Re: [PATCH] net/sched: avoid unused-label warning

[Arnd Bergmann]

On Wed, Mar 6, 2019 at 1:28 AM Cong Wang <xiyou.wangc...@gmail.com> wrote:
> On Mon, Mar 4, 2019 at 12:40 PM Arnd Bergmann <a...@arndb.de> wrote:
> > diff --git a/net/sched/act_tunnel_key.c b/net/sched/act_tunnel_key.c
> > index 2a5f215ae876..3beb4717d3b7 100644
> > --- a/net/sched/act_tunnel_key.c
> > +++ b/net/sched/act_tunnel_key.c
> > @@ -392,8 +392,8 @@ static int tunnel_key_init(struct net *net, struct 
> > nlattr *nla,
> >  #ifdef CONFIG_DST_CACHE
> >         if (metadata)
> >                 dst_cache_destroy(&metadata->u.tun_info.dst_cache);
> > -#endif
> >  release_tun_meta:
> > +#endif
>
> These #ifdef's are ugly, either we should select DST_CACHE
> or provide a nop for these dst_cache_*() APIs when it is not
> enabled.

I agree that would be nicer, or alternatively convert the preprocessor
conditionals to C conditionals like

diff --git a/net/sched/act_tunnel_key.c b/net/sched/act_tunnel_key.c
index 3beb4717d3b7..586343a5accc 100644
--- a/net/sched/act_tunnel_key.c
+++ b/net/sched/act_tunnel_key.c
@@ -327,11 +327,11 @@ static int tunnel_key_init(struct net *net,
struct nlattr *nla,
                        goto err_out;
                }

-#ifdef CONFIG_DST_CACHE
-               ret = dst_cache_init(&metadata->u.tun_info.dst_cache,
GFP_KERNEL);
-               if (ret)
-                       goto release_tun_meta;
-#endif
+               if (IS_ENABLED(CONFIG_DST_CACHE)) {
+                       ret =
dst_cache_init(&metadata->u.tun_info.dst_cache, GFP_KERNEL);
+                       if (ret)
+                               goto release_tun_meta;
+               }

                if (opts_len) {
                        ret = tunnel_key_opts_set(tb[TCA_TUNNEL_KEY_ENC_OPTS],
@@ -389,11 +389,9 @@ static int tunnel_key_init(struct net *net,
struct nlattr *nla,
        return ret;

 release_dst_cache:
-#ifdef CONFIG_DST_CACHE
        if (metadata)
                dst_cache_destroy(&metadata->u.tun_info.dst_cache);
 release_tun_meta:
-#endif
        if (metadata)
                dst_release(&metadata->dst);

Usually, you'd want to do that consistently though, and change all the
related checks at the same time, so I would keep that separate from
the trivial bugfix.

     Arnd