Programmer’s Guide

• 1. Introduction
  • 1.1. Documentation Roadmap
  • 1.2. Related Publications
• 2. Overview
  • 2.1. Development Environment
  • 2.2. Environment Abstraction Layer
  • 2.3. Core Components
    • 2.3.1. Ring Manager (librte_ring)
    • 2.3.2. Memory Pool Manager (librte_mempool)
    • 2.3.3. Network Packet Buffer Management (librte_mbuf)
    • 2.3.4. Timer Manager (librte_timer)
  • 2.4. Ethernet* Poll Mode Driver Architecture
  • 2.5. Packet Forwarding Algorithm Support
  • 2.6. librte_net
• 3. Source Organization
  • 3.1. Libraries
  • 3.2. Drivers
  • 3.3. Applications
• 4. Environment Abstraction Layer
  • 4.1. EAL in a Linux-userland Execution Environment
    • 4.1.1. Initialization and Core Launching
    • 4.1.2. Shutdown and Cleanup
    • 4.1.3. Multi-process Support
    • 4.1.4. Memory Mapping Discovery and Memory Reservation
    • 4.1.5. Support for Externally Allocated Memory
    • 4.1.6. Per-lcore and Shared Variables
    • 4.1.7. Logs
    • 4.1.8. CPU Feature Identification
    • 4.1.9. User Space Interrupt Event
    • 4.1.10. Block list
    • 4.1.11. Misc Functions
    • 4.1.12. Lock annotations
    • 4.1.13. IOVA Mode Detection
    • 4.1.14. IOVA Mode Configuration
    • 4.1.15. Max SIMD bitwidth
  • 4.2. Memory Segments and Memory Zones (memzone)
  • 4.3. Multiple pthread
    • 4.3.1. EAL pthread and lcore Affinity
    • 4.3.2. non-EAL pthread support
    • 4.3.3. Public Thread API
    • 4.3.4. Control Thread API
    • 4.3.5. Known Issues
    • 4.3.6. Signal Safety
    • 4.3.7. cgroup control
  • 4.4. Malloc
    • 4.4.1. Alignment and NUMA Constraints
    • 4.4.2. Use Cases
    • 4.4.3. Internal Implementation
• 5. Log Library
  • 5.1. Log Levels
    • 5.1.1. Setting Global Log Level
    • 5.1.2. Setting Log Level for a Component
  • 5.2. Using Logging APIs to Generate Log Messages
• 6. Command-line Library
  • 6.1. Library Features
  • 6.2. Adding Command-line to an Application
    • 6.2.1. Creating a Command List File
    • 6.2.2. Running the Generator Script
    • 6.2.3. Providing the Function Callbacks
    • 6.2.4. Integrating with the Application
    • 6.2.5. Limitations of the Script Approach
  • 6.3. Worked Example of Adding Command-line to an Application
    • 6.3.1. Defining Command Result Structure
    • 6.3.2. Providing Field Initializers
    • 6.3.3. Defining Callback Function
    • 6.3.4. Associating Callback and Command
    • 6.3.5. Adding Command to Command-line Context
    • 6.3.6. Creating a Command-line Instance
    • 6.3.7. Multiplexing Multiple Commands to a Single Function
  • 6.4. Examples of Command-line Use in DPDK
• 7. Service Cores
  • 7.1. Service Core Initialization
  • 7.2. Enabling Services on Cores
  • 7.3. Service Core Statistics
  • 7.4. Service Core Tracing
• 8. Trace Library
  • 8.1. Overview
  • 8.2. DPDK tracing library features
  • 8.3. How to add a tracepoint?
