diff options
| author | Jens Axboe <axboe@kernel.dk> | 2024-01-29 04:11:55 +0100 |
|---|---|---|
| committer | Jens Axboe <axboe@kernel.dk> | 2024-02-08 21:27:06 +0100 |
| commit | 521223d7c229f83915619f888c99e952f24dc39f (patch) | |
| tree | b19885749949ac462562ebc474b8964edc85331f /include | |
| parent | io_uring: expand main struct io_kiocb flags to 64-bits (diff) | |
| download | linux-521223d7c229f83915619f888c99e952f24dc39f.tar.xz linux-521223d7c229f83915619f888c99e952f24dc39f.zip | |
io_uring/cancel: don't default to setting req->work.cancel_seq
Just leave it unset by default, avoiding dipping into the last
cacheline (which is otherwise untouched) for the fast path of using
poll to drive networked traffic. Add a flag that tells us if the
sequence is valid or not, and then we can defer actually assigning
the flag and sequence until someone runs cancelations.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Diffstat (limited to 'include')
| -rw-r--r-- | include/linux/io_uring_types.h | 3 |
1 files changed, 3 insertions, 0 deletions
diff --git a/include/linux/io_uring_types.h b/include/linux/io_uring_types.h index 56bf733d3ee6..e19698daae1a 100644 --- a/include/linux/io_uring_types.h +++ b/include/linux/io_uring_types.h @@ -463,6 +463,7 @@ enum { REQ_F_SUPPORT_NOWAIT_BIT, REQ_F_ISREG_BIT, REQ_F_POLL_NO_LAZY_BIT, + REQ_F_CANCEL_SEQ_BIT, /* not a real bit, just to check we're not overflowing the space */ __REQ_F_LAST_BIT, @@ -535,6 +536,8 @@ enum { REQ_F_HASH_LOCKED = IO_REQ_FLAG(REQ_F_HASH_LOCKED_BIT), /* don't use lazy poll wake for this request */ REQ_F_POLL_NO_LAZY = IO_REQ_FLAG(REQ_F_POLL_NO_LAZY_BIT), + /* cancel sequence is set and valid */ + REQ_F_CANCEL_SEQ = IO_REQ_FLAG(REQ_F_CANCEL_SEQ_BIT), }; typedef void (*io_req_tw_func_t)(struct io_kiocb *req, struct io_tw_state *ts); |