[PATCH 00/10] ath9k / mac80211: power save fixes

[PATCH 00/10] ath9k / mac80211: power save fixes

[Luis R. Rodriguez]

Here are a bundle of joy of fixes for power save. Some of
these are regressions, which sucks ass to find out about them
so late. We *really* need a regression testbed for mac80211 and
ath9k. Some stable fixes trickle down to 2.6.32 even.

I'll soon send links for the respective ports of these patches to each
kernels. For now I'd just like them merged upstream and ensure we get
this out for 2.6.36-rc stuff.

The offchannel needs more work to be really useful, for example
we should just be sending nullunc frames prior to actualloy probing
the AP. We can work on this later for 37.

Luis R. Rodriguez (9):
  ath9k: fix power save race conditions
  ath9k: fix regression on beacon loss after bgscan
  mac80211: add helper for reseting the connection monitor
  mac80211: reset probe send counter upon connection timer reset
  mac80211: reset connection idle when going offchannel
  mac80211: make the beacon monitor available externally
  mac80211: disable beacon monitor while going offchannel
  mac80211: send last 3/5 probe requests as unicast
  ath9k: fix regression which disabled ps on ath9k

Senthil Balasubramanian (1):
  ath9k: fix regression which prevents chip sleep after CAB data

 drivers/net/wireless/ath/ath9k/init.c |    3 +-
 drivers/net/wireless/ath/ath9k/main.c |   11 ++++++--
 drivers/net/wireless/ath/ath9k/recv.c |    5 +++-
 net/mac80211/ieee80211_i.h            |    2 +
 net/mac80211/mlme.c                   |   40 +++++++++++++++++++++++---------
 net/mac80211/offchannel.c             |    7 +++++
 6 files changed, 52 insertions(+), 16 deletions(-)

Re: [PATCH 00/10] ath9k / mac80211: power save fixes

[Luis R. Rodriguez]

On Mon, Sep 13, 2010 at 10:30 PM, Luis R. Rodriguez
<lrodriguez@atheros.com> wrote:
> Here are a bundle of joy of fixes for power save. Some of
> these are regressions, which sucks ass to find out about them
> so late. We *really* need a regression testbed for mac80211 and
> ath9k. Some stable fixes trickle down to 2.6.32 even.
>
> I'll soon send links for the respective ports of these patches to each
> kernels. For now I'd just like them merged upstream and ensure we get
> this out for 2.6.36-rc stuff.
>
> The offchannel needs more work to be really useful, for example
> we should just be sending nullunc frames prior to actualloy probing
> the AP. We can work on this later for 37.
>
> Luis R. Rodriguez (9):
>  ath9k: fix power save race conditions
>  ath9k: fix regression on beacon loss after bgscan
>  mac80211: add helper for reseting the connection monitor
>  mac80211: reset probe send counter upon connection timer reset
>  mac80211: reset connection idle when going offchannel
>  mac80211: make the beacon monitor available externally
>  mac80211: disable beacon monitor while going offchannel
>  mac80211: send last 3/5 probe requests as unicast
>  ath9k: fix regression which disabled ps on ath9k
>
> Senthil Balasubramanian (1):
>  ath9k: fix regression which prevents chip sleep after CAB data

The all in one git am'able file can be downloaded here:

http://bombadil.infradead.org/~mcgrof/patches/ath9k/2010/09/all-ps-fixes-09-14-v1.patch

  Luis

Re: [PATCH 00/10] ath9k / mac80211: power save fixes

[Ben Greear]

On 09/14/2010 08:07 AM, Luis R. Rodriguez wrote:
> On Mon, Sep 13, 2010 at 10:30 PM, Luis R. Rodriguez
> <lrodriguez@atheros.com>  wrote:
>> Here are a bundle of joy of fixes for power save. Some of
>> these are regressions, which sucks ass to find out about them
>> so late. We *really* need a regression testbed for mac80211 and
>> ath9k. Some stable fixes trickle down to 2.6.32 even.

We're working on supporting ath9k for the virtual interfaces
in our system, and we do our own regression testing.

Aside from normal traffic passing, do you have any test
cases you'd like to see run?

> The all in one git am'able file can be downloaded here:
>
> http://bombadil.infradead.org/~mcgrof/patches/ath9k/2010/09/all-ps-fixes-09-14-v1.patch

Is there any way to get automated emails for commits going into wireless-testing?

If these go in soon, I'd rather not deal with merging them into my own repo,
but if it's going to take a while, it might be worth the effort.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com

Re: [PATCH 00/10] ath9k / mac80211: power save fixes

[Luis R. Rodriguez]

On Tue, Sep 14, 2010 at 8:53 AM, Ben Greear <greearb@candelatech.com> wrote:
> On 09/14/2010 08:07 AM, Luis R. Rodriguez wrote:
>>
>> On Mon, Sep 13, 2010 at 10:30 PM, Luis R. Rodriguez
>> <lrodriguez@atheros.com>  wrote:
>>>
>>> Here are a bundle of joy of fixes for power save. Some of
>>> these are regressions, which sucks ass to find out about them
>>> so late. We *really* need a regression testbed for mac80211 and
>>> ath9k. Some stable fixes trickle down to 2.6.32 even.
>
> We're working on supporting ath9k for the virtual interfaces
> in our system, and we do our own regression testing.

