[PATCH v2] cfg80211: avoid restoring the regulatory during disconnect

[PATCH v2] cfg80211: avoid restoring the regulatory during disconnect

[Rajkumar Manoharan]

From: Sachin Ahuja <sahuja@codeaurora.org>

When STA disconnects, cfg80211 restores the regulatory setting
to world domain. For the drivers that ignore country ie, cfg80211
should remain in the default country code which is configured
at driver load and should do not be restored to world reg domain.

Signed-off-by: Sachin Ahuja <sahuja@codeaurora.org>
Signed-off-by: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
---
v2: Removed is_all_countryie_ignore
 net/wireless/sme.c | 22 +++++++++++++++++++++-
 1 file changed, 21 insertions(+), 1 deletion(-)

diff --git a/net/wireless/sme.c b/net/wireless/sme.c
index 4bc21a2..1b771a8 100644
--- a/net/wireless/sme.c
+++ b/net/wireless/sme.c
@@ -558,10 +558,30 @@ static bool cfg80211_is_all_idle(void)
 	return is_all_idle;
 }
 
+static bool cfg80211_is_all_countryie_ignore(void)
+{
+	struct cfg80211_registered_device *rdev;
+	struct wireless_dev *wdev;
+
+	list_for_each_entry(rdev, &cfg80211_rdev_list, list) {
+		list_for_each_entry(wdev, &rdev->wdev_list, list) {
+			wdev_lock(wdev);
+			if (!(wdev->wiphy->regulatory_flags &
+			      REGULATORY_COUNTRY_IE_IGNORE)) {
+				wdev_unlock(wdev);
+				return false;
+			}
+			wdev_unlock(wdev);
+		}
+	}
+	return true;
+}
+
 static void disconnect_work(struct work_struct *work)
 {
 	rtnl_lock();
-	if (cfg80211_is_all_idle())
+	if (cfg80211_is_all_idle() &&
+	    !cfg80211_is_all_countryie_ignore())
 		regulatory_hint_disconnect();
 	rtnl_unlock();
 }
-- 
1.9.2

Re: [PATCH v2] cfg80211: avoid restoring the regulatory during disconnect

[Johannes Berg]

On Fri, 2014-05-09 at 15:58 +0530, Rajkumar Manoharan wrote:

> +static bool cfg80211_is_all_countryie_ignore(void)
> +{
> +	struct cfg80211_registered_device *rdev;
> +	struct wireless_dev *wdev;
> +
> +	list_for_each_entry(rdev, &cfg80211_rdev_list, list) {
> +		list_for_each_entry(wdev, &rdev->wdev_list, list) {
> +			wdev_lock(wdev);
> +			if (!(wdev->wiphy->regulatory_flags &
> +			      REGULATORY_COUNTRY_IE_IGNORE)) {
> +				wdev_unlock(wdev);

The locking here seems completely pointless.

Also, maybe you can do something about the long lines, maybe with a
temporary variable?

johannes

Re: [PATCH v2] cfg80211: avoid restoring the regulatory during disconnect

[Rajkumar Manoharan]

On Mon, May 12, 2014 at 04:00:03PM +0200, Johannes Berg wrote:
> On Fri, 2014-05-09 at 15:58 +0530, Rajkumar Manoharan wrote:
> 
> > +static bool cfg80211_is_all_countryie_ignore(void)
> > +{
> > +	struct cfg80211_registered_device *rdev;
> > +	struct wireless_dev *wdev;
> > +
> > +	list_for_each_entry(rdev, &cfg80211_rdev_list, list) {
> > +		list_for_each_entry(wdev, &rdev->wdev_list, list) {
> > +			wdev_lock(wdev);
> > +			if (!(wdev->wiphy->regulatory_flags &
> > +			      REGULATORY_COUNTRY_IE_IGNORE)) {
> > +				wdev_unlock(wdev);
> 
> The locking here seems completely pointless.
>
locking is handled in same way of cfg80211_is_all_idle. Still
want to remove locks?

> Also, maybe you can do something about the long lines, maybe with a
> temporary variable?
>

-Rajkumar

Re: [PATCH v2] cfg80211: avoid restoring the regulatory during disconnect

[Rajkumar Manoharan]

On Mon, May 12, 2014 at 10:51:37PM +0530, Rajkumar Manoharan wrote:
> On Mon, May 12, 2014 at 04:00:03PM +0200, Johannes Berg wrote:
> > On Fri, 2014-05-09 at 15:58 +0530, Rajkumar Manoharan wrote:
> > 
> > > +static bool cfg80211_is_all_countryie_ignore(void)
> > > +{
> > > +	struct cfg80211_registered_device *rdev;
> > > +	struct wireless_dev *wdev;
> > > +
> > > +	list_for_each_entry(rdev, &cfg80211_rdev_list, list) {
> > > +		list_for_each_entry(wdev, &rdev->wdev_list, list) {
> > > +			wdev_lock(wdev);
> > > +			if (!(wdev->wiphy->regulatory_flags &
> > > +			      REGULATORY_COUNTRY_IE_IGNORE)) {
> > > +				wdev_unlock(wdev);
> > 
> > The locking here seems completely pointless.
> >
> locking is handled in same way of cfg80211_is_all_idle. Still
> want to remove locks?
> 
> > Also, maybe you can do something about the long lines, maybe with a
> > temporary variable?
> >
sorry for the noise. sent v3 patch.

-Rajkumar