BCM43228 not working (no access points shown)

BCM43228 not working (no access points shown)

[Falco Preiseni]

Hello everyone,

my Broadcom BCM43228 wifi card is not able to show any access points (neither
with "iw wlpx scan" nor with NetworkManager). Connecting with the ESSID does not
work, either. According to the LED and the output of "ip link" the card is
active. Furthermore "lspci -k" tells that "bcma-pci-bridge" is the kernel driver
in use despite I modprobed b43. According to [3] this could be a sign that b43
ignores my device. Or is this label normal?

I followed the instructions on the ArchLinux Wiki[1] (which should align with
the official instructions [2]), installed b43-fwcutter and b43-firmware in order
to obtain the firmware version 5.100.138 and I also tried a newer firmware
(6.30.163.46). Currently I have blacklisted almost everything (b43legacy,
bcma-pci-bridge, bcma, brcmsmac, ssb and b43) and activated "option b43
allhwsupport=1" and modprobed b43 by hand.

I did a lot of reasearch but wasn't able to solve my problem. At least the label
"bcma-pci-bridge" looks like a bug but I don't know. Any idea what's wrong?

Best wishes,
Falco

[1] https://wiki.archlinux.org/index.php/B43#b43.2Fb43legacy
[2] http://wireless.kernel.org/en/users/Drivers/b43
[3]
http://www.linuxforums.org/forum/wireless-internet/188346-b43-requiring-bcma-ssb.html#post889324

$ uname -a
Linux hugo 3.17.6-1-ARCH #1 SMP PREEMPT Sun Dec 7 23:43:32 UTC 2014 x86_64
GNU/Linux
(current ArchLinux kernel from the official repo)

$ lspci -vvn|grep 43 -A7
02:00.0 0280: 14e4:4359
	DeviceName: WLAN
	Subsystem: 14e4:05e2
	Physical Slot: 0
	Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping-
SERR- FastB2B- DisINTx-
	Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort-
<MAbort- >SERR- <PERR- INTx+
	Latency: 0, Cache Line Size: 64 bytes
	Interrupt: pin A routed to IRQ 54

$ lspci -k
...
02:00.0 Network controller: Broadcom Corporation BCM43228 802.11a/b/g/n
	DeviceName: WLAN
	Subsystem: Broadcom Corporation Device 05e2
	Kernel driver in use: bcma-pci-bridge
	Kernel modules: bcma
...

$ dmesg
pasted here: http://paste.opensuse.org/79451668

BCM43228 not working (no access points shown)

[Rafał Miłecki]

On 4 January 2015 at 18:13, Falco Preiseni
<apfelkomplott2014@mailbox.org> wrote:
> my Broadcom BCM43228 wifi card is not able to show any access points (neither
> with "iw wlpx scan" nor with NetworkManager). Connecting with the ESSID does not
> work, either. According to the LED and the output of "ip link" the card is
> active. Furthermore "lspci -k" tells that "bcma-pci-bridge" is the kernel driver
> in use despite I modprobed b43. According to [3] this could be a sign that b43
> ignores my device. Or is this label normal?
>
> I followed the instructions on the ArchLinux Wiki[1] (which should align with
> the official instructions [2]), installed b43-fwcutter and b43-firmware in order
> to obtain the firmware version 5.100.138 and I also tried a newer firmware
> (6.30.163.46). Currently I have blacklisted almost everything (b43legacy,
> bcma-pci-bridge, bcma, brcmsmac, ssb and b43) and activated "option b43
> allhwsupport=1" and modprobed b43 by hand.
>
> I did a lot of reasearch but wasn't able to solve my problem. At least the label
> "bcma-pci-bridge" looks like a bug but I don't know. Any idea what's wrong?

Hi Falco,

Sorry for the late reply, it took me some time to get my dev machine running.

It appears my BCM43228 identifies itself as the same model as yours.
I've tested it and it works fine with 3.17 and 3.19-rc2. I switched to
the same firmware you use and it was still OK. Performance varies (for
most of time it's about 20Mb/s but sometimes drops down to 8Mb/s, but
it works).

