Re: [PATCH 1/2] ap: Remove opt-out DisableHT for opt-in EnableHT

From: James Prestwood <prestwoj@gmail.com>
To: Denis Kenzior <denkenz@gmail.com>, iwd@lists.linux.dev
Subject: Re: [PATCH 1/2] ap: Remove opt-out DisableHT for opt-in EnableHT
Date: Tue, 6 Jun 2023 09:11:30 -0700	[thread overview]
Message-ID: <4318c465-abde-2057-0974-007a504fdc16@gmail.com> (raw)
In-Reply-To: <5a290730-f50b-cb27-e488-3ed605aa831a@gmail.com>

On 6/6/23 8:40 AM, Denis Kenzior wrote:
> Hi James,
> 
>> This deserves further investigation but its looking like more
>> often than not users need to add DisableHT. This isn't so bad
>> for AP profiles since the user has to create a profile anyways,
>> but for the Start command its best if it "just works". To error
>> on the side of caution DisableHT is being renamed to EnableHT
>> making it opt-in rather than opt-out.
> 
> Our philosophy has always been to have the user perform the least amount 
> of work for the best possible result.  Most modern hardware should 
> support HT these days, so the 'just works' approach is to enable HT 
> whenever possible and not make it 'opt-in'.  The current settings naming 
> is just fine in my opinion.
> 
> The ideal solution to the above problem would be to figure out why HT 
> isn't working on certain drivers and fix that.

I agree.

> 
> If not possible, the next best approach would be to create a list of 
> drivers we know do not work (or vice-versa) and do not attempt HT on the 
> problematic ones and print a warning to the log.
> 
> Do we know which drivers don't work by the way?  I might take a look 
> when / if I have time.

So far complaints have been with brcmfmac (raspi), mwifiex (NXP driver 
out of tree), and one other I'm not sure of (I'll ask).

I did dig into the NXP driver somewhat. Its out of tree so take with a 
grain of salt, but I was able to get it more or less working by first 
issuing NL80211_CMD_SET_WIPHY with the freq/channel width before 
START_AP (and leave both attributes out of START_AP). This is how 
hostapd behaves as well depending on the configuration file. Since this 
seemed like a one-off or maybe legacy behavior so I didn't even bother 
sending a patch, but maybe more investigation is needed.

I tested on iwlwifi, realtek and ath9k.

When I have some time I'll get my raspi set up and see if I can 
reproduce and compare with hostapd.

> 
> Regards,
> -Denis