在 2021/11/25 上午11:48, Hao Xu 写道:
在 2021/11/10 下午11:49, Pavel Begunkov 写道:When no of queued for the batch completion requests need to post an CQE, see IOSQE_CQE_SKIP_SUCCESS, avoid grabbing ->completion_lock and other commit/post. Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx> --- fs/io_uring.c | 26 +++++++++++++++++--------- 1 file changed, 17 insertions(+), 9 deletions(-) diff --git a/fs/io_uring.c b/fs/io_uring.c index 22572cfd6864..0c0ea3bbb50a 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -321,6 +321,7 @@ struct io_submit_state { bool plug_started; bool need_plug; + bool flush_cqes; unsigned short submit_nr; struct blk_plug plug; };@@ -1525,8 +1526,11 @@ static void io_prep_async_link(struct io_kiocb *req)static inline void io_req_add_compl_list(struct io_kiocb *req) { + struct io_ring_ctx *ctx = req->ctx; struct io_submit_state *state = &req->ctx->submit_state; + if (!(req->flags & REQ_F_CQE_SKIP)) + ctx->submit_state.flush_cqes = true;Should we init it to false in submit_state_start()?
Never mind, I saw it in __io_submit_flush_completions() so there is no problem..
