0 RO Stream
8'h01: 2015.3 and 2015.4
8'h02: 2016.1
8'h03: 2016.2
8'h04: 2016.3
8'h05: 2016.4
H2C SGDMA Descriptor Low Address (0x80)
Table 110: H2C SGDMA Descriptor Low Address (0x80)

31:0 32’h0 RW dsc_adr[31:0]
Lower bits of start descriptor address. Dsc_adr[63:0] is the
first descriptor address that is fetched after the Control
register Run bit is set.
H2C SGDMA Descriptor High Address (0x84)
Table 111: H2C SGDMA Descriptor High Address (0x84)

31:0 32’h0 RW dsc_adr[63:32]
Upper bits of start descriptor address.
Dsc_adr[63:0] is the first descriptor address that is fetched
after the Control register Run bit is set.
H2C SGDMA Descriptor Adjacent (0x88)
Table 112: H2C SGDMA Descriptor Adjacent (0x88)

5:0 6’h0 RW dsc_adj[5:0]
Number of extra adjacent descriptors after the start
descriptor address.
PG195 (v4.1) November 24, 2023

Send Feedback
H2C SGDMA Descriptor Credits (0x8C)
Table 113: H2C SGDMA Descriptor Credits (0x8C)

9:0 10'h0 RW h2c_dsc_credit[9:0]
Writes to this register will add descriptor credits for the
channel. This register will only be used if it is enabled via the
channel's bits in the Descriptor Credit Mode register.
Credits are automatically cleared on the falling edge of the
channels Control register Run bit or if Descriptor Credit
Mode is disabled for the channel. The register can be read
to determine the number of current remaining credits for
the channel.
C2H SGDMA Registers (0x5)

The C2H SGDMA registers are described in this section.
Table 114: C2H SGDMA Registers (0x5)

0x00 C2H SGDMA Identifier (0x00)
0x80 C2H SGDMA Descriptor Low Address (0x80)
0x84 C2H SGDMA Descriptor High Address (0x84)
0x88 C2H SGDMA Descriptor Adjacent (0x88)
0x8C C2H SGDMA Descriptor Credits (0x8C)
C2H SGDMA Identifier (0x00)
Table 115: C2H SGDMA Identifier (0x00)

19:16 4’h5 RO C2H DMA Target
15 1’b0 RO Stream
8'h01: 2015.3 and 2015.4
8'h02: 2016.1
8'h03: 2016.2
8'h04: 2016.3
8'h05: 2016.4
PG195 (v4.1) November 24, 2023

Send Feedback
C2H SGDMA Descriptor Low Address (0x80)
Table 116: C2H SGDMA Descriptor Low Address (0x80)

31:0 32’h0 RW dsc_adr[31:0]
Lower bits of start descriptor address. Dsc_adr[63:0] is the
first descriptor address that is fetched after the Control
register Run bit is set.
C2H SGDMA Descriptor High Address (0x84)
Table 117: C2H SGDMA Descriptor High Address (0x84)

31:0 32’h0 RW dsc_adr[63:32]
Upper bits of start descriptor address.
Dsc_adr[63:0] is the first descriptor address that is fetched
after the Control register Run bit is set.
C2H SGDMA Descriptor Adjacent (0x88)
Table 118: C2H SGDMA Descriptor Adjacent (0x88)

5:0 6’h0 RW dsc_adj[5:0]
Number of extra adjacent descriptors after the start
descriptor address.
C2H SGDMA Descriptor Credits (0x8C)
Table 119: C2H SGDMA Descriptor Credits (0x8C)

9:0 10'h0 RW c2h_dsc_credit[9:0]
Writes to this register will add descriptor credits for the
channel. This register is only used if it is enabled through
the channel's bits in the Descriptor Credit Mode register.
Credits are automatically cleared on the falling edge of the
channels Control register Run bit or if Descriptor Credit
Mode is disabled for the channel. The register can be read
to determine the number of current remaining credits for
the channel.
PG195 (v4.1) November 24, 2023

Send Feedback
SGDMA Common Registers (0x6)

Table 120: SGDMA Common Registers (0x6)

0x00 SGDMA Identifier Registers (0x00)
0x10 SGDMA Descriptor Control Register (0x10)
0x20 SGDMA Descriptor Credit Mode Enable (0x20)
0x24 SG Descriptor Mode Enable Register (0x24)
0x28 SG Descriptor Mode Enable Register (0x28)
SGDMA Identifier Registers (0x00)
Table 121: SGDMA Identifier Registers (0x00)

19:16 4’h6 RO SGDMA Target
8'h01: 2015.3 and 2015.4
8'h02: 2016.1
8'h03: 2016.2
8'h04: 2016.3
8'h05: 2016.4
SGDMA Descriptor Control Register (0x10)
Table 122: SGDMA Descriptor Control Register (0x10)

19:16 4’h0 RW c2h_dsc_halt[3:0]
One bit per C2H channel. Set to one to halt descriptor
fetches for corresponding channel.
15:4 Reserved
3:0 4’h0 RW h2c_dsc_halt[3:0]
One bit per H2C channel. Set to one to halt descriptor
fetches for corresponding channel.
PG195 (v4.1) November 24, 2023

Send Feedback

W1S Bit descriptions are the same as in SGDMA Descriptor
Control Register (0x10).

W1C Bit descriptions are the same as in SGDMA Descriptor
Control Register (0x10).
SGDMA Descriptor Credit Mode Enable (0x20)
Table 125: SGDMA Descriptor Credit Mode Enable (0x20)

3:0 0x0 RW h2c_dsc_credit_enable [3:0]
One bit per H2C channel. Set to 1 to enable descriptor
crediting. For each channel, the descriptor fetch engine will
limit the descriptors fetched to the number of descriptor
credits it is given through writes to the channel's Descriptor
Credit Register.
15:4 Reserved
19:16 0x0 RW c2h_dsc_credit_enable [3:0]
One bit per C2H channel. Set to 1 to enable descriptor
crediting. For each channel, the descriptor fetch engine will
limit the descriptors fetched to the number of descriptor
credits it is given through writes to the channel's Descriptor
Credit Register.
SG Descriptor Mode Enable Register (0x24)
Table 126: SG Descriptor Mode Enable Register (0x24)

W1S Bit descriptions are the same as in SGDMA Descriptor Credit
Mode Enable (0x20).
PG195 (v4.1) November 24, 2023

Send Feedback
SG Descriptor Mode Enable Register (0x28)
Table 127: SG Descriptor Mode Enable Register (0x28)

W1C Bit descriptions are the same as in SGDMA Descriptor Credit
Mode Enable (0x20).
MSI-X Vector Table and PBA (0x8)

The MSI-X Vector table and PBA are described in the following table. MSI-X table offsets start at
0x8000. The table below shows two MSI-X vector entries (MSI-X table has 32 vector entries).
PBA address offsets start at 0x8FE0. Address offsets are fixed values.
Note: The MSI-X enable in configuration control register should be asserted before writing to MSI-X table.
If not, the MSI-X table will not work as expected.
Table 128: MSI-X Vector Table and PBA (0x00–0xFE0)
Byte Offset Bit Index Default Access Type Description

0x00 31:0 32’h0 RW MSIX_Vector0_Address[31:0]
MSI-X vector0 message lower address.
0x04 31:0 32’h0 RW MSIX_Vector0_Address[63:32]
MSI-X vector0 message upper address.
0x08 31:0 32’h0 RW MSIX_Vector0_Data[31:0]
MSI-X vector0 message data.
0x0C 31:0 32'hFFFFFFFF RW MSIX_Vector0_Control[31:0]
MSI-X vector0 control.
Bit Position:
31:1: Reserved.
0: Mask. When set to 1, this MSI-X vector is not
used to generate a message. When reset to 0, this
MSI-X Vector is used to generate a message.
0x1F0 31:0 32’h0 RW MSIX_Vector31_Address[31:0]
MSI-X vector31 message lower address.
0x1F4 31:0 32’h0 RW MSIX_Vector31_Address[63:32]
MSI-X vector31 message upper address.
0x1F8 31:0 32’h0 RW MSIX_Vector31_Data[31:0]
MSI-X vector31 message data.
0x1FC 31:0 32'hFFFFFFFF RW MSIX_Vector31_Control[31:0]
MSI-X vector31 control.
Bit Position:
31:1: Reserved.
0: Mask. When set to one, this MSI-X vector is not
used to generate a message. When reset to 0, this
MSI-X Vector is used to generate a message.
0xFE0 31:0 32’h0 RW Pending_Bit_Array[31:0]
MSI-X Pending Bit Array. There is one bit per
vector. Bit 0 corresponds to vector0, etc.
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 4: Designing with the Subsystem
Chapter 4
Designing with the Subsystem

This section includes guidelines and additional information to facilitate designing with the
subsystem.
Clocking and Resets

Clocking
The axi_aclk output is the clock used for all AXI interfaces and should drive all corresponding
AXI Interconnect aclk signals. axi_aclkis not a free running clock. This is a derived clock and
will be valid after signal axi_aresetn is de-asserted
Note: The axi_aclk output should not be used for the system clock for your design. The axi_aclk is
not a free-run clock output. As noted, axi_aclk may not be present at all times.
Resets
For the DMA/ Bridge Subsystem for PCIe in AXI Bridge mode, there is an optional
dma_bridge_resetn input pin which allows you to reset all internal Bridge engines and
registers as well as all AXI peripherals driven by axi_aresetn pin. When the following
parameter is set, dma_bridge_resetn does not need to be asserted during initial link up
operation because it will be done automatically by the IP. You must terminate all transactions
before asserting this pin. After being asserted, the pin must be kept asserted for a minimum
duration of at least equal to the Completion Timeout value (typically 50 ms) to clear any pending
transfer that may currently be queued in the data path. To set this parameter, type the following
command at the Tcl command line:
set_property -dict [list CONFIG.soft_reset_en {true} [get_ips <ip_name>]
For information about clocking and resets, see the applicable PCIe® integrated block product
guide:
• 7 Series FPGAs Integrated Block for PCI Express LogiCORE IP Product Guide (PG054)
• Virtex-7 FPGA Integrated Block for PCI Express LogiCORE IP Product Guide (PG023)
• UltraScale Devices Gen3 Integrated Block for PCI Express LogiCORE IP Product Guide (PG156)
PG195 (v4.1) November 24, 2023

Send Feedback
• UltraScale+ Devices Integrated Block for PCI Express LogiCORE IP Product Guide (PG213)
Tandem Configuration
Tandem Configuration features are available for the AMD DMA Subsystem for PCI Express® for
all AMD UltraScale™ and AMD UltraScale+™ device with PCIe hard blocks. Tandem
Configuration uses a two-stage methodology that enables the IP to meet the configuration time
requirements indicated in the PCI Express Specification. Multiple use cases are supported with
this technology:
• Tandem PROM: Load the single two-stage bitstream from the flash.
• Tandem PCIe: Load the first stage bitstream from flash, and deliver the second stage bitstream
over the PCIe link to the MCAP.
• Tandem with Field Updates: After a Tandem PROM (UltraScale only) or Tandem PCIe initial
configuration, update the entire user design while the PCIe link remains active. The update
region (floorplan) and design structure are predefined, and Tcl scripts are provided.
• Tandem PCIe + Dynamic Function eXchange: This is a more general case of Tandem
Configuration followed by Dynamic Function eXchange (DFX) of any size or number of
dynamic regions.
• Dynamic Function eXchange over PCIe: This is a standard configuration followed by DFX,
using the PCIe/MCAP as the delivery path of partial bitstreams.
For information on Dynamic Function eXchange, see the Vivado Design Suite User Guide: Dynamic
Function eXchange (UG909).
Customizing the Subsystem for Tandem

Configuration
UltraScale Devices
To enable any of the Tandem Configuration capabilities for AMD UltraScale™ devices, select the
appropriate AMD Vivado™ IP catalog option when customizing the subsystem. In the Basic tab:
1. Change the Mode to Advanced.

2. Change the Tandem Configuration or Dynamic Function eXchange option according to your
particular case:
• Tandem: For Tandem PROM, Tandem PCIe or Tandem PCIe + Dynamic Function eXchange
use cases.
• Tandem with Field Updates: ONLY for the predefined Field Updates use case.
PG195 (v4.1) November 24, 2023

Send Feedback
• DFX over PCIe: To enable the MCAP link for Dynamic Function eXchange, without
enabling Tandem Configuration.
Figure 10: Tandem Configuration or Dynamic Function eXchangeOptions for

UltraScale Devices
For complete information about Tandem Configuration, including required PCIe block locations,
design flow examples, requirements, restrictions and other considerations, see Tandem
Configuration in the UltraScale Devices Gen3 Integrated Block for PCI Express LogiCORE IP Product
Guide (PG156).
UltraScale+ Devices
To enable any of the Tandem Configuration capabilities for AMD UltraScale+™ devices, select the
appropriate IP catalog option when customizing the subsystem. In the Basic tab:
1. Change the Mode to Advanced.

2. Change the Tandem Configuration or Dynamic Function eXchange option according to your
particular case:
• Tandem PROM: For the Tandem PROM use case.
• Tandem PCIe: For Tandem PCIe or Tandem + Dynamic Function eXchange use cases.
• Tandem PCIe with Field Updates: ONLY for the predefined Field Updates use case.
• DFX over PCIe: To enable the MCAP link for Dynamic Function eXchange, without
enabling Tandem Configuration.
Figure 11: Tandem Configuration or Dynamic Function eXchange Option
PG195 (v4.1) November 24, 2023

Send Feedback
For complete information about Tandem Configuration, including required PCIe block locations,
design flow examples, requirements, restrictions and other considerations, see Tandem
Configuration in the UltraScale+ Devices Integrated Block for PCI Express LogiCORE IP Product Guide
(PG213).
Supported Devices
The DMA/Bridge Subsystem for PCIe® and AMD Vivado™ tool flow support implementations
targeting AMD reference boards and specific part/package combinations. Tandem configuration
supports the configurations found in the following tables.
UltraScale Devices
The following table lists the Tandem PROM/PCIe supported configurations for AMD UltraScale™
devices.
HDL Verilog Only

PCIe All configurations (max: X8Gen3)
Configuration
AMD Reference KCU105 Evaluation Board for AMD Kintex™ UltraScale™ FPGA
Board Support VCU108 Evaluation Board for AMD Virtex™ UltraScale™ FPGA
Part1 PCIe Block PCIe Reset Tandem Tandem with
Device Support Location Location Configuration Field Updates
XCKU025 PCIE_3_1_X0Y0 IOB_X1Y103 Production Production
Kintex UltraScale XCKU060 PCIE_3_1_X0Y0 IOB_X2Y103 Production Production
XCVU065 PCIE_3_1_X0Y0 IOB_X1Y103 Production Production
Virtex UltraScale XCVU125 PCIE_3_1_X0Y0 IOB_X1Y103 Production Production
Notes:
1. Only production silicon is officially supported. Bitstream generation is disabled for all engineering sample silicon
(ES2) devices.
PG195 (v4.1) November 24, 2023

Send Feedback
UltraScale+ Devices
The following table lists the Tandem PROM/PCIe supported configurations for UltraScale+
devices.
HDL Verilog Only

PCIe Configuration All configurations (max: X16Gen3 or X8Gen4)
AMD Reference KCU116 Evaluation Board for Kintex UltraScale+ FPGA

