Hi,
CC'ing lots of people who touched or used this in the past in hopes of
triggering some reaction somewhere ...
I'm trying to do some cleanup in IBSS and following that some other
cleanups wrt. station allocation etc., but OCB pretty much copied the
IBSS code in this area, and I don't know how to use it, how to test it,
who's using it, if anyone is actually maintaining it, etc.
Also, it only ever got implemented for ath9k, so I'm guessing it's not
getting any traction in new products/devices.
So I'm probably going to remove it.
Any takers to help maintain it instead?
johannes
On 9/28/2023 8:38 AM, Johannes Berg wrote:
> Hi,
>
> CC'ing lots of people who touched or used this in the past in hopes of
> triggering some reaction somewhere ...
>
> I'm trying to do some cleanup in IBSS and following that some other
> cleanups wrt. station allocation etc., but OCB pretty much copied the
> IBSS code in this area, and I don't know how to use it, how to test it,
> who's using it, if anyone is actually maintaining it, etc.
>
> Also, it only ever got implemented for ath9k, so I'm guessing it's not
> getting any traction in new products/devices.
>
> So I'm probably going to remove it.
>
> Any takers to help maintain it instead?
Do you have an estimate on the footprint of this logic? Do you have a
proposed patch?
In addition to ath9k I know there is out-of-tree usage of this
functionality, and I'm trying to see if I can find someone to maintain
this. But how would that actually work?
/jeff
Hi Carlo,
Thanks for your response! I'm quoting it fully below for the benefit of
the list which dropped your response since it was HTML.
On Fri, 2023-09-29 at 16:40 +0200, Carlo Augusto Grazia wrote:
> Dear Johannes,
> Thank you for this email!
> I'm in the "people who use/used" it.
:)
> I've tested the OCB performance in some scientific papers and used the OCB mode in
> the laboratory part of my course on Automotive Connectivity at the University of Modena and Reggio Emilia (Italy).
> I'm using it in "low-cost laboratory" mode with Arduino Yuns and a proper version of LEDE/OpenWRT, indeed, as you said,
> the Yun device mounts ath9k.
Right.
> I don't know if will be ever implemented a DSRC/WAVE/802.11p mode on Linux, allowing low-cost hardware to perform as On Board Unit
> for my tests, I'll probably always have to use Codha Wireless devices or SDR properly configured for doing this...
>
> I've always tried to see if the OCB on ath9k was going to be ported at least on ath9k_htc, but nothing...so I got your point (maybe), an IBSS code well maintained for more
> drivers is better than an OCB code working only for ath9k.
>
> Also, I've never maintained a driver, so I don't know if I can help, but I'm really interested in the direction this story will take.
OK, fair enough. I was mostly just putting out feelers.
Honestly the thing isn't even necessarily maintaining it, but as it
stands, I don't even know of any way to _run_ the code, which is a bit
sad.
Is all the setup that you had very complex?
johannes
On Fri, 2023-10-06 at 16:15 -0700, Jeff Johnson wrote:
> On 9/28/2023 8:38 AM, Johannes Berg wrote:
> > Hi,
> >
> > CC'ing lots of people who touched or used this in the past in hopes of
> > triggering some reaction somewhere ...
> >
> > I'm trying to do some cleanup in IBSS and following that some other
> > cleanups wrt. station allocation etc., but OCB pretty much copied the
> > IBSS code in this area, and I don't know how to use it, how to test it,
> > who's using it, if anyone is actually maintaining it, etc.
> >
> > Also, it only ever got implemented for ath9k, so I'm guessing it's not
> > getting any traction in new products/devices.
> >
> > So I'm probably going to remove it.
> >
> > Any takers to help maintain it instead?
>
> Do you have an estimate on the footprint of this logic? Do you have a
> proposed patch?
Neither. It's actually not really big, and I was just putting out
feelers.
The particular reason was that we had a patch a long time back:
https://patchwork.kernel.org/project/linux-wireless/patch/[email protected]/
which was rightfully rejected. But I was looking at it recently again,
thinking I should revive it (and I may still), and then I thought if I
do that, I can remove the gfp_t argument to sta_info_alloc(), but OCB
identically copied this code ...
Now, I suppose OCB doesn't have much in the way of station
discovery/connection process (like auth) since that's kind of the
_point_, but maybe doing something else would work here?
> In addition to ath9k I know there is out-of-tree usage of this
> functionality, and I'm trying to see if I can find someone to maintain
> this. But how would that actually work?
Well for starters it'd be nice to actually have some tests with hwsim, I
guess, and someone who knows how to use it, how to test things if there
are relevant changes, etc.
johannes