Re: [Simulavr-devel] two things

[John Regehr]

> Isn't it true that the irq should not even be put into the pending list
> unless the irq enable bit is set? I thought that was one of the
> conditions that must be met before an irq can be raised. Maybe I'm
> missing something...

If the interrupt is of the flag-bit type, then the interrupt should become
pending when a device decides to signal it, and then the interrupt should
fire later on if it is still pending and its enable bit becomes set.

I don't know a lot about simulavr's architecture, but it seems the right
implementation here is events attached to the flag bits: any time this bit
becomes set the interrupt should go on the pending list, and when the bit
becomes unset the interrupt should be removed from the pending list.

So signaling and firing should be decoupled I think.

John