IOS/KD: Implement NWC24_CHECK_MAIL_NOW

[noahpistilli]

This PR implements both a scheduler for WiiConnect24 as well as the NWC24_CHECK_MAIL_NOW ioctl.

I have decided to split #11746 into separate PR's for easier reviewing.

[iwubcode]

I think a slight optimization could be done here? If m_response_headers supported heterogeneous lookup you could use the name as the key in the map. I don't think this is blocking but something to consider.

[iwubcode]

Code LGTM. Untested

[AdmiralCurtiss]

I assume the file not existing is normal operation (eg. on initial system menu boot), but if it exists but isn't large enough should there be a error log? We have a lot of users with questionable NANDs out there...

[noahpistilli]

There probably should, if the file is corrupted the Wii Menu will display the System Files are corrupted error on any attempt to launch the message board.

[AdmiralCurtiss]

Untested, but this looks pretty good to me now. Is there a good way to test this as-is?

[noahpistilli]

You would have to use a NAND that is patched with RiiConnect24 mail, as well as a friend linked who can send you mail.

This link can patch if you supply the nwc24msg.cfg file.

I would have recommended WiiLink but their patcher is dependant on #11906.

[noahpistilli]

On another note, I am thinking that we should block starting the scheduler if the user is using the default console ID.

[AdmiralCurtiss]

Yes, blocking the Mail features for the default NAND makes sense.

[noahpistilli]

On another note, I am thinking that we should block starting the scheduler if the user is using the default console ID.

I wanted to future proof this for when we support downloading in the scheduler, so instead of blocking the scheduler starting, I just block dispatching the mail task.

[AdmiralCurtiss]

Tangential to this PR technically, but probably worth fixing before we do threading here...

dolphin/Source/Core/Core/IOS/Network/KD/NetKDRequest.cpp

Lines 516 to 520 in f063fb3

	// On a real Wii, GetSchedulerStat copies memory containing a list of error codes recorded by
	// KD among other things. In most instances there will never be more than one error code
	// recorded as we do not have a scheduler.
	const u32 out_size = std::min(request.buffer_out_size, 256U);
	memory.CopyToEmu(request.buffer_out, m_scheduler_buffer.data(), out_size);

This also doesn't seem to lock the m_scheduler_buffer_lock. The comment is also likely outdated with this PR.

[AdmiralCurtiss]

Okay, I think I'm good now. I don't have a setup for testing this so if you don't mind, please give it a go after the recent changes to see if everything still works as expected from your perspective, and if yes give me a poke to merge.