ath11k: QCA6390 hw2.0: firmware crashing on ath.git master branch

[Aloka Dixit]

On 10/23/2023 9:35 AM, Kalle Valo wrote:
> Aloka Dixit <quic_alokad at quicinc.com> writes:
> 
>> On 10/23/2023 7:34 AM, Kalle Valo wrote:
>>
>>> + aloka
>>> Kalle Valo <kvalo at kernel.org> writes:
>>> I found a reliable way to reproduce the firmware crash. I connect to
>>> my
>>> home network, then do 'sudo service connman stop' and boom, the firmware
>>> has crashed. I suspect it has nothing to do with connman, most likely a
>>> simple disconnect will suffice but I didn't bother to test that.
>>> I did a git bisect and it points to Aloka's recent commit. Luckily
>>> the
>>> commit is only in ath-next so far and not in any release. My plan is to
>>> submit a revert by tomorrow so that we get this fixed before pulling
>>> ath-next into wireless-next.
>>> e149353e6562f3e3246f75dfc4cca6a0cc5b4efc
>>> commit e149353e6562f3e3246f75dfc4cca6a0cc5b4efc
>>> Author:     Aloka Dixit <quic_alokad at quicinc.com>
>>> AuthorDate: Mon Oct 9 10:13:54 2023 +0300
>>> Commit:     Kalle Valo <quic_kvalo at quicinc.com>
>>> CommitDate: Tue Oct 10 10:23:02 2023 +0300
>>>       wifi: ath11k: call ath11k_mac_fils_discovery() without
>>> condition
>>>
>>
>> Oh, I'm so sorry about that.
> 
> No worries, this is why we do testing :)
> 
> Once again this proves why eating your own dog food is so important[1].
> 
>> Please revert the commit, I will check what is happening and get back.
>> You haven't faced this issue with the same change in ath12k right?
> 
> My ath12k testing is still quite limited. I also didn't see this on my
> ath11k test box with WCN6855 but I haven't investigated this in detail.
> So far I have only seen this with QCA6390.
> 
> [1] https://en.wikipedia.org/wiki/Eating_your_own_dog_food
> 

I can send ath12k patch to revert that change as well until I figure out 
why it fails in ath11k. Let me know.
Thanks.