[PATCH 0/2] two small fixes

[PATCH 0/2] two small fixes

[Pavel Begunkov]

The first correctly attributes failed iopoll requests.
2/2 fixes a for-next bug.

Pavel Begunkov (2):
  io_uring: fail iopoll links if can't retry
  io_uring: fix async_data checks for msg setup

 fs/io_uring.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

-- 
2.33.1

[PATCH 1/2] io_uring: fail iopoll links if can't retry

[Pavel Begunkov]

If io_rw_should_reissue() fails in iopoll path and we can't reissue we
fail the request. Don't forget to also mark it as failed, so links are
broken.

Cc: stable@vger.kernel.org
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 0d7613c7355c..40b1697e7354 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -2687,6 +2687,7 @@ static void io_complete_rw_iopoll(struct kiocb *kiocb, long res, long res2)
 			req->flags |= REQ_F_REISSUE;
 			return;
 		}
+		req_set_fail(req);
 		req->result = res;
 	}
 
-- 
2.33.1

[PATCH 2/2] io_uring: fix async_data checks for msg setup

[Pavel Begunkov]

We should use req_has_async_data() instead of directly testing
->async_data to see if async_data has been allocated, there is one spot
slipped that doesn't follow the rule.

Fixes: 016e0451e9ed0 ("io_uring: control ->async_data with a REQ_F flag")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 40b1697e7354..7ead1507be9b 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -4683,9 +4683,9 @@ static int io_sync_file_range(struct io_kiocb *req, unsigned int issue_flags)
 static int io_setup_async_msg(struct io_kiocb *req,
 			      struct io_async_msghdr *kmsg)
 {
-	struct io_async_msghdr *async_msg = req->async_data;
+	struct io_async_msghdr *async_msg;
 
-	if (async_msg)
+	if (req_has_async_data(req))
 		return -EAGAIN;
 	if (io_alloc_async_data(req)) {
 		kfree(kmsg->free_iov);
-- 
2.33.1

Re: [PATCH 1/2] io_uring: fail iopoll links if can't retry

[Greg KH]

On Sun, Oct 17, 2021 at 08:30:55PM +0000, Pavel Begunkov wrote:
> If io_rw_should_reissue() fails in iopoll path and we can't reissue we
> fail the request. Don't forget to also mark it as failed, so links are
> broken.
> 
> Cc: stable@vger.kernel.org
> Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
> ---
>  fs/io_uring.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 0d7613c7355c..40b1697e7354 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -2687,6 +2687,7 @@ static void io_complete_rw_iopoll(struct kiocb *kiocb, long res, long res2)
>  			req->flags |= REQ_F_REISSUE;
>  			return;
>  		}
> +		req_set_fail(req);
>  		req->result = res;
>  	}
>  
> -- 
> 2.33.1
> 

<formletter>

This is not the correct way to submit patches for inclusion in the
stable kernel tree.  Please read:
    https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
for how to do this properly.

</formletter>