[v2,5/6] io_uring: add memory region registration

Commit Message

Pavel Begunkov Nov. 14, 2024, 5:38 p.m. UTC

Regions will serve multiple purposes. First, with it we can decouple
ring/etc. object creation from registration / mapping of the memory they
will be placed in. We already have hacks that allow to put both SQ and
CQ into the same huge page, in the future we should be able to:

region = create_region(io_ring);
create_pbuf_ring(io_uring, region, offset=0);
create_pbuf_ring(io_uring, region, offset=N);

The second use case is efficiently passing parameters. The following
patch enables back on top of regions IORING_ENTER_EXT_ARG_REG, which
optimises wait arguments. It'll also be useful for request arguments
replacing iovecs, msghdr, etc. pointers. Eventually it would also be
handy for BPF as well if it comes to fruition.

Signed-off-by: Pavel Begunkov <[e-mail]>
---
 include/linux/io_uring_types.h |  3 +++
 include/uapi/linux/io_uring.h  |  8 ++++++++
 io_uring/io_uring.c            |  1 +
 io_uring/register.c            | 37 ++++++++++++++++++++++++++++++++++
 4 files changed, 49 insertions(+)

Patch

diff --git a/include/linux/io_uring_types.h b/include/linux/io_uring_types.h
index 1d3a37234ace..e1d69123e164 100644
--- a/include/linux/io_uring_types.h
+++ b/include/linux/io_uring_types.h
@@ -429,6 +429,9 @@  struct io_ring_ctx {
 	unsigned short			n_sqe_pages;
 	struct page			**ring_pages;
 	struct page			**sqe_pages;
+
+	/* used for optimised request parameter and wait argument passing  */
+	struct io_mapped_region		param_region;
 };
 
 struct io_tw_state {
diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
index 5cbfd330c688..1ee35890125b 100644
--- a/include/uapi/linux/io_uring.h
+++ b/include/uapi/linux/io_uring.h
@@ -627,6 +627,8 @@  enum io_uring_register_op {
 	/* resize CQ ring */
 	IORING_REGISTER_RESIZE_RINGS		= 33,
 
+	IORING_REGISTER_MEM_REGION		= 34,
+
 	/* this goes last */
 	IORING_REGISTER_LAST,
 
@@ -661,6 +663,12 @@  struct io_uring_region_desc {
 	__u64 __resv[4];
 };
 
+struct io_uring_mem_region_reg {
+	__u64 region_uptr; /* struct io_uring_region_desc * */
+	__u64 flags;
+	__u64 __resv[2];
+};
+
 /*
  * Register a fully sparse file space, rather than pass in an array of all
  * -1 file descriptors.
diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index 286b7bb73978..c640b8a4ceee 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -2709,6 +2709,7 @@  static __cold void io_ring_ctx_free(struct io_ring_ctx *ctx)
 	io_alloc_cache_free(&ctx->msg_cache, io_msg_cache_free);
 	io_futex_cache_free(ctx);
 	io_destroy_buffers(ctx);
+	io_free_region(ctx, &ctx->param_region);
 	mutex_unlock(&ctx->uring_lock);
 	if (ctx->sq_creds)
 		put_cred(ctx->sq_creds);
diff --git a/io_uring/register.c b/io_uring/register.c
index 3c5a3cfb186b..2cbac3d9b288 100644
--- a/io_uring/register.c
+++ b/io_uring/register.c
@@ -570,6 +570,37 @@  static int io_register_resize_rings(struct io_ring_ctx *ctx, void __user *arg)
 	return ret;
 }
 
+static int io_register_mem_region(struct io_ring_ctx *ctx, void __user *uarg)
+{
+	struct io_uring_mem_region_reg __user *reg_uptr = uarg;
+	struct io_uring_mem_region_reg reg;
+	struct io_uring_region_desc __user *rd_uptr;
+	struct io_uring_region_desc rd;
+	int ret;
+
+	if (io_region_is_set(&ctx->param_region))
+		return -EBUSY;
+	if (copy_from_user(&reg, reg_uptr, sizeof(reg)))
+		return -EFAULT;
+	rd_uptr = u64_to_user_ptr(reg.region_uptr);
+	if (copy_from_user(&rd, rd_uptr, sizeof(rd)))
+		return -EFAULT;
+
+	if (memchr_inv(&reg.__resv, 0, sizeof(reg.__resv)))
+		return -EINVAL;
+	if (reg.flags)
+		return -EINVAL;
+
+	ret = io_create_region(ctx, &ctx->param_region, &rd);
+	if (ret)
+		return ret;
+	if (copy_to_user(rd_uptr, &rd, sizeof(rd))) {
+		io_free_region(ctx, &ctx->param_region);
+		return -EFAULT;
+	}
+	return 0;
+}
+
 static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
 			       void __user *arg, unsigned nr_args)
 	__releases(ctx->uring_lock)
@@ -764,6 +795,12 @@  static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
 			break;
 		ret = io_register_resize_rings(ctx, arg);
 		break;
+	case IORING_REGISTER_MEM_REGION:
+		ret = -EINVAL;
+		if (!arg || nr_args != 1)
+			break;
+		ret = io_register_mem_region(ctx, arg);
+		break;
 	default:
 		ret = -EINVAL;
 		break;