BLE(ATT) mouse doesn't register wakeup source in Linux kernel during BLE mouse connecting

BLE(ATT) mouse doesn't register wakeup source in Linux kernel during BLE mouse connecting

[Ge, Ethan]

Hi,

In file drivers/hid/hid-input.c, function hidinput_setup_battery() will return during connect a BLE mouse because of filed->usage->hid != HID_DC_BATTERYSTRENGTH. And BLE(ATT) HID doesn't register wakeup source in Linux kernel. 

It will cause BLE HID mouse can't wake DUT from S3. Issue can be reproduced by IBUFFALO BSMBB09DS / Dell WM615 mouse / Microsoft arc touch, these all are BLE mouse. 

static bool hidinput_setup_battery(struct hid_device *dev, unsigned report_type, struct hid_field *field)
{
    struct power_supply_desc *psy_desc = NULL;
    struct power_supply_config psy_cfg = { .drv_data = dev, };
    unsigned quirks;
    s32 min, max;

    if (field->usage->hid != HID_DC_BATTERYSTRENGTH)
        return false;    /* no match */

Re: BLE(ATT) mouse doesn't register wakeup source in Linux kernel during BLE mouse connecting

[Luiz Augusto von Dentz]

Hi Ethan,

On Fri, Sep 11, 2015 at 1:12 PM, Ge, Ethan <ethan.ge@intel.com> wrote:
> Hi,
>
> In file drivers/hid/hid-input.c, function hidinput_setup_battery() will return during connect a BLE mouse because of filed->usage->hid != HID_DC_BATTERYSTRENGTH. And BLE(ATT) HID doesn't register wakeup source in Linux kernel.
>
> It will cause BLE HID mouse can't wake DUT from S3. Issue can be reproduced by IBUFFALO BSMBB09DS / Dell WM615 mouse / Microsoft arc touch, these all are BLE mouse.
>
> static bool hidinput_setup_battery(struct hid_device *dev, unsigned report_type, struct hid_field *field)
> {
>     struct power_supply_desc *psy_desc = NULL;
>     struct power_supply_config psy_cfg = { .drv_data = dev, };
>     unsigned quirks;
>     s32 min, max;
>
>     if (field->usage->hid != HID_DC_BATTERYSTRENGTH)
>         return false;    /* no match */

But this is checked in case the device does not support battery
strength support, I thought any input device would register a wakeup
source regardless.


-- 
Luiz Augusto von Dentz

RE: BLE(ATT) mouse doesn't register wakeup source in Linux kernel during BLE mouse connecting

[Ge, Ethan]
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Re: BLE(ATT) mouse doesn't register wakeup source in Linux kernel during BLE mouse connecting

[Ethan]

Hi Luiz,

2015-09-13 20:26 GMT-07:00 Ge, Ethan <ethan.ge@intel.com>:
> Hi Luiz
>
> -----Original Message-----
> From: Luiz Augusto von Dentz [mailto:luiz.dentz@gmail.com]
> Sent: Friday, September 11, 2015 9:36 PM
> To: Ge, Ethan
> Cc: linux-bluetooth@vger.kernel.org; jeremy@goop.org; Marcel Holtmann
> Subject: Re: BLE(ATT) mouse doesn't register wakeup source in Linux kernel during BLE mouse connecting
>
> Hi Ethan,
>
> On Fri, Sep 11, 2015 at 1:12 PM, Ge, Ethan <ethan.ge@intel.com> wrote:
>> Hi,
>>
>> In file drivers/hid/hid-input.c, function hidinput_setup_battery() will return during connect a BLE mouse because of filed->usage->hid != HID_DC_BATTERYSTRENGTH. And BLE(ATT) HID doesn't register wakeup source in Linux kernel.
>>
>> It will cause BLE HID mouse can't wake DUT from S3. Issue can be reproduced by IBUFFALO BSMBB09DS / Dell WM615 mouse / Microsoft arc touch, these all are BLE mouse.
>>
>> static bool hidinput_setup_battery(struct hid_device *dev, unsigned
>> report_type, struct hid_field *field) {
>>     struct power_supply_desc *psy_desc = NULL;
>>     struct power_supply_config psy_cfg = { .drv_data = dev, };
>>     unsigned quirks;
>>     s32 min, max;
>>
>>     if (field->usage->hid != HID_DC_BATTERYSTRENGTH)
>>         return false;    /* no match */
>
>> But this is checked in case the device does not support battery strength support, I thought any input device would register a wakeup source regardless.
>> Luiz Augusto von Dentz
>
> I said that because I verified it by ftrace. I found if it return hidinput_setup_battery(), and input device can't register a wakeup source. Below are the code flow. hidinput_setup_battery->__power_supply_register-> device_init_wakeup.
> In other hand, the BR/EDR mouse can register a wakeup source because it won't return hidinput_setup_battery() and can go through next function. I am trying to check why BLE mouse usage->hid != HID_DC_BATTERYSTRENGTH and can not wakeup system in S3.
>
> static bool hidinput_setup_battery(struct hid_device *dev, unsigned report_type, struct hid_field *field)
> {
>     struct power_supply_desc *psy_desc = NULL;
>     struct power_supply_config psy_cfg = { .drv_data = dev, };
>     unsigned quirks;
>     s32 min, max;
>
>     if (field->usage->hid != HID_DC_BATTERYSTRENGTH)
>         return false;    /* no match */
> ...
>     dev->battery = power_supply_register(&dev->dev, psy_desc, &psy_cfg);
> ..
> }
>
> static struct power_supply *__must_check
> __power_supply_register(struct device *parent,
>                    const struct power_supply_desc *desc,
>                    const struct power_supply_config *cfg,
>                    bool ws)
> {
>  ....
>
>     spin_lock_init(&psy->changed_lock);
>     rc = device_init_wakeup(dev, ws);
>     if (rc)
>         goto wakeup_init_failed;
> ....
> }
>
> int device_init_wakeup(struct device *dev, bool enable)
> {
>     int ret = 0;
>
>     if (!dev)
>         return -EINVAL;
>
>     if (enable) {
>         device_set_wakeup_capable(dev, true);
>         ret = device_wakeup_enable(dev);
> ...
> }
> Ethan
> --

I want to correct my description in previous mail, this issue doesn't
cause from hidinput_setup_battery.  And I found
/sys/bus/usb/devices/1-5/power/wakeup is enabled after BR/EDR HID
connected, but it's disabled after BLE HID connected. If I echo
enabled > /sys/bus/usb/devices/1-5/power/wakeup, and issue will
disappear. The wakeup value is configured from
wake_store->device_set_wakeup_enable(dev, 1) for BR/EDR HID case. I am
trying to figure out why BLE HID won't process it.

Ethan