Re: [patch net-next 1/2] rtnetlink: allow to register ops without ops->setup set

[ebiederm@xmission.com (Eric W. Biederman)]

Jiri Pirko <jiri@resnulli.us> writes:

> So far, it is assumed that ops->setup is filled up. But there might be
> case that ops might make sense even without ->setup. In that case,
> forbid to newlink and dellink.

This patch does not handle all cases of dellink, and so you have created
a way to oops the kernel.  One case that is not handled is
default_device_exit_batch.

Further using !setup to prevent dellink does not make any kind of
logical sense.  Which leads to a maintenance hazard.  Aka how can we
further update this code without running into weird corner cases.

Further you have not given a description why anyone would want to have
something that is almost completely unlike rtnl link operations
implemented with the rtnl link operations.


> Signed-off-by: Jiri Pirko <jiri@resnulli.us>
> ---
>  net/core/rtnetlink.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
> index 1063996..84affd7 100644
> --- a/net/core/rtnetlink.c
> +++ b/net/core/rtnetlink.c
> @@ -1777,7 +1777,7 @@ static int rtnl_dellink(struct sk_buff *skb, struct nlmsghdr *nlh)
>  		return -ENODEV;
>  
>  	ops = dev->rtnl_link_ops;
> -	if (!ops)
> +	if (!ops || !ops->setup)
>  		return -EOPNOTSUPP;
>  
>  	ops->dellink(dev, &list_kill);
> @@ -2038,6 +2038,9 @@ replay:
>  			return -EOPNOTSUPP;
>  		}
>  
> +		if (!ops->setup)
> +			return -EOPNOTSUPP;
> +
>  		if (!ifname[0])
>  			snprintf(ifname, IFNAMSIZ, "%s%%d", ops->kind);