L-Tile and H-Tile Avalon® Memory-mapped Intel® FPGA IP for PCI Express* User Guide

Download PDF
ID 683667
Date 9/13/2024
Public
Document Table of Contents
Document Table of Contents x
1. Introduction 2. Quick Start Guide 3. Interface Overview 4. Parameters 5. Designing with the IP Core 6. Block Descriptions 7. Registers 8. Programming Model for the DMA Descriptor Controller 9. Programming Model for the Avalon® -MM Root Port 10. Avalon-MM Testbench and Design Example 11. Document Revision History A. PCI Express Core Architecture B. Root Port Enumeration C. Troubleshooting and Observing the Link Status
1. Introduction x
1.1. Avalon-MM Interface for PCIe 1.2. Features 1.3. Release Information 1.4. Device Family Support 1.5. Recommended Fabric Speed Grades 1.6. Performance and Resource Utilization 1.7. Transceiver Tiles 1.8. PCI Express IP Core Package Layout 1.9. Channel Availability
2. Quick Start Guide x
2.1. Design Components 2.2. Hardware and Software Requirements 2.3. Directory Structure 2.4. Generating the Design Example 2.5. Simulating the Design Example 2.6. Compiling the Design Example and Programming the Device 2.7. Installing the Linux Kernel Driver 2.8. Running the Design Example Application
3. Interface Overview x
3.1. Avalon-MM DMA Interfaces when Descriptor Controller Is Internally Instantiated 3.2. Avalon-MM DMA Interfaces when Descriptor Controller is Externally Instantiated 3.3. Other Avalon-MM Interfaces 3.4. Clocks and Reset 3.5. System Interfaces
3.3. Other Avalon-MM Interfaces x
3.3.1. Avalon-MM Master Interfaces 3.3.2. Avalon-MM Slave Interfaces 3.3.3. Control Register Access (CRA) Avalon-MM Slave
4. Parameters x
4.1. Avalon-MM Settings 4.2. Base Address Registers 4.3. Device Identification Registers 4.4. PCI Express and PCI Capabilities Parameters 4.5. Configuration, Debug and Extension Options 4.6. PHY Characteristics 4.7. Example Designs
4.4. PCI Express and PCI Capabilities Parameters x
4.4.1. Device Capabilities 4.4.2. Link Capabilities 4.4.3. MSI and MSI-X Capabilities 4.4.4. Slot Capabilities 4.4.5. Power Management 4.4.6. Vendor Specific Extended Capability (VSEC)
5. Designing with the IP Core x
5.1. Generation 5.2. Simulation 5.3. IP Core Generation Output ( Quartus® Prime Pro Edition) 5.4. Channel Layout and PLL Usage
6. Block Descriptions x
6.1. Interfaces
6.1. Interfaces x
6.1.1. Stratix® 10 DMA Avalon-MM DMA Interface to the Application Layer 6.1.2. Avalon-MM Interface to the Application Layer 6.1.3. Clocks and Reset 6.1.4. Interrupts 6.1.5. Flush Requests 6.1.6. Serial Data, PIPE, Status, Reconfiguration, and Test Interfaces
6.1.1. Stratix® 10 DMA Avalon-MM DMA Interface to the Application Layer x
6.1.1.1. Descriptor Controller Interfaces when Instantiated Internally 6.1.1.2. Write DMA Avalon-MM Master Port 6.1.1.3. Descriptor Controller Interfaces when Instantiated Externally
6.1.1.1. Descriptor Controller Interfaces when Instantiated Internally x
6.1.1.1.1. Read Data Mover 6.1.1.1.2. Read Descriptor Controller Avalon-MM Master interface 6.1.1.1.3. Write Descriptor Controller Avalon-MM Master Interface 6.1.1.1.4. Read Descriptor Table Avalon-MM Slave Interface 6.1.1.1.5. Write Descriptor Table Avalon-MM Slave Interface
6.1.1.3. Descriptor Controller Interfaces when Instantiated Externally x
6.1.1.3.1. Avalon® -ST Descriptor Source
