Patchwork sysfs: Don't allow the creation of symlinks we can't remove

login
register
mail settings
Submitter Johannes Berg
Date July 22, 2010, 10:41 a.m.
Message ID <1279795286.12439.8.camel@jlt3.sipsolutions.net>
Download mbox | patch
Permalink /patch/59560/
State RFC
Delegated to: David Miller
Headers show

Comments

Johannes Berg - July 22, 2010, 10:41 a.m.
On Thu, 2010-07-22 at 03:35 -0700, Eric W. Biederman wrote:

> >> The warning patch just makes things fail faster.  Although I get some of the
> >> wireless interfaces for hwsim when I use this one.
> >
> > Hmm, I didn't.
> 
> To be clear I just get hwsim0.  Not wlan0 or wlan1.

Ah, yes, but that's just a regular netdev, you can pretty much ignore
it. It just shows all hwsim traffic as it is on the "air" for sniffing.

> > Right, it actually starts working again with that patch you sent.
> > However, netns support is really broken:
> >
> > <create net namespace, put phy0/wlan0 into it>
> 
> Do we have a convenient command line tool to do this?
> I remember there being a different netlink message from
> normal network devices.

iw phy0 set netns <pid>

http://git.sipsolutions.net/iw.git

> > root@kvm:~# ip link
> > 3: wlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
> >     link/ether 02:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
> > 7: lo: <LOOPBACK> mtu 16436 qdisc noop state DOWN 
> >     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> > root@kvm:~# ls /sys/class/net/
> > eth0  hwsim0  lo  wlan1  wlan2
> 
> I think this is actually the output of something working.
> 
> I expect after you created a new netns you didn't mount
> a new instance of /sys.  /sys remembers which netns you
> had when you mounted it.  So you have to mount /sys again
> so you can see the /sys/class/net for the network namespace
> you are in.

Ohh, oops! I saw all the "current->" references in the code and somehow
expected the same instance of sysfs to show the right thing.

Yes, it works now. But the patch below doesn't seem to work, am I
missing something?

johannes

---
 include/linux/netdevice.h |    2 ++
 net/core/net-sysfs.c      |    3 ++-
 net/wireless/sysfs.c      |    9 +++++++++
 3 files changed, 13 insertions(+), 1 deletion(-)



--
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
Eric W. Biederman - July 22, 2010, 11:27 a.m.
Johannes Berg <johannes@sipsolutions.net> writes:

> On Thu, 2010-07-22 at 03:35 -0700, Eric W. Biederman wrote:
>
>> >> The warning patch just makes things fail faster.  Although I get some of the
>> >> wireless interfaces for hwsim when I use this one.
>> >
>> > Hmm, I didn't.
>> 
>> To be clear I just get hwsim0.  Not wlan0 or wlan1.
>
> Ah, yes, but that's just a regular netdev, you can pretty much ignore
> it. It just shows all hwsim traffic as it is on the "air" for sniffing.
>
>> > Right, it actually starts working again with that patch you sent.
>> > However, netns support is really broken:
>> >
>> > <create net namespace, put phy0/wlan0 into it>
>> 
>> Do we have a convenient command line tool to do this?
>> I remember there being a different netlink message from
>> normal network devices.
>
> iw phy0 set netns <pid>
>
> http://git.sipsolutions.net/iw.git
>
>> > root@kvm:~# ip link
>> > 3: wlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
>> >     link/ether 02:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
>> > 7: lo: <LOOPBACK> mtu 16436 qdisc noop state DOWN 
>> >     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
>> > root@kvm:~# ls /sys/class/net/
>> > eth0  hwsim0  lo  wlan1  wlan2
>> 
>> I think this is actually the output of something working.
>> 
>> I expect after you created a new netns you didn't mount
>> a new instance of /sys.  /sys remembers which netns you
>> had when you mounted it.  So you have to mount /sys again
>> so you can see the /sys/class/net for the network namespace
>> you are in.
>
> Ohh, oops! I saw all the "current->" references in the code and somehow
> expected the same instance of sysfs to show the right thing.
>
> Yes, it works now. But the patch below doesn't seem to work, am I
> missing something?

You are trying to move the phy devices as well?

My guess is that at least part of the problem is that you don't have a
ieee80211 directory under hwsim.

My apologies for not thinking about the peculiarities of the wireless
drivers.

