A lockdep circular locking dependency warning can be triggered
reproducibly when using a force-feedback gamepad with uinput (for
example, playing ELDEN RING under Wine with a Flydigi Vader 5
controller):
ff->mutex -> udev->mutex -> input_mutex -> dev->mutex -> ff->mutex
The cycle is caused by four lock acquisition paths:
1. ff upload: input_ff_upload() holds ff->mutex and calls
uinput_dev_upload_effect() -> uinput_request_submit() ->
uinput_request_send(), which acquires udev->mutex.
2. device create: uinput_ioctl_handler() holds udev->mutex and calls
uinput_create_device() -> input_register_device(), which acquires
input_mutex.
3. device register: input_register_device() holds input_mutex and
calls kbd_connect() -> input_register_handle(), which acquires
dev->mutex.
4. evdev release: evdev_release() calls input_flush_device() under
dev->mutex, which calls input_ff_flush() acquiring ff->mutex.
Fix this by introducing a new state_lock spinlock to protect
udev->state and udev->dev access in uinput_request_send() instead of
acquiring udev->mutex. The function only needs to atomically check
device state and queue an input event into the ring buffer via
uinput_dev_event() -- both operations are safe under a spinlock
(ktime_get_ts64() and wake_up_interruptible() do not sleep). This
breaks the ff->mutex -> udev->mutex link since a spinlock is a leaf in
the lock ordering and cannot form cycles with mutexes.
To keep state transitions visible to uinput_request_send(), protect
writes to udev->state in uinput_create_device() and
uinput_destroy_device() with the same state_lock spinlock.
Additionally, move init_completion(&request->done) from
uinput_request_send() to uinput_request_submit() before
uinput_request_reserve_slot(). Once the slot is allocated,
uinput_flush_requests() may call complete() on it at any time from
the destroy path, so the completion must be initialised before the
request becomes visible.
Lock ordering after the fix:
ff->mutex -> state_lock (spinlock, leaf)
udev->mutex -> state_lock (spinlock, leaf)
udev->mutex -> input_mutex -> dev->mutex -> ff->mutex (no back-edge)
Fixes: ff462551235d ("Input: uinput - switch to the new FF interface")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/all/CABXGCsMoxag+kEwHhb7KqhuyxfmGGd0P=tHZyb1uKE0pLr8Hkg@mail.gmail.com/
Signed-off-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Link: https://patch.msgid.link/20260407075031.38351-1-mikhail.v.gavrilov@gmail.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
struct input_dev *dev;
struct mutex mutex;
enum uinput_state state;
+ spinlock_t state_lock;
wait_queue_head_t waitq;
unsigned char ready;
unsigned char head;
static int uinput_request_send(struct uinput_device *udev,
struct uinput_request *request)
{
- int retval;
+ int retval = 0;
- retval = mutex_lock_interruptible(&udev->mutex);
- if (retval)
- return retval;
+ spin_lock(&udev->state_lock);
if (udev->state != UIST_CREATED) {
retval = -ENODEV;
goto out;
}
- init_completion(&request->done);
-
/*
* Tell our userspace application about this new request
* by queueing an input event.
uinput_dev_event(udev->dev, EV_UINPUT, request->code, request->id);
out:
- mutex_unlock(&udev->mutex);
+ spin_unlock(&udev->state_lock);
return retval;
}
{
int retval;
+ /*
+ * Initialize completion before allocating the request slot.
+ * Once the slot is allocated, uinput_flush_requests() may
+ * complete it at any time, so it must be initialized first.
+ */
+ init_completion(&request->done);
+
retval = uinput_request_reserve_slot(udev, request);
if (retval)
return retval;
struct input_dev *dev = udev->dev;
enum uinput_state old_state = udev->state;
+ /*
+ * Update state under state_lock so that concurrent
+ * uinput_request_send() sees the state change before we
+ * flush pending requests and tear down the device.
+ */
+ spin_lock(&udev->state_lock);
udev->state = UIST_NEW_DEVICE;
+ spin_unlock(&udev->state_lock);
if (dev) {
name = dev->name;
if (error)
goto fail2;
+ spin_lock(&udev->state_lock);
udev->state = UIST_CREATED;
+ spin_unlock(&udev->state_lock);
return 0;
return -ENOMEM;
mutex_init(&newdev->mutex);
+ spin_lock_init(&newdev->state_lock);
spin_lock_init(&newdev->requests_lock);
init_waitqueue_head(&newdev->requests_waitq);
init_waitqueue_head(&newdev->waitq);