[Bluez-devel] Re: Re: Bluetooth SCO message sending (sco_send_frame)

[Bluez-devel] Re: Re: Bluetooth SCO message sending (sco_send_frame)

[Youngwoo Park]

I apologize my inexperience..

This is unified diff result..


--- linux-2.6.12/net/bluetooth/sco.c 2005-06-18 04:48:29.000000000 +0900
+++ linux-2.6.12-patch/net/bluetooth/sco.c 2005-08-24 14:33:54.517717440
+0900
@@ -234,35 +234,50 @@
return err;
}

+/**
+ * sco_send_frame
+ *
+ * Modified 2005 by Youngwoo Park
+ *
+ * Send large sco frame sequentially even if frame size is larger than MTU
+ *
+ **/
static inline int sco_send_frame(struct sock *sk, struct msghdr *msg,
int len)
{
- struct sco_conn *conn = sco_pi(sk)->conn;
- struct sk_buff *skb;
- int err, count;
+ struct sco_conn *conn = sco_pi(sk)->conn;
+ struct sk_buff *skb;
+ int err, count;

- /* Check outgoing MTU */
- if (len > conn->mtu)
- return -EINVAL;
+ do {

- BT_DBG("sk %p len %d", sk, len);
+ BT_DBG("sk %p len %d", sk, len);

- count = min_t(unsigned int, conn->mtu, len);
- if (!(skb = bt_skb_send_alloc(sk, count, msg->msg_flags &
MSG_DONTWAIT, &err)))
- return err;
+ count = min_t(unsigned int, conn->mtu, len);
+ BT_DBG("sk %p len %d", sk, len);

- if (memcpy_fromiovec(skb_put(skb, count), msg->msg_iov, count)) {
- err = -EFAULT;
- goto fail;
- }
+ if (!(skb = bt_skb_send_alloc(sk, count, msg->msg_flags &
MSG_DONTWAIT, &err))) {
+ BT_DBG("Faile to alloc");
+ return err;
+ }
+
+ if (memcpy_fromiovec(skb_put(skb, count), msg->msg_iov, count)) {
+ err = -EFAULT;
+ goto fail;
+ }

- if ((err = hci_send_sco(conn->hcon, skb)) < 0)
- goto fail;
+ if ((err = hci_send_sco(conn->hcon, skb)) < 0)
+ goto fail;

- return count;
+ len -= count;
+
+ } while(len > 0);
+
+ return len;

fail:
- kfree_skb(skb);
- return err;
+ BT_DBG("Fail To Send", sk, len);
+ kfree_skb(skb);
+ return err;
}

static inline void sco_recv_frame(struct sco_conn *conn, struct sk_buff
*skb)



If my diff format or coding style still has a problem.
Please, let me give a hint about that.

Thank you.


Regards

Park.

-- 
Youngwoo Park

M.S. Student
CORE(COmputer engineering REsearch) Lab
Dept. EECS, KAIST, Daejeon 305-701, Korea
Phone: +82-42-869-5425 Fax: +82-42-8695425
e-mail: ywpark@core.kaist.ac.kr

--




















-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Re: Re: Bluetooth SCO message sending (sco_send_frame)

[Marcel Holtmann]

Hi Youngwoo,

> I apologize my inexperience..
> 
> This is unified diff result..

you screwed it up again. When I talk about coding style, I really mean
an indentation with a tab that is 8 characters width. Your code is too
hard to read and so I won't do it.

Regards

Marcel




-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

[Bluez-devel] Re: Re: Bluetooth SCO message sending (sco_send_frame)

[Young-woo Park]

I'm really apologize my mistake.
The problem was my e-mail sending program.
Of course, I follow up the bluez-coding style.
However the e-mail program sent message as a plain text removing tab!!!
I change my mail server now.


It's all my fault that I don't confirm my message.


Here is my patch..



