Re: [PATCH 05/19] irqchip: atmel-aic: use simple constant to