[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-sctp
Subject:    Re: [PATCH net-next] sctp: move trace_sctp_probe_path into sctp_outq_sack
From:       Kevin Kou <qdkevin.kou () gmail ! com>
Date:       2019-12-27 2:47:51
Message-ID: 647b9a55-6236-5d67-6ba8-8a62e828f89c () gmail ! com
[Download RAW message or body]



On 2019/12/27 10:28, Marcelo Ricardo Leitner wrote:
> On Thu, Dec 26, 2019 at 03:38:35PM -0800, David Miller wrote:
>> From: Kevin Kou <qdkevin.kou@gmail.com>
>> Date: Fri, 27 Dec 2019 07:09:07 +0800
>>
>>>
>>>
>>>> From: Kevin Kou <qdkevin.kou@xxxxxxxxx>
>>>> Date: Thu, 26 Dec 2019 12:29:17 +0000
>>>>
>>>>> This patch is to remove trace_sctp_probe_path from the TP_fast_assign
>>>>> part of TRACE_EVENT(sctp_probe) to avoid the nest of entry function,
>>>>> and trigger sctp_probe_path_trace in sctp_outq_sack.
>>>> ...
>>>>
>>>> Applied, but why did you remove the trace enabled check, just out of
>>>> curiosity?
>>>
>>> Actually, the check in trace_sctp_probe_path_enabled also done in
>>> trace_sctp_probe_path according to the Macro definition, both check
>>> if (static_key_false(&__tracepoint_##name.key)).
>>
>> Indeed, thanks for the explanation.
> 
> It was duplicated, yes, but it was also a small optimization:
> 
> if (enabled) {
>    for (X times) {
>      if (enabled) {
>      }
>    }
> }
> 
> So it wouldn't traverse the list if not needed.  But X is usually 1 or
> 2 and this list is already traversed multiple times in this code path.
> 

Yes, It is a small optimization indeed. let me changed a little, do 
testing and resend another patch. Thanks for your comments, Marcelo.


[prev in list] [next in list] [prev in thread] [next in thread]