arm/arm64: KVM: Properly account for guest CPU time
authorChristoffer Dall <christoffer.dall@linaro.org>
Thu, 28 May 2015 18:49:10 +0000 (19:49 +0100)
committerMarc Zyngier <marc.zyngier@arm.com>
Wed, 17 Jun 2015 08:46:28 +0000 (09:46 +0100)
commit1b3d546daf85ed2bc9966e12cee3e6435fb65eca
tree70660cf7577fe60cc374fbf39595bdb1d883e8d4
parentea2c6d9745c6698d9f820bc230aa1a80d9e908ac
arm/arm64: KVM: Properly account for guest CPU time

Until now we have been calling kvm_guest_exit after re-enabling
interrupts when we come back from the guest, but this has the
unfortunate effect that CPU time accounting done in the context of timer
interrupts occurring while the guest is running doesn't properly notice
that the time since the last tick was spent in the guest.

Inspired by the comment in the x86 code, move the kvm_guest_exit() call
below the local_irq_enable() call and change __kvm_guest_exit() to
kvm_guest_exit(), because we are now calling this function with
interrupts enabled.  We have to now explicitly disable preemption and
not enable preemption before we've called kvm_guest_exit(), since
otherwise we could be preempted and everything happening before we
eventually get scheduled again would be accounted for as guest time.

At the same time, move the trace_kvm_exit() call outside of the atomic
section, since there is no reason for us to do that with interrupts
disabled.

Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
arch/arm/kvm/arm.c
This page took 0.025516 seconds and 5 git commands to generate.