[ath9k-devel] ath9k driver instability

[ath9k-devel] ath9k driver instability

[Cloud Strife]

I'm using kernel 2.6.34

My card is
phy0: Atheros AR5416 MAC/BB Rev:2 AR2133 RF Rev:81 mem=0xffffc90012120000, irq=21
[   43.935433] ath9k: Two wiphys trying to scan at the same time
[   44.134441] wlan0: authenticate with 00:22:75:a1:14:88 (try 1)
[   44.136353] wlan0: authenticated
[   44.136364] wlan0: associate with 00:22:75:a1:14:88 (try 1)
[   44.140575] wlan0: RX AssocResp from 00:22:75:a1:14:88 (capab=0x411 status=0 aid=2)
[   44.140577] wlan0: associated
eq
When i run downloads my speed is very unstable! I'll go up to 340 K/s down to 8 K/s. The speed is completely instable
This thing is MIMO AND it has a Rosewill Mega Antenna set connected to it and running @ 85%-89% reception ! I'm probably a maximum of 10feet away from my Belkin N+ router and I'm using channel 11+ extension channel 7 in mixed mode with WPA2 PSK  @ bandwitdths 20/40MHz

My broadcom air force one card has no issues like this at all, so it is the Atheros drivers and not the router. Just thought I should rule that out for ya ;)

Sometimes the card stops working all together (cannot scan, connect, etc.), I have to ifconfig wlan0 down; rmmod ath9k; modprobe ath9k; ifconfig wlan0 up to get it started again. There is also no interference in my area as well. Honestly the version in 2.6.33 worked better with speed stability. I probably got around 260K/s although I should have been recieving around 340K/s .
So now I can reach my top speed that I should be acquiring but It doesn't last and flucuates too much to rely on.

darksurf at DarkGlow ~ $ ping yahoo.com
PING yahoo.com (209.191.122.70) 56(84) bytes of data.
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=1 ttl=54 time=162 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=2 ttl=54 time=147 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=3 ttl=54 time=76.6 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=4 ttl=54 time=1158 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=5 ttl=54 time=303 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=6 ttl=54 time=93.2 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=7 ttl=54 time=81.3 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=8 ttl=54 time=180 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=10 ttl=54 time=870 ms
64 bytes from ir1.fp.vip.mud.yahoo.com (209.191.122.70): icmp_seq=11 ttl=54 time=256 ms
^C
--- yahoo.com ping statistics ---
11 packets transmitted, 10 received, 9% packet loss, time 15333ms
rtt min/avg/max/mdev = 76.633/333.143/1158.779/353.651 ms, pipe 2

any ideas?



      

[ath9k-devel] ath9k driver instability

[Peter Stuge]

Cloud Strife wrote:
> phy0: Atheros AR5416 MAC/BB

That's an AR5008 card. I have one too.


> speed is completely instable

> Sometimes the card stops working all together

> any ideas?

Unfortunately Atheros does not support ath9k for AR5008, only for the
newest generation hardware, so you and I are both SOL. I wish I knew
that before I bought my card, or at least immediately after I tried
to debug the driver and seek help from Atheros experts.

I also experience many performance problems with the card and driver,
typically with scanning and associating. I never load my link with
more than 10Mbps, and can usually get that reliably at one site.
(Linux AP running madwifi-ng.) In particular, any attempts at using
AR5008+ath9k to communicate with Apple products is an absolute
travesty in my experience.

Great work is done by community members, so the situation may still
improve for us, but it would certainly be nice if Atheros could more
actively support also their older products.


//Peter

[ath9k-devel] ath9k driver instability

[Alan]

In my case, I have 2 different AR5008 based cards: a pci-e d-link
dwa-556 with 3x3 mimo that works with ath9k somehow stable (but I have
not been able to get more than 20 mbit/s, and can't use or don't know
how to use 802.11n), and a bunch of mini pci based cards with 3x3 mimo
that only work with madwifi (and crash with ath9k), and I'm unable to
exceed 20 mbit/s with these too (and of course, madwifi doesn't
support 802.11n).

But your case seems far worse than mine, I would suggest you to try
madwifi and see if at least you can get more stability and maybe
speed, and if it doesn't work, get an intel based card, I've never had
a bad experience with intel cards.


