Re: [PATCHv4 0/2] cfg80211: mac80211: BTCOEX feature support

Re: [PATCHv4 0/2] cfg80211: mac80211: BTCOEX feature support

[Arend Van Spriel]

+ linux-bluetooth, Marcel

On 11-4-2017 12:15, c_traja@qti.qualcomm.com wrote:
> From: Tamizh chelvam <c_traja@qti.qualcomm.com>
> 
> This patchset add support for BTCOEX feature to enable/disable and
> modifying btcoex priority value via nl80211

So you make a distinction between WMM ACs, but what about the different
types/profiles of BT traffic?

Regards,
Arend

> Tamizh chelvam (2):
>   cfg80211: Add support to enable or disable btcoex and set
>     btcoex_priority
>   mac80211: Add support to enable or disable btcoex and set btcoex
>     priority value
> 
> v4 :
>   * Moved btcoex_priority_support_flag enum to nl80211.h and renamed it.
>   * fixed typo.
> 
> v3 :
>   * Introduced NL80211_EXT_FEATURE_BTCOEX_PRIORITY to expose
>     btcoex priority support and removed bool variable.
> 
> v2 :
>   * Introduced NL80211_CMD_SET_BTCOEX to enable/disable btcoex and
>     to set/modify btcoex_priority.
>   * Added bool variable in wiphy structure to advertise btcoex_priority
>     feature and removed BITMAP calculation for btcoex_priority value.
> 
>  include/net/cfg80211.h       |    6 +++++
>  include/net/mac80211.h       |    5 +++++
>  include/uapi/linux/nl80211.h |   50 ++++++++++++++++++++++++++++++++++++++++++
>  net/mac80211/cfg.c           |    9 ++++++++
>  net/mac80211/driver-ops.h    |   15 +++++++++++++
>  net/mac80211/trace.h         |   20 +++++++++++++++++
>  net/wireless/nl80211.c       |   45 +++++++++++++++++++++++++++++++++++++
>  net/wireless/rdev-ops.h      |   13 +++++++++++
>  net/wireless/trace.h         |   17 ++++++++++++++
>  9 files changed, 180 insertions(+)
> 

RE: [PATCHv4 0/2] cfg80211: mac80211: BTCOEX feature support

[Tamizh Chelvam Raja]

> + linux-bluetooth, Marcel
>=20
> On 11-4-2017 12:15, c_traja@qti.qualcomm.com wrote:
> > From: Tamizh chelvam <c_traja@qti.qualcomm.com>
> >
> > This patchset add support for BTCOEX feature to enable/disable and
> > modifying btcoex priority value via nl80211
>=20
> So you make a distinction between WMM ACs, but what about the different
> types/profiles of BT traffic?
>
[Tamizh] There will be BT high and BT low traffic. It will be decided by BT=
 module. Firmware internally checks BT low traffic with wlan traffic. If we=
 enable some of
wlan frames as high priority, those frames will have more priority than BT =
low traffic.
=20
> Regards,
> Arend
>=20
> > Tamizh chelvam (2):
> >   cfg80211: Add support to enable or disable btcoex and set
> >     btcoex_priority
> >   mac80211: Add support to enable or disable btcoex and set btcoex
> >     priority value
> >
> > v4 :
> >   * Moved btcoex_priority_support_flag enum to nl80211.h and renamed it=
.
> >   * fixed typo.
> >
> > v3 :
> >   * Introduced NL80211_EXT_FEATURE_BTCOEX_PRIORITY to expose
> >     btcoex priority support and removed bool variable.
> >
> > v2 :
> >   * Introduced NL80211_CMD_SET_BTCOEX to enable/disable btcoex and
> >     to set/modify btcoex_priority.
> >   * Added bool variable in wiphy structure to advertise btcoex_priority
> >     feature and removed BITMAP calculation for btcoex_priority value.
> >
> >  include/net/cfg80211.h       |    6 +++++
> >  include/net/mac80211.h       |    5 +++++
> >  include/uapi/linux/nl80211.h |   50
> ++++++++++++++++++++++++++++++++++++++++++
> >  net/mac80211/cfg.c           |    9 ++++++++
> >  net/mac80211/driver-ops.h    |   15 +++++++++++++
> >  net/mac80211/trace.h         |   20 +++++++++++++++++
> >  net/wireless/nl80211.c       |   45
> +++++++++++++++++++++++++++++++++++++
> >  net/wireless/rdev-ops.h      |   13 +++++++++++
> >  net/wireless/trace.h         |   17 ++++++++++++++
> >  9 files changed, 180 insertions(+)
> >

Re: [PATCHv4 0/2] cfg80211: mac80211: BTCOEX feature support

[Johannes Berg]

On Wed, 2017-04-12 at 07:08 +0000, Tamizh Chelvam Raja wrote:
> > 
> > So you make a distinction between WMM ACs, but what about the
> > different
> > types/profiles of BT traffic?
> > 
> 
> [Tamizh] There will be BT high and BT low traffic. It will be decided
> by BT module. Firmware internally checks BT low traffic with wlan
> traffic. If we enable some of wlan frames as high priority, those
> frames will have more priority than BT low traffic.

That ("firmware internally..." etc.) really sounds more like an
argument *not* to apply this patch ...

Everyone has their favourite BT coex control. We could possibly
implement this in our driver, but I'm not sure we'd *want* to, since
it's so far from what we actually do today.

Do we really need more than toggling it on/off?

Actually, I probably should've asked this much earlier - but what use
cases do you see for this? What can a user, or userspace application
like NM, really try to set here? It seems the use cases for this would
be rather constrained?

johannes