Re: [Qemu-devel] [PATCH RFC] migration: kvmclock: save and load the PVCLOCK_TSC_UNSTABLE_BIT flag when migration

[Jay Zhou <jianjay.zhou@huawei.com>]

Hi Radim,

On 2017/6/13 4:03, Radim Krčmář wrote:
> 2017-06-12 21:30+0800, Jay Zhou:
>> Guest using kvmclock will be hanged when migrating from unstable
>> tsc host to stable tsc host occasionally.
>> Sometimes, the tsc timestamp saved at the source side will be
>> backward when the guest stopped, and this value is transferred
>> to the destination side. The guest at the destination side thought
>> kvmclock is stable, so the protection mechanism against time
>> going backwards is not used.
>> When the first time vcpu0 enters the guest at the destination
>> side to update the wall clock, the result of
>> pvclock_clocksource_read will be backward occasionally,
>> which results in the wall clock drift.
>>
>> Signed-off-by: Jay Zhou <jianjay.zhou@huawei.com>
>> ---
>> diff --git a/hw/i386/kvm/clock.c b/hw/i386/kvm/clock.c
>>
>>       if (running) {
>>           struct kvm_clock_data data = {};
>> +        uint8_t flags_at_migration;
>>
>>           /*
>>            * If the host where s->clock was read did not support reliable
>>            * KVM_GET_CLOCK, read kvmclock value from memory.
>>            */
>>           if (!s->clock_is_reliable) {
>
> 'clock_is_reliable = true' on all newer KVMs (v4.9+), so I don't see a
> reason to add a feature that can't be used.

After rereading the codes, yes, I agree. Sorry to disturb.

>
>> -            uint64_t pvclock_via_mem = kvmclock_current_nsec(s);
>> +            uint64_t pvclock_via_mem = kvmclock_current_nsec(s,
>> +                                                    &flags_at_migration);
>
> kvmclock_current_nsec() was introduced to work around the problem with
> backward time, so we should understand why it returns a time that is
> backwards if we want to do something for old KVMs ...
>
> Is pvclock_via_mem < s->clock?

Please see the other thread I replied to you when the bug occured.


Regards,
Jay Zhou