Re: [PATCH v2] HID: Replace semaphore driver_lock with mutex

[Arnd Bergmann]

On Tue, Jun 13, 2017 at 12:09 PM, Binoy Jayan <binoy.jayan@xxxxxxxxxx> wrote:
> Hi,
>
> On 13 June 2017 at 15:26, Benjamin Tissoires
> <benjamin.tissoires@xxxxxxxxxx> wrote:
>
>>> Looks good to me, but I see you didn't include David and Andrew on
>>> Cc, it would be good for at least one of them to provide an Ack as well.
>>
>> Please also CC linux-input@
>
> Will do that.
>> (one more nitpick below too)
>> A little bit below, there is:
>>         bool io_started;                                                /* Protected by driver_lock. If IO has started */
>>
>> You should probably remove the mention to driver_lock here.
>
> Will remove the reference too.
>
> Thank you for noticing that, initially I missed it as I thought
> 'io_started' somehow
> influences the increment of the semaphore, but its anyway used only in
> hid-core.c

It is also used in hid_device_io_start()  and hid_device_io_stop(), but
what's important here is that these are only ever called from inside of
hid_device_probe() and other functions called by that, so no
synchronization across CPUs is required here.

I think in theory, it could be accessed from below hid_device_remove
as well, but I did not find any instance of that.

        Arnd