Bug in the b43 driver (BCM4311 on Ubuntu 3.11.0-13-generic)?

["Christoph Rüßler" <christoph.ruessler@googlemail.com>]

Am 27.11.2013 21:32, schrieb Larry Finger:
> On 11/27/2013 09:21 AM, Christoph R??ler wrote:
>> Hello,
>>
>> I have a Broadcom BCM4311 wifi card and I'm using the b43 kernel
>> driver with it
>> (Ubuntu 3.11.0-13-generic). Unfortunately my system seems to have
>> problems
>> loading the driver correctly at boot time. In most cases it does not
>> properly
>> load the driver on boot making it necessary to use the radio
>> killswitch to
>> manually stop and restart the wifi which then works perfectly and
>> without any
>> further disruptions.
>>
>> The bug is not reproducible in a way that that I am not able to
>> predict if it
>> will happen after the next boot. I am also not able to reproduce it
>> once I
>> have successfully enabled the wifi after a failed boot. Since I only
>> started
>> using the b43 driver with the kernel I'm still using I do not know if
>> it would
>> have worked with previous kernels which were using the Broadcom driver.
>>
>> The only hint I have so far found in debugging is a small difference
>> in the
>> dmesg output between working and non-working boot sequences. When the
>> wifi
>> fails to start dmesg contains the following lines:
>>> [   16.169637] b43-phy0: Broadcom 4311 WLAN found (core revision 13)
>> [?]
>>> [   16.217018] b43-phy0: Found PHY: Analog 4, Type 2 (G), Revision 9
>> [?]
>>> [   16.246721] Broadcom 43xx driver loaded [ Features: PNL ]
>>
>> Using the radio killswitch under these circumstances yields the following
>> output:
>>> [  496.421727] pci 0000:30:00.0: [14e4:4312] type 00 class 0x028000
>>> [  496.421782] pci 0000:30:00.0: reg 0x10: [mem 0x00000000-0x00003fff
>>> 64bit]
>>> [  496.421930] pci 0000:30:00.0: supports D1 D2
>>> [  496.422004] pci 0000:30:00.0: System wakeup disabled by ACPI
>>> [  496.428993] pci 0000:30:00.0: BAR 0: assigned [mem
>>> 0xc8000000-0xc8003fff
>> 64bit]
>>> [  496.577397] b43-pci-bridge 0000:30:00.0: enabling device (0000 ->
>>> 0002)
>>> [  496.596096] ssb: Found chip with id 0x4311, rev 0x02 and package 0x00
>>> [  496.596111] ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x13, vendor
>> 0x4243)
>>> [  496.596123] ssb: Core 1 found: IEEE 802.11 (cc 0x812, rev 0x0D,
>>> vendor
>> 0x4243)
>>> [  496.596136] ssb: Core 2 found: USB 1.1 Host (cc 0x817, rev 0x04,
>>> vendor
>> 0x4243)
>>> [  496.596148] ssb: Core 3 found: PCI-E (cc 0x820, rev 0x05, vendor
>>> 0x4243)
>>> [  496.660259] b43-phy1: Broadcom 4311 WLAN found (core revision 13)
>>> [  496.704083] b43-phy1: Found PHY: Analog 4, Type 2 (G), Revision 9
>>> [  496.728352] ssb: Sonics Silicon Backplane found on PCI device
>> 0000:30:00.0
>>> [  496.729194] ieee80211 phy1: Selected rate control algorithm
>>> 'minstrel_ht'
>>> [  496.972075] b43-phy1: Loading firmware version 666.2 (2011-02-23
>>> 01:15:07)
>>
>> When everything is working as expected I can find the following in the
>> dmesg
>> output:
>>> [   16.115666] b43-phy0: Broadcom 4311 WLAN found (core revision 13)
>> [?]
>>> [   16.160089] b43-phy0: Found PHY: Analog 4, Type 2 (G), Revision 9
>> [?]
>>> [   16.184214] Broadcom 43xx driver loaded [ Features: PNL ]
>> [?]
>>> [   24.738141] pci 0000:30:00.0: [14e4:4312] type 00 class 0x028000
>>> [   24.738203] pci 0000:30:00.0: reg 0x10: [mem 0x00000000-0x00003fff
>>> 64bit]
>>> [   24.738350] pci 0000:30:00.0: supports D1 D2
>>> [   24.738425] pci 0000:30:00.0: System wakeup disabled by ACPI
>>> [   24.755897] pci 0000:30:00.0: BAR 0: assigned [mem
>>> 0xc8000000-0xc8003fff
>> 64bit]
>>> [   24.809549] b43-pci-bridge 0000:30:00.0: enabling device (0000 ->
>>> 0002)
>>> [   24.896511] ssb: Found chip with id 0x4311, rev 0x02 and package 0x00
>>> [   24.896538] ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x13, vendor
>> 0x4243)
>>> [   24.896551] ssb: Core 1 found: IEEE 802.11 (cc 0x812, rev 0x0D,
>>> vendor
>> 0x4243)
>>> [   24.896563] ssb: Core 2 found: USB 1.1 Host (cc 0x817, rev 0x04,
>>> vendor
>> 0x4243)
>>> [   24.896574] ssb: Core 3 found: PCI-E (cc 0x820, rev 0x05, vendor
>>> 0x4243)
>>> [   24.960281] b43-phy1: Broadcom 4311 WLAN found (core revision 13)
>>> [   25.004095] b43-phy1: Found PHY: Analog 4, Type 2 (G), Revision 9
>>> [   25.028328] ssb: Sonics Silicon Backplane found on PCI device
>> 0000:30:00.0
>>> [   25.029050] ieee80211 phy1: Selected rate control algorithm
>>> 'minstrel_ht'
>>> [   25.236092] b43-phy1: Loading firmware version 666.2 (2011-02-23
>>> 01:15:07)
>>
>> Am I right in suspecting that this may be a bug in the driver? What
>> can I do
>> in order to help debugging? And if it's not a bug in the driver what
>> could be
>> the reason for this behaviour?
> 
> I do not see this problem with b43 on my BCM4311. What is probably
> different is that I use openSUSE (12.2, 12.3, and 13.1) with
> NetworkManager. I suspect that you are getting a timing error between
> the initial attempt to bring the interface up, and the availability of
> the user-space code needed to load the firmware.
> 
> Larry
> 
> 

Thanks for the suggestion, that seems very plausible. As I am using
NetworkManager too, could this be a NetworkManager problem? If I
understand it correctly this is a case where my wifi should "just work".
Anyway, I will try to delay NetworkManager's startup on boot to give the
device more time to be ready.

Thanks again,
Christoph