Re: [PATCH] Prevent nested interrupts when the IRQ stack is nearoverflowing v2

From: Linus Torvalds
Date: Thu Mar 25 2010 - 16:58:27 EST




On Thu, 25 Mar 2010, Linus Torvalds wrote:
>
> And I just checked: even the 8250 serial driver doesn't use IRQF_DISABLED
> any more, so doing that shouldn't cause any latency issues (sure, the
> serial driver may interrupt another irq, but another irq can also
> interrupt the serial driver as things stand now, so the original latency
> issue with fast irq handlers doesn't actually work these days _anyway_).

Btw, if we really do decide that everybody is IRQF_DISABLED, that really
should make the whole mask-and-ack issue for the irq controllers go away.

We'd still need it, but only for the very special IDE controller case and
others who _explicitly_ re-enable interrupts. Those would really have to
cause the irq to be masked so that we don't get issues with recursive irqs
of the same kind.

Linus
--
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/