Re: [RFC PATCH v2 1/4] net: wire up support for file_operations->uring_cmd()

[Pavel Begunkov <asml.silence@xxxxxxxxx>]

On 6/14/23 16:15, David Ahern wrote:
> On 6/14/23 5:07 AM, Breno Leitao wrote:
> > diff --git a/include/linux/net.h b/include/linux/net.h
> > index 8defc8f1d82e..58dea87077af 100644
> > --- a/include/linux/net.h
> > +++ b/include/linux/net.h
> > @@ -182,6 +182,8 @@ struct proto_ops {
> >   	int	 	(*compat_ioctl) (struct socket *sock, unsigned int cmd,
> >   				      unsigned long arg);
> >   #endif
> > +	int		(*uring_cmd)(struct socket *sock, struct io_uring_cmd *cmd,
> > +				     unsigned int issue_flags);
> >   	int		(*gettstamp) (struct socket *sock, void __user *userstamp,
> >   				      bool timeval, bool time32);
> >   	int		(*listen)    (struct socket *sock, int len);
> > diff --git a/include/net/sock.h b/include/net/sock.h
> > index 62a1b99da349..a49b8b19292b 100644
> > --- a/include/net/sock.h
> > +++ b/include/net/sock.h
> > @@ -111,6 +111,7 @@ typedef struct {
> >   struct sock;
> >   struct proto;
> >   struct net;
> > +struct io_uring_cmd;
> >   
> >   typedef __u32 __bitwise __portpair;
> >   typedef __u64 __bitwise __addrpair;
> > @@ -1259,6 +1260,9 @@ struct proto {
> >   
> >   	int			(*ioctl)(struct sock *sk, int cmd,
> >   					 int *karg);
> > +	int			(*uring_cmd)(struct sock *sk,
> > +					     struct io_uring_cmd *cmd,
> > +					     unsigned int issue_flags);
> >   	int			(*init)(struct sock *sk);
> >   	void			(*destroy)(struct sock *sk);
> >   	void			(*shutdown)(struct sock *sk, int how);
> > @@ -1934,6 +1938,8 @@ int sock_common_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
> >   			int flags);
> >   int sock_common_setsockopt(struct socket *sock, int level, int optname,
> >   			   sockptr_t optval, unsigned int optlen);
> > +int sock_common_uring_cmd(struct socket *sock, struct io_uring_cmd *cmd,
> > +			  unsigned int issue_flags);
> >   
> >   void sk_common_release(struct sock *sk);
> >   
> > diff --git a/net/core/sock.c b/net/core/sock.c
> > index 1df7e432fec5..339fa74db60f 100644
> > --- a/net/core/sock.c
> > +++ b/net/core/sock.c
> > @@ -3668,6 +3668,18 @@ int sock_common_setsockopt(struct socket *sock, int level, int optname,
> >   }
> >   EXPORT_SYMBOL(sock_common_setsockopt);
> >   
> > +int sock_common_uring_cmd(struct socket *sock, struct io_uring_cmd *cmd,
> > +			  unsigned int issue_flags)
> > +{
> > +	struct sock *sk = sock->sk;
> > +
> > +	if (!sk->sk_prot || !sk->sk_prot->uring_cmd)
> > +		return -EOPNOTSUPP;
> > +
> > +	return sk->sk_prot->uring_cmd(sk, cmd, issue_flags);
> > +}
> > +EXPORT_SYMBOL(sock_common_uring_cmd);
> > +
>
>
> io_uring is just another in-kernel user of sockets. There is no reason
> for io_uring references to be in core net code. It should be using
> exposed in-kernel APIs and doing any translation of its op codes in
> io_uring/  code.

That callback is all about file dependent operations, just like ioctl.
And as the patch in question is doing socket specific stuff, I think
architecturally it fits well. I also believe Breno wants to extend it
later to support more operations.

Sockets are a large chunk of use cases, it can be implemented as a
separate io_uring request type if nothing else works, but in general
that might not be as scalable.

--
Pavel Begunkov