Re: linux-next: Tree for June 4 (mac80211)

Re: linux-next: Tree for June 4 (mac80211)

[Randy Dunlap]

On Fri, 4 Jun 2010 15:30:06 +1000 Stephen Rothwell wrote:

> Hi all,
> 
> Changes since 20100603:
> 
> My fixes tree contains:
>       v4l-dvb: update gfp/slab.h includes
>       arm: update gfp/slab.h includes
>       davinci: update gfp/slab.h includes
>       ocfs2: update gfp/slab.h includes
>       acpi: update gfp/slab.h includes
> 
> The wireless tree gained a conflict against the net tree.
> 
> The rr tree gained a build failure for which I applied a patch.


when CONFIG_INET is not enabled:


ERROR: "register_inetaddr_notifier" [net/mac80211/mac80211.ko] undefined!
ERROR: "unregister_inetaddr_notifier" [net/mac80211/mac80211.ko] undefined!

or

(.text+0xb9585): undefined reference to `unregister_inetaddr_notifier'
(.text+0xb9b84): undefined reference to `register_inetaddr_notifier'


---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

[PATCH] mac80211: make ARP filtering depend on CONFIG_INET

[John W. Linville]

Signed-off-by: John W. Linville <linville@tuxdriver.com>
---
 net/mac80211/main.c |    6 ++++++
 net/mac80211/mlme.c |    2 ++
 2 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/net/mac80211/main.c b/net/mac80211/main.c
index 045ead9..5706156 100644
--- a/net/mac80211/main.c
+++ b/net/mac80211/main.c
@@ -329,6 +329,7 @@ static void ieee80211_recalc_smps_work(struct work_struct *work)
 	mutex_unlock(&local->iflist_mtx);
 }
 
+#ifdef CONFIG_INET
 int ieee80211_set_arp_filter(struct ieee80211_sub_if_data *sdata)
 {
 	struct in_device *idev;
@@ -380,6 +381,7 @@ static int ieee80211_ifa_changed(struct notifier_block *nb,
 
 	return NOTIFY_DONE;
 }
+#endif
 
 struct ieee80211_hw *ieee80211_alloc_hw(size_t priv_data_len,
 					const struct ieee80211_ops *ops)
@@ -669,10 +671,12 @@ int ieee80211_register_hw(struct ieee80211_hw *hw)
 		goto fail_pm_qos;
 	}
 
+#ifdef CONFIG_INET
 	local->ifa_notifier.notifier_call = ieee80211_ifa_changed;
 	result = register_inetaddr_notifier(&local->ifa_notifier);
 	if (result)
 		goto fail_ifa;
+#endif
 
 	return 0;
 
@@ -707,7 +711,9 @@ void ieee80211_unregister_hw(struct ieee80211_hw *hw)
 
 	pm_qos_remove_notifier(PM_QOS_NETWORK_LATENCY,
 			       &local->network_latency_notifier);
+#ifdef CONFIG_INET
 	unregister_inetaddr_notifier(&local->ifa_notifier);
+#endif
 
 	rtnl_lock();
 
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index 7e72013..3623bb7 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -2078,6 +2078,7 @@ static enum work_done_result ieee80211_assoc_done(struct ieee80211_work *wk,
 			cfg80211_send_assoc_timeout(wk->sdata->dev,
 						    wk->filter_ta);
 			return WORK_DONE_DESTROY;
+#ifdef CONFIG_INET
 		} else {
 			mutex_unlock(&wk->sdata->u.mgd.mtx);
 
@@ -2088,6 +2089,7 @@ static enum work_done_result ieee80211_assoc_done(struct ieee80211_work *wk,
 			rtnl_lock();
 			ieee80211_set_arp_filter(wk->sdata);
 			rtnl_unlock();
+#endif
 		}
 	}
 
-- 
1.6.6.1

Re: [PATCH] mac80211: make ARP filtering depend on CONFIG_INET

[Juuso Oikarinen]

Hi,

On Fri, 2010-06-04 at 20:28 +0200, ext John W. Linville wrote:
> Signed-off-by: John W. Linville <linville@tuxdriver.com>
> ---
>  net/mac80211/main.c |    6 ++++++
>  net/mac80211/mlme.c |    2 ++
>  2 files changed, 8 insertions(+), 0 deletions(-)



> diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
> index 7e72013..3623bb7 100644
> --- a/net/mac80211/mlme.c
> +++ b/net/mac80211/mlme.c
> @@ -2078,6 +2078,7 @@ static enum work_done_result ieee80211_assoc_done(struct ieee80211_work *wk,
>  			cfg80211_send_assoc_timeout(wk->sdata->dev,
>  						    wk->filter_ta);
>  			return WORK_DONE_DESTROY;
> +#ifdef CONFIG_INET
>  		} else {
>  			mutex_unlock(&wk->sdata->u.mgd.mtx);
>  
> @@ -2088,6 +2089,7 @@ static enum work_done_result ieee80211_assoc_done(struct ieee80211_work *wk,
>  			rtnl_lock();
>  			ieee80211_set_arp_filter(wk->sdata);
>  			rtnl_unlock();
> +#endif
>  		}

Here, if CONFIG_INET is not set, and the previous condition is not true,
the mutex will not be unlocked.

The arp-filter code here is causing some locking issues, so I'll need to
move it elsewhere anyway. I can also fix the unlocking while at it.

-Juuso

>  	}
>