iwl3945 regression: associates but no data is transfered

iwl3945 regression: associates but no data is transfered

[Kalle Valo]

Hello,

I upgraded to the latest wireless-testing and my iwl3945 is now
broken. It can associate but no data is transfered after that. Tried
with open and wep modes. With dhcp I cannot get ip and with static ip
ping to the gateway doesn't work.

Any hints?

-- 
Kalle Valo

Re: iwl3945 regression: associates but no data is transfered

[John W. Linville]

On Thu, Jan 22, 2009 at 08:26:14PM +0200, Kalle Valo wrote:
> Hello,
> 
> I upgraded to the latest wireless-testing and my iwl3945 is now
> broken. It can associate but no data is transfered after that. Tried
> with open and wep modes. With dhcp I cannot get ip and with static ip
> ping to the gateway doesn't work.
> 
> Any hints?

Lots of iwlwifi stuff merged yesterday...

Chatre, Reinette (1):
      iwl3945: add debugging for wrong command queue

Helmut Schaa (2):
      iwl3945: fix some warnings when compiled without debug
      iwlagn: fix hw-rfkill while the interface is down

Jay Sternberg (4):
      iwlwifi: correct Kconfig to prevent following entries from not indenting
      iwlwifi: remove static from 5000 structures
      iwlwifi: add recognition of Intel WiFi Link 6000 and 6050 Series
      iwlwifi: add recognition of Intel WiFi Link 100 Series

Reinette Chatre (1):
      iwlwifi: return NETDEV_TX_OK from _tx ops

Samuel Ortiz (3):
      iwl3945: Use iwl_txq_update_write_ptr
      iwlwifi: Add TFD library operations
      iwl3945: Use iwl-hcmd host command routines

Winkler, Tomas (6):
      iwlwifi: make iwl-power.c more readable
      iwlwifi: fix iwl_mac_set_key and iwl3945_mac_set_key
      iwlwifi: kill iwl3945_scan_cancel and iwl3945_scan_cancel_timeout
      iwlwifi: kill scan39
      iwlwifi: remove unused or twice defined members in iwl_priv
      iwlwifi: eliminate power_data_39.

-- 
John W. Linville		Someday the world will need a hero, and you
linville@tuxdriver.com			might be all we have.  Be ready.

Re: iwl3945 regression: associates but no data is transfered

[Sedat Dilek]

Hi Kalle,

I noticed the same regression here with iwl3945 and static-IP with
wl.git master 2009-01-21.

I tried to apply the iwlwifi-updates from 20-Jan-2009 (5 patches,
without the iwl3945-rfkill one, see below).
As far as I can see the patchset as of 19-Jan-2009 (15 patches) seem
to be all included in latest master.

There were two patches specially for 2.6.29 [1], but they were not
applicable in my case, so I took the ones from the older patchset.
But no luck here.

Johannes and Mohamed are discussing mac80211 disassociation on [2].

Hope these infos help.

Kind Regards,
Sedat

[1] http://marc.info/?l=linux-wireless&m=123256418517083&w=2
[2] http://marc.info/?l=linux-wireless&m=123264308904962&w=2


----- BEGIN series -----

# 00xx-series: All patches against wireless-testing (master-2009-01-21)
0000-patches-against-wl-v2.6.29-rc2-11925-g88e5942
#
# 01xx-series: Patches from <kernel.org>
0100-patch-2.6.29-rc2-git1
#
# 02xx-series: Patches from wl-mailing-list
0200-0_5-iwlwifi-driver-updates-01_20_2009.patch
0201-1_5-iwlwifi-fix-rs_get_rate-WARN_ON.patch
0202-2_5-iwlwifi-allow-user-to-set-max-rate.patch
0203-3_5-iwlagn-fix-agn-rate-scaling.patch
# Disabled - included in wl (master-2009-01-21)
#0204-4_5-iwl3945-Use-iwl-rfkill-v2.patch
0205-5_5-iwl3945-fix-deep-sleep-when-removing-the-driver.patch