Eric


> ---
>  include/linux/netdevice.h |    2 ++
>  net/core/net-sysfs.c      |    3 ++-
>  net/wireless/sysfs.c      |    9 +++++++++
>  3 files changed, 13 insertions(+), 1 deletion(-)
>
> --- wireless-testing.orig/include/linux/netdevice.h	2010-07-22 10:01:22.000000000 +0200
> +++ wireless-testing/include/linux/netdevice.h	2010-07-22 10:11:00.000000000 +0200
> @@ -2148,6 +2148,8 @@ extern void dev_seq_stop(struct seq_file
>  extern int netdev_class_create_file(struct class_attribute *class_attr);
>  extern void netdev_class_remove_file(struct class_attribute *class_attr);
>  
> +extern struct kobj_ns_type_operations net_ns_type_operations;
> +
>  extern char *netdev_drivername(const struct net_device *dev, char *buffer, int len);
>  
>  extern void linkwatch_run_queue(void);
> --- wireless-testing.orig/net/core/net-sysfs.c	2010-07-22 10:01:22.000000000 +0200
> +++ wireless-testing/net/core/net-sysfs.c	2010-07-22 10:11:51.000000000 +0200
> @@ -785,12 +785,13 @@ static const void *net_netlink_ns(struct
>  	return sock_net(sk);
>  }
>  
> -static struct kobj_ns_type_operations net_ns_type_operations = {
> +struct kobj_ns_type_operations net_ns_type_operations = {
>  	.type = KOBJ_NS_TYPE_NET,
>  	.current_ns = net_current_ns,
>  	.netlink_ns = net_netlink_ns,
>  	.initial_ns = net_initial_ns,
>  };
> +EXPORT_SYMBOL_GPL(net_ns_type_operations);
>  
>  static void net_kobj_ns_exit(struct net *net)
>  {
> --- wireless-testing.orig/net/wireless/sysfs.c	2010-07-22 10:01:22.000000000 +0200
> +++ wireless-testing/net/wireless/sysfs.c	2010-07-22 10:13:08.000000000 +0200
> @@ -110,6 +110,13 @@ static int wiphy_resume(struct device *d
>  	return ret;
>  }
>  
> +static const void *wiphy_namespace(struct device *d)
> +{
> +	struct wiphy *wiphy = container_of(d, struct wiphy, dev);
> +
> +	return wiphy_net(wiphy);
> +}
> +
>  struct class ieee80211_class = {
>  	.name = "ieee80211",
>  	.owner = THIS_MODULE,
> @@ -120,6 +127,8 @@ struct class ieee80211_class = {
>  #endif
>  	.suspend = wiphy_suspend,
>  	.resume = wiphy_resume,
> +	.ns_type = &net_ns_type_operations,
> +	.namespace = wiphy_namespace,
>  };
>  
>  int wiphy_sysfs_init(void)
--
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
Johannes Berg - July 22, 2010, 11:30 a.m.
On Thu, 2010-07-22 at 04:27 -0700, Eric W. Biederman wrote:

> >> Do we have a convenient command line tool to do this?
> >> I remember there being a different netlink message from
> >> normal network devices.
> >
> > iw phy0 set netns <pid>
> >
> > http://git.sipsolutions.net/iw.git
> >
> >> > root@kvm:~# ip link
> >> > 3: wlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
> >> >     link/ether 02:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
> >> > 7: lo: <LOOPBACK> mtu 16436 qdisc noop state DOWN 
> >> >     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> >> > root@kvm:~# ls /sys/class/net/
> >> > eth0  hwsim0  lo  wlan1  wlan2
> >> 
> >> I think this is actually the output of something working.
> >> 
> >> I expect after you created a new netns you didn't mount
> >> a new instance of /sys.  /sys remembers which netns you
> >> had when you mounted it.  So you have to mount /sys again
> >> so you can see the /sys/class/net for the network namespace
> >> you are in.
> >
> > Ohh, oops! I saw all the "current->" references in the code and somehow
> > expected the same instance of sysfs to show the right thing.
> >
> > Yes, it works now. But the patch below doesn't seem to work, am I
> > missing something?
> 
> You are trying to move the phy devices as well?

Yes. The intent is that each wireless phy lives in a netns along with
all of its child devices.

> My guess is that at least part of the problem is that you don't have a
> ieee80211 directory under hwsim.

But I should have? 'ieee80211' is a class just like 'net', no?

> My apologies for not thinking about the peculiarities of the wireless
> drivers.

No worries.

johannes

--
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

Patch

--- wireless-testing.orig/include/linux/netdevice.h	2010-07-22 10:01:22.000000000 +0200
+++ wireless-testing/include/linux/netdevice.h	2010-07-22 10:11:00.000000000 +0200
@@ -2148,6 +2148,8 @@  extern void dev_seq_stop(struct seq_file
 extern int netdev_class_create_file(struct class_attribute *class_attr);
 extern void netdev_class_remove_file(struct class_attribute *class_attr);
 
+extern struct kobj_ns_type_operations net_ns_type_operations;
+
 extern char *netdev_drivername(const struct net_device *dev, char *buffer, int len);
 
 extern void linkwatch_run_queue(void);
--- wireless-testing.orig/net/core/net-sysfs.c	2010-07-22 10:01:22.000000000 +0200
+++ wireless-testing/net/core/net-sysfs.c	2010-07-22 10:11:51.000000000 +0200
@@ -785,12 +785,13 @@  static const void *net_netlink_ns(struct
 	return sock_net(sk);
 }
 
-static struct kobj_ns_type_operations net_ns_type_operations = {
+struct kobj_ns_type_operations net_ns_type_operations = {
 	.type = KOBJ_NS_TYPE_NET,
 	.current_ns = net_current_ns,
 	.netlink_ns = net_netlink_ns,
 	.initial_ns = net_initial_ns,
 };
+EXPORT_SYMBOL_GPL(net_ns_type_operations);
 
 static void net_kobj_ns_exit(struct net *net)
 {
--- wireless-testing.orig/net/wireless/sysfs.c	2010-07-22 10:01:22.000000000 +0200
+++ wireless-testing/net/wireless/sysfs.c	2010-07-22 10:13:08.000000000 +0200
@@ -110,6 +110,13 @@  static int wiphy_resume(struct device *d
 	return ret;
 }
 
+static const void *wiphy_namespace(struct device *d)
+{
+	struct wiphy *wiphy = container_of(d, struct wiphy, dev);
+
+	return wiphy_net(wiphy);
+}
+
 struct class ieee80211_class = {
 	.name = "ieee80211",
 	.owner = THIS_MODULE,
@@ -120,6 +127,8 @@  struct class ieee80211_class = {
 #endif
 	.suspend = wiphy_suspend,
 	.resume = wiphy_resume,
+	.ns_type = &net_ns_type_operations,
+	.namespace = wiphy_namespace,
 };
 
 int wiphy_sysfs_init(void)