Board Support VCU118 Evaluation Board for Virtex UltraScale+ FPGA
Part1 PCIe Block Location Tandem Tandem PCIe with
Device Support
Configuration Field Updates
AU7P PCIE40E4_X0Y0 Not supported Not supported
AU10P PCIE40E4_X0Y0 Production Production
Artix UltraScale+ AU15P PCIE40E4_X0Y0 Production Production
KU3P PCIE40E4_X0Y0 Production Production
Kintex UltraScale+ KU11P PCIE40E4_X1Y0 Production Production
KU19P2 PCIE4CE4_X0Y0 Tandem PROM only Not supported
VU3P PCIE40E4_X1Y0 Production Production
VU19P PCIE4CE4_X0Y2 Production Production
Virtex UltraScale+ VU27P PCIE40E4_X0Y0 Production Production
PG195 (v4.1) November 24, 2023

Send Feedback
ZU3T PCIE40E4_X0Y0 Not supported Not supported

ZU4CG/EG/EV PCIE40E4_X0Y1 Production Production
AMD Zynq™
UltraScale+™ MPSoC
ZU11EG PCIE40E4_X1Y0 Production Production
ZU21DR PCIE40E4_X0Y0 Tandem PROM only Not supported
ZU25DR PCIE40E4_X0Y0 Not supported Not supported
AMD Zynq™
ZU39DR PCIE4CE4_X0Y0 Not supported Not supported
UltraScale+™ RFSoC2
ZU48DR PCIE4CE4_X0Y0 Tandem PROM only Not supported
Notes:
1. Only production silicon is officially supported. Bitstream generation is disabled for all engineering sample silicon (ES1,
ES2) devices.
2. Zynq RFSoC devices and Kintex UltraScale+ KU19P do not have MCAP-enabled PCIe block locations. Explicit support is
added to accommodate Tandem PROM configuration only for these devices.
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 5: Design Flow Steps
Chapter 5
Design Flow Steps

This section describes customizing and generating the subsystem, constraining the subsystem,
and the simulation, synthesis, and implementation steps that are specific to this IP subsystem.
More detailed information about the standard AMD Vivado™ design flows and the IP integrator
can be found in the following Vivado Design Suite user guides:
• Vivado Design Suite User Guide: Designing IP Subsystems using IP Integrator (UG994)
• Vivado Design Suite User Guide: Designing with IP (UG896)
• Vivado Design Suite User Guide: Getting Started (UG910)
• Vivado Design Suite User Guide: Logic Simulation (UG900)
Customizing and Generating the Subsystem

This section includes information about using AMD tools to customize and generate the
subsystem in the AMD Vivado™ Design Suite.
If you are customizing and generating the subsystem in the Vivado IP integrator, see the Vivado
Design Suite User Guide: Designing IP Subsystems using IP Integrator (UG994) for detailed
information. IP integrator might auto-compute certain configuration values when validating or
generating the design. To check whether the values do change, see the description of the
parameter in this chapter. To view the parameter value, run the validate_bd_design
command in the Tcl console.
You can customize the IP for use in your design by specifying values for the various parameters
associated with the IP subsystem using the following steps:
1. Select the IP from the IP catalog.

2. Double-click the selected IP or select the Customize IP command from the toolbar or right-
click menu.
For details, see the Vivado Design Suite User Guide: Designing with IP (UG896) and the Vivado
Design Suite User Guide: Getting Started (UG910).
Figures in this chapter are illustrations of the Vivado IDE. The layout depicted here might vary
from the current version.
PG195 (v4.1) November 24, 2023

Send Feedback
Basic Tab
The Basic tab for the DMA mode (Functional Mode option) is shown in the following figure.
Figure 12: Basic Tab for DMA Functional Mode
The options are defined as follows:
PG195 (v4.1) November 24, 2023

Send Feedback
• Functional Mode: Allows you to select between the following:
• DMA (DMA Subsystem for PCIe).

• AXI Bridge (AXI Bridge Subsystem for PCIe). The AXI Bridge option is valid only for AMD
UltraScale+™ devices. For details about PCIe Bridge mode operation, see AXI Bridge for PCI
Express Gen3 Subsystem Product Guide (PG194). This document covers DMA mode
operation only.
• Mode: Allows you to select the Basic or Advanced mode of the configuration of subsystem.
• Device /Port Type: Only PCI Express® Endpoint device mode is supported.
• PCIe Block Location: Selects from the available integrated blocks to enable generation of
location-specific constraint files and pinouts. This selection is used in the default example
design scripts. This option is not available if an AMD Development Board is selected.
• Lane Width: The subsystem requires the selection of the initial lane width. For supported lane
widths and link speeds, see the 7 Series FPGAs Integrated Block for PCI Express LogiCORE IP
Product Guide (PG054), Virtex-7 FPGA Integrated Block for PCI Express LogiCORE IP Product
Guide (PG023), UltraScale Devices Gen3 Integrated Block for PCI Express LogiCORE IP Product
Guide (PG156), or the UltraScale+ Devices Integrated Block for PCI Express LogiCORE IP Product
Guide (PG213) Higher link speed cores are capable of training to a lower link speed if
connected to a lower link speed capable device.
• Maximum Link Speed: The subsystem requires the selection of the PCIe Gen speed.
• Reference Clock Frequency: The default is 100 MHz, but 125 MHz and 250 MHz are also
supported.
• Reset Source: You can choose between User Reset and Phy ready.
• User reset comes from the PCIe core once link is established. When PCIe link goes down,
User Reset is asserted and XDMA goes to reset mode. And when the link comes back up,
User Reset is deasserted.
• When the Phy ready option is selected, XDMA is not affected by PCIe link status.
• GT DRP Clock Selection: Select either internal clock (default) or external clock.
• GT Selection, Enable GT Quad Selection: Select the Quad in which lane 0 is located.
• AXI Address Width: Currently, only 64-bit width is supported.
• AXI Data Width: Select 64, 128, 256-bit, or 512-bit (only for UltraScale+). The subsystem
allows you to select the Interface Width, as defined in the 7 Series FPGAs Integrated Block for
PCI Express LogiCORE IP Product Guide (PG054), Virtex-7 FPGA Integrated Block for PCI Express
LogiCORE IP Product Guide (PG023), UltraScale Devices Gen3 Integrated Block for PCI Express
LogiCORE IP Product Guide (PG156), or the UltraScale+ Devices Integrated Block for PCI Express
LogiCORE IP Product Guide (PG213)
PG195 (v4.1) November 24, 2023

Send Feedback
• AXI Clock Frequency: Select 62.5 MHz, 125 MHz or 250 MHz depending on the lane width/
speed.
• DMA Interface Option: Select AXI4 Memory Mapped and AXI4-Stream.
• AXI4-Lite Slave Interface: Select to enable the AXI4-Lite slave Interface.
• Data Protection: By default parity checking is disabled.
• When Check Parity is enabled, XDMA checks for parity on read data from the PCIe and
generates parity for write data to the PCIe.
• When Propagate Parity is enabled, XDMA propagates parity to the user AXI interface. The
user is responsible for checking and generating parity on the user AXI interface.
• Tandem Configuration or Dynamic Function eXchange: Select the tandem configuration or

Dynamic Function eXchange feature, if application to your design.
RECOMMENDED: AMD recommends that you select the correct GT starting quad before setting the link
rate and width. Selecting line rate and width prior to selecting GT quad can have adverse effects.
Related Information
Tandem Configuration
PCIe ID Tab
The PCIe ID tab is shown in the following figure.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 13: PCIe ID Tab
• Enable PCIe-ID Interface: By enabling this option, PCIe_ID port is given as an input port, you
are expected to connect to a proper value as desired.
For a description of these options, see the “Design Flow Steps” chapter in the respective product
guide listed below:
PG195 (v4.1) November 24, 2023

Send Feedback
PCIe BARs Tab

The PCIe BARs tab is shown in the following figure.
Figure 14: PCIe BARs Tab
• Base Address Register Overview: In Endpoint configuration, the core supports up to six 32-bit
BARs or three 64-bit BARs, and the Expansion read-only memory (ROM) BAR. BARs can be
one of two sizes:
• 32-bit BARs: The address space can be as small as 128 bytes or as large as 2 GB. Used for
DMA, AXI Lite Master or AXI Bridge Master.
• 64-bit BARs: The address space can be as small as 128 bytes or as large as 8 Exabytes.
Used for DMA, AXI Lite Master or AXI Bridge Master.
All BAR registers share these options.
IMPORTANT! The DMA requires a large amount of space to support functions and queues. By default,
64-bit BAR space is selected for the DMA BAR. This applies for PF and VF bars. You must calculate your
design needs first before selecting between 64-bit and 32-bit BAR space.
BAR selections are configurable. By default DMA is at BAR 0 (64 bit), AXI4-Lite Master is at BAR
2 (64 bit). These selections can be changed according to user needs.
PG195 (v4.1) November 24, 2023

Send Feedback
• BAR: Click the checkbox to enable the BAR. Deselect the checkbox to disable the BAR.
• Type: Select from DMA (by default in BAR0), AXI Lite Master (by default in BAR1, if enabled),
or AXI Bridge Master (by default in BAR2, if enabled). All other BARs, you can select between
AXI List Master and AXI Bridge Master. Expansion ROM can be enabled by selecting BAR6
For 64-bit BAR (default selection), DMA (by default in BAR0), AXI Lite Master (by default in
BAR2, if enabled), and AXI Bridge Master (by default in BAR4, if enabled). Expansion ROM
can be enabled by selection BAR6.
• DMA: DMA by default is assigned to BAR0 space and for all PFs. DMA option can be
selected in any available BAR (only one BAR can have DMA option). If you select DMA
Mailbox Management rather than DMA; however, DMA Mailbox Management will not
allow you to perform any DMA operations. After selecting the DMA Mailbox Management
option, the host has access to the extended Mailbox space. For details about this space, see
the QDMA_PF_MAILBOX (0x22400) register space.
• AXI Lite Master: Select the AXI Lite Master interface option for any BAR space. The Size,
scale and address translation are configurable.
• AXI Bridge Master: Select the AXI Bridge Master interface option for any BAR space. The
Size, scale and address translation are configurable.
• Expansion ROM: When enabled, this space is accessible on the AXI4-Lite Master. This is a
read-only space. The size, scale, and address translation are configurable.
• Size: The available Size range depends on the 32-bit or 64-bit bar selected. The DMA requires
256 KB of space, which is the fixed default selection. Other BAR size selections are available,
but must be specified.
• Scale: Select between Byte, Kilobytes and Megabytes.
• Value: The value assigned to the BAR based on the current selections.
Note: For best results, disable unused base address registers to conserve system resources. A base address
register is disabled by deselecting unused BARs in the Customize IP dialog box.
PCIe MISC Tab

The PCIe Miscellaneous tab is shown in the following figure.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 15: PCIe Misc Tab
• Number of User Interrupt Request: Up to 16 user interrupt requests can be selected.
• Legacy Interrupt Settings: Select one of the Legacy Interrupts: INTA, INTB, INTC, or INTD.
• MSI Capabilities: By default, MSI Capabilities is enabled, and 1 vector is enabled. You can
choose up to 32 vectors. In general, Linux uses only 1 vector for MSI. This option can be
disabled.
• MSI RX PIN EN: This option is valid only in AXI Bridge Root Port mode.
• MSI-X Capabilities: Select a MSI-X event. For more information, see MSI-X Vector Table and
PBA (0x8).
PG195 (v4.1) November 24, 2023

Send Feedback
• Finite Completion Credits: On systems which support finite completion credits, this option
can be enabled for better performance.
• Extended Tag Field: By default, 6-bit completion tags are used. For AMD UltraScale™ and
AMD Virtex™ 7 devices, the Extended Tag option gives 64 tags. For AMD UltraScale+™
devices, the Extended Tag option gives 256 tags. If the Extended Tag option is not selected,
DMA uses 32 tag for all devices.
• Configuration Extend Interface: PCIe extended interface can be selected for more
configuration space. When Configuration Extend Interface is selected, you are responsible for
adding logic to extend the interface to make it work properly.
• Configuration Management Interface: PCIe Configuration Management interface can be

brought to the top level when this options is selected.
• Link Status Register: By default, Enable Slot Clock Configuration is selected. This means that
the slot configuration bit is enabled in the link status register.
PCIe DMA Tab

The PCIe DMA tab is shown in the following figure.
Figure 16: PCIe DMA Tab
• Number of DMA Read Channels: Available selection is from 1 to 4.
• Number of DMA Write Channels: Available selection is from 1 to 4.
• Number of Request IDs for Read channel: Select the max number of outstanding request per
channel. Available selection is from 2 to 64.
PG195 (v4.1) November 24, 2023

Send Feedback
• Number of Request IDs for Write channel: Select max number of outstanding request per
channel. Available selection is from 2 to 32.
• Descriptor Bypass for Read (H2C): Available for all selected read channels. Each binary digits
corresponds to a channel. LSB corresponds to Channel 0. Value of one in bit position means
corresponding channels has Descriptor bypass enabled.
• Descriptor Bypass for Write (C2H): Available for all selected write channels. Each binary digits
corresponds to a channel. LSB corresponds to Channel 0. Value of one in bit position means
corresponding channels has Descriptor bypass enabled.
• AXI ID Width: The default is 4-bit wide. You can also select 2 bits.
• DMA Status port: DMA status ports are available for all channels.
Debug Options Tab

The Debug Options tab is shown in the following figure.
Figure 17: Debug Options Tab
• JTAG Debugger: This option enables JTAG debugging.
• LTSSM State Debug Logic: This option shows all the LTSSM state transitions that have been
made starting from link up.
PG195 (v4.1) November 24, 2023

Send Feedback
• In System IBERT: This option is used to check and see the eye diagram of the serial link at the
desired link speed. For more information on In System IBERT, refer to In-System IBERT
LogiCORE IP Product Guide (PG246).
IMPORTANT! This option is used mainly for hardware debug purposes. Simulations are not supported
when this option is used.
• Add Mark Debug Utility: This option adds predefined PCIe signals to with mark_debug
attribute so these signals can be added in ILA for debug purpose. Following is the list of
signals:
• m_axis_cq_tdata
• s_axis_cc_tdata
• s_axis_rq_tdata
• m_axis_rc_tdata
• m_axis_cq_tuser
• s_axis_cc_tuser
• m_axis_cq_tlast
• s_axis_rq_tlast
• m_axis_rc_tlast
• s_axis_cc_tlast
• pcie_cq_np_req
• pcie_cq_np_req_count
• s_axis_rq_tuser
• m_axis_rc_tuser
• m_axis_cq_tkeep
• s_axis_cc_tkeep
• s_axis_rq_tkeep
• m_axis_rc_tkeep
• m_axis_cq_tvalid
• s_axis_cc_tvalid
• s_axis_rq_tvalid
• m_axis_rc_tvalid
• m_axis_cq_tready
• s_axis_cc_tready
PG195 (v4.1) November 24, 2023

Send Feedback
• s_axis_rq_tready
• m_axis_rc_tready
• Enable Descrambler: This option integrates encrypted version of the descrambler module
inside the PCIe core, which is used to descrambler the PIPE data to/from PCIe integrated
block in Gen3/Gen4 link speed mode only.
• PCIe Debug Ports: With this option enabled, the following ports are available:
• cfg_negotiated_width: cfg_negotiated_width_o
• cfg_current_speed: cfg_current_speed_o
• cfg_ltssm_state: cfg_ltssm_state_o
• cfg_err_cor: cfg_err_cor_o
• cfg_err_fatal: cfg_err_fatal_o
• cfg_err_nonfatal: cfg_err_nonfatal_o
• cfg_local_error: cfg_local_error_o
• cfg_local_error_valid: cfg_local_error_valid_o
Shared Logic Tab

The Shared Logic tab for IP in an AMD UltraScale™ device is shown in the following figure.
Figure 18: Shared Logic (UltraScale Devices)
The Shared Logic tab for IP in an AMD UltraScale+™ device is shown in the following figure.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 19: Shared Logic (UltraScale+ Devices)
For a description of these options, see Chapter 4, “Design Flow Steps” in the respective product
guide listed below:
GT Settings Tab
The GT Settings tab is shown in the following figure.
Figure 20: GT Settings Tab
PG195 (v4.1) November 24, 2023

