i folks, In this proposed patch, I am introducing a worker fired by kvmclock that updates guest wallclock periodically to reflect changes in the host's wallclock. With this patch, a large pool of VMs will no longer have to run NTP in all of its guests.
The worker does that at a configurable interval, with a minimum granularity of 1 second. So, although not exactly cheap, the msr write needed to get an updated wallclock value won't pose a heavy burden on the system. It is also possible to disable it completely if this behaviour is undesired for a specific scenario. Changes from v1: * disabled by default * adjust clock in a loop, to prevent agaist host scheduling. diffstat follows: arch/x86/include/asm/kvm_para.h | 6 +++ arch/x86/kernel/kvmclock.c | 85 ++++++++++++++++++++++++++++++++++---- kernel/sysctl.c | 13 ++++++ 3 files changed, 95 insertions(+), 9 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
