diff mbox

[v2] wireless: Support can-scan-one logic.

Message ID 1305586296-23147-1-git-send-email-greearb@candelatech.com (mailing list archive)
State Not Applicable, archived
Headers show

Commit Message

Ben Greear May 16, 2011, 10:51 p.m. UTC
From: Ben Greear <greearb@candelatech.com>

Enable this by passing a -1 for a scan frequency.

When enabled, the system will only scan the current active
channel if at least one VIF is actively using it.  If no
VIFS are active or this flag is disabled, then default
behaviour is used.

This helps when using multiple STA interfaces that otherwise might
constantly be trying to scan all channels.

Signed-off-by: Ben Greear <greearb@candelatech.com>
---

v2:  Remove verbose printk commands, fix up comment syntax.

:100644 100644 c32e683... c0be322... M	include/net/cfg80211.h
:100644 100644 489b6ad... d00287f... M	net/mac80211/scan.c
:100644 100644 0a199a1... c0b9cad... M	net/wireless/nl80211.c
:100644 100644 e17b0be... 9044120... M	net/wireless/sme.c
 include/net/cfg80211.h |    3 +++
 net/mac80211/scan.c    |   42 ++++++++++++++++++++++++++++++++++++++++++
 net/wireless/nl80211.c |   22 ++++++++++++++++++++--
 net/wireless/sme.c     |    6 ++++++
 4 files changed, 71 insertions(+), 2 deletions(-)

Comments

Johannes Berg May 18, 2011, 11:02 p.m. UTC | #1
On Mon, 2011-05-16 at 15:51 -0700, greearb@candelatech.com wrote:
> From: Ben Greear <greearb@candelatech.com>
> 
> Enable this by passing a -1 for a scan frequency.

I still don't think we should do that, especially not with -1. That's
totally non-netlink like inband signalling. I'll also reply to your
other mail though since I don't think it makes sense to have this sort
of convenience function in the kernel.

johannes

--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Ben Greear May 18, 2011, 11:30 p.m. UTC | #2
On 05/18/2011 04:02 PM, Johannes Berg wrote:
> On Mon, 2011-05-16 at 15:51 -0700, greearb@candelatech.com wrote:
>> From: Ben Greear<greearb@candelatech.com>
>>
>> Enable this by passing a -1 for a scan frequency.
>
> I still don't think we should do that, especially not with -1. That's
> totally non-netlink like inband signalling. I'll also reply to your
> other mail though since I don't think it makes sense to have this sort
> of convenience function in the kernel.

It's virtually impossible (as far as I can tell) to carry an
out-of-tree netlink patch that uses a new netlink message
and still keep things backwards-compat when someone adds a
new message to the upstream kernel.  So, the -1 hack works
well for me.

If it were to go into the kernel proper, then we could
add a proper flag to the netlink API and start using
that.

If you just don't like the feature, thats OK...it is a pretty
specialized feature, and easy enough to carry in my own tree.

Thanks,
Ben
Luis Rodriguez June 2, 2011, 7:12 p.m. UTC | #3
On Wed, May 18, 2011 at 4:30 PM, Ben Greear <greearb@candelatech.com> wrote:
> On 05/18/2011 04:02 PM, Johannes Berg wrote:
>>
>> On Mon, 2011-05-16 at 15:51 -0700, greearb@candelatech.com wrote:
>>>
>>> From: Ben Greear<greearb@candelatech.com>
>>>
>>> Enable this by passing a -1 for a scan frequency.
>>
>> I still don't think we should do that, especially not with -1. That's
>> totally non-netlink like inband signalling. I'll also reply to your
>> other mail though since I don't think it makes sense to have this sort
>> of convenience function in the kernel.
>
> It's virtually impossible (as far as I can tell) to carry an
> out-of-tree netlink patch that uses a new netlink message
> and still keep things backwards-compat when someone adds a
> new message to the upstream kernel.  So, the -1 hack works
> well for me.
>
> If it were to go into the kernel proper, then we could
> add a proper flag to the netlink API and start using
> that.
>
> If you just don't like the feature, thats OK...it is a pretty
> specialized feature, and easy enough to carry in my own tree.

Since this shit was merged can you add a respective documentation
extension for the command for nl80211.h?

  Luis
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Johannes Berg June 2, 2011, 7:14 p.m. UTC | #4
On Thu, 2011-06-02 at 12:12 -0700, Luis R. Rodriguez wrote:

> Since this shit was merged can you add a respective documentation
> extension for the command for nl80211.h?