    • 8.3.1. Create the tracepoint header file
    • 8.3.2. Register the tracepoint
  • 8.4. Fast path tracepoint
  • 8.5. Event record mode
  • 8.6. Trace file location
  • 8.7. View and analyze the recorded events
    • 8.7.1. Use the babeltrace command-line tool
    • 8.7.2. Use the tracecompass GUI tool
  • 8.8. Quick start
  • 8.9. Implementation details
    • 8.9.1. Trace metadata creation
    • 8.9.2. Trace memory
    • 8.9.3. Trace memory layout
  • 8.10. Limitations
• 9. RCU Library
  • 9.1. What is Quiescent State
  • 9.2. Factors affecting the RCU mechanism
  • 9.3. RCU in DPDK
  • 9.4. How to use this library
  • 9.5. Resource reclamation framework for DPDK
• 10. Ring Library
  • 10.1. References for Ring Implementation in FreeBSD*
  • 10.2. Lockless Ring Buffer in Linux*
  • 10.3. Additional Features
    • 10.3.1. Name
  • 10.4. Use Cases
  • 10.5. Anatomy of a Ring Buffer
    • 10.5.1. Single Producer Enqueue
    • 10.5.2. Single Consumer Dequeue
    • 10.5.3. Multiple Producers Enqueue
    • 10.5.4. Modulo 32-bit Indexes
  • 10.6. Producer/consumer synchronization modes
    • 10.6.1. MP/MC (default one)
    • 10.6.2. SP/SC
    • 10.6.3. MP_RTS/MC_RTS
    • 10.6.4. MP_HTS/MC_HTS
  • 10.7. Ring Peek API
  • 10.8. Ring Peek Zero Copy API
  • 10.9. References
• 11. Stack Library
  • 11.1. Implementation
    • 11.1.1. Lock-based Stack
    • 11.1.2. Lock-free Stack
• 12. Mempool Library
  • 12.1. Cookies
  • 12.2. Stats
  • 12.3. Memory Alignment Constraints on x86 architecture
  • 12.4. Local Cache
  • 12.5. Mempool Handlers
  • 12.6. Use Cases
• 13. Mbuf Library
  • 13.1. Design of Packet Buffers
  • 13.2. Buffers Stored in Memory Pools
  • 13.3. Constructors
  • 13.4. Allocating and Freeing mbufs
  • 13.5. Manipulating mbufs
  • 13.6. Meta Information
    • 13.6.1. Dynamic fields and flags
  • 13.7. Direct and Indirect Buffers
  • 13.8. Debug
  • 13.9. Use Cases
• 14. Poll Mode Driver
  • 14.1. Requirements and Assumptions
  • 14.2. Design Principles
  • 14.3. Logical Cores, Memory and NIC Queues Relationships
  • 14.4. Device Identification, Ownership and Configuration
    • 14.4.1. Device Identification
    • 14.4.2. Port Ownership
    • 14.4.3. Device Configuration
    • 14.4.4. On-the-Fly Configuration
    • 14.4.5. Configuration of Transmit Queues
    • 14.4.6. Free Tx mbuf on Demand
    • 14.4.7. Hardware Offload
  • 14.5. Poll Mode Driver API
    • 14.5.1. Generalities
    • 14.5.2. Generic Packet Representation
    • 14.5.3. Ethernet Device API
    • 14.5.4. Ethernet Device Standard Device Arguments
    • 14.5.5. Extended Statistics API
    • 14.5.6. NIC Reset API
    • 14.5.7. Proactive Error Handling Mode
• 15. Generic flow API (rte_flow)
  • 15.1. Overview
  • 15.2. Flow rule
    • 15.2.1. Description
    • 15.2.2. Attributes
    • 15.2.3. Pattern item
    • 15.2.4. Matching pattern
    • 15.2.5. Meta item types
    • 15.2.6. Data matching item types
    • 15.2.7. Actions
    • 15.2.8. Action types
    • 15.2.9. Negative types
  • 15.3. Rules management
    • 15.3.1. Validation
    • 15.3.2. Creation
    • 15.3.3. Destruction
    • 15.3.4. Update
    • 15.3.5. Flush
    • 15.3.6. Query
  • 15.4. Flow engine configuration
    • 15.4.1. Configuration
    • 15.4.2. Group Miss Actions
    • 15.4.3. Flow templates
  • 15.5. Asynchronous operations
    • 15.5.1. Enqueue creation operation
    • 15.5.2. Enqueue creation by index operation
    • 15.5.3. Enqueue destruction operation
    • 15.5.4. Enqueue update operation
    • 15.5.5. Enqueue indirect action creation operation
    • 15.5.6. Enqueue indirect action destruction operation
    • 15.5.7. Enqueue indirect action update operation
    • 15.5.8. Enqueue indirect action query operation
    • 15.5.9. Push enqueued operations
    • 15.5.10. Pull enqueued operations
    • 15.5.11. Calculate hash
  • 15.6. Flow isolated mode
  • 15.7. Verbose error reporting
  • 15.8. Helpers
    • 15.8.1. Error initializer
