Editorial suggestions

[littledan]

Consider reusing sync generator internal slots, and using a special kind field to differentiate (I suggested something similar for SharedArrayBuffer, though it apparently caused problems; I think it'd be good to reduce spec code duplication, but maybe not worth it given the hazards encountered there).

[littledan]

The await fulfilled/rejected functions don't say what they return. Other places in ECMA262 return a value after "resuming" to another execution context. Consider making them explicitly return undefined. (FWIW, V8's implementation of async/await is structured this way as well, and I had to include an explicit 'return undefined' after resuming to the other context).

[domenic]

Fixed the return values.

For the reuse of slots, as far as I can tell that would not be much of an improvement. It would allow consolidating a single [[AsyncGeneratorState]] vs. [[GeneratorState]] usage in this spec, but would require updating the existing spec's [[GeneratorState]] checks to instead be something like [[GeneratorState]] + no [[AsyncGeneratorQueue]]. Let me know if you see things differently.