RE: [PATCH v2 rtw-next 1/2] wifi: rtw89: add dev_id_quirks to driver_info for per-device quirk control

[Johnson Tsai <wenjie.tsai@realtek.com>]

On Tuesday, May 26, 2026 9:10, Ping-Ke Shih wrote:
> Greg KH <gregkh@linuxfoundation.org> wrote:
> > On Mon, May 25, 2026 at 04:51:47PM +0800, Ping-Ke Shih wrote:
> > > --- a/drivers/net/wireless/realtek/rtw89/rtw8851be.c
> > > +++ b/drivers/net/wireless/realtek/rtw89/rtw8851be.c
> > > @@ -73,6 +73,7 @@ static const struct rtw89_driver_info
> rtw89_8851be_info = {
> > >       .chip = &rtw8851b_chip_info,
> > >       .variant = NULL,
> > >       .quirks = NULL,
> > > +     .dev_id_quirks = 0,
> >
> > Why are you manually adding a "= 0" for these when that's the default?
> > No need to do so, only do it when it is set to something, right?
> 
> I'd ask authors spending a little time to fill these default values because
> Realtek WiFi chips have many specific attributes for many chips, and
> somehow it gets messed. For example,
>  - miss to fill value for a certain chip
>    (people might only force on one chip he wants to fix)
>  - for some cases, it should set 0 explicitly as the value
>    (such as a threshold value of hardware register)
>  - messed up the order of attributes
>    People add a field along the existing field across all chips.
>    If a field is omitted, somehow people might add the field at wrong place.
>  - simply coding rule
>    People might forget why, so I'd have a simple rule for them.
> 
> I encountered these cases when I processed earlier Realtek WiFi drivers
> (rtlwifi, rtl8xxxu, rtw88), so I'd explicitly fill values to avoid them.
> 

Just a gentle follow-up on Ping-Ke's explanation regarding the explicit "= 0" initialization. 

If there are no further concerns on this, I will keep the current approach for this part and send out the v3 patch.


Thanks,

Johnson