Send Feedback
For a description of these options, see Chapter 4, “Design Flow Steps” in the respective product
guide listed below:
Output Generation
For details, see Vivado Design Suite User Guide: Designing with IP (UG896).
Constraining the Subsystem

This section contains information about constraining the subsystem in the AMD Vivado™ Design
Suite.
Required Constraints
The DMA/Bridge Subsystem for PCI Express® requires the specification of timing and other
physical implementation constraints to meet specified performance requirements for PCI
Express. These constraints are provided in a Xilinx Design Constraints (XDC) file. Pinouts and
hierarchy names in the generated XDC correspond to the provided example design.
IMPORTANT! If the example design top file is not used, copy the IBUFDS_GTE3 (for UltraScale+
IBUFDS_GTE4) instance for the reference clock, IBUF Instance for sys_rst and also the location and
timing constraints associated with them into your local design top.
To achieve consistent implementation results, an XDC containing these original, unmodified

constraints must be used when a design is run through the AMD tools. For additional details on
the definition and use of an XDC or specific constraints, see Vivado Design Suite User Guide: Using
Constraints (UG903).
Constraints provided with the integrated block solution have been tested in hardware and
provide consistent results. Constraints can be modified, but modifications should only be made
with a thorough understanding of the effect of each constraint. Additionally, support is not
provided for designs that deviate from the provided constraints.
Device, Package, and Speed Grade Selections
The device selection portion of the XDC informs the implementation tools which part, package,
and speed grade to target for the design.
PG195 (v4.1) November 24, 2023

Send Feedback
IMPORTANT! Because Gen2 and Gen3 Integrated Block for PCIe cores are designed for specific part and
package combinations, this section should not be modified.
The device selection section always contains a part selection line, but can also contain part or
package-specific options. An example part selection line follows:
CONFIG PART = XCKU040-ffva1156-3-e-es1
Clock Frequencies, Clock Management, and Clock Placement
For detailed information about clock requirements, see the respective product guide listed below:
Banking
This section is not applicable for this IP subsystem.
Transceiver Placement
I/O Standard and Placement
Relocating the Integrated Block Core

By default, the IP core-level constraints lock block RAMs, transceivers, and the PCIe block to the
recommended location. To relocate these blocks, you must override the constraints for these
blocks in the XDC constraint file. To do so:
1. Copy the constraints for the block that needs to be overwritten from the core-level XDC
constraint file.
2. Place the constraints in the user XDC constraint file.
3. Update the constraints with the new location.
The user XDC constraints are usually scoped to the top-level of the design; therefore, ensure that
the cells referred by the constraints are still valid after copying and pasting them. Typically, you
need to update the module path with the full hierarchy name.
PG195 (v4.1) November 24, 2023

Send Feedback
Note: If there are locations that need to be swapped (that is, the new location is currently being occupied
by another module), there are two ways to do this.
• If there is a temporary location available, move the first module out of the way to a new temporary
location first. Then, move the second module to the location that was occupied by the first module.
Next, move the first module to the location of the second module. These steps can be done in XDC
constraint file.
• If there is no other location available to be used as a temporary location, use the reset_property
command from Tcl command window on the first module before relocating the second module to this
location. The reset_property command cannot be done in XDC constraint file and must be called
from the Tcl command file or typed directly into the Tcl Console.
Simulation
This section contains information about simulating IP in the AMD Vivado™ Design Suite.
For comprehensive information about AMD Vivado™ simulation components, as well as

information about using supported third-party tools, see the Vivado Design Suite User Guide: Logic
Simulation (UG900).
Basic Simulation
Simulation models for AXI-MM and AXI-ST options can be generates and simulated. These are
very basic simulation model options on which you can develop complicated designs.
AXI-MM Mode
The example design for the AXI4 Memory Mapped (AXI-MM) mode has 4 KB block RAM on the
user side, so data can be written to the block RAM and read from block RAM to the Host. The
first H2C transfer is started and the DMA reads data from the Host memory and writes to the
block RAM. Then, the C2H transfer is started and the DMA reads data from the block RAM and
writes to the Host memory. The original data is compared with the C2H write data.
H2C and C2H are setup with one descriptor each, and the total transfer size is 64 bytes.
AXI-ST Mode
The example design for the AXI4-Stream (AXI_ST) mode is a loopback design. On the user side
the H2C ports are looped back to the C2H ports. First, the C2H transfer is started and the C2H
DMA engine waits for data on the user side. Then, the H2C transfer is started and the DMA
engine reads data from the Host memory and writes to the user side. Because it is a loopback,
design data from H2C is directed to C2H and ends up in the host destination address.
H2C and C2H are setup with one descriptor each, and the total transfer size is 64 bytes.
PG195 (v4.1) November 24, 2023

Send Feedback
Interrupts are not used in AMD Vivado™ Design Suite simulations. Instead, descriptor completed
count register is polled to determine transfer complete.
Descriptor Bypass
Simulation models for the descriptor bypass mode is available only for channel 0. This design can
be expanded to support other channels.
PIPE Mode Simulation

The DMA/Bridge Subsystem for PCI Express® supports the PIPE mode simulation where the
PIPE interface of the core is connected to the PIPE interface of the link partner. This mode
increases the simulation speed.
Use the Enable PIPE Simulation option on the Basic page of the Customize IP dialog box to
enable PIPE mode simulation in the current AMD Vivado™ Design Suite solution example design,
in either Endpoint mode or Root Port mode. The External PIPE Interface signals are generated at
the core boundary for access to the external device. Enabling this feature also provides the
necessary hooks to use third-party PCI Express VIPs/BFMs instead of the Root Port model
provided with the example design. See also PIPE Mode Simulation Using Integrated Endpoint PCI
Express Block in Gen3 x8 and Gen2 x8 Configurations Application Note (XAPP1184).
The following tables describe the PIPE bus signals available at the top level of the core and their
corresponding mapping inside the EP core (pcie_top) PIPE signals.
IMPORTANT! The xil_sig2pipe.v file is delivered in the simulation directory, and the file replaces
phy_sig_gen.v . BFM/VIPs should interface with the xil_sig2pipe instance in board.v.
PIPE mode simulations are not supported for this core when VHDL is the selected target
language.
Table 131: Common In/Out Commands and Endpoint PIPE Signals Mappings
Endpoint PIPE Endpoint PIPE

In Commands Signals Mapping Out Commands Signals Mapping
common_commands_in[25:0] not used common_commands_out[0] pipe_clk1

common_commands_out[2:1] pipe_tx_rate_gt2
common_commands_out[3] pipe_tx_rcvr_det_gt
common_commands_out[6:4] pipe_tx_margin_gt
common_commands_out[7] pipe_tx_swing_gt
common_commands_out[8] pipe_tx_reset_gt
common_commands_out[9] pipe_tx_deemph_gt
PG195 (v4.1) November 24, 2023

Send Feedback
Table 131: Common In/Out Commands and Endpoint PIPE Signals Mappings (cont'd)

In Commands Signals Mapping Out Commands Signals Mapping
common_commands_out[16:10] not used3

Notes:
1. pipe_clk is an output clock based on the core configuration. For Gen1 rate, pipe_clk is 125 MHz. For Gen2 and
Gen3, pipe_clk is 250 MHz.
2. pipe_tx_rate_gt indicates the pipe rate (2’b00-Gen1, 2’b01-Gen2, and 2’b10-Gen3).
3. The functionality of this port has been deprecated and it can be left unconnected.
Table 132: Input/Output Bus with Endpoint PIPE Signals Mapping

Input Bus Output Bus
Signals Mapping Signals Mapping
pipe_rx_0_sigs[31:0] pipe_rx0_data_gt pipe_tx_0_sigs[31: 0] pipe_tx0_data_gt

pipe_rx_0_sigs[33:32] pipe_rx0_char_is_k_gt pipe_tx_0_sigs[33:32] pipe_tx0_char_is_k_gt
pipe_rx_0_sigs[34] pipe_rx0_elec_idle_gt pipe_tx_0_sigs[34] pipe_tx0_elec_idle_gt
pipe_rx_0_sigs[35] pipe_rx0_data_valid_gt pipe_tx_0_sigs[35] pipe_tx0_data_valid_gt
pipe_rx_0_sigs[36] pipe_rx0_start_block_gt pipe_tx_0_sigs[36] pipe_tx0_start_block_gt
pipe_rx_0_sigs[38:37] pipe_rx0_syncheader_gt pipe_tx_0_sigs[38:37] pipe_tx0_syncheader_gt
pipe_rx_0_sigs[83:39] not used pipe_tx_0_sigs[39] pipe_tx0_polarity_gt
pipe_tx_0_sigs[41:40] pipe_tx0_powerdown_gt
pipe_tx_0_sigs[69:42] not used1
Notes:
1. The functionality of this port has been deprecated and it can be left unconnected.
Parameters for Custom PIPE Simulation

For PIPE simulation, certain parameters are required, and might need to be manually set. These
required parameters are provided in the example design. When you generate an example design
from the Vivado IP catalog, all required parameters are set, and no additional action is required.
However, custom designs will require that you add the following parameters to your design test
bench file.
defparam board.AXI_PCIE_EP.xdma_0_i.inst.pcie4_ip_i.inst.PL_SIM_FAST_LINK_TRAINING=2'h3;
localparam EXT_PIPE_SIM = "TRUE";
defparam board.AXI_PCIE_EP.xdma_0_i.inst.pcie4_ip_i.inst.EXT_PIPE_SIM = EXT_PIPE_SIM;
defparam board.RP.pcie_4_0_rport.pcie_4_0_int_inst.EXT_PIPE_SIM = "TRUE";
defparam board.RP.EXT_PIPE_SIM = "TRUE";
PG195 (v4.1) November 24, 2023

Send Feedback
Synthesis and Implementation

For details about synthesis and implementation, see the Vivado Design Suite User Guide: Designing
with IP (UG896).
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 6: Example Design
Chapter 6
Example Design
This chapter contains information about the example designs provided in the AMD Vivado™
Design Suite.
Available Example Designs

The example designs are as follows:
• AXI4 Memory Mapped Default Example Design

• AXI4 Memory Mapped with PCIe to AXI4-Lite Master and PCIe to DMA Bypass Example
Design
• AXI4 Memory Mapped with AXI4-Lite Slave Interface Example Design
• AXI4-Stream Example Design
• AXI4 Memory Mapped with Descriptor Bypass Example
• Vivado IP Integrator-Based Example Design
• User IRQ Example Design
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4 Memory Mapped Default Example Design

The following figure shows the AXI4 Memory Mapped (AXI-MM) interface as the default design.
The example design gives 4 kilobytes (KB) block RAM on user design with AXI4 MM interface.
For H2C transfers, the DMA/Bridge Subsystem for PCI Express® reads data from host and writes
to block RAM in the user side. For C2H transfers, the DMA/Bridge Subsystem for PCI Express®
reads data from block RAM and writes to host memory. The example design from the IP catalog
has only 4 KB block RAM; you can regenerate the subsystem for larger block RAM size, if
wanted.
Figure 21: AXI-MM Default Example Design
FPGA
CQ
CC
PCIe AXI-MM Block

Host DMA RAM
IP*
RQ
RC
* may include wrapper as necessary

X15052-010115
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4 Memory Mapped with PCIe to AXI4-Lite Master

and PCIe to DMA Bypass Example Design
The following figure shows a system where the PCIe to AXI4-Lite Master (BAR0) and PCIe to
DMA Bypass (BAR2) are selected. 4K block RAM is connected to the PCIe to DMA Bypass
interfaces. The host can use DMA Bypass interface to read/write data to the user space using
the AXI4 MM interface. This interface bypasses DMA engines. The host can also use the PCIe to
AXI4-Lite Master (BAR0 address space) to write/read user logic. The example design connects
4K block RAM to the PCIe to AXI4-Lite Master interface so the user application can perform
read/writes.
Figure 22: AXI-MM Example with PCIe to DMA Bypass Interface and PCIe to AXI-Lite
Master Enabled
FPGA
CQ AXI4-Lite Master
Block
RAM
CC
PCIe DMA Block

Host DMA
IP* AXI-MM RAM
RQ
DMA bypass Block
RC RAM
AXI-MM

X15047-010115
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4 Memory Mapped with AXI4-Lite Slave Interface

Example Design
When the PCIe® to AXI4-Lite master and AXI4-Lite slave interface are enabled, the generated
example design (shown in the following figure) has a loopback from AXI4-Lite master to AXI4-
Lite slave. Typically, the user logic can use a AXI4-Lite slave interface to read/write DMA/Bridge
Subsystem for PCI Express® registers. With this example design, the host can use PCIe to AXI4-
Lite Master (BAR0 address space) and read/write DMA/Bridge Subsystem for PCI Express®
registers, which is the same as using PCIe to DMA (BAR1 address space). This example design
also shows PCIe to DMA bypass Interface (BAR2) enabled.
Figure 23: AXI-MM Example with AXI-Lite Slave Enabled
FPGA
AXI4-Lite Master
CQ
AXI4-Lite Slave
CC
PCIe DMA Block

Host DMA
IP* AXI-MM RAM
RQ
DMA bypass Block
RC RAM
AXI-MM

X15045-010115
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4-Stream Example Design

When the AXI4-Stream interface is enabled, each H2C streaming channels is looped back to C2H
channel. As shown in the following figure, the example design gives a loopback design for AXI4
streaming. The limitation is that you need to select an equal number of H2C and C2H channels
for proper operation. This example design also shows PCIe to DMA bypass interface and PCIe to
AXI-Lite Master selected.
Figure 24: AXI4-Stream Example with PCIe to DMA Bypass Interface and PCIe to AXI-
Lite Master Enabled
FPGA
AXI4-Lite Master Block
CQ RAM
CC DMA ST H2C
PCIe AXI-ST
Host DMA
IP*
RQ DMA ST C2H
DMA bypass Block

RC
AXI-MM RAM

X15046-010115
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4 Memory Mapped with Descriptor Bypass

Example
When Descriptor bypass mode is enabled, the user logic is responsible for making descriptors
and transferring them in descriptor bypass interface. The following figure shows AXI4 Memory
Mapped design with descriptor bypass mode enabled. You can select which channels will have
descriptor bypass mode. When Channel 0 of H2C and Channel 0 C2H are selected for Descriptor
bypass mode, the generated AMD Vivado™ example design has descriptor bypass ports of H2C0
and C2H0 connected to logic that will generate only one descriptor of 64 bytes. The user is
responsible for developing codes for other channels and expanding the descriptor itself.
The following figure shows the AXI-MM example with Descriptor Bypass Mode enabled.
Figure 25: AXI-MM Example With Descriptor Bypass Mode Enabled
FPGA
CQ
AXI-MM Block
RAM
CC
PCIe
Host DMA Des
IP* bypass
RQ
interface
Descriptor
RC

X17931-010115
PG195 (v4.1) November 24, 2023

Send Feedback
Vivado IP Integrator-Based Example Design

In addition to the RTL-based example designs, the IP also supports a AMD Vivado™ IP integrator-
based example design. To use the example design:
1. Create an IP integrator block diagram.

