Re: [PATCH] ath9k: Add RX inactivity detection and reset chip when it occurs

["Toke Høiland-Jørgensen" <toke@toke.dk>]

Hamdi Issam <ih@simonwunderlich.de> writes:

> On 11/6/24 17:03, Simon Wunderlich wrote:
>> On Wednesday, November 6, 2024 3:12:59 PM CET Toke Høiland-Jørgensen wrote:
>>> Sven Eckelmann <se@simonwunderlich.de> writes:
>>>> Hi,
>>>>
>>>> Thank you for submitting the patch.
>>>>
>>>> On Wednesday, 6 November 2024 13:41:44 CET Toke Høiland-Jørgensen wrote:
>>>>> Since this is based on ideas by all three people, but not actually
>>>>> directly derived from any of the patches, I'm including Suggested-by
>>>>> tags from Simon, Sven and Felix below, which should hopefully serve as
>>>>> proper credit.
>>>> At least for me, this is more than enough. Thanks.
>>>>
>>>> I don't have the setup at the moment to test it again - maybe Issam can do
>>>> this. One concern I would have (because I don't find the notes regarding
>>>> this problem), is whether this check is now breaking because we count
>>>> more things. In the past, rxlp/rxok was used for the check. And now I
>>>> don't know whether the count for the other ones were still increasing.
>>>>
>>>> * RXHP (rather sure that "high priority frame" wasn't increasing)
>>>> * RXEOL ("no RX descriptors available" - I would guess no, but I can't say
>>>> for>
>>>>    sure)
>>>>
>>>> * RXORN ("FIFO overrun" I would guess no, but I can't say for sure)
>>>>
>>>> Reviewed-by: Sven Eckelmann <se@simonwunderlich.de>
>>> Great, thanks for the review! I'll let it sit in patchwork for a little
>>> while to give people a chance to test it out before sending it over to
>>> Kalle to be applied :)
>>>
>>> -Toke
>> Hi Toke,
>>
>> this looks good to me in general. I'm not sure either about the particular RX
>> interrupts. We can test this by putting the AP in a shield box and verify that
>> the counters are actually increasing, and that should be good enough.
>>
>> Acked-by: Simon Wunderlich <sw@simonwunderlich.de>
>>
>> Thank you!
>>        Simon
>
> Hi Toke,
>
> I have tested this patch in mesh mode, and it functions as expected.
>
> I conducted the test by placing one node inside a shield box and the 
> other outside, then verified whether a reset occurred due to RX path 
> inactivity.
>
> Tested-by: Issam Hamdi <ih@simonwunderlich.de>

Great, thanks for testing! :)

-Toke