[PATCH 1/8] irq: choose irq_domain type for generic-chip based on irq_base

[robherring2@gmail.com (Rob Herring)]

On 02/11/2012 11:14 AM, Shawn Guo wrote:
> Choosing irq_domain type for generic-chip based on device_node forces
> non-DT users to use legacy irq_domain, which makes no sense, because
> linear irq_domain is actually encouraged to be used by both DT and
> non-DT users.

NAK

Encouraged by who? As we discussed in person at Connect, I said you
should use legacy domain for non-DT case. Grant and I have discussed
this as well and agree.

> The patch changes it to make the decision based on irq_base.  If users
> pass in a negative irq_base value, a linear irq_domain will be created,
> otherwise a legacy irq_domain will be created.

This allows DT users to decide as well and possibly use legacy domains
which is explicitly what I was trying to prevent. DT users must use
linear domains and non-DT must use legacy. Otherwise, we'll get more
half done DT conversions like mx5 is.

Getting linear domains to work for non-DT to work would be a worthwhile
goal if we didn't plan to remove non-DT boards. I don't think we really
need linear to work for non-DT case.

Rob

> Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
> Cc: Rob Herring <rob.herring@calxeda.com>
> Cc: Grant Likely <grant.likely@secretlab.ca>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> ---
>  kernel/irq/generic-chip.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/kernel/irq/generic-chip.c b/kernel/irq/generic-chip.c
> index 3ac7fa1..839f882 100644
> --- a/kernel/irq/generic-chip.c
> +++ b/kernel/irq/generic-chip.c
> @@ -346,7 +346,7 @@ int irq_setup_generic_chip_domain(const char *name, struct device_node *node,
>  		irq_setup_generic_chip(gc[i], 0, flags, clr, set);
>  	}
>  
> -	if (node)
> +	if ((int) irq_base < 0)
>  		d = irq_domain_add_linear(node, hwirq_cnt,
>  					  &irq_gc_irq_domain_ops, gc);
>  	else