io_uring: optimise extra io_get_cqe null check

If the cached cqe check passes in io_get_cqe*() it already means that
the cqe we return is valid and non-zero, however the compiler is unable
to optimise null checks like in io_fill_cqe_req().

Do a bit of trickery, return success/fail boolean from io_get_cqe*()
and store cqe in the cqe parameter. That makes it do the right thing,
erasing the check together with the introduced indirection.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/322ea4d3377d3d4efd8ae90ab8ed28a99f518210.1692916914.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Author: isilence

io_uring/io_uring.c

@@ -683,10 +683,10 @@ static void __io_cqring_overflow_flush(struct io_ring_ctx *ctx)
 
 	io_cq_lock(ctx);
 	while (!list_empty(&ctx->cq_overflow_list)) {
-		struct io_uring_cqe *cqe = io_get_cqe_overflow(ctx, true);
+		struct io_uring_cqe *cqe;
 		struct io_overflow_cqe *ocqe;
 
-		if (!cqe)
+		if (!io_get_cqe_overflow(ctx, &cqe, true))
 			break;
 		ocqe = list_first_entry(&ctx->cq_overflow_list,
 					struct io_overflow_cqe, list);
@@ -862,8 +862,7 @@ static bool io_fill_cqe_aux(struct io_ring_ctx *ctx, u64 user_data, s32 res,
 	 * submission (by quite a lot). Increment the overflow count in
 	 * the ring.
 	 */
-	cqe = io_get_cqe(ctx);
-	if (likely(cqe)) {
+	if (likely(io_get_cqe(ctx, &cqe))) {
 		trace_io_uring_complete(ctx, NULL, user_data, res, cflags, 0, 0);
 
 		WRITE_ONCE(cqe->user_data, user_data);

io_uring/io_uring.h

@@ -109,28 +109,27 @@ static inline void io_req_task_work_add(struct io_kiocb *req)
 #define io_for_each_link(pos, head) \
 	for (pos = (head); pos; pos = pos->link)
 
-static inline struct io_uring_cqe *io_get_cqe_overflow(struct io_ring_ctx *ctx,
-						       bool overflow)
+static inline bool io_get_cqe_overflow(struct io_ring_ctx *ctx,
+					struct io_uring_cqe **ret,
+					bool overflow)
 {
-	struct io_uring_cqe *cqe;
-
 	io_lockdep_assert_cq_locked(ctx);
 
 	if (unlikely(ctx->cqe_cached >= ctx->cqe_sentinel)) {
 		if (unlikely(!io_cqe_cache_refill(ctx, overflow)))
-			return NULL;
+			return false;
 	}
-	cqe = ctx->cqe_cached;
+	*ret = ctx->cqe_cached;
 	ctx->cached_cq_tail++;
 	ctx->cqe_cached++;
 	if (ctx->flags & IORING_SETUP_CQE32)
 		ctx->cqe_cached++;
-	return cqe;
+	return true;
 }
 
-static inline struct io_uring_cqe *io_get_cqe(struct io_ring_ctx *ctx)
+static inline bool io_get_cqe(struct io_ring_ctx *ctx, struct io_uring_cqe **ret)
 {
-	return io_get_cqe_overflow(ctx, false);
+	return io_get_cqe_overflow(ctx, ret, false);
 }
 
 static inline bool io_fill_cqe_req(struct io_ring_ctx *ctx, struct io_kiocb *req)
@@ -142,8 +141,7 @@ static inline bool io_fill_cqe_req(struct io_ring_ctx *ctx, struct io_kiocb *req
 	 * submission (by quite a lot). Increment the overflow count in
 	 * the ring.
 	 */
-	cqe = io_get_cqe(ctx);
-	if (unlikely(!cqe))
+	if (unlikely(!io_get_cqe(ctx, &cqe)))
 		return false;
 
 	if (trace_io_uring_complete_enabled())