concurrency advertising -- driver maintainers

concurrency advertising -- driver maintainers

[Johannes Berg]

Please excuse the long CC list. I want to make you aware that I plan to
remove the WIPHY_FLAG_ENFORCE_COMBINATIONS flag again soon, so that all
drivers that don't advertise valid combinations will not be able to have
multiple virtual interfaces.

The affected drivers, as far as I can tell, are:
 * ath5k
 * ath9k
 * carl9170
 * hwsim (I'll look at this)
 * iwlagn (I'm handling that of course)
 * mwl8k
 * rt2x00 (see RFC patch)

I'm happy to help out with the restrictions code, please let me know.

I'd like to remove the flag again before it ever goes upstream, but I
understand the timing is bad, unless we hold my patch out of 2.6.40.

For completeness, I checked all drivers, and the following drivers don't
support more than a single virtual interface anyway:
 * adm8211
 * at76c50x-usb -- but it is broken because it has no checks
 * b43
 * b43legacy
 * iwlegacy
 * libertas_tf
 * p54
 * rtl8180
 * rtl8187
 * rtlwifi
 * wl1251
 * wl12xx
 * zd1211
(Staging drivers:)
 * brcm80211 is broken, allows multiple but can't handle it
 * winbond is also broken the same way

johannes

Re: concurrency advertising -- driver maintainers

[Johannes Berg]

On Fri, 2011-05-13 at 14:44 +0200, Johannes Berg wrote:

> The affected drivers, as far as I can tell, are:

I forgot ath9k_htc, which needs to advertise something afaict (two
interfaces seem to be supported)

johannes

Re: concurrency advertising -- driver maintainers

[Sujith]

Johannes Berg wrote:
> On Fri, 2011-05-13 at 14:44 +0200, Johannes Berg wrote:
> 
> > The affected drivers, as far as I can tell, are:
> 
> I forgot ath9k_htc, which needs to advertise something afaict (two
> interfaces seem to be supported)

The limitation of two is artificial, more can actually be supported.
The firmware has to be updated, though.

Sujith

Re: concurrency advertising -- driver maintainers

[Johannes Berg]

On Fri, 2011-05-13 at 20:02 +0530, Sujith wrote:
> Johannes Berg wrote:
> > On Fri, 2011-05-13 at 14:44 +0200, Johannes Berg wrote:
> > 
> > > The affected drivers, as far as I can tell, are:
> > 
> > I forgot ath9k_htc, which needs to advertise something afaict (two
> > interfaces seem to be supported)
> 
> The limitation of two is artificial, more can actually be supported.
> The firmware has to be updated, though.

Ok, but no matter what the limitation is or whether it can be changed it
should be advertised with the new advertising feature I submitted,
that's what I wanted you to do :)

johannes

Re: concurrency advertising -- driver maintainers

[Sujith]

Johannes Berg wrote:
> Ok, but no matter what the limitation is or whether it can be changed it
> should be advertised with the new advertising feature I submitted,
> that's what I wanted you to do :)

Yep, I got that. :)

Sujith

Re: concurrency advertising -- driver maintainers

[Gertjan van Wingerde]

On Fri, May 13, 2011 at 2:44 PM, Johannes Berg
<johannes@sipsolutions.net> wrote:
> Please excuse the long CC list. I want to make you aware that I plan to
> remove the WIPHY_FLAG_ENFORCE_COMBINATIONS flag again soon, so that all
> drivers that don't advertise valid combinations will not be able to have
> multiple virtual interfaces.
>
> The affected drivers, as far as I can tell, are:
>  * ath5k
>  * ath9k
>  * carl9170
>  * hwsim (I'll look at this)
>  * iwlagn (I'm handling that of course)
>  * mwl8k
>  * rt2x00 (see RFC patch)
>
> I'm happy to help out with the restrictions code, please let me know.
>
> I'd like to remove the flag again before it ever goes upstream, but I
> understand the timing is bad, unless we hold my patch out of 2.6.40.
>
> For completeness, I checked all drivers, and the following drivers don't
> support more than a single virtual interface anyway:
>  * adm8211
>  * at76c50x-usb -- but it is broken because it has no checks
>  * b43
>  * b43legacy
>  * iwlegacy
>  * libertas_tf
>  * p54
>  * rtl8180
>  * rtl8187
>  * rtlwifi
>  * wl1251
>  * wl12xx
>  * zd1211
> (Staging drivers:)
>  * brcm80211 is broken, allows multiple but can't handle it
>  * winbond is also broken the same way
>

I seem to have missed the patch that introduces this advertisement of
valid combinations
in mac80211. Do you have a pointer towards this patch and/or some of
the documentation
for it?

That would make it a lot easier to see if your RFC patch for rt2x00
actually makes sense ;-)

---
Gertjan

Re: concurrency advertising -- driver maintainers

[Johannes Berg]

On Fri, 2011-05-13 at 16:44 +0200, Gertjan van Wingerde wrote:

> I seem to have missed the patch that introduces this advertisement of
> valid combinations
> in mac80211. Do you have a pointer towards this patch and/or some of
> the documentation
> for it?
> 
> That would make it a lot easier to see if your RFC patch for rt2x00
> actually makes sense ;-)

:-)

I submitted it a little bit earlier, here:
http://thread.gmane.org/gmane.linux.kernel.wireless.general/69917

johannes