https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=250516
rai...@ultra-secure.de changed:
What|Removed |Added
CC||rai...@ultra-secure.de
---
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=256781
Roger Pau Monné changed:
What|Removed |Added
CC||roy...@freebsd.org
--- Comment #
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=256781
--- Comment #20 from Colin Percival ---
(In reply to Roger Pau Monné from comment #19)
> So this system reports invariant TSC, but the TSCs are not synchronized
> across CPUs?
My understanding is that it's even worse than that: The TSC i
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=256781
--- Comment #21 from Roger Pau Monné ---
(In reply to Colin Percival from comment #20)
That sounds like a bug in their implementation? As the hypervisor should keep
the TSC of the virtual CPU synchronised across physical CPU movements. Have
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=256781
--- Comment #22 from Colin Percival ---
They're the people who told me that we should use pvclock instead. ;-)
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=256781
--- Comment #23 from Roger Pau Monné ---
(In reply to Colin Percival from comment #22)
Oh OK. Maybe they don't do the vCPU TSC adjustments in order to avoid some of
the overhead of vCPU scheduling. Not sure exactly what are the
requirement/
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=256781
--- Comment #24 from martin.stoya...@vertalo.com ---
(In reply to Roger Pau Monné from comment #19)
That seems to help, although one has to edit /boot/loader.conf and reboot the
instance to apply it... Also my only test is just running date
