On Fri, 3 Jan 2025 11:02:30 +0100
Wolfram Sang <wsa+renesas@xxxxxxxxxxxxxxxxxxxx> wrote:
> > > > > - (dynaddr_parity(address) ? DAT_0_DYNADDR_PARITY : 0);
> > > > > + (parity8(address) ? 0 : DAT_0_DYNADDR_PARITY);
>
> ...
>
> > The old code is:
> > > -static inline bool dynaddr_parity(unsigned int addr)
> > > -{
> > > - addr |= 1 << 7;
> > > - addr += addr >> 4;
> > > - addr += addr >> 2;
> > > - addr += addr >> 1;
> > > - return (addr & 1);
> > > -}
> >
> > So:
> > 1) it always sets 0x80.
>
> Right, this is why the arguments of the ternary operator above are
> exchanged. The old function was basically 'is_odd'.
Provided the high bit isn't already set - which it may not be.
> > 2) it uses addition not exclusive or.
>
> True, but it will work nonetheless because we are only interested in bit
> 0 of the result. For one bit, XOR and addition are interchangable. The
> overflow to other bits is not important.
add: 00010001 => xxxx0010 => xx10 => x1
xor: 00010001 => xxxx0000 => 00xx => x0
>
> > So just not the same definition of 'parity'.
>
> I think it is. I mean, I3C wants odd parity, otherwise it will not work.
> And Jarkko kindly confirmed it still works.
I bet the target isn't checking...
So you might be fixing a bug.
David