Neat, so does Google, they talked about this at the Wireless summit.
They have stuffed their code into their version of autotest:

http://code.google.com/p/autotest/

The upstream autotest:

http://autotest.kernel.org/

I have yet to check this code out though.

> Aside from normal traffic passing, do you have any test
> cases you'd like to see run?

What I think we need is to take advantage of the random users of
wireless-testing / compat-wireless out there and their different APs
and get collection of test results of some basis tests. A very simple
series of test cases like authentication, association, enabling power
save and keeping that connection alive for a period of time, and some
simple iperf runs would go long way.

On very specific testsbeds though like yours and Google's we can focus
more on more complex items like roaming. We can actually still
accomplish automated roaming tests by using Orbit's mobile robot, as
it goes around a grid of APs. It would be very helpful if regression
tests had published results publicly. It would be really helpful if we
had more coordination between those who are doing regression tests.

>> The all in one git am'able file can be downloaded here:
>>
>>
>> http://bombadil.infradead.org/~mcgrof/patches/ath9k/2010/09/all-ps-fixes-09-14-v1.patch
>
> Is there any way to get automated emails for commits going into
> wireless-testing?

You can subscribe to the RSS feed via http://git.kernel.org for it maybe.

> If these go in soon, I'd rather not deal with merging them into my own repo,
> but if it's going to take a while, it might be worth the effort.

Right.

  Luis

Re: [PATCH 00/10] ath9k / mac80211: power save fixes

[Ben Greear]

On 09/14/2010 09:09 AM, Luis R. Rodriguez wrote:
> On Tue, Sep 14, 2010 at 8:53 AM, Ben Greear<greearb@candelatech.com>  wrote:
>> On 09/14/2010 08:07 AM, Luis R. Rodriguez wrote:
>>>
>>> On Mon, Sep 13, 2010 at 10:30 PM, Luis R. Rodriguez
>>> <lrodriguez@atheros.com>    wrote:
>>>>
>>>> Here are a bundle of joy of fixes for power save. Some of
>>>> these are regressions, which sucks ass to find out about them
>>>> so late. We *really* need a regression testbed for mac80211 and
>>>> ath9k. Some stable fixes trickle down to 2.6.32 even.
>>
>> We're working on supporting ath9k for the virtual interfaces
>> in our system, and we do our own regression testing.
>
> Neat, so does Google, they talked about this at the Wireless summit.
> They have stuffed their code into their version of autotest:
>
> http://code.google.com/p/autotest/
>
> The upstream autotest:
>
> http://autotest.kernel.org/
>
> I have yet to check this code out though.
>
>> Aside from normal traffic passing, do you have any test
>> cases you'd like to see run?
>
> What I think we need is to take advantage of the random users of
> wireless-testing / compat-wireless out there and their different APs
> and get collection of test results of some basis tests. A very simple
> series of test cases like authentication, association, enabling power
> save and keeping that connection alive for a period of time, and some
> simple iperf runs would go long way.
>
> On very specific testsbeds though like yours and Google's we can focus
> more on more complex items like roaming. We can actually still
> accomplish automated roaming tests by using Orbit's mobile robot, as
> it goes around a grid of APs. It would be very helpful if regression
> tests had published results publicly. It would be really helpful if we
> had more coordination between those who are doing regression tests.

Our normal testing efforts would be good at long-term traffic generation
(1+ days) using our tool which is somewhat similar to iperf.

We haven't bothered with roaming before, but might can work that in for
at least limited testing.  Maybe a laptop running hostapd strapped on a roomba
would be fun :)

When we can get the virtual station (and hopefully virtual APs) stuff
stable, we'll start doing some tests and try to figure out how to publish
it in a useful manner for others.

>>> The all in one git am'able file can be downloaded here:
>>>
>>>
>>> http://bombadil.infradead.org/~mcgrof/patches/ath9k/2010/09/all-ps-fixes-09-14-v1.patch
>>
>> Is there any way to get automated emails for commits going into
>> wireless-testing?
>
> You can subscribe to the RSS feed via http://git.kernel.org for it maybe.

I was hoping someone had a post-receive trigger sending email to some list
that I could subscribe to.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com

Re: [PATCH 00/10] ath9k / mac80211: power save fixes

[Johannes Berg]

On Tue, 2010-09-14 at 09:09 -0700, Luis R. Rodriguez wrote:

> Neat, so does Google, they talked about this at the Wireless summit.
> They have stuffed their code into their version of autotest:
> 
> http://code.google.com/p/autotest/

Which really is at
http://git.chromium.org/cgi-bin/gitweb.cgi?p=autotest.git

johannes