Re: [PATCH] tracing/uprobes: Support ftrace_event_file base multibuffer

[Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>]

(2013/06/15 2:25), Paul E. McKenney wrote:
> On Fri, Jun 14, 2013 at 12:33:27PM -0400, Steven Rostedt wrote:
>> On Fri, 2013-06-14 at 09:21 -0700, Paul E. McKenney wrote:
>>
>>>>>> @@ -548,15 +556,35 @@ static void uprobe_trace_print(struct trace_uprobe *tu,
>>>>>>  /* uprobe handler */
>>>>>>  static int uprobe_trace_func(struct trace_uprobe *tu, struct pt_regs *regs)
>>>>>>  {
>>>>>> -	if (!is_ret_probe(tu))
>>>>>> -		uprobe_trace_print(tu, 0, regs);
>>>>>> +	struct ftrace_event_file **file;
>>>>>> +
>>>>>> +	if (is_ret_probe(tu))
>>>>>> +		return 0;
>>>>>> +
>>>>>> +	file = rcu_dereference_raw(tu->files);
>>>>
>>>> Why are you using rcu_dereference_raw() and not rcu_dereference(). The
>>>> _raw() is a bit special, and unless you know what you are doing with RCU
>>>> here, don't use it.
>>>>
>>>> As I see you using rcu_dereference_raw() all over this patch, along with
>>>> mutexes, I believe that you are not using RCU correctly here.
>>>
>>> If irqs and preempt are disabled, I suggest using rcu_dereference_sched().
>>> That is what it is there for.  ;-)
>>
>> I believe this just copied what kprobes did, where irqs and preemption
>> is disabled. I don't believe that these probes have that same luxury.
>>
>> But that begs the question. Should we convert the rcu_dereference_raw()
>> in the kprobe code to rcu_dereference_sched()?
> 
> It makes a lot of sense to me, at least assuming no issues with the
> interrupts being disabled, but the checks not spotting this.  Here
> is the check:
> 
> 	preempt_count() != 0 || irqs_disabled()
> 
> (With additional elaboration for if lockdep is enabled.)

OK, I see. So I'll convert all the rcu_dereference_raw() to
rcu_dereference_sched() except kprobe handler, because in the
kprobe handler above check always be true. :)

Thank you,


-- 
Masami HIRAMATSU
IT Management Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@hitachi.com