2. Open the IP integrator workspace, as shown in the following figure.
3. In order to add the DMA/Bridge IP to the canvas, search for DMA/Bridge (xdma) IP in the IP
catalog.
After adding the IP to the canvas, the green Designer Assistance information bar appears at
the top of the canvas.
4. Click Run Block Automation from the Designer Assistance information bar.
PG195 (v4.1) November 24, 2023

Send Feedback
This opens a Run Block Automation dialog box (shown in the following figure) which lists all
the IP currently in the design eligible for block automation (left pane), and any options
associated with a particular automation (right pane). In this case, there is only the XDMA IP in
the hierarchy in the left pane. The right pane has a description and options available. The
Options can be used to configure the IP as well as decide the level of automation for block
automation.
The Run Block Automation dialog box has an Automation Level option, which can be set to IP
Level or Subsystem Level.
• IP Level: When you select IP level automation, the Block Automation inserts the utility buffer
for the sys_clk input, connects the sys_rst_n input and pcie_mgt output interface for
the XDMA IP, as shown in the following figure.
PG195 (v4.1) November 24, 2023

Send Feedback
• Subsystem Level: When you select subsystem level automation, the Block Automation inserts
the necessary sub IPs on the canvas and makes the necessary connections. In addition to
connecting the sys_clk and sys_rst_n inputs it also connects the pcie_mgt output
interface and user_lnk_up, user_clk_heartbeat and user_resetn outputs. It inserts
the AXI interconnect to connect the Block Memory with the XDMA IP through the AXI BRAM
controller. The AXI interconnect has one master interface and multiple slave interfaces when
the AXI4-Lite master and AXI-MM Bypass interfaces are enabled in the Run Block Automation
dialog box. The block automation also inserts Block Memories and AXI BRAM Controllers
when the AXI4-Lite master and AXI-MM Bypass interfaces are enabled.
PG195 (v4.1) November 24, 2023

Send Feedback
User IRQ Example Design

The user IRQ example design enables the host to connect to the AXI4-Lite Master interface
along with the default DMA/Bridge Subsystem for PCI Express® example design. In the example
design, the User Interrupt generator module and an external block RAM is integrated on this
AXI4-Lite interface. The host can use this interface to generate the user IRQ by writing to the
register space of the User Interrupt generator module and can also read/write to the external 1K
block RAM. The following figure shows the example design.
The example design can be generated using the following Tcl command.
set_property -dict [list CONFIG.usr_irq_exdes {true}] [get_ips <ip_name>]
Figure 26: User IRQ Example Design
FPGA
RQ
AXI-MM Block
RAM
RC
User
PCIe Interrupt
Host DMA 0x0000 –
Endpoint 0x000F Generator
CQ
De-mux
AXI4-Lite Module
0x0800 –
CC 0x0BFF Block
RAM
(1K x 32)
X21787-052019
The register description is found in the following table.
Table 133: Example Design Registers
Register
Register Name Access Type Description
Offset
0x00 Scratch Pad RW Scratch Pad
PG195 (v4.1) November 24, 2023

Send Feedback
Table 133: Example Design Registers (cont'd)
Register
Register Name Access Type Description
Offset
User Memory Size connected to XDMA.
Memory size = (2[7:4]) ([3:0]Byte)
[7:4] – denotes the size in powers of 2.
0–1
1–2
2–4
…
8 – 256
0x04 DMA BRAM Size RO
9 – 512
[3:0] – denotes unit.
0 – Byte
1 – KB
2 – MB
3 – GB
For example, if the register value is 21, the size is 4
KB. If the register value is 91, the size is 512 KB.
0x08 Interrupt Control Register RW Interrupt control register (write 1 to generate
interrupt).
Interrupt Status register corresponding bit must be
1 (ready) to generate interrupt. Also, reset the
corresponding bit after ISR is served.
0x0C Interrupt Status Register RO Interrupt Status.
1: ready
0: Interrupt generation in progress
Note: In case of Legacy interrupt, the Interrupt Control Register (0x08) value for the corresponding
interrupt bit should only be cleared after the ISR is served as this can be used by the host to determine the
interrupt source.
Customizing and Generating the Example

Design
In the Customize IP dialog box, use the default parameter values for the IP example design.
After reviewing the IP parameters:
1. Right-click the component name.
PG195 (v4.1) November 24, 2023

Send Feedback
2. Select Open IP Example Design.
This opens an example design in a separate Vivado Project.
PG195 (v4.1) November 24, 2023

Send Feedback
Chapter 7: Test Bench
Chapter 7
Test Bench
This chapter contains information about the test bench provided in the AMD Vivado™ Design
Suite.
Root Port Model Test Bench for Endpoint

The PCI Express® Root Port Model is a basic test bench environment that provides a test
program interface that can be used with the provided PIO design or with your design. The
purpose of the Root Port Model is to provide a source mechanism for generating downstream
PCI Express TLP traffic to stimulate the customer design, and a destination mechanism for
receiving upstream PCI Express TLP traffic from the customer design in a simulation
environment. Source code for the Root Port Model is included to provide the model for a starting
point for your test bench. All the significant work for initializing the core configuration space,
creating TLP transactions, generating TLP logs, and providing an interface for creating and
verifying tests are complete, allowing you to dedicate efforts to verifying the correct
functionality of the design rather than spending time developing an Endpoint core test bench
infrastructure.
Source code for the Root Port Model is included to provide the model for a starting point for
your test bench. All the significant work for initializing the core configuration space, creating TLP
transactions, generating TLP logs, and providing an interface for creating and verifying tests are
complete, allowing you to dedicate efforts to verifying the correct functionality of the design
rather than spending time developing an Endpoint core testbench infrastructure.
The Root Port Model consists of:
• Test Programming Interface (TPI), which allows you to stimulate the Endpoint device for the
PCI Express.
• Example tests that illustrate how to use the test program TPI.
• Verilog source code for all Root Port Model components, which allow you to customize the
test bench.
The following figure shows the Root Port Module with DMA Subsystem for PCIe.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 27: Root Port Module with DMA Subsystem for PCIe
Root Port model
Output log Usrapp_com
Usrapp_rx Usrapp_tx
dsport
PCI Express fabric
Endpoint subsystem
X15051-010115
Architecture
The Root Port Model, illustrated in the previous figure, consists of these blocks:
• dsport (Root Port)

• usrapp_tx
• usrapp_rx
• usrapp_com (Verilog only)
The usrapp_tx and usrapp_rx blocks interface with the dsport block for transmission and
reception of TLPs to/from the EndPoint DUT. The Endpoint DUT consists of the DMA Subsystem
for PCIe.
The usrapp_tx block sends TLPs to the dsport block for transmission across the PCI Express
Link to the Endpoint DUT. In turn, the Endpoint DUT device transmits TLPs across the PCI
Express Link to the dsport block, which are subsequently passed to the usrapp_rx block. The
dsport and core are responsible for the data link layer and physical link layer processing when
communicating across the PCI Express logic. Both usrapp_tx and usrapp_rx utilize the
usrapp_com block for shared functions, for example, TLP processing and log file outputting.
PG195 (v4.1) November 24, 2023

Send Feedback
PIO write and read are initiated by usrapp_tx.
The DMA Subsystem for PCIe uses the 7 series Gen2 Integrated Block for PCIe, the 7 series
Gen3 Integrated Block for PCIe, the AMD UltraScale™ Devices Gen3 Integrate Block for PCIe,
and the AMD UltraScale+™ Devices Integrate Block for PCIe. See the “Test Bench” chapter in the
appropriate guide:
Test Case
The DMA Subsystem for PCIe can be configured as AXI4 Memory Mapped (AXI-MM) or AXI4-
Stream (AXI-ST) interface. The simulation test case reads configuration register to determine if a
AXI4 Memory Mapped or AXI4-Stream configuration. The test case, based on the AXI settings,
performs simulation for either configuration.
Table 134: Test Case Descriptions
Test Case Name Description

AXI4 Memory Mapped interface simulation. Reads data from host memory and writes to block
Dma_test0 RAM (H2C). Then, reads data from block RAM and write to host memory (C2H). The test case at
the end compares data for correctness.
AXI4-Stream interface simulation. Reads data from host memory and sends to AXI4-Stream user
Dma_stream0
interface (H2C), and the data is looped back to host memory (C2H).
Simulation
Simulation is set up to transfer one descriptor in H2C and one descriptor in C2H direction.
Transfer size is set to 128 bytes in each descriptor. For both AXI-MM and AXI-Stream, data is
read from Host and sent to Card (H2C). Then data is read from Card and sent to Host (C2H). Data
read from Card is compared with original data for data validity.
Limitations:
• Simulation does not support Interrupts. Test case just reads status and complete descriptor
count registers to decide if transfer is completed.
• Simulations are done only for Channel 0. In a future release, multi channels simulations will be
enabled.
• Transfer size is limited to 128 bytes and only one descriptor.
• Root port simulation model is not a complete BFM. Simulation supports one descriptor
transfer which shows a basic DMA procedure.
PG195 (v4.1) November 24, 2023

Send Feedback
• By default, post-synthesis simulation is not supported for the example design. To enable post-
synthesis simulation, generate the IP using the following Tcl command:
set_property -dict [list CONFIG.post_synth_sim_en {true}] [get_ips
<ip_name>]
Note: With this feature, functional simulation time increases to approximately 2.5 ms.
AXI4 Memory Mapped Interface

First, the test case starts the H2C engine. The H2C engine reads data from host memory and
writes to block RAM on user side. Then, the test case starts the C2H engine. The C2H engine
reads data from block RAM and writes to host memory. The following are the simulation steps:
1. The test case sets up one descriptor for the H2C engine.
2. The H2C descriptor is created in the Host memory. The H2C descriptor gives data length 128
bytes, source address (host), and destination address (Card).
3. The test case writes data (incremental 128 bytes of data) in the source address space.
4. The test case also sets up one descriptor for the C2H engine.
5. The C2H descriptor gives data length 128 bytes, source address (Card), and destination
address (host).
6. Write H2C descriptor starting address to register (0x4080 and 0x4084).
7. Write to H2C control register to start H2C transfer address (0x0004). Bit 0 is set to 1 to start
the transfer. For details of control register, refer to H2C Channel Control (0x04).
8. The DMA transfer takes the data host source address and sends to the block RAM
destination address.
9. The test case then starts the C2H transfer.
10. Write C2H descriptor starting address to register (0x5080 and0x5084).
11. Write to C2H control register to start the C2H transfer (0x1004). Bit 0 is set to 1 to start the
transfer. For details of control the register, see C2H Channel Control (0x04).
12. The DMA transfer takes data from the block RAM source address and sends data to the host
13. The test case then compares the data for correctness.
14. The test case checks for the H2C and C2H descriptor completed count (value of 1).
15. The test case then disables transfer by deactivating the Run bit (bit0) in the Control registers
(0x0004 and 0x1004) for the H2C and C2H engines.
PG195 (v4.1) November 24, 2023

Send Feedback
AXI4-Stream Interface
For AXI4-Stream, the example design is a loopback design. Channel H2C_0 is looped back to
C2H_0 (and so on) for all other channels. First, the test case starts the C2H engine. The C2H
engine waits for data that is transmitted by the H2C engine. Then, the test case starts the H2C
engine. The H2C engine reads data from host and sends to the Card, which is looped back to the
C2H engine. The C2H engine then takes the data, and writes back to host memory. The following
are the simulation steps:
1. The test case sets up one descriptor for the H2C engine.
2. The H2C descriptor is created in the Host memory. The H2C descriptor gives the data length
128 bytes, Source address (host), and Destination address (Card).
3. The test case writes data (incremental 128 bytes of data) in the Host source address space.
4. The test case also sets up one descriptor for the C2H engine in Host memory.
5. The C2H descriptor gives data length 128 bytes, source address (Card), and destination
address (host).
6. Write C2H descriptor starting address to register (0x5080 and 0x5084).
7. Write to the C2H control register to start the C2H transfer first.
8. The C2H engine starts and waits for data to come from the H2C ports.
9. Write H2C descriptor starting address to register (0x4080 and 0x4084).
10. Write to the H2C control register to start H2C transfer.
11. The H2C engine takes data from the host source address to the Card destination address.
12. The data is looped back to the C2H engine.
13. The C2H engine read data from the Card and writes it back to the Host memory destination
address.
15. The test case then compares the data for correctness.
16. The test case then disables transfer by deactivating the Run bit (bit 0) in the Control registers
0x0004 and 0x1004 for the H2C and C2H engines.
Descriptor Bypass Mode

Simulation for Descriptor bypass mode is possible when Channel 0 of both H2C and C2H are
selected for descriptor bypass option. The example design generated has one descriptor ready to
pump in the Descriptor bypass mode interface.
PG195 (v4.1) November 24, 2023

Send Feedback
AXI-MM Descriptor Bypass Mode Simulation

1. The example design has a predefined descriptor for the H2C and C2H engine.
2. The H2C descriptor has 128 bytes of data, source address (Host) and destination address
(Card).
3. The C2H descriptor has 128 bytes of data, source address (Card) and destination address
(Host).
4. The test case writes incremental 128 bytes of data to the Host memory source address.
5. The PIO writes to the H2C engine Control register to start the transfer (0x0004).
6. The DMA reads data from the Host address and sends it to the Card block RAM destination
address.
7. The PIO writes to the C2H engine Control register to start the transfer (0x1004).
8. The DMA reads data from the Card block RAM source address and sends it to the Host
9. The test case compares data for correctness.
11. The test case then disables the transfer by deasserting the Run bit (bit 0) in the Control
register for the H2C and C2H engine (0x0004 and 0x1004).
AXI-Stream Descriptor Bypass Mode Simulation with Loopback

Design
1. The example design has a predefined descriptor for the H2C and C2H engine.
2. The H2C descriptor has 128 bytes of data, source address (Host) and destination address
(Card).
3. The C2H descriptor has 128 bytes of data, source address (Card) and destination address
(Host).
4. The test case writes incremental 128 bytes of data to Host memory source address.
5. The PIO writes to the C2H engine Control register to start the transfer (0x1004).
6. The C2H engine starts the DMA transfer but waits for data (loopback design).
7. The PIO writes to the H2C engine Control register to start the transfer (0x0004).
8. The H2C engine reads data from the Host address and sends it to Card.
9. The data is looped back to the C2H engine.
10. The C2H engine reads data from the Card and sends it to the Host destination address.
11. The test case compares data for correctness.
PG195 (v4.1) November 24, 2023

Send Feedback
13. The test case then disables the transfer by deasserting the Run bit (bit 0) in the Control
register for the H2C and C2H engine (0x0004 and 0x1004).
When the transfer is started, one H2C and one C2H descriptor are transferred in Descriptor
bypass interface and after that DMA transfers are performed as explained in above section.
Descriptor is setup for 64 bytes transfer only.
Simulation Updates
Following is an overview of how existing root port tasks can be modified to exercise multi-
channels, and multi descriptor cases.
Multi-Channels Simulation, Example Channel 1 H2C and C2H

1. Create an H2C Channel 1 descriptor in the Host memory address that is different than the
H2C and C2H Channel 0 descriptor.
2. Create a C2H Channel 1 descriptor in the Host memory address that is different than the
H2C and C2H Channel 0 and H2C Channel 1 descriptor.
3. Create transfer data (128 Bytes) for the H2C Channel 1 transfer in the Host memory which
does not overwrite any of the 4 descriptors in the Host memory (H2C and C2H Channel 0
and Channel 1 descriptors), and H2C Channel 0 data.
4. Also make sure the H2C data in the Host memory does not overlap the C2H data transfer
space for both C2H Channel 0 and 1.
5. Write the descriptor starting address to H2C Channel 0 and 1. Write to H2C SGDMA
Descriptor Low Address (0x80) and H2C SGDMA Descriptor High Address (0x84) registers.
6. Write the descriptor starting address to C2H Channel 0 and 1. Write to C2H SGDMA
Descriptor Low Address (0x80) and H2C SGDMA Descriptor High Address (0x84) registers.
7. Enable multi-channel transfer by writing to control register (bit 0) of H2C Channel 0 and 1.
Write to H2C Channel Control (0x04) register for Channel 0 and Channel 1.
8. Enable multi-channel transfer by writing to control register (bit 0) of C2H Channel 0 and 1.
Write to C2H Channel Control (0x04) register for Channel 0 and Channel 1.
9. Compare the data for correctness.
The same procedure applies for AXI4-Stream configuration. Refer to the above section for
detailed explanation of the AXI4-Stream transfer.
Multi Descriptor Simulation

