Hi Grant,

On 1/6/2012 10:22 PM, Grant Likely wrote:

On Tue, Dec 20, 2011 at 02:39:55PM +0100, Benoit Cousson wrote:

[...]

quoted

+	/*
+	 * XXX: Use a 0 irq_base for the moment since the legacy devices
+	 * created statically are expected a hwirq = irq mapping.
+	 * A proper offset will be added later, when IRQ resource creation
+	 * will be handled by DT.
+	 */
+	domain.irq_base = 0;
+	domain.nr_irq = nr_irqs;
+	domain.ops =&irq_domain_simple_ops;
+	irq_domain_add(&domain);

The omap doesn't need anything special from irq_domain.  You can use
irq_domain_add_simple().  A recent patch was posted that makes
irq_domain_add_simple() accepts a nr_irqs parameter and returns the
allocated irq domain.

Yes, I saw that, but in fact as discussed previously with Rob, even the 
irq_domain_add_simple should become useless when Rob's series to convert 
generic-chip to use irq_domain will be merged.

I have a patch ready for that, but like Nico for AT91, I decided to 
avoid that dependency for 3.3, since that series was not yet stable.

However, that isn't enough reason for me to reject this patch if there
is any opportunity to get it into v3.3 provided you promise to fix it
up later.

Yes, I do :-)

Acked-by: Grant Likely<redacted>

Thanks,
Benoit