----- END series -----

On Thu, Jan 22, 2009 at 7:26 PM, Kalle Valo <kalle.valo@iki.fi> wrote:
> Hello,
>
> I upgraded to the latest wireless-testing and my iwl3945 is now
> broken. It can associate but no data is transfered after that. Tried
> with open and wep modes. With dhcp I cannot get ip and with static ip
> ping to the gateway doesn't work.
>
> Any hints?
>
> --
> Kalle Valo
> --
> 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
>

Re: iwl3945 regression: associates but no data is transfered

[samuel]

On Thu, 22 Jan 2009 13:34:48 -0500, "John W. Linville"

<linville@tuxdriver.com> wrote:

> On Thu, Jan 22, 2009 at 08:26:14PM +0200, Kalle Valo wrote:

>> Hello,

>> 

>> I upgraded to the latest wireless-testing and my iwl3945 is now

>> broken. It can associate but no data is transfered after that. Tried

>> with open and wep modes. With dhcp I cannot get ip and with static ip

>> ping to the gateway doesn't work.

>> 

>> Any hints?

> 

> Lots of iwlwifi stuff merged yesterday...

The bad commit seems to be e32b1b9a33759e8a83ac566c4c43f23ed5d6343b, which

is not iwlwifi related.

Even though it's an mlme commit, it doesnt seem to affect other drivers (at

least b43 and ath5k).

I'll come up soon with a proper fix, but for now iwl3945 users should

revert that commit locally.



Cheers,

Samuel.



> 

> Chatre, Reinette (1):

>       iwl3945: add debugging for wrong command queue

> 

> Helmut Schaa (2):

>       iwl3945: fix some warnings when compiled without debug

>       iwlagn: fix hw-rfkill while the interface is down

> 

> Jay Sternberg (4):

>       iwlwifi: correct Kconfig to prevent following entries from not

>       indenting

>       iwlwifi: remove static from 5000 structures

>       iwlwifi: add recognition of Intel WiFi Link 6000 and 6050 Series

>       iwlwifi: add recognition of Intel WiFi Link 100 Series

> 

> Reinette Chatre (1):

>       iwlwifi: return NETDEV_TX_OK from _tx ops

> 

> Samuel Ortiz (3):

>       iwl3945: Use iwl_txq_update_write_ptr

>       iwlwifi: Add TFD library operations

>       iwl3945: Use iwl-hcmd host command routines

> 

> Winkler, Tomas (6):

>       iwlwifi: make iwl-power.c more readable

>       iwlwifi: fix iwl_mac_set_key and iwl3945_mac_set_key

>       iwlwifi: kill iwl3945_scan_cancel and iwl3945_scan_cancel_timeout

>       iwlwifi: kill scan39

>       iwlwifi: remove unused or twice defined members in iwl_priv

>       iwlwifi: eliminate power_data_39.

Re: iwl3945 regression: associates but no data is transfered

[Larry Finger]

samuel@sortiz.org wrote:

> The bad commit seems to be e32b1b9a33759e8a83ac566c4c43f23ed5d6343b, which
> is not iwlwifi related.
> Even though it's an mlme commit, it doesnt seem to affect other drivers (at
> least b43 and ath5k).
> I'll come up soon with a proper fix, but for now iwl3945 users should
> revert that commit locally.

Something between 2.6.29-rc1 and -rc2 killed b43 and p54usb on my WPA2 network. My rtl8187
devices work fine.

I'm in the process of bisecting right now. Only 1500 commits to go.

Larry

Re: iwl3945 regression: associates but no data is transfered

[Sedat Dilek]

Hi,

BTW, v2 of Alina's fix is *NOT* fixing the problem (not on my machine).