1. Create a transfer of 256 bytes data (incremental or any data). Split the data into two 128
bytes of data section. First, the data starts at address S1, and second, 128 bytes starts at
address S2.
2. Create a new descriptor (named DSC_H2C_1) in the Host memory address at DSC1.
PG195 (v4.1) November 24, 2023

Send Feedback
3. The DSC_H2C_1 descriptor has 128 bytes for DMA transfer, Host address S1 (source) and
destination address D1 (card).
4. Create a new descriptor (named DSC_H2C_2) in the Host memory at address DSC2 that is
different from DSC_H2C_1 Descriptor.
5. The DSC_H2C_2 descriptor has 128 bytes for DMA transfer, Host address S2 (source) and
destination address D2 (card).
6. Link these two descriptors by adding next descriptor address in DSC_H2C_1. Write DSC2 in
next descriptor field.
7. Wire the descriptor starting address to H2C Channel 0.
8. Enable DMA transfer for H2C Channel 0 by writing the Run bit in Control register 0x0004.
Test Tasks
Table 135: Test Tasks
Name Description
TSK_INIT_DATA_H2C This task generates one descriptor for H2C engine and initializes source data in
host memory.
TSK_INIT_DATA_C2H This task generates one descriptor for C2H engine.
TSK_XDMA_REG_READ This task reads the DMA Subsystem for PCIe register.
TSK_XDMA_REG_WRITE This task writes the DMA Subsystem for PCIe register.
COMPARE_DATA_H2C This task compares source data in the host memory to destination data written
to block RAM. This task is used in AXI4 Memory Mapped simulation.
COMPARE_DATA_C2H This task compares the original data in the host memory to the data C2H engine
writing to host. This task is used in AXI4 Memory Mapped simulation.
TSK_XDMA_FIND_BAR This task finds XDMA configuration space between different enabled BARs (BAR0
to BAR6).
For other PCIe-related tasks, see the “Test Bench” chapter in the 7 Series FPGAs Integrated Block
for PCI Express LogiCORE IP Product Guide (PG054), Virtex-7 FPGA Integrated Block for PCI Express
LogiCORE IP Product Guide (PG023), UltraScale Devices Gen3 Integrated Block for PCI Express
LogiCORE IP Product Guide (PG156), or UltraScale+ Devices Integrated Block for PCI Express
LogiCORE IP Product Guide (PG213).
PG195 (v4.1) November 24, 2023

Send Feedback
Appendix A: GT Locations
Appendix A
GT Locations
For more information on GT Locations, see GT Locations appendix in UltraScale+ Devices
Integrated Block for PCI Express LogiCORE IP Product Guide (PG213).
PG195 (v4.1) November 24, 2023

Send Feedback
Appendix B: Application Software Development
Appendix B
Application Software Development

This section provides details about the Linux device driver that is provided with the core.
Device Drivers
Figure 28: Device Drivers
Test App
User Space
Kernel Space
DMA Driver
X.86 Linux Host
PCIe
AMD Device
(DMA Example Design)
Linux Kernel Driver

Usage model
X28828-110923
The above figure shows the usage model of Linux XDMA software drivers. The DMA/Bridge
Subsystem for PCIe example design is implemented on an AMD FPGA , which is connected to an
X86 host through PCI Express. In this mode, the XDMA driver in kernel space runs on Linux,
whereas the test application runs in user space.
Linux Device Driver

The Linux device driver has the following character device interfaces:
• User character device for access to user components.
PG195 (v4.1) November 24, 2023

Send Feedback
• Control character device for controlling DMA/Bridge Subsystem for PCI Express®
components.
• Events character device for waiting for interrupt events.
• SGDMA character devices for high performance transfers.
The user accessible devices are as follows:
• XDMA0_control: Used to access DMA/Bridge Subsystem for PCI Express® registers.
• XDMA0_user: Used to access AXI-Lite master interface.
• XDMA0_bypass: Used to access DMA Bypass interface.
• XDMA0_events_*: Used to recognize user interrupts.
Using the Driver

The XDMA drivers can be downloaded from the DMA IP Drivers page.
Interrupt Processing
Legacy Interrupts
There are four types of legacy interrupts: A, B, C and D. You can select any interrupts in the PCIe
Misc tab under Legacy Interrupt Settings. You must program the corresponding values for both
the IRQ Block Channel Vector (see IRQ Block Channel Vector Number (0xA0)) and the IRQ Block
User Vector (see IRQ Block User Vector Number (0x80)). Values for each legacy interrupts are A
= 0, B = 1, C = 2 and D = 3. The host recognizes interrupts only based on these values.
MSI Interrupts
For MSI interrupts, you can select from 1 to 32 vectors in the PCIe Misc tab under MSI
Capabilities, which consists of a maximum of 16 usable DMA interrupt vectors and a maximum of
16 usable user interrupt vectors. The Linux operating system (OS) supports only 1 vector. Other
operating systems might support more vectors and you can program different vectors values in
the IRQ Block Channel Vector (see IRQ Block Channel Vector Number (0xA0)) and in the IRQ
Block User Vector (see IRQ Block User Vector Number (0x80)) to represent different interrupt
sources. The AMD Linux driver supports only 1 MSI vector.
PG195 (v4.1) November 24, 2023

Send Feedback
MSI-X Interrupts
The DMA supports up to 32 different interrupt source for MSI-X, which consists of a maximum
of 16 usable DMA interrupt vectors and a maximum of 16 usable user interrupt vectors. The
DMA has 32 MSI-X tables, one for each source (see MSI-X Vector Table and PBA (0x00–0xFE0)).
For MSI-X channel interrupt processing the driver should use the Engine’s Interrupt Enable Mask
for H2C and C2H (see H2C Channel Interrupt Enable Mask (0x90) or Table C2H Channel
Interrupt Enable Mask (0x90)) to disable and enable interrupts.
User Interrupts
The user logic must hold usr_irq_req active-High even after receiving usr_irq_ack (acks)
to keep the interrupt pending register asserted. This enables the Interrupt Service Routine (ISR)
within the driver to determine the source of the interrupt. Once the driver receives user
interrupts, the driver or software can reset the user interrupts source to which hardware should
respond by deasserting usr_irq_req.
Example H2C Flow

In the example H2C flow, loaddriver.sh loads devices for all available channels. The
dma_to_device user program transfers data from host to Card.
The example H2C flow sequence is as follows:
1. Open the H2C device and initialize the DMA.

2. The user program reads the data file, allocates a buffer pointer, and passes the pointer to
write function with the specific device (H2C) and data size.
3. The driver creates a descriptor based on input data/size and initializes the DMA with
descriptor start address, and if there are any adjacent descriptor.
4. The driver writes a control register to start the DMA transfer.
5. The DMA reads descriptor from the host and starts processing each descriptor.
6. The DMA fetches data from the host and sends the data to the user side. After all data is
transferred based on the settings, the DMA generates an interrupt to the host.
7. The ISR driver processes the interrupt to find out which engine is sending the interrupt and
checks the status to see if there are any errors. It also checks how many descriptors are
processed.
8. After the status is good, the drive returns transfer byte length to user side so it can check for
the same.
PG195 (v4.1) November 24, 2023

Send Feedback
Example C2H Flow

In the example C2H flow, loaddriver.sh loads the devices for all available channels. The
dma_from_device user program transfers data from Card to host.
The example C2H flow sequence is as follows:
1. Open device C2H and initialize the DMA.

2. The user program allocates buffer pointer (based on size), passes pointer to read function
with specific device (C2H) and data size.
3. The driver creates descriptor based on size and initializes the DMA with descriptor start
address. Also if there are any adjacent descriptor.
4. The driver writes control register to start the DMA transfer.
5. The DMA reads descriptor from host and starts processing each descriptor.
6. The DMA fetches data from Card and sends data to host. After all data is transferred based
on the settings, the DMA generates an interrupt to host.
7. The ISR driver processes the interrupt to find out which engine is sending the interrupt and
checks the status to see if there are any errors and also checks how many descriptors are
processed.
8. After the status is good, the drive returns transfer byte length to user side so it can check for
the same.
PG195 (v4.1) November 24, 2023

Send Feedback
Appendix C: Upgrading
Appendix C
Upgrading
This appendix contains information about upgrading to a more recent version of the IP.
New Parameters
The following new parameters are added in the IP in the current release.
Table 136: New Parameters
Name Display Name Description Default Value

shared_logic_gtc_7xG2 Include Shared Logic (Transceiver When selected, includes False
GT_COMMON) in example design GT_COMMON block in example
design.
shared_logic_clk_7xG2 Include Shared Logic (Clocking) in When selected, includes Clock False
example design block in example design.
shared_logic_both_7xG2 Include Shared Logic in core When selected, includes both False
GT_COMMON and Clock blocks in
core.
ecc_en Enable ECC Enables ECC. Requires one of the False
Parity option to be turned on.
aspm_support ASPM Support optionality Indicates ASPM support is Enabled No_ASPM
or Disabled.
New Ports
The ports in the following table appear at the boundary when the Internal Shared GT_COMMON
and Clocking option is selected in the Shared Logic tab for 7 series Gen2 devices.
Table 137: Ports For Shared Logic (Internal Shared GT_COMMON and Clocking Option)
Name Direction Width

int_dclk_out O 1 bit
int_oobclk_out O 1 bit
int_pclk_sel_slave I 1 bit
int_pclk_out_slave O 1 bit
PG195 (v4.1) November 24, 2023

Send Feedback
Table 137: Ports For Shared Logic (Internal Shared GT_COMMON and Clocking Option)
(cont'd)

int_pipe_rxusrclk_out O 1 bit
int_qplllock_out O 2 bits
int_qplloutclk_out O 2 bits
int_qplloutrefclk_out O 2 bits
int_rxoutclk_out O 1 bit
int_userclk1_out O 1 bit
int_userclk2_out O 1 bit
The ports in the following table appear at the boundary when the Shared GT_COMMON option is
selected in the Share Logic tab for 7 series Gen2 devices.
Table 138: Ports For Shared Logic (Shared GT_COMMON Option)

qpll_drp_crscode I 12 bits
qpll_drp_fsm I 18 bits
qpll_drp_done I 2 bits
qpll_drp_reset I 2 bits
qpll_qplllock I 2 bits
qpll_qplloutclk I 2 bits
qpll_qplloutrefclk I 2 bits
qpll_qplld O 1 bit
qpll_qpllreset O 2 bits
qpll_drp_clk O 1 bit
qpll_drp_rst_n O 1 bit
qpll_drp_ovrd O 1 bit
qpll_drp_gen3 O 1 bit
qpll_drp_start O 1 bit
The ports in the following table appear at the boundary when the Shared Clocking option is
selected in the Share Logic tab for 7 series Gen2 devices.
Table 139: Ports For Shared Logic (Shared Clocking Option)

pipe_pclk_in I 1 bit
pipe_rxusrclk_in I 1 bit
pipe_rxoutclk_in I 1 bit
pipe_dclk_in I 1 bit
PG195 (v4.1) November 24, 2023

Send Feedback
Table 139: Ports For Shared Logic (Shared Clocking Option) (cont'd)

pipe_userclk1_in I 1 bit
pipe_userclk2_in I 1 bit
pipe_oobclk_in I 1 bit
pipe_mmcm_lock_in I 1 bit
pipe_mmcm_rst_n I 1 bit
pipe_txoutclk_out O 1 bit
pipe_rxoutclk_out O 1 bit
pipe_pclk_sel_out O 1 bit
pipe_gen3_out O 1 bit
The following table shows the new port added in this version of the IP. This port is available at
the boundary when MSI-X feature is enabled and the device type is PCIe Endpoint.
Table 140: New Port

msix_en O 1 bit
PG195 (v4.1) November 24, 2023

Send Feedback
Appendix D: Debugging
Appendix D
Debugging
This appendix includes details about resources available on the AMD Support website and
debugging tools.
Finding Help with AMD Adaptive Computing

Solutions
To help in the design and debug process when using the subsystem, the Support web page
contains key resources such as product documentation, release notes, answer records,
information about known issues, and links for obtaining further product support. The Community
Forums are also available where members can learn, participate, share, and ask questions about
AMD Adaptive Computing solutions.
Documentation
This product guide is the main document associated with the subsystem. This guide, along with
documentation related to all products that aid in the design process, can be found on the Support
web page or by using the AMD Adaptive Computing Documentation Navigator. Download the
Documentation Navigator from the Downloads page. For more information about this tool and
the features available, open the online help after installation.
Debug Guide
For more information on PCIe debug, see PCIe Debug K-Map.
Answer Records
Answer Records include information about commonly encountered problems, helpful information
on how to resolve these problems, and any known issues with an AMD Adaptive Computing
product. Answer Records are created and maintained daily to ensure that users have access to
the most accurate information available.
Answer Records for this subsystem can be located by using the Search Support box on the main
Support web page. To maximize your search results, use keywords such as:
PG195 (v4.1) November 24, 2023

Send Feedback
• Product name
• Tool message(s)
• Summary of the issue encountered
A filter search is available after results are returned to further target the results.
Master Answer Record for the Subsystem

AR 65443
Technical Support
AMD Adaptive Computing provides technical support on the Community Forums for this AMD
LogiCORE™ IP product when used as described in the product documentation. AMD Adaptive
Computing cannot guarantee timing, functionality, or support if you do any of the following:
• Implement the solution in devices that are not defined in the documentation.
• Customize the solution beyond that allowed in the product documentation.
• Change any section of the design labeled DO NOT MODIFY.
To ask questions, navigate to the Community Forums.
Debug Tools
There are many tools available to address DMA/Bridge Subsystem for PCIe design issues. It is
important to know which tools are useful for debugging various situations.
Vivado Design Suite Debug Feature

The AMD Vivado™ Design Suite debug feature inserts logic analyzer and virtual I/O cores
directly into your design. The debug feature also allows you to set trigger conditions to capture
application and integrated block port signals in hardware. Captured signals can then be analyzed.
This feature in the Vivado IDE is used for logic debugging and validation of a design running in
AMD devices.
The Vivado logic analyzer is used to interact with the logic debug LogiCORE IP cores, including:
• ILA 2.0 (and later versions)

• VIO 2.0 (and later versions)
See the Vivado Design Suite User Guide: Programming and Debugging (UG908).
PG195 (v4.1) November 24, 2023

Send Feedback
Reference Boards
Various AMD development boards support the DMA/Bridge Subsystem for PCIe core. These
boards can be used to prototype designs and establish that the core can communicate with the
system.
• 7 series FPGA evaluation boards

○ VC709
○ KC705
• AMD UltraScale™ FPGA Evaluation boards

○ KCU105
○ VCU108
• AMD UltraScale+™
○ KCU116
○ VCU118
○ ZCU106
Hardware Debug
Hardware issues can range from link bring-up to problems seen after hours of testing. This
section provides debug steps for common issues. The AMD Vivado™ debug feature is a valuable
resource to use in hardware debug. The signal names mentioned in the following individual
sections can be probed using the debug feature for debugging the specific problems.
General Checks
Ensure that all the timing constraints for the core were properly incorporated from the example
design and that all constraints were met during implementation.
• Does it work in post-place and route timing simulation? If problems are seen in hardware but
not in timing simulation, this could indicate a PCB issue. Ensure that all clock sources are
active and clean.
• If using MMCMs in the design, ensure that all MMCMs have obtained lock by monitoring the
locked port.
• If your outputs go to 0, check your licensing.
PG195 (v4.1) November 24, 2023

