Re: [PATCH rtw-next] wifi: rtw89: usb: Avoid crash with dynamically added device ID

[Bitterblue Smith <rtl8821cerfe2@gmail.com>]

On 17/08/2026 06:38, Ping-Ke Shih wrote:
> Bitterblue Smith <rtl8821cerfe2@gmail.com> wrote:
>> Adding a device ID via sysfs causes a crash when the device is plugged
>> in, because the driver_info pointer is null.
> 
> I don't know about this before. Does it looks like?
> 
>    echo "1234 5678" | sudo tee /sys/bus/usb/drivers/my_driver/new_id
> 

Yes, like that. A few people have tried to use it recently to test new
devices and got crashes.

>>
>> Add a wrapper around rtw89_usb_probe() in each driver to check if
>> driver_info is null and provide a reasonable default value for it.
>>
>> Signed-off-by: Bitterblue Smith <rtl8821cerfe2@gmail.com>
> 
> Acked-by: Ping-Ke Shih <pkshih@realtek.com>
> 
> I'd give my acked-by in advance. But you still can consider my opinion below.
> 
>> ---
>> The PCI side has the same problem, but new device IDs are a lot less
>> likely there.
> 
> Thanks for the info.
> 
> [...]
> 
>> +static int rtw8851bu_probe(struct usb_interface *intf,
>> +                          const struct usb_device_id *id)
>> +{
>> +       const struct rtw89_driver_info *info;
>> +
>> +       if (id->driver_info)
>> +               info = (const struct rtw89_driver_info *)id->driver_info;
>> +       else
>> +               info = &rtw89_8851bu_info;
>> +
>> +       return rtw89_usb_probe(intf, info);
>> +}
>> +
> 
> If you want to simply specific probe(), here might be 
> 
> static int rtw8851bu_probe(struct usb_interface *intf,
>                            const struct usb_device_id *id)
> {
>         return rtw89_usb_probe(intf, id, &rtw89_8851bu_info);
> }
> 
> Then
> 
> int rtw89_usb_probe(struct usb_interface *intf,
>                     const struct usb_device_id *id,
>                     const struct rtw89_driver_info *default_info)
> {
>         ...
> 
>         info = (const struct rtw89_driver_info *)id->driver_info;
>         if (!info)
>                 info = default_info;
> }
> 
> 

Ah, yes, that is better.