6.1.2. Avalon-MM Interface to the Application Layer x
6.1.2.1. Bursting and Non-Bursting Avalon® -MM Module Signals 6.1.2.2. Non-Bursing Slave Module 6.1.2.3. 32-Bit Control Register Access (CRA) Slave Signals 6.1.2.4. Bursting Slave Module
6.1.3. Clocks and Reset x
6.1.3.1. Clocks 6.1.3.2. Resets
6.1.4. Interrupts x
6.1.4.1. MSI Interrupts for Endpoints 6.1.4.2. Legacy Interrupts
6.1.6. Serial Data, PIPE, Status, Reconfiguration, and Test Interfaces x
6.1.6.1. Serial Data Interface 6.1.6.2. PIPE Interface 6.1.6.3. Hard IP Status Interface 6.1.6.4. Hard IP Reconfiguration 6.1.6.5. Test Interface
7. Registers x
7.1. Configuration Space Registers 7.2. Avalon-MM DMA Bridge Registers
7.1. Configuration Space Registers x
7.1.1. Register Access Definitions 7.1.2. PCI Configuration Header Registers 7.1.3. PCI Express Capability Structures 7.1.4. Intel Defined VSEC Capability Header 7.1.5. Uncorrectable Internal Error Status Register 7.1.6. Uncorrectable Internal Error Mask Register 7.1.7. Correctable Internal Error Status Register 7.1.8. Correctable Internal Error Mask Register
7.1.4. Intel Defined VSEC Capability Header x
7.1.4.1. Intel Defined Vendor Specific Header 7.1.4.2. Intel Marker 7.1.4.3. JTAG Silicon ID 7.1.4.4. User Configurable Device and Board ID
7.2. Avalon-MM DMA Bridge Registers x
7.2.1. PCI Express Avalon-MM Bridge Register Address Map 7.2.2. DMA Descriptor Controller Registers
7.2.1. PCI Express Avalon-MM Bridge Register Address Map x
7.2.1.1. Avalon-MM to PCI Express Interrupt Status Registers 7.2.1.2. Avalon-MM to PCI Express Interrupt Enable Registers 7.2.1.3. Address Mapping for High-Performance Avalon-MM 32-Bit Slave Modules 7.2.1.4. PCI Express to Avalon-MM Interrupt Status and Enable Registers for Endpoints 7.2.1.5. PCI Express Configuration Information Registers
7.2.2. DMA Descriptor Controller Registers x
7.2.2.1. Read DMA Internal Descriptor Controller Registers 7.2.2.2. Write DMA Internal Descriptor Controller Registers
8. Programming Model for the DMA Descriptor Controller x
8.1. Read DMA Example 8.2. Write DMA Example 8.3. Software Program for Simultaneous Read and Write DMA 8.4. Read DMA and Write DMA Descriptor Format
9. Programming Model for the Avalon® -MM Root Port x
9.1. Root Port TLP Data Control and Status Registers 9.2. Sending a TLP 9.3. Receiving a Non-Posted Completion TLP 9.4. Example of Reading and Writing BAR0 Using the CRA Interface
10. Avalon-MM Testbench and Design Example x
10.1. Avalon-MM Endpoint Testbench 10.2. Endpoint Design Example 10.3. Avalon® -MM Test Driver Module 10.4. Root Port BFM 10.5. BFM Procedures and Functions
10.2. Endpoint Design Example x
10.2.1. BAR Setup
10.4. Root Port BFM x
10.4.1. Overview 10.4.2. Issuing Read and Write Transactions to the Application Layer 10.4.3. Configuration of Root Port and Endpoint 10.4.4. Configuration Space Bus and Device Numbering 10.4.5. BFM Memory Map
10.5. BFM Procedures and Functions x
10.5.1. ebfm_barwr Procedure 10.5.2. ebfm_barwr_imm Procedure 10.5.3. ebfm_barrd_wait Procedure 10.5.4. ebfm_barrd_nowt Procedure 10.5.5. ebfm_cfgwr_imm_wait Procedure 10.5.6. ebfm_cfgwr_imm_nowt Procedure 10.5.7. ebfm_cfgrd_wait Procedure 10.5.8. ebfm_cfgrd_nowt Procedure 10.5.9. BFM Configuration Procedures 10.5.10. BFM Shared Memory Access Procedures 10.5.11. BFM Log and Message Procedures 10.5.12. Verilog HDL Formatting Functions
