From mboxrd@z Thu Jan  1 00:00:00 1970
From: robherring2@gmail.com (Rob Herring)
Date: Tue, 20 Nov 2012 17:42:47 -0600
Subject: [PATCH 02/16] arm: add set_handle_irq() to register the parent
 IRQ controller handler function
In-Reply-To: <1353448867-15008-3-git-send-email-thomas.petazzoni@free-electrons.com>
References: <1353448867-15008-1-git-send-email-thomas.petazzoni@free-electrons.com>
 <1353448867-15008-3-git-send-email-thomas.petazzoni@free-electrons.com>
Message-ID: <50AC1577.4080007@gmail.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On 11/20/2012 04:00 PM, Thomas Petazzoni wrote:
> In order to allow irqchip drivers to register their IRQ handling
> function as the parent IRQ controller handler function, we provide a
> convenience function. This will avoid poking directly into the global
> handle_arch_irq variable, and ensures that the user gets warned if
> we're trying to register a second IRQ controller as the parent one.
> 
> Suggested by Arnd Bergmann.
> 
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
> ---
>  arch/arm/include/asm/mach/irq.h |    1 +
>  arch/arm/kernel/irq.c           |   10 ++++++++++
>  2 files changed, 11 insertions(+)
> 
> diff --git a/arch/arm/include/asm/mach/irq.h b/arch/arm/include/asm/mach/irq.h
> index 15cb035..18c8830 100644
> --- a/arch/arm/include/asm/mach/irq.h
> +++ b/arch/arm/include/asm/mach/irq.h
> @@ -22,6 +22,7 @@ extern int show_fiq_list(struct seq_file *, int);
>  
>  #ifdef CONFIG_MULTI_IRQ_HANDLER
>  extern void (*handle_arch_irq)(struct pt_regs *);
> +extern void set_handle_irq(void (*handle_irq)(struct pt_regs *));
>  #endif
>  
>  /*
> diff --git a/arch/arm/kernel/irq.c b/arch/arm/kernel/irq.c
> index 8961650..f135dab 100644
> --- a/arch/arm/kernel/irq.c
> +++ b/arch/arm/kernel/irq.c
> @@ -117,6 +117,16 @@ void __init init_IRQ(void)
>  	machine_desc->init_irq();
>  }
>  
> +#ifdef CONFIG_MULTI_IRQ_HANDLER
> +void __init set_handle_irq(void (*handle_irq)(struct pt_regs *))
> +{
> +	if (WARN_ON(handle_arch_irq))
> +		return;

I wonder if we should just return silently without warning and rely on
the top level controller being initialized 1st. For the DT case, this is
guaranteed. In the non-DT case, we don't really know if we are the
top-level controller at least in a consistent way. AFAIK, it happens
that the 1st instance of the VIC or GIC is always the top level and we
can use that (i.e. gic_nr == 0). However, theoretically you could have a
VIC connected to a top-level GIC or even vice-versa.

Rob

> +
> +	handle_arch_irq = handle_irq;
> +}
> +#endif
> +
>  #ifdef CONFIG_SPARSE_IRQ
>  int __init arch_probe_nr_irqs(void)
>  {
>