Re: [bpf PATCH 1/3] net: add a UID to use for ULP socket assignment

On 01/26/2018 07:52 AM, Dave Watson wrote:

On 01/25/18 04:27 PM, John Fastabend wrote:

quoted

I did not however retest TLS with the small change to ULP layer.
Mostly because I don't have a TLS setup. I plan/hope to get around
to writing either a sample or preferably a selftest for this case
as well (assuming I didn't miss one).

quoted

@Dave Watson can you take a quick look to verify the changes are
good on TLS ULP side.

Looks reasonable, and passes my test suite.  One comment below

Tested-by: Dave Watson <redacted>

quoted

Signed-off-by: John Fastabend <john.fastabend@gmail.com>
---
 include/net/tcp.h  |    6 ++++++
 net/ipv4/tcp_ulp.c |   53 +++++++++++++++++++++++++++++++++++++++++++++++-----
 net/tls/tls_main.c |    2 ++
 3 files changed, 56 insertions(+), 5 deletions(-)

diff --git a/net/ipv4/tcp_ulp.c b/net/ipv4/tcp_ulp.c
index 6bb9e14..8ef437d 100644
--- a/net/ipv4/tcp_ulp.c
+++ b/net/ipv4/tcp_ulp.c

@@ -133,3 +157,22 @@ int tcp_set_ulp(struct sock *sk, const char *name)
 	icsk->icsk_ulp_ops = ulp_ops;
 	return 0;
 }
+
+int tcp_set_ulp_id(struct sock *sk, int ulp)
+{
+	struct inet_connection_sock *icsk = inet_csk(sk);
+	const struct tcp_ulp_ops *ulp_ops;
+	int err;
+
+	if (icsk->icsk_ulp_ops)
+		return -EEXIST;
+
+	ulp_ops = __tcp_ulp_lookup(ulp);
+	if (!ulp_ops)
+		return -ENOENT;
+
+	err = ulp_ops->init(sk);
+	if (!err)
+		icsk->icsk_ulp_ops = ulp_ops;

Does this need module_put on error, similar to tcp_set_ulp?

Not needed in current use because its only being used with an in-kernel
ULP. However, best to fix it so that we don't have a (hard to detect)
bug first time someone uses this with a module ULP.

Thanks I'll spin a v2 this morning.

quoted

+	return err;
+}