10.5.9. BFM Configuration Procedures x
10.5.9.1. ebfm_cfg_rp_ep Procedure 10.5.9.2. ebfm_cfg_decode_bar Procedure
10.5.10. BFM Shared Memory Access Procedures x
10.5.10.1. Shared Memory Constants 10.5.10.2. shmem_write Task 10.5.10.3. shmem_read Function 10.5.10.4. shmem_display Verilog HDL Function 10.5.10.5. shmem_fill Procedure 10.5.10.6. shmem_chk_ok Function
10.5.11. BFM Log and Message Procedures x
10.5.11.1. ebfm_display Verilog HDL Function 10.5.11.2. ebfm_log_stop_sim Verilog HDL Function 10.5.11.3. ebfm_log_set_suppressed_msg_mask Task 10.5.11.4. ebfm_log_set_stop_on_msg_mask Verilog HDL Task 10.5.11.5. ebfm_log_open Verilog HDL Function 10.5.11.6. ebfm_log_close Verilog HDL Function
10.5.12. Verilog HDL Formatting Functions x
10.5.12.1. himage1 10.5.12.2. himage2 10.5.12.3. himage4 10.5.12.4. himage8 10.5.12.5. himage16 10.5.12.6. dimage1 10.5.12.7. dimage2 10.5.12.8. dimage3 10.5.12.9. dimage4 10.5.12.10. dimage5 10.5.12.11. dimage6 10.5.12.12. dimage7
11. Document Revision History x
11.1. Document Revision History for the L-Tile and H-Tile Avalon® Memory-mapped Intel® FPGA IP for PCI Express* User Guide
A. PCI Express Core Architecture x
A.1. Transaction Layer A.2. Data Link Layer A.3. Physical Layer
C. Troubleshooting and Observing the Link Status x
C.1. Troubleshooting C.2. PCIe Link Inspector Overview
C.1. Troubleshooting x
C.1.1. Simulation Fails To Progress Beyond Polling.Active State C.1.2. Hardware Bring-Up Issues C.1.3. Link Training C.1.4. Use Third-Party PCIe Analyzer
C.2. PCIe Link Inspector Overview x
C.2.1. PCIe* Link Inspector Hardware
C.2.1. PCIe* Link Inspector Hardware x
C.2.1.1. Enabling the PCIe* Link Inspector C.2.1.2. Launching the PCIe* Link Inspector C.2.1.3. The PCIe* Link Inspector LTSSM Monitor C.2.1.4. Accessing the Configuration Space and Transceiver Registers C.2.1.5. Additional Status Commands
C.2.1.3. The PCIe* Link Inspector LTSSM Monitor x
C.2.1.3.1. LTSSM Monitor Registers C.2.1.3.2. ltssm_save2file <file_name> C.2.1.3.3. ltssm_file2console C.2.1.3.4. ltssm_debug_check C.2.1.3.5. ltssm_display <num_states> C.2.1.3.6. ltssm_save_oldstates
C.2.1.4. Accessing the Configuration Space and Transceiver Registers x
C.2.1.4.1. PCIe* Link Inspector Commands C.2.1.4.2. NPDME PLL and Channel Commands C.2.1.4.3. Register Address Map
C.2.1.5. Additional Status Commands x
C.2.1.5.1. Displaying PLL Lock and Calibration Status Registers
1. Introduction
2. Quick Start Guide
3. Interface Overview
4. Parameters
5. Designing with the IP Core
6. Block Descriptions
7. Registers
8. Programming Model for the DMA Descriptor Controller
9. Programming Model for the Avalon® -MM Root Port
10. Avalon-MM Testbench and Design Example
11. Document Revision History
A. PCI Express Core Architecture
B. Root Port Enumeration
C. Troubleshooting and Observing the Link Status

8.3. Software Program for Simultaneous Read and Write DMA

Program the following steps to implement a simultaneous DMA transfer:

  1. Allocate PCIe* system memory for the Read and Write DMA descriptor tables. If, for example, each table supports up to 128, eight-DWORD descriptors and 128, one-DWORD status entries for a total of 1152 DWORDs. Total memory for the Read and Write DMA descriptor tables is 2304 DWORDs.
