> > Level-triggered interrupts are persistent conditions, not queueable 
> > events.  They typically require device-driver level intervention to be
> > 
> > cleared.  This is a major error in the PCI design (no surprises
> > there).
> > 
>       [ML]  Whoa there!  That's the MAJOR advantage of PCI design.
> Open collector, active low, level triggered interrupts are the only
> possibility for interupt line sharing without programmatically
> accessible registers on card which say "yes, I am still interrupting".

For a simplistic bus, perhaps.  But an arbitrated token-delivery
interrupt buslet with a bus-standardised interrupt state acknowledgement
protocol would be much more efficient.

PCI makes too many compromises to the PC's architecture; we're just 
about ready for a new bus again.

>       Active high, edge triggered interrupts are an abomination (there
> is no way to reliably share the interrupt line and you cannot even wire
> or it).  They are the reason why one never has enough interrupt lines on
> ISA.

I don't believe I ever suggested that this was the only alternative 
technique, and I'm certainly on record as not liking it either.

-- 
\\  The mind's the standard       \\  Mike Smith
\\  of the man.                   \\  [email protected]
\\    -- Joseph Merrick           \\  [email protected]




To Unsubscribe: send mail to [email protected]
with "unsubscribe freebsd-hackers" in the body of the message

Reply via email to