Re: [Qemu-devel] [PATCH RESEND v2 01/18] net/colo: Add notifier/callback related helpers for filter

[Jason Wang <jasowang@redhat.com>]

On 2017年04月26日 16:14, Hailiang Zhang wrote:
> On 2017/4/25 19:40, Jason Wang wrote:
>>
>> On 2017年04月22日 16:35, zhanghailiang wrote:
>>> We will use this notifier to help COLO to notify filter object
>>> to do something, like do checkpoint, or process failover event.
>>>
>>> Cc: Jason Wang<jasowang@redhat.com>
>>> Signed-off-by: zhanghailiang<zhang.zhanghailiang@huawei.com>
>>> Signed-off-by: Zhang Chen<zhangchen.fnst@cn.fujitsu.com>
>>> Signed-off-by: Li Zhijian<lizhijian@cn.fujitsu.com>
>>> ---
>>>    net/colo.c | 105 
>>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>>    net/colo.h |  19 +++++++++++
>>>    2 files changed, 124 insertions(+)
>> Went through this series and I feel that the code duplicates (at least
>> part) functionality of IOThread (iothread.c) . So I come to an idea 
>> that,
>> switch to use IOThread for COLO comparing thread then you can use aio bh
>> to do the inter process communication.
>>
>> Thoughts?
>
> Interesting idea, but after investigation, we found it will make 
> things quite complex
> if we use iothread for net compare (Actually, i don't think we can use 
> it directly for COLO),
> It seems to be special for block, you can see some codes call block 
> helpers.

You mean iothread_stop_all() which was called by main()? It looks just 
use to change aio context back to main context. And I don't think you 
need to care about that.

> Besides, we still have to realize a way to communicate with COLO frame 
> for filters  (I mean patch 16/17/18),
> even we use it after reconstruct those iothread part. ;)
>

I think you can use bh, and e.g go through all the filter in the bh.

Thanks

> Thanks,
> Hailiang
>
>> Thanks
>>
>> .
>>
>
>
>