cvs commit: src/sys/dev/acpica acpi_quirks acpi_timer.c acpivar.h

Nate Lawson nate at root.org
Sat Oct 9 20:33:08 PDT 2004


M. Warner Losh wrote:
> In message: <4168A67A.2090805 at root.org>
>             Nate Lawson <nate at root.org> writes:
> : M. Warner Losh wrote:
> : 
> : > In message: <4167178E.3050303 at root.org>
> : >             Nate Lawson <nate at root.org> writes:
> : > : Some slots force irq sharing but if you have devices that can't handle 
> : > : that irq, they fail.  This is especially on older systems that have an 
> : > : ISA and PCI slot that share an irq line.
> : > 
> : > Are there such systems like this that actually work?  We've been
> : > telling people that they can't share PCI and ISA interrupts because
> : > ISA is edge triggered and PCI is level triggered.  
> : 
> : My Abit BP6 has such a configuration for the last slot.  You can only 
> : put a PCI card OR ISA card in, not both.
> 
> And you MUST use an IRQ that is used by PCI for that ISA slot?  The
> ISA slots don't share an IRQ with PCI at all.  They might share the
> slot, but they don't share interrupts because the ISA card dispatches
> the interrupts directly through the IRQ lines on the ISA bus, and the
> PCI dispatches it through #INT[A-D] through the bridges between the
> device and CPU.

I think I'm confusing two different things.  For the shared slot, you 
can't put more than one card in at a time.  The IRQ conflict was between 
AGP and PCI slot 1 and had nothing to do with ISA.

-Nate


More information about the cvs-src mailing list