2020-04-02 15:26:28

by Sathish Narasimman

[permalink] [raw]
Subject: [PATCH v6 1/2] bluetooth: add support to notify using SCO air_mode

From: Sathish Narsimman <[email protected]>

notifying using HCI_NOTIFY_CONN_ADD for SCO connection is generic in
case of msbc audio. To differnetiate SCO air mode introducing
HCI_NOTIFY_ENABLE_SCO_CVSD and HCI_NOTIFY_ENABLE_SCO_TRANSP.

Signed-off-by: Sathish Narsimman <[email protected]>
---
include/net/bluetooth/hci.h | 3 +++
include/net/bluetooth/hci_core.h | 3 +++
net/bluetooth/hci_conn.c | 19 +++++++++++++++----
net/bluetooth/hci_event.c | 15 ++++++++++++++-
4 files changed, 35 insertions(+), 5 deletions(-)

diff --git a/include/net/bluetooth/hci.h b/include/net/bluetooth/hci.h
index 5f60e135aeb6..9ff2f7a9e131 100644
--- a/include/net/bluetooth/hci.h
+++ b/include/net/bluetooth/hci.h
@@ -53,6 +53,9 @@
#define HCI_NOTIFY_CONN_ADD 1
#define HCI_NOTIFY_CONN_DEL 2
#define HCI_NOTIFY_VOICE_SETTING 3
+#define HCI_NOTIFY_ENABLE_SCO_CVSD 4
+#define HCI_NOTIFY_ENABLE_SCO_TRANSP 5
+#define HCI_NOTIFY_DISABLE_SCO 6

/* HCI bus types */
#define HCI_VIRTUAL 0
diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h
index d4e28773d378..1d1eb802f720 100644
--- a/include/net/bluetooth/hci_core.h
+++ b/include/net/bluetooth/hci_core.h
@@ -1662,4 +1662,7 @@ void hci_copy_identity_address(struct hci_dev *hdev, bdaddr_t *bdaddr,
#define SCO_AIRMODE_CVSD 0x0000
#define SCO_AIRMODE_TRANSP 0x0003

+#define SCO_CODED_CVSD 0x02
+#define SCO_CODED_TRANSP 0x03
+
#endif /* __HCI_CORE_H */
diff --git a/net/bluetooth/hci_conn.c b/net/bluetooth/hci_conn.c
index e245bc155cc2..613b6b86f170 100644
--- a/net/bluetooth/hci_conn.c
+++ b/net/bluetooth/hci_conn.c
@@ -122,8 +122,13 @@ static void hci_conn_cleanup(struct hci_conn *conn)

hci_conn_hash_del(hdev, conn);

- if (hdev->notify)
- hdev->notify(hdev, HCI_NOTIFY_CONN_DEL);
+ if (conn->type == SCO_LINK || conn->type == ESCO_LINK) {
+ if (hdev->notify)
+ hdev->notify(hdev, HCI_NOTIFY_DISABLE_SCO);
+ } else {
+ if (hdev->notify)
+ hdev->notify(hdev, HCI_NOTIFY_CONN_DEL);
+ }

hci_conn_del_sysfs(conn);

@@ -577,8 +582,14 @@ struct hci_conn *hci_conn_add(struct hci_dev *hdev, int type, bdaddr_t *dst,
hci_dev_hold(hdev);

hci_conn_hash_add(hdev, conn);
- if (hdev->notify)
- hdev->notify(hdev, HCI_NOTIFY_CONN_ADD);
+
+ /* As HCI_NOTIFY_ENABLE_SCO_CVSD & HCI_NOTIFY_ENABLE_SCO_TRANSP is
+ * used to notify SCO connections during sync_conn_complete event
+ */
+ if (conn->type != SCO_LINK && conn->type != ESCO_LINK) {
+ if (hdev->notify)
+ hdev->notify(hdev, HCI_NOTIFY_CONN_ADD);
+ }

