Re: [syzbot] [input?] possible deadlock in evdev_cleanup (2)

[Hillf Danton]

On Sat, 18 May 2024 07:18:36 -0700
> syzbot found the following issue on:
> 
> HEAD commit:    fda5695d692c Merge branch 'for-next/core' into for-kernelci
> git tree:       git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-kernelci
> C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=134d3182980000

#syz test https://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git  for-kernelci

--- l/drivers/input/misc/uinput.c
+++ u/drivers/input/misc/uinput.c
@@ -311,6 +311,7 @@ static int uinput_create_device(struct u
 	struct input_dev *dev = udev->dev;
 	int error, nslot;
 
+	lockdep_assert_held(&udev->mutex);
 	if (udev->state != UIST_SETUP_COMPLETE) {
 		printk(KERN_DEBUG "%s: write device info first\n", UINPUT_NAME);
 		return -EINVAL;
@@ -362,7 +363,9 @@ static int uinput_create_device(struct u
 
 	input_set_drvdata(udev->dev, udev);
 
-	error = input_register_device(udev->dev);
+	mutex_unlock(&udev->mutex);
+	error = input_register_device(dev);
+	mutex_lock(&udev->mutex);
 	if (error)
 		goto fail2;
 
--