[PATCH 1/1] mac80211: tell driver when dtim change detected

[PATCH 1/1] mac80211: tell driver when dtim change detected

[wey-yi.w.guy]

From: Wey-Yi Guy <wey-yi.w.guy@intel.com>

In current implementation, mac80211 send dtim_period update to driver during
association, but if no NetworkManager or similar application perform scan
operation, plus tim_ie is not part of probe response; mac80211 will not get
beacon with dtim information later, then  mac80211 will not pass the 
information to driver for update.

Call ieee80211_hw_config() with IEEE80211_CONF_CHANGE_PS flag set to
allow driver make correct dtim adjustment if dtim_period change detected.

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
---
 net/mac80211/scan.c |   19 ++++++++++++++++++-
 1 files changed, 18 insertions(+), 1 deletions(-)

diff --git a/net/mac80211/scan.c b/net/mac80211/scan.c
index 71e10ca..a5508f5 100644
--- a/net/mac80211/scan.c
+++ b/net/mac80211/scan.c
@@ -64,6 +64,7 @@ ieee80211_bss_info_update(struct ieee80211_local *local,
 	struct ieee80211_bss *bss;
 	int clen;
 	s32 signal = 0;
+	struct ieee80211_sub_if_data *sdata;
 
 	if (local->hw.flags & IEEE80211_HW_SIGNAL_DBM)
 		signal = rx_status->signal * 100;
@@ -87,7 +88,23 @@ ieee80211_bss_info_update(struct ieee80211_local *local,
 	if (elems->tim) {
 		struct ieee80211_tim_ie *tim_ie =
 			(struct ieee80211_tim_ie *)elems->tim;
-		bss->dtim_period = tim_ie->dtim_period;
+		if (bss->dtim_period != tim_ie->dtim_period) {
+			bss->dtim_period = tim_ie->dtim_period;
+			mutex_lock(&local->iflist_mtx);
+			list_for_each_entry(sdata, &local->interfaces, list) {
+				if (!netif_running(sdata->dev))
+					continue;
+				if (sdata->vif.type ==
+				    NL80211_IFTYPE_STATION &&
+				    sdata->u.mgd.associated == bss) {
+					sdata->vif.bss_conf.dtim_period =
+						bss->dtim_period;
+					ieee80211_hw_config(local,
+						IEEE80211_CONF_CHANGE_PS);
+				}
+			}
+			mutex_unlock(&local->iflist_mtx);
+		}
 	}
 
 	/* set default value for buggy AP/no TIM element */
-- 
1.5.6.3

Re: [PATCH 1/1] mac80211: tell driver when dtim change detected

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 992 bytes --]

On Thu, 2010-01-21 at 07:51 -0800, wey-yi.w.guy@intel.com wrote:
> From: Wey-Yi Guy <wey-yi.w.guy@intel.com>
> 
> In current implementation, mac80211 send dtim_period update to driver during
> association, but if no NetworkManager or similar application perform scan
> operation, plus tim_ie is not part of probe response; mac80211 will not get
> beacon with dtim information later, then  mac80211 will not pass the 
> information to driver for update.

Good catch.

> Call ieee80211_hw_config() with IEEE80211_CONF_CHANGE_PS flag set to
> allow driver make correct dtim adjustment if dtim_period change detected.

I don't think this is the right place or the right thing to do though,
this seems like an implementation detail from iwlwifi -- it only (and
maybe only it) uses the DTIM interval for power saving.

However, the internal power saving stuff from mac80211 should also be
recalculated if this happens, and then we can probably call it a PS
change?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 801 bytes --]

Re: [PATCH 1/1] mac80211: tell driver when dtim change detected

[Guy, Wey-Yi]

Hi Johannes,

Agree, only iwlwifi use the dtim information now, but not sure it will
be always true. Yes, we should recalculate the internal power save when
it happen, I will make the changes and re-submit

On Thu, 2010-01-21 at 08:32 -0800, Johannes Berg wrote:
> On Thu, 2010-01-21 at 07:51 -0800, wey-yi.w.guy@intel.com wrote:
> > From: Wey-Yi Guy <wey-yi.w.guy@intel.com>
> > 
> > In current implementation, mac80211 send dtim_period update to driver during
> > association, but if no NetworkManager or similar application perform scan
> > operation, plus tim_ie is not part of probe response; mac80211 will not get
> > beacon with dtim information later, then  mac80211 will not pass the 
> > information to driver for update.
> 
> Good catch.
> 
> > Call ieee80211_hw_config() with IEEE80211_CONF_CHANGE_PS flag set to
> > allow driver make correct dtim adjustment if dtim_period change detected.
> 
> I don't think this is the right place or the right thing to do though,
> this seems like an implementation detail from iwlwifi -- it only (and
> maybe only it) uses the DTIM interval for power saving.
> 
> However, the internal power saving stuff from mac80211 should also be
> recalculated if this happens, and then we can probably call it a PS
> change?
> 
> johannes