Re: [PATCH 3/7] vm-event: introduce vm_event_vcpu_enter

[Julien Grall <julien.grall@arm.com>]

Hello,

On 17/06/16 12:40, Corneliu ZUZU wrote:
> On 6/17/2016 11:55 AM, Julien Grall wrote:
>> On 16/06/16 15:08, Corneliu ZUZU wrote:
>>> diff --git a/xen/arch/arm/domain.c b/xen/arch/arm/domain.c
>>> index d31f821..ba248c8 100644
>>> --- a/xen/arch/arm/domain.c
>>> +++ b/xen/arch/arm/domain.c
>>> @@ -19,6 +19,7 @@
>>>   #include <xen/errno.h>
>>>   #include <xen/bitops.h>
>>>   #include <xen/grant_table.h>
>>> +#include <xen/vm_event.h>
>>>
>>>   #include <asm/current.h>
>>>   #include <asm/event.h>
>>> @@ -251,6 +252,8 @@ static void schedule_tail(struct vcpu *prev)
>>>
>>>       ctxt_switch_to(current);
>>>
>>> +    vm_event_vcpu_enter(current);
>>> +
>>>       local_irq_enable();
>>>
>>>       context_saved(prev);
>>> @@ -296,7 +299,7 @@ void context_switch(struct vcpu *prev, struct
>>> vcpu *next)
>>>
>>>   void continue_running(struct vcpu *same)
>>>   {
>>> -    /* Nothing to do */
>>> +    vm_event_vcpu_enter(same);
>>>   }
>>>
>>>   void sync_local_execstate(void)
>>
>> From my understanding of the commit message, vm_event_vcpu_enter
>> should be called before returning to the guest. The scheduling
>> functions are not called every-time Xen is returning to the guest. So
>> if you want to do every time Xen is re-entering to the guest, then
>> this should be done in leave_hypervisor_tail.
>>
>> Regards,
>>
>
> Which is also done, vm_event_vcpu_enter is called from
> leave_hypervisor_tail as well.

Sorry I did not spot this one.

> Are you saying that calling from leave_hypervisor_tail is enough and
> that the schedule_tail call is not necessary?

leave_hypervisor_tail will always be called before returning to the 
guest. So a call to vm_event_vcpu_enter in this function will be sufficient.

Regards,

-- 
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel