[rt2x00-users] rt2800pci works nearly fine with 2.6.39rc7 kernel and compat-wireless-2.6.39-rc6-1 - problems during rekeying of PTK

[Gertjan van Wingerde]

On 05/28/11 08:07, Andreas Hartmann wrote:
> Andreas Hartmann schrieb:
>> Andreas Hartmann schrieb:
>>> A test with PTK rekeying disabled will follow.
>>
>> I couldn't see any problems with disabled PTK rekeying over about 15
>> minutes firing with two supplicants with netperf in different always
>> changing directions on each supplicant.
>>
>> This test should be done over a longer time, but I didn't have the time
>> for it :-(.
> 
> I switched to compat-wireless-2011-05-26 and applied the patches (no
> problem).
> 
> Testing as described above over 2 hours with disabled PTK rekeying
> didn't show up any of the warnings or errors we're searching for.
> The AP was/is stable.
> 
> This was the same as with enabled PTK rekeying.
> With compat-wireless-2001-05-05, I encountered a slowdown of the data
> throughput from AP to supplicant until 0.
> This problem didn't show up with 2011-05-26: the throughput was up to
> 50% reduced for short time, but always found back to 100% after a short
> period of time after the broken PTK rekeying.
> 
> All tests with enabled PTK rekeying where done with the AR9285 based
> supplicant. They couldn't be done with the WUSB600Nv2, because
> reauthentication with rt3572sta is somewhat broken. That's why I
> appreciate your support for rt3572-chips :-).
> 
> 
> Summarized:
> Your 3 patches solve the problem with "[ 3369.612961] phy0 ->
> rt2x00queue_write_tx_frame: Error - Arrived at non-free entry in the
> non-full queue 2. " and "phy0 -> rt2800pci_txdone: Warning - Got TX
> status for an empty queue 0, dropping". I never saw them anymore since
> applying your patch. The device didn't hang anymore and just worked as
> expected :-). This is good news, isn't it?
> 

Yes that it good news. I just submitted the patches for formal inclusion.
At least one problem that bites the dust.

<snip>

> 
> 
> Now, one problem is still left open: the broken rekeying under high
> load. If this would be repaired too, it would be great!
> 

Yeah, that one seems to be trickier, and maybe not even rt2x00 related,
as you reported it is also broken on other chipsets with other drivers.
I guess we should go to the linux-wireless mailing list for this problem.

In the mean time could you increase the log level of hostapd and see if
something strange pops up in the logs at the time that the PTK rekeying
breaks?

---
Gertjan