On Tue, 16 May 2000, Jan Kasprzak wrote:
>
> One minor issue, though: I think the IRQs should be allocated
> in a more clever way. Currently almost everything in my computer
> is given an IRQ 11 (in 2.2 each device had its own IRQ, so different
> wiring is definitely possible):
In 2.2, the external PCMCIA package used the legacy mode irq routing,
which is indeed more programmable, but it has some rather bad behaviour
like the fact that it is not sharable and it's edge-triggered (think ISA
here). We also don't generally have good enough information about it to do
a really good job of choosing an interrupt.
In short, the legacy mode ISA interrupts definitely have some problems. On
my VAIO, for example, they would cause lost interrupts with the tulip
driver.
So the new yenta driver always uses the PCI interrupt for all 32-bit
true CardBus cards. Which is imho definitely the right thing to do, but
because of the way CardBus PCI interrupts work you will always find all
cardbus devices sharing the same interrupt with the actual cardbus
controller.
Note that the controller only sends interrupts for exceptional events
(card insert, remove etc), and the controller irq handler is very simple,
so this is unlikely to actually impact performance in any negative way.
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue May 23 2000 - 21:00:10 EST