You're kidding, right? The -1 crap was merged? Breaking all drivers
other than mac80211?

johannes

--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Ben Greear June 2, 2011, 7:15 p.m. UTC | #5
On 06/02/2011 12:12 PM, Luis R. Rodriguez wrote:
> On Wed, May 18, 2011 at 4:30 PM, Ben Greear<greearb@candelatech.com>  wrote:
>> On 05/18/2011 04:02 PM, Johannes Berg wrote:
>>>
>>> On Mon, 2011-05-16 at 15:51 -0700, greearb@candelatech.com wrote:
>>>>
>>>> From: Ben Greear<greearb@candelatech.com>
>>>>
>>>> Enable this by passing a -1 for a scan frequency.
>>>
>>> I still don't think we should do that, especially not with -1. That's
>>> totally non-netlink like inband signalling. I'll also reply to your
>>> other mail though since I don't think it makes sense to have this sort
>>> of convenience function in the kernel.
>>
>> It's virtually impossible (as far as I can tell) to carry an
>> out-of-tree netlink patch that uses a new netlink message
>> and still keep things backwards-compat when someone adds a
>> new message to the upstream kernel.  So, the -1 hack works
>> well for me.
>>
>> If it were to go into the kernel proper, then we could
>> add a proper flag to the netlink API and start using
>> that.
>>
>> If you just don't like the feature, thats OK...it is a pretty
>> specialized feature, and easy enough to carry in my own tree.
>
> Since this shit was merged can you add a respective documentation
> extension for the command for nl80211.h?

You un-merged it almost immediately, as far as I can tell.

What tree is it merged in?

Thanks,
Ben

>
>    Luis
Luis Rodriguez June 2, 2011, 7:15 p.m. UTC | #6
On Thu, Jun 2, 2011 at 12:14 PM, Johannes Berg
<johannes@sipsolutions.net> wrote:
> On Thu, 2011-06-02 at 12:12 -0700, Luis R. Rodriguez wrote:
>
>> Since this shit was merged can you add a respective documentation
>> extension for the command for nl80211.h?
>
> You're kidding, right? The -1 crap was merged? Breaking all drivers
> other than mac80211?
commit 718ab883ccd1258c1e5c150edfb407880df9eeaa
Author: Ben Greear <greearb@candelatech.com>
Date:   Mon May 16 15:51:36 2011 -0700

    wireless: Support can-scan-one logic.

    Enable this by passing a -1 for a scan frequency.

    When enabled, the system will only scan the current active
    channel if at least one VIF is actively using it.  If no
    VIFS are active or this flag is disabled, then default
    behaviour is used.

    This helps when using multiple STA interfaces that otherwise might
    constantly be trying to scan all channels.

    Signed-off-by: Ben Greear <greearb@candelatech.com>
    Signed-off-by: John W. Linville <linville@tuxdriver.com>

My baby just turned.

  Luis
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Luis Rodriguez June 2, 2011, 7:16 p.m. UTC | #7
On Thu, Jun 2, 2011 at 12:15 PM, Ben Greear <greearb@candelatech.com> wrote:
> What tree is it merged in?

Ah it was reverted. Sorry.

  Luis
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h
index c32e683..c0be322 100644
--- a/include/net/cfg80211.h
+++ b/include/net/cfg80211.h
@@ -780,6 +780,8 @@  struct cfg80211_ssid {
  * @wiphy: the wiphy this was for
  * @dev: the interface
  * @aborted: (internal) scan request was notified as aborted
+ * @can_scan_one:  If true, only scan active channel if at least one
+ *       vif is already associated.
  */
 struct cfg80211_scan_request {
 	struct cfg80211_ssid *ssids;
@@ -792,6 +794,7 @@  struct cfg80211_scan_request {
 	struct wiphy *wiphy;
 	struct net_device *dev;
 	bool aborted;
+	bool can_scan_one;
 
