On 13. 02. 25, 14:38, Stuart Hayhurst wrote:
corsair_void_process_receiver can be called from an interrupt context, locking battery_mutex in it was causing a kernel panic. Fix it by moving the critical section into its own work, sharing this work with battery_add_work and battery_remove_work to remove the need for any locking
Closes: https://bugzilla.suse.com/show_bug.cgi?id=1236843 Fixes: 6ea2a6fd3872 ("HID: corsair-void: Add Corsair Void headset family driver") Cc: stable@vger.kernel.org Signed-off-by: Stuart Hayhurst stuart.a.hayhurst@gmail.com
Reviewed-by: Jiri Slaby jirislaby@kernel.org
v2 -> v3:
- Use an enum instead of a define for battery flag values
- Use an integer instead of BIT() for the bit index
Good catch :).
- Drop unhelpful comments
- Simplify corsair_void_battery_work_handler logic
- Remove extra newline in commit message
v1 -> v2:
- Actually remove the mutex
drivers/hid/hid-corsair-void.c | 83 ++++++++++++++++++---------------- 1 file changed, 43 insertions(+), 40 deletions(-)
diff --git a/drivers/hid/hid-corsair-void.c b/drivers/hid/hid-corsair-void.c index 56e858066c3c..afbd67aa9719 100644 --- a/drivers/hid/hid-corsair-void.c +++ b/drivers/hid/hid-corsair-void.c
...
@@ -583,16 +567,42 @@ static void corsair_void_battery_add_work_handler(struct work_struct *work) drvdata->battery = new_supply; } +static void corsair_void_battery_work_handler(struct work_struct *work) +{
- struct corsair_void_drvdata *drvdata = container_of(work,
struct corsair_void_drvdata, battery_work);- bool add_battery = test_and_clear_bit(CORSAIR_VOID_ADD_BATTERY,
&drvdata->battery_work_flags);- bool remove_battery = test_and_clear_bit(CORSAIR_VOID_REMOVE_BATTERY,
&drvdata->battery_work_flags);- bool update_battery = test_and_clear_bit(CORSAIR_VOID_UPDATE_BATTERY,
&drvdata->battery_work_flags);- if (add_battery && !remove_battery) {
corsair_void_add_battery(drvdata);- } else if (remove_battery && !add_battery && drvdata->battery) {
power_supply_unregister(drvdata->battery);drvdata->battery = NULL;- }
Now I think, what is actually expected to happen if both add_battery and remove_battery is set? Do nothing as the code does?
- if (update_battery && drvdata->battery)
power_supply_changed(drvdata->battery);+}
thanks,