Re: [PATCH 3.7 1/2] cfg80211: fix antenna gain handling

[Felix Fietkau <nbd@openwrt.org>]

On 2012-10-06 10:01 AM, Arend van Spriel wrote:
> On 10/06/2012 01:15 AM, Felix Fietkau wrote:
>> On 2012-10-05 11:02 PM, Arend van Spriel wrote:
>>> On 10/05/2012 10:22 PM, John W. Linville wrote:
>>>> Anyone object?  I suspect that Johannes may be a little more
>>>> busy/distracted than usual for the next few days/weeks...
>>>>
>>>> John
>>>>
>>>> On Wed, Oct 03, 2012 at 07:51:46PM +0200, Felix Fietkau wrote:
>>>>> When not intersecting, orig->chan_mag is initialized to zero. Because of
>>>>> that, the regulatory antenna gain limit never gets set.
>>>>>
>>>>> Signed-off-by: Felix Fietkau <nbd@openwrt.org>
>>>>> Cc: stable@vger.kernel.org
>>>>> ---
>>>>>    net/wireless/reg.c |    3 +--
>>>>>    1 file changed, 1 insertion(+), 2 deletions(-)
>>>>>
>>>>> diff --git a/net/wireless/reg.c b/net/wireless/reg.c
>>>>> index 3b8cbbc..1a16f19 100644
>>>>> --- a/net/wireless/reg.c
>>>>> +++ b/net/wireless/reg.c
>>>>> @@ -915,8 +915,7 @@ static void handle_channel(struct wiphy *wiphy,
>>>>>
>>>>>    	chan->beacon_found = false;
>>>>>    	chan->flags = flags | bw_flags | map_regdom_flags(reg_rule->flags);
>>>>> -	chan->max_antenna_gain = min(chan->orig_mag,
>>>>> -		(int) MBI_TO_DBI(power_rule->max_antenna_gain));
>>>
>>> Just reading the commit message, I would think it better to initialize
>>> chan->orig_mag to INT_MAX. Just my 0.02$
>> I'm not sure it's useful for the driver to be able to add initial max
>> antenna gain restrictions that way (which IMHO is the only point to
>> using chan->orig_mag).
>> I think it's better (and more predictable) to just stick to the power
>> rules from the regulatory information.
>>
>> - Felix
>>
> 
> Ok,
> 
> I was not understanding the usage scenario here. Your patch will ignore 
> any driver initialized max antenna gain, which does not sound right to 
> me. What if chan->orig_mag is a sensible (non-zero) value, which is 
> lower than the value in regulatory power rule? Just seems that the patch 
> changes more than the commit message it saying.
I did take another close look at the code and it seems you're right. I
will change the code so that it initializes chan->orig_mag to INT_MAX in
wiphy_register, and I will make the patch description a bit more verbose
as well.
That way the chan->max_antenna_gain value supplied by the driver gets
ignored, which is a good thing, considering that only one driver
actually uses max_antenna_gain, and no driver initializes it to
something useful.

- Felix