number of peers exceeded after a short time

[Michal Kazior]

On 7 January 2015 at 06:28, Janusz Dziedzic <janusz.dziedzic at tieto.com> wrote:
> On 6 January 2015 at 21:30, Conrad Kostecki <ck at conrad-kostecki.de> wrote:
>> Hi!
>>
>> I am using hostapd 2.3 with 2x ath10k (WLE900VX, FW 10.2). Kernel is 3.18.0.
>> Four SSIDs are being broadcasted. Two directly from ath10k and another two via a vif, based on the real ath10k-interfaces.
>> Two SSIDs are using WPA2 and two other are using WPA2-Enterprise.
>>
>> This problem started for me since upgrading vom Kernel 3.17.x (FW 10.1) to Kernel 3.18.0 (FW 10.2). When I downgrade, it happens much less..
>>
>> After a very short time, about 5-15 minutes, my dmesg is beeing logged full with those messages:
>> [2055815.205855] ath10k_pci 0000:0c:00.0: number of peers exceeded: peers number 127 (max peers 127)
>> [2055818.796016] ath10k_pci 0000:0c:00.0: number of peers exceeded: peers number 127 (max peers 127)
>> [2055822.392301] ath10k_pci 0000:0c:00.0: number of peers exceeded: peers number 127 (max peers 127)
>> [...]
>>
>> How is this possible? This is my private home router. I've only a few wifi devices.
>> I've noticed, that this problem mainly happens, when my Nokia Lumia 930 tries to connect with my WPA2-Enterprise network.
>>
>> Is this a ath10k problem, which is not correctly processing the peers or hostapd?

I guess this might be ath10k issue. I'd love to get my hands on logs
with debug_mask=0xffffff3f (ath10k_core parameter) though.

Perhaps Lumia randomizes its hw addr (though I doubt that).


> I suspect Michal already fix this,
> ath10k: fix station count enforcement
>
> Best use kernel from ath10k github.

The patch refined accounting to be more accurate so I don't think
it'll fix what Conrad is seeing. Nevertheless it's a good idea to
check up with github.com/kvalo/ath master branch.


Michał