2017-12-04 22:02:42

by Hin-Tak Leung

[permalink] [raw]
Subject: Re: [PATCH] rtl8187: add master mode

--------------------------------------------
On Mon, 4/12/17, Gabriele Gristina <[email protected]> wrote:

> I use these few
> byte from years and hostapd work like a charm.I
> don't understand why you write "completely
> wrong" but you are free to refuse this
> patch.
> Have a nice
> day,Gabriele

I think the issue was to do with hardware support of certain features - power-management (allowing the clients to sleep) and maintaining states across low-power states, and somewhat large number of clients. That it can work to some extent, and even to a very good extent, to serve a small number of clients - at full power, without power managrment - is not a surprise.

My impression that software (i.e. non-hardware) support was to improve to make it possible. Larry and others can comment on the current state.




2017-12-04 22:29:05

by Larry Finger

[permalink] [raw]
Subject: Re: [PATCH] rtl8187: add master mode

On 12/04/2017 04:02 PM, Hin-Tak Leung wrote:
> --------------------------------------------
> On Mon, 4/12/17, Gabriele Gristina <[email protected]> wrote:
>
>> I use these few
>> byte from years and hostapd work like a charm.I
>> don't understand why you write "completely
>> wrong" but you are free to refuse this
>> patch.
>> Have a nice
>> day,Gabriele
>
> I think the issue was to do with hardware support of certain features - power-management (allowing the clients to sleep) and maintaining states across low-power states, and somewhat large number of clients. That it can work to some extent, and even to a very good extent, to serve a small number of clients - at full power, without power managrment - is not a surprise.
>
> My impression that software (i.e. non-hardware) support was to improve to make it possible. Larry and others can comment on the current state.

Gabriele,

I agree that merely informing mac80211 that you support AP mode is unlikely to
provide full support for master mode, but I have not looked at that code for
many years.

If you can show that your "fix" does support many clients, and that it does not
go belly-up when one of them tries to go to low power, then I will agree to
merge your patch. Until then, merging it now would suggest that we are offering
a service that the driver does not really support.

BTW, why are you trying to create an AP with a 802.11G device? The performance
will be on par with what was expected 10 or 12 years ago.

Larry