diff mbox series

[2/5] io_uring/core: move io_cqe->fd over from io_cqe->flags to io_cqe->res

Message ID c4a5afc7fec8314032074cc60b9f27b6e0af9f39.1663363798.git.metze@samba.org (mailing list archive)
State New
Headers show
Series IORING_OP_SEND_ZC improvements | expand

Commit Message

Stefan Metzmacher Sept. 16, 2022, 9:36 p.m. UTC
Both flags and res have the same lifetime currently.
io_req_set_res() sets both of them.

Callers of io_req_set_res() like req_fail_link_node(),
io_req_complete_failed() and io_req_task_queue_fail()
set io_cqe->res to their callers value and force flags to 0.

The motivation for this change is the next commit,
it will let us keep io_cqe->flags even on error.
For IORING_OP_SEND_ZC it is needed to keep IORING_CQE_F_MORE
even on a generic failure, userspace needs to know that
a IORING_CQE_F_NOTIF will follow. Otherwise the buffers
might be reused too early.

Fixes: b48c312be05e8 ("io_uring/net: simplify zerocopy send user API")
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Cc: Pavel Begunkov <asml.silence@gmail.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: io-uring@vger.kernel.org
---
 include/linux/io_uring_types.h |  6 +++---
 io_uring/io_uring.c            | 10 ++++++++--
 2 files changed, 11 insertions(+), 5 deletions(-)
diff mbox series

Patch

diff --git a/include/linux/io_uring_types.h b/include/linux/io_uring_types.h
index 677a25d44d7f..37925db42ae9 100644
--- a/include/linux/io_uring_types.h
+++ b/include/linux/io_uring_types.h
@@ -473,12 +473,12 @@  struct io_task_work {
 
 struct io_cqe {
 	__u64	user_data;
-	__s32	res;
-	/* fd initially, then cflags for completion */
+	/* fd initially, then res for completion */
 	union {
-		__u32	flags;
 		int	fd;
+		__s32	res;
 	};
+	__u32	flags;
 };
 
 /*
diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index b9640ad5069f..ae69cff94664 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -837,8 +837,6 @@  static void io_preinit_req(struct io_kiocb *req, struct io_ring_ctx *ctx)
 	req->ctx = ctx;
 	req->link = NULL;
 	req->async_data = NULL;
-	/* not necessary, but safer to zero */
-	req->cqe.res = 0;
 }
 
 static void io_flush_cached_locked_reqs(struct io_ring_ctx *ctx,
@@ -1574,6 +1572,12 @@  static int io_issue_sqe(struct io_kiocb *req, unsigned int issue_flags)
 	if (!def->audit_skip)
 		audit_uring_entry(req->opcode);
 
+	/*
+	 * req->cqe.fd was resolved by io_assign_file
+	 * now make sure its alias req->cqe.res is reset,
+	 * so we don't use that value by accident.
+	 */
+	req->cqe.res = -1;
 	ret = def->issue(req, issue_flags);
 
 	if (!def->audit_skip)
@@ -1902,6 +1906,8 @@  static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
 			state->need_plug = false;
 			blk_start_plug_nr_ios(&state->plug, state->submit_nr);
 		}
+	} else {
+		req->cqe.fd = -1;
 	}
 
 	personality = READ_ONCE(sqe->personality);