I’ve been playing around with the low-level access to the Windows networking stack that is provided by \Device\Afd. This provides a ‘readiness’ interface rather than the ‘completion’ interface provided by traditional IOCP designs. I’ve been meaning to do some comparative performance tests, much in the same way that I did for my investigations of the RIO API, but I’ve been too busy.
Instead, what has happened is that the AFD code has made its way into the version of The Server Framework that I’m using with my Online Game Company client.
Last time I looked at way of using \Device\Afd to perform individual socket polling for readiness. This differed from the previous approach to using \Device\Afd, which batched up the sockets and issued a single poll for multiple sockets.
The individual socket polling approach appeals to me as it would appear to scale more easily, and putting together an echo server that supports multiple connections is now much easier. It doesn’t map as well to the way other operating systems do things though, so if that’s your primary goal, then you’re probably better off continuing with the ‘set of sockets’ approach.
Recently, I’be been exploring socket readiness notifications on Windows using the \Device\Afd interface. My initial forays into this were from a Linux epoll direction as the use of the \Device\Afd API provides a similar interface to the epoll API and makes it possible to build something almost the same on Windows which can help when writing cross platform code.
I’d got to the point where I had a simple client and server working nicely and had taken a pause before the next step which was to add support for multiple sockets.
For a while now I’ve been playing around with the low-level access to the Windows networking stack that is provided by \Device\Afd. Recently I had a comment from a reader, Matt James, who pointed me in the direction of his work over at https://github.com/Emjayen/napi which does a lot more with the \Device\Afd interface than I have. It’s well worth checking out.
This code includes, as reference, a set of NT Native API headers from the Process Hacker project, which in itself is interesting stuff, but Matt’s code goes far beyond my megre investigations and provides enough information to produce a full featured network API that appears to be able to everything you might need through the \Device\Afd interface.
The only difference between a client and a server is the way in which the connection is established. For a client, you create a socket and call “connect” on it. For a server, we have a socket that is “listening” for connections, “accepts” new incoming connections and returns a socket that is then indistinguishable from a client connection.
In the past, I’ve created bad abstractions at this point. A socket connection and a listening socket are both represented by the operating system API as the same type, and the only differences are the calls that you make on the type.
Now that I have a reasonably easy to use event-driven socket class I can start to use it. The first thing to build is a simple client. This will connect to a server and expect that server to echo anything that is sent to it. The client will send data and read it back and compare it with what it sent.
A simple, single-threaded, client Last time, I put together a tcp_socket object that made it easier to work with the \Device\Afd interface that I’ve been playing with.
I’ve been investigating the ‘sparsely documented’ \Device\Afd interface that lies below the Winsock2 layer. Today I use a test-driven method for building some code to make using this API a little easier.
Using the API In my previous posts on the \Device\Afd interface I focused on exploring and understanding the API itself. Since it’s ‘sparsely documented’ I used unit tests to explore and record my findings. This has left me with some code that is easy to come back to and pick up, which is lucky since it’s been quite a while since I last had some time to tinker with it.
I’ve been investigating the ‘sparsely documented’ \Device\Afd interface that lies below the Winsock2 layer. Today I use a test driven method for understanding and documenting the API.
TDU - Test Driven Understanding When trying to understand a new API I always like to end up with executable documentation in the form of tests that show the behaviour of the API. I write these tests in the same way that I write any tests; writing a test that fails and then adjusting so that it passes.
I’ve been playing around with Rust recently and whilst investigating asynchronous programming in Rust I was looking at Tokio, an async runtime. From there I started looking at Mio, the cross-platform, low-level, I/O code that Tokio uses.
For Windows platforms Mio uses wepoll, which is a Windows implementation of the Linux epoll API on Windows and is based on the code that is used by libuv for Node.js. This uses networking code that is NOT your standard high-performance Windows networking code using I/O completion ports and instead uses the ‘sparsely documented’ \Device\Afd interface that lies below the Winsock2 layer.
