Add into_async for ParlIO

[bugadani]

Forgot to update parl_io in #2430. Still not sure what to do about TxOnly/RxOnly.

[Dominaezzz]

Still not sure what to do about TxOnly/RxOnly.

At some point we should support converting each half of the DMA channel to and from blocking mode (on chips that an support it), which should solve this I think.

[bugadani]

We can do some runtime checks on chips that don't support it (unless the presence of the async handler poses a problem in blocking mode).

[Dominaezzz]

unless the presence of the async handler poses a problem in blocking mode

It does, users could end up overriding the async handler, rendering the async driver broken.
Unless we implement some handler sharing logic like GPIO does but I don't like this direction.

[bugadani]

It does, users could end up overriding the async handler, rendering the async driver broken.

That's what I mean by runtime checks, I'd just add a panic to prevent this.

Unless we implement some handler sharing logic like GPIO does but I don't like this direction.

Me neither.

[Dominaezzz]

That's what I mean by runtime checks, I'd just add a panic to prevent this.

Oh right, got it. I suppose there's a risk that the async handler would clear the interrupt bits that the blocking driver is polling for maybe.

Also, if it's gonna be a panic, might as well use the type system to prevent it. i.e. use cfg to remove the APIs on chips that don't support it.

[bugadani]

Also, if it's gonna be a panic, might as well use the type system to prevent it. i.e. use cfg to remove the APIs on chips that don't support it.

"If it's gonna be a panic, don't use a panic" is a weird thing to say.

[JurajSadel]

Thanks!