netdev - Re: [PATCH net-next] net: llc: remove init

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1512072766.19682.27.camel@gmail.com>
Date:   Thu, 30 Nov 2017 12:12:46 -0800
From:   Eric Dumazet <eric.dumazet@...il.com>
To:     Alexander Aring <aring@...atatu.com>, davem@...emloft.net
Cc:     hadi@...atatu.com, netdev@...r.kernel.org
Subject: Re: [PATCH net-next] net: llc: remove init_net check

On Thu, 2017-11-30 at 14:45 -0500, Alexander Aring wrote:
> Commit e730c15519d09 ("[NET]: Make packet reception network namespace
> safe") added several init_net compares to protocols with unknown
> net namespace support. The idea was as users started using the
> different
> protocols they would test and enable them.
> In our use-case we generate STP enabled Linux bridges in different
> namespaces. The bridges are connected via veth interfaces.
> This patch makes STP work in such a setup.
> 
> Signed-off-by: Alexander Aring <aring@...atatu.com>
> ---
>  net/llc/llc_input.c | 4 ----
>  1 file changed, 4 deletions(-)
> 
> diff --git a/net/llc/llc_input.c b/net/llc/llc_input.c
> index 82cb93f66b9b..09b530d90610 100644
> --- a/net/llc/llc_input.c
> +++ b/net/llc/llc_input.c
> @@ -14,7 +14,6 @@
>  #include <linux/netdevice.h>
>  #include <linux/slab.h>
>  #include <linux/export.h>
> -#include <net/net_namespace.h>
>  #include <net/llc.h>
>  #include <net/llc_pdu.h>
>  #include <net/llc_sap.h>
> @@ -162,9 +161,6 @@ int llc_rcv(struct sk_buff *skb, struct
> net_device *dev,
>  	void (*sta_handler)(struct sk_buff *skb);
>  	void (*sap_handler)(struct llc_sap *sap, struct sk_buff
> *skb);
>  
> -	if (!net_eq(dev_net(dev), &init_net))
> -		goto drop;
> -
>  	/*
>  	 * When the interface is in promisc. mode, drop all the crap
> that it
>  	 * receives, do not try to analyse it.


Well, we use different netns for isolation.

You need more changes than simply removing this check, I guess.

__llc_sap_find() would need a per netns list, or proper netns checks.