    • 15.8.2. Object conversion
    • 15.8.3. Tunneled traffic offload
  • 15.9. Caveats
  • 15.10. PMD interface
  • 15.11. Device compatibility
    • 15.11.1. Global bit-masks
    • 15.11.2. Unsupported layer types
    • 15.11.3. ANY pattern item
    • 15.11.4. Unsupported actions
    • 15.11.5. Flow rules priority
• 16. Switch Representation within DPDK Applications
  • 16.1. Introduction
  • 16.2. Sub Function
  • 16.3. Port Representors
  • 16.4. Basic SR-IOV
  • 16.5. Controlled SR-IOV
    • 16.5.1. Initialization
    • 16.5.2. VF Representors
    • 16.5.3. Traffic Steering
  • 16.6. Flow API (rte_flow)
    • 16.6.1. Extensions
    • 16.6.2. Traffic Direction
    • 16.6.3. Transferring Traffic
    • 16.6.4. Pattern Items And Actions
    • 16.6.5. Actions Order and Repetition
  • 16.7. Switching Examples
    • 16.7.1. Associating VF 1 with Physical Port 0
    • 16.7.2. Sharing Broadcasts
    • 16.7.3. Encapsulating VF 2 Traffic in VXLAN
• 17. Traffic Metering and Policing API
  • 17.1. Overview
  • 17.2. Configuration steps
  • 17.3. Run-time processing
  • 17.4. API walk-through
  • 17.5. Protocol based input color selection
• 18. Traffic Management API
  • 18.1. Overview
  • 18.2. Capability API
  • 18.3. Scheduling Algorithms
  • 18.4. Traffic Shaping
  • 18.5. Congestion Management
  • 18.6. Packet Marking
  • 18.7. Steps to Setup the Hierarchy
    • 18.7.1. Initial Hierarchy Specification
    • 18.7.2. Hierarchy Commit
    • 18.7.3. Run-Time Hierarchy Updates
• 19. Wireless Baseband Device Library
  • 19.1. Design Principles
  • 19.2. Device Management
    • 19.2.1. Device Creation
    • 19.2.2. Device Identification
    • 19.2.3. Device Configuration
    • 19.2.4. Queues Configuration
    • 19.2.5. Device & Queues Management
    • 19.2.6. Logical Cores, Memory and Queues Relationships
  • 19.3. Device Operation Capabilities
    • 19.3.1. Capabilities Discovery
    • 19.3.2. Capabilities details for LDPC Decoder
    • 19.3.3. Capabilities details for FFT function
    • 19.3.4. Other optional capabilities exposed during device discovery
  • 19.4. Operation Processing
    • 19.4.1. Enqueue / Dequeue Burst APIs
    • 19.4.2. Operation Representation
    • 19.4.3. Operation Management and Allocation
    • 19.4.4. BBDEV Inbound/Outbound Memory
    • 19.4.5. BBDEV Turbo Encode Operation
    • 19.4.6. BBDEV Turbo Decode Operation
    • 19.4.7. BBDEV LDPC Encode Operation
    • 19.4.8. BBDEV LDPC Decode Operation
    • 19.4.9. BBDEV FFT Operation
    • 19.4.10. BBDEV MLD-TS Operation
  • 19.5. Sample code
    • 19.5.1. BBDEV Device API
• 20. Cryptography Device Library
  • 20.1. Design Principles
  • 20.2. Device Management
    • 20.2.1. Device Creation
    • 20.2.2. Device Identification
    • 20.2.3. Device Configuration
    • 20.2.4. Configuration of Queue Pairs
    • 20.2.5. Logical Cores, Memory and Queues Pair Relationships
  • 20.3. Device Features and Capabilities
    • 20.3.1. Device Features
    • 20.3.2. Device Operation Capabilities
    • 20.3.3. Capabilities Discovery
  • 20.4. Operation Processing
    • 20.4.1. Private data
    • 20.4.2. User callback APIs
    • 20.4.3. Enqueue / Dequeue Burst APIs
    • 20.4.4. Operation Representation
    • 20.4.5. Operation Management and Allocation
  • 20.5. Symmetric Cryptography Support
    • 20.5.1. Session and Session Management
    • 20.5.2. Transforms and Transform Chaining
    • 20.5.3. Symmetric Operations
  • 20.6. Synchronous mode
    • 20.6.1. Cryptodev Raw Data-path APIs
  • 20.7. Sample code
  • 20.8. Asymmetric Cryptography
    • 20.8.1. Session and Session Management
    • 20.8.2. Asymmetric Sessionless Support
    • 20.8.3. Transforms and Transform Chaining
    • 20.8.4. Asymmetric Operations
    • 20.8.5. Private user data
  • 20.9. Asymmetric crypto Sample code
    • 20.9.1. Asymmetric Crypto Device API
  • 20.10. Device Statistics
• 21. Compression Device Library
  • 21.1. Device Management
    • 21.1.1. Device Creation
