cfg80211: fix locking in netlink owner interface destruction

From: Johannes Berg <johannes.berg@intel.com>

Harald Arnesen reported [1] a deadlock at reboot time, and after
he captured a stack trace a picture developed of what's going on:

The distribution he's using is using iwd (not wpa_supplicant) to
manage wireless. iwd will usually use the "socket owner" option
when it creates new interfaces, so that they're automatically
destroyed when it quits (unexpectedly or otherwise). This is also
done by wpa_supplicant, but it doesn't do it for the normal one,
only for additional ones, which is different with iwd.

Anyway, during shutdown, iwd quits while the netdev is still UP,
i.e. IFF_UP is set. This causes the stack trace that Linus so
nicely transcribed from the pictures:

cfg80211_destroy_iface_wk() takes wiphy_lock
 -> cfg80211_destroy_ifaces()
  ->ieee80211_del_iface
    ->ieeee80211_if_remove
      ->cfg80211_unregister_wdev
        ->unregister_netdevice_queue
          ->dev_close_many
            ->__dev_close_many
              ->raw_notifier_call_chain
                ->cfg80211_netdev_notifier_call
and that last call tries to take wiphy_lock again.

In commit a05829a7222e ("cfg80211: avoid holding the RTNL when
calling the driver") I had taken into account the possibility of
recursing from cfg80211 into cfg80211_netdev_notifier_call() via
the network stack, but only for NETDEV_UNREGISTER, not for what
happens here, NETDEV_GOING_DOWN and NETDEV_DOWN notifications.

Additionally, while this worked still back in commit 78f22b6a3a92
("cfg80211: allow userspace to take ownership of interfaces"), it
missed another corner case: unregistering a netdev will cause
dev_close() to be called, and thus stop wireless operations (e.g.
disconnecting), but there are some types of virtual interfaces in
wifi that don't have a netdev - for that we need an additional
call to cfg80211_leave().

So, to fix this mess, change cfg80211_destroy_ifaces() to not
require the wiphy_lock(), but instead make it acquire it, but
only after it has actually closed all the netdevs on the list,
and then call cfg80211_leave() as well before removing them
from the driver, to fix the second issue. The locking change in
this requires modifying the nl80211 call to not get the wiphy
lock passed in, but acquire it by itself after flushing any
potentially pending destruction requests.

[1] https://lore.kernel.org/r/09464e67-f3de-ac09-28a3-e27b7914ee7d@skogtun.org

Cc: stable@vger.kernel.org # 5.12
Reported-by: Harald Arnesen <harald@skogtun.org>
Fixes: 776a39b8196d ("cfg80211: call cfg80211_destroy_ifaces() with wiphy lock held")
Fixes: 78f22b6a3a92 ("cfg80211: allow userspace to take ownership of interfaces")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
---
Linus, I'll send this the regular way, just CC'ing you since
you were involved in the debug.
---
 net/wireless/core.c    | 21 +++++++++++++++++----
 net/wireless/nl80211.c | 24 +++++++++++++++++++-----
 2 files changed, 36 insertions(+), 9 deletions(-)

