netdev - Re: [RFC PATCH 1/1] IPVS netns shutdown/startup dead-lock

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 for Android: free password hash cracker in your pocket

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

Date:	Mon, 13 Jun 2011 20:11:29 +0900
From:	Simon Horman <horms@...ge.net.au>
To:	Hans Schillstrom <hans.schillstrom@...csson.com>
Cc:	ja@....bg, wensong@...ux-vs.org, lvs-devel@...r.kernel.org,
	netdev@...r.kernel.org, netfilter-devel@...r.kernel.org,
	hans@...illstrom.com
Subject: Re: [RFC PATCH 1/1] IPVS netns shutdown/startup dead-lock

On Mon, Jun 13, 2011 at 11:32:10AM +0200, Hans Schillstrom wrote:
> ip_vs_mutext is used by both netns shutdown code and startup
> and both implicit uses sk_lock-AF_INET mutex.
> 
> cleanup CPU-1         startup CPU-2
> ip_vs_dst_event()     ip_vs_genl_set_cmd()
>  sk_lock-AF_INET     __ip_vs_mutex
>                      sk_lock-AF_INET
> __ip_vs_mutex
> * DEAD LOCK *
> 
> This can be solved by have the ip_vs_mutex per netns
> or avid locking when starting/stoping sync-threads.
> i.e. just add a starting/stoping flag.
> 
> ip_vs_mutex per name-space seems to be a more future proof solution.
> 
> Which one should be used ?

I don't feel strongly either way.

> Signed-off-by: Hans Schillstrom <hans.schillstrom@...csson.com>
> ---
>  include/net/ip_vs.h             |    2 ++
>  net/netfilter/ipvs/ip_vs_ctl.c  |   15 ++++++++++-----
>  net/netfilter/ipvs/ip_vs_sync.c |   30 +++++++++++++++++++++++++-----
>  3 files changed, 37 insertions(+), 10 deletions(-)
> 

[snip]

> diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
> index 699c79a..21c541f 100644
> --- a/net/netfilter/ipvs/ip_vs_ctl.c
> +++ b/net/netfilter/ipvs/ip_vs_ctl.c

[snip]

> @@ -3305,12 +3309,13 @@ static int ip_vs_genl_set_cmd(struct sk_buff *skb, struct genl_info *info)
>  			ret = -EINVAL;
>  			goto out;
>  		}
> -
> +		/* Unlock since a global socket lock will be taken later */
> +		mutex_unlock(&__ip_vs_mutex);
>  		if (cmd == IPVS_CMD_NEW_DAEMON)
>  			ret = ip_vs_genl_new_daemon(net, daemon_attrs);
>  		else
>  			ret = ip_vs_genl_del_daemon(net, daemon_attrs);
> -		goto out;
> +		goto out_nounlock;

I'm not a huge fan of labels that only return.
So I think it would be slightly easier on the eyes
to just return here, not add out_nounlock,
and possibly rename out as out_unlock.

>  	} else if (cmd == IPVS_CMD_ZERO &&
>  		   !info->attrs[IPVS_CMD_ATTR_SERVICE]) {
>  		ret = ip_vs_zero_all(net);

[snip]
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html