23.03.2026 05:22, Kirill A. Korinsky пишет:
> On Sun, 22 Mar 2026 22:00:50 +0100,
> Stefan Sperling <stsp@stsp.name> wrote:
>>
>> On Sun, Mar 22, 2026 at 09:56:18PM +0100, Stefan Sperling wrote:
>>> On Sun, Mar 22, 2026 at 09:07:37PM +0100, Kirill A. Korinsky wrote:
>>>> And in beacon I see: Channel Width: 80 MHz, 160 MHz or 80+80 MHz BSS
>>>> Bandwidth (1).
>>>
>>> I checked our net80211 headers first, which still list value 2 as 160 MHz.
>>>
>>> Turns out the 802.11 spec has since deprecated values 2 and 3. Wireshark is
>>> correct that this value by itself doesn't mean anything and can be set to 1
>>> in all cases. We need to check the two bytes following this value specify
>>> the center frequency of two 80 MHz portions of a 160 MHz channel.
>>
>> And in case it wasn't clear, this means that the function
>> ieee80211_node_supports_vht_chan160() will always return false at present
>> because it checks for the deprecated value 2 here:
>>
>> 	return (op_chan_width == IEEE80211_VHTOP0_CHAN_WIDTH_160);
>>
>> Which would explain why 160 MHz is not actually being used for transmit
>> with your patch.
>>
> 
> Thanks for direction, it allows to cook a better diff.
> 
> With it my unifi network says that this laptop is connected to network with
> Rx Rate 1.56 Gbps and Tx Rate 1.30 Gbps.
> 
> I had added instruments to decode PHY width/NSS/MCS/SGI and let say on my
> iperf tests like 95% of frames were with 160Mhz window.
> 
> But I still limited for ~400 Mbits/sec (without this diff it is ~350).
> 
> My main thought that we have hardcoded MPDU as 3895, when Linux driver uses
> 11454. Plus it may need a bit more bits to set inside vhtcaps. I only
> touched antenna patterns and TX/RX STBC with following Linux logic.
> 
> Frankly, I think that optimization of speed should be the next diff, and
> keep this one is focused on 160Mhz window.

This version works better:

Notebook
1733 Мбит/с WPA2
11ac 2x2 160 МГц
	[  5]   0.00-10.00  sec   542 MBytes   455 Mbits/sec   13            sender
	[  5]   0.00-10.01  sec   540 MBytes   453 Mbits/sec                  receiver

Thanks!