2020-02-25 00:02:18

by Abhishek Pandit-Subedi

[permalink] [raw]
Subject: [RFC PATCH v3 1/5] Bluetooth: Add mgmt op set_wake_capable

When the system is suspended, only some connected Bluetooth devices
cause user input that should wake the system (mostly HID devices). Add
a list to keep track of devices that can wake the system and add
a management API to let userspace tell the kernel whether a device is
wake capable or not. For LE devices, the wakeable property is added to
the connection parameter and can only be modified after calling
add_device.

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

Changes in v3: None
Changes in v2: None

include/net/bluetooth/hci_core.h | 2 ++
include/net/bluetooth/mgmt.h | 7 +++++
net/bluetooth/hci_core.c | 1 +
net/bluetooth/mgmt.c | 51 ++++++++++++++++++++++++++++++++
4 files changed, 61 insertions(+)

diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h
index dcc0dc6e2624..9d9ada5bc9d4 100644
--- a/include/net/bluetooth/hci_core.h
+++ b/include/net/bluetooth/hci_core.h
@@ -394,6 +394,7 @@ struct hci_dev {
struct list_head mgmt_pending;
struct list_head blacklist;
struct list_head whitelist;
+ struct list_head wakeable;
struct list_head uuids;
struct list_head link_keys;
struct list_head long_term_keys;
@@ -575,6 +576,7 @@ struct hci_conn_params {

struct hci_conn *conn;
bool explicit_connect;
+ bool wakeable;
};

extern struct list_head hci_dev_list;
diff --git a/include/net/bluetooth/mgmt.h b/include/net/bluetooth/mgmt.h
index a90666af05bd..283ba5320bdb 100644
--- a/include/net/bluetooth/mgmt.h
+++ b/include/net/bluetooth/mgmt.h
@@ -671,6 +671,13 @@ struct mgmt_cp_set_blocked_keys {
} __packed;
#define MGMT_OP_SET_BLOCKED_KEYS_SIZE 2

+#define MGMT_OP_SET_WAKE_CAPABLE 0x0047
+#define MGMT_SET_WAKE_CAPABLE_SIZE 8
+struct mgmt_cp_set_wake_capable {
+ struct mgmt_addr_info addr;
+ u8 wake_capable;
+} __packed;
+
#define MGMT_EV_CMD_COMPLETE 0x0001
struct mgmt_ev_cmd_complete {
__le16 opcode;
diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
index cbbc34a006d1..2fceaf76644a 100644
--- a/net/bluetooth/hci_core.c
+++ b/net/bluetooth/hci_core.c
@@ -3299,6 +3299,7 @@ struct hci_dev *hci_alloc_dev(void)
INIT_LIST_HEAD(&hdev->mgmt_pending);
INIT_LIST_HEAD(&hdev->blacklist);
INIT_LIST_HEAD(&hdev->whitelist);
+ INIT_LIST_HEAD(&hdev->wakeable);
INIT_LIST_HEAD(&hdev->uuids);
INIT_LIST_HEAD(&hdev->link_keys);
INIT_LIST_HEAD(&hdev->long_term_keys);
diff --git a/net/bluetooth/mgmt.c b/net/bluetooth/mgmt.c
index 3074363c68df..9a873097000b 100644
--- a/net/bluetooth/mgmt.c
+++ b/net/bluetooth/mgmt.c
@@ -107,6 +107,7 @@ static const u16 mgmt_commands[] = {
MGMT_OP_READ_EXT_INFO,
MGMT_OP_SET_APPEARANCE,
MGMT_OP_SET_BLOCKED_KEYS,
+ MGMT_OP_SET_WAKE_CAPABLE,
};

static const u16 mgmt_events[] = {
@@ -4663,6 +4664,48 @@ static int set_fast_connectable(struct sock *sk, struct hci_dev *hdev,
return err;
}

