On 22/03/2020 19:09, Pavel Begunkov wrote:
> On 19/03/2020 21:56, Jens Axboe wrote:
>> We always punt async buffered writes to an io-wq helper, as the core
>> kernel does not have IOCB_NOWAIT support for that. Most buffered async
>> writes complete very quickly, as it's just a copy operation. This means
>> that doing multiple locking roundtrips on the shared wqe lock for each
>> buffered write is wasteful. Additionally, buffered writes are hashed
>> work items, which means that any buffered write to a given file is
>> serialized.
>>
>> When looking for a new work item, build a chain of identicaly hashed
>> work items, and then hand back that batch. Until the batch is done, the
>> caller doesn't have to synchronize with the wqe or worker locks again.

I have an idea, how to do it a bit better. Let me try it.


-- 
Pavel Begunkov