Were you using this card successfully with wl.ko? I don't really know
what may be wrong with it :|

BCM43228 not working (no access points shown)

[Falco Preiseni]

Hi Rafa?,

thanks for your efforts :) I tried out the wl driver on ArchLinux but it didn't
work either. However, on Debian the wl driver works with my card:

$ uname -a
Linux noname 3.2.0-4-amd64 #1 SMP Debian 3.2.63-2+deb7u1 x86_64 GNU/Linux

$ dmesg
[   97.463846] wl 0000:02:00.0: setting latency timer to 64
[   97.492386] INFO @wl_cfg80211_attach : Registered CFG80211 phy
[   97.493105] wlan0: Broadcom BCM4359 802.11 Hybrid Wireless Controller
5.100.82.112

On ArchLinux I noticed that there is always (for b43 and wl) the same error
(INVALID_DEVICE_REQUEST) in the dmesg output after the broadcom stuff:
[  587.974839] wl0: online cpus 1
[  587.975382] wlan0: Broadcom BCM4359 802.11 Hybrid Wireless Controller
6.30.223.248 (r487574)
[  587.983443] wl 0000:02:00.0 wlo1: renamed from wlan0
[  588.134778] AMD-Vi: Event logged [INVALID_DEVICE_REQUEST device=00:00.1
address=0x000000fdf80c0020 flags=0x0a00]

I found only one entry on google [1] with the same error and interestingly
enough that user has the same wifi card and laptop model (HP Elitebook 755 G2
with AMD processor). Unfortunately the given solution (add *intremap=off* as
kernel parameter) does not work for me :(

Howsoever: It's not a b43 problem. I don't know...

Regards,
Falco

[1] http://forums.fedoraforum.org/showthread.php?t=300746

> Rafa? Mi?ecki <zajec5@gmail.com> hat am 16. Januar 2015 um 20:03 geschrieben:
> 
> 
> On 4 January 2015 at 18:13, Falco Preiseni
> <apfelkomplott2014@mailbox.org> wrote:
> > my Broadcom BCM43228 wifi card is not able to show any access points
> > (neither
> > with "iw wlpx scan" nor with NetworkManager). Connecting with the ESSID does
> > not
> > work, either. According to the LED and the output of "ip link" the card is
> > active. Furthermore "lspci -k" tells that "bcma-pci-bridge" is the kernel
> > driver
> > in use despite I modprobed b43. According to [3] this could be a sign that
> > b43
> > ignores my device. Or is this label normal?
> >
> > I followed the instructions on the ArchLinux Wiki[1] (which should align
> > with
> > the official instructions [2]), installed b43-fwcutter and b43-firmware in
> > order
> > to obtain the firmware version 5.100.138 and I also tried a newer firmware
> > (6.30.163.46). Currently I have blacklisted almost everything (b43legacy,
> > bcma-pci-bridge, bcma, brcmsmac, ssb and b43) and activated "option b43
> > allhwsupport=1" and modprobed b43 by hand.
> >
> > I did a lot of reasearch but wasn't able to solve my problem. At least the
> > label
> > "bcma-pci-bridge" looks like a bug but I don't know. Any idea what's wrong?
> 
> Hi Falco,
> 
> Sorry for the late reply, it took me some time to get my dev machine running.
> 
> It appears my BCM43228 identifies itself as the same model as yours.
> I've tested it and it works fine with 3.17 and 3.19-rc2. I switched to
> the same firmware you use and it was still OK. Performance varies (for
> most of time it's about 20Mb/s but sometimes drops down to 8Mb/s, but
> it works).
> 
> Were you using this card successfully with wl.ko? I don't really know
> what may be wrong with it :|

BCM43228 not working (no access points shown)

[Rafał Miłecki]

