On Monday 26 March 2007 21:29, Eric W. Biederman wrote:
quoted text
> "Luck, Tony" <tony.luck@intel.com> writes:
> 
> >> What I'm proposing we do is move the irq allocation code out of
> >> pci_enable_device and the irq freeing code out of pci_disable_device
> >> in the future.
> >
> > Sounds rational ... in a world that wasn't dominated by PCI it would
> > seem to be the logical approach (since the irq code would have much
> > more utility independent of the PCI code).
> 
> Right.  We can even do this earlier in the pci code.  Just doing this
> on demand when the device driver needs it is problematic.  As devices
> drivers like to keep the requested over a pci_disable_device pci_enable_device
> pair.
> 
> The big practical issue is that we will like wind up allocating an irq
> number to all usable irqs on ia64.  Which means we will like need many
> more irq numbers...  Although I guess if we keep it at the pci layer
> we should be fairly safe.

The main reason we wait until pci_enable_device() to allocate an
IRQ number is that ia64 currently only has about 180 device vectors,
and there are machines with more PCI slots than that.

I also think it's nice that we don't do anything with a device until
we have a driver to claim it.  But there certainly have been cases
where delaying IRQ allocation has caused troubles.

I really like the idea of moving to the IRQ == GSI model for ia64.
But of course, we'll have to get rid of the 180-vector limit to
make that work, too.

Bjorn
-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/