  2. Allocate PCIe* system memory and initialize it with data for the Read Data Mover to read.
  3. Allocate PCIe* system memory for the Write Data Mover to write.
  4. Create all the descriptors for the read DMA descriptor table. Assign the DMA Descriptor IDs sequentially, starting with 0 to a maximum of 127. For the read DMA, the source address is the memory space allocated in Step 2. The destination address is the Avalon‑MM address that the Read Data Mover module writes. Specify the DMA length in DWORDs. Each descriptor transfers contiguous memory. Assuming a base address of 0, for the Read DMA, the following assignments illustrate construction of a read descriptor:
    1. RD_LOW_SRC_ADDR = 0x0000 (The base address for the read descriptor table in the PCIe system memory.)
    2. RD_HIGH_SRC_ADDR = 0x0004
    3. RD_CTRL_LOW_DEST_ADDR 0x0008
    4. RD_CTRL_HIGH_DEST_ADDR = 0x000C
    5. RD_DMA_LAST_PTR = 0x0010
    Writing the RD_DMA_LAST_PTR register starts operation.
  5. For the Write DMA, the source address is the Avalon‑MM address that the Write Data Mover module should read. The destination address is the PCIe* system memory space allocated in Step 3. Specify the DMA size in DWORDs. Assuming a base address of 0x100, for the Write Data Mover, the following assignments illustrate construction of a write descriptor:
    1. WD_LOW_SRC_ADDR = 0x0100 (The base address for the write descriptor table in the PCIe* system memory.)
    2. WD_HIGH_SRC_ADDR = 0x0104
    3. WD_CTRL_LOW_DEST_ADDR 0x0108
    4. WD_CTRL_HIGH_DEST_ADDR = 0x010C
    5. WD_DMA_LAST_PTR = 0x0110
    Writing the WD_DMA_LAST_PTR register starts operation.
  6. To improve throughput, the Read DMA module copies the descriptor table to the Avalon-MM memory before beginning operation. Specify the memory address by writing to the Descriptor Table Base (Low) and (High) registers.
  7. An MSI interrupt is sent for each WD_DMA_LAST_PTR or RD_DMA_LAST_PTR that completes. These completions result in updates to the Update bits. Host software can then read Update bits to determine which DMA operations are complete.
Note: If the transfer size of the read DMA is greater than the maximum read request size, the Read DMA creates multiple read requests. For example, if maximum read request size is 512 bytes, the Read Data Mover breaks a 4 KB read request into 8 requests with 8 different tags. The Read Completions can come back in any order. The Read Data Mover's Avalon-MM master port writes the data received in the Read Completions to the correct locations in Avalon® -MM memory, based on the tags in the same order as it receives the Completions. This order is not necessarily in increasing address order;. The data mover does not include an internal reordering buffer. If system allows out of order read completions, then status for the latest entry is latest only in number, but potentially earlier than other completions chronologically
Level Two Title