--- linux-2.6.12/net/bluetooth/sco.c=092005-06-18 04:48:29.000000000 +0900
+++ linux-2.6.12-patch/net/bluetooth/sco.c=092005-08-25 10:58:44.537536256 =
+0900
@@ -234,33 +234,48 @@
 =09return err;
 }
=20
+/**
+ *  sco_send_frame
+ *
+ *  Modified 2005 by Youngwoo Park
+ *
+ *  Send large sco frame sequentially even if frame size is larger than MT=
U
+ *
+ **/
 static inline int sco_send_frame(struct sock *sk, struct msghdr *msg, int =
len)
 {
 =09struct sco_conn *conn =3D sco_pi(sk)->conn;
 =09struct sk_buff *skb;
 =09int err, count;
=20
-=09/* Check outgoing MTU */
-=09if (len > conn->mtu)
-=09=09return -EINVAL;
+=09do {
=20
-=09BT_DBG("sk %p len %d", sk, len);
+=09=09BT_DBG("sk %p len %d", sk, len);
=20
-=09count =3D min_t(unsigned int, conn->mtu, len);
-=09if (!(skb =3D bt_skb_send_alloc(sk, count, msg->msg_flags &
MSG_DONTWAIT, &err)))
-=09=09return err;
+=09=09count =3D min_t(unsigned int, conn->mtu, len);
+=09=09BT_DBG("sk %p len %d", sk, len);
=20
-=09if (memcpy_fromiovec(skb_put(skb, count), msg->msg_iov, count)) {
-=09=09err =3D -EFAULT;
-=09=09goto fail;
-=09}
+=09=09if (!(skb =3D bt_skb_send_alloc(sk, count, msg->msg_flags &
MSG_DONTWAIT, &err))) {
+=09=09=09BT_DBG("Faile to alloc");
+=09=09=09return err;
+=09=09}
+
+=09=09if (memcpy_fromiovec(skb_put(skb, count), msg->msg_iov, count)) {
+=09=09=09err =3D -EFAULT;
+=09=09=09goto fail;
+=09=09}
+
+=09=09if ((err =3D hci_send_sco(conn->hcon, skb)) < 0)
+=09=09=09goto fail;
+
+=09=09len -=3D count;
=20
-=09if ((err =3D hci_send_sco(conn->hcon, skb)) < 0)
-=09=09goto fail;
+=09} while(len > 0);
=20
-=09return count;
+=09return len;
=20
 fail:
+=09BT_DBG("Fail To Send", sk, len);
 =09kfree_skb(skb);
 =09return err;
 }

Regards

Park


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Re: Re: Bluetooth SCO message sending (sco_send_frame)

[Marcel Holtmann]

Hi Young-woo,

> Here is my patch..

so the question is, why can't you do the fragmentation inside your
userspace program? I don't see any need to make the kernel deal with it.

Regards

Marcel




-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Re: Re: Bluetooth SCO message sending (sco_send_frame)

[Youngwoo Park]

I used the "btsco" program.
Do you think that "btsco' program is changed to solve my problem?

I think btsco program read data from bluetooth h4 device, and just write
data to bluetooth.
Also, Application has no information about socket and MTU size.
Is it hard to handling data fragmentation in application program?

I think framgmentation is handled by kernel if we need.
Otherwise h4 device driver must guarantee that data frame size is
smaller than MTU.


Regards

Park

-- 
Youngwoo Park

M.S. Student
CORE(COmputer engineering REsearch) Lab
Dept. EECS, KAIST, Daejeon 305-701, Korea
Phone: +82-42-869-5425 Fax: +82-42-8695425
e-mail: ywpark@core.kaist.ac.kr

--




















-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Re: Re: Bluetooth SCO message sending (sco_send_frame)

[Brad Midgley]

Youngwoo

> I used the "btsco" program.
> Do you think that "btsco' program is changed to solve my problem?
> 
> I think btsco program read data from bluetooth h4 device, and just write
> data to bluetooth.
> Also, Application has no information about socket and MTU size.
> Is it hard to handling data fragmentation in application program?
> 
> I think framgmentation is handled by kernel if we need.
> Otherwise h4 device driver must guarantee that data frame size is
> smaller than MTU.

When does this fragmentation problem show up? btsco is not doing
anything special so we expect the kernel to take care of it...

Brad


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel