Interaction issue of intel wifi and broadcom bluetooth - it appears that nobody feels responsible for doing something

Interaction issue of intel wifi and broadcom bluetooth - it appears that nobody feels responsible for doing something

[Jonas Thiem]

Hi *,

this is my first post to this mailing list, sorry if it's not supposed
to go here. (also CC in responses would be nice since I'm not
subscribed)

I filed a bug about an intel centrino wifi interaction with broadcom's BCM2045B:
https://bugzilla.kernel.org/show_bug.cgi?id=97101

In short, the two seem to kinda fight over the wireless spectrum and
both drop connections all the time - unless the 'iwlwifi' module is
loaded with 11n_disabled=1.

Despite this easy workaround which I would assume also could give some
sort of hint where the problem might be, all I got so far is a
short-worded "Intel can't do much here. Removing Intel. You can ask
for support from Broadcom. Good luck." from the intel person and
infinite silence from whomever is responsible for bluetooth in the
kernel.

Therefore I wanted to ask whom I should file this bug to, and which
sort of sleeping dragon I might need to poke so that more than either
silence or rejection happens.

After all, both wifi and bluetooth headphones working at the same time
is a neat thing. Especially since there's a workaround available, but
the drivers are in their current state apparently not smart enough to
do something like that on their own.

Regards,
Jonas Thiem

Re: Interaction issue of intel wifi and broadcom bluetooth - it appears that nobody feels responsible for doing something

[Jeremiah Mahler]

Jonas,

On Fri, Jun 05, 2015 at 01:00:32AM +0200, Jonas Thiem wrote:
> Hi *,
> 
> this is my first post to this mailing list, sorry if it's not supposed
> to go here. (also CC in responses would be nice since I'm not
> subscribed)
> 
> I filed a bug about an intel centrino wifi interaction with broadcom's BCM2045B:
> https://bugzilla.kernel.org/show_bug.cgi?id=97101
> 
Those are some unhelpful replies :-(

> In short, the two seem to kinda fight over the wireless spectrum and
> both drop connections all the time - unless the 'iwlwifi' module is
> loaded with 11n_disabled=1.
> 

I don't have a solution but I think the problem is interesting.

Both Bluetooth and 11n share the same frequency band near 2.4 GHz so it
is possible that they could conflict.  If you had two laptops, and you
ran just Bluetooth on one and just 11n on the other, would they
both have problems?  This would tell as whether it was something inside
the kernel or if it was really wireless interference.

[...]
> 
> Regards,
> Jonas Thiem

-- 
- Jeremiah Mahler

Re: Interaction issue of intel wifi and broadcom bluetooth - it appears that nobody feels responsible for doing something

[Jonas Thiem]

[-- Attachment #1: Type: text/plain, Size: 1975 bytes --]

Hi Jeremiah,

thanks for responding!

I did have my mobile phone very nearby also connected to the bluetooth
headphones while my laptop was still using 11n wifi. I didn't have any
noticeable issues with bluetooth there.

But I got the feeling that my phone's android drivers + hardware for
bluetooth are tuned better than the laptop ones, so maybe that just
means the phone is just better at jumping frequencies to avoid.

I guess the best test would be the same laptop model in direct
proximity, but sadly I only own that laptop once. ;)

I hope wireless interference wouldn't rule out that some driver work
would be considered to make it work better - after all, both chips are
in the same laptop and as per the intel comment, bluetooth is supposed
to work despite of wifi activity.

Regards,
Jonas Thiem

On 06/05/2015 06:45 AM, Jeremiah Mahler wrote:
> Jonas,
> 
> On Fri, Jun 05, 2015 at 01:00:32AM +0200, Jonas Thiem wrote:
>> Hi *,
>>
>> this is my first post to this mailing list, sorry if it's not supposed
>> to go here. (also CC in responses would be nice since I'm not
>> subscribed)
>>
>> I filed a bug about an intel centrino wifi interaction with broadcom's BCM2045B:
>> https://bugzilla.kernel.org/show_bug.cgi?id=97101
>>
> Those are some unhelpful replies :-(
> 
>> In short, the two seem to kinda fight over the wireless spectrum and
>> both drop connections all the time - unless the 'iwlwifi' module is
>> loaded with 11n_disabled=1.
>>
> 
> I don't have a solution but I think the problem is interesting.
> 
> Both Bluetooth and 11n share the same frequency band near 2.4 GHz so it
> is possible that they could conflict.  If you had two laptops, and you
> ran just Bluetooth on one and just 11n on the other, would they
> both have problems?  This would tell as whether it was something inside
> the kernel or if it was really wireless interference.
> 
> [...]
>>
>> Regards,
>> Jonas Thiem
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: Interaction issue of intel wifi and broadcom bluetooth - it appears that nobody feels responsible for doing something

[Arend van Spriel]

+ Marcel, wireless list, and bt list.

