Re: [PATCH v1] netlink: Remove implicit 32-bit trunction in nla_memcmp

[David Laight <david.laight.linux@gmail.com>]

On Tue,  9 Dec 2025 14:41:58 -0800
Ian Rogers <irogers@google.com> wrote:

> 64-bit truncation to 32-bit can result in the sign of the truncated
> value changing. The nla_memcmp function subtracts a 16-bit nla_len
> from a size_t size and so this shouldn't occur, but the code looks
> hazardous so change it to make the comparisons explicit and avoid a
> truncated subtract.
> 
> Signed-off-by: Ian Rogers <irogers@google.com>
> ---
>  lib/nlattr.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/lib/nlattr.c b/lib/nlattr.c
> index be9c576b6e2d..6e1c40dd11e3 100644
> --- a/lib/nlattr.c
> +++ b/lib/nlattr.c
> @@ -854,12 +854,12 @@ EXPORT_SYMBOL(nla_memcpy);
>  int nla_memcmp(const struct nlattr *nla, const void *data,
>  			     size_t size)
>  {
> -	int d = nla_len(nla) - size;
> +	u16 l = nla_len(nla);

Don't use u16 for locals.
Generates unnecessarily bad code.

	David

>  
> -	if (d == 0)
> -		d = memcmp(nla_data(nla), data, size);
> +	if (l == size)
> +		return memcmp(nla_data(nla), data, size);
>  
> -	return d;
> +	return l > size ? 1 : -1;
>  }
>  EXPORT_SYMBOL(nla_memcmp);
>