Re: [PATCH bpf-next v2 2/4] bpf: bpf_csum_diff: optimize and homogenize for all archs

[Daniel Borkmann]

On 10/25/24 12:11 PM, Puranjay Mohan wrote:
[...]
> This file has a lot of such sparse warnings. Specifically, to fix the
> warning introduced by me, I can apply the following diff:
>
> --- >8 ---
>
> diff --git a/net/core/filter.c b/net/core/filter.c
> index e00bec7de9ed..b94037f29b2a 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -2019,16 +2019,18 @@ BPF_CALL_5(bpf_csum_diff, __be32 *, from, u32, from_size,
>           * Even for diffing, from_size and to_size don't need to be equal.
>           */
>
> +       __wsum ret = seed;
> +
>          if (from_size && to_size)
> -               return csum_from32to16(csum_sub(csum_partial(to, to_size, seed),
> -                                               csum_partial(from, from_size, 0)));
> +               ret = csum_sub(csum_partial(to, to_size, seed), csum_partial(from, from_size, 0));

Lets also pass ret into csum_partial() instead of seed given the arg
is of type __wsum there too, otherwise lgtm and yes lets fix these.

> +
>          if (to_size)
> -               return csum_from32to16(csum_partial(to, to_size, seed));
> +               ret =  csum_partial(to, to_size, seed);
>
>          if (from_size)
> -               return csum_from32to16(~csum_partial(from, from_size, ~seed));
> +               ret = ~csum_partial(from, from_size, ~seed);
>
> -       return seed;
> +       return csum_from32to16((__force unsigned int)ret);
>   }
>
> --- 8< ---
>
> If others feel that fixing these warnings is useful, I can send another
> version with above diff. I will then also send a separate patch to fix
> all other such warnings in this file.

That would be great, as separate patch, agree.

Thanks,
Daniel