Re: [PATCH v2 2/2] wifi: rtlwifi: rtl8821ae: phy: fix an undefined bitwise shift behavior

[Ping-Ke Shih <pkshih@xxxxxxxxxxx>]

On Fri, 2023-11-24 at 18:06 +0800, Su Hui wrote:
> 
> On 2023/11/24 16:51, Ping-Ke Shih wrote:
> > Subject: [PATCH v2 2/2] wifi: rtlwifi: rtl8821ae: phy: fix an undefined bitwise shift behavior
> > 
> > [...]
> > > diff --git a/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/phy.c
> > > b/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/phy.c
> > > index 6df270e29e66..52ab1b0761c0 100644
> > > --- a/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/phy.c
> > > +++ b/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/phy.c
> > > @@ -31,7 +31,12 @@ static u32 _rtl8821ae_phy_calculate_bit_shift(u32 bitmask)
> > >   {
> > >          u32 i = ffs(bitmask);
> > > 
> > > -       return i ? i - 1 : 32;
> > > +       if (!i) {
> > > +               WARN_ON_ONCE(1);
> > > +               return 0;
> > > +       }
> > > +
> > > +       return i - 1;
> > >   }
> > Personally, I prefer to use __ffs(), because in normal case no need additional '-1',
> > and abnormal cases should not happen.
> 
> Hi,  Ping-Ke
> 
> Replace _rtl8821ae_phy_calculate_bit_shift() by __ffs(bitmask) is better,
> but I'm not sure what callers should do when callers check bitmask is 0 before calling.
> Maybe this check is useless?
> 
> I can send a v3 patch if using  __ffs(bitmask) and no check for bitmask is fine.
> Or could you send this patch if you have a better idea?
> Thanks for your suggestion!
> 

Can this work to you? 

static u32 _rtl8821ae_phy_calculate_bit_shift(u32 bitmask)
{
	if (WARN_ON_ONCE(!bitmask))
		return 0;

	return __ffs(bitmask);
}