+static int set_wake_capable(struct sock *sk, struct hci_dev *hdev, void *data,
+ u16 len)
+{
+ struct mgmt_cp_set_wake_capable *cp = data;
+ struct hci_conn_params *params;
+ int err;
+ u8 status = MGMT_STATUS_FAILED;
+ u8 addr_type = cp->addr.type == BDADDR_BREDR ?
+ cp->addr.type :
+ le_addr_type(cp->addr.type);
+
+ BT_DBG("Set wake capable %pMR (type 0x%x) = 0x%x\n", &cp->addr.bdaddr,
+ addr_type, cp->wake_capable);
+
+ if (cp->addr.type == BDADDR_BREDR) {
+ if (cp->wake_capable)
+ err = hci_bdaddr_list_add(&hdev->wakeable,
+ &cp->addr.bdaddr, addr_type);
+ else
+ err = hci_bdaddr_list_del(&hdev->wakeable,
+ &cp->addr.bdaddr, addr_type);
+
+ if (!err || err == -EEXIST || err == -ENOENT)
+ status = MGMT_STATUS_SUCCESS;
+
+ goto done;
+ }
+
+ /* Add wakeable param to le connection parameters */
+ params = hci_conn_params_lookup(hdev, &cp->addr.bdaddr, addr_type);
+ if (params) {
+ params->wakeable = cp->wake_capable;
+ status = MGMT_STATUS_SUCCESS;
+ }
+
+done:
+ err = mgmt_cmd_complete(sk, hdev->id, MGMT_OP_SET_WAKE_CAPABLE, status,
+ cp, sizeof(*cp));
+
+ return err;
+}
+
static void set_bredr_complete(struct hci_dev *hdev, u8 status, u16 opcode)
{
struct mgmt_pending_cmd *cmd;
@@ -5791,6 +5834,13 @@ static int remove_device(struct sock *sk, struct hci_dev *hdev,
err = hci_bdaddr_list_del(&hdev->whitelist,
&cp->addr.bdaddr,
cp->addr.type);
+
+ /* Don't check result since it either succeeds or device
+ * wasn't there (not wakeable or invalid params as
+ * covered by deleting from whitelist).
+ */
+ hci_bdaddr_list_del(&hdev->wakeable, &cp->addr.bdaddr,
+ cp->addr.type);
if (err) {
err = mgmt_cmd_complete(sk, hdev->id,
MGMT_OP_REMOVE_DEVICE,
@@ -6990,6 +7040,7 @@ static const struct hci_mgmt_handler mgmt_handlers[] = {
{ set_phy_configuration, MGMT_SET_PHY_CONFIGURATION_SIZE },
{ set_blocked_keys, MGMT_OP_SET_BLOCKED_KEYS_SIZE,
HCI_MGMT_VAR_LEN },
+ { set_wake_capable, MGMT_SET_WAKE_CAPABLE_SIZE },
};

void mgmt_index_added(struct hci_dev *hdev)
--
2.25.0.265.gbab2e86ba0-goog


2020-03-02 22:17:49

by Marcel Holtmann

[permalink] [raw]
Subject: Re: [RFC PATCH v3 1/5] Bluetooth: Add mgmt op set_wake_capable

Hi Abhishek,

> I seem to have forgotten to update the series changes here. In series
> 3, I added a wakeable property to le_conn_param so that the wakeable
> list is only used for BR/EDR as requested in the previous revision.

are you sending a v4?

Regards

Marcel

2020-03-04 01:11:39

by Abhishek Pandit-Subedi

[permalink] [raw]
Subject: Re: [RFC PATCH v3 1/5] Bluetooth: Add mgmt op set_wake_capable

Hi Marcel,

Yes, I've sent up a v4. While enabling and testing these changes, I
found a bug (which I had originally implemented but missed during
various merges) where powered down and powering down adapters weren't
working.

I've ran the most recent series through a set of automated tests (on
4.19, 4.14, 4.4 and 3.18 kernels) and they are all passing now.

Thanks
Abhishek

On Mon, Mar 2, 2020 at 2:17 PM Marcel Holtmann <[email protected]> wrote:
>
> Hi Abhishek,
>
> > I seem to have forgotten to update the series changes here. In series
> > 3, I added a wakeable property to le_conn_param so that the wakeable
> > list is only used for BR/EDR as requested in the previous revision.
>
> are you sending a v4?
>
> Regards
>
> Marcel
>