Kind Regards,
Sedat

[1] [PATCH v2] mac80211: Fixed BSSID handling revisited
http://marc.info/?l=linux-wireless&m=123259183328417&w=2

On Fri, Jan 23, 2009 at 12:41 AM,  <samuel@sortiz.org> wrote:
>
> On Thu, 22 Jan 2009 13:34:48 -0500, "John W. Linville"
> <linville@tuxdriver.com> wrote:
>> On Thu, Jan 22, 2009 at 08:26:14PM +0200, Kalle Valo wrote:
>>> Hello,
>>>
>>> I upgraded to the latest wireless-testing and my iwl3945 is now
>>> broken. It can associate but no data is transfered after that. Tried
>>> with open and wep modes. With dhcp I cannot get ip and with static ip
>>> ping to the gateway doesn't work.
>>>
>>> Any hints?
>>
>> Lots of iwlwifi stuff merged yesterday...
> The bad commit seems to be e32b1b9a33759e8a83ac566c4c43f23ed5d6343b, which
> is not iwlwifi related.
> Even though it's an mlme commit, it doesnt seem to affect other drivers (at
> least b43 and ath5k).
> I'll come up soon with a proper fix, but for now iwl3945 users should
> revert that commit locally.
>
> Cheers,
> Samuel.
>
>>
>> Chatre, Reinette (1):
>>       iwl3945: add debugging for wrong command queue
>>
>> Helmut Schaa (2):
>>       iwl3945: fix some warnings when compiled without debug
>>       iwlagn: fix hw-rfkill while the interface is down
>>
>> Jay Sternberg (4):
>>       iwlwifi: correct Kconfig to prevent following entries from not
>>       indenting
>>       iwlwifi: remove static from 5000 structures
>>       iwlwifi: add recognition of Intel WiFi Link 6000 and 6050 Series
>>       iwlwifi: add recognition of Intel WiFi Link 100 Series
>>
>> Reinette Chatre (1):
>>       iwlwifi: return NETDEV_TX_OK from _tx ops
>>
>> Samuel Ortiz (3):
>>       iwl3945: Use iwl_txq_update_write_ptr
>>       iwlwifi: Add TFD library operations
>>       iwl3945: Use iwl-hcmd host command routines
>>
>> Winkler, Tomas (6):
>>       iwlwifi: make iwl-power.c more readable
>>       iwlwifi: fix iwl_mac_set_key and iwl3945_mac_set_key
>>       iwlwifi: kill iwl3945_scan_cancel and iwl3945_scan_cancel_timeout
>>       iwlwifi: kill scan39
>>       iwlwifi: remove unused or twice defined members in iwl_priv
>>       iwlwifi: eliminate power_data_39.
> --
> 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
>

Re: iwl3945 regression: associates but no data is transfered

[samuel]

On Thu, 22 Jan 2009 17:55:39 -0600, Larry Finger

<Larry.Finger@lwfinger.net>

wrote:

> samuel@sortiz.org wrote:

> 

>> The bad commit seems to be e32b1b9a33759e8a83ac566c4c43f23ed5d6343b,

>> which

>> is not iwlwifi related.

>> Even though it's an mlme commit, it doesnt seem to affect other drivers

>> (at

>> least b43 and ath5k).

>> I'll come up soon with a proper fix, but for now iwl3945 users should

>> revert that commit locally.

> 

> Something between 2.6.29-rc1 and -rc2 killed b43 and p54usb on my WPA2

> network. My rtl8187

> devices work fine.

> 

> I'm in the process of bisecting right now. Only 1500 commits to go.

Have you tried reverting that commit ?

I actually found that it does break ath5k as well...



Cheers,

Samuel.

Re: iwl3945 regression: associates but no data is transfered

[Larry Finger]

samuel@sortiz.org wrote:
> Have you tried reverting that commit ?
> I actually found that it does break ath5k as well...

No - as I was bisecting, I thought I'd keep going; however, that sounds like a good idea
that might save some time.