hci_conn_init_sysfs(conn);

diff --git a/net/bluetooth/hci_event.c b/net/bluetooth/hci_event.c
index 0a591be8b0ae..d6fd99029ca6 100644
--- a/net/bluetooth/hci_event.c
+++ b/net/bluetooth/hci_event.c
@@ -2607,8 +2607,12 @@ static void hci_conn_complete_evt(struct hci_dev *hdev, struct sk_buff *skb)
if (ev->status) {
hci_connect_cfm(conn, ev->status);
hci_conn_del(conn);
- } else if (ev->link_type != ACL_LINK)
+ } else if (ev->link_type == SCO_LINK) {
+ if (hdev->notify)
+ hdev->notify(hdev, HCI_NOTIFY_ENABLE_SCO_CVSD);
+
hci_connect_cfm(conn, ev->status);
+ }

unlock:
hci_dev_unlock(hdev);
@@ -4307,6 +4311,15 @@ static void hci_sync_conn_complete_evt(struct hci_dev *hdev,
break;
}

+ BT_DBG("sco connected with air mode: %02x", ev->air_mode);
+ if (ev->air_mode == SCO_CODED_CVSD) {
+ if (hdev->notify)
+ hdev->notify(hdev, HCI_NOTIFY_ENABLE_SCO_CVSD);
+ } else if (ev->air_mode == SCO_CODED_TRANSP) {
+ if (hdev->notify)
+ hdev->notify(hdev, HCI_NOTIFY_ENABLE_SCO_TRANSP);
+ }
+
hci_connect_cfm(conn, ev->status);
if (ev->status)
hci_conn_del(conn);
--
2.17.1


2020-04-02 15:28:10

by Sathish Narasimman

[permalink] [raw]
Subject: [PATCH v7 2/2] bluetooth:btusb: handle msbc audio over USB Endpoints

From: Sathish Narasimman <[email protected]>

For msbc encoded audio stream over usb transport, btusb driver
to be set to alternate settings 6 as per BT core spec 5.0. The
type of air mode is used to differenting which alt setting to be
used.
The changes are made considering some discussion over the similar
patch submitted earlier from Kuba Pawlak(link below)
https://www.spinics.net/lists/linux-bluetooth/msg64577.html

Reported-by: kbuild test robot <[email protected]>
Signed-off-by: Sathish Narasimman <[email protected]>
Signed-off-by: Chethan T N <[email protected]>
Signed-off-by: Hsin-Yu Chao <[email protected]>
Signed-off-by: Amit K Bag <[email protected]>
---
drivers/bluetooth/btusb.c | 155 +++++++++++++++++++++++++++++---------
1 file changed, 118 insertions(+), 37 deletions(-)

diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
index 3bdec42c9612..be207fd49e6f 100644
--- a/drivers/bluetooth/btusb.c
+++ b/drivers/bluetooth/btusb.c
@@ -492,6 +492,8 @@ struct btusb_data {
__u8 cmdreq;

unsigned int sco_num;
+ unsigned int air_mode;
+ bool usb_alt6_packet_flow;
int isoc_altsetting;
int suspend_count;

