Re: [ath5k-devel] Connecting BlackBerry to hotspot
On 9 June 2011 20:00, Bob Copeland m...@bobcopeland.com wrote: http://linuxwireless.org/en/users/Documentation/Bluetooth-coexistence From that page: Apart from AFS and channel skipping techniques Bluetooth coexistence is typically tested with bundled 802.11 and Bluetooth devices. This becomes more evident with 2-wire and 3-wire which relies on GPIO pins for signaling. In other words, the devices have to be in the same system so that the 802.11 device can tell the BT device to stop transmitting by asserting some signals on the shared lines. This is how I understand ath9k's bt coex to work, at least. Right, and the original poster: Today after installing compat-wireless modules with still no luck as a last resort i decided to turn of BT and take out BT dongles from hotspot's USB HUB. Guess even if AR5212 did have bluetooth coexistence support, there's no IO lines coming from the bluetooth device which are twiddling the btcoex GPIOs. Damn. :-) Adrian ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
On Thu, Jun 09, 2011 at 08:17:59PM +0800, Adrian Chadd wrote: Right, and the original poster: Today after installing compat-wireless modules with still no luck as a last resort i decided to turn of BT and take out BT dongles from hotspot's USB HUB. Oh, yeah, I did miss that completely, heh. I kept thinking the only BT was on the phone since the other phones didn't have issues. -- Bob Copeland %% www.bobcopeland.com ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
On 9 June 2011 04:34, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: Sorry for no new packet dumps yet but i wanted to tell you that i've found a possible reason. Bluetooth. Our hotspot has 'a lot of bluetooth going on' on it. Today after installing compat-wireless modules with still no luck as a last resort i decided to turn of BT and take out BT dongles from hotspot's USB HUB. Magically BB started showing pages on atheros. Doe's anyone know a solution for that (except not using WiFi or BT:) ) ? I guess I blamed ath5k driver without a reason, sorry for that. Hm, is there BT coexistance support for ath5k? Adrian ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
On Thu, Jun 09, 2011 at 07:08:43AM +0800, Adrian Chadd wrote: On 9 June 2011 04:34, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: Doe's anyone know a solution for that (except not using WiFi or BT:) ) ? I guess I blamed ath5k driver without a reason, sorry for that. Hm, is there BT coexistance support for ath5k? No, but my understanding is that BT coex support is only for co-located devices, so unless the phone is using ath5k it wouldn't help here. I guess one could try using a different channel for the AP or fiddling with the transmit power in hopes of making the channel avoidance on the BT device work better, but I don't know if there's anything more that could be done. -- Bob Copeland %% www.bobcopeland.com ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
On 9 June 2011 09:01, Bob Copeland m...@bobcopeland.com wrote: Hm, is there BT coexistance support for ath5k? No, but my understanding is that BT coex support is only for co-located devices, so unless the phone is using ath5k it wouldn't help here. From my understanding, bluetooth coexistance is so devices don't tx/rx interfere with each other. Ie, what you don't want is the wifi device RX'ing whilst the bluetooth device is TX'ing; you don't want the wifi device TX'ing whilst the bluetooth device is also TX'ing. I could be off though, I haven't yet written any btcoex code for FreeBSD (as I just don't currently have any hardware that supports it.) http://linuxwireless.org/en/users/Documentation/Bluetooth-coexistence I guess one could try using a different channel for the AP or fiddling with the transmit power in hopes of making the channel avoidance on the BT device work better, but I don't know if there's anything more that could be done. There's mention on that page of bluetooth devices being able to be told about BT channel ranges which they shouldn't use; you could then program the ap and BT device to use non-conflicting channel ranges. How this is done though is an exercise left to the reader. Also, I've no idea if the AR5212 era devices support bluetooth coexistance. I'd have to do some further digging into the historical archives to check. FreeBSD (which uses the older HAL based code) certainly doesn't have btcoex code for the chipsets. Adrian ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
W dniu 05.06.2011 21:13, Bob Copeland pisze: On Fri, May 27, 2011 at 5:27 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: Sorry, I pressed reply instead of reply all. W dniu 26.05.2011 14:56, Bob Copeland pisze: Please don't top-post. On Thu, May 26, 2011 at 2:48 AM, Piotr Kaczorek piotrkaczo...@bluetrace.euwrote: It seems to me i'ts ath5k issue. Since I started this thread I've tried few more chipsets: VIA (vt6655) - has problems with sending packets, sometimes they are 'broken'. Most images don't reach mobile phone correctly. Ralink (rt2561t - rt61pci driver) - works correctly (and also uses nl80211). Range is poor - maybe it's a matter of hardware i've bought. Prism (ISL3886 - p54 driver) - works corretly (and also uses nl80211). But thanks for pointing this out. It might be issue with ath5k- nl80211 cooperation, as well. Should I post a bug on kernel's bugzilla? I attach gziped packet dumps (hope it will work - I'm using mailing lists for the first time). There are two cap files - one from my SE P1i (on which requests work as they should), and one from BB Curve 8520 (on which I can associate but can't open website). Both contain association, and two requests to 'made up' url's. All requests should respond with the same content as my setup is quite untypical - requests are intercepted with iptables, and static website placed on hotspot is served in response to any request. There is also dnsmasq responding with my hotspot's IP to all DNS queries. Well, one thing missing in both of these captures is beacons which are kind of important since they have the traffic indication maps. The BB is sending the correct AID for PS-poll requests. I do see some successful PS-polls like packet #3089-3091 - where the AP delivers an arp reply. I'll keep looking for some discrepancy but it would help if we could get a capture with beacons. FWIW, I've attached the script I use to collect data -- not perfect and may need some tweaking for different setups, but it grabs a lot of useful data: - monitor mode captures from both the sta and the secondary card - dmesg, preferably with mac80211 debugging on - tracepoints from ath5k and mac80211 (must have enabled them in kernel config and have added ath5k_trace.py[1] to ~/.trace-cmd/plugins/) Once you start setup.sh then you need to associate with the AP in the usual way (script assumes they live on channel 11) and it grabs 5 minutes of data or until you ctrl-c it. [1] http://bobcopeland.com/srcs/ath5k_trace.py This script should be run on my hotspot or on 'external sniffer' (e.g. my laptop)? What does it require? I get: File ath5k_trace.py, line 1, in module import tracecmd ImportError: No module named tracecmd What setup.sh are you writing about? ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
On Mon, Jun 6, 2011 at 11:20 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: W dniu 05.06.2011 21:13, Bob Copeland pisze: On Fri, May 27, 2011 at 5:27 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: Sorry, I pressed reply instead of reply all. W dniu 26.05.2011 14:56, Bob Copeland pisze: Please don't top-post. On Thu, May 26, 2011 at 2:48 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: It seems to me i'ts ath5k issue. Since I started this thread I've tried few more chipsets: VIA (vt6655) - has problems with sending packets, sometimes they are 'broken'. Most images don't reach mobile phone correctly. Ralink (rt2561t - rt61pci driver) - works correctly (and also uses nl80211). Range is poor - maybe it's a matter of hardware i've bought. Prism (ISL3886 - p54 driver) - works corretly (and also uses nl80211). But thanks for pointing this out. It might be issue with ath5k- nl80211 cooperation, as well. Should I post a bug on kernel's bugzilla? I attach gziped packet dumps (hope it will work - I'm using mailing lists for the first time). There are two cap files - one from my SE P1i (on which requests work as they should), and one from BB Curve 8520 (on which I can associate but can't open website). Both contain association, and two requests to 'made up' url's. All requests should respond with the same content as my setup is quite untypical - requests are intercepted with iptables, and static website placed on hotspot is served in response to any request. There is also dnsmasq responding with my hotspot's IP to all DNS queries. Well, one thing missing in both of these captures is beacons which are kind of important since they have the traffic indication maps. The BB is sending the correct AID for PS-poll requests. I do see some successful PS-polls like packet #3089-3091 - where the AP delivers an arp reply. I'll keep looking for some discrepancy but it would help if we could get a capture with beacons. FWIW, I've attached the script I use to collect data -- not perfect and may need some tweaking for different setups, but it grabs a lot of useful data: - monitor mode captures from both the sta and the secondary card - dmesg, preferably with mac80211 debugging on - tracepoints from ath5k and mac80211 (must have enabled them in kernel config and have added ath5k_trace.py[1] to ~/.trace-cmd/plugins/) Once you start setup.sh then you need to associate with the AP in the usual way (script assumes they live on channel 11) and it grabs 5 minutes of data or until you ctrl-c it. [1] http://bobcopeland.com/srcs/ath5k_trace.py This script should be run on my hotspot or on 'external sniffer' (e.g. my laptop)? What does it require? I get: File ath5k_trace.py, line 1, in module import tracecmd ImportError: No module named tracecmd What setup.sh are you writing about? setup.sh was attached to the email. I run it on my laptop. I mostly use this script for debugging STA issues, so in my case I get a capture from ath5k acting as a STA, and a USB ar9170 on the same machine to act as the external sniffer. Since you are interested in ath5k as an AP, some things would have to change (e.g., I wouldn't bother with a monitor mode capture on the AP, but it might be worthwhile to get the driver trace) -- you could do something like connect both the laptop and the hotspot via the wired interface and start/stop the driver trace via ssh. ath5k_trace.py needs to be added to ~/.trace-cmd/plugins/ and run via trace-cmd, not directly (as invoked by setup.sh). It extracts data from the ath5k tracepoints and writes out a pcap file and some general debugging info. YMMV - this is just an example of the kind of data I collect for my own debugging. -- Bob Copeland %% www.bobcopeland.com ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
On Fri, May 27, 2011 at 5:27 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: Sorry, I pressed reply instead of reply all. W dniu 26.05.2011 14:56, Bob Copeland pisze: Please don't top-post. On Thu, May 26, 2011 at 2:48 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: It seems to me i'ts ath5k issue. Since I started this thread I've tried few more chipsets: VIA (vt6655) - has problems with sending packets, sometimes they are 'broken'. Most images don't reach mobile phone correctly. Ralink (rt2561t - rt61pci driver) - works correctly (and also uses nl80211). Range is poor - maybe it's a matter of hardware i've bought. Prism (ISL3886 - p54 driver) - works corretly (and also uses nl80211). But thanks for pointing this out. It might be issue with ath5k- nl80211 cooperation, as well. Should I post a bug on kernel's bugzilla? I attach gziped packet dumps (hope it will work - I'm using mailing lists for the first time). There are two cap files - one from my SE P1i (on which requests work as they should), and one from BB Curve 8520 (on which I can associate but can't open website). Both contain association, and two requests to 'made up' url's. All requests should respond with the same content as my setup is quite untypical - requests are intercepted with iptables, and static website placed on hotspot is served in response to any request. There is also dnsmasq responding with my hotspot's IP to all DNS queries. Well, one thing missing in both of these captures is beacons which are kind of important since they have the traffic indication maps. The BB is sending the correct AID for PS-poll requests. I do see some successful PS-polls like packet #3089-3091 - where the AP delivers an arp reply. I'll keep looking for some discrepancy but it would help if we could get a capture with beacons. FWIW, I've attached the script I use to collect data -- not perfect and may need some tweaking for different setups, but it grabs a lot of useful data: - monitor mode captures from both the sta and the secondary card - dmesg, preferably with mac80211 debugging on - tracepoints from ath5k and mac80211 (must have enabled them in kernel config and have added ath5k_trace.py[1] to ~/.trace-cmd/plugins/) Once you start setup.sh then you need to associate with the AP in the usual way (script assumes they live on channel 11) and it grabs 5 minutes of data or until you ctrl-c it. [1] http://bobcopeland.com/srcs/ath5k_trace.py -- Bob Copeland %% www.bobcopeland.com setup.sh Description: Bourne shell script ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