    • 21.1.2. Device Identification
    • 21.1.3. Device Configuration
    • 21.1.4. Configuration of Queue Pairs
    • 21.1.5. Logical Cores, Memory and Queue Pair Relationships
  • 21.2. Device Features and Capabilities
    • 21.2.1. Capabilities
    • 21.2.2. Capabilities Discovery
  • 21.3. Compression Operation
    • 21.3.1. Operation Representation
    • 21.3.2. Operation Management and Allocation
    • 21.3.3. Passing source data as mbuf-chain
    • 21.3.4. Operation Status
    • 21.3.5. Operation status after enqueue / dequeue
    • 21.3.6. Produced, Consumed And Operation Status
  • 21.4. Transforms
  • 21.5. Compression API Hash support
  • 21.6. Compression API Stateless operation
    • 21.6.1. priv_xform in Stateless operation
    • 21.6.2. Stateless and OUT_OF_SPACE
    • 21.6.3. Hash in Stateless
    • 21.6.4. Checksum in Stateless
  • 21.7. Compression API Stateful operation
    • 21.7.1. Stream in Stateful operation
    • 21.7.2. Stateful and OUT_OF_SPACE
    • 21.7.3. Hash in Stateful
    • 21.7.4. Checksum in Stateful
  • 21.8. Burst in compression API
    • 21.8.1. Enqueue / Dequeue Burst APIs
  • 21.9. Sample code
    • 21.9.1. Compression Device API
• 22. RegEx Device Library
  • 22.1. Design Principles
  • 22.2. Device Management
    • 22.2.1. Device Creation
    • 22.2.2. Device Identification
    • 22.2.3. Device Configuration
    • 22.2.4. Configuration of Rules Database
    • 22.2.5. Configuration of Queue Pairs
    • 22.2.6. Logical Cores, Memory and Queues Pair Relationships
  • 22.3. Device Features and Capabilities
    • 22.3.1. Enqueue / Dequeue Burst APIs
• 23. Machine Learning Device Library
  • 23.1. Design Principles
  • 23.2. Device Operations
    • 23.2.1. Device Creation
    • 23.2.2. Device Identification
    • 23.2.3. Device Features and Capabilities
    • 23.2.4. Device Configuration
    • 23.2.5. Configuration of Queue Pairs
    • 23.2.6. Logical Cores, Memory and Queues Pair Relationships
    • 23.2.7. Configuration of Machine Learning models
    • 23.2.8. Enqueue / Dequeue
    • 23.2.9. Quantize and Dequantize
• 24. DMA Device Library
  • 24.1. Design Principles
  • 24.2. Device Management
    • 24.2.1. Device Creation
    • 24.2.2. Device Identification
  • 24.3. Device Features and Capabilities
    • 24.3.1. Enqueue / Dequeue APIs
    • 24.3.2. Querying Device Statistics
• 25. General-Purpose Graphics Processing Unit Library
  • 25.1. Features
  • 25.2. API Overview
    • 25.2.1. Child Device
    • 25.2.2. Memory Allocation
    • 25.2.3. Memory Registration
    • 25.2.4. CPU mapping
    • 25.2.5. Memory Barrier
    • 25.2.6. Communication Flag
    • 25.2.7. Communication list
  • 25.3. CUDA Example
• 26. Security Library
  • 26.1. Design Principles
    • 26.1.1. Inline Crypto
    • 26.1.2. Inline protocol offload
    • 26.1.3. Lookaside protocol offload
    • 26.1.4. PDCP Flow Diagram
    • 26.1.5. DOCSIS Protocol
    • 26.1.6. MACSEC Protocol
    • 26.1.7. TLS-Record Protocol
  • 26.2. Device Features and Capabilities
    • 26.2.1. Device Capabilities For Security Operations
    • 26.2.2. Capabilities Discovery
    • 26.2.3. Security Session Create/Free
    • 26.2.4. Security session configuration
    • 26.2.5. Security API
    • 26.2.6. Flow based Security Session
  • 26.3. Telemetry support
• 27. Rawdevice Library
  • 27.1. Introduction
  • 27.2. Design
    • 27.2.1. Device Identification
• 28. Link Bonding Poll Mode Driver Library
  • 28.1. Link Bonding Modes Overview
  • 28.2. Implementation Details
    • 28.2.1. Link Status Change Interrupts / Polling
    • 28.2.2. Requirements / Limitations
    • 28.2.3. Configuration
  • 28.3. Using Link Bonding Devices
    • 28.3.1. Using the Poll Mode Driver from an Application
    • 28.3.2. Using Link Bonding Devices from the EAL Command Line
  • 28.4. Testpmd driver specific commands
    • 28.4.1. create bonding device
    • 28.4.2. add bonding member
    • 28.4.3. remove bonding member