Send Feedback
Initial Debug of the DMA/Bridge Subsystem for PCIe

Status bits out of each engine can be used for initial debug of the subsystem. Per channel
interface provides important status to the user application.
Table 141: Initial Debug of the Subsystem
Bit Index Field Description

7 NA Reserved
6 Run Channel control register run bit.
5 IRQ_Pending Asserted when the channel has interrupt pending.
4 Packet_Done On an AXIST interface this bit indicates the last data indicated by the
EOP bit has been posted.
3 Descriptor_Done A descriptor has finished transferring data from the source and
posted it to the destination.
2 Descriptor_Stop Descriptor_Done and Stop bit set in the descriptor.
1 Descriptor_Completed Descriptor_Done and Completed bit set in the descriptor.
0 Busy Channel descriptor buffer is not empty or DMA requests are
outstanding.
Read H2C/C2H Channel status register 0x40 to see if there are any errors registered. Also check
if bit [0] busy bit is set to 1. If it is set to 1, then DMA is waiting for some user event to happen.
Read H2C/C2H Channel Completed Descriptor Count (0x48) to see how many descriptors are
fetched by the DMA, and compare it with what it should be, based on the DMA transfer. If MSIX
interrupt mode transfers have issues, try in poll mode to see if transfers go through.
Software Debug
Using AMD driver can help you debug some issues. Compile the driver in DEBUG mode which
enables verbose mode and prints more information for a transfer. Check the dmesg for a run to
see all the steps are listed for a dma transfer.
Descriptor used for a transfer is printed so check the descriptor to make sure source/destination
address is listed correctly. Check the transfer length for that descriptor is correct. Make sure that
the descriptor magic word is correct. Follow the dmesg log to see if there are any errors
recorded.
ILA Debug
You can add ILA on the output and input of the IP on the AXI side to see if any abnormal
transfers are happening. You can check if the packet sent into the IP is proper and it matches
with what is expected. You can add ILA in the PCIe core CQ/CC and RQ/RC interface, this lets
you know how the DMA IP is fetching the descriptor and data. You need to use all the above
suggestions to figure out the issue.
PG195 (v4.1) November 24, 2023

Send Feedback
Related Information
Channel 0-3 Status Ports
PG195 (v4.1) November 24, 2023

Send Feedback
Appendix E: Using the Xilinx Virtual Cable to Debug
Appendix E
Using the Xilinx Virtual Cable to

Debug
The Xilinx Virtual Cable (XVC) allows the AMD Vivado™ Design Suite to connect to FPGA debug
cores through non-JTAG interfaces. The standard AMD Vivado™ Design Suite debug feature uses
JTAG to connect to physical hardware FPGA resources and perform debug through Vivado. This
section focuses on using XVC to perform debug over a PCIe® link rather than the standard JTAG
debug interface. This is referred to as XVC-over-PCIe and allows for Vivado ILA waveform
capture, VIO debug control, and interaction with other AMD debug cores using the PCIe link as
the communication channel.
XVC-over-PCIe should be used to perform FPGA debug remotely using the Vivado Design Suite
debug feature when JTAG debug is not available. This is commonly used for data center
applications where the FPGA is connected to a PCIe Host system without any other connections
to the hardware device.
Using debug over XVC requires software, driver, and FPGA hardware design components. Since
there is an FPGA hardware design component to XVC-over-PCIe debug, you cannot perform
debug until the FPGA is already loaded with an FPGA hardware design that implements XVC-
over-PCIe and the PCIe link to the Host PC is established. This is normally accomplished by
loading an XVC-over-PCIe enabled design into the configuration flash on the board prior to
inserting the card into the data center location. Since debug using XVC-over-PCIe is dependent
on the PCIe communication channel this should not be used to debug PCIe link related issue.
IMPORTANT! XVC only provides connectivity to the debug cores within the FPGA. It does not provide the
ability to program the device or access device JTAG and configuration registers. These operations can be
performed through other standard AMD interfaces or peripherals such as the PCIe MCAP VSEC and
HWICAP IP.
Overview
The main components that enable XVC-over-PCIe debug are as follows:
• Host PC XVC-Server application

• Host PC PCIe-XVC driver
PG195 (v4.1) November 24, 2023

Send Feedback
• XVC-over-PCIe enabled FPGA design
These components are provided as a reference on how to create XVC connectivity for AMD
FPGA designs. These three components are shown in the following figure and connect to the
Vivado Design Suite debug feature through a TCP/IP socket.
Figure 29: XVC-over-PCIe Software and Hardware Components
Vivado Design TCP/IP

PCIe-XVC
PCIe
XVC-over-PCIe
Suite Debug XVC-Server
Driver FPGA Design
Feature
Running on local or remote Running on Host PC connected to AMD FPGA card Running on AMD FPGA
Host
X18837-032119
Host PC XVC-Server Application

The hw_server application is launched by Vivado Design Suite when using the debug feature.
Through the Vivado IDE you can connect hw_server to local or remote FPGA targets. This same
interface is used to connect to local or remote PCIe-XVC targets as well. The Host PCIe XVC-
Server application connects to the AMD hw_server using TCP/IP socket. This allows Vivado
(using hw_server) and the XVC-Server application to be running on the same PC or separate
PCs connected through Ethernet. The XVC-Server application needs to be run on a PC that is
directly connected to the FPGA hardware resource. In this scenario the FPGA hardware is
connected through PCIe® to a Host PC. The XVC-Server application connects to the FPGA
hardware device through the PCIe-XVC driver that is also running on the Host PC.
Host PC XVC-over-PCIe Driver

The XVC-over-PCIe driver provides connectivity to the PCIe enabled FPGA hardware resource
that is connected to the Host PC. As such this is provided as a Linux kernel mode driver to access
the PCIe hardware device, which is available in the following location,
<Vivado_Installation_Path>/data/xicom/driver/pcie/xvc_pcie.zip. The
necessary components of this driver must be added to the driver that is created for a specific
FPGA platform. The driver implements the basic functions needed by the XVC-Server application
to communicate with the FPGA via PCIe.
PG195 (v4.1) November 24, 2023

Send Feedback
XVC-over-PCIe Enabled FPGA Design

Traditionally AMD Vivado™ debug is performed over JTAG. By default, Vivado tool automation
connects the AMD debug cores to the JTAG BSCAN resource within the FPGA to perform
debug. In order to perform XVC-over-PCIe debug, this information must be transmitted over the
PCIe link rather than over the JTAG cable interface. The AMD Debug Bridge IP allows you to
connect the debug network to PCIe through either the PCIe extended configuration interface
(PCIe-XVC-VSEC) or through a PCIe BAR via an AXI4-Lite Memory Mapped interface (AXI-XVC).
The Debug Bridge IP, when configured for From PCIe to BSCAN or From AXI to BSCAN,
provides a connection point for the AMD debug network from either the PCIe Extended
Capability or AXI4-Lite interfaces respectively. Vivado tool automation connects this instance of
the Debug Bridge to the AMD debug cores found in the design rather than connecting them to
the JTAG BSCAN interface. There are design trade-offs to connecting the debug bridge to the
PCIe Extended Configuration Space or AXI4-Lite. The following sections describe the
implementation considerations and register map for both implementations.
XVC-over-PCIe Through PCIe Extended Configuration

Space (PCIe-XVC-VSEC)
Using the PCIe-XVC-VSEC approach, the Debug Bridge IP uses a PCIe Vendor Specific Extended
Capability (VSEC) to implement the connection from PCIe to the Debug Bridge IP. The PCIe
extended configuration space is set up as a linked list of extended capabilities that are
discoverable from a Host PC. This is specifically valuable for platforms where one version of the
design implements the PCIe-XVC-VSEC and another design implementation does not. The linked
list can be used to detect the existence or absence of the PCIe-XVC-VSEC and respond
accordingly.
The PCIe Extended Configuration Interface uses PCIe configuration transactions rather than PCIe
memory BAR transactions. While PCIe configuration transactions are much slower, they do not
interfere with PCIe memory BAR transactions at the PCIe IP boundary. This allows for separate
data and debug communication paths within the FPGA. This is ideal if you expect to debug the
datapath. Even if the datapath becomes corrupt or halted, the PCIe Extended Configuration
Interface can remain operational to perform debug. The following figure describes the
connectivity between the PCIe IP and the Debug Bridge IP to implement the PCIe-XVC-VSEC.
PG195 (v4.1) November 24, 2023

Send Feedback
Figure 30: XVC-over-PCIe with PCIe Extended Capability Interface
Note: Although the previous figure shows the AMD UltraScale+™ Devices Integrated Block for PCIe IP,
other PCIe IP (that is, the AMD UltraScale™ Devices Integrated Block for PCIe, AXI Bridge for PCIe, or PCIe
DMA IP) can be used interchangeably in this diagram.
XVC-over-PCIe Through AXI (AXI-XVC)

Using the AXI-XVC approach, the Debug Bridge IP connects to the PCIe IP through an AXI
Interconnect IP. The Debug Bridge IP connects to the AXI Interconnect like other AXI4-Lite Slave
IPs and similarly requires that a specific address range be assigned to it. Traditionally the
debug_bridge IP in this configuration is connected to the control path network rather than the
system datapath network. The following figure describes the connectivity between the DMA
Subsystem for PCIe IP and the Debug Bridge IP for this implementation.
Figure 31: XVC over PCIe with AXI4-Lite Interface
Note: Although the previous figure shows the PCIe DMA IP, any AXI-enabled PCIe IP can be used
interchangeably in this diagram.
The AXI-XVC implementation allows for higher speed transactions. However, XVC debug traffic
passes through the same PCIe ports and interconnect as other PCIe control path traffic, making it
more difficult to debug transactions along this path. As result the AXI-XVC debug should be used
to debug a specific peripheral or a different AXI network rather than attempting to debug
datapaths that overlap with the AXI-XVC debug communication path.
PG195 (v4.1) November 24, 2023

Send Feedback
XVC-over-PCIe Register Map
The PCIe-XVC-VSEC and AXI-XVC have a slightly different register map that must be taken into
account when designing XVC drivers and software. The register maps in the following tables
show the byte-offset from the base address.
• The PCIe-XVC-VSEC base address must fall within the valid range of the PCIe Extended
Configuration space. This is specified in the Debug Bridge IP configuration.
• The base address of an AXI-XVC Debug Bridge is the offset for the Debug Bridge IP peripheral
that was specified in the Vivado Address Editor.
The following tables describe the register map for the Debug Bridge IP as an offset from the base
address when configured for the From PCIe-Ext to BSCAN or From AXI to BSCAN modes.
Table 142: Debug Bridge for XVC-PCIe-VSEC Register Map
Register
Register Name Description Register Type
Offset
0x00 PCIe Ext Capability Header PCIe defined fields for VSEC use. Read Only
0x04 PCIe VSEC Header PCIe defined fields for VSEC use. Read Only
0x08 XVC Version Register IP version and capabilities information. Read Only
0x0C XVC Shift Length Register Shift length. Read Write
0x10 XVC TMS Register TMS data. Read Write
0x14 XVC TDIO Register TDO/TDI data. Read Write
0x18 XVC Control Register General control register. Read Write
0x1C XVC Status Register General status register. Read Only
Table 143: Debug Bridge for AXI-XVC Register Map
Register
Register Name Description Register Type
Offset
0x00 XVC Shift Length Register Shift length. Read Write
0x04 XVC TMS Register TMS data. Read Write
0x08 XVC TDI Register TDI data. Read Write
0x0C XVC TDO Register TDO data. Read Only
0x10 XVC Control Register General control register. Read Write
0x14 XVC Status Register General status register. Read Only
0x18 XVC Version Register IP version and capabilities information. Read Only
PG195 (v4.1) November 24, 2023

Send Feedback
PCIe Ext Capability Header

This register is used to identify the PCIe-XVC-VSEC added to a PCIe design. The fields and values
in the PCIe Ext Capability Header are defined by PCI-SIG and are used to identify the format of
the extended capability and provide a pointer to the next extended capability, if applicable.
When used as a PCIe-XVC-VSEC, the appropriate PCIe ID fields should be evaluated prior to
interpretation. These can include PCIe Vendor ID, PCIe Device ID, PCIe Revision ID, Subsystem
Vendor ID, and Subsystem ID. The provided drivers specifically check for a PCIe Vendor ID that
matches AMD (0x10EE) before interpreting this register. The following table describes the fields
within this register.
Table 144: PCIe Ext Capability Header Register Description
Bit Initial
Field Description Type
Location Value
PCIe
This field is a PCI-SIG defined ID number that indicates the nature
Extended
15:0 and format of the Extended Capability. The Extended Capability ID 0x000B Read Only
Capability
for a VSEC is 0x000B
ID
This field is a PCI-SIG defined version number that indicates the
Capability
19:16 version of the capability structure present. Must be 0x1 for this 0x1 Read Only
Version
version of the specification.
This field is passed in from the user and contains the offset to the
next PCI Express Capability structure or 0x000 if no other items
Next
exist in the linked list of capabilities. For Extended Capabilities
31:20 Capability 0x000 Read Only
implemented in the PCIe extended configuration space, this value
Offset
must always be within the valid range of the PCIe Extended
Configuration space.
PCIe VSEC Header (PCIe-XVC-VSEC only)

This register is used to identify the PCIe-XVC-VSEC when the Debug Bridge IP is in this mode.
The fields are defined by PCI-SIG, but the values are specific to the Vendor ID (0x10EE for AMD).
The PCIe Ext Capability Header register values should be qualified prior to interpreting this
register.
Table 145: PCIe XVC VSEC Header Register Description
Bit Initial
Field Description Type
Location Value
This field is the ID value that can be used to identify the PCIe-XVC-
15:0 VSEC ID 0x0008 Read Only
VSEC and is specific to the Vendor ID (0x10EE for AMD).
This field is the Revision ID value that can be used to identify the
19:16 VSEC Rev 0x0 Read Only
PCIe-XVC-VSEC revision.
This field indicates the number of bytes in the entire PCIe-XVC-
VSEC
31:20 VSEC structure, including the PCIe Ext Capability Header and PCIe 0x020 Read Only
Length
VSEC Header registers.
PG195 (v4.1) November 24, 2023

Send Feedback
XVC Version Register (PCIe-XVC-VSEC only)

This register is populated by the AMD tools and is used by the Vivado Design Suite to identify
the specific features of the Debug Bridge IP that is implemented in the hardware design.
XVC Shift Length Register
This register is used to set the scan chain shift length within the debug scan chain.
XVC TMS Register
This register is used to set the TMS data within the debug scan chain.
XVC TDO/TDI Data Register(s)
This register is used for TDO/TDI data access. When using PCIePCI-XVC-VSEC, these two
registers are combined into a single field. When using AXI-XVC, these are implemented as two
separate registers.
XVC Control Register
This register is used for XVC control data.
XVC Status Register

This register is used for XVC status information.
XVC Driver and Software

Example XVC driver and software has been provided with the Vivado Design Suite installation,
which is available at the following location: <Vivado_Installation_Path>/data/xicom/
driver/pcie/xvc_pcie.zip. This should be used for reference when integrating the XVC
capability into AMD FPGA platform design drivers and software. The provided Linux kernel mode
driver and software implement XVC-over-PCIe debug for both PCIe-XVC-VSEC and AXI-XVC
debug bridge implementations.
When operating in PCIe-XVC-VSEC mode, the driver will initiate PCIe configuration transactions
to interface with the FPGA debug network. When operating in AXI-XVC mode, the driver will
initiate 32-bit PCIe Memory BAR transactions to interface with the FPGA debug network. By
default, the driver will attempt to discover the PCIe-XVC-VSEC and use AXI-XVC if the PCIe-
XVC-VSEC is not found in the PCIe configuration extended capability linked list.
PG195 (v4.1) November 24, 2023