@@ -983,6 +985,42 @@ static void btusb_isoc_complete(struct urb *urb)
}
}

+static inline void __fill_isoc_descriptor_msbc(struct urb *urb, int len,
+ int mtu, struct btusb_data *data)
+{
+ int i, offset = 0;
+ unsigned int interval;
+
+ BT_DBG("len %d mtu %d", len, mtu);
+
+ /* For msbc ALT 6 setting the host will send the packet at continuous
+ * flow. As per core spec 5, vol 4, part B, table 2.1. For ALT setting
+ * 6 the HCI PACKET INTERVAL should be 7.5ms for every usb packets.
+ * To maintain the rate we send 63bytes of usb packets alternatively for
+ * 7ms and 8ms to maintain the rate as 7.5ms.
+ */
+ if (data->usb_alt6_packet_flow) {
+ interval = 7;
+ data->usb_alt6_packet_flow = false;
+ } else {
+ interval = 6;
+ data->usb_alt6_packet_flow = true;
+ }
+
+ for (i = 0; i < interval; i++) {
+ urb->iso_frame_desc[i].offset = offset;
+ urb->iso_frame_desc[i].length = offset;
+ }
+
+ if (len && i < BTUSB_MAX_ISOC_FRAMES) {
+ urb->iso_frame_desc[i].offset = offset;
+ urb->iso_frame_desc[i].length = len;
+ i++;
+ }
+
+ urb->number_of_packets = i;
+}
+
static inline void __fill_isoc_descriptor(struct urb *urb, int len, int mtu)
{
int i, offset = 0;
@@ -1386,9 +1424,13 @@ static struct urb *alloc_isoc_urb(struct hci_dev *hdev, struct sk_buff *skb)

urb->transfer_flags = URB_ISO_ASAP;

- __fill_isoc_descriptor(urb, skb->len,
- le16_to_cpu(data->isoc_tx_ep->wMaxPacketSize));
-
+ if (data->isoc_altsetting == 6)
+ __fill_isoc_descriptor_msbc(urb, skb->len,
+ le16_to_cpu(data->isoc_tx_ep->wMaxPacketSize),
+ data);
+ else
+ __fill_isoc_descriptor(urb, skb->len,
+ le16_to_cpu(data->isoc_tx_ep->wMaxPacketSize));
skb->dev = (void *)hdev;

return urb;
@@ -1484,6 +1526,7 @@ static void btusb_notify(struct hci_dev *hdev, unsigned int evt)

if (hci_conn_num(hdev, SCO_LINK) != data->sco_num) {
data->sco_num = hci_conn_num(hdev, SCO_LINK);
+ data->air_mode = evt;
schedule_work(&data->work);
}
}
@@ -1531,11 +1574,65 @@ static inline int __set_isoc_interface(struct hci_dev *hdev, int altsetting)
return 0;
}

+static int bt_switch_alt_setting(struct hci_dev *hdev, int new_alts)
+{
+ struct btusb_data *data = hci_get_drvdata(hdev);
+ int err;
+
+ if (data->isoc_altsetting != new_alts) {
+ unsigned long flags;
+
+ clear_bit(BTUSB_ISOC_RUNNING, &data->flags);
+ usb_kill_anchored_urbs(&data->isoc_anchor);
+
+ /* When isochronous alternate setting needs to be
+ * changed, because SCO connection has been added
+ * or removed, a packet fragment may be left in the
+ * reassembling state. This could lead to wrongly
+ * assembled fragments.
+ *
+ * Clear outstanding fragment when selecting a new
+ * alternate setting.
+ */
+ spin_lock_irqsave(&data->rxlock, flags);
+ kfree_skb(data->sco_skb);
+ data->sco_skb = NULL;
+ spin_unlock_irqrestore(&data->rxlock, flags);
+
+ err = __set_isoc_interface(hdev, new_alts);
+ if (err < 0)
+ return err;
+ }
+ if (!test_and_set_bit(BTUSB_ISOC_RUNNING, &data->flags)) {
+ if (btusb_submit_isoc_urb(hdev, GFP_KERNEL) < 0)
+ clear_bit(BTUSB_ISOC_RUNNING, &data->flags);
+ else
+ btusb_submit_isoc_urb(hdev, GFP_KERNEL);
+ }
+
+ return 0;
+}
+
+static struct usb_host_interface *btusb_find_altsetting(struct btusb_data *data, int alt)
+{
+ struct usb_interface *intf = data->isoc;
+ int i;
+
+ BT_DBG("Looking for Alt no :%d", alt);
+
+ for (i = 0; i < intf->num_altsetting; i++) {
+ if (intf->altsetting[i].desc.bAlternateSetting == alt)
+ return &intf->altsetting[i];
+ }
+
+ return NULL;
+}
+
static void btusb_work(struct work_struct *work)
{
struct btusb_data *data = container_of(work, struct btusb_data, work);
struct hci_dev *hdev = data->hdev;
- int new_alts;
+ int new_alts = 0;
int err;

if (data->sco_num > 0) {
@@ -1550,44 +1647,28 @@ static void btusb_work(struct work_struct *work)
set_bit(BTUSB_DID_ISO_RESUME, &data->flags);
}