    • 28.4.4. set bonding mode
    • 28.4.5. set bonding primary
    • 28.4.6. set bonding mac
    • 28.4.7. set bonding balance_xmit_policy
    • 28.4.8. set bonding mon_period
    • 28.4.9. set bonding lacp dedicated_queue
    • 28.4.10. set bonding agg_mode
    • 28.4.11. show bonding config
• 29. Timer Library
  • 29.1. Implementation Details
  • 29.2. Use Cases
  • 29.3. References
• 30. Hash Library
  • 30.1. Hash API Overview
  • 30.2. Multi-process support
  • 30.3. Multi-thread support
  • 30.4. Extendable Bucket Functionality support
  • 30.5. Implementation Details (non Extendable Bucket Case)
  • 30.6. Implementation Details (with Extendable Bucket)
  • 30.7. Entry distribution in hash table
  • 30.8. Use Case: Flow Classification
  • 30.9. References
• 31. Toeplitz Hash Library
  • 31.1. Toeplitz hash function API
  • 31.2. Predictable RSS
    • 31.2.1. Thash context
    • 31.2.2. Thash helper
    • 31.2.3. Calculation of the complementary bits to adjust the subtuple
    • 31.2.4. Adjust tuple API
  • 31.3. Use case example
• 32. Elastic Flow Distributor Library
  • 32.1. Introduction
  • 32.2. Flow Based Distribution
    • 32.2.1. Computation Based Schemes
    • 32.2.2. Flow-Table Based Schemes
    • 32.2.3. EFD Based Scheme
  • 32.3. Example of EFD Library Usage
  • 32.4. Library API Overview
    • 32.4.1. EFD Table Create
    • 32.4.2. EFD Insert and Update
    • 32.4.3. EFD Lookup
    • 32.4.4. EFD Delete
  • 32.5. Library Internals
    • 32.5.1. Insert Function Internals
    • 32.5.2. Lookup Function Internals
    • 32.5.3. Group Rebalancing Function Internals
  • 32.6. References
• 33. Membership Library
  • 33.1. Introduction
  • 33.2. Vector of Bloom Filters
  • 33.3. Hash-Table based Set-Summaries
    • 33.3.1. Set-Summaries with False Negative Probability
  • 33.4. Library API Overview
    • 33.4.1. Set-summary Create
    • 33.4.2. Set-summary Element Insertion
    • 33.4.3. Set-summary Element Lookup
    • 33.4.4. Set-summary Element Delete
  • 33.5. References
• 34. LPM Library
  • 34.1. LPM API Overview
  • 34.2. Implementation Details
    • 34.2.1. Addition
    • 34.2.2. Deletion
    • 34.2.3. Lookup
    • 34.2.4. Limitations in the Number of Rules
    • 34.2.5. Use Case: IPv4 Forwarding
    • 34.2.6. References
• 35. LPM6 Library
  • 35.1. LPM6 API Overview
    • 35.1.1. Implementation Details
    • 35.1.2. Addition
    • 35.1.3. Lookup
    • 35.1.4. Limitations in the Number of Rules
  • 35.2. Use Case: IPv6 Forwarding
• 36. FIB Library
  • 36.1. FIB API Overview
  • 36.2. Implementation details
  • 36.3. Dataplane Algorithms
    • 36.3.1. Dummy
    • 36.3.2. DIR-24-8
  • 36.4. Use cases
• 37. RIB Library
  • 37.1. Implementation details
  • 37.2. RIB API Overview
  • 37.3. Extensions usage example
• 38. Packet Distributor Library
  • 38.1. Distributor Core Operation
  • 38.2. Worker Operation
• 39. Reorder Library
  • 39.1. Operation
  • 39.2. Implementation Details
  • 39.3. Use Case: Packet Distributor
• 40. IP Fragmentation and Reassembly Library
  • 40.1. Packet fragmentation
  • 40.2. Packet reassembly
    • 40.2.1. IP Fragment Table
    • 40.2.2. Packet Reassembly
    • 40.2.3. Debug logging and Statistics Collection
• 41. Generic Receive Offload Library
  • 41.1. Overview
  • 41.2. Two Sets of API
    • 41.2.1. Lightweight Mode API
    • 41.2.2. Heavyweight Mode API
  • 41.3. Reassembly Algorithm
    • 41.3.1. Challenges
    • 41.3.2. Key-based Reassembly Algorithm
  • 41.4. TCP-IPv4/IPv6 GRO
  • 41.5. VxLAN GRO
  • 41.6. GRO Library Limitations
• 42. Generic Segmentation Offload Library
  • 42.1. Overview
  • 42.2. Limitations
  • 42.3. Packet Segmentation
    • 42.3.1. GSO Output Segment Format
  • 42.4. Supported GSO Packet Types
    • 42.4.1. TCP/IPv4 GSO
    • 42.4.2. UDP/IPv4 GSO
    • 42.4.3. VXLAN IPv4 GSO
    • 42.4.4. GRE TCP/IPv4 GSO
  • 42.5. How to Segment a Packet
• 43. IPsec Packet Processing Library
  • 43.1. SA level API
