Re: In kernel PIC support: kernel patch

[Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>]

Gregory Haskins wrote:
> On Fri, 2007-06-22 at 10:09 +0800, Dong, Eddie wrote:
>   
>> Avi Kivity wrote:
>>     
>>> Eddie, is that what you were driving at in your simplification
>>> attempt? 
>>>
>>>       
>> Yes, some minor thing:
>> KVM_ISA_INTERRUPT: Per VM I/F, and need an irq line and irq level.
>> KVM_IOAPIC_INTERRUPT: Per VM I/F, may also need irq level for level
>> triggered irq.
>> KVM_APIC_MESSAGE: Per VCPU I/F
>>     
>
> Note that in my implementation the APIC_MESSAGE is also VM I/F.   In
> fact, IMHO the only API that should be VCPU I/F is the original API,
> KVM_INTERRUPT.  The reason is, the PIC related interrupts are targeted
> at the vm-global ISA resources, and the APIC_MESSAGE is targeted at the
> vm-global apic *bus*, not a particular lapic (yet).  Its the bus that
> then determines the actual routing to its final destination.  I think
> this design distinction is important.
>
>   

Good point.  I agree.

>   
>> Per step by step approach way, KVM_EXTINT (assume for LINT0/LINT1) can
>> be deferred to future implementation.
>>     
>
> Unless I misunderstood Avi, I think he was suggesting that use use
> KVM_EXTINT/ISA_INTERRUPT as a way to distinguish between the dual modes
> of ISA_INTERRUPT as I have today (e.g.
> ISA_INTERRUPT(level-1)/ISA_INTERRUPT(level-2)).  This doesn't really
> have anything to do with LINT0/1 (directly, anyway).  Hope this helps to
> clarify. 
>   

Well, we don't need to think in terms of modes.  Each ioctl controls
interrupts at a different point in the architecture.

KVM_INTERRUPT is at the interface between the processor core and the
lapic/system bus.

KVM_APIC_MESSAGE is at the apic bus.

KVM_EXTINT is at the interface between the lapic and the system bus.  If
the guest disabled the local apic, it's equivalent to the KVM_INTERRUPT
(injecting a vector directly); otherwise it is handled by the lapic
(LINT0/1 handling).

KVM_ISA_INTERRUPT (maybe KVM_IRQ_LINE, to signify it doesn't carry a
vector) is at the boundary between the interrupt controller chips and
the rest of the system.

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.


-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/