- if (hdev->voice_setting & 0x0020) {
- static const int alts[3] = { 2, 4, 5 };
-
- new_alts = alts[data->sco_num - 1];
- } else {
- new_alts = data->sco_num;
- }
-
- if (data->isoc_altsetting != new_alts) {
- unsigned long flags;
+ if (data->air_mode == HCI_NOTIFY_ENABLE_SCO_CVSD) {
+ if (hdev->voice_setting & 0x0020) {
+ static const int alts[3] = { 2, 4, 5 };

- clear_bit(BTUSB_ISOC_RUNNING, &data->flags);
- usb_kill_anchored_urbs(&data->isoc_anchor);
-
- /* When isochronous alternate setting needs to be
- * changed, because SCO connection has been added
- * or removed, a packet fragment may be left in the
- * reassembling state. This could lead to wrongly
- * assembled fragments.
- *
- * Clear outstanding fragment when selecting a new
- * alternate setting.
- */
- spin_lock_irqsave(&data->rxlock, flags);
- kfree_skb(data->sco_skb);
- data->sco_skb = NULL;
- spin_unlock_irqrestore(&data->rxlock, flags);
+ new_alts = alts[data->sco_num - 1];
+ } else {
+ new_alts = data->sco_num;
+ }
+ } else if (data->air_mode == HCI_NOTIFY_ENABLE_SCO_TRANSP) {

- if (__set_isoc_interface(hdev, new_alts) < 0)
- return;
- }
+ data->usb_alt6_packet_flow = true;

- if (!test_and_set_bit(BTUSB_ISOC_RUNNING, &data->flags)) {
- if (btusb_submit_isoc_urb(hdev, GFP_KERNEL) < 0)
- clear_bit(BTUSB_ISOC_RUNNING, &data->flags);
+ /* Check if Alt 6 is supported for Transparent audio*/
+ if (btusb_find_altsetting(data, 6))
+ new_alts = 6;
else
- btusb_submit_isoc_urb(hdev, GFP_KERNEL);
+ bt_dev_err(hdev, "(%s) Device does not support ALT setting 6",
+ hdev->name);
}
+
+ if (bt_switch_alt_setting(hdev, new_alts) < 0)
+ bt_dev_err(hdev, "(%s) set USB alt:(%d) failed!", hdev->name, new_alts);
} else {
clear_bit(BTUSB_ISOC_RUNNING, &data->flags);
usb_kill_anchored_urbs(&data->isoc_anchor);
--
2.17.1

2020-04-03 17:54:24

by Marcel Holtmann

[permalink] [raw]
Subject: Re: [PATCH v7 2/2] bluetooth:btusb: handle msbc audio over USB Endpoints

Hi Sathish,

> For msbc encoded audio stream over usb transport, btusb driver
> to be set to alternate settings 6 as per BT core spec 5.0. The
> type of air mode is used to differenting which alt setting to be
> used.
> The changes are made considering some discussion over the similar
> patch submitted earlier from Kuba Pawlak(link below)
> https://www.spinics.net/lists/linux-bluetooth/msg64577.html
>
> Reported-by: kbuild test robot <[email protected]>
> Signed-off-by: Sathish Narasimman <[email protected]>
> Signed-off-by: Chethan T N <[email protected]>
> Signed-off-by: Hsin-Yu Chao <[email protected]>
> Signed-off-by: Amit K Bag <[email protected]>
> ---
> drivers/bluetooth/btusb.c | 155 +++++++++++++++++++++++++++++---------
> 1 file changed, 118 insertions(+), 37 deletions(-)

I decided to look at these to patches and fix up a few things.