    • 43.1.1. RTE_SECURITY_ACTION_TYPE_NONE
    • 43.1.2. RTE_SECURITY_ACTION_TYPE_CPU_CRYPTO
    • 43.1.3. RTE_SECURITY_ACTION_TYPE_INLINE_CRYPTO
    • 43.1.4. RTE_SECURITY_ACTION_TYPE_INLINE_PROTOCOL
    • 43.1.5. RTE_SECURITY_ACTION_TYPE_LOOKASIDE_PROTOCOL
  • 43.2. SA database API
    • 43.2.1. Create/destroy
    • 43.2.2. Add/delete rules
    • 43.2.3. Lookup
  • 43.3. Supported features
  • 43.4. Telemetry support
  • 43.5. Limitations
• 44. PDCP Protocol Processing Library
  • 44.1. PDCP entity API
  • 44.2. PDCP PDU (Protocol Data Unit) API
    • 44.2.1. PDCP packet processing API for control PDU
    • 44.2.2. PDCP packet processing API for data PDU
  • 44.3. Supported features
    • 44.3.1. Supported ciphering algorithms
    • 44.3.2. Supported integrity protection algorithms
  • 44.4. Timers
  • 44.5. Sample API usage
• 45. Packet Capture Next Generation Library
  • 45.1. Usage
• 46. Packet Capture Library
  • 46.1. Operation
  • 46.2. Implementation Details
  • 46.3. Use Case: Packet Capturing
• 47. Multi-process Support
  • 47.1. Memory Sharing
  • 47.2. Deployment Models
    • 47.2.1. Symmetric/Peer Processes
    • 47.2.2. Asymmetric/Non-Peer Processes
    • 47.2.3. Running Multiple Independent DPDK Applications
    • 47.2.4. Running Multiple Independent Groups of DPDK Applications
  • 47.3. Multi-process Limitations
  • 47.4. Communication between multiple processes
    • 47.4.1. Registering for incoming messages
    • 47.4.2. Sending messages
    • 47.4.3. Sending requests
    • 47.4.4. Receiving and responding to messages
    • 47.4.5. Misc considerations
• 48. Thread Safety of DPDK Functions
  • 48.1. Fast-Path APIs
  • 48.2. Performance Insensitive API
  • 48.3. Library Initialization
  • 48.4. Interrupt Thread
• 49. Event Device Library
  • 49.1. Event struct
    • 49.1.1. Event Metadata
    • 49.1.2. Event Payload
    • 49.1.3. Event Vector
    • 49.1.4. Queues
    • 49.1.5. Ports
  • 49.2. API Walk-through
    • 49.2.1. Init and Config
    • 49.2.2. Setting up Queues
    • 49.2.3. Setting up Ports
    • 49.2.4. Linking Queues and Ports
    • 49.2.5. Linking Queues to Ports with link profiles
    • 49.2.6. Starting the EventDev
    • 49.2.7. Ingress of New Events
    • 49.2.8. Forwarding of Events
    • 49.2.9. Egress of Events
    • 49.2.10. Quiescing Event Ports
    • 49.2.11. Stopping the EventDev
  • 49.3. Summary
• 50. Event Ethernet Rx Adapter Library
  • 50.1. API Walk-through
    • 50.1.1. Creating an Adapter Instance
    • 50.1.2. Event device configuration for service based adapter
    • 50.1.3. Adding Rx Queues to the Adapter Instance
    • 50.1.4. Querying Adapter Capabilities
    • 50.1.5. Configuring the Service Function
    • 50.1.6. Starting the Adapter Instance
    • 50.1.7. Getting Adapter Statistics
    • 50.1.8. Getting Adapter queue config
    • 50.1.9. Set/Get adapter runtime configuration parameters
    • 50.1.10. Getting and resetting Adapter queue stats
    • 50.1.11. Getting Adapter Instance ID
    • 50.1.12. Interrupt Based Rx Queues
    • 50.1.13. Rx Callback for SW Rx Adapter
    • 50.1.14. Rx event vectorization
    • 50.1.15. Rx event vectorization for SW Rx adapter
• 51. Event Ethernet Tx Adapter Library
  • 51.1. API Walk-through
    • 51.1.1. Creating an Adapter Instance
    • 51.1.2. Event device configuration for service based adapter
    • 51.1.3. Adding Tx Queues to the Adapter Instance
    • 51.1.4. Querying Adapter Capabilities
    • 51.1.5. Linking a Queue to the Adapter’s Event Port
    • 51.1.6. Configuring the Service Function
    • 51.1.7. Starting the Adapter Instance
    • 51.1.8. Enqueuing Packets to the Adapter
    • 51.1.9. Getting Adapter Statistics
    • 51.1.10. Getting Adapter Instance ID
    • 51.1.11. Tx event vectorization
    • 51.1.12. Queue start/stop
    • 51.1.13. Set/Get adapter runtime configuration parameters
• 52. Event Timer Adapter Library
  • 52.1. Event Timer struct
    • 52.1.1. Timer Expiry Event