Best regards,

Alan

[ath9k-devel] ath9k driver instability

[Steven Barrett]

I also have an ath9k card:

02:00.0 Network controller [0280]: Atheros Communications Inc. AR9285
Wireless Network Adapter (PCI-Express) [168c:002b] (rev 01)

I cannot complete any large download.  For example, if I try to
download the full kernel source from
http://www.kernel.org/pub/linux/kernel/v2.6/linux-2.6.34.tar.bz2, the
first one or two seconds will download at about 300 to 400 kB/s, then
it seems that the download simply stops, and no more bits are received
for many seconds.  The download stops and starts and usually even
fails.  Eventually, I need to just stop all together and find an
ethernet jack to plug into.

The ath9k is really unreliable, I basically can't use it to do
anything, even surfing web pages for a few minutes.

Is there anything I can do to help debug this connection interruption?
 I've been testing the newest compat-wireless-2.6 packs on
wireless.kernel.org, nothing is helping.  Whatever the problem is, it
seems to be common between ath5k and ath9k.

On Tue, Jun 1, 2010 at 12:20 AM, Alan <lameventanas@gmail.com> wrote:
> In my case, I have 2 different AR5008 based cards: a pci-e d-link
> dwa-556 with 3x3 mimo that works with ath9k somehow stable (but I have
> not been able to get more than 20 mbit/s, and can't use or don't know
> how to use 802.11n), and a bunch of mini pci based cards with 3x3 mimo
> that only work with madwifi (and crash with ath9k), and I'm unable to
> exceed 20 mbit/s with these too (and of course, madwifi doesn't
> support 802.11n).
>
> But your case seems far worse than mine, I would suggest you to try
> madwifi and see if at least you can get more stability and maybe
> speed, and if it doesn't work, get an intel based card, I've never had
> a bad experience with intel cards.
>
>
> Best regards,
>
> Alan
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>

[ath9k-devel] ath9k driver instability

[Ben Gamari]

On Sat, 29 May 2010 22:16:13 -0700 (PDT), Cloud Strife <piroisl33t@yahoo.com> wrote:
> Sometimes the card stops working all together (cannot scan, connect,
> etc.), I have to ifconfig wlan0 down; rmmod ath9k; modprobe ath9k;
> ifconfig wlan0 up to get it started again. There is also no
> interference in my area as well. 
> 
I too have been experiencing this sort of failure with my AR5008
cards. I have a AR5008 rev. 1 card in my server running in master mode,
which dies about once per day. During this period all stations
attempting to associate with the access point timeout. Unfortunately,
there is no useful error message produced by the driver, so I have no
idea what might be going wrong. The only way to bring the card back is
to reload the driver.

I have another similar card in my personal laptop, which generally runs
in managed mode. Here, I experience similar types of failures several
times a day. The symptoms are almost identical, with inexplicable timeouts

I'm aware that the Atheros developers aren't actively supporting
AR5008 hardware but even some hints as to where to begin in debugging
would be appreciated. I've more than once stated my complete willingness
to do whatever necessary to troubleshoot this issue yet Atheros has
remained entirely silent.

I would be more than willing to buy hardware based on a newer Atheros
chipset if there were products available. However, despite the dozens of
products listed on the Ath9k products page[1], I have not seen a single
AR92xx-based full-sized mini-PCIe card available to consumers (most of
the hardware is produce for OEM it appears, a market to which most
individuals have no access). Meanwhile, AR5008 hardware is cheap,
available, and works, albeit not with ath9k. However, I would be most
appreciative of anyone who could point me in the direction of where to
acquire a newer card compatible with my hardware.

Any input Atheros could provide would be greatly appreciated. Thanks,

- Ben


[1] http://wireless.kernel.org/en/users/Drivers/ath9k/products/external

[ath9k-devel] ath9k driver instability

[Felix Fietkau]

On 2010-06-14 1:58 PM, Ben Gamari wrote:
> On Sat, 29 May 2010 22:16:13 -0700 (PDT), Cloud Strife <piroisl33t@yahoo.com> wrote:
>> Sometimes the card stops working all together (cannot scan, connect,
>> etc.), I have to ifconfig wlan0 down; rmmod ath9k; modprobe ath9k;
>> ifconfig wlan0 up to get it started again. There is also no
>> interference in my area as well. 
>> 
> I too have been experiencing this sort of failure with my AR5008
> cards. I have a AR5008 rev. 1 card in my server running in master mode,
> which dies about once per day. During this period all stations
> attempting to associate with the access point timeout. Unfortunately,
> there is no useful error message produced by the driver, so I have no
> idea what might be going wrong. The only way to bring the card back is
> to reload the driver.
I would recommend testing the latest bleeding-edge compat-wireless
tarball. That's what I use in OpenWrt, which is getting a lot of testing
specifically in AP mode.

