[ovs-dev] [PATCH v14 0/5] dpcls func ptrs & optimizations

Ilya Maximets i.maximets at samsung.com
Fri Jul 19 11:10:08 UTC 2019


On 19.07.2019 14:00, Stokes, Ian wrote:
> 
> 
>> -----Original Message-----
>> From: Ilya Maximets [mailto:i.maximets at samsung.com]
>> Sent: Friday, July 19, 2019 11:57 AM
>> To: Stokes, Ian <ian.stokes at intel.com>; Van Haaren, Harry
>> <harry.van.haaren at intel.com>; dev at openvswitch.org
>> Subject: Re: [ovs-dev] [PATCH v14 0/5] dpcls func ptrs & optimizations
>>
>> On 19.07.2019 11:27, Ian Stokes wrote:
>>> On 7/19/2019 9:08 AM, Ilya Maximets wrote:
>>>> On 19.07.2019 9:58, Ian Stokes wrote:
>>>>> On 7/18/2019 3:30 PM, Ian Stokes wrote:
>>>>>> On 7/18/2019 2:03 PM, Harry van Haaren wrote:
>>>>>>> Hey Folks,
>>>>>>>
>>>>>>> Here a v14 of the DPCLS Function Pointer patchset, as has been
>>>>>>> presented at OVS Conf in Nov '18, and discussed on the ML since
>> then.
>>>>>>> I'm aware of the soft-freeze for 2.12, I feel this patchset has had
>>>>>>> enough reviews/versions/testing to be merged in 2.12.
>>>>>>>
>>>>>>> Thanks Ilya for input and suggestions on v13. Only change is that
>>>>>>> the implementation of the blocks array memory allocation is now
>>>>>>> using DEFINE_PER_THREAD_MALLOCED_DATA() macro, allowing for proper
>>>>>>> de-allocation of the allocated memory after a thread exits.
>>>>>>>
>>>>>>> Regards, -Harry
>>>>>>
>>>>>> Thanks Harry for the v14. Just testing it and it seems in good shape
>> to me.
>>>>>>
>>>>>> Thanks Ilya for the suggestions around the per thread allocation for
>> scratch as well, this feedback has helped and is resolving multithread
>> issues spotted in the earlier series.
>>>>>>
>>>>>> I feel the patchset is now in a state that it should be considered
>> for the 2.12 release.
>>>>>>
>>>>>> We've had testing from a number of groups over the series (Red Hat,
>> Arm, Intel, Samsung) with positive results across the board.
>>>>>>
>>>>>> It also opens up opportunities for further improving the dpcls in the
>> 2.13 release next year.
>>>>>>
>>>>>> As such I think it's worth an exception to merge before Mondays
>> feature freeze so that it benefits from the 4 week settling period before
>> release.
>>>>>>
>>>>>> Are there any objections to this?
>>>>>>
>>>>>> Thanks
>>>>>> Ian
>>>>>
>>>>> Hi All,,
>>>>>
>>>>> I haven't heard any objections to this since yesterday. My intentions
>> are to merge this today.
>>>>
>>>> Hi. Sorry, it was only 2.5 hours of my working time since your previous
>> mail.
>>>> Are you at office 24/7?
>>>
>>> Really? I thought my last email was yesterday evening, maybe there was a
>> delay in it going out on my side, it's OVS release crunch time so  tend to
>> lose track of time, I'm pretty sure I slept for a few hours since the last
>> mail :)
>>>
>>>> I'm testing the v14 now and will reply with results in a couple of
>> hours.
>>>> Is it OK for you?
>>>>
>>>
>>> Thanks for testing on the latest revsion Ilya, much appreciated. If you
>> come across anything untowards today let us know :).
>>
>> I finished testing v14. No issues observed. Slight performance difference
>> with a previous versions, but nothing significant. So, it's OK for me.
> 
> Thanks Ilya, can I add your ack for the series with the amendment on patch 5 you flagged?

OK. Sure.

> 
> Regards
> Ian
> 


More information about the dev mailing list