[PATCH for-6.1 3/4] io_uring: fix multishot accept request leaks

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Having REQ_F_POLLED set doesn't guarantee that the request is
executed as a multishot from the polling path. Fortunately for us, if
the code thinks it's multishot issue when it's not, it can only ask to
skip completion so leaking the request. Use issue_flags to mark
multipoll issues.

Cc: stable@xxxxxxxxxxxxxxx
Fixes: 390ed29b5e425 ("io_uring: add IORING_ACCEPT_MULTISHOT for accept")
Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx>
---
 include/linux/io_uring.h | 3 +++
 io_uring/io_uring.c      | 2 +-
 io_uring/io_uring.h      | 4 ++--
 io_uring/net.c           | 7 ++-----
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/include/linux/io_uring.h b/include/linux/io_uring.h
index 43bc8a2edccf..0ded9e271523 100644
--- a/include/linux/io_uring.h
+++ b/include/linux/io_uring.h
@@ -16,6 +16,9 @@ enum io_uring_cmd_flags {
 	IO_URING_F_SQE128		= 4,
 	IO_URING_F_CQE32		= 8,
 	IO_URING_F_IOPOLL		= 16,
+
+	/* the request is executed from poll, it should not be freed */
+	IO_URING_F_MULTISHOT		= 32,
 };
 
 struct io_uring_cmd {
diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index 4a1e482747cc..8840cf3e20f2 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -1768,7 +1768,7 @@ int io_poll_issue(struct io_kiocb *req, bool *locked)
 	io_tw_lock(req->ctx, locked);
 	if (unlikely(req->task->flags & PF_EXITING))
 		return -EFAULT;
-	return io_issue_sqe(req, IO_URING_F_NONBLOCK);
+	return io_issue_sqe(req, IO_URING_F_NONBLOCK|IO_URING_F_MULTISHOT);
 }
 
 struct io_wq_work *io_wq_free_work(struct io_wq_work *work)
diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h
index e99a79f2df9b..cef5ff924e63 100644
--- a/io_uring/io_uring.h
+++ b/io_uring/io_uring.h
@@ -17,8 +17,8 @@ enum {
 	IOU_ISSUE_SKIP_COMPLETE	= -EIOCBQUEUED,
 
 	/*
-	 * Intended only when both REQ_F_POLLED and REQ_F_APOLL_MULTISHOT
-	 * are set to indicate to the poll runner that multishot should be
+	 * Intended only when both IO_URING_F_MULTISHOT is passed
+	 * to indicate to the poll runner that multishot should be
 	 * removed and the result is set on req->cqe.res.
 	 */
 	IOU_STOP_MULTISHOT	= -ECANCELED,
diff --git a/io_uring/net.c b/io_uring/net.c
index 15dea91625e2..a390d3ea486c 100644
--- a/io_uring/net.c
+++ b/io_uring/net.c
@@ -1289,8 +1289,7 @@ int io_accept(struct io_kiocb *req, unsigned int issue_flags)
 			 * return EAGAIN to arm the poll infra since it
 			 * has already been done
 			 */
-			if ((req->flags & IO_APOLL_MULTI_POLLED) ==
-			    IO_APOLL_MULTI_POLLED)
+			if (issue_flags & IO_URING_F_MULTISHOT)
 				ret = IOU_ISSUE_SKIP_COMPLETE;
 			return ret;
 		}
@@ -1315,9 +1314,7 @@ int io_accept(struct io_kiocb *req, unsigned int issue_flags)
 		goto retry;
 
 	io_req_set_res(req, ret, 0);
-	if (req->flags & REQ_F_POLLED)
-		return IOU_STOP_MULTISHOT;
-	return IOU_OK;
+	return (issue_flags & IO_URING_F_MULTISHOT) ? IOU_STOP_MULTISHOT : IOU_OK;
 }
 
 int io_socket_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
-- 
2.38.1




[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux