Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 10 Feb 2022 14:14:27 +0100
From:   Andrew Lunn <andrew@lunn.ch>
To:     Horatiu Vultur <horatiu.vultur@microchip.com>
Cc:     Jakub Kicinski <kuba@kernel.org>, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, UNGLinuxDriver@microchip.com,
        davem@davemloft.net, kernel test robot <lkp@intel.com>
Subject: Re: [PATCH net-next] net: lan966x: Fix when CONFIG_IPV6 is not set
Message-ID: <YgUPswHG60Fi9rjA@lunn.ch>
References: <20220209101823.1270489-1-horatiu.vultur@microchip.com>
 <YgPHjxpo0N4ND1ch@lunn.ch>
 <20220209180620.3699bf25@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
 <20220210083612.4mszzwgcrvmn67rn@soft-dev3-1.localhost>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220210083612.4mszzwgcrvmn67rn@soft-dev3-1.localhost>
Precedence: bulk

> What do you think if I do something like this in the lan966x_main.h
> 
> ---
> #if IS_ENABLED(CONFIG_IPV6)
> static inline bool lan966x_hw_offload_ipv6(struct sk_buff *skb)
> {
> 	if (skb->protocol == htons(ETH_P_IPV6) &&
> 	    ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr) &&
> 	    !ipv6_mc_check_mld(skb))
> 		return false;
> 
> 	return true;
> }
> #else
> static inline bool lan966x_hw_offload_ipv6(struct sk_buff *skb)
> {
> 	return false;
> }
> #endif
> ---

The reason we prefer not to use #if is that it reduced compile testing
coverage. The block of code inside gets compiled a lot less.

> And then in lan966x_main.c just call this function.
> 
> > 
> > If it's linking we can do:
> > 
> >         if (IS_ENABLED(CONFIG_IPV6) &&
> >             skb->protocol == htons(ETH_P_IPV6) &&
> >             ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr) &&
> >             !ipv6_mc_check_mld(skb))
> >                 return false;

Jakub solution results in the code always being compiled, but the
IS_ENABLED(CONFIG_IPV6) gets turned into a constant 0 or 1. The
optimizer can then remove the whole block of code in the 0 case.

> I was also looking at other drivers on how they use 'ipv6_mc_check_mld'.
> Then I have seen that drivers/net/amt.c and net/bridge/br_multicast.c
> they wrap this function with #if.
> But then there is net/batman-adv/multicast.c which doesn't do that and
> it can compile and link without CONFIG_IPV6 and I just don't see how
> that is working.

Maybe it is to do with this at the end of net/ip6/Makefile

ifneq ($(CONFIG_IPV6),)
obj-$(CONFIG_NET_UDP_TUNNEL) += ip6_udp_tunnel.o
obj-y += mcast_snoop.o
endif