Re: [PATCH] comedi: fix device lifecycle handling in comedi_dev_kref_release()

[Ian Abbott]

On 30/01/2026 22:35, Salah Triki wrote:
> `comedi_dev_kref_release()` calls `put_device()` on `dev->class_dev` and
> then immediately frees the `comedi_device` structure with `kfree()`.
> If the device reference count reaches zero, the device model release
> callback may be invoked, potentially accessing or freeing parts of
> the structure. Calling `kfree()` directly after `put_device()` can
> cause a use-after-free or double-free.
>
> After `device_initialize()` and once the struct device is in the device
> model, all cleanup must go through `put_device()` and the release
> callback. The kref release should not free the structure directly,
> ensuring that `comedi_device` lifetime is properly managed through
> device reference counting and preventing potential memory corruption.
>
> Fixes: 5b13ed94a7d24 ("staging: comedi: add a kref to comedi device")
>
> Signed-off-by: Salah Triki <salah.triki@xxxxxxxxx>
> ---
>   drivers/comedi/comedi_fops.c | 1 -
>   1 file changed, 1 deletion(-)
>
> diff --git a/drivers/comedi/comedi_fops.c b/drivers/comedi/comedi_fops.c
> index 2c3eb9e89571..24bb06fe6b45 100644
> --- a/drivers/comedi/comedi_fops.c
> +++ b/drivers/comedi/comedi_fops.c
> @@ -118,7 +118,6 @@ static void comedi_dev_kref_release(struct kref *kref)
>   
>   	mutex_destroy(&dev->mutex);
>   	put_device(dev->class_dev);
> -	kfree(dev);
>   }
>   
>   /**

NAK

This isn't freeing the `struct device`, it is freeing the `struct 
comedi_device`.  The `struct device` will be freed by 
`device_create_release()` in "drivers/base/core.c".

--
-=( Ian Abbott <abbotti@xxxxxxxxx> || MEV Ltd. is a company  )=-
-=( registered in England & Wales.  Regd. number: 02862268.  )=-
-=( Regd. addr.: S11 & 12 Building 67, Europa Business Park, )=-
-=( Bird Hall Lane, STOCKPORT, SK3 0XA, UK. || www.mev.co.uk )=-