svn commit: r325108 - head/sys/amd64/vmm/io

[Peter Grehan]

Hi Ian,

> In fact I did check maintainers, and I just now checked it again, and
> seeing nothing on-point, wrote a really snarky reply.
> 
> Luckily, before hitting send I realized you wouldn't have said anything
> unless there was an entry in there, so I checked like 4 more times
> before I noticed the line that begins vmm(4).  I had been looking for a
> line that began with sys/amd64.
> 
> So... sorry about that.  Should I revert it and open a review?

  No, it's fine. It's a good idea in general for code in that area since 
there are often unintended side-effects with guest behaviour, and 
putting up a review can at least flush out folk who might have more 
insight into that.

  I'll update the maintainer's file to include the directory.

> Longer-term, I was hoping to find some time over the next few weeks to
> further explore the roundoff errors in the timers and attempt a better
> fix, and I was definitely intending to put that up for review
> (especially because it involves math, and I'm always better off asking
> for help with that).

  I know Rod Grimes is going to be doing some testing this week with 
clock drift in guests, and I'll do some dtrace work on HPET usage with 
non-FreeBSD guests.

  Timer accuracy has long been an issue with bhyve so any help in this 
area is much appreciated.

later,

Peter.