CCing Marcelo, who wrote kvm_update_clock() and kvmclock_pre_save(). On Thu, Aug 29, 2019 at 06:07:11PM -0300, Maxiwell S. Garcia wrote: > The clock move makes the guest knows about the paused time between the > 'stop' and 'migrate' commands. This is an issue in an already-paused > VM because some side effects, like process stalls, could happen > after migration. > > So, this patch checks the runstate of guest in the pre_save handler and > do not re-reads the clock in case of paused state (cold migration). > > Signed-off-by: Maxiwell S. Garcia <maxiw...@linux.ibm.com> > --- > hw/i386/kvm/clock.c | 15 +++++++++++---- > 1 file changed, 11 insertions(+), 4 deletions(-) > > diff --git a/hw/i386/kvm/clock.c b/hw/i386/kvm/clock.c > index 80c133a724..2c59b6894b 100644 > --- a/hw/i386/kvm/clock.c > +++ b/hw/i386/kvm/clock.c > @@ -41,6 +41,9 @@ typedef struct KVMClockState { > uint64_t clock; > bool clock_valid; > > + /* whether the 'clock' value was obtained in the 'paused' state */ > + bool runstate_paused; > + > /* whether machine type supports reliable KVM_GET_CLOCK */ > bool mach_use_reliable_get_clock; > > @@ -202,6 +205,8 @@ static void kvmclock_vm_state_change(void *opaque, int > running, > return; > } > > + s->runstate_paused = runstate_check(RUN_STATE_PAUSED); > + > kvm_synchronize_all_tsc(); > > kvm_update_clock(s); > @@ -260,9 +265,9 @@ static int kvmclock_pre_load(void *opaque) > } > > /* > - * When migrating, read the clock just before migration, > - * so that the guest clock counts during the events > - * between: > + * When migrating a running guest, read the clock just > + * before migration, so that the guest clock counts > + * during the events between: > * > * * vm_stop() > * * > @@ -277,7 +282,9 @@ static int kvmclock_pre_save(void *opaque) > { > KVMClockState *s = opaque; > > - kvm_update_clock(s); > + if (!s->runstate_paused) { > + kvm_update_clock(s); > + } > > return 0; > } > -- > 2.20.1 >
-- Eduardo