    • 52.1.2. Timeout Ticks
    • 52.1.3. State
    • 52.1.4. User Metadata
  • 52.2. API Overview
    • 52.2.1. Create and Configure an Adapter Instance
    • 52.2.2. Retrieve Event Timer Adapter Contextual Information
    • 52.2.3. Configuring the Service Component
    • 52.2.4. Starting the Adapter Instance
    • 52.2.5. Arming Event Timers
    • 52.2.6. Canceling Event Timers
  • 52.3. Processing Timer Expiry Events
  • 52.4. Summary
• 53. Event Crypto Adapter Library
  • 53.1. Adapter Mode
    • 53.1.1. RTE_EVENT_CRYPTO_ADAPTER_OP_NEW mode
    • 53.1.2. RTE_EVENT_CRYPTO_ADAPTER_OP_FORWARD mode
  • 53.2. API Overview
    • 53.2.1. Create an adapter instance
    • 53.2.2. Event device configuration for service based adapter
    • 53.2.3. Querying adapter capabilities
    • 53.2.4. Adding queue pair to the adapter instance
    • 53.2.5. Configure the service function
    • 53.2.6. Set event request/response information
    • 53.2.7. Enable event vectorization
    • 53.2.8. Start the adapter instance
    • 53.2.9. Get adapter statistics
    • 53.2.10. Set/Get adapter runtime configuration parameters
• 54. Event DMA Adapter Library
  • 54.1. Adapter Modes
    • 54.1.1. RTE_EVENT_DMA_ADAPTER_OP_NEW mode
    • 54.1.2. RTE_EVENT_DMA_ADAPTER_OP_FORWARD mode
  • 54.2. API Overview
    • 54.2.1. Create an adapter instance
    • 54.2.2. Event device configuration for service based adapter
    • 54.2.3. Querying adapter capabilities
    • 54.2.4. Adding vchan to the adapter instance
    • 54.2.5. Configuring service function
    • 54.2.6. Set event response information
    • 54.2.7. Start the adapter instance
    • 54.2.8. Get adapter statistics
    • 54.2.9. Set/Get adapter runtime configuration parameters
• 55. Dispatcher Library
  • 55.1. Overview
  • 55.2. Dispatcher Creation
  • 55.3. Event Port Binding
  • 55.4. Event Handlers
    • 55.4.1. Handler Registration
    • 55.4.2. Event Matching
    • 55.4.3. Event Delivery
    • 55.4.4. Event Clustering
  • 55.5. Finalize
  • 55.6. Service
    • 55.6.1. Multi Service Dispatcher Lcores
• 56. Quality of Service (QoS) Framework
  • 56.1. Packet Pipeline with QoS Support
  • 56.2. Hierarchical Scheduler
    • 56.2.1. Overview
    • 56.2.2. Scheduling Hierarchy
    • 56.2.3. Application Programming Interface (API)
    • 56.2.4. Implementation
    • 56.2.5. Worst Case Scenarios for Performance
  • 56.3. Droppers
    • 56.3.1. Configuration
    • 56.3.2. Enqueue Operation
    • 56.3.3. Queue Empty Operation
    • 56.3.4. Source Files Location
    • 56.3.5. Integration with the DPDK QoS Scheduler
    • 56.3.6. Integration with the DPDK QoS Scheduler Sample Application
    • 56.3.7. Application Programming Interface (API)
  • 56.4. Traffic Metering
    • 56.4.1. Functional Overview
    • 56.4.2. Implementation Overview
• 57. Power Management
  • 57.1. CPU Frequency Scaling
  • 57.2. Core-load Throttling through C-States
  • 57.3. Per-core Turbo Boost
  • 57.4. Use of Power Library in a Hyper-Threaded Environment
  • 57.5. API Overview of the Power Library
  • 57.6. User Cases
  • 57.7. Ethernet PMD Power Management API
    • 57.7.1. Abstract
    • 57.7.2. API Overview for Ethernet PMD Power Management
  • 57.8. Intel Uncore API
    • 57.8.1. Abstract
    • 57.8.2. API Overview for Intel Uncore
  • 57.9. References
• 58. Packet Classification and Access Control
  • 58.1. Overview
    • 58.1.1. Rule definition
    • 58.1.2. RT memory size limit
    • 58.1.3. Classification methods
  • 58.2. Application Programming Interface (API) Usage
    • 58.2.1. Classify with Multiple Categories
• 59. Packet Framework
  • 59.1. Design Objectives
  • 59.2. Overview
  • 59.3. Port Library Design
    • 59.3.1. Port Types
    • 59.3.2. Port Interface
  • 59.4. Table Library Design
    • 59.4.1. Table Types
    • 59.4.2. Table Interface
    • 59.4.3. Hash Table Design
  • 59.5. Pipeline Library Design
    • 59.5.1. Connectivity of Ports and Tables
