io_uring/rsrc: add an empty io_rsrc_node for sparse buffer entries

Rather than allocate an io_rsrc_node for an empty/sparse buffer entry,
add a const entry that can be used for that. This just needs checking
for writing the tag, and the put check needs to check for that sparse
node rather than NULL for validity.

This avoids allocating rsrc nodes for sparse buffer entries.

Signed-off-by: Jens Axboe <[email protected]>

Author: axboe

io_uring/io_uring.c

@@ -2032,8 +2032,8 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
 	req->flags = (__force io_req_flags_t) sqe_flags;
 	req->cqe.user_data = READ_ONCE(sqe->user_data);
 	req->file = NULL;
-	req->rsrc_nodes[IORING_RSRC_FILE] = NULL;
-	req->rsrc_nodes[IORING_RSRC_BUFFER] = NULL;
+	req->rsrc_nodes[IORING_RSRC_FILE] = rsrc_empty_node;
+	req->rsrc_nodes[IORING_RSRC_BUFFER] = rsrc_empty_node;
 	req->task = current;
 	req->cancel_seq_set = false;
 

io_uring/notif.c

@@ -117,8 +117,8 @@ struct io_kiocb *io_alloc_notif(struct io_ring_ctx *ctx)
 	notif->file = NULL;
 	notif->task = current;
 	io_get_task_refs(1);
-	notif->rsrc_nodes[IORING_RSRC_FILE] = NULL;
-	notif->rsrc_nodes[IORING_RSRC_BUFFER] = NULL;
+	notif->rsrc_nodes[IORING_RSRC_FILE] = rsrc_empty_node;
+	notif->rsrc_nodes[IORING_RSRC_BUFFER] = rsrc_empty_node;
 
 	nd = io_notif_to_data(notif);
 	nd->zc_report = false;

io_uring/rsrc.c

@@ -38,6 +38,11 @@ static const struct io_mapped_ubuf dummy_ubuf = {
 	.len = UINT_MAX,
 };
 
+const struct io_rsrc_node empty_node = {
+	.type = IORING_RSRC_BUFFER,
+	.buf = (struct io_mapped_ubuf *) &dummy_ubuf,
+};
+
 int __io_account_mem(struct user_struct *user, unsigned long nr_pages)
 {
 	unsigned long page_limit, cur_pages, new_pages;
@@ -144,7 +149,8 @@ static void io_rsrc_data_free(struct io_rsrc_data *data)
 	for (i = 0; i < data->nr; i++) {
 		struct io_rsrc_node *node = data->nodes[i];
 
-		io_put_rsrc_node(node);
+		if (node)
+			io_put_rsrc_node(node);
 	}
 	kvfree(data->nodes);
 	kfree(data);
@@ -229,7 +235,8 @@ static int __io_sqe_files_update(struct io_ring_ctx *ctx,
 				break;
 			}
 			ctx->file_table.nodes[i] = node;
-			node->tag = tag;
+			if (tag)
+				node->tag = tag;
 			io_fixed_file_set(node, file);
 			io_file_bitmap_set(&ctx->file_table, i);
 		}
@@ -281,10 +288,12 @@ static int __io_sqe_buffers_update(struct io_ring_ctx *ctx,
 			err = PTR_ERR(node);
 			break;
 		}
-		io_put_rsrc_node(ctx->user_bufs[i]);
+		if (ctx->user_bufs[i])
+			io_put_rsrc_node(ctx->user_bufs[i]);
 
 		ctx->user_bufs[i] = node;
-		node->tag = tag;
+		if (tag)
+			node->tag = tag;
 		if (ctx->compat)
 			user_data += sizeof(struct compat_iovec);
 		else
@@ -600,8 +609,10 @@ static void __io_sqe_buffers_unregister(struct io_ring_ctx *ctx)
 	lockdep_assert_held(&ctx->uring_lock);
 
 	for (i = 0; i < ctx->nr_user_bufs; i++) {
-		io_put_rsrc_node(ctx->user_bufs[i]);
-		ctx->user_bufs[i] = NULL;
+		if (ctx->user_bufs[i]) {
+			io_put_rsrc_node(ctx->user_bufs[i]);
+			ctx->user_bufs[i] = NULL;
+		}
 	}
 	kvfree(ctx->user_bufs);
 	ctx->user_bufs = NULL;
@@ -799,11 +810,6 @@ static struct io_rsrc_node *io_sqe_buffer_register(struct io_ring_ctx *ctx,
 		return ERR_PTR(-ENOMEM);
 	node->buf = NULL;
 
