On 01/03/2020 19:18, Pavel Begunkov wrote: > There are several independent parts in the patchset, but bundled > to make a point. > 1-2: random stuff, that implicitly used later. > 3-5: restore @nxt propagation > 6-8: optimise locking in io_worker_handle_work() > 9: optimise io_uring refcounting > > The next propagation bits are done similarly as it was before, but > - nxt stealing is now at top-level, but not hidden in handlers > - ensure there is no with REQ_F_DONT_STEAL_NEXT > > [6-8] is the reason to dismiss the previous @nxt propagation appoach, > I didn't found a good way to do the same. Even though it looked > clearer and without new flag. > > Performance tested it with link-of-nops + IOSQE_ASYNC: > > link size: 100 > orig: 501 (ns per nop) > 0-8: 446 > 0-9: 416 > > link size: 10 > orig: 826 > 0-8: 776 > 0-9: 756 BTW, that's basically QD1, and with contention for wqe->lock the gap should be even wider. > > Pavel Begunkov (9): > io_uring: clean up io_close > io-wq: fix IO_WQ_WORK_NO_CANCEL cancellation > io_uring: make submission ref putting consistent > io_uring: remove @nxt from handlers > io_uring: get next req on subm ref drop > io-wq: shuffle io_worker_handle_work() code > io-wq: io_worker_handle_work() optimise locking > io-wq: optimise double lock for io_get_next_work() > io_uring: pass submission ref to async > > fs/io-wq.c | 162 ++++++++++++---------- > fs/io_uring.c | 366 ++++++++++++++++++++++---------------------------- > 2 files changed, 258 insertions(+), 270 deletions(-) > -- Pavel Begunkov
