[pkg-wpa-devel] Bug#659059: Bug#659059: hostapd: Clients diconnected every few minutes

Tue Feb 7 23:10:28 UTC 2012

tags 659059 + moreinfo
thanks

Hi

On Tuesday 07 February 2012, Marcus Meyer wrote:
> Package: hostapd
> Version: 1:0.6.10-2
> Severity: normal
> Tags: upstream
> 
> 
> I'm running hostapd in bridge mode with a TP-Link Model No. TL-WN951N
> wifi card.
> lspci says:
> 00:0a.0 Network controller: Atheros Communications Inc. AR5008 Wireless Network Adapter (rev 01)
> 
> There are a few devices connected through this wifi router. (Logitech
> Squuezeboxes, Smartphones, Netbooks, Laptops and so on...)
> 
> But only a Samsung Galaxy SII and a ASUS eeePC has problems with this
> configuration.

There are unfortunately a couple of potential reasons for this 
behaviour, so it would be nice to isolate this a bit more:

- please test "wmm_enabled=0" in your configuration, see #655129 which
  is unlikely to be fix{ed,able} before hostapd 1.0~ gets uploaded
  (at this moment it's not decided yet if that can happen in time for 
  wheezy).
  http://bugs.debian.org/655129
- can you determine which chipsets/ kernel versions (and/or drivers)
  are used on your failing devices (Samsung Galaxy SII, ASUS eeePC)?
- please try to disable powersaving on your failing clients. While 
  probably no (easily) possible on your smartphone, this should be as 
  easy as "iwconfig wlan0 power off" on your EeePC (assuming it runs 
  linux).
- you could try hostapd 1:0.7.3-2 from snapshots.d.o, which should 
  install on squeeze without further dependencies or configuration 
  changes - don't try to install any newer hostapd versions from 
  testing/ unstable on squeeze though!
  http://snapshot.debian.org/package/hostapd/1%3A0.7.3-2/
- while not as easy as upgrading hostapd on squeeze, it would be worth
  testing a newer kernel for your TL-WN951N; 3.2.1-2~bpo60+1 should be
  available from backports/squeeze.

> The main problem seems to me the reconnection.
> Here are a short log when the Galaxy SII is unable to connect or 
> better to say it lost the connection to the access point:
> 
> Feb  7 21:00:27 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: EAPOL-Key timeout
> Feb  7 21:00:27 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: sending 1/2 msg of Group Key Handshake
> Feb  7 21:00:28 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: EAPOL-Key timeout
> Feb  7 21:00:28 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: sending 1/2 msg of Group Key Handshake
> Feb  7 21:00:29 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: EAPOL-Key timeout
> Feb  7 21:00:29 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: sending 1/2 msg of Group Key Handshake
> Feb  7 21:00:30 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: EAPOL-Key timeout
> Feb  7 21:00:30 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.1X: unauthorizing port
> Feb  7 21:00:30 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: deauthenticated due to local deauth request
> Feb  7 21:11:07 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: authentication OK (open system)
> Feb  7 21:11:07 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-AUTHENTICATE.indication(44:a7:cf:c2:97:76, OPEN_SYSTEM)
> Feb  7 21:11:07 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DELETEKEYS.request(44:a7:cf:c2:97:76)
> Feb  7 21:11:07 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: authenticated
> Feb  7 21:15:27 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: Not in PTKINITDONE; skip Group Key update
> Feb  7 21:16:08 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: disassociated due to inactivity
> Feb  7 21:16:08 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DISASSOCIATE.indication(44:a7:cf:c2:97:76, 4)
> Feb  7 21:16:08 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DELETEKEYS.request(44:a7:cf:c2:97:76)
> Feb  7 21:16:09 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: deauthenticated due to inactivity
> Feb  7 21:16:09 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DEAUTHENTICATE.indication(44:a7:cf:c2:97:76, 2)
> Feb  7 21:16:09 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DELETEKEYS.request(44:a7:cf:c2:97:76)
> Feb  7 21:17:03 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: authentication OK (open system)
> Feb  7 21:17:03 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-AUTHENTICATE.indication(44:a7:cf:c2:97:76, OPEN_SYSTEM)
> Feb  7 21:17:03 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DELETEKEYS.request(44:a7:cf:c2:97:76)
> Feb  7 21:17:03 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: authenticated
> Feb  7 21:20:27 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 WPA: Not in PTKINITDONE; skip Group Key update
> Feb  7 21:22:04 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: disassociated due to inactivity
> Feb  7 21:22:04 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DISASSOCIATE.indication(44:a7:cf:c2:97:76, 4)
> Feb  7 21:22:04 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DELETEKEYS.request(44:a7:cf:c2:97:76)
> Feb  7 21:22:05 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 IEEE 802.11: deauthenticated due to inactivity
> Feb  7 21:22:05 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DEAUTHENTICATE.indication(44:a7:cf:c2:97:76, 2)
> Feb  7 21:22:05 router hostapd: wlan0: STA 44:a7:cf:c2:97:76 MLME: MLME-DELETEKEYS.request(44:a7:cf:c2:97:76)
> 
> After that I must disable and enable the wifi connection on the
> Galaxy.
> 
> With the eeePC I had the same problem, but with this netbook it 
> happend after a hour or so that the hole hostapd goes down.
> The daemon is still running but every device is unable to connect 
> and I must do a complete reboot of the router.
> 
> If you need more information or more logs let me know.
[...]

Personally I don't think hostapd is at fault here (unless we're talking
about WMM/ #655129), but rather suspect kernel and/or the hardware. The 
AR5008 is a very early 802.11n-draft chipset, developed and put on the 
market before the final 802.11n specification was released. As such it 
has several known hardware quirks, which match exactly your observed 
behaviour - some of them got worked around in more recent kernels, some
are still pending and/ or might never be completely fixable in 
software. So if you have a chance to test different hardware with 
hostapd (ideally Atheros AR92xx or newer, but RaLink rt2800pci - USB 
devices don't support powersaving clients -  is also said to have a 
good chance of working (in kernel >= 2.6.37)), the results would give 
an interesting comparison.

Regards
	Stefan Lippers-Hollmann
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.alioth.debian.org/pipermail/pkg-wpa-devel/attachments/20120208/50829fac/attachment.pgp>