Implement UDS service with example

[Jhynjhiruu]

This PR adds basic local networking functionality with the UDS service.
It's very much not feature-complete, but enough is implemented to be able to interface with the devkitPRO local networking example - i.e. joining a network, creating a network, sending and pulling packets, and basic configuration. This is probably enough functionality for most homebrew applications, and I plan to try and implement a Download Play sniffer with it.

Todo: finish implementing the rest of the functions; test properly; cleanup code?

An example is included (local-networking.rs; this is a mostly-but-not-entirely-complete port of the devkitPRO local networking example).

[Meziu]

Great PR! A bit long for my liking, but I saw you followed our standards for the public API. Sadly, I have no way to thoroughly test the changes since I can't make a local network (I only have one 3DS). Is there any way I can test things out, maybe with an emulator?

[Meziu]

@AzureMarker Ehm, do you remember why a closure is used here? It doesn't seem to be necessary.

[AzureMarker]

This was done so I could use ?, like the unstable try block.

[Jhynjhiruu]

This change was the easiest way to fix a compile error introduced by my implementation of FromResidual. If I handle errors a different way, this change can be reverted.

[Meziu]

I believe so, the internal ctru_sys::udsConnectionStatus contains informations specific to an existing network.

[Jhynjhiruu]

It seems like calling it when not connected actually returns success. On New 3DS, I get

ConnectionStatus {
    status: 3,
    unk_x4: 0,
    cur_node_id: 2,
    unk_xa: 0,
    unk_xc: [0, 0, 0, 0, 0, 0, 0, 0],
    total_nodes: 0,
    max_nodes: 0,
    node_bitmask: 0,
}

and on Old 3DS I get

ConnectionStatus {
    status: 3,
    unk_x4: 0,
    cur_node_id: 0,
    unk_xa: 0,
    unk_xc: [0, 0, 0, 0, 0, 0, 0, 0],
    total_nodes: 0,
    max_nodes: 0,
    node_bitmask: 0,
}

consistently (they're identical besides cur_node_id).

[Meziu]

Interesting. At this point I wonder whether the status member itself specifies whether or not the client is connected to a network. We should probably try to understand its meaning.

[Jhynjhiruu]

libctru doesn't describe it at all, so it would very much be a reverse-engineering effort. From my 20 seconds of experimentation, it looks like a value of 6 represents connected. Presumably it's a bitfield.

[ian-h-chamberlain]

I think I found some details, seems like it's more of a weird enum with sparse value choices? https://www.3dbrew.org/wiki/NWMUDS:GetConnectionStatus#Status_values

[Meziu]

The TODO doc comment should be removed now that it's fixed.

[Jhynjhiruu]

Is there any way I can test things out, maybe with an emulator?

I've looked a little bit into it and I think it's maybe possible to connect to a 3DS's network with another device, but it'd require more knowledge about how Wi-Fi works than I have.
I think Citra supports relaying local networking data for multiplayer, but I don't know how to set it up.

[Jhynjhiruu]

Functions left to implement:

• udsGenerateNodeInfo (not really intended for use standalone, I think, and doesn't really need to be in the safe API)
• udsGetNodeInfoUsername (the safe API replaces the NodeInfo type and that decodes the string, so there's no need for this)
• udsCheckNodeInfoInitialized (I have no idea why this is supposed to be useful, since you always know this already)
• udsGenerateDefaultNetworkStruct (also not really intended for use standalone, used as part of Uds::new())
• udsWaitDataAvailable (would be good to have, but not really needed since you can just check whether Uds::pull_packet() returns Some or None)
• udsEjectClient
• udsEjectSpectator
• udsUpdateNetworkAttribute (definitely not really intended for use standalone, other functions call this internally)
• udsSetNewConnectionsBlocked
• udsAllowSpectators
• udsGetNodeInformation

So, only about 5½ that would be actually useful, and they're mostly trivial.

[ian-h-chamberlain]

I think I found some details, seems like it's more of a weird enum with sparse value choices? https://www.3dbrew.org/wiki/NWMUDS:GetConnectionStatus#Status_values

[ian-h-chamberlain]

What does this status represent? it still seems like it would be an error of some kind but we
return Ok for this?

[Jhynjhiruu]

See libctru and the UDS example. There's probably a nicer way to handle this, but this does work.

[ian-h-chamberlain]

Ok, so this error code basically means a packet was still sent but some other error occurred, I guess? Seems fine

[Jhynjhiruu]

Yeah, something like that. Maybe it'd be a good idea to hardcode it as a constant somewhere rather than calling it in the function, though - MAKERESULT presumably isn't const so it'd have to be a magic number with some explanation.

[ian-h-chamberlain]

If it isn't const, we probably can/should make it const, same for any of those other helper functions that just do bit operations on Result

[Jhynjhiruu]

Presumably, this would need to be changed in ctru_sys. Once somebody does that, I'll change this to be a little more elegant.

[ian-h-chamberlain]

Wow, this is a pretty intense example but it seems necessary given how many features there are in the uds service. I was wondering if some of this functionality could go onto Uds itself instead of just in this example, but honestly on first read I don't see any obvious things that make sense, so 👍 good with me

[adryzz]

This looks a lot like a manually implemented future state machine. could this have worked better with async?

[Jhynjhiruu]

Very possibly! I'm scared of async.

[Jhynjhiruu]

I've done some basic reading about futures, and I don't understand how this could be done with async. Could you explain how that would work? It would definitely be nice to clean up the code if it's possible.

[Meziu]

The TODO doc comment should be removed now that it's fixed.