net-next tree broken with CONFIG_NETFILTER_INGRESS=n

net-next tree broken with CONFIG_NETFILTER_INGRESS=n

[Marcel Holtmann]

Hi Dave,

the net-next tree is broken since a few days now when CONFIG_NETFILTER_INGRESS=n is set.

  CC      net/netfilter/core.o
In file included from ./include/linux/linkage.h:4:0,
                 from ./include/linux/kernel.h:6,
                 from net/netfilter/core.c:10:
net/netfilter/core.c: In function ‘nf_set_hooks_head’:
net/netfilter/core.c:96:30: error: ‘struct net_device’ has no member named ‘nf_hooks_ingress’
   rcu_assign_pointer(reg->dev->nf_hooks_ingress, entry);
                              ^
./include/linux/compiler.h:299:17: note: in definition of macro ‘WRITE_ONCE’
  union { typeof(x) __val; char __c[1]; } __u = \
                 ^
Regards

Marcel

Re: net-next tree broken with CONFIG_NETFILTER_INGRESS=n

[Marcel Holtmann]

Hi Dave,

> the net-next tree is broken since a few days now when CONFIG_NETFILTER_INGRESS=n is set.
> 
>  CC      net/netfilter/core.o
> In file included from ./include/linux/linkage.h:4:0,
>                 from ./include/linux/kernel.h:6,
>                 from net/netfilter/core.c:10:
> net/netfilter/core.c: In function ‘nf_set_hooks_head’:
> net/netfilter/core.c:96:30: error: ‘struct net_device’ has no member named ‘nf_hooks_ingress’
>   rcu_assign_pointer(reg->dev->nf_hooks_ingress, entry);
>                              ^
> ./include/linux/compiler.h:299:17: note: in definition of macro ‘WRITE_ONCE’
>  union { typeof(x) __val; char __c[1]; } __u = \

I really wonder what is going on here. Normally if you break the build, you fix it. However it seems nobody really cares enough. If I am not mistaken, then this patch broke it:

commit e3b37f11e6e4e6b6f02cc762f182ce233d2c1c9d
Author: Aaron Conole <aconole@bytheb.org>
Date:   Wed Sep 21 11:35:07 2016 -0400

    netfilter: replace list_head with single linked list
    
    The netfilter hook list never uses the prev pointer, and so can be trimmed to
    be a simple singly-linked list.

It is funny that it would have been obvious to test CONFIG_NETFILTER_INGRESS=n since there is an ifdef right in the first hunk of the patch.

        struct netdev_queue __rcu *ingress_queue;
 #ifdef CONFIG_NETFILTER_INGRESS
-       struct list_head        nf_hooks_ingress;
+       struct nf_hook_entry __rcu *nf_hooks_ingress;
 #endif

Regards

Marcel

Re: net-next tree broken with CONFIG_NETFILTER_INGRESS=n

[David Miller]

From: Marcel Holtmann <marcel@holtmann.org>
Date: Wed, 5 Oct 2016 09:48:42 +0200

> I really wonder what is going on here. Normally if you break the
> build, you fix it. However it seems nobody really cares enough. If I
> am not mistaken, then this patch broke it:

The people who can work on and merge the fix are travelling and
conferencing in Tokyo right now, that's why this wasn't fixed
instantaneously.

Re: net-next tree broken with CONFIG_NETFILTER_INGRESS=n

[Florian Westphal]

David Miller <davem@davemloft.net> wrote:
> From: Marcel Holtmann <marcel@holtmann.org>
> Date: Wed, 5 Oct 2016 09:48:42 +0200
> 
> > I really wonder what is going on here. Normally if you break the
> > build, you fix it. However it seems nobody really cares enough. If I
> > am not mistaken, then this patch broke it:
> 
> The people who can work on and merge the fix are travelling and
> conferencing in Tokyo right now, that's why this wasn't fixed
> instantaneously.

Seems the fix is rotting in nf-next....