> I have another similar card in my personal laptop, which generally runs
> in managed mode. Here, I experience similar types of failures several
> times a day. The symptoms are almost identical, with inexplicable timeouts
> 
> I'm aware that the Atheros developers aren't actively supporting
> AR5008 hardware but even some hints as to where to begin in debugging
> would be appreciated. I've more than once stated my complete willingness
> to do whatever necessary to troubleshoot this issue yet Atheros has
> remained entirely silent.
There is just too much stuff to do, which makes it hard to find the time
to take care of stability on old hardware as well, especially since
debugging such issues can be very time consuming.

> I would be more than willing to buy hardware based on a newer Atheros
> chipset if there were products available. However, despite the dozens of
> products listed on the Ath9k products page[1], I have not seen a single
> AR92xx-based full-sized mini-PCIe card available to consumers (most of
> the hardware is produce for OEM it appears, a market to which most
> individuals have no access). Meanwhile, AR5008 hardware is cheap,
> available, and works, albeit not with ath9k. However, I would be most
> appreciative of anyone who could point me in the direction of where to
> acquire a newer card compatible with my hardware.
I got myself an Ubiquiti SR-71e, but unfortunately the card seems to
have disappeared from the manufacturer's website. Maybe you can still
find it in a shop somewhere. It does work well for me with ath9k.

- Felix

[ath9k-devel] ath9k driver instability

[RHS Linux User]

Hi All,

    I notice that the wireless driver version in xpud works reliably,
while the version in puppy does not? The puppy version comes up OK, but
then dies after a short time (minutes)? 

    I wonder if the wireless driver can (easily?) be arranged to be kernel
version independent? It would sure help with debug. Maybe some sort of
compatibility layer or special conversion/debug interface driver with a
"common" interface to the CPU ?

    warm regards to all,
    wiz

On Tue, 15 Jun 2010, Felix Fietkau wrote:

> On 2010-06-14 1:58 PM, Ben Gamari wrote:
> > On Sat, 29 May 2010 22:16:13 -0700 (PDT), Cloud Strife <piroisl33t@yahoo.com> wrote:
> >> Sometimes the card stops working all together (cannot scan, connect,
> >> etc.), I have to ifconfig wlan0 down; rmmod ath9k; modprobe ath9k;
> >> ifconfig wlan0 up to get it started again. There is also no
> >> interference in my area as well. 
> >> 
> > I too have been experiencing this sort of failure with my AR5008
> > cards. I have a AR5008 rev. 1 card in my server running in master mode,
> > which dies about once per day. During this period all stations
> > attempting to associate with the access point timeout. Unfortunately,
> > there is no useful error message produced by the driver, so I have no
> > idea what might be going wrong. The only way to bring the card back is
> > to reload the driver.
> I would recommend testing the latest bleeding-edge compat-wireless
> tarball. That's what I use in OpenWrt, which is getting a lot of testing
> specifically in AP mode.
> 
> > I have another similar card in my personal laptop, which generally runs
> > in managed mode. Here, I experience similar types of failures several
> > times a day. The symptoms are almost identical, with inexplicable timeouts
> > 
> > I'm aware that the Atheros developers aren't actively supporting
> > AR5008 hardware but even some hints as to where to begin in debugging
> > would be appreciated. I've more than once stated my complete willingness
> > to do whatever necessary to troubleshoot this issue yet Atheros has
> > remained entirely silent.
> There is just too much stuff to do, which makes it hard to find the time
> to take care of stability on old hardware as well, especially since
> debugging such issues can be very time consuming.
> 
> > I would be more than willing to buy hardware based on a newer Atheros
> > chipset if there were products available. However, despite the dozens of
> > products listed on the Ath9k products page[1], I have not seen a single
> > AR92xx-based full-sized mini-PCIe card available to consumers (most of
> > the hardware is produce for OEM it appears, a market to which most
> > individuals have no access). Meanwhile, AR5008 hardware is cheap,
> > available, and works, albeit not with ath9k. However, I would be most
> > appreciative of anyone who could point me in the direction of where to
> > acquire a newer card compatible with my hardware.
> I got myself an Ubiquiti SR-71e, but unfortunately the card seems to
> have disappeared from the manufacturer's website. Maybe you can still
> find it in a shop somewhere. It does work well for me with ath9k.
> 
> - Felix
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
> 

