Re: [PATCH v2 1/1] io_uring: allocate memory for overflowed CQEs

[Pavel Begunkov <asml.silence@gmail.com>]

On 24/02/2021 03:18, Jens Axboe wrote:
> On 2/23/21 8:06 PM, Hao Xu wrote:
>> 在 2021/2/23 下午8:40, Pavel Begunkov 写道:
>>> Instead of using a request itself for overflowed CQE stashing, allocate
>>> a separate entry. The disadvantage is that the allocation may fail and
>>> it will be accounted as lost (see rings->cq_overflow), so we lose
>>> reliability in case of memory pressure. However, it opens a way for for
>>> multiple CQEs per an SQE and even generating SQE-less CQEs >
>>> Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
>>> ---
>> Hi Pavel,
>> Allow me to ask a stupid question, why do we need to support multiple 
>> CQEs per SQE or even SQE-less CQEs in the future?
> 
> Not a stupid question at all, since it's not something we've done
> before. There's been discussion about this in the past, in the presence
> of the zero copy IO where we ideally want to post two CQEs for an SQE.
> Most recently I've been playing with multishot poll support, where a
> POLL_ADD will stay active after triggering. Hence you could be posting
> many CQEs for that SQE, over the life time of the request.

Yep, in addition should be useful for eBPF requests.

-- 
Pavel Begunkov