Re: [PATCHv2 net] ipv6: sr: fix invalid unregister error path

[Hangbin Liu <liuhangbin@gmail.com>]

On Wed, May 08, 2024 at 11:33:18AM +0200, Sabrina Dubroca wrote:
> > ---
> > v2: define label out_unregister_genl in CONFIG_IPV6_SEG6_LWTUNNEL(Sabrina Dubroca)
> > ---
> >  net/ipv6/seg6.c | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> > 
> > diff --git a/net/ipv6/seg6.c b/net/ipv6/seg6.c
> > index 35508abd76f4..6a80d93399ce 100644
> > --- a/net/ipv6/seg6.c
> > +++ b/net/ipv6/seg6.c
> > @@ -551,8 +551,8 @@ int __init seg6_init(void)
> >  #endif
> >  #ifdef CONFIG_IPV6_SEG6_LWTUNNEL
> >  out_unregister_genl:
> > -	genl_unregister_family(&seg6_genl_family);
> >  #endif
> > +	genl_unregister_family(&seg6_genl_family);
> 
> Sorry, I didn't notice when you answered my comment yesterday, but
> this will create unreachable code after return when
> CONFIG_IPV6_SEG6_LWTUNNEL=n and CONFIG_IPV6_SEG6_HMAC=n:

Oh.. Didn't notice this...

> 
> out:
> 	return err;
> 	genl_unregister_family(&seg6_genl_family);
> out_unregister_pernet:
> 	unregister_pernet_subsys(&ip6_segments_ops);
> 	goto out;
> 
> 
> (stragely, gcc doesn't complain about it, I thought it would)

Yes, I also complied the patch with not complain, so I just posted it.

> 
> 
> The only solution I can think of if we want to avoid it is ugly:
> 
>  #ifdef CONFIG_IPV6_SEG6_LWTUNNEL
>  out_unregister_genl:
>  #endif
> +#if IS_ENABLED(CONFIG_IPV6_SEG6_LWTUNNEL) || IS_ENABLED(CONFIG_IPV6_SEG6_HMAC)
>  	genl_unregister_family(&seg6_genl_family);
> +#endif
>  out_unregister_pernet:
>  	unregister_pernet_subsys(&ip6_segments_ops);
>  	goto out;
> 
> (on top of v2)
> 
> For all other cases your patch looks correct.

Thanks, I will check if there are any other workaround. If not, I will do
like what you said.

Hangbin