2013-01-02 14:47:35

by Vinicius Costa Gomes

Subject: [RFC] Bluetooth: Allow getting SCO options for not connected sockets

Now, that we have proper support for deferred setup for SCO sockets it
is very convenient that we are able to get the socket options for
sockets that are not in the 'connected' state.

And makes the behaviour more consistent with what L2CAP does, for
example.

Signed-off-by: Vinicius Costa Gomes <[email protected]>
---

Sending this as a RFC, because even though I couldn't think of any cases
that this new behaviour would surprise userspace applications, there
may be still be some.

net/bluetooth/sco.c | 5 -----
1 file changed, 5 deletions(-)

diff --git a/net/bluetooth/sco.c b/net/bluetooth/sco.c
index 531a93d..271320c 100644
--- a/net/bluetooth/sco.c
+++ b/net/bluetooth/sco.c
@@ -730,11 +730,6 @@ static int sco_sock_getsockopt_old(struct socket *sock, int optname, char __user

switch (optname) {
case SCO_OPTIONS:
- if (sk->sk_state != BT_CONNECTED) {
- err = -ENOTCONN;
- break;
- }
-
opts.mtu = sco_pi(sk)->conn->mtu;

BT_DBG("mtu %d", opts.mtu);
--
1.8.0.2

2013-01-09 21:41:55

by Gustavo Padovan

Subject: Re: [RFC] Bluetooth: Allow getting SCO options for not connected sockets

Hi Vinicius,

* Vinicius Costa Gomes <[email protected]> [2013-01-02 11:47:35 -0300]:

> Now, that we have proper support for deferred setup for SCO sockets it
> is very convenient that we are able to get the socket options for
> sockets that are not in the 'connected' state.
>
> And makes the behaviour more consistent with what L2CAP does, for
> example.
>
> Signed-off-by: Vinicius Costa Gomes <[email protected]>
> ---
>
> Sending this as a RFC, because even though I couldn't think of any cases
> that this new behaviour would surprise userspace applications, there
> may be still be some.
>
> net/bluetooth/sco.c | 5 -----
> 1 file changed, 5 deletions(-)
>
> diff --git a/net/bluetooth/sco.c b/net/bluetooth/sco.c
> index 531a93d..271320c 100644
> --- a/net/bluetooth/sco.c
> +++ b/net/bluetooth/sco.c
> @@ -730,11 +730,6 @@ static int sco_sock_getsockopt_old(struct socket *sock, int optname, char __user
>
> switch (optname) {
> case SCO_OPTIONS:
> - if (sk->sk_state != BT_CONNECTED) {
> - err = -ENOTCONN;
> - break;
> - }
> -
> opts.mtu = sco_pi(sk)->conn->mtu;
>
> BT_DBG("mtu %d", opts.mtu);

I'm ok with this patch, if someone else has something against it, please tell
me.

Gustavo