[ath9k-devel] Issues with ar5418

[ath9k-devel] Issues with ar5418

[Jacob]

I have been running the latest wireless-testing for about 24 hours and generally
the card has been very stable. I've tested it with and without WPA/1.

There are a few issues though,  I also had the same issues with 2.6.27.1...

1) Whenever i run iperf, I get this many of these in the logs:
ath_tx_prepare: TX queue: 1 is full, depth: 492
ath_tx_prepare: TX queue: 1 is full, depth: 492

2) There is a big difference in speed upstream and downstream throughout, the
figures below are fairly constant:

Client connecting to XX, TCP port 5001
TCP window size: 16.0 KByte (default)
------------------------------------------------------------
[  5] local 192.168.0.97 port 40407 connected with 192.168.0.3 port 5001
[  4] local 192.168.0.97 port 5001 connected with 192.168.0.3 port 41205
[ ID] Interval       Transfer     Bandwidth
[  5]  0.0-10.0 sec  27.0 MBytes  22.6 Mbits/sec
[  4]  0.0-12.3 sec  64.0 KBytes  42.8 Kbits/sec

3) If I run tcpdump the speeds drop considerably, below is a before and after
running tcpdump -i wla0

------------------------------------------------------------
[  3] local 192.168.0.97 port 57664 connected with 192.168.0.3 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec  28.6 MBytes  23.9 Mbits/sec

== after==

------------------------------------------------------------
Client connecting to isola, TCP port 5001
TCP window size: 16.0 KByte (default)
------------------------------------------------------------
[  3] local 192.168.0.97 port 57665 connected with 192.168.0.3 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-12.1 sec  72.0 KBytes  48.9 Kbits/sec


I also had this in my logs:
tcpdump uses obsolete (PF_INET,SOCK_PACKET)


I hope to try out 802.11n later this week.

Cheers,
Jacob

--

02:00.0 Network controller: Atheros Communications Inc. AR5418 802.11abgn
Wireless PCI Express Adapter (rev 01)
	Subsystem: Apple Computer Inc. Device 0087
	Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping-
SERR- FastB2B- DisINTx-
	Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort-
<MAbort- >SERR- <PERR- INTx-
	Latency: 0, Cache Line Size: 256 bytes
	Interrupt: pin A routed to IRQ 17
	Region 0: Memory@90100000 (64-bit, non-prefetchable) [size=64K]
	Capabilities: <access denied>
	Kernel driver in use: ath9k
	Kernel modules: ath9k

wlan0     IEEE 802.11abgn  ESSID:"XX"  
          Mode:Managed  Frequency:2.412 GHz  Access Point: 00:1C:B3:AE:F8:30   
          Bit Rate=0 kb/s   Tx-Power=20 dBm   
          Retry min limit:7   RTS thr:off   Fragment thr=2352 B   
          Power Management:off
          Link Quality=59/100  Signal level:-56 dBm  Noise level=-94 dBm
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:0  Invalid misc:0   Missed beacon:0

[ath9k-devel] Issues with ar5418

[Benoit PAPILLAULT]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jacob a ?crit :
> I have been running the latest wireless-testing for about 24 hours and generally
> the card has been very stable. I've tested it with and without WPA/1.
> 
> There are a few issues though,  I also had the same issues with 2.6.27.1...
> 
> 1) Whenever i run iperf, I get this many of these in the logs:
> ath_tx_prepare: TX queue: 1 is full, depth: 492
> ath_tx_prepare: TX queue: 1 is full, depth: 492
> 
> 2) There is a big difference in speed upstream and downstream throughout, the
> figures below are fairly constant:
> 
> Client connecting to XX, TCP port 5001
> TCP window size: 16.0 KByte (default)
> ------------------------------------------------------------
> [  5] local 192.168.0.97 port 40407 connected with 192.168.0.3 port 5001
> [  4] local 192.168.0.97 port 5001 connected with 192.168.0.3 port 41205
> [ ID] Interval       Transfer     Bandwidth
> [  5]  0.0-10.0 sec  27.0 MBytes  22.6 Mbits/sec
> [  4]  0.0-12.3 sec  64.0 KBytes  42.8 Kbits/sec
> 
> 3) If I run tcpdump the speeds drop considerably, below is a before and after
> running tcpdump -i wla0
> 
> ------------------------------------------------------------
> [  3] local 192.168.0.97 port 57664 connected with 192.168.0.3 port 5001
> [ ID] Interval       Transfer     Bandwidth
> [  3]  0.0-10.0 sec  28.6 MBytes  23.9 Mbits/sec
> 
> == after==
> 
> ------------------------------------------------------------
> Client connecting to isola, TCP port 5001
> TCP window size: 16.0 KByte (default)
> ------------------------------------------------------------
> [  3] local 192.168.0.97 port 57665 connected with 192.168.0.3 port 5001
> [ ID] Interval       Transfer     Bandwidth
> [  3]  0.0-12.1 sec  72.0 KBytes  48.9 Kbits/sec
> 
> 
> I also had this in my logs:
> tcpdump uses obsolete (PF_INET,SOCK_PACKET)
> 
> 
> I hope to try out 802.11n later this week.
> 
> Cheers,
> Jacob
> 
> --
> 
> 02:00.0 Network controller: Atheros Communications Inc. AR5418 802.11abgn
> Wireless PCI Express Adapter (rev 01)
> 	Subsystem: Apple Computer Inc. Device 0087
> 	Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping-
> SERR- FastB2B- DisINTx-
> 	Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort-
> <MAbort- >SERR- <PERR- INTx-
> 	Latency: 0, Cache Line Size: 256 bytes
> 	Interrupt: pin A routed to IRQ 17
> 	Region 0: Memory at 90100000 (64-bit, non-prefetchable) [size=64K]
> 	Capabilities: <access denied>
> 	Kernel driver in use: ath9k
> 	Kernel modules: ath9k
> 
> wlan0     IEEE 802.11abgn  ESSID:"XX"  
>           Mode:Managed  Frequency:2.412 GHz  Access Point: 00:1C:B3:AE:F8:30   
>           Bit Rate=0 kb/s   Tx-Power=20 dBm   
>           Retry min limit:7   RTS thr:off   Fragment thr=2352 B   
>           Power Management:off
>           Link Quality=59/100  Signal level:-56 dBm  Noise level=-94 dBm
>           Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
>           Tx excessive retries:0  Invalid misc:0   Missed beacon:0

Hi Jacob,

I got similar results using WRT350N from Linksys instead of the Apple
AP. I'm really wondering if 2.4GHz is the appropriate band for 802.11n
since I got the same results with another 802.11n USB thing from Ralink
and I heard nearby users complaining about no Internet access while I
was doing my test.

In the 2.4GHz band, each channel is 5MHz apart and 40MHz means you are
using 8 channels out of 11 simultaneously!

I'd like to know what other people think about this point?

Regards,
Benoit

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFI+ZfQOR6EySwP7oIRAso+AJ4zhQFRWO6Sb3wzwTqXSQYQGlCfEgCg20oV
RwOVmB1ur0xTTtY/CgCW5vY=
=KoEu
-----END PGP SIGNATURE-----

[ath9k-devel] Issues with ar5418

[Luis R. Rodriguez]

On Sat, Oct 18, 2008 at 01:01:21AM -0700, Benoit PAPILLAULT wrote:
> Hi Jacob,
> 
> I got similar results using WRT350N from Linksys instead of the Apple
> AP. I'm really wondering if 2.4GHz is the appropriate band for 802.11n
> since I got the same results with another 802.11n USB thing from Ralink
> and I heard nearby users complaining about no Internet access while I
> was doing my test.
> 
> In the 2.4GHz band, each channel is 5MHz apart and 40MHz means you are
> using 8 channels out of 11 simultaneously!
> 
> I'd like to know what other people think about this point?

If you can use 5 GHz I'd recommend it as there is less noise there
compared to 2.4 Ghz. You can also use HT20 on 2 GHz. Remember that
the original goal for 11n was to try to achieve 100 Mbps but we
obviously surpassed that. Another question to is how much throughput
do you really need.

If you want to make use of aggregation sujith today posted a patch
which enables it for ath9k. Check Linux-wireless mailing list for that.

Anyway -- regardless any throughput issues we have need to be addressed.
My feeling is we need to review locking on TX path a bit more carefully
on ath9k. I'm going to try to look into that today.

  Luis