Re: [PATCH mptcp-next v4 1/2] mptcp: refactor MP_FAIL response timeout

[Mat Martineau <mathew.j.martineau@linux.intel.com>]

On Thu, 9 Jun 2022, Paolo Abeni wrote:

> On Thu, 2022-06-09 at 08:00 +0800, Geliang Tang wrote:
>> mptcp_mp_fail_no_response shouldn't be invoked on each worker run, it
>> should be invoked only when MP_FAIL response timeout occurs.
>>
>> This patch refactors the MP_FAIL response timeout logic. Add fail_tout
>> in mptcp_sock to record the MP_FAIL timestamp. Check it in mptcp_worker()
>> before invoking mptcp_mp_fail_no_response(). Drop the code to reuse
>> sk_timer for MP_FAIL response.
>
> I'm sorry for not noticing it on the previous version - likely for the
> lack of clarity on my side: we still need to start the timer for
> mp_fail response: otherwise we are not assured that the mptcp_worker
> will be triggered.
>

(capturing this on the mailing list in addition to the recent meeting)

Yes, I agree. I saw that the timer stop calls were removed but missed the 
timer reset removal. It is imporant for the timer to still be started.

> Additionally we need some more care to manipulate the timer reset. I
> think it's easier if I send a squash-to patch - assuming I'll have some
> time for that early next week.
>

Thanks Paolo!

--
Mat Martineau
Intel