I can confirm that the machine reboots with this patch applied.
Harald Arnesen


Johannes Berg [27.04.2021 11:49]:
> From: Johannes Berg <johannes.berg@intel.com>
> 
> Harald Arnesen reported [1] a deadlock at reboot time, and after
> he captured a stack trace a picture developed of what's going on:
> 
> The distribution he's using is using iwd (not wpa_supplicant) to
> manage wireless. iwd will usually use the "socket owner" option
> when it creates new interfaces, so that they're automatically
> destroyed when it quits (unexpectedly or otherwise). This is also
> done by wpa_supplicant, but it doesn't do it for the normal one,
> only for additional ones, which is different with iwd.
> 
> Anyway, during shutdown, iwd quits while the netdev is still UP,
> i.e. IFF_UP is set. This causes the stack trace that Linus so
> nicely transcribed from the pictures:
> 
> cfg80211_destroy_iface_wk() takes wiphy_lock
>  -> cfg80211_destroy_ifaces()
>   ->ieee80211_del_iface
>     ->ieeee80211_if_remove
>       ->cfg80211_unregister_wdev
>         ->unregister_netdevice_queue
>           ->dev_close_many
>             ->__dev_close_many
>               ->raw_notifier_call_chain
>                 ->cfg80211_netdev_notifier_call
> and that last call tries to take wiphy_lock again.
> 
> In commit a05829a7222e ("cfg80211: avoid holding the RTNL when
> calling the driver") I had taken into account the possibility of
> recursing from cfg80211 into cfg80211_netdev_notifier_call() via
> the network stack, but only for NETDEV_UNREGISTER, not for what
> happens here, NETDEV_GOING_DOWN and NETDEV_DOWN notifications.
> 
> Additionally, while this worked still back in commit 78f22b6a3a92
> ("cfg80211: allow userspace to take ownership of interfaces"), it
> missed another corner case: unregistering a netdev will cause
> dev_close() to be called, and thus stop wireless operations (e.g.
> disconnecting), but there are some types of virtual interfaces in
> wifi that don't have a netdev - for that we need an additional
> call to cfg80211_leave().
> 
> So, to fix this mess, change cfg80211_destroy_ifaces() to not
> require the wiphy_lock(), but instead make it acquire it, but
> only after it has actually closed all the netdevs on the list,
> and then call cfg80211_leave() as well before removing them
> from the driver, to fix the second issue. The locking change in
> this requires modifying the nl80211 call to not get the wiphy
> lock passed in, but acquire it by itself after flushing any
> potentially pending destruction requests.
> 
> [1] https://lore.kernel.org/r/09464e67-f3de-ac09-28a3-e27b7914ee7d@skogtun.org
> 
> Cc: stable@vger.kernel.org # 5.12
> Reported-by: Harald Arnesen <harald@skogtun.org>
> Fixes: 776a39b8196d ("cfg80211: call cfg80211_destroy_ifaces() with wiphy lock held")
> Fixes: 78f22b6a3a92 ("cfg80211: allow userspace to take ownership of interfaces")
> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
> ---
> Linus, I'll send this the regular way, just CC'ing you since
> you were involved in the debug.
> ---
>  net/wireless/core.c    | 21 +++++++++++++++++----
>  net/wireless/nl80211.c | 24 +++++++++++++++++++-----
>  2 files changed, 36 insertions(+), 9 deletions(-)
> 
> diff --git a/net/wireless/core.c b/net/wireless/core.c
> index a2785379df6e..589ee5a69a2e 100644
> --- a/net/wireless/core.c
> +++ b/net/wireless/core.c
> @@ -332,14 +332,29 @@ static void cfg80211_event_work(struct work_struct *work)
>  void cfg80211_destroy_ifaces(struct cfg80211_registered_device *rdev)
>  {
>  	struct wireless_dev *wdev, *tmp;
> +	bool found = false;
>  
>  	ASSERT_RTNL();
> -	lockdep_assert_wiphy(&rdev->wiphy);
>  
> +	list_for_each_entry(wdev, &rdev->wiphy.wdev_list, list) {
> +		if (wdev->nl_owner_dead) {
> +			if (wdev->netdev)
> +				dev_close(wdev->netdev);
> +			found = true;
> +		}
> +	}
> +
> +	if (!found)
> +		return;
> +
> +	wiphy_lock(&rdev->wiphy);
>  	list_for_each_entry_safe(wdev, tmp, &rdev->wiphy.wdev_list, list) {
> -		if (wdev->nl_owner_dead)
> +		if (wdev->nl_owner_dead) {
> +			cfg80211_leave(rdev, wdev);
>  			rdev_del_virtual_intf(rdev, wdev);
> +		}
>  	}
> +	wiphy_unlock(&rdev->wiphy);
>  }
>  
>  static void cfg80211_destroy_iface_wk(struct work_struct *work)
> @@ -350,9 +365,7 @@ static void cfg80211_destroy_iface_wk(struct work_struct *work)
>  			    destroy_work);
>  
>  	rtnl_lock();
> -	wiphy_lock(&rdev->wiphy);
>  	cfg80211_destroy_ifaces(rdev);
> -	wiphy_unlock(&rdev->wiphy);
>  	rtnl_unlock();
>  }
>  
> diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
> index b1df42e4f1eb..a5224da63832 100644
> --- a/net/wireless/nl80211.c
> +++ b/net/wireless/nl80211.c
> @@ -3929,7 +3929,7 @@ static int nl80211_set_interface(struct sk_buff *skb, struct genl_info *info)
>  	return err;
>  }
>  
> -static int nl80211_new_interface(struct sk_buff *skb, struct genl_info *info)
> +static int _nl80211_new_interface(struct sk_buff *skb, struct genl_info *info)
>  {
>  	struct cfg80211_registered_device *rdev = info->user_ptr[0];
>  	struct vif_params params;
> @@ -3938,9 +3938,6 @@ static int nl80211_new_interface(struct sk_buff *skb, struct genl_info *info)
>  	int err;
>  	enum nl80211_iftype type = NL80211_IFTYPE_UNSPECIFIED;
>  
> -	/* to avoid failing a new interface creation due to pending removal */
> -	cfg80211_destroy_ifaces(rdev);
> -
>  	memset(&params, 0, sizeof(params));
>  
>  	if (!info->attrs[NL80211_ATTR_IFNAME])
> @@ -4028,6 +4025,21 @@ static int nl80211_new_interface(struct sk_buff *skb, struct genl_info *info)
>  	return genlmsg_reply(msg, info);
>  }
>  
> +static int nl80211_new_interface(struct sk_buff *skb, struct genl_info *info)
> +{
> +	struct cfg80211_registered_device *rdev = info->user_ptr[0];
> +	int ret;
> +
> +	/* to avoid failing a new interface creation due to pending removal */
> +	cfg80211_destroy_ifaces(rdev);
> +
> +	wiphy_lock(&rdev->wiphy);
> +	ret = _nl80211_new_interface(skb, info);
> +	wiphy_unlock(&rdev->wiphy);
> +
> +	return ret;
> +}
> +
>  static int nl80211_del_interface(struct sk_buff *skb, struct genl_info *info)
>  {
>  	struct cfg80211_registered_device *rdev = info->user_ptr[0];
> @@ -15040,7 +15052,9 @@ static const struct genl_small_ops nl80211_small_ops[] = {
>  		.doit = nl80211_new_interface,
>  		.flags = GENL_UNS_ADMIN_PERM,
>  		.internal_flags = NL80211_FLAG_NEED_WIPHY |
> -				  NL80211_FLAG_NEED_RTNL,
> +				  NL80211_FLAG_NEED_RTNL |
> +				  /* we take the wiphy mutex later ourselves */
> +				  NL80211_FLAG_NO_WIPHY_MTX,
>  	},
>  	{
>  		.cmd = NL80211_CMD_DEL_INTERFACE,
>

On Tue, 2021-04-27 at 14:07 +0200, Harald Arnesen wrote:
> I can confirm that the machine reboots with this patch applied.

Great, thanks!

johannes

On Tue, Apr 27, 2021 at 2:59 AM Johannes Berg <johannes@sipsolutions.net> wrote:
>
> Though then again, I'm not sure I have a good pathway into the tree
> right now (pre rc1), so if you want to throw it in sooner that's fine
> too.

Ok, applied with Harald's tested-by added.

Thanks,
           Linus