Re: [PATCH v2] HID: betop: fix slab-out-of-bounds Write in betop_probe

[Pavel Skripkin <paskripkin@gmail.com>]

On 8/22/21 4:43 PM, F.A.Sulaiman wrote:
> Syzbot reported slab-out-of-bounds Write bug in hid-betopff driver.
> The problem is the driver assumes the device must have an input report but
> some malicious devices violate this assumption.
> 
> So this patch checks hid_device's input is non empty before it's been used.
> 
> Reported-by: syzbot+07efed3bc5a1407bd742@syzkaller.appspotmail.com
> Signed-off-by: F.A. SULAIMAN <asha.16@itfac.mrt.ac.lk>
> ---
>   drivers/hid/hid-betopff.c | 5 +++++
>   1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/hid/hid-betopff.c b/drivers/hid/hid-betopff.c
> index 0790fbd3fc9a..2d62bde21413 100644
> --- a/drivers/hid/hid-betopff.c
> +++ b/drivers/hid/hid-betopff.c
> @@ -116,6 +116,11 @@ static int betop_probe(struct hid_device *hdev, const struct hid_device_id *id)
>   {
>   	int ret;
>   
> +	if (list_empty(&hdev->inputs)) {
> +		hid_err(hdev, "no inputs found\n");
> +		return -ENODEV;
> +	}
> +
>   	if (id->driver_data)
>   		hdev->quirks |= HID_QUIRK_MULTI_INPUT;
>   
> 

I am still able to trigger reported slab-out-bound with this patch 
applied, please move this sanity check inside betopff_init().


Jiri, does it make sense to add proper error handling of betopff_init()? 
Nowadays betop_probe() just ignores betopff_init() return value. It 
looks wrong to me.


I think, Asha can prepare a patch series with these 2 changes



With regards,
Pavel Skripkin