Re: [gentoo-user] Broadcom wireless problem after a certain time

[Mick]

On Sunday 01 Jan 2017 11:54:22 Hogren wrote:
> On 31/12/2016 00:21, Mick wrote:
> > On Friday 30 Dec 2016 21:22:17 Hogren wrote:
> >> Hello,
> >>
> >> I write to the list for a little strange problem.
> >>
> >> I have a wireless card :
> >> # lspci |grep -i 802
> >> 24:00.0 Network controller: Broadcom Corporation BCM43228 802.11a/b/g/n
> >>
> >>
> >> Globally, the card works fine. I use wpa_supplicant. I have a systemd
> >> service to start the connection when I want. The DHCP client works fine
> >> too. etc. All is right.
> >>
> >> But, because there is always a "but", After a certain quantity of
> >> downloaded data, the card's stopping to work fine. I can't receive
> >> anything. I need to stop an restart the systemd service.
> >>
> >> I herd that there is many problems with Broadcom wireless devices
> >> because the driver is only based on reverse engineering.
> >>
> >> That is the reason ? And there is a solution for this kind of problem ?
> >>
> >> Thank you very much !
> >>
> >> Hogren
> >
> > I don't know what problems are being reported on the model you are using,
> > but previous models were experiencing problems with power saving.  Use
> > modinfo to find out what power saving parameters the module has and try
> > disabling these, in case power management is the cause of the
> > interruptions you are experiencing.
>
> I'm not enough experienced to see that and deactivate it.
>
> My modinfo output :
> # modinfo b43
> filename:
> /lib/modules/4.4.26-gentoo/kernel/drivers/net/wireless/b43/b43.ko
> firmware:       b43/ucode9.fw
> firmware:       b43/ucode5.fw
> firmware:       b43/ucode16_mimo.fw
> firmware:       b43/ucode15.fw
> firmware:       b43/ucode14.fw
> firmware:       b43/ucode13.fw
> firmware:       b43/ucode11.fw
> license:        GPL
> author:         Rafał Miłecki
> author:         Gábor Stefanik
> author:         Michael Buesch
> author:         Stefano Brivio
> author:         Martin Langer
> description:    Broadcom B43 wireless driver
> alias:          ssb:v4243id0812rev10*
> alias:          ssb:v4243id0812rev0F*
> alias:          ssb:v4243id0812rev0D*
> alias:          ssb:v4243id0812rev0C*
> alias:          ssb:v4243id0812rev0B*
> alias:          ssb:v4243id0812rev0A*
> alias:          ssb:v4243id0812rev09*
> alias:          ssb:v4243id0812rev07*
> alias:          ssb:v4243id0812rev06*
> alias:          ssb:v4243id0812rev05*
> alias:          bcma:m04BFid0812rev2Acl*
> alias:          bcma:m04BFid0812rev28cl*
> alias:          bcma:m04BFid0812rev1Ecl*
> alias:          bcma:m04BFid0812rev1Dcl*
> alias:          bcma:m04BFid0812rev1Ccl*
> alias:          bcma:m04BFid0812rev18cl*
> alias:          bcma:m04BFid0812rev17cl*
> alias:          bcma:m04BFid0812rev15cl*
> alias:          bcma:m04BFid0812rev11cl*
> depends:
> intree:         Y
> vermagic:       4.4.26-gentoo SMP mod_unload
> parm:           bad_frames_preempt:enable(1) / disable(0) Bad Frames
> Preemption (int)
> parm:           fwpostfix:Postfix for the .fw files to load. (string)
> parm:           hwpctl:Enable hardware-side power control (default off)
> (int)
[snip ...]

The above parameter is the only one I see related to power management.  So, 
let's see what the card is configured to use at present:

cat /sys/class/net/wlp36s0b1/power/control

The default setting would be off and since this parameter is specified as an 
integer it is likely to be 0.  There is also a possibility of it being set at 
'auto'.  In any case you can try switching this on, to see if it is makes any 
difference, or if it is set at 'auto' force it to off.

I see you are running systemd.  I am not well versed with it, so others will 
hopefully chip in to correct me.  On a normal Gentoo system with openrc I 
would use this file to set options:

 /etc/modprobe.d/b43.conf

So create it and add in it:

options b43 hwpctl=1

to switch power management on, or:

options b43 hwpctl=0

to force it off.

Then unload/reload the module using modprobe, or reboot and see if with either 
setting the problem goes away.

PS.  I am using the same module, but with a different card.  I noticed on mine 
I had set qos=0 for some reason.  I must have found this advice somewhere on 
the interwebs and set it so.  You may want to try this too to see if it makes 
any odds.

PPS.  You may also want to increase verbosity in the logs in the first instance 
to see if something is reported as a result of the fault you are getting.  Use 
this setting:

options b43 verbose=3

to increase verbosity.
-- 
Regards,
Mick