I'll report ASAP.

Larry

Re: iwl3945 regression: associates but no data is transfered

[Sedat Dilek]

[-- Attachment #1: Type: text/plain, Size: 2612 bytes --]

Thanks Samuel, this works.

-SD

----- BEGIN series -----

# 00xx-series: All patches against wireless-testing (master-2009-01-21)
0000-patches-against-wl-v2.6.29-rc2-11925-g88e5942
#
# 01xx-series: Patches from <kernel.org>
0100-patch-2.6.29-rc2-git1
#
# 02xx-series: Patches from wl-mailing-list
0200-Revert-mac80211-Fixed-BSSID-handling-revisited.patch

----- END series -----

On Fri, Jan 23, 2009 at 12:41 AM,  <samuel@sortiz.org> wrote:
>
> On Thu, 22 Jan 2009 13:34:48 -0500, "John W. Linville"
> <linville@tuxdriver.com> wrote:
>> On Thu, Jan 22, 2009 at 08:26:14PM +0200, Kalle Valo wrote:
>>> Hello,
>>>
>>> I upgraded to the latest wireless-testing and my iwl3945 is now
>>> broken. It can associate but no data is transfered after that. Tried
>>> with open and wep modes. With dhcp I cannot get ip and with static ip
>>> ping to the gateway doesn't work.
>>>
>>> Any hints?
>>
>> Lots of iwlwifi stuff merged yesterday...
> The bad commit seems to be e32b1b9a33759e8a83ac566c4c43f23ed5d6343b, which
> is not iwlwifi related.
> Even though it's an mlme commit, it doesnt seem to affect other drivers (at
> least b43 and ath5k).
> I'll come up soon with a proper fix, but for now iwl3945 users should
> revert that commit locally.
>
> Cheers,
> Samuel.
>
>>
>> Chatre, Reinette (1):
>>       iwl3945: add debugging for wrong command queue
>>
>> Helmut Schaa (2):
>>       iwl3945: fix some warnings when compiled without debug
>>       iwlagn: fix hw-rfkill while the interface is down
>>
>> Jay Sternberg (4):
>>       iwlwifi: correct Kconfig to prevent following entries from not
>>       indenting
>>       iwlwifi: remove static from 5000 structures
>>       iwlwifi: add recognition of Intel WiFi Link 6000 and 6050 Series
>>       iwlwifi: add recognition of Intel WiFi Link 100 Series
>>
>> Reinette Chatre (1):
>>       iwlwifi: return NETDEV_TX_OK from _tx ops
>>
>> Samuel Ortiz (3):
>>       iwl3945: Use iwl_txq_update_write_ptr
>>       iwlwifi: Add TFD library operations
>>       iwl3945: Use iwl-hcmd host command routines
>>
>> Winkler, Tomas (6):
>>       iwlwifi: make iwl-power.c more readable
>>       iwlwifi: fix iwl_mac_set_key and iwl3945_mac_set_key
>>       iwlwifi: kill iwl3945_scan_cancel and iwl3945_scan_cancel_timeout
>>       iwlwifi: kill scan39
>>       iwlwifi: remove unused or twice defined members in iwl_priv
>>       iwlwifi: eliminate power_data_39.
> --
> 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
>

[-- Attachment #2: 0200-Revert-mac80211-Fixed-BSSID-handling-revisited.patch --]
[-- Type: text/x-diff, Size: 4880 bytes --]

diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index ca9e415..b9e4b93 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -1614,7 +1614,6 @@ static int ieee80211_sta_join_ibss(struct ieee80211_sub_if_data *sdata,
 
 	ieee80211_sta_def_wmm_params(sdata, bss);
 
-	ifsta->flags |= IEEE80211_STA_PREV_BSSID_SET;
 	ifsta->state = IEEE80211_STA_MLME_IBSS_JOINED;
 	mod_timer(&ifsta->timer, jiffies + IEEE80211_IBSS_MERGE_INTERVAL);
 
