wiimote_cmd_wait() in wiimod_battery_get_property() may signal that the
task of getting specific battery property was interrupted or timed out.
There is no need to do any further computation in such cases, so just
return the error.
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
Fixes: dcf392313817 ("HID: wiimote: convert BATTERY to module")
Signed-off-by: Nikita Zhandarovich <[email protected]>
---
drivers/hid/hid-wiimote-modules.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/hid/hid-wiimote-modules.c b/drivers/hid/hid-wiimote-modules.c
index dbccdfa63916..9755718d9856 100644
--- a/drivers/hid/hid-wiimote-modules.c
+++ b/drivers/hid/hid-wiimote-modules.c
@@ -220,8 +220,10 @@ static int wiimod_battery_get_property(struct power_supply *psy,
wiiproto_req_status(wdata);
spin_unlock_irqrestore(&wdata->state.lock, flags);
- wiimote_cmd_wait(wdata);
+ ret = wiimote_cmd_wait(wdata);
wiimote_cmd_release(wdata);
+ if (ret)
+ return ret;
spin_lock_irqsave(&wdata->state.lock, flags);
state = wdata->state.cmd_battery;
--
2.25.1
Hi
On Mon, 20 Mar 2023 at 16:34, Nikita Zhandarovich
<[email protected]> wrote:
>
> wiimote_cmd_wait() in wiimod_battery_get_property() may signal that the
> task of getting specific battery property was interrupted or timed out.
> There is no need to do any further computation in such cases, so just
> return the error.
>
> Found by Linux Verification Center (linuxtesting.org) with static
> analysis tool SVACE.
>
> Fixes: dcf392313817 ("HID: wiimote: convert BATTERY to module")
> Signed-off-by: Nikita Zhandarovich <[email protected]>
> ---
> drivers/hid/hid-wiimote-modules.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/hid/hid-wiimote-modules.c b/drivers/hid/hid-wiimote-modules.c
> index dbccdfa63916..9755718d9856 100644
> --- a/drivers/hid/hid-wiimote-modules.c
> +++ b/drivers/hid/hid-wiimote-modules.c
> @@ -220,8 +220,10 @@ static int wiimod_battery_get_property(struct power_supply *psy,
> wiiproto_req_status(wdata);
> spin_unlock_irqrestore(&wdata->state.lock, flags);
>
> - wiimote_cmd_wait(wdata);
> + ret = wiimote_cmd_wait(wdata);
> wiimote_cmd_release(wdata);
> + if (ret)
> + return ret;
The current code returns cached battery-information in case a
synchronous update did not succeed. Battery information is likely
updated regularly, anyway, so the synchronous update is usually not
required.
I don't think bailing out and returning the error to the caller is
required or gains us anything but more complexity. Or am I missing
something here?
Thanks
David
On 3/20/23 12:08, David Rheinsberg wrote:
> Hi
>
> On Mon, 20 Mar 2023 at 16:34, Nikita Zhandarovich
> <[email protected]> wrote:
>>
>> wiimote_cmd_wait() in wiimod_battery_get_property() may signal that the
>> task of getting specific battery property was interrupted or timed out.
>> There is no need to do any further computation in such cases, so just
>> return the error.
>>
>> Found by Linux Verification Center (linuxtesting.org) with static
>> analysis tool SVACE.
>>
>> Fixes: dcf392313817 ("HID: wiimote: convert BATTERY to module")
>> Signed-off-by: Nikita Zhandarovich <[email protected]>
>> ---
>> drivers/hid/hid-wiimote-modules.c | 4 +++-
>> 1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/hid/hid-wiimote-modules.c b/drivers/hid/hid-wiimote-modules.c
>> index dbccdfa63916..9755718d9856 100644
>> --- a/drivers/hid/hid-wiimote-modules.c
>> +++ b/drivers/hid/hid-wiimote-modules.c
>> @@ -220,8 +220,10 @@ static int wiimod_battery_get_property(struct power_supply *psy,
>> wiiproto_req_status(wdata);
>> spin_unlock_irqrestore(&wdata->state.lock, flags);
>>
>> - wiimote_cmd_wait(wdata);
>> + ret = wiimote_cmd_wait(wdata);
>> wiimote_cmd_release(wdata);
>> + if (ret)
>> + return ret;
>
> The current code returns cached battery-information in case a
> synchronous update did not succeed. Battery information is likely
> updated regularly, anyway, so the synchronous update is usually not
> required.
>
> I don't think bailing out and returning the error to the caller is
> required or gains us anything but more complexity. Or am I missing
> something here?
>
> Thanks
> David
Hi. I think you are right, my change is not that essential to begin with
and there is no urgency to patch this.
Thanks for your patience,
Nikita