diff options
| author | Pavel Begunkov <asml.silence@gmail.com> | 2023-04-06 14:20:12 +0100 |
|---|---|---|
| committer | Jens Axboe <axboe@kernel.dk> | 2023-04-06 16:23:28 -0600 |
| commit | 8751d15426a31baaf40f7570263c27c3e5d1dc44 (patch) | |
| tree | ce78034f5b032052374c63e97c403143511365a2 /io_uring/io_uring.h | |
| parent | 5150940079a3ce94d7474f6f5b0d6276569dc1de (diff) | |
io_uring: reduce scheduling due to tw
Every task_work will try to wake the task to be executed, which causes
excessive scheduling and additional overhead. For some tw it's
justified, but others won't do much but post a single CQE.
When a task waits for multiple cqes, every such task_work will wake it
up. Instead, the task may give a hint about how many cqes it waits for,
io_req_local_work_add() will compare against it and skip wake ups
if #cqes + #tw is not enough to satisfy the waiting condition. Task_work
that uses the optimisation should be simple enough and never post more
than one CQE. It's also ignored for non DEFER_TASKRUN rings.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/d2b77e99d1e86624d8a69f7037d764b739dcd225.1680782017.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Diffstat (limited to 'io_uring/io_uring.h')
| -rw-r--r-- | io_uring/io_uring.h | 9 |
1 files changed, 9 insertions, 0 deletions
diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h index cb4309a2acdc..ef449e43d493 100644 --- a/io_uring/io_uring.h +++ b/io_uring/io_uring.h @@ -18,6 +18,15 @@ enum { /* don't use deferred task_work */ IOU_F_TWQ_FORCE_NORMAL = 1, + + /* + * A hint to not wake right away but delay until there are enough of + * tw's queued to match the number of CQEs the task is waiting for. + * + * Must not be used wirh requests generating more than one CQE. + * It's also ignored unless IORING_SETUP_DEFER_TASKRUN is set. + */ + IOU_F_TWQ_LAZY_WAKE = 2, }; enum { |