Message ID | 1245263058.31588.38.camel@johannes.local |
---|---|
State | RFC, archived |
Delegated to: | David Miller |
Headers | show |
Johannes Berg <johannes@sipsolutions.net> writes: > This makes wireless extensions netns aware. > > Signed-off-by: Johannes Berg <johannes@sipsolutions.net> > --- > Is this ok, or is this racy? I guess what I'm asking is -- will > for_each_net() stop iterating over a netns that is going away before the > pernet exit op is called? If yes, this should be fine. for_each_net requires the rtnl_lock or the net_mutex to be safe. You aren't taking either so your code is racy. A dying network namespace will be removed from the net_namespace_list (aka for_each_net) before the per net exit methods are called. Is grabbing the rtnl_lock safe in your workqueue and is that something we want to do? Eric > include/net/net_namespace.h | 3 +++ > net/wireless/wext.c | 30 ++++++++++++++++++++++-------- > 2 files changed, 25 insertions(+), 8 deletions(-) > > --- wireless-testing.orig/include/net/net_namespace.h 2009-06-17 20:20:47.000000000 +0200 > +++ wireless-testing/include/net/net_namespace.h 2009-06-17 20:20:51.000000000 +0200 > @@ -78,6 +78,9 @@ struct net { > #ifdef CONFIG_XFRM > struct netns_xfrm xfrm; > #endif > +#ifdef CONFIG_WIRELESS_EXT > + struct sk_buff_head wext_nlevents; > +#endif > struct net_generic *gen; > }; > > --- wireless-testing.orig/net/wireless/wext.c 2009-06-17 20:20:47.000000000 +0200 > +++ wireless-testing/net/wireless/wext.c 2009-06-17 20:20:51.000000000 +0200 > @@ -1273,11 +1273,25 @@ int compat_wext_handle_ioctl(struct net > * Jean II > */ > > -static struct sk_buff_head wireless_nlevent_queue; > +static int __net_init wext_pernet_init(struct net *net) > +{ > + skb_queue_head_init(&net->wext_nlevents); > + return 0; > +} > + > +static void __net_exit wext_pernet_exit(struct net *net) > +{ > + skb_queue_purge(&net->wext_nlevents); > +} > + > +static struct pernet_operations wext_pernet_ops = { > + .init = wext_pernet_init, > + .exit = wext_pernet_exit, > +}; > > static int __init wireless_nlevent_init(void) > { > - skb_queue_head_init(&wireless_nlevent_queue); > + return register_pernet_subsys(&wext_pernet_ops); > return 0; > } > > @@ -1286,9 +1300,12 @@ subsys_initcall(wireless_nlevent_init); > static void wireless_nlevent_process(unsigned long data) > { > struct sk_buff *skb; > + struct net *net; > > - while ((skb = skb_dequeue(&wireless_nlevent_queue))) > - rtnl_notify(skb, &init_net, 0, RTNLGRP_LINK, NULL, GFP_ATOMIC); > + for_each_net(net) > + while ((skb = skb_dequeue(&net->wext_nlevents))) > + rtnl_notify(skb, net, 0, RTNLGRP_LINK, NULL, > + GFP_ATOMIC); > } > > static DECLARE_TASKLET(wireless_nlevent_tasklet, wireless_nlevent_process, 0); > @@ -1341,9 +1358,6 @@ static void rtmsg_iwinfo(struct net_devi > struct sk_buff *skb; > int err; > > - if (!net_eq(dev_net(dev), &init_net)) > - return; > - > skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC); > if (!skb) > return; > @@ -1356,7 +1370,7 @@ static void rtmsg_iwinfo(struct net_devi > } > > NETLINK_CB(skb).dst_group = RTNLGRP_LINK; > - skb_queue_tail(&wireless_nlevent_queue, skb); > + skb_queue_tail(&dev_net(dev)->wext_nlevents, skb); > tasklet_schedule(&wireless_nlevent_tasklet); > } > > > > -- > To unsubscribe from this list: send the line "unsubscribe netdev" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, 2009-06-17 at 13:46 -0700, Eric W. Biederman wrote: > Johannes Berg <johannes@sipsolutions.net> writes: > > > This makes wireless extensions netns aware. > > > > Signed-off-by: Johannes Berg <johannes@sipsolutions.net> > > --- > > Is this ok, or is this racy? I guess what I'm asking is -- will > > for_each_net() stop iterating over a netns that is going away before the > > pernet exit op is called? If yes, this should be fine. > > for_each_net requires the rtnl_lock or the net_mutex to be safe. > You aren't taking either so your code is racy. Ah ok! > A dying network namespace will be removed from the net_namespace_list > (aka for_each_net) before the per net exit methods are called. > > Is grabbing the rtnl_lock safe in your workqueue and is that something > we want to do? Ok, thanks. Well, it's a tasklet currently, but can easily be converted to a work struct and then take the rtnl. johannes