Send Feedback
The driver is provided in the data directory of the Vivado installation as a .zip file. This .zip
file should be copied to the Host PC connected through PCIe to the AMD FPGA and extracted
for use. README.txt files have been included; review these files for instructions on installing
and running the XVC drivers and software.
Special Considerations for Tandem or Dynamic

Function eXchange Designs
Tandem Configuration and Dynamic Function eXchange (DFX) designs may require additional
considerations as these flows partition the physical resources into separate regions. These
physical partitions should be considered when adding debug IPs to a design, such as VIO, ILA,
MDM, and MIG-IP. A Debug Bridge IP configured for From PCIe-ext to BSCAN or From AXI to
BSCAN should only be placed into the static partition of the design. When debug IPs are used
inside of a DFX or Tandem Field Updates region, an additional debug BSCAN interface should be
added to the dynamic region module definition and left unconnected in the dynamic region
module instantiation.
To add the BSCAN interface to the Reconfigurable Partition definition the appropriate ports and
port attributes should be added to the Reconfigurable Partition definition. The sample Verilog
provided below can be used as a template for adding the BSCAN interface to the port
declaration.
...
// BSCAN interface definition and attributes.
// This interface should be added to the DFX module definition
// and left unconnected in the DFX module instantiation.
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN drck" *)
(* DEBUG="true" *)
input S_BSCAN_drck,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN shift" *)
(* DEBUG="true" *)
input S_BSCAN_shift,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN tdi" *)
(* DEBUG="true" *)
input S_BSCAN_tdi,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN update" *)
(* DEBUG="true" *)
input S_BSCAN_update,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN sel" *)
(* DEBUG="true" *)
input S_BSCAN_sel,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN tdo" *)
(* DEBUG="true" *)
output S_BSCAN_tdo,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN tms" *)
(* DEBUG="true" *)
input S_BSCAN_tms,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN tck" *)
(* DEBUG="true" *)
input S_BSCAN_tck,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN runtest" *)
(* DEBUG="true" *)
input S_BSCAN_runtest,
PG195 (v4.1) November 24, 2023

Send Feedback
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN reset" *)

(* DEBUG="true" *)
input S_BSCAN_reset,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN capture" *)
(* DEBUG="true" *)
input S_BSCAN_capture,
(* X_INTERFACE_INFO = "xilinx.com:interface:bscan:1.0 S_BSCAN bscanid_en" *)
(* DEBUG="true" *)
input S_BSCAN_bscanid_en,
....
When link_design is run, the exposed ports are connected to the static portion of the debug
network through tool automation. The ILAs are also connected to the debug network as required
by the design. There might also be an additional dbg_hub cell that is added at the top level of
the design. For Tandem PCIe with Field Updates designs, the dbg_hub and tool inserted clock
buffer(s) must be added to the appropriate design partition. The following is an example of the
Tcl commands that can be run after opt_design to associate the dbg_hub primitives with the
appropriate design partitions.
# Add the inserted dbg_hub cell to the appropriate design partition.

set_property HD.TANDEM_IP_PBLOCK Stage1_Main [get_cells dbg_hub]
# Add the clock buffer to the appropriate design partition.
set_property HD.TANDEM_IP_PBLOCK Stage1_Config_IO [get_cells
dma_pcie_0_support_i/
pcie_ext_cap_i/vsec_xvc_inst/vsec_xvc_dbg_bridge_inst/inst/bsip/ins
t/USE_SOFTBSCAN.U_TAP_TCKBUFG]
Using the PCIe-XVC-VSEC Example Design

The PCIe-XVC-VSEC has been integrated into the PCIe example design as part of the Advanced
settings for the AMD UltraScale+™ Integrated Block for PCIe IP. This section provides instruction
of how to generate the PCIe example design with the PCIe-XVC-VSEC, and then debug the
FPGA through PCIe using provided XVC drivers and software. This is an example for using XVC in
customer applications. The FPGA design, driver, and software elements will need to be
integrated into customer designs.
Generating a PCIe-XVC-VSEC Example Design
The PCIe-XVC-VSEC can be added to the AMD UltraScale+™ PCIe example design by selecting
the following options.
1. Configure the core to the desired configuration.

2. On the Basic tab, select the Advanced Mode.
PG195 (v4.1) November 24, 2023

Send Feedback
3. On the Adv. Options-3 tab:

a. Select the PCI Express Extended Configuration Space Enable checkbox to enable the PCI
Express extended configuration interface. This is where additional extended capabilities
can be added to the PCI Express core.
b. Select the Add the PCIe-XVC-VSEC to the Example Design checkbox to enable the PCIe-
XVC-VSEC in the example design generation.
4. Verify the other configuration selections for the PCIe IP. The following selections are needed
to configure the driver for your hardware implementation:
• PCIe Vendor ID (0x10EE for AMD)
• PCIe Device ID (dependent on user selection)
5. Click OK to finalize the selection and generate the IP.
6. Generate the output products for the IP as desired for your application.
7. In the Sources window, right-click the IP and select Open IP Example Design.
8. Select a directory for generating the example design, and select OK.
After being generated, the example design shows that:
• the PCIe IP is connected to xvc_vsec within the support wrapper, and
• an ILA IP is added to the user application portion of the design.
This demonstrates the desired connectivity for the hardware portion of the FPGA design.
Additional debug cores can be added as required by your application.
Note: Although the previous figure shows to the UltraScale+ Devices Integrated Block for PCIe IP, the
example design hierarchy is the same for other PCIe IPs.
PG195 (v4.1) November 24, 2023

Send Feedback
9. Double-click the Debug Bridge IP identified as xvc_vsec to view the configuration option
for this IP. Make note of the following configuration parameters because they will be used to
configure the driver.
• PCIe XVC VSEC ID (default 0x0008)
• PCIe XVC VSEC Rev ID (default 0x0)
IMPORTANT! Do not modify these parameter values when using an AMD Vendor ID or provided XVC
drivers and software. These values are used to detect the XVC extended capability. (See the PCIe
specification for additional details.)
10. In the Flow Navigator, click Generate Bitstream to generate a bitstream for the example
design project. This bitstream will be then be loaded onto the FPGA board to enable XVC
debug over PCIe.
After the XVC-over-PCIe hardware design has been completed, an appropriate XVC enabled
PCIe driver and associated XVC-Server software application can be used to connect the Vivado
Design Suite to the PCIe connected FPGA. Vivado can connect to an XVC-Server application that
is running local on the same Machine or remotely on another machine using a TCP/IP socket.
System Bring-Up
The first step is to program the FPGA and power on the system such that the PCIe link is
detected by the host system. This can be accomplished by either:
• Programming the design file into the flash present on the FPGA board, or
• Programming the device directly via JTAG.
If the card is powered by the Host PC, it will need to be powered on to perform this
programming using JTAG and then re-started to allow the PCIe link to enumerate. After the
system is up and running, you can use the Linux lspci utility to list out the details for the
FPGA-based PCIe device.
Compiling and Loading the Driver

The provided PCIe drivers and software should be customized to a specific platform. To
accomplish this, drivers and software are normally developed to verify the Vendor ID, Device ID,
Revision ID, Subsystem Vendor ID, and Subsystem ID before attempting to access device-
extended capabilities or peripherals like the PCIe-XVC-VSEC or AXI-XVC. Because the provided
driver is generic, it only verifies the Vendor ID and Device ID for compatibility before attempting
to identify the PCIe-XVC-VSEC or AXI-XVC peripheral.
The XVC driver and software are provide as a ZIP file included with the Vivado Design Suite
installation.
PG195 (v4.1) November 24, 2023

Send Feedback
1. Copy the ZIP file from the Vivado install directory to the FPGA connected Host PC and
extract (unzip) its contents. This file is located at the following path within the Vivado
installation directory.
XVC Driver and SW Path: …/data/xicom/driver/pcie/xvc_pcie.zip
The README.txt files within the driver_* and xvcserver directories identify how to
compile, install, and run the XVC drivers and software, and are summarized in the following
steps. Follow the following steps after the driver and software files have been copied to the
Host PC and you are logged in as a user with root permissions.
2. Modify the variables within the driver_*/xvc_pcie_user_config.h file to match your
hardware design and IP settings. Consider modifying the following variables:
• PCIE_VENDOR_ID: The PCIe Vendor ID defined in the PCIe® IP customization.
• PCIE_DEVICE_ID: The PCIe Device ID defined in the PCIe® IP customization.
• Config_space: Allows for the selection between using a PCIe-XVC-VSEC or an AXI-XVC
peripheral. The default value of AUTO first attempts to discover the PCIe-XVC-VSEC, then
attempts to connect to an AXI-XVC peripheral if the PCIe-XVC-VSEC is not found. A value
of CONFIG or BAR can be used to explicitly select between PCIe®-XVC-VSEC and AXI-
XVC implementations, as desired.
• config_vsec_id: The PCIe XVC VSEC ID (default 0x0008) defined in the Debug Bridge IP
when the Bridge Type is configured for From PCIE to BSCAN. This value is only used for
detection of the PCIe®-XVC-VSEC.
• config_vsec_rev: The PCIe XVC VSEC Rev ID (default 0x0) defined in the Debug Bridge IP
when the Bridge Type is configured for From PCIe to BSCAN. This value is only used for
detection of the PCIe-XVC-VSEC.
• bar_index: The PCIe BAR index that should be used to access the Debug Bridge IP when
the Bridge Type is configured for From AXI to BSCAN. This BAR index is specified as a
combination of the PCIe IP customization and the addressable AXI peripherals in your
system design. This value is only used for detection of an AXI-XVC peripheral.
• bar_offset: PCIe BAR Offset that should be used to access the Debug Bridge IP when the
Bridge Type is configured for From AXI to BSCAN. This BAR offset is specified as a
combination of the PCIe IP customization and the addressable AXI peripherals in your
system design. This value is only used for detection of an AXI-XVC peripheral.
3. Move the source files to the directory of your choice. For example, use:
/home/username/xil_xvc or /usr/local/src/xil_xvc
4. Make sure you have root permissions and change to the directory containing the driver files.
# cd /driver_*/
5. Compile the driver module:

# make install
The kernel module object file will be installed as:
PG195 (v4.1) November 24, 2023

Send Feedback
/lib/modules/[KERNEL_VERSION]/kernel/drivers/pci/pcie/Xilinx/
xil_xvc_driver.ko
6. Run the depmod command to pick up newly installed kernel modules:
# depmod -a
7. Make sure no older versions of the driver are loaded:

# modprobe -r xil_xvc_driver
8. Load the module:

# modprobe xil_xvc_driver
If you run the dmesg command, you will see the following message:
kernel: xil_xvc_driver: Starting…
Note: You can also use insmod on the kernel object file to load the module:
# insmod xil_xvc_driver.ko
However, this is not recommended unless necessary for compatibility with older kernels.
9. The resulting character file, /dev/xil_xvc/cfg_ioc0, is owned by user root and group
root, and it will need to have permissions of 660. Change permissions on this file if it does
not allow the application to interact with the driver.
# chmod 660 /dev/xil_xvc/cfg_ioc0
10. Build the simple test program for the driver:

# make test
11. Run the test program:

# ./driver_test/verify_xil_xvc_driver
You should see various successful tests of differing lengths, followed by the following
message:
"XVC PCIE Driver Verified Successfully!"
Compiling and Launching the XVC-Server Application

The XVC-Server application provides the connection between the Vivado HW server and the
XVC enabled PCIe device driver. The Vivado Design Suite connects to the XVC-Server using
TCP/IP. The desired port number will need to be exposed appropriately through the firewalls for
your network. The following steps can be used to compile and launch the XVC software
application, using the default port number of 10200.
1. Make sure the firewall settings on the system expose the port that will be used to connect to
the Vivado Design Suite. For this example, port 10200 is used.
PG195 (v4.1) November 24, 2023

Send Feedback
2. Make note of the host name or IP address. The host name and port number will be required
to connect Vivado to the xvcserver application. See the OS help pages for information
regarding the firewall port settings for your OS.
3. Move the source files to the directory of your choice. For example, use:
/home/username/xil_xvc or /usr/local/src/xil_xvc
4. Change to the directory containing the application source files:
# cd ./xvcserver/
5. Compile the application:

# make
6. Start the XVC-Server application:

# ./bin/xvc_pcie -s TCP::10200
After the Vivado Design Suite has connected to the XVC-server application you should see
the following message from the XVC-server.
Enable verbose by setting VERBOSE evn var.
Opening /dev/xil_xvc/cfg_ioc0
Connecting the Vivado Design Suite to the XVC-

Server Application
The Vivado Design Suite can be run on the computer that is running the XVC-server application,
or it can be run remotely on another computer that is connected over an Ethernet network. The
port however must be accessible to the machine running Vivado. To connect Vivado to the XVC-
Server application follow the steps should be used and are shown using the default port number.
1. Launch the Vivado Design Suite.

2. Select Open HW Manager.
3. In the Hardware Manager, select Open target → Open New Target.
4. Click Next.
5. Select Local server, and click Next.
This launches hw_server on the local machine, which then connects to the xvcserver
application.
6. Select Add Xilinx Virtual Cable (XVC).
7. In the Add Virtual Cable dialog box, type in the appropriate Host name or IP address, and
Port to connect to the xvcserver application. Click OK.
PG195 (v4.1) November 24, 2023

Send Feedback
8. Select the newly added XVC target from the Hardware Targets table, and click Next.
PG195 (v4.1) November 24, 2023

Send Feedback
9. Click Finish.
10. In the Hardware Device Properties panel, select the debug bridge target, and assign the
appropriate probes .ltx file.
PG195 (v4.1) November 24, 2023

Send Feedback
Vivado now recognizes your debug cores and debug signals, and you can debug your design
through the Vivado hardware tools interface using the standard debug approach.
This allows you to debug AMD FPGA designs through the PCIe connection rather than JTAG
using the Xilinx Virtual Cable technology. You can terminate the connection by closing the
hardware server from Vivado using the right-click menu. If the PCIe connection is lost or the
XVC-Server application stops running, the connection to the FPGA and associated debug cores
will also be lost.
Run Time Considerations

The Vivado connection to an XVC-Server Application should not be running when a device is
programmed. The XVC-Server Application along with the associated connection to Vivado
should only be initiated after the device has been programmed and the hardware PCIe interface
is active.
For DFX designs, it is important to terminate the connection during DFX operations. During a
DFX operation where debug cores are present inside the dynamic region, a portion of the debug
tree is expected to be reprogrammed. Vivado debug tools should not be actively communicating
with the FPGA through XVC during a DFX operation.
PG195 (v4.1) November 24, 2023

Send Feedback
Appendix F: Additional Resources and Legal Notices
Appendix F
Additional Resources and Legal

Notices
Finding Additional Documentation

Documentation Portal
The AMD Adaptive Computing Documentation Portal is an online tool that provides robust
search and navigation for documentation using your web browser. To access the Documentation
Portal, go to https://docs.xilinx.com.
Documentation Navigator
Documentation Navigator (DocNav) is an installed tool that provides access to AMD Adaptive
Computing documents, videos, and support resources, which you can filter and search to find
information. To open DocNav:
• From the AMD Vivado™ IDE, select Help → Documentation and Tutorials.
• On Windows, click the Start button and select Xilinx Design Tools → DocNav.
• At the Linux command prompt, enter docnav.
Note: For more information on DocNav, refer to the Documentation Navigator User Guide (UG968).
Design Hubs
AMD Design Hubs provide links to documentation organized by design tasks and other topics,
which you can use to learn key concepts and address frequently asked questions. To access the
Design Hubs:
• In DocNav, click the Design Hubs View tab.

