From mboxrd@z Thu Jan  1 00:00:00 1970
From: jason@lakedaemon.net (Jason Cooper)
Date: Mon, 23 Jun 2014 09:07:05 -0400
Subject: [PATCH v3 1/7] genirq: generic chip: export
 irq_map_generic_chip function
In-Reply-To: <53A68CF0.8040303@free-electrons.com>
References: <1403276487-8792-1-git-send-email-boris.brezillon@free-electrons.com>
 <1403276487-8792-2-git-send-email-boris.brezillon@free-electrons.com>
 <20140621235108.GJ21711@titan.lakedaemon.net>
 <53A68CF0.8040303@free-electrons.com>
Message-ID: <20140623130705.GM21711@titan.lakedaemon.net>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On Sun, Jun 22, 2014 at 09:59:44AM +0200, Boris BREZILLON wrote:
> 
> On 22/06/2014 01:51, Jason Cooper wrote:
> > On Fri, Jun 20, 2014 at 05:01:21PM +0200, Boris BREZILLON wrote:
> >> Export the generic irq map function in order to provide irq_domain ops with
> >> generic mapping and specific of xlate function (needed by the new atmel
> >> AIC driver).
> >>
> >> Signed-off-by: Boris BREZILLON <boris.brezillon@free-electrons.com>
> >> ---
> >>  include/linux/irq.h       | 2 ++
> >>  kernel/irq/generic-chip.c | 5 +++--
> >>  2 files changed, 5 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/include/linux/irq.h b/include/linux/irq.h
> >> index 0d998d8..62af592 100644
> >> --- a/include/linux/irq.h
> >> +++ b/include/linux/irq.h
> >> @@ -771,6 +771,8 @@ void irq_gc_eoi(struct irq_data *d);
> >>  int irq_gc_set_wake(struct irq_data *d, unsigned int on);
> >>  
> >>  /* Setup functions for irq_chip_generic */
> >> +int irq_map_generic_chip(struct irq_domain *d, unsigned int virq,
> >> +			 irq_hw_number_t hw_irq);
> >>  struct irq_chip_generic *
> >>  irq_alloc_generic_chip(const char *name, int nr_ct, unsigned int irq_base,
> >>  		       void __iomem *reg_base, irq_flow_handler_t handler);
> >> diff --git a/kernel/irq/generic-chip.c b/kernel/irq/generic-chip.c
> >> index 452d6f2..cf80e7b 100644
> >> --- a/kernel/irq/generic-chip.c
> >> +++ b/kernel/irq/generic-chip.c
> >> @@ -341,8 +341,8 @@ static struct lock_class_key irq_nested_lock_class;
> >>  /*
> >>   * irq_map_generic_chip - Map a generic chip for an irq domain
> >>   */
> >> -static int irq_map_generic_chip(struct irq_domain *d, unsigned int virq,
> >> -				irq_hw_number_t hw_irq)
> >> +int irq_map_generic_chip(struct irq_domain *d, unsigned int virq,
> >> +			 irq_hw_number_t hw_irq)
> >>  {
> >>  	struct irq_data *data = irq_get_irq_data(virq);
> >>  	struct irq_domain_chip_generic *dgc = d->gc;
> >> @@ -394,6 +394,7 @@ static int irq_map_generic_chip(struct irq_domain *d, unsigned int virq,
> >>  	irq_modify_status(virq, dgc->irq_flags_to_clear, dgc->irq_flags_to_set);
> >>  	return 0;
> >>  }
> >> +EXPORT_SYMBOL_GPL(irq_map_generic_chip);
> >>  
> >>  struct irq_domain_ops irq_generic_chip_ops = {
> >>  	.map	= irq_map_generic_chip,
> > Why can't you use irq_generic_chip_ops.map in your code and avoid this

s/code/declaration/, sorry for the misunderstanding.

> > patch entirely?
> 
> Because in this case I'll have to remove constness from my
> irq_domain_ops struct and initialize it in my init function.
> This is not a big concern, but in general I tend to declare ops struct
> with a const constraint.

/*
 * We're the only user of irq_map_generic_chip() who
 * doesn't also use irq_domain_xlate_onetwocell()
 */
static const struct irq_domain_ops aic_irq_ops = {
	.map    = irq_generic_chip_ops.map,
        .xlate  = aic_irq_domain_xlate,
};

Wouldn't work?

> Tell me if you still want me to drop this patch and rework my init process.

If the above is satisfactory for you, please do so.  If we get a few
more chips that need one function, but not the other, then we'll revisit
what we're exporting.  The comment should bring the issue up in a
future grep.

thx,

Jason.