[aon_timer,dv] Fix how we wait for intr_state to stop being busy

[rswarbrick]

The existing code doesn't actually work because there is nothing that consumes time and the EDA tool doesn't know when to re-evaluate the expression.

That causes warnings from VCS (at least). Avoid them by waiting more explicitly.

[antmarzam]

I presume you refer to the following warning?

  Function calls inside wait/event expression are sensitive only to arguments.
  this.ral.intr_state.is_busy() is called without any arguments which may not 
  unblock wait/event expression

Not sure why VCS complains about this, I've seen it supported in multiple simulators and for the UVM RAL it seems the recommendation is to wait until is_busy=0 . is_busy only wraps around m_is_busy from uvm_reg.

I think your suggestion would work, but would also force to wait until the next clock tick, when m_is_busy can be released before the next tick.

How about we replace: wait (ral.intr_state.is_busy() == 0);
With: wait (ral.intr_state.m_is_busy == 0) ? It seems simpler and doesn't have to wait until the next tick?

In uvm_reg, you'll see m_is_busy was intended to be a local variable but the code has been commented out:
/*local*/ bit m_is_busy; I wonder if the warning is related to why they haven't yet uncommented the local keyword

[rswarbrick]

I think this is handled more cleanly in @antmarzam's #26167. Let's do it that way :-)