2021-01-29 21:56:12

by Yu Liu

Subject: [PATCH v3] Bluetooth: Skip eSCO 2M params when not supported

If a peer device doesn't support eSCO 2M we should skip the params that
use it when setting up sync connection since they will always fail.

Signed-off-by: Yu Liu <[email protected]>
Reviewed-by: Abhishek Pandit-Subedi <[email protected]>
---

Changes in v3:
- Use pkt_type instead of adding new field

Changes in v2:
- Fix title

Changes in v1:
- Initial change

include/net/bluetooth/hci_core.h | 1 +
net/bluetooth/hci_conn.c | 20 ++++++++++++++++++--
2 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h
index 239ab72f16c6e..71468a9ea798a 100644
--- a/include/net/bluetooth/hci_core.h
+++ b/include/net/bluetooth/hci_core.h
@@ -1237,6 +1237,7 @@ void hci_conn_del_sysfs(struct hci_conn *conn);
#define lmp_le_capable(dev) ((dev)->features[0][4] & LMP_LE)
#define lmp_sniffsubr_capable(dev) ((dev)->features[0][5] & LMP_SNIFF_SUBR)
#define lmp_pause_enc_capable(dev) ((dev)->features[0][5] & LMP_PAUSE_ENC)
+#define lmp_esco_2m_capable(dev) ((dev)->features[0][5] & LMP_EDR_ESCO_2M)
#define lmp_ext_inq_capable(dev) ((dev)->features[0][6] & LMP_EXT_INQ)
#define lmp_le_br_capable(dev) (!!((dev)->features[0][6] & LMP_SIMUL_LE_BR))
#define lmp_ssp_capable(dev) ((dev)->features[0][6] & LMP_SIMPLE_PAIR)
diff --git a/net/bluetooth/hci_conn.c b/net/bluetooth/hci_conn.c
index 07c34c55fc508..357ce0cfbc5c9 100644
--- a/net/bluetooth/hci_conn.c
+++ b/net/bluetooth/hci_conn.c
@@ -278,6 +278,20 @@ static void hci_add_sco(struct hci_conn *conn, __u16 handle)
hci_send_cmd(hdev, HCI_OP_ADD_SCO, sizeof(cp), &cp);
}

+static bool find_next_esco_param(struct hci_conn *conn,
+ const struct sco_param *esco_param, int size)
+{
+ for (; conn->attempt <= size; conn->attempt++) {
+ if (lmp_esco_2m_capable(conn->link) ||
+ (esco_param[conn->attempt - 1].pkt_type & ESCO_2EV3))
+ break;
+ BT_DBG("hcon %p skipped attempt %d, eSCO 2M not supported",
+ conn, conn->attempt);
+ }
+
+ return conn->attempt <= size;
+}
+
bool hci_setup_sync(struct hci_conn *conn, __u16 handle)
{
struct hci_dev *hdev = conn->hdev;
@@ -299,13 +313,15 @@ bool hci_setup_sync(struct hci_conn *conn, __u16 handle)

switch (conn->setting & SCO_AIRMODE_MASK) {
case SCO_AIRMODE_TRANSP:
- if (conn->attempt > ARRAY_SIZE(esco_param_msbc))
+ if (!find_next_esco_param(conn, esco_param_msbc,
+ ARRAY_SIZE(esco_param_msbc)))
return false;
param = &esco_param_msbc[conn->attempt - 1];
break;
case SCO_AIRMODE_CVSD:
if (lmp_esco_capable(conn->link)) {
- if (conn->attempt > ARRAY_SIZE(esco_param_cvsd))
+ if (!find_next_esco_param(conn, esco_param_cvsd,
+ ARRAY_SIZE(esco_param_cvsd)))
return false;
param = &esco_param_cvsd[conn->attempt - 1];
} else {
--
2.30.0.365.g02bc693789-goog

2021-02-01 16:08:11

by Marcel Holtmann

Subject: Re: [PATCH v3] Bluetooth: Skip eSCO 2M params when not supported

Hi Yu,

> If a peer device doesn't support eSCO 2M we should skip the params that
> use it when setting up sync connection since they will always fail.
>
> Signed-off-by: Yu Liu <[email protected]>
> Reviewed-by: Abhishek Pandit-Subedi <[email protected]>
> ---
>
> Changes in v3:
> - Use pkt_type instead of adding new field
>
> Changes in v2:
> - Fix title
>
> Changes in v1:
> - Initial change
>
> include/net/bluetooth/hci_core.h | 1 +
> net/bluetooth/hci_conn.c | 20 ++++++++++++++++++--
> 2 files changed, 19 insertions(+), 2 deletions(-)

patch has been applied to bluetooth-next tree.

Regards

Marcel