SWBA errors and no ssid

[Ann Lo]

Hi Michal,

This is good information.

Thanks,
Ann

On Thu, Feb 4, 2016 at 1:41 AM, Michal Kazior <michal.kazior at tieto.com> wrote:
> On 4 February 2016 at 10:33, Ann Lo <annlo.tech at gmail.com> wrote:
>> Hi Michal,
>>
>> Thanks for your information and suggestion. A limitation we have is
>> that we deal with board support package kernel, and we need to worry
>> about other drivers as well. We are hoping to make minimal changes.
>> There appear to be a number of changes in this area, including:
>>
>> ath10k: implement and use new beacon method
>> http://lists.infradead.org/pipermail/ath10k/2014-January/000940.html
>>
>> the best ath10k driver
>> http://lists.infradead.org/pipermail/ath10k/2015-August/005815.html
>>
>> Is it possible to apply a number of patches to Kernel 3.14.28? Or,
>> port ath10k from Kernel 3.19 which is newer but not as new as Kernel
>> 4.4? Is Qualcomm firmware 10.2.4.70.2 the only solution?
>
> Backports allow you to use newer drivers with older kernel.
>
> If that, for some reason, doesn't suit you you're probably left with
> approach taken with the driver for chromium:
>
>   https://groups.google.com/a/chromium.org/forum/#!topic/chromium-os-reviews/qUageXhYDb8
>
> (This was done based on what originally Intel guys did with their iwl driver)
>
> Cherry-picking changes manually is going to be a major pain with an
> insane cascade of changes (possibly conflicting with some custom
> changes in the kernel you use).
>
>
> Michał
>
>
>>
>> Thanks,
>> Ann
>>
>> On Wed, Feb 3, 2016 at 11:43 PM, Michal Kazior <michal.kazior at tieto.com> wrote:
>>> On 3 February 2016 at 20:33, Ann Lo <annlo.tech at gmail.com> wrote:
>>>> Hello,
>>>>
>>>> We are seeking for expert opinions on "SWBA errors" and any possible
>>>> workarounds. If we run AP+client mode, and the client scans for an
>>>> nonexistent ssid, then "SWBA errors" happen easily. From this point
>>>> on, the ssid of the AP is no longer visible. We use kernel 3.14.28,
>>>> Qualcomm firmware 10.1.167.3-1, and WLE900VX. Qualcomm firmware
>>>> 999.999.0.636 has the same problem except that it does not report
>>>> "SWBA errors". Is it possible that we can fix this problem by porting
>>>> some patches?
>>>>
>>>> Details of errors
>>>> ============
>>>> [ 3455.101199] ath10k: SWBA overrun on vdev 0
>>>> [ 3455.203616] ath10k: SWBA overrun on vdev 0
>>>> [ 3455.305992] ath10k: SWBA overrun on vdev 0
>>>> [ 3455.408389] ath10k: SWBA overrun on vdev 0
>>>> [ 3455.510768] ath10k: SWBA overrun on vdev 0
>>>> [ 3455.613164] ath10k: SWBA overrun on vdev 0
>>>> [ 3455.715557] ath10k: SWBA overrun on vdev 0
>>>> [ 3455.817992] ath10k: SWBA overrun on vdev 0
>>>> [ 3455.920360] ath10k: SWBA overrun on vdev 0
>>>> [ 3456.022768] ath10k: SWBA overrun on vdev 0
>>>> [ 3457.946735] uap0: failed to remove key (0, a4:2b:8c:04:6d:a8) from
>>>> hardware (-11)
>>>> [ 3460.118597] net_ratelimit: 40 callbacks suppressed
>>>> [ 3460.123463] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.221003] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.323398] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.425785] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.528186] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.630568] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.732971] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.835370] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.937767] ath10k: SWBA overrun on vdev 0
>>>> [ 3460.946679] ath10k: could not remove peer wep key 0 (-11)
>>>
>>> Looks like firmware has hanged. I recall there were weird issues a
>>> long time ago.
>>>
>>> SWBA is an event reported by firmware to host prompting it to submit a
>>> new beacon frame that is to be sent out. The "overrun" means that the
>>> driver is unable to submit the beacon because firmware command pipe is
>>> busy.
>>>
>>> You're using quite an old kernel. Can you perhaps try 4.4? If you
>>> can't replace the kernel for some reason did you think about using
>>> backports[1]?
>>>
>>> [1]: https://wireless.wiki.kernel.org/en/users/drivers/ath10k/backports
>>>
>>>
>>> Michał