Re: [PATCH] staging: rtl8192e: Use min_t/max_t macros for variable comparison

[Deepak R Varma <drv@mailo.com>]

On Thu, Nov 03, 2022 at 11:53:45AM +0300, Dan Carpenter wrote:
> On Thu, Nov 03, 2022 at 08:24:15AM +0000, David Laight wrote:
> > > --- a/drivers/staging/rtl8192e/rtl819x_HTProc.c
> > > +++ b/drivers/staging/rtl8192e/rtl819x_HTProc.c
> > > @@ -587,17 +587,12 @@ void HTOnAssocRsp(struct rtllib_device *ieee)
> > >  			else
> > >  				pHTInfo->CurrentAMPDUFactor = HT_AGG_SIZE_64K;
> > >  		} else {
> > > -			if (pPeerHTCap->MaxRxAMPDUFactor < HT_AGG_SIZE_32K)
> > > -				pHTInfo->CurrentAMPDUFactor =
> > > -						 pPeerHTCap->MaxRxAMPDUFactor;
> > > -			else
> > > -				pHTInfo->CurrentAMPDUFactor = HT_AGG_SIZE_32K;
> > > +			pHTInfo->CurrentAMPDUFactor = min_t(u32, pPeerHTCap->MaxRxAMPDUFactor,
> > > +							    HT_AGG_SIZE_32K);
> >
> > For min() to fail there must be a signed v unsigned mismatch.
> > Maybe that ought to be fixed.
> >
>
> u32 is the right choice here.
>
> I'm having a hard time understanding your email.  You might be saying
> we could declare HT_AGG_SIZE_32K as a u32 so then we could use min()
> instead of min_t()?  HT_AGG_SIZE_32K is an enum.
>
> pPeerHTCap->MaxRxAMPDUFactor is a bitfield.
>
> 	u8 MaxRxAMPDUFactor:2;
>
> We will never be able to use min().

I think we could do min((u32)a, (u32)b), but it is just unwrapped min_t
if I understand David's comment.

>
> > >  		}
> > >  	}
> > > -	if (pHTInfo->MPDU_Density > pPeerHTCap->MPDUDensity)
> > > -		pHTInfo->current_mpdu_density = pHTInfo->MPDU_Density;
> > > -	else
> > > -		pHTInfo->current_mpdu_density = pPeerHTCap->MPDUDensity;
> > > +	pHTInfo->current_mpdu_density = max_t(u8, pHTInfo->MPDU_Density,
> > > +					      pPeerHTCap->MPDUDensity);
> >
> > Using u8 with max_t() really doesn't make any sense.
>
> Using u8 looks wrong because you would worry that one of the types is
> larger than U8_MAX.  But it's actually fine.  The types are u8 vs another
> bitfield.  I would probably have gone with u32 here as well.
I will take your advise and upgrade the type to u32 as a revision.
>
> > The value will get promoted to signed int prior to the comparison.
> >
>
> That's sort of true-ish but I don't understand what you are saying?
> #confused

Yes, I too did not understand David's comment. I tried to dig dipper into max_t but
it gets very complex. Can you please elaborate how you determined the promotion
to signed int?

>
> regards,
> dan carpenter
>
>