Re: [PATCH] Revert "ath: add support for special 0x0 regulatory domain"

["Alvin Šipraga" <alsi@bang-olufsen.dk>]

On 8/27/20 12:12 PM, Kalle Valo wrote:
> Alvin Šipraga <alsi@bang-olufsen.dk> writes:
> 
>> Hi Kalle,
>>
>> On 7/30/20 2:49 PM, Alvin Šipraga wrote:
>>> This reverts commit 2dc016599cfa9672a147528ca26d70c3654a5423.
>>>
>>> Per Atheros documentation to manufacturers, the EEPROM regulatory domain
>>> code 0x0 must always map to "US". In particular, it should not map to a
>>> custom world regulatory domain. For references, see [1] and [2] below.
>>> Furthermore, __ath_regd_init() has a specific condition to set the
>>> country code to "US" in this case, which emits the following log
>>> message:
>>>
>>> [    7.814307] ath: EEPROM indicates default country code should be used
>>>
>>> The patch being reverted mistakenly maps 0x0 to the custom world
>>> regulatory domain 0x64 - the most restrictive of the world regulatory
>>> domains. The premise of the patch is that in the case of EEPROM
>>> regulatory domain code 0x0, ath_is_world_regd() should return true. But,
>>> as stated above, 0x0 should not map to a world regulatory domain, and so
>>> the function should return false. The original behaviour, whereby
>>> NL80211_REGDOM_SET_BY_COUNTRY_IE is ignored, was correct according to
>>> the manufacturer's intent and should not have been changed.
>>>
>>> [1] https://wireless.wiki.kernel.org/en/users/drivers/ath#the_0x0_regulatory_domain
>>> [2] http://article.gmane.org/gmane.linux.kernel.wireless.general/38410
>>>
>>> Fixes: 2dc016599cfa ("ath: add support for special 0x0 regulatory domain")
>>> Cc: Wen Gong <wgong@codeaurora.org>
>>> Cc: Luis R. Rodriguez <mcgrof@kernel.org>
>>> Cc: linux-wireless@vger.kernel.org
>>> Tested-on: QCA9880 hw2.0 PCI 10.2.4-1.0-00047
>>> Signed-off-by: Alvin Šipraga <alsi@bang-olufsen.dk>
>>> ---
>>>    drivers/net/wireless/ath/regd.c | 10 +++++-----
>>>    1 file changed, 5 insertions(+), 5 deletions(-)
>>
>> Do you have any feedback on this patch? No problem if you simply have
>> not looked yet - I am not sure what kind of lead time to expect on the
>> list. But without the patch, a (correctly) programmed 0x0 (US) card
>> will not be able to operate on 5GHz channels without some hacking. I
>> have cited some references to justify reverting this patch, so I would
>> like to know if anything further should be done to get this into
>> future kernels?
>>
>> I wonder also if Wen Gong could comment, whose patch I am reverting in
>> the first place. Maybe there is something I am missing?
> 
> I'm working on it, I just need to check something internally first.
> 
> BTW, Brian submitted an identical revert first so I'm planning to use
> his patch instead of yours:
> 
> https://patchwork.kernel.org/patch/11573585/

Hi Kalle,

Thank you for the update, glad to hear it. Sorry that I did not notice 
Brian's patch - I guess you can close my one on patchwork then.

Kind regards,
Alvin