    • 59.5.2. Port Actions
    • 59.5.3. Table Actions
  • 59.6. Multicore Scaling
    • 59.6.1. Shared Data Structures
  • 59.7. Interfacing with Accelerators
  • 59.8. The Software Switch (SWX) Pipeline
• 60. Vhost Library
  • 60.1. Vhost API Overview
  • 60.2. Vhost-user Implementations
  • 60.3. Guest memory requirement
  • 60.4. Vhost supported vSwitch reference
  • 60.5. Vhost data path acceleration (vDPA)
  • 60.6. Vhost asynchronous data path
• 61. Metrics Library
  • 61.1. Initializing the library
  • 61.2. Registering metrics
  • 61.3. Updating metric values
  • 61.4. Querying metrics
  • 61.5. Deinitialising the library
  • 61.6. Bit-rate statistics library
    • 61.6.1. Initialization
    • 61.6.2. Controlling the sampling rate
  • 61.7. Latency statistics library
    • 61.7.1. Initialization
    • 61.7.2. Triggering statistic updates
    • 61.7.3. Library shutdown
    • 61.7.4. Timestamp and latency calculation
• 62. Telemetry Library
  • 62.1. Creating Callback Functions
    • 62.1.1. Function Type
    • 62.1.2. Formatting Data
  • 62.2. Registering Commands
  • 62.3. Using Commands
• 63. Berkeley Packet Filter Library
  • 63.1. Packet data load instructions
  • 63.2. Not currently supported eBPF features
• 64. Graph Library and Inbuilt Nodes
  • 64.1. Features
  • 64.2. Advantages of Graph architecture
  • 64.3. Performance tuning parameters
  • 64.4. Programming model
    • 64.4.1. Anatomy of Node:
    • 64.4.2. Node creation and registration
    • 64.4.3. Link the Nodes to create the graph topology
    • 64.4.4. Create the graph object
    • 64.4.5. Graph models
    • 64.4.6. In fast path
    • 64.4.7. Context update when graph walk in action
    • 64.4.8. Get the node statistics using graph cluster
    • 64.4.9. Node writing guidelines
    • 64.4.10. Static nodes
    • 64.4.11. Intermediate nodes
  • 64.5. Graph object memory layout
  • 64.6. Inbuilt Nodes
    • 64.6.1. ethdev_rx
    • 64.6.2. ethdev_tx
    • 64.6.3. pkt_drop
    • 64.6.4. ip4_lookup
    • 64.6.5. ip4_rewrite
    • 64.6.6. ip4_reassembly
    • 64.6.7. ip6_lookup
    • 64.6.8. ip6_rewrite
    • 64.6.9. null
    • 64.6.10. kernel_tx
    • 64.6.11. kernel_rx
    • 64.6.12. ip4_local
    • 64.6.13. udp4_input
• 65. Installing DPDK Using the meson build system
  • 65.1. Summary
  • 65.2. Getting the Tools
  • 65.3. Configuring the Build
  • 65.4. Performing the Build
  • 65.5. Installing the Compiled Files
  • 65.6. Cross Compiling DPDK
  • 65.7. Using the DPDK within an Application
• 66. Running DPDK Unit Tests with Meson
  • 66.1. Grouping of test cases
  • 66.2. Dealing with skipped test cases
• 67. Building Your Own Application
  • 67.1. Compiling a Sample Application in the Development Kit Directory
  • 67.2. Build Your Own Application Outside the Development Kit
  • 67.3. Customizing Makefiles
    • 67.3.1. Application Makefile
• 68. Performance Optimization Guidelines
  • 68.1. Introduction
• 69. Writing Efficient Code
  • 69.1. Memory
    • 69.1.1. Memory Copy: Do not Use libc in the Data Plane
    • 69.1.2. Memory Allocation
    • 69.1.3. Concurrent Access to the Same Memory Area
    • 69.1.4. NUMA
    • 69.1.5. Distribution Across Memory Channels
    • 69.1.6. Locking memory pages
  • 69.2. Communication Between lcores
  • 69.3. PMD
    • 69.3.1. Lower Packet Latency
  • 69.4. Locks and Atomic Operations
    • 69.4.1. Locks
    • 69.4.2. Atomic Operations: Use C11 Atomic Builtins
  • 69.5. Coding Considerations
    • 69.5.1. Inline Functions
    • 69.5.2. Branch Prediction
  • 69.6. Setting the Target CPU Type
• 70. Link Time Optimization
• 71. Profile Your Application
  • 71.1. Profiling on x86
    • 71.1.1. Profiling with VTune
  • 71.2. Profiling on ARM64
    • 71.2.1. Using Linux perf
    • 71.2.2. Low-resolution generic counter
    • 71.2.3. High-resolution cycle counter
• 72. Running AddressSanitizer
  • 72.1. Example heap-buffer-overflow error
  • 72.2. Example use-after-free error
• 73. Glossary