Re: [PATCH for-5.7] io_uring: fix zero len do_splice()

From: Pavel Begunkov
Date: Mon May 04 2020 - 16:15:35 EST


On 04/05/2020 23:00, Pavel Begunkov wrote:
> do_splice() doesn't expect len to be 0. Just always return 0 in this
> case as splice(2) do.

There is a thing, splice/tee will always return success on len=0 even with
invalid fds. Fast return for len=0, should really has been done after basic
validation, but I don't want to break userspace.

Any ideas?

>
> Fixes: 7d67af2c0134 ("io_uring: add splice(2) support")
> Reported-by: Jann Horn <jannh@xxxxxxxxxx>
> Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx>
> ---
> fs/io_uring.c | 11 +++++++----
> 1 file changed, 7 insertions(+), 4 deletions(-)
>
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 65458eda2127..d53a1ef2a205 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -2772,16 +2772,19 @@ static int io_splice(struct io_kiocb *req, bool force_nonblock)
> struct file *out = sp->file_out;
> unsigned int flags = sp->flags & ~SPLICE_F_FD_IN_FIXED;
> loff_t *poff_in, *poff_out;
> - long ret;
> + long ret = 0;
>
> if (force_nonblock)
> return -EAGAIN;
>
> poff_in = (sp->off_in == -1) ? NULL : &sp->off_in;
> poff_out = (sp->off_out == -1) ? NULL : &sp->off_out;
> - ret = do_splice(in, poff_in, out, poff_out, sp->len, flags);
> - if (force_nonblock && ret == -EAGAIN)
> - return -EAGAIN;
> +
> + if (sp->len) {
> + ret = do_splice(in, poff_in, out, poff_out, sp->len, flags);
> + if (force_nonblock && ret == -EAGAIN)
> + return -EAGAIN;
> + }
>
> io_put_file(req, in, (sp->flags & SPLICE_F_FD_IN_FIXED));
> req->flags &= ~REQ_F_NEED_CLEANUP;
>

--
Pavel Begunkov