• Go to the Design Hubs web page.
PG195 (v4.1) November 24, 2023

Send Feedback
Support Resources
For support resources such as Answers, Documentation, Downloads, and Forums, see Support.
References
These documents provide supplemental material useful with this guide:
1. AMBA AXI4-Stream Protocol Specification (ARM IHI 0051A)

2. PCI-SIG Documentation (www.pcisig.com/specifications)
3. Vivado Design Suite: AXI Reference Guide (UG1037)
4. AXI Bridge for PCI Express Gen3 Subsystem Product Guide (PG194)
5. 7 Series FPGAs Integrated Block for PCI Express LogiCORE IP Product Guide (PG054)
6. Virtex-7 FPGA Integrated Block for PCI Express LogiCORE IP Product Guide (PG023)
7. UltraScale Devices Gen3 Integrated Block for PCI Express LogiCORE IP Product Guide (PG156)
8. UltraScale+ Devices Integrated Block for PCI Express LogiCORE IP Product Guide (PG213)
9. Versal Adaptive SoC DMA and Bridge Subsystem for PCI Express Product Guide (PG344)
10. Vivado Design Suite User Guide: Designing IP Subsystems using IP Integrator (UG994)
11. Vivado Design Suite User Guide: Designing with IP (UG896)
12. Vivado Design Suite User Guide: Getting Started (UG910)
13. Vivado Design Suite User Guide: Using Constraints (UG903)
14. Vivado Design Suite User Guide: Logic Simulation (UG900)
15. Vivado Design Suite User Guide: Dynamic Function eXchange (UG909)
16. ISE to Vivado Design Suite Migration Guide (UG911)
17. Vivado Design Suite User Guide: Programming and Debugging (UG908)
18. Vivado Design Suite User Guide: Implementation (UG904)
19. AXI Interconnect LogiCORE IP Product Guide (PG059)
PG195 (v4.1) November 24, 2023

Send Feedback
Revision History
The following table shows the revision history for this document.
Section Revision Summary

11/24/2023 Version 4.1
UltraScale+ Devices Updated device support table.
Debug Options Tab Added list of signals.
11/16/2022 Version 4.1
Chapter 2: Overview Updated Bridge Between PCIe and AXI Memory section.
UltraScale+ Devices Updated the configuration table.
06/10/2022 Version 4.1
General updates Entire document.
04/29/2021 Version 4.1
General Updates Added references and links to Versal Adaptive SoC DMA and
Bridge Subsystem for PCI Express Product Guide (PG344) for
more information.
UltraScale+ Devices Updated supported devices.
09/21/2020 Version 4.1
General Updates Made clarifications throughout.
Tandem Configuration Updated Partial Reconfiguration references to Dynamic
Function eXchange.
Debug Options Tab Added new debug options.
Parameters for Custom PIPE Simulation Added guidance for required PIPE Simulation parameters.
Appendix B: Application Software Development Updated link for additional driver information.
11/22/2019 Version 4.1
Tandem Configuration Updated Supported Devices.
MSI-X Vector Table and PBA (0x8) Added MSI-X table offset and PBA table offset values.
06/20/2019 Version 4.1
Tandem Configuration Updated Supported Devices.
DMA C2H Stream Clarified that C2H Stream descriptor length size must be a
multiple of 64 Bytes.
IRQ Block Registers (0x2) Clarified MSI-X Interrupt register description for AXI Bridge
mode.
Customizing and Generating the Subsystem Updated screen captures.
Basic Tab Added GT DRP Clock Selection, and Data Protection options.
PCIe MISC Tab Added MSI RX PIN EN option.
PCIe DMA Tab Removed Parity Checking option (moved to Basic tab).
Appendix B: Application Software Development Appendix renamed from Device Driver to Application
Software Development.
PG195 (v4.1) November 24, 2023

Send Feedback

12/05/2018 Version 4.1
Chapter 3: Product Specification In the Minimum Device Requirements table:
• Added -2 supported speed grade for Gen3 x16 for
architecture devices (PCIE4).
• Added Gen4 link speed details for Virtex AMD
UltraScale+™ devices with high bandwidth memory
(HBM).
• Added information about behavior for multiple channels
to the H2C Channel section.
• Added information about access restrictions to the AXI4-
Lite Slave.
• Updated the cfg_ext_read_received description.
Chapter 4: Designing with the Subsystem Updated the production support details in the Tandem
PROM/PCIeSupported Configurations (UltraScale+ Devices)
table.
• Updated the Descriptor Bypass Mode description to
reflect that H2C and C2H descriptors have 128 bytes of
data.
• Added the Tcl command for Post-Synthesis simulation

for the example design.
Chapter 6: Example Design Added the User IRQ Example Design

04/04/2018 Version 4.1
General Updates Clarified that Tandem Configuration is not yet supported for
Bridge mode in UltraScale+ devices.
Chapter 2: Overview Added limitation: For 7 series, PCIe access from Host system
must be limited to 1DW (4 Bytes) transaction only.
Chapter 3: Product Specification Added clarifying text to the IRQ Module configuration
component (Legacy Interrupts, and MSI and MSI-X
Interrupts sections).
Editorial updates in the H2C Channel 0-3 AXI4-Stream
Interface Signals tables, and C2H Channel 0-3 AXI4-Stream
Interface Signals tables.
Added the dma_bridge_resetn signal to the Top-Level
Interface Signals table.
Updated Register Name: IRQ Block Channel Interrupt
Pending (0x4C)
Added Virtex UltraScale+ Devices with HBM (PCIE4C)
minimum device requirements information.
Chapter 4: Designing with the Subsystem Added Virtex UltraScale+ parts to the Tandem PROM/PCIe
Supported Configurations (UltraScale+ Devices) table.
Added Shared Logic support for 7 series Gen2 family
devices.
Device Driver Appendix Added clarifying text to the MSI Interrupt, MSI-X Interrupts,
and User Interrupts sections.
PG195 (v4.1) November 24, 2023

Send Feedback

12/20/2017 Version 4.0
General Updates Updated Minimum Device Requirements table for Gen 3 x8
support.
Added detail to the h2c_dsc_byp_ctl[15:0], and
c2h_dsc_byp_ctl[15:0] port descriptions.
Added Timing Diagram for Descriptor Bypass mode.
Updated description for 11:8 bit index (Channel ID[3:0] field)
in the PCIe to DMA Address Field Descriptions table.
Added new c_s_axi_supports_narrow_burst parameter to the
“Upgrading” appendix.
10/04/2017 Version 4.0
General Updates PCIe AXI Bridge mode operation removed from this guide,
and moved to AXI Bridge for PCI Express Gen3 Subsystem
Product Guide (PG194). This document (PG195) only covers
DMA mode operation.
In the Tandem Configuration section, added instruction and
device support information for UltraScale+ devices, and
added device support information for UltraScale devices.
Updated the “Upgrading” appendix according to port and
parameter changes for this version of the core.
Added Appendix D, “Using Xilinx Virtual Cable to Debug”.
06/07/2017 Version 3.1
General Updates Updated the [NUM_USR_INT-1:0] bit description details.
Updated the PCI Extended Tag parameter description.
Added a quick start for DMA C2H and H2C transfers in the
Product Specification chapter.
04/05/2017 Version 3.1
General Updates Updated driver support, Windows driver is in pre-
production.
Updated Identifier Version.
Added new GUI parameters: Reset Source, MSI-X
Implementation
Location, and AXI outstanding transactions.
Added Vivado IP integrator-based example design.
Updated the Simulation and Descriptor Bypass Mode
sections in the Test Bench chapter.
Added new parameters and ports to the Upgrading
appendix.
02/21/2017 Version 3.0
General Updates Updated supported UltraScale+ device speed grades in
Minimum Device Requirements table.
11/30/2016 Version 3.0
General Updates Updated the core name to reflect two core functional
modes: AXI Bridge Subsystem for PCIe (UltraScale+ only),
and DMA Subsystem for PCIe (all other supported devices).
Organized the Customizing and Generating the Subsystem
section (Chapter 4) according to the options available for
the two functional modes.
Added Debug Options tab in the Vivado IDE to enable
debugging options in the core.
Updated Identifier Version.
PG195 (v4.1) November 24, 2023

Send Feedback

10/12/2016 Version 3.0
General Updates Added AMD Artix™ 7 and Zynq 7000 SoC device restriction
that 7A15T and 7A25T are the only ones not supported.
10/05/2016 Version 3.0
General Updates Added additional device family support.
Add support for use with the AMD Gen2 integrated block for
PCIe core.
Added performance data to an Answer Record on the web.
Updated datapath width and restriction in the Address
Alignment and
Length Granularity tables in the DMA Operations section.
Updated Port Descriptions:
• Added support for Parity ports.
• Added support for the Configuration Extend ports.
Updated Register Space descriptions:
• Updated Identifier Version.
• Added H2C SGDMA Descriptor Credits (0x8C), C2H

SGDMA Descriptor Credits (0x8C0, SGDMA Descriptor
Credit Mode Enable (0x20), SG Descriptor Mode Enable
Register (0x24), SG Descriptor Mode Enable Register
(0x28).
Updated Vivado IP catalog description (2016.3):
• Updated PCIe: BARs tab, PCIe: Misc tab, and PCIe: DMA
tab.
• Added Shared Logic tab.
Added Basic Vivado Simulation section.

Added AXI-MM Example with Descriptor Bypass Mode
section.
Added additional supported 7 series evaluation board in
Debugging appendix).
06/08/2016 Version 2.0
General Updates Identifier Version update
AXI4-Stream Writeback Disable Control bit documented
04/06/2016 Version 2.0
Initial Xilinx release. N/A
PG195 (v4.1) November 24, 2023

Send Feedback
Please Read: Important Legal Notices

The information presented in this document is for informational purposes only and may contain
technical inaccuracies, omissions, and typographical errors. The information contained herein is
subject to change and may be rendered inaccurate for many reasons, including but not limited to
product and roadmap changes, component and motherboard version changes, new model and/or
product releases, product differences between differing manufacturers, software changes, BIOS
flashes, firmware upgrades, or the like. Any computer system has risks of security vulnerabilities
that cannot be completely prevented or mitigated. AMD assumes no obligation to update or
otherwise correct or revise this information. However, AMD reserves the right to revise this
information and to make changes from time to time to the content hereof without obligation of
AMD to notify any person of such revisions or changes. THIS INFORMATION IS PROVIDED "AS
IS." AMD MAKES NO REPRESENTATIONS OR WARRANTIES WITH RESPECT TO THE
CONTENTS HEREOF AND ASSUMES NO RESPONSIBILITY FOR ANY INACCURACIES,
ERRORS, OR OMISSIONS THAT MAY APPEAR IN THIS INFORMATION. AMD SPECIFICALLY
DISCLAIMS ANY IMPLIED WARRANTIES OF NON-INFRINGEMENT, MERCHANTABILITY, OR
FITNESS FOR ANY PARTICULAR PURPOSE. IN NO EVENT WILL AMD BE LIABLE TO ANY
PERSON FOR ANY RELIANCE, DIRECT, INDIRECT, SPECIAL, OR OTHER CONSEQUENTIAL
DAMAGES ARISING FROM THE USE OF ANY INFORMATION CONTAINED HEREIN, EVEN IF
AMD IS EXPRESSLY ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
AUTOMOTIVE APPLICATIONS DISCLAIMER
AUTOMOTIVE PRODUCTS (IDENTIFIED AS "XA" IN THE PART NUMBER) ARE NOT

WARRANTED FOR USE IN THE DEPLOYMENT OF AIRBAGS OR FOR USE IN APPLICATIONS
THAT AFFECT CONTROL OF A VEHICLE ("SAFETY APPLICATION") UNLESS THERE IS A
SAFETY CONCEPT OR REDUNDANCY FEATURE CONSISTENT WITH THE ISO 26262
AUTOMOTIVE SAFETY STANDARD ("SAFETY DESIGN"). CUSTOMER SHALL, PRIOR TO USING
OR DISTRIBUTING ANY SYSTEMS THAT INCORPORATE PRODUCTS, THOROUGHLY TEST
SUCH SYSTEMS FOR SAFETY PURPOSES. USE OF PRODUCTS IN A SAFETY APPLICATION
WITHOUT A SAFETY DESIGN IS FULLY AT THE RISK OF CUSTOMER, SUBJECT ONLY TO
APPLICABLE LAWS AND REGULATIONS GOVERNING LIMITATIONS ON PRODUCT
LIABILITY.
Copyright
© Copyright 2016-2023 Advanced Micro Devices, Inc. AMD, the AMD Arrow logo, Artix, Kintex,
UltraScale+, Versal, Virtex, Vivado, Zynq, and combinations thereof are trademarks of Advanced
Micro Devices, Inc. PCI, PCIe, and PCI Express are trademarks of PCI-SIG and used under license.
AMBA, AMBA Designer, Arm, ARM1176JZ-S, CoreSight, Cortex, PrimeCell, Mali, and MPCore
are trademarks of Arm Limited in the US and/or elsewhere. Other product names used in this
publication are for identification purposes only and may be trademarks of their respective
companies.
PG195 (v4.1) November 24, 2023

Send Feedback

DMA With Bridge (XDMA)

Uploaded by

Copyright:

Available Formats

DMA With Bridge (XDMA)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DMA With Bridge (XDMA)

Uploaded by

Copyright:

Available Formats

DMA/Bridge Subsystem for

PCI Express v4.1

PG195 (v4.1) November 24, 2023

AMD Adaptive Computing is creating an environment where

Chapter 3: Product Specification......................................................................... 12

Chapter 4: Designing with the Subsystem..................................................... 77

Chapter 5: Design Flow Steps.................................................................................83

Chapter 6: Example Design................................................................................... 102

PG195 (v4.1) November 24, 2023

Chapter 7: Test Bench...............................................................................................114

Appendix B: Application Software Development..................................... 123

Appendix D: Debugging.......................................................................................... 130

Appendix E: Using the Xilinx Virtual Cable to Debug............................. 135

Appendix F: Additional Resources and Legal Notices............................ 152

PG195 (v4.1) November 24, 2023

This document covers DMA mode operation only.

PG195 (v4.1) November 24, 2023

• Legacy, MSI, and MSI-X interrupts

PG195 (v4.1) November 24, 2023

This document covers DMA mode operation only.

PG195 (v4.1) November 24, 2023

Figure 1: DMA/Bridge Subsystem for PCI Express® Overview

DMA Subsystem for PCIe

Integrated Block PCIe RX

Cfg Master CQ/CC

PG195 (v4.1) November 24, 2023

• Data communications networks

PG195 (v4.1) November 24, 2023

Table 1: Supported PCIe Transaction Types

PG195 (v4.1) November 24, 2023

PCIe to DMA Bypass Master

User Interrupt in MSI-X Mode

Licensing and Ordering

PG195 (v4.1) November 24, 2023

PG195 (v4.1) November 24, 2023

• AMBA AXI4-Stream Protocol Specification (ARM IHI 0051A)

Performance and Resource Utilization

For DMA Resource Utilization, see Resource Utilization web page.

Minimum Device Requirements

Table 2: Minimum Device Requirements

Capability Link Speed Capability Link Width Supported Speed Grades

PG195 (v4.1) November 24, 2023

Table 2: Minimum Device Requirements (cont'd)

Capability Link Speed Capability Link Width Supported Speed Grades

Configurable Components of the Subsystem

PG195 (v4.1) November 24, 2023

The type of channel configured determines the transactions on which bus:

Table 3: 32-Bit BARs

PCIe BARs Selection

PG195 (v4.1) November 24, 2023

Table 4: 64-Bit BARs

PCIe BARs Selection

PCIe BARs Tab

PG195 (v4.1) November 24, 2023

Host-to-Card Bypass Master

PG195 (v4.1) November 24, 2023

PG195 (v4.1) November 24, 2023

Figure 2: Legacy Interrupts

MSI and MSI-X Interrupts

Figure 3: MSI Interrupts