Skip to content

Morcules/SwiftNet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

290 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SwiftNet - Networking Library

SwiftNet is a lightweight networking library built on top of pcap, operating at Layer 2. It is designed for developers who value simplicity, readability, and performance.

Use Cases

  • High performance server side networking with kernel bypass (DPDK / AF_XDP)
  • Performance critical applications requiring minimal OS overhead
  • Multiplayer game networking with fully custom transport protocols

Supported Platforms

  • Apple Silicon (macOS arm64)
  • Linux arm64 (Release build works, but testing is disabled due to TSAN issues)

Features

  • 💡 Ease of Use: Simple API designed to get up and running quickly, without needing to deal directly with scokets, ip/eth headers or complicated libraries.
  • 📂 Lightweight: No dependencies except PCAP and a small footprint.

Why Use SwiftNet?

  • Straightforward API: Get up and running with minimal setup.
  • Open Source and Collaborative: Contributions are welcome to make it even better.
  • Compile time feature choosing Compile only specific features of the library.
  • Lightweight Static library build is under 1MB.
  • Future-proof Designed to scale with optional features enabled at compile time.
  • Server side performance The main focus is to make linux servers easily scalable, by enabling AF DXP, DPDK or other ways of accelerating packets without operating system involvement.

Future Version Goals

  • 0.5.0: Performance improvements in critical paths, ready for some real world usage

Goals

Upcoming goals

  • Optimize the most obvious parts of the codebase

  • Stabilize the API to avoid breaking changes in future releases

  • AF_XDP Linux support

  • Long-term: optional hardware acceleration (FPGA/ASIC) with external NIC to bypass OS overhead for server side processing

Performance

  • Hardware & Conditions
    Tested on a MacBook Air M3 (my daily driver machine, running multiple background applications)

  • Interface
    Real Wi-Fi interface: en0

  • Throughput
    18 MB/s

  • Memory Usage
    Peak memory footprint: 10 million bytes

  • CPU Usage
    Average utilization: ~8%

  • Detailed Timing (from /usr/bin/time -l)

    • Real (wall-clock) time: 5.01 seconds
    • User CPU time: 0.58 seconds
    • System CPU time: 0.23 seconds
    • Total CPU time: 0.81 seconds
  • CPU Cycles
    2,273,393,860 cycles

Installation

Follow these steps to install SwiftNet:

VCPKG

vcpkg install morcules-swiftnet

From Source

  1. Clone the repository to your local machine:
git clone https://github.com/morcules/SwiftNet
  1. Navigate to the build directory inside the SwiftNet directory:
cd SwiftNet/build
  1. Compile:
./build_for_release.sh
  1. To use SwiftNet in your project:
  • Include the SwiftNet.h header from the src directory in your main source file (e.g., main.c).
  • Link against the static library libswiftnet.a using your compiler.

Important note

  • To run the library successfully, you're required to run the app with sudo.
  • Pcap requires sudo

Example

struct SwiftNetClientConnection* const client_conn = swiftnet_create_client("127.0.0.1", 8080, 1000);
if (client_conn == NULL) {
    printf("Failed to create client connection\n");
    return -1;
}

swiftnet_client_set_message_handler(client_conn, on_client_packet, NULL);

struct SwiftNetPacketBuffer buf = swiftnet_client_create_packet_buffer(sizeof(int));

int code = 0xFF;

swiftnet_client_append_to_packet(&code, sizeof(code), &buf);
swiftnet_client_send_packet(client_conn, &buf);
swiftnet_client_destroy_packet_buffer(&buf);

Contributing

Contributions are very welcome no matter how small. Every single PR, issue, question, or typo fix is admired and appreciated.

  • Check out open issues — Start with those labeled good first issue or help wanted.
  • Found a bug or have a feature idea? Open an issue to discuss it first (especially for larger changes).
  • Want to fix something yourself? Fork the repo, create a branch, and submit a pull request.

Rules

  • AI - You're free to use anything you want to your advantage, but AI written slop that wasn't even ran locally won't be merged. If you don't understand the code please do not subbmit anything.

License

This project is licensed under the Apache License 2.0