Re: [PATCH net-next] net: dsa: yt921x: Fix missing type casting to u64

[David Laight]

On Fri, 24 Oct 2025 16:49:13 +0800
David Yang <mmyangfl@xxxxxxxxx> wrote:

> Reported by the following Smatch static checker warning:
> 
>   drivers/net/dsa/yt921x.c:702 yt921x_read_mib()
>   warn: was expecting a 64 bit value instead of '(~0)'
> 
> Reported-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> Closes: https://lore.kernel.org/netdev/aPsjYKQMzpY0nSXm@stanley.mountain/
> Signed-off-by: David Yang <mmyangfl@xxxxxxxxx>
> ---
>  drivers/net/dsa/yt921x.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/dsa/yt921x.c b/drivers/net/dsa/yt921x.c
> index ab762ffc4661..8baed8107512 100644
> --- a/drivers/net/dsa/yt921x.c
> +++ b/drivers/net/dsa/yt921x.c
> @@ -699,7 +699,7 @@ static int yt921x_read_mib(struct yt921x_priv *priv, int port)
>  			if (val < (u32)val)

That check is wrong as well, probably (val0 < (u32)val) is right.
But the code is confusing.

>  				/* overflow */
>  				val += (u64)U32_MAX + 1;
> -			val &= ~U32_MAX;
> +			val &= ~(u64)U32_MAX;
>  			val |= val0;

How about:
		if (desc->size <= 1) {
			u64 old_val = *valp;
			val = upper32_bits(old_val) | val0;
			if (val < old_val)
				val += 1ull << 32;
		}

There is also an inconsistency with the read of *valp and the
WRITE_ONCE() lower down.
If there is a READ_ONCE() elsewhere then it not going to work on
32bit architectures - since both the read and write are still
likely to be two memory cycles.

	David

>  		} else {
>  			res = yt921x_reg_read(priv, reg + 4, &val1);