 	/* keep last */
 	struct ieee80211_channel *channels[0];
diff --git a/net/mac80211/scan.c b/net/mac80211/scan.c
index 489b6ad..d00287f 100644
--- a/net/mac80211/scan.c
+++ b/net/mac80211/scan.c
@@ -353,6 +353,48 @@  static int ieee80211_start_sw_scan(struct ieee80211_local *local)
 	 * nullfunc frames and probe requests will be dropped in
 	 * ieee80211_tx_h_check_assoc().
 	 */
+	int associated_station_vifs = 0;
+	int running_station_vifs = 0; /* not necessarily associated */
+	int running_other_vifs = 0; /* AP, etc */
+	struct ieee80211_sub_if_data *sdata;
+
+	if (local->scan_req->can_scan_one && local->scan_req->n_channels >= 1) {
+		struct sta_info *sta;
+		mutex_lock(&local->iflist_mtx);
+		list_for_each_entry(sdata, &local->interfaces, list) {
+			if (!ieee80211_sdata_running(sdata))
+				continue;
+
+			if (sdata->vif.type == NL80211_IFTYPE_STATION)
+				running_station_vifs++;
+			else
+				running_other_vifs++;
+		}
+		mutex_unlock(&local->iflist_mtx);
+
+		rcu_read_lock();
+		list_for_each_entry_rcu(sta, &local->sta_list, list) {
+			if (!ieee80211_sdata_running(sta->sdata))
+				continue;
+			if (sta->sdata->vif.type != NL80211_IFTYPE_STATION)
+				continue;
+			if (test_sta_flags(sta, WLAN_STA_ASSOC))
+				associated_station_vifs++;
+		}
+		rcu_read_unlock();
+
+		/*
+		 * If one sta is associated, we don't want another to start
+		 * scanning on all channels, as that will interfere with the
+		 * one already associated.
+		 */
+		if ((running_other_vifs > 0) ||
+		    (associated_station_vifs > 1)) {
+			local->scan_req->channels[0] = local->hw.conf.channel;
+			local->scan_req->n_channels = 1;
+		}
+	}
+
 	drv_sw_scan_start(local);
 
 	local->leave_oper_channel_time = 0;
diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
index 0a199a1..c0b9cad 100644
--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -3176,6 +3176,9 @@  static int validate_scan_freqs(struct nlattr *freqs)
 	int n_channels = 0, tmp1, tmp2;
 
 	nla_for_each_nested(attr1, freqs, tmp1) {
+		if (nla_get_u32(attr1) == 0xFFFFFFFF)
+			continue; /* skip can-scan-one flag */
+
 		n_channels++;
 		/*
 		 * Some hardware has a limited channel list for
@@ -3207,6 +3210,7 @@  static int nl80211_trigger_scan(struct sk_buff *skb, struct genl_info *info)
 	int err, tmp, n_ssids = 0, n_channels, i;
 	enum ieee80211_band band;
 	size_t ie_len;
+	bool do_all_chan = true;
 
 	if (!is_valid_ie_attr(info->attrs[NL80211_ATTR_IE]))
 		return -EINVAL;
@@ -3223,8 +3227,9 @@  static int nl80211_trigger_scan(struct sk_buff *skb, struct genl_info *info)
 		n_channels = validate_scan_freqs(
 				info->attrs[NL80211_ATTR_SCAN_FREQUENCIES]);
 		if (!n_channels)
-			return -EINVAL;
+			goto auto_channels;
 	} else {
+auto_channels:
 		n_channels = 0;
 
 		for (band = 0; band < IEEE80211_NUM_BANDS; band++)
@@ -3270,6 +3275,17 @@  static int nl80211_trigger_scan(struct sk_buff *skb, struct genl_info *info)
 		nla_for_each_nested(attr, info->attrs[NL80211_ATTR_SCAN_FREQUENCIES], tmp) {
 			struct ieee80211_channel *chan;
 
+			/*
+			 * Special hack:  channel -1 means 'scan only active
+			 * channel if any VIFs on this device are associated
+			 * on the channel.
+			 */
+			if (nla_get_u32(attr) == 0xFFFFFFFF) {
+				request->can_scan_one = true;
+				continue;
+			}
+
+			do_all_chan = false;
 			chan = ieee80211_get_channel(wiphy, nla_get_u32(attr));
 
 			if (!chan) {
@@ -3284,7 +3300,9 @@  static int nl80211_trigger_scan(struct sk_buff *skb, struct genl_info *info)
 			request->channels[i] = chan;
 			i++;
 		}
-	} else {
+	}
+
+	if (do_all_chan) {
 		/* all channels */
 		for (band = 0; band < IEEE80211_NUM_BANDS; band++) {
 			int j;
diff --git a/net/wireless/sme.c b/net/wireless/sme.c
index e17b0be..9044120 100644
--- a/net/wireless/sme.c
+++ b/net/wireless/sme.c
@@ -105,6 +105,12 @@  static int cfg80211_conn_scan(struct wireless_dev *wdev)
 	if (!request)
 		return -ENOMEM;
 
+	/*
+	 * If at least one VIF on this hardware is already associated, then
+	 * only scan on the active channel.
+	 */
+	request->can_scan_one = true;
+
 	if (wdev->conn->params.channel)
 		request->channels[0] = wdev->conn->params.channel;
 	else {