[ath9k-devel] ath9k driver instability

[Felix Fietkau]

On 2010-06-16 1:32 AM, RHS Linux User wrote:
> 
> Hi All,
> 
>     I notice that the wireless driver version in xpud works reliably,
> while the version in puppy does not? The puppy version comes up OK, but
> then dies after a short time (minutes)? 
> 
>     I wonder if the wireless driver can (easily?) be arranged to be kernel
> version independent? It would sure help with debug. Maybe some sort of
> compatibility layer or special conversion/debug interface driver with a
> "common" interface to the CPU ?
Compat-wireless takes care of the backporting stuff...

- Felix

[ath9k-devel] ath9k driver instability

[Alphan Ulusoy]

Why don't you give compat-wireless packages a try? I've upgraded my
kernel only after trying out a 2.6.34 compat-wireless package. You can
find them on the linuxwireless website or google for it.

On 6/16/10, RHS Linux User <xxx@nei.mv.com> wrote:
>
> Hi All,
>
>     I notice that the wireless driver version in xpud works reliably,
> while the version in puppy does not? The puppy version comes up OK, but
> then dies after a short time (minutes)?
>
>     I wonder if the wireless driver can (easily?) be arranged to be kernel
> version independent? It would sure help with debug. Maybe some sort of
> compatibility layer or special conversion/debug interface driver with a
> "common" interface to the CPU ?
>
>     warm regards to all,
>     wiz
>
> On Tue, 15 Jun 2010, Felix Fietkau wrote:
>
>> On 2010-06-14 1:58 PM, Ben Gamari wrote:
>> > On Sat, 29 May 2010 22:16:13 -0700 (PDT), Cloud Strife
>> > <piroisl33t@yahoo.com> wrote:
>> >> Sometimes the card stops working all together (cannot scan, connect,
>> >> etc.), I have to ifconfig wlan0 down; rmmod ath9k; modprobe ath9k;
>> >> ifconfig wlan0 up to get it started again. There is also no
>> >> interference in my area as well.
>> >>
>> > I too have been experiencing this sort of failure with my AR5008
>> > cards. I have a AR5008 rev. 1 card in my server running in master mode,
>> > which dies about once per day. During this period all stations
>> > attempting to associate with the access point timeout. Unfortunately,
>> > there is no useful error message produced by the driver, so I have no
>> > idea what might be going wrong. The only way to bring the card back is
>> > to reload the driver.
>> I would recommend testing the latest bleeding-edge compat-wireless
>> tarball. That's what I use in OpenWrt, which is getting a lot of testing
>> specifically in AP mode.
>>
>> > I have another similar card in my personal laptop, which generally runs
>> > in managed mode. Here, I experience similar types of failures several
>> > times a day. The symptoms are almost identical, with inexplicable
>> > timeouts
>> >
>> > I'm aware that the Atheros developers aren't actively supporting
>> > AR5008 hardware but even some hints as to where to begin in debugging
>> > would be appreciated. I've more than once stated my complete willingness
>> > to do whatever necessary to troubleshoot this issue yet Atheros has
>> > remained entirely silent.
>> There is just too much stuff to do, which makes it hard to find the time
>> to take care of stability on old hardware as well, especially since
>> debugging such issues can be very time consuming.
>>
>> > I would be more than willing to buy hardware based on a newer Atheros
>> > chipset if there were products available. However, despite the dozens of
>> > products listed on the Ath9k products page[1], I have not seen a single
>> > AR92xx-based full-sized mini-PCIe card available to consumers (most of
>> > the hardware is produce for OEM it appears, a market to which most
>> > individuals have no access). Meanwhile, AR5008 hardware is cheap,
>> > available, and works, albeit not with ath9k. However, I would be most
>> > appreciative of anyone who could point me in the direction of where to
>> > acquire a newer card compatible with my hardware.
>> I got myself an Ubiquiti SR-71e, but unfortunately the card seems to
>> have disappeared from the manufacturer's website. Maybe you can still
>> find it in a shop somewhere. It does work well for me with ath9k.
>>
>> - Felix
>> _______________________________________________
>> ath9k-devel mailing list
>> ath9k-devel at lists.ath9k.org
>> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>>
>
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>


-- 
-- alphan