@@ -1758,8 +1757,7 @@ static void ieee80211_rx_bss_info(struct ieee80211_sub_if_data *sdata,
 		       (unsigned long long)(rx_timestamp - beacon_timestamp),
 		       jiffies);
 #endif /* CONFIG_MAC80211_IBSS_DEBUG */
-		if (beacon_timestamp > rx_timestamp &&
-		    memcmp(sdata->u.sta.bssid, mgmt->bssid, ETH_ALEN) != 0) {
+		if (beacon_timestamp > rx_timestamp) {
 #ifdef CONFIG_MAC80211_IBSS_DEBUG
 			printk(KERN_DEBUG "%s: beacon TSF higher than "
 			       "local TSF - IBSS merge with BSSID %pM\n",
@@ -2179,18 +2177,19 @@ static int ieee80211_sta_create_ibss(struct ieee80211_sub_if_data *sdata,
 	int i;
 	int ret;
 
-	if (sdata->u.sta.flags & IEEE80211_STA_BSSID_SET) {
-		memcpy(bssid, ifsta->bssid, ETH_ALEN);
-	} else {
-		/* Generate random, not broadcast, locally administered BSSID. Mix in
-		 * own MAC address to make sure that devices that do not have proper
-		 * random number generator get different BSSID. */
-		get_random_bytes(bssid, ETH_ALEN);
-		for (i = 0; i < ETH_ALEN; i++)
-			bssid[i] ^= sdata->dev->dev_addr[i];
-		bssid[0] &= ~0x01;
-		bssid[0] |= 0x02;
-	}
+#if 0
+	/* Easier testing, use fixed BSSID. */
+	memset(bssid, 0xfe, ETH_ALEN);
+#else
+	/* Generate random, not broadcast, locally administered BSSID. Mix in
+	 * own MAC address to make sure that devices that do not have proper
+	 * random number generator get different BSSID. */
+	get_random_bytes(bssid, ETH_ALEN);
+	for (i = 0; i < ETH_ALEN; i++)
+		bssid[i] ^= sdata->dev->dev_addr[i];
+	bssid[0] &= ~0x01;
+	bssid[0] |= 0x02;
+#endif
 
 	printk(KERN_DEBUG "%s: Creating new IBSS network, BSSID %pM\n",
 	       sdata->dev->name, bssid);
@@ -2251,9 +2250,6 @@ static int ieee80211_sta_find_ibss(struct ieee80211_sub_if_data *sdata,
 		    memcmp(ifsta->ssid, bss->ssid, bss->ssid_len) != 0
 		    || !(bss->capability & WLAN_CAPABILITY_IBSS))
 			continue;
-		if ((ifsta->flags & IEEE80211_STA_BSSID_SET) &&
-		    memcmp(ifsta->bssid, bss->bssid, ETH_ALEN) != 0)
-			continue;
 #ifdef CONFIG_MAC80211_IBSS_DEBUG
 		printk(KERN_DEBUG "   bssid=%pM found\n", bss->bssid);
 #endif /* CONFIG_MAC80211_IBSS_DEBUG */
@@ -2270,9 +2266,7 @@ static int ieee80211_sta_find_ibss(struct ieee80211_sub_if_data *sdata,
 		       "%pM\n", bssid, ifsta->bssid);
 #endif /* CONFIG_MAC80211_IBSS_DEBUG */
 
-	if (found &&
-	    ((!(ifsta->flags & IEEE80211_STA_PREV_BSSID_SET)) ||
-	     memcmp(ifsta->bssid, bssid, ETH_ALEN) != 0)) {
+	if (found && memcmp(ifsta->bssid, bssid, ETH_ALEN) != 0) {
 		int ret;
 		int search_freq;
 