It seems to me i'ts ath5k issue. Since I started this thread I've tried few more chipsets: VIA (vt6655) - has problems with sending packets, sometimes they are 'broken'. Most images don't reach mobile phone correctly. Ralink (rt2561t - rt61pci driver) - works correctly (and also uses nl80211). Range is poor - maybe it's a matter of hardware i've bought. Prism (ISL3886 - p54 driver) - works corretly (and also uses nl80211). But thanks for pointing this out. It might be issue with ath5k- nl80211 cooperation, as well. Should I post a bug on kernel's bugzilla? W dniu 17.05.2011 07:23, Adrian Chadd pisze: It may also be the different 802.11 layer; it doesn't have to be ath5k specific. I know there's an outstanding bug with FreeBSD (which uses net80211, like madwifi does) with power save queue management timers being off-whack and that appears like this, also I found this comment in the source: /* * NB: We used to deauth the station but it turns out * the Blackberry Curve 8230 (and perhaps other devices) * sometimes send the wrong AID when WME is negotiated. * Being more lenient here seems ok as we already check * the station is associated and we only return frames * queued for the station (i.e. we don't use the AID). */ That's in hostap_recv_pspoll(), which is handling the power-save polling stuff. Good luck! Adrian On 12 May 2011 05:14, Piotr Kaczorekpiotrkaczo...@bluetrace.eu wrote: W dniu 11.05.2011 06:05, Bob Copeland pisze: On Tue, May 10, 2011 at 9:42 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: Is there any known problem with BlackBerry phones? I've set up hotspot (ath5k + hostapd). Any device can connect to it and use it (SE P1i, few laptops, iPhone) except BlackBerry (Curve 8520) phone. It connects to Wifi but i cannot access any website (it displays can't find server message). Maybe it's an issue with power management My guess is this. Probably blackberry is going to sleep and missing (or ath5k is not sending) the beacons along with all of the buffered traffic. (http://thread.gmane.org/gmane.os.freebsd.current/110707)? A timeout? I have no idea where to look further for a solution. Try to get a packet capture with a second device, and see if the AP is sending when the station is in power save mode. Thanks for your suggestions, i'll probably try that if i will stay with ath5k. Confirmed this is definately a matter of ath5k driver. I tried madwifi and BlackBerry worked. Of course i've got the Stuck beacon problem on madwifi. Now I'll have to make a decision - try to fix ath5k, try to find working madwifi or leave atheros at all and try different wifi card/chipset. ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
Please don't top-post. On Thu, May 26, 2011 at 2:48 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: It seems to me i'ts ath5k issue. Since I started this thread I've tried few more chipsets: VIA (vt6655) - has problems with sending packets, sometimes they are 'broken'. Most images don't reach mobile phone correctly. Ralink (rt2561t - rt61pci driver) - works correctly (and also uses nl80211). Range is poor - maybe it's a matter of hardware i've bought. Prism (ISL3886 - p54 driver) - works corretly (and also uses nl80211). But thanks for pointing this out. It might be issue with ath5k- nl80211 cooperation, as well. Should I post a bug on kernel's bugzilla? It's fine to post here -- but you most likely need to get a packet dump before anyone will be able to help (not all of us have Blackberrys to test with). -- Bob Copeland %% www.bobcopeland.com ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
It may also be the different 802.11 layer; it doesn't have to be ath5k specific. I know there's an outstanding bug with FreeBSD (which uses net80211, like madwifi does) with power save queue management timers being off-whack and that appears like this, also I found this comment in the source: /* * NB: We used to deauth the station but it turns out * the Blackberry Curve 8230 (and perhaps other devices) * sometimes send the wrong AID when WME is negotiated. * Being more lenient here seems ok as we already check * the station is associated and we only return frames * queued for the station (i.e. we don't use the AID). */ That's in hostap_recv_pspoll(), which is handling the power-save polling stuff. Good luck! Adrian On 12 May 2011 05:14, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: W dniu 11.05.2011 06:05, Bob Copeland pisze: On