io_uring: get rid of intermediate aux cqe caches

io_post_aux_cqe(), which is used for multishot requests, delays
completions by putting CQEs into a temporary array for the purpose
completion lock/flush batching.

DEFER_TASKRUN doesn't need any locking, so for it we can put completions
directly into the CQ and defer post completion handling with a flag.
That leaves !DEFER_TASKRUN, which is not that interesting / hot for
multishot requests, so have conditional locking with deferred flush
for them.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Tested-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/b1d05a81fd27aaa2a07f9860af13059e7ad7a890.1710799188.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

include/linux/io_uring_types.h

@@ -205,6 +205,7 @@ struct io_submit_state {
 
 	bool			plug_started;
 	bool			need_plug;
+	bool			cq_flush;
 	unsigned short		submit_nr;
 	unsigned int		cqes_count;
 	struct blk_plug		plug;
@@ -341,8 +342,6 @@ struct io_ring_ctx {
 		unsigned		cq_last_tm_flush;
 	} ____cacheline_aligned_in_smp;
 
-	struct io_uring_cqe	completion_cqes[16];
-
 	spinlock_t		completion_lock;
 
 	/* IRQ completion list, under ->completion_lock */

io_uring/io_uring.c

@@ -630,6 +630,12 @@ static inline void __io_cq_lock(struct io_ring_ctx *ctx)
 		spin_lock(&ctx->completion_lock);
 }
 
+static inline void __io_cq_unlock(struct io_ring_ctx *ctx)
+{
+	if (!ctx->lockless_cq)
+		spin_unlock(&ctx->completion_lock);
+}
+
 static inline void io_cq_lock(struct io_ring_ctx *ctx)
 	__acquires(ctx->completion_lock)
 {
@@ -882,31 +888,6 @@ static bool io_fill_cqe_aux(struct io_ring_ctx *ctx, u64 user_data, s32 res,
 	return false;
 }
 
-static void __io_flush_post_cqes(struct io_ring_ctx *ctx)
-	__must_hold(&ctx->uring_lock)
-{
-	struct io_submit_state *state = &ctx->submit_state;
-	unsigned int i;
-
-	lockdep_assert_held(&ctx->uring_lock);
-	for (i = 0; i < state->cqes_count; i++) {
-		struct io_uring_cqe *cqe = &ctx->completion_cqes[i];
-
-		if (!io_fill_cqe_aux(ctx, cqe->user_data, cqe->res, cqe->flags)) {
-			if (ctx->lockless_cq) {
-				spin_lock(&ctx->completion_lock);
-				io_cqring_event_overflow(ctx, cqe->user_data,
-							cqe->res, cqe->flags, 0, 0);
-				spin_unlock(&ctx->completion_lock);
-			} else {
-				io_cqring_event_overflow(ctx, cqe->user_data,
-							cqe->res, cqe->flags, 0, 0);
-			}
-		}
-	}
-	state->cqes_count = 0;
-}
-
 bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags)
 {
 	bool filled;
@@ -927,31 +908,16 @@ bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags
 bool io_req_post_cqe(struct io_kiocb *req, s32 res, u32 cflags)
 {
 	struct io_ring_ctx *ctx = req->ctx;
-	u64 user_data = req->cqe.user_data;
-	struct io_uring_cqe *cqe;
+	bool posted;
 
 	lockdep_assert(!io_wq_current_is_worker());
 	lockdep_assert_held(&ctx->uring_lock);
 
-	if (ctx->submit_state.cqes_count == ARRAY_SIZE(ctx->completion_cqes)) {
-		__io_cq_lock(ctx);
-		__io_flush_post_cqes(ctx);
-		/* no need to flush - flush is deferred */
-		__io_cq_unlock_post(ctx);
-	}
-
-	/* For defered completions this is not as strict as it is otherwise,
-	 * however it's main job is to prevent unbounded posted completions,
-	 * and in that it works just as well.
-	 */
-	if (test_bit(IO_CHECK_CQ_OVERFLOW_BIT, &ctx->check_cq))
-		return false;
-
-	cqe = &ctx->completion_cqes[ctx->submit_state.cqes_count++];
-	cqe->user_data = user_data;
-	cqe->res = res;
-	cqe->flags = cflags;
-	return true;
+	__io_cq_lock(ctx);
+	posted = io_fill_cqe_aux(ctx, req->cqe.user_data, res, cflags);
+	ctx->submit_state.cq_flush = true;
+	__io_cq_unlock_post(ctx);
+	return posted;
 }
 
 static void __io_req_complete_post(struct io_kiocb *req, unsigned issue_flags)
@@ -1545,9 +1511,6 @@ void __io_submit_flush_completions(struct io_ring_ctx *ctx)
 	struct io_wq_work_node *node;
 
 	__io_cq_lock(ctx);
-	/* must come first to preserve CQE ordering in failure cases */
-	if (state->cqes_count)
-		__io_flush_post_cqes(ctx);
 	__wq_list_for_each(node, &state->compl_reqs) {
 		struct io_kiocb *req = container_of(node, struct io_kiocb,
 					    comp_list);
@@ -1569,6 +1532,7 @@ void __io_submit_flush_completions(struct io_ring_ctx *ctx)
 		io_free_batch_list(ctx, state->compl_reqs.first);
 		INIT_WQ_LIST(&state->compl_reqs);
 	}
+	ctx->submit_state.cq_flush = false;
 }
 
 static unsigned io_cqring_events(struct io_ring_ctx *ctx)

io_uring/io_uring.h

@@ -156,7 +156,7 @@ static inline void io_req_task_work_add(struct io_kiocb *req)
 static inline void io_submit_flush_completions(struct io_ring_ctx *ctx)
 {
 	if (!wq_list_empty(&ctx->submit_state.compl_reqs) ||
-	    ctx->submit_state.cqes_count)
+	    ctx->submit_state.cq_flush)
 		__io_submit_flush_completions(ctx);
 }