On 3/3/2020 9:54 AM, Pavel Begunkov wrote:
> On 03/03/2020 07:26, Jens Axboe wrote:
>> On 3/2/20 1:45 PM, Pavel Begunkov wrote:
>>> Get next request when dropping the submission reference. However, if
>>> there is an asynchronous counterpart (i.e. read/write, timeout, etc),
>>> that would be dangerous to do, so ignore them using new
>>> REQ_F_DONT_STEAL_NEXT flag.
>>
>> Hmm, not so sure I like this one. It's not quite clear to me where we
>> need REQ_F_DONT_STEAL_NEXT. If we have an async component, then we set
>> REQ_F_DONT_STEAL_NEXT. So this is generally the case where our
>> io_put_req() for submit is not the last drop. And for the other case,
>> the put is generally in the caller anyway. So I don't really see what
>> this extra flag buys us?
>
> Because io_put_work() holds a reference, no async handler can achive req->refs
> == 0, so it won't return next upon dropping the submission ref (i.e. by
> put_find_nxt()). And I want to have next before io_put_work(), to, instead of as
> currently:
>
> run_work(work);
> assign_cur_work(NULL); // spinlock + unlock worker->lock
> new_work = put_work(work);
> assign_cur_work(new_work); // the second time
>
> do:
>
> new_work = run_work(work);
> assign_cur_work(new_work); // need new_work here
> put_work(work);
>
>
> The other way:
>
> io_wq_submit_work() // for all async handlers
> {
> ...
> // Drop submission reference.
> // One extra ref will be put in io_put_work() right
> // after return, and it'll be done in the same thread
> if (atomic_dec_and_get(req) == 1)
> steal_next(req);
> }
>
> Maybe cleaner, but looks fragile as well. Would you prefer it?
Any chance you've measured your next-work fix? I wonder how much does it
hurt performance, and whether we need a terse patch for 5.6.
>> Few more comments below.
>>
>>> +static void io_put_req_async_submission(struct io_kiocb *req,
>>> + struct io_wq_work **workptr)
>>> +{
>>> + static struct io_kiocb *nxt;
>>> +
>>> + nxt = io_put_req_submission(req);
>>> + if (nxt)
>>> + io_wq_assign_next(workptr, nxt);
>>> +}
>>
>> This really should be called io_put_req_async_completion() since it's
>> called on completion. The naming is confusing.
>
> Ok
>
>>> @@ -2581,14 +2598,11 @@ static void __io_fsync(struct io_kiocb *req)
>>> static void io_fsync_finish(struct io_wq_work **workptr)
>>> {
>>> struct io_kiocb *req = container_of(*workptr, struct io_kiocb, work);
>>> - struct io_kiocb *nxt = NULL;
>>>
>>> if (io_req_cancelled(req))
>>> return;
>>> __io_fsync(req);
>>> - io_put_req(req); /* drop submission reference */
>>> - if (nxt)
>>> - io_wq_assign_next(workptr, nxt);
>>> + io_put_req_async_submission(req, workptr);
>>> }
>>>
>>> static int io_fsync(struct io_kiocb *req, bool force_nonblock)
>>> @@ -2617,14 +2631,11 @@ static void __io_fallocate(struct io_kiocb *req)
>>> static void io_fallocate_finish(struct io_wq_work **workptr)
>>> {
>>> struct io_kiocb *req = container_of(*workptr, struct io_kiocb, work);
>>> - struct io_kiocb *nxt = NULL;
>>>
>>> if (io_req_cancelled(req))
>>> return;
>>> __io_fallocate(req);
>>> - io_put_req(req); /* drop submission reference */
>>> - if (nxt)
>>> - io_wq_assign_next(workptr, nxt);
>>> + io_put_req_async_submission(req, workptr);
>>> }
>>
>> All of these cleanups are nice (except the naming, as mentioned).
>>
>>> @@ -3943,7 +3947,10 @@ static int io_poll_add(struct io_kiocb *req)
>>> if (mask) {
>>> io_cqring_ev_posted(ctx);
>>> io_put_req(req);
>>> + } else {
>>> + req->flags |= REQ_F_DONT_STEAL_NEXT;
>>> }
>>> +
>>> return ipt.error;
>>> }
>>
>> Is this racy? I guess it doesn't matter since we're still holding the
>> completion reference.
>
> It's done by the same thread, that uses it. There could be a race if the async
> counterpart is going to change req->flags, but we tolerate false negative (i.e.
> put_req() will handle it).
>
--
Pavel Begunkov
