With 2.6.24-rc5, a b43 user reports a problem at bootup. The b43 module, which should be loaded by
the ssb module, fails with the following type of message:
ssb: Sonics Silicon Backplane found on PCI device 0000:0c:00.0
b43: disagrees about version of symbol ssb_device_is_enabled
b43: Unknown symbol ssb_device_is_enabled
b43: disagrees about version of symbol ssb_pcicore_dev_irqvecs_enable
b43: Unknown symbol ssb_pcicore_dev_irqvecs_enable
b43: disagrees about version of symbol ssb_bus_may_powerdown
b43: Unknown symbol ssb_bus_may_powerdown
< and more similar lines....>
If the user issues the commands
rmmod ssb; modprobe b43
then the modules are loaded correctly and the wireless device works.
What could be causing such a problem?
Larry
> What could be causing such a problem?
I would guess that user has "ssb" loaded from his initrd for some reason
and upgraded the modules by hand.
johannes
On Sunday 23 December 2007 20:39:18 Larry Finger wrote:
> With 2.6.24-rc5, a b43 user reports a problem at bootup. The b43 module, which should be loaded by
> the ssb module, fails with the following type of message:
>
> ssb: Sonics Silicon Backplane found on PCI device 0000:0c:00.0
> b43: disagrees about version of symbol ssb_device_is_enabled
> b43: Unknown symbol ssb_device_is_enabled
> b43: disagrees about version of symbol ssb_pcicore_dev_irqvecs_enable
> b43: Unknown symbol ssb_pcicore_dev_irqvecs_enable
> b43: disagrees about version of symbol ssb_bus_may_powerdown
> b43: Unknown symbol ssb_bus_may_powerdown
This is a module versioning warning.
I think it's caused by a b43 compiled against a different
kernel tree than the currently running ssb module.
Different may mean that only a tiny little thing was changed, which
might not affect the API at all. modversion will complain then.
I don't use module versioning, so I can't help you how to fix this.
I'd simply suggest turning off module versioning.
--
Greetings Michael.
Larry Finger wrote:
> With 2.6.24-rc5, a b43 user reports a problem at bootup. The b43
> module, which should be loaded by the ssb module, fails with the
> following type of message:
>
> ssb: Sonics Silicon Backplane found on PCI device 0000:0c:00.0 b43:
> disagrees about version of symbol ssb_device_is_enabled b43: Unknown
> symbol ssb_device_is_enabled b43: disagrees about version of symbol
> ssb_pcicore_dev_irqvecs_enable b43: Unknown symbol
> ssb_pcicore_dev_irqvecs_enable b43: disagrees about version of symbol
> ssb_bus_may_powerdown b43: Unknown symbol ssb_bus_may_powerdown
>
> < and more similar lines....>
>
> If the user issues the commands
>
> rmmod ssb; modprobe b43
>
> then the modules are loaded correctly and the wireless device works.
>
> What could be causing such a problem?
>
> Larry
>
Larry, thanks for pursuing this. I upgraded to 2.6.24-rc6 last night and
I don't see this issue anymore. One might think that the clean remake
solved the problem (which I am pretty sure I did with rc5 as well); but
still does not explain with 2.6.24-rc5, the module would not load on
first attempt but then load after the manual:
# rmmod ssb;modprobe b43
Best regards,
Shourya
Shourya Sarcar wrote:
> Larry Finger wrote:
>> With 2.6.24-rc5, a b43 user reports a problem at bootup. The b43
>> module, which should be loaded by the ssb module, fails with the
>> following type of message:
>>
>> ssb: Sonics Silicon Backplane found on PCI device 0000:0c:00.0 b43:
>> disagrees about version of symbol ssb_device_is_enabled b43: Unknown
>> symbol ssb_device_is_enabled b43: disagrees about version of symbol
>> ssb_pcicore_dev_irqvecs_enable b43: Unknown symbol
>> ssb_pcicore_dev_irqvecs_enable b43: disagrees about version of symbol
>> ssb_bus_may_powerdown b43: Unknown symbol ssb_bus_may_powerdown
>>
>> < and more similar lines....>
>>
>> If the user issues the commands
>>
>> rmmod ssb; modprobe b43
>>
>> then the modules are loaded correctly and the wireless device works.
>>
>> What could be causing such a problem?
>>
>> Larry
>>
>
> Larry, thanks for pursuing this. I upgraded to 2.6.24-rc6 last night and
> I don't see this issue anymore. One might think that the clean remake
> solved the problem (which I am pretty sure I did with rc5 as well); but
> still does not explain with 2.6.24-rc5, the module would not load on
> first attempt but then load after the manual:
>
> # rmmod ssb;modprobe b43
Johannes may have been right and that you had a stale initrd. I was going to have you try a 'sudo
mkinitrd', but it is now fixed.
Larry