Re: [Qemu-devel] [PATCH v4 0/7] trace: Generic event state description

[Stefan Hajnoczi]

On Mon, Jun 11, 2012 at 2:52 PM, Lluís Vilanova <address@hidden> wrote:
> Stefan Hajnoczi writes:
>
>> On Mon, Jun 11, 2012 at 12:12 PM, Lluís Vilanova <address@hidden> wrote:
>>> Stefan Hajnoczi writes:
>>>
>>>> On Tue, May 8, 2012 at 3:38 PM, Lluís Vilanova <address@hidden> wrote:
>>>>> Provides a generic event state description and a more detailed event 
>>>>> control and
>>>>> query interface.
>>>>>
>>>>> Signed-off-by: Lluís Vilanova <address@hidden>
>>>>> ---
>>> [...]
>>>> What is the point of sstate vs dstate?  It seems the dynamic state is
>>>> what can be toggled and sstate is the "disable" keyword.  Why have
>>>> sstate since there is already a macro?
>>>
>>> 'sstate' is there just in case you query the tracing state of an event 
>>> through
>>> the 'TraceEvent' structure instead of through the corresponding macro.
>
>> If this is not used by a tracer today we should hold off until it's needed.
>
> I double-checked the code for uses of that and it appears to be there just for
> the sake of completeness.
>
> Still, I realized I should have added a check in 
> 'trace_event_set_state_dynamic'
> and assert that the event is statically enabled (otherwise dynamically 
> enabling
> an event that is statically disabled just does not make sense).
>
> Whatever you prefer, although the code is pretty simple.

I'd like to avoid adding things that are unused - even if they are small.

Stefan