@@ -2610,16 +2604,16 @@ int ieee80211_sta_set_ssid(struct ieee80211_sub_if_data *sdata, char *ssid, size
 		memset(ifsta->ssid, 0, sizeof(ifsta->ssid));
 		memcpy(ifsta->ssid, ssid, len);
 		ifsta->ssid_len = len;
+		ifsta->flags &= ~IEEE80211_STA_PREV_BSSID_SET;
 	}
 
-	ifsta->flags &= ~IEEE80211_STA_PREV_BSSID_SET;
-
 	if (len)
 		ifsta->flags |= IEEE80211_STA_SSID_SET;
 	else
 		ifsta->flags &= ~IEEE80211_STA_SSID_SET;
 
-	if (sdata->vif.type == NL80211_IFTYPE_ADHOC) {
+	if (sdata->vif.type == NL80211_IFTYPE_ADHOC &&
+	    !(ifsta->flags & IEEE80211_STA_BSSID_SET)) {
 		ifsta->ibss_join_req = jiffies;
 		ifsta->state = IEEE80211_STA_MLME_IBSS_SEARCH;
 		return ieee80211_sta_find_ibss(sdata, ifsta);
@@ -2639,18 +2633,36 @@ int ieee80211_sta_get_ssid(struct ieee80211_sub_if_data *sdata, char *ssid, size
 int ieee80211_sta_set_bssid(struct ieee80211_sub_if_data *sdata, u8 *bssid)
 {
 	struct ieee80211_if_sta *ifsta;
+	int res;
+	bool valid;
 
 	ifsta = &sdata->u.sta;
+	valid = is_valid_ether_addr(bssid);
+
+	if (memcmp(ifsta->bssid, bssid, ETH_ALEN) != 0) {
+		if(valid)
+			memcpy(ifsta->bssid, bssid, ETH_ALEN);
+		else
+			memset(ifsta->bssid, 0, ETH_ALEN);
+		res = 0;
+		/*
+		 * Hack! See also ieee80211_sta_set_ssid.
+		 */
+		if (netif_running(sdata->dev))
+			res = ieee80211_if_config(sdata, IEEE80211_IFCC_BSSID);
+		if (res) {
+			printk(KERN_DEBUG "%s: Failed to config new BSSID to "
+			       "the low-level driver\n", sdata->dev->name);
+			return res;
+		}
+	}
 
-	if (is_valid_ether_addr(bssid)) {
-		memcpy(ifsta->bssid, bssid, ETH_ALEN);
+	if (valid)
 		ifsta->flags |= IEEE80211_STA_BSSID_SET;
-	} else {
-		memset(ifsta->bssid, 0, ETH_ALEN);
+	else
 		ifsta->flags &= ~IEEE80211_STA_BSSID_SET;
-	}
 
-	return ieee80211_sta_set_ssid(sdata, ifsta->ssid, ifsta->ssid_len);
+	return 0;
 }
 
 int ieee80211_sta_set_extra_ie(struct ieee80211_sub_if_data *sdata, char *ie, size_t len)

Re: iwl3945 regression: associates but no data is transfered

[Harvey Harrison]

On Thu, 2009-01-22 at 17:55 -0600, Larry Finger wrote:
> samuel@sortiz.org wrote:
> 
> > The bad commit seems to be e32b1b9a33759e8a83ac566c4c43f23ed5d6343b, which
> > is not iwlwifi related.
> > Even though it's an mlme commit, it doesnt seem to affect other drivers (at
> > least b43 and ath5k).
> > I'll come up soon with a proper fix, but for now iwl3945 users should
> > revert that commit locally.
> 
> Something between 2.6.29-rc1 and -rc2 killed b43 and p54usb on my WPA2 network. My rtl8187
> devices work fine.
> 
> I'm in the process of bisecting right now. Only 1500 commits to go.

I wish I'd said something sooner, I thought it was just the university's network
being particularly crappy causing wpa2 to be so flaky since going to -rc2. (b43)

Harvey