This patch enables vCPU notification to wake it up
when new async event comes in replay mode.

The motivation of this patch is the following.
Consider recorded block async event. It is saved into the log
with one of the checkpoints. This checkpoint may be passed in
vCPU loop. In replay mode when this async event is read from
the log, and block thread task is not finished yet, vCPU thread
goes to sleep. That is why this patch adds waking up the vCPU
to process this finished event.

Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgalyuk@ispras.ru>
---
 replay/replay-events.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/replay/replay-events.c b/replay/replay-events.c
index a1c6bb934e..92dc800219 100644
--- a/replay/replay-events.c
+++ b/replay/replay-events.c
@@ -126,6 +126,7 @@ void replay_add_event(ReplayAsyncEventKind event_kind,
 
     g_assert(replay_mutex_locked());
     QTAILQ_INSERT_TAIL(&events_list, event, events);
+    qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
 }
 
 void replay_bh_schedule_event(QEMUBH *bh)

On 29/03/21 09:43, Pavel Dovgalyuk wrote:
> This patch enables vCPU notification to wake it up
> when new async event comes in replay mode.
> 
> The motivation of this patch is the following.
> Consider recorded block async event. It is saved into the log
> with one of the checkpoints. This checkpoint may be passed in
> vCPU loop. In replay mode when this async event is read from
> the log, and block thread task is not finished yet, vCPU thread
> goes to sleep. That is why this patch adds waking up the vCPU
> to process this finished event.
> 
> Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgalyuk@ispras.ru>
> ---
>   replay/replay-events.c |    1 +
>   1 file changed, 1 insertion(+)
> 
> diff --git a/replay/replay-events.c b/replay/replay-events.c
> index a1c6bb934e..92dc800219 100644
> --- a/replay/replay-events.c
> +++ b/replay/replay-events.c
> @@ -126,6 +126,7 @@ void replay_add_event(ReplayAsyncEventKind event_kind,
>   
>       g_assert(replay_mutex_locked());
>       QTAILQ_INSERT_TAIL(&events_list, event, events);
> +    qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
>   }

I still do not understand why the qemu_clock_notify function is used to 
notify the CPU.  It would be nice to have a description of the call 
chain in a comment or to have a more appropriate API.  I can queue the 
patch since it's deep in replay-events.c and not in common code, but if 
you post a short description here I'll integrate it.

Paolo

On 29.03.2021 13:29, Paolo Bonzini wrote:
> On 29/03/21 09:43, Pavel Dovgalyuk wrote:
>> This patch enables vCPU notification to wake it up
>> when new async event comes in replay mode.
>>
>> The motivation of this patch is the following.
>> Consider recorded block async event. It is saved into the log
>> with one of the checkpoints. This checkpoint may be passed in
>> vCPU loop. In replay mode when this async event is read from
>> the log, and block thread task is not finished yet, vCPU thread
>> goes to sleep. That is why this patch adds waking up the vCPU
>> to process this finished event.
>>
>> Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgalyuk@ispras.ru>
>> ---
>>   replay/replay-events.c |    1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/replay/replay-events.c b/replay/replay-events.c
>> index a1c6bb934e..92dc800219 100644
>> --- a/replay/replay-events.c
>> +++ b/replay/replay-events.c
>> @@ -126,6 +126,7 @@ void replay_add_event(ReplayAsyncEventKind 
>> event_kind,
>>       g_assert(replay_mutex_locked());
>>       QTAILQ_INSERT_TAIL(&events_list, event, events);
>> +    qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
>>   }
> 
> I still do not understand why the qemu_clock_notify function is used to 
> notify the CPU.  It would be nice to have a description of the call 
> chain in a comment or to have a more appropriate API.  I can queue the 
> patch since it's deep in replay-events.c and not in common code, but if 
> you post a short description here I'll integrate it.

Maybe I missed something.
This call is needed for the case when vCPU is waiting in rr_wait_io_event.
You mean that we can use qemu_cpu_kick instead?

Pavel Dovgalyuk

On 29/03/21 13:23, Pavel Dovgalyuk wrote:
>>
>> I still do not understand why the qemu_clock_notify function is used 
>> to notify the CPU.  It would be nice to have a description of the call 
>> chain in a comment or to have a more appropriate API.  I can queue the 
>> patch since it's deep in replay-events.c and not in common code, but 
>> if you post a short description here I'll integrate it.
> 
> Maybe I missed something.
> This call is needed for the case when vCPU is waiting in rr_wait_io_event.
> You mean that we can use qemu_cpu_kick instead?

I have no idea.  Using qemu_clock_notify however is definitely weird.

Paolo