Re: [RFC net-next PATCH] ipv6: New define for reoccurring code

[Eric Dumazet <eric.dumazet@gmail.com>]

On 1/17/20 1:56 PM, Jeff Kirsher wrote:
> Through out the kernel, sizeof() is used to determine the size of the IPv6
> address structure, so create a define for the commonly used code.
> 
> s/sizeof(struct in6_addr)/ipv6_addr_size/g
> 
> This is just a portion of the instances in the kernel and before cleaning
> up all the occurrences, wanted to make sure that this was a desired change
> or if this obfuscates the code.
> 
> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
> ---
...

>  };
> +#define ipv6_addr_size		sizeof(struct in6_addr)
>  #endif /* __UAPI_DEF_IN6_ADDR */
>  
>  #if __UAPI_DEF_SOCKADDR_IN6
> diff --git a/net/core/filter.c b/net/core/filter.c
> index ef01c5599501..eabf42893b60 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -5053,7 +5053,7 @@ BPF_CALL_4(bpf_lwt_seg6_action, struct sk_buff *, skb,
>  	case SEG6_LOCAL_ACTION_END_X:
>  		if (!seg6_bpf_has_valid_srh(skb))
>  			return -EBADMSG;
> -		if (param_len != sizeof(struct in6_addr))
> +		if (param_len != ipv6_addr_size)

Hmm...

I vote seeing sizeof(struct in6_addr) rather than dealing
with yet another thing to remember and additional backports conflicts.

And I prefer not seeing dozens of followup trivial patches because
people will forget about this new pseudo variable.