Re: [Patch bpf-next v4 03/11] skmsg: introduce skb_send_sock() for sock_map

On Thu, Mar 11, 2021 at 3:42 AM Jakub Sitnicki [off-list ref] wrote:

On Wed, Mar 10, 2021 at 06:32 AM CET, Cong Wang wrote:

quoted

From: Cong Wang <redacted>

We only have skb_send_sock_locked() which requires callers
to use lock_sock(). Introduce a variant skb_send_sock()
which locks on its own, callers do not need to lock it
any more. This will save us from adding a ->sendmsg_locked
for each protocol.

To reuse the code, pass function pointers to __skb_send_sock()
and build skb_send_sock() and skb_send_sock_locked() on top.

Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jakub Sitnicki <jakub@cloudflare.com>
Cc: Lorenz Bauer <redacted>
Signed-off-by: Cong Wang <redacted>
---
 include/linux/skbuff.h |  1 +
 net/core/skbuff.c      | 52 ++++++++++++++++++++++++++++++++++++------
 2 files changed, 46 insertions(+), 7 deletions(-)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 0503c917d773..2fc8c3657c53 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h

@@ -3626,6 +3626,7 @@ int skb_splice_bits(struct sk_buff *skb, struct sock *sk, unsigned int offset,
                  unsigned int flags);
 int skb_send_sock_locked(struct sock *sk, struct sk_buff *skb, int offset,
                       int len);
+int skb_send_sock(struct sock *sk, struct sk_buff *skb, int offset, int len);
 void skb_copy_and_csum_dev(const struct sk_buff *skb, u8 *to);
 unsigned int skb_zerocopy_headlen(const struct sk_buff *from);
 int skb_zerocopy(struct sk_buff *to, struct sk_buff *from,

diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 545a472273a5..396586bd6ae3 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c

@@ -2500,9 +2500,12 @@ int skb_splice_bits(struct sk_buff *skb, struct sock *sk, unsigned int offset,
 }
 EXPORT_SYMBOL_GPL(skb_splice_bits);

-/* Send skb data on a socket. Socket must be locked. */
-int skb_send_sock_locked(struct sock *sk, struct sk_buff *skb, int offset,
-                      int len)
+typedef int (*sendmsg_func)(struct sock *sk, struct msghdr *msg,
+                         struct kvec *vec, size_t num, size_t size);
+typedef int (*sendpage_func)(struct sock *sk, struct page *page, int offset,
+                        size_t size, int flags);
+static int __skb_send_sock(struct sock *sk, struct sk_buff *skb, int offset,
+                        int len, sendmsg_func sendmsg, sendpage_func sendpage)
 {
      unsigned int orig_len = len;
      struct sk_buff *head = skb;

@@ -2522,7 +2525,7 @@ int skb_send_sock_locked(struct sock *sk, struct sk_buff *skb, int offset,
              memset(&msg, 0, sizeof(msg));
              msg.msg_flags = MSG_DONTWAIT;

-             ret = kernel_sendmsg_locked(sk, &msg, &kv, 1, slen);
+             ret = sendmsg(sk, &msg, &kv, 1, slen);

Maybe use INDIRECT_CALLABLE_DECLARE() and INDIRECT_CALL_2() since there
are just two possibilities? Same for sendpage below.

Yeah. Actually I wanted to call __skb_send_sock() in espintcp for
tcp_sendmsg(), but it actually could be TCP over IPv6 too, so I decided
not to  touch it.

Thanks.