On 17 January 2015 at 20:13, Falco Preiseni
<apfelkomplott2014@mailbox.org> wrote:
> thanks for your efforts :) I tried out the wl driver on ArchLinux but it didn't
> work either. However, on Debian the wl driver works with my card:
>
> $ uname -a
> Linux noname 3.2.0-4-amd64 #1 SMP Debian 3.2.63-2+deb7u1 x86_64 GNU/Linux
>
> $ dmesg
> [   97.463846] wl 0000:02:00.0: setting latency timer to 64
> [   97.492386] INFO @wl_cfg80211_attach : Registered CFG80211 phy
> [   97.493105] wlan0: Broadcom BCM4359 802.11 Hybrid Wireless Controller
> 5.100.82.112
>
> On ArchLinux I noticed that there is always (for b43 and wl) the same error
> (INVALID_DEVICE_REQUEST) in the dmesg output after the broadcom stuff:
> [  587.974839] wl0: online cpus 1
> [  587.975382] wlan0: Broadcom BCM4359 802.11 Hybrid Wireless Controller
> 6.30.223.248 (r487574)
> [  587.983443] wl 0000:02:00.0 wlo1: renamed from wlan0
> [  588.134778] AMD-Vi: Event logged [INVALID_DEVICE_REQUEST device=00:00.1
> address=0x000000fdf80c0020 flags=0x0a00]
>
> I found only one entry on google [1] with the same error and interestingly
> enough that user has the same wifi card and laptop model (HP Elitebook 755 G2
> with AMD processor). Unfortunately the given solution (add *intremap=off* as
> kernel parameter) does not work for me :(
>
> Howsoever: It's not a b43 problem. I don't know...

Wait a second. So are all these tests coming from the same machine? Do
I get it right that:
1) Debian (3.2.0) + wl.ko (5.100.82.112) = GOOD
2) ArchLinux + wl.ko (6.30.223.248) = BAD
3) ArchLinux + b43.ko = BAD
?

If so it would mean that:
1) Upgrading kernel broke something
2) Upgrading wl.ko broken something

Do you still have this Debian? Could you try following setup:
Debian (3.2.0) + wl.ko (6.30.223.248)
and let us know if it works?

-- 
Rafa?

BCM43228 not working (no access points shown)

[Falco Preiseni]

Sorry for the late reply..

> On 19 January 2015 at 09:06 Rafa? Mi?ecki <zajec5@gmail.com> wrote:
> Wait a second. So are all these tests coming from the same machine? Do
> I get it right that:
> 1) Debian (3.2.0) + wl.ko (5.100.82.112) = GOOD
> 2) ArchLinux + wl.ko (6.30.223.248) = BAD
> 3) ArchLinux + b43.ko = BAD
> ?
Yes, that's correct and it's the same machine.

> Do you still have this Debian? Could you try following setup:
> Debian (3.2.0) + wl.ko (6.30.223.248)
> and let us know if it works?
This also works. I installed the broadcom-sta-dkms package from testing as it
contains the above version...

BCM43228 not working (no access points shown)

[Rafał Miłecki]

On 26 January 2015 at 15:36, Falco Preiseni
<apfelkomplott2014@mailbox.org> wrote:
> Sorry for the late reply..
>
>> On 19 January 2015 at 09:06 Rafa? Mi?ecki <zajec5@gmail.com> wrote:
>> Wait a second. So are all these tests coming from the same machine? Do
>> I get it right that:
>> 1) Debian (3.2.0) + wl.ko (5.100.82.112) = GOOD
>> 2) ArchLinux + wl.ko (6.30.223.248) = BAD
>> 3) ArchLinux + b43.ko = BAD
>> ?
> Yes, that's correct and it's the same machine.
>
>> Do you still have this Debian? Could you try following setup:
>> Debian (3.2.0) + wl.ko (6.30.223.248)
>> and let us know if it works?
> This also works. I installed the broadcom-sta-dkms package from testing as it
> contains the above version...

So this could be some kernel regression. Probably related to your
machine. Too bad, as it means I won't be able to resolve this problem.

You should bisect kernel marking 3.2 as GOOD and whatever version
ArchLinux uses as BAD. This will of course require compiling kernel
for many times (10? 20?) and trying version by version with the
Broadcom's wl driver.

-- 
Rafa?

BCM43228 not working (no access points shown)

[Falco Preiseni]

Ok, thank you for your detailed help :)