[sclug] Initialising wireless network interface problem

[Will Dickson]

Neil Haughton wrote:
> 
>>
>>
>> Just a random thought: some older VIA kit in particular has dodgy 
>> ACPI. Have you tried setting pci=noacpi as a kernel parameter, or even 
>> acpi=off ?
> 
> I've now tried pci=noacpi and so far the symptoms haven' t returned - 
> although it's probably too early to tell if it has cured the problem. 
> But I'm interested - why should a power management protocol effect 
> initialising a wireless network interface, when it doesn't affect the 
> conventional ethernet interface, or any other on board or plugged-in 
> devices?

AIUI ACPI has some kind of effect on the way in which PCI interrupts are 
dealt with. Whether this is because it's handled by the same chip or for 
some other reason I've no idea - I've never gotten round to following 
the thing up. All I do know is that VIA Athlon mobos generally won't 
boot unless this is set. (VIA Athlon64 mobos seem to be OK though).

> Good question, and those were my initial thoughts too - but I don't know 
> enough about 'low level' Linux to be able to tell how this could be 
> done. Can anyone tell me? Presumably a simple 'pause for n secs' in the 
> startup scripts just before the network interfaces are initialised is 
> not what you mean.

Well, actually it was :-) Lots of stuff is forking off - just slowing 
everything down a little and allowing stuff to catch a breath might 
help. I'm guessing here.

Will.