On 06/05/15 07:12, Jonas Thiem wrote:
> Hi Jeremiah,
>
> thanks for responding!
>
> I did have my mobile phone very nearby also connected to the bluetooth
> headphones while my laptop was still using 11n wifi. I didn't have any
> noticeable issues with bluetooth there.
>
> But I got the feeling that my phone's android drivers + hardware for
> bluetooth are tuned better than the laptop ones, so maybe that just
> means the phone is just better at jumping frequencies to avoid.
>
> I guess the best test would be the same laptop model in direct
> proximity, but sadly I only own that laptop once. ;)
>
> I hope wireless interference wouldn't rule out that some driver work
> would be considered to make it work better - after all, both chips are
> in the same laptop and as per the intel comment, bluetooth is supposed
> to work despite of wifi activity.
>
> Regards,
> Jonas Thiem
>
> On 06/05/2015 06:45 AM, Jeremiah Mahler wrote:
>> Jonas,
>>
>> On Fri, Jun 05, 2015 at 01:00:32AM +0200, Jonas Thiem wrote:
>>> Hi *,
>>>
>>> this is my first post to this mailing list, sorry if it's not supposed
>>> to go here. (also CC in responses would be nice since I'm not
>>> subscribed)
>>>
>>> I filed a bug about an intel centrino wifi interaction with broadcom's BCM2045B:
>>> https://bugzilla.kernel.org/show_bug.cgi?id=97101
>>>
>> Those are some unhelpful replies :-(

Indeed. Now my reply maybe of a similar nature as my knowledge of the 
bluetooth part is next to nil.

>>> In short, the two seem to kinda fight over the wireless spectrum and
>>> both drop connections all the time - unless the 'iwlwifi' module is
>>> loaded with 11n_disabled=1.
>>>

I guess you don't have a choice about the channel your access point is 
operating on. If you can you might move wifi channel to 5GHz although 
that will likely affect your range.

To get more info on your setup here a couple of questions.

What type of device are we talking about here?
Bugzilla mentions Thinkpad X220, is that correct?
Is your AP setup for 40MHz bandwidth?

I suppose both wifi and bt have their own antennas. When sharing 
antennas it needs to be coordinated who is using the antennas. This is 
called bt-coex. As said I don't think this is the case for your device, 
but if so I think there is still major development task needed as I saw 
email from Marcel about plans to develop a bt-coex subsystem in the 
kernel. Not sure if that also covers the case where bt and wifi do not 
share antennas.

Regards,
Arend

>> I don't have a solution but I think the problem is interesting.
>>
>> Both Bluetooth and 11n share the same frequency band near 2.4 GHz so it
>> is possible that they could conflict.  If you had two laptops, and you
>> ran just Bluetooth on one and just 11n on the other, would they
>> both have problems?  This would tell as whether it was something inside
>> the kernel or if it was really wireless interference.
>>
>> [...]
>>>
>>> Regards,
>>> Jonas Thiem
>>
>

Re: Interaction issue of intel wifi and broadcom bluetooth - it appears that nobody feels responsible for doing something

[Marcel Holtmann]

Hi Arend,

>> thanks for responding!
>> 
>> I did have my mobile phone very nearby also connected to the bluetooth
>> headphones while my laptop was still using 11n wifi. I didn't have any
>> noticeable issues with bluetooth there.
>> 
>> But I got the feeling that my phone's android drivers + hardware for
>> bluetooth are tuned better than the laptop ones, so maybe that just
>> means the phone is just better at jumping frequencies to avoid.
>> 
>> I guess the best test would be the same laptop model in direct
>> proximity, but sadly I only own that laptop once. ;)
>> 
>> I hope wireless interference wouldn't rule out that some driver work
>> would be considered to make it work better - after all, both chips are
>> in the same laptop and as per the intel comment, bluetooth is supposed
>> to work despite of wifi activity.
>> 
>> Regards,
>> Jonas Thiem
>> 
>> On 06/05/2015 06:45 AM, Jeremiah Mahler wrote:
>>> Jonas,
>>> 
>>> On Fri, Jun 05, 2015 at 01:00:32AM +0200, Jonas Thiem wrote:
>>>> Hi *,
>>>> 
>>>> this is my first post to this mailing list, sorry if it's not supposed
>>>> to go here. (also CC in responses would be nice since I'm not
>>>> subscribed)
>>>> 
>>>> I filed a bug about an intel centrino wifi interaction with broadcom's BCM2045B:
>>>> https://bugzilla.kernel.org/show_bug.cgi?id=97101
>>>> 
>>> Those are some unhelpful replies :-(
> 
> Indeed. Now my reply maybe of a similar nature as my knowledge of the bluetooth part is next to nil.
> 
>>>> In short, the two seem to kinda fight over the wireless spectrum and
>>>> both drop connections all the time - unless the 'iwlwifi' module is
>>>> loaded with 11n_disabled=1.
>>>> 
> 
> I guess you don't have a choice about the channel your access point is operating on. If you can you might move wifi channel to 5GHz although that will likely affect your range.
> 
> To get more info on your setup here a couple of questions.
> 
> What type of device are we talking about here?
> Bugzilla mentions Thinkpad X220, is that correct?
> Is your AP setup for 40MHz bandwidth?
> 
> I suppose both wifi and bt have their own antennas. When sharing antennas it needs to be coordinated who is using the antennas. This is called bt-coex. As said I don't think this is the case for your device, but if so I think there is still major development task needed as I saw email from Marcel about plans to develop a bt-coex subsystem in the kernel. Not sure if that also covers the case where bt and wifi do not share antennas.

from the report, we have to make the assumption that in this design it is two independent antennas. Meaning Bluetooth is going to win the spectrum war. It should play nice due to AFH support in almost every 4.0 device in the market, but there are limits. If your spectrum is busy, then it becomes really hard for AFH to function properly.

So in the case of two independent antennas, the rfres subsystem I am proposing would help indeed. We could tell the Bluetooth controller to block out the frequency that the WiFi chip is currently operating on.

Main problem is that these design are so rare these days that nobody actually cared enough to add this subsystem. If Bluetooth and WiFi are sharing the same antenna, the coex issue is solved by a direct connection between both chips.

Regards

Marcel