Re: [PATCH] kernel: trace: remove __init from race_selftest_startup_function()and trace_selftest_startup_function_graph()

[Chen Gang]

On 07/12/2013 03:20 PM, Chen Gang wrote:
> On 07/12/2013 11:04 AM, Chen Gang wrote:
>> On 07/12/2013 10:38 AM, Steven Rostedt wrote:
>>> On Fri, 2013-07-12 at 09:58 +0800, Chen Gang wrote:
>>>> On 07/12/2013 09:41 AM, Steven Rostedt wrote:
>>>>> On Fri, 2013-07-12 at 07:51 +0800, Chen Gang wrote:
>>>>>
>>>>>>> Hmm, can all trace_selftest_startup_* (*selftest* in trace_selftest.c)
>>>>>>> use '__init', so not waste memory keeping them around ?
>>>>> Yeah, they should all be set to __init, but that's pretty low on the
>>>>> totem poll, as distros don't enable selftests in their main kernels.
>>>>
>>>> Excuse me, my English is not quite well, I guess your meaning is:
>>>>
>>>>   they should all be set to '__init', although it is minor in real world.
>>>>
>>>> Is it correct ?
>>>
>>> Correct.
>>>
>>>>
>>>>
>>>> For me, I recommend to let all *selftest* as the same: "all add '
>>>> __init' or none add '__init'" (if choose add, all report warnings).
>>>>
>>>> Is it suitable to still send new related patch for it ? If so, could
>>>> you provide your suggesting choice (all add, or none add) ?
>>>
>>> Does this patch fix your warning?
>>>
>>> -- Steve
>>>
>>
> 
> After the test, they will not report the related warning.
> 
> Hmm..., but that will let another none *selftest* functions miss
> '__read_mostly'.
> 
> Do the original *selftest* intend to have no '__init' so can fit other
> none *selftest* with '__read_mostly', and without warnings ?
> 
> Welcome any members' suggestions or completions.
> 
> Thanks.
> 

Hello Frederic and Ingo:

Could you provide your suggestions or completions for it ?

The trace_selftest_startup_* funcitons are mostly added by you without
'__init', do you have additional considerations about it (intend to have
no '__init') ?

If no reply, I recommend to keep no '__init': apply this patch or
regress part of the patch "f1ed7c7 ftrace: Do not run selftest if
command line parameter is set" (at least, it can avoid related warnings
and treat all *selftest* fair).


Thanks.


>> I guess it can (although I do not give a compiling test), it seems a
>> better fixing.
>>
>> And is it suitable to let all *selftest* as the same ? (all add, or none
>> add '__init').
>>
>> Thanks.
>>
>>> diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
>>> index a4ed382..5e794d1 100644
>>> --- a/kernel/trace/trace.h
>>> +++ b/kernel/trace/trace.h
>>> @@ -680,6 +680,15 @@ extern int trace_selftest_startup_sched_switch(struct tracer *trace,
>>>  					       struct trace_array *tr);
>>>  extern int trace_selftest_startup_branch(struct tracer *trace,
>>>  					 struct trace_array *tr);
>>> +/*
>>> + * Tracer data references selftest functions that only occur
>>> + * on boot up. These can be __init functions. Thus, when selftests
>>> + * are enabled, then the tracers need to reference __init functions.
>>> + */
>>> +#define __tracer_data		__refdata
>>> +#else
>>> +/* Tracers are seldom changed. Optimize when selftests are disabled. */
>>> +#define __tracer_data		__read_mostly
>>>  #endif /* CONFIG_FTRACE_STARTUP_TEST */
>>>  
>>>  extern void *head_page(struct trace_array_cpu *data);
>>> diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c
>>> index b863f93..38fe148 100644
>>> --- a/kernel/trace/trace_functions.c
>>> +++ b/kernel/trace/trace_functions.c
>>> @@ -199,7 +199,7 @@ static int func_set_flag(u32 old_flags, u32 bit, int set)
>>>  	return 0;
>>>  }
>>>  
>>> -static struct tracer function_trace __read_mostly =
>>> +static struct tracer function_trace __tracer_data =
>>>  {
>>>  	.name		= "function",
>>>  	.init		= function_trace_init,
>>>
>>>
>>>
>>>
>>
>>
> 
> 


-- 
Chen Gang
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/