[PATCH 0/4] io-wq/io_uring locking optimisation

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

[1-3] are shedding excessive locking of @wqe's and @worker's spinlocks
from io_worker_handle_work().

[4] removes an extra pair of refcount get/put by making former io_put_work()
to own the submission reference. It also changes io-wq get/put API
and renames io_put_work() into io_free_work() to reflect it.

Pavel Begunkov (4):
  io-wq: shuffle io_worker_handle_work() code
  io-wq: optimise locking in io_worker_handle_work()
  io-wq: optimise out *next_work() double lock
  io_uring/io-wq: forward submission ref to async

 fs/io-wq.c    | 148 ++++++++++++++++++++++++++------------------------
 fs/io-wq.h    |   6 +-
 fs/io_uring.c |  31 ++++-------
 3 files changed, 90 insertions(+), 95 deletions(-)

-- 
2.24.0