KVM: x86: Restrict writeback of VCPU state
commitea64305139357e89f58fc05ff5d48dc233d44d87
authorJan Kiszka <jan.kiszka@siemens.com>
Mon, 1 Mar 2010 18:10:31 +0000 (1 19:10 +0100)
committerMarcelo Tosatti <mtosatti@redhat.com>
Thu, 4 Mar 2010 03:29:30 +0000 (4 00:29 -0300)
tree0b3ebd229d76fc2f770dc3228dcd81379a362a14
parentea375f9ab8c76686dca0af8cb4f87a4eb569cad3
KVM: x86: Restrict writeback of VCPU state

Do not write nmi_pending, sipi_vector, and mpstate unless we at least go
through a reset. And TSC as well as KVM wallclocks should only be
written on full sync, otherwise we risk to drop some time on state
read-modify-write.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
target-i386/kvm.c