1) Use conn->setting & SCO_AIRMODE_MASK to select the option and with that also ensure that we don’t send this for ulaw and alaw air modes if they ever get established (even if unlikely).

2) Fix a bunch of coding style violation

Please test these two patches if they still work according and then I can apply them.

Regards

Marcel

2020-04-06 09:35:51

by Sathish Narasimman

[permalink] [raw]
Subject: Re: [PATCH v7 2/2] bluetooth:btusb: handle msbc audio over USB Endpoints

Hi Marcel


On Fri, Apr 3, 2020 at 11:24 PM Marcel Holtmann <[email protected]> wrote:
>
> Hi Sathish,
>
> > For msbc encoded audio stream over usb transport, btusb driver
> > to be set to alternate settings 6 as per BT core spec 5.0. The
> > type of air mode is used to differenting which alt setting to be
> > used.
> > The changes are made considering some discussion over the similar
> > patch submitted earlier from Kuba Pawlak(link below)
> > https://www.spinics.net/lists/linux-bluetooth/msg64577.html
> >
> > Reported-by: kbuild test robot <[email protected]>
> > Signed-off-by: Sathish Narasimman <[email protected]>
> > Signed-off-by: Chethan T N <[email protected]>
> > Signed-off-by: Hsin-Yu Chao <[email protected]>
> > Signed-off-by: Amit K Bag <[email protected]>
> > ---
> > drivers/bluetooth/btusb.c | 155 +++++++++++++++++++++++++++++---------
> > 1 file changed, 118 insertions(+), 37 deletions(-)
>
> I decided to look at these to patches and fix up a few things.
>
> 1) Use conn->setting & SCO_AIRMODE_MASK to select the option and with that also ensure that we don’t send this for ulaw and alaw air modes if they ever get established (even if unlikely).
>
> 2) Fix a bunch of coding style violation
>
> Please test these two patches if they still work according and then I can apply them.
>
> Regards
>
> Marcel
>
Tested these 2 patches and they are working fine

Thanks
Sathish N

2020-04-06 11:52:18

by Marcel Holtmann

[permalink] [raw]
Subject: Re: [PATCH v7 2/2] bluetooth:btusb: handle msbc audio over USB Endpoints

Hi Sathish,

>>> For msbc encoded audio stream over usb transport, btusb driver
>>> to be set to alternate settings 6 as per BT core spec 5.0. The
>>> type of air mode is used to differenting which alt setting to be
>>> used.
>>> The changes are made considering some discussion over the similar
>>> patch submitted earlier from Kuba Pawlak(link below)
>>> https://www.spinics.net/lists/linux-bluetooth/msg64577.html
>>>
>>> Reported-by: kbuild test robot <[email protected]>
>>> Signed-off-by: Sathish Narasimman <[email protected]>
>>> Signed-off-by: Chethan T N <[email protected]>
>>> Signed-off-by: Hsin-Yu Chao <[email protected]>
>>> Signed-off-by: Amit K Bag <[email protected]>
>>> ---
>>> drivers/bluetooth/btusb.c | 155 +++++++++++++++++++++++++++++---------
>>> 1 file changed, 118 insertions(+), 37 deletions(-)
>>
>> I decided to look at these to patches and fix up a few things.
>>
>> 1) Use conn->setting & SCO_AIRMODE_MASK to select the option and with that also ensure that we don’t send this for ulaw and alaw air modes if they ever get established (even if unlikely).
>>
>> 2) Fix a bunch of coding style violation
>>
>> Please test these two patches if they still work according and then I can apply them.
>>
>> Regards
>>
>> Marcel
>>
> Tested these 2 patches and they are working fine

good since Johan already applied them to bluetooth-next tree.

Regards

Marcel