-	if (!iov->iov_base) {
-		node->buf = (struct io_mapped_ubuf *) &dummy_ubuf;
-		return node;
-	}
-
 	ret = -ENOMEM;
 	pages = io_pin_pages((unsigned long) iov->iov_base, iov->iov_len,
 				&nr_pages);
@@ -927,7 +933,8 @@ int io_sqe_buffers_register(struct io_ring_ctx *ctx, void __user *arg,
 			ret = PTR_ERR(node);
 			break;
 		}
-		node->tag = tag;
+		if (tag)
+			node->tag = tag;
 		ctx->user_bufs[i] = node;
 	}
 
@@ -1028,18 +1035,18 @@ static int io_clone_buffers(struct io_ring_ctx *ctx, struct io_ring_ctx *src_ctx
 		goto out_free_data;
 
 	for (i = 0; i < nbufs; i++) {
-		struct io_mapped_ubuf *imu = src_ctx->user_bufs[i]->buf;
+		struct io_rsrc_node *src_node = src_ctx->user_bufs[i];
 		struct io_rsrc_node *dst_node;
 
-		dst_node = io_rsrc_node_alloc(ctx, IORING_RSRC_BUFFER);
-		if (!dst_node)
-			goto out_put_free;
-
-		if (imu == &dummy_ubuf) {
-			dst_node->buf = (struct io_mapped_ubuf *) &dummy_ubuf;
+		if (src_node == rsrc_empty_node) {
+			dst_node = rsrc_empty_node;
 		} else {
-			refcount_inc(&imu->refs);
-			dst_node->buf = imu;
+			dst_node = io_rsrc_node_alloc(ctx, IORING_RSRC_BUFFER);
+			if (!dst_node)
+				goto out_put_free;
+
+			refcount_inc(&src_node->buf->refs);
+			dst_node->buf = src_node->buf;
 		}
 		user_bufs[i] = dst_node;
 	}

io_uring/rsrc.h

@@ -70,9 +70,12 @@ int io_register_rsrc_update(struct io_ring_ctx *ctx, void __user *arg,
 int io_register_rsrc(struct io_ring_ctx *ctx, void __user *arg,
 			unsigned int size, unsigned int type);
 
+extern const struct io_rsrc_node empty_node;
+#define rsrc_empty_node	(struct io_rsrc_node *) &empty_node
+
 static inline void io_put_rsrc_node(struct io_rsrc_node *node)
 {
-	if (node && !--node->refs)
+	if (node != rsrc_empty_node && !--node->refs)
 		io_free_rsrc_node(node);
 }
 
@@ -85,8 +88,10 @@ static inline void io_req_put_rsrc_nodes(struct io_kiocb *req)
 static inline void io_req_assign_rsrc_node(struct io_kiocb *req,
 					   struct io_rsrc_node *node)
 {
-	node->refs++;
-	req->rsrc_nodes[node->type] = node;
+	if (node != rsrc_empty_node) {
+		node->refs++;
+		req->rsrc_nodes[node->type] = node;
+	}
 }
 
 int io_files_update(struct io_kiocb *req, unsigned int issue_flags);

io_uring/splice.c

@@ -35,7 +35,7 @@ static int __io_splice_prep(struct io_kiocb *req,
 	if (unlikely(sp->flags & ~valid_flags))
 		return -EINVAL;
 	sp->splice_fd_in = READ_ONCE(sqe->splice_fd_in);
-	sp->rsrc_node = NULL;
+	sp->rsrc_node = rsrc_empty_node;
 	req->flags |= REQ_F_FORCE_ASYNC;
 	return 0;
 }