Re: Cleanups for APIC

[Maciej W. Rozycki]

On Thu, 27 May 2004, Ingo Molnar wrote:

> >  The I/O APIC need not be hooked to PCI ;-) -- I'm not sure about the
> > i82093AA, but that's definitely true for the i82489DX.  The call to
> > io_apic_sync() is needed for masking to make sure interrupts won't be
> > dispatched after returning from the call -- this is not needed for
> > unmasking as a delay here is harmless.
> 
> well, an APIC message could be on the way to the CPU even with this
> synchronization. Does it matter whether it's a newly dispatched one due
> to POST delays or an in-fly one due to APIC bus delays?

 Well, if you'd mask, sync, ack (send EOI) in a handler, then the sync
would assure the ack wouldn't be in effect before masking, so no further
interrupt would arrive till unmasking.  It would work for level-triggered
interrupts and the i82093AA, but OTOH for the i82489DX, which uses
level-deassert messages, it wouldn't.

 Too much hassle for an unreliable result...  Just scrap it.

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@xxxxxxxxxxxxx, PGP key available        +
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/