Tue, May 10, 2011 at 9:42 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: Is there any known problem with BlackBerry phones? I've set up hotspot (ath5k + hostapd). Any device can connect to it and use it (SE P1i, few laptops, iPhone) except BlackBerry (Curve 8520) phone. It connects to Wifi but i cannot access any website (it displays can't find server message). Maybe it's an issue with power management My guess is this. Probably blackberry is going to sleep and missing (or ath5k is not sending) the beacons along with all of the buffered traffic. (http://thread.gmane.org/gmane.os.freebsd.current/110707)? A timeout? I have no idea where to look further for a solution. Try to get a packet capture with a second device, and see if the AP is sending when the station is in power save mode. Thanks for your suggestions, i'll probably try that if i will stay with ath5k. Confirmed this is definately a matter of ath5k driver. I tried madwifi and BlackBerry worked. Of course i've got the Stuck beacon problem on madwifi. Now I'll have to make a decision - try to fix ath5k, try to find working madwifi or leave atheros at all and try different wifi card/chipset. ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
[ath5k-devel] Connecting BlackBerry to hotspot
Just joined this list an hour ago ... The 2 packets you show are a PING REQUEST (TYPE=8) to your hotspot, and a PING REPLY (TYPE=0) from your hotspot! AND, they have the same ICMP ID # so the responce is from that request and not a previous one. smime.p7s Description: S/MIME Cryptographic Signature ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
Re: [ath5k-devel] Connecting BlackBerry to hotspot
W dniu 11.05.2011 06:05, Bob Copeland pisze: On Tue, May 10, 2011 at 9:42 AM, Piotr Kaczorek piotrkaczo...@bluetrace.eu wrote: Is there any known problem with BlackBerry phones? I've set up hotspot (ath5k + hostapd). Any device can connect to it and use it (SE P1i, few laptops, iPhone) except BlackBerry (Curve 8520) phone. It connects to Wifi but i cannot access any website (it displays can't find server message). Maybe it's an issue with power management My guess is this. Probably blackberry is going to sleep and missing (or ath5k is not sending) the beacons along with all of the buffered traffic. (http://thread.gmane.org/gmane.os.freebsd.current/110707)? A timeout? I have no idea where to look further for a solution. Try to get a packet capture with a second device, and see if the AP is sending when the station is in power save mode. Thanks for your suggestions, i'll probably try that if i will stay with ath5k. Confirmed this is definately a matter of ath5k driver. I tried madwifi and BlackBerry worked. Of course i've got the Stuck beacon problem on madwifi. Now I'll have to make a decision - try to fix ath5k, try to find working madwifi or leave atheros at all and try different wifi card/chipset. ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
[ath5k-devel] Connecting BlackBerry to hotspot
Is there any known problem with BlackBerry phones? I've set up hotspot (ath5k + hostapd). Any device can connect to it and use it (SE P1i, few laptops, iPhone) except BlackBerry (Curve 8520) phone. It connects to Wifi but i cannot access any website (it displays can't find server message). I tried to debug packets sent on wifi interface - it looks like DNS packets don't reach hotspot at all (well they do but thats 1 of 100). I can see ARP 'who has' and ARP 'is at' packets (sniffing with scapy) but nothing else. Maybe it's a matter of some timeout? I tried also to ping my hotspot from BB - when i ping it as 'WiFi gateway' then one request reaches hotspot and returns (in about 1600ms). When i try to ping hotspot with it's IP no request reaches back BB (however they return from hotspot mmediately) May 10 15:00:13 localhost kernel: [ 7909.164681] IN=wlan0 OUT= MAC=00:0e:8e:25:a8:fc:30:69:4b:04:6a:7e:08:00 SRC=192.168.100.146 DST=192.168.100.1 LEN=60 TOS=0x00 PREC=0x00 TTL=128 ID=1236 DF PROTO=ICMP TYPE=8 CODE=0 ID=2910 SEQ=0 May 10 15:00:13 localhost kernel: [ 7909.165014] xxxOUTxxxIN= OUT=wlan0 SRC=192.168.100.1 DST=192.168.100.146 LEN=60 TOS=0x00 PREC=0x00 TTL=64 ID=38548 PROTO=ICMP TYPE=0 CODE=0 ID=2910 SEQ=0 Maybe it's an issue with power management (http://thread.gmane.org/gmane.os.freebsd.current/110707)? A timeout? I have no idea where to look further for a solution. I'm using ath5k from kernel 2.6.38 hostapd 0.7.3 (driver nl80211) dnsmasq 2.45 ___ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel