[Asterisk-Users] Intel Desktop MotherBoards Unsuitable for DigiumBoards

[George Pajari]

Kevin P. Fleming wrote:

> 'lspci -vb' does not understand IO-APIC mode, as best I can tell, so 
> the interrupt number that it reports is totally useless.
>
> On my desktop machine, with an nVidia graphics card in a PCI-Express 
> slot, /proc/interrupts shows it using interrupt 185, and 'lspci -vb' 
> shows it using IRQ 10. It is most definitely _not_ on IRQ 10, since 
> /proc/interrupts show the actual mapping from the IRQ controller (the 
> APIC) to the driver servicing the device.
>
> At best, 'lspci -vb' is showing what interrupt the device was on 
> _before_ the kernel reassigned it using APIC mode.

Mr. Fleming:

I am aware of all of the above but does the fact you are posting this 
mean that Digium is now aware of this?

I have had a Digium tech ssh to a machine, run an lspci -vb, report 
"your interrupts are shared" and refuse to work on the case until the 
lspci -vb showed unique IRQs.

Does your post mean that this policy has now changed at Digium?

As Piotr A. Sygula wrote in another related post:

>I.e. although APIC is splitting up IRQ's rather nicely, the tech support guy
>is saying that it doesn't matter what the APIC layer says.  Would someone
>out there "break the tie"?  I'd like an educated opinion/statement on
>whether APIC support solves the IRQ sharing issue, or simply masks it.
>

Which suggests other customers have been told by Digium that lspci -vb 
is the final word on whether IRQs are being shared and whether or not 
your hardware configuration is supported by Digium.

So what's the real story -- does APIC count or not? Will Digium support 
systems with unique APIC interrupts but not unique lspci -vb interrupts?

-- 
George Pajari, netVOICE communications    604 484 VOIP (484 8647 x102)
Open Source VoIP/Telephony Specialists  1 877 NET VOIP (638 8647 x102)
                  www.netvoice.ca  www.ip-centrex.ca
      www.digium.ca www.grandstream.ca www.sipura.ca www.snom.ca