On Wed, 2009-06-17 at 13:46 -0700, Eric W. Biederman wrote: > Johannes Berg <johannes@sipsolutions.net> writes: > > > This makes wireless extensions netns aware. > > > > Signed-off-by: Johannes Berg <johannes@sipsolutions.net> > > --- > > Is this ok, or is this racy? I guess what I'm asking is -- will > > for_each_net() stop iterating over a netns that is going away before the > > pernet exit op is called? If yes, this should be fine. > > for_each_net requires the rtnl_lock or the net_mutex to be safe. > You aren't taking either so your code is racy. So it looks like I can also use rcu_read_lock(), but there's no for_each_net_rcu(), should there be? johannes
Johannes Berg <johannes@sipsolutions.net> writes: > On Wed, 2009-06-17 at 13:46 -0700, Eric W. Biederman wrote: >> Johannes Berg <johannes@sipsolutions.net> writes: >> >> > This makes wireless extensions netns aware. >> > >> > Signed-off-by: Johannes Berg <johannes@sipsolutions.net> >> > --- >> > Is this ok, or is this racy? I guess what I'm asking is -- will >> > for_each_net() stop iterating over a netns that is going away before the >> > pernet exit op is called? If yes, this should be fine. >> >> for_each_net requires the rtnl_lock or the net_mutex to be safe. >> You aren't taking either so your code is racy. > > So it looks like I can also use rcu_read_lock(), but there's no > for_each_net_rcu(), should there be? I'm not using rcu safe list manipulation. What makes it look like rcu_read_lock() is safe? Eric -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
--- wireless-testing.orig/include/net/net_namespace.h 2009-06-17 20:20:47.000000000 +0200 +++ wireless-testing/include/net/net_namespace.h 2009-06-17 20:20:51.000000000 +0200 @@ -78,6 +78,9 @@ struct net { #ifdef CONFIG_XFRM struct netns_xfrm xfrm; #endif +#ifdef CONFIG_WIRELESS_EXT + struct sk_buff_head wext_nlevents; +#endif struct net_generic *gen; }; --- wireless-testing.orig/net/wireless/wext.c 2009-06-17 20:20:47.000000000 +0200 +++ wireless-testing/net/wireless/wext.c 2009-06-17 20:20:51.000000000 +0200 @@ -1273,11 +1273,25 @@ int compat_wext_handle_ioctl(struct net * Jean II */ -static struct sk_buff_head wireless_nlevent_queue; +static int __net_init wext_pernet_init(struct net *net) +{ + skb_queue_head_init(&net->wext_nlevents); + return 0; +} + +static void __net_exit wext_pernet_exit(struct net *net) +{ + skb_queue_purge(&net->wext_nlevents); +} + +static struct pernet_operations wext_pernet_ops = { + .init = wext_pernet_init, + .exit = wext_pernet_exit, +}; static int __init wireless_nlevent_init(void) { - skb_queue_head_init(&wireless_nlevent_queue); + return register_pernet_subsys(&wext_pernet_ops); return 0; } @@ -1286,9 +1300,12 @@ subsys_initcall(wireless_nlevent_init); static void wireless_nlevent_process(unsigned long data) { struct sk_buff *skb; + struct net *net; - while ((skb = skb_dequeue(&wireless_nlevent_queue))) - rtnl_notify(skb, &init_net, 0, RTNLGRP_LINK, NULL, GFP_ATOMIC); + for_each_net(net) + while ((skb = skb_dequeue(&net->wext_nlevents))) + rtnl_notify(skb, net, 0, RTNLGRP_LINK, NULL, + GFP_ATOMIC); } static DECLARE_TASKLET(wireless_nlevent_tasklet, wireless_nlevent_process, 0); @@ -1341,9 +1358,6 @@ static void rtmsg_iwinfo(struct net_devi struct sk_buff *skb; int err; - if (!net_eq(dev_net(dev), &init_net)) - return; - skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC); if (!skb) return; @@ -1356,7 +1370,7 @@ static void rtmsg_iwinfo(struct net_devi } NETLINK_CB(skb).dst_group = RTNLGRP_LINK; - skb_queue_tail(&wireless_nlevent_queue, skb); + skb_queue_tail(&dev_net(dev)->wext_nlevents, skb); tasklet_schedule(&wireless_nlevent_tasklet); }
This makes wireless extensions netns aware. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> --- Is this ok, or is this racy? I guess what I'm asking is -- will for_each_net() stop iterating over a netns that is going away before the pernet exit op is called? If yes, this should be fine. include/net/net_namespace.h | 3 +++ net/wireless/wext.c | 30 ++++++++++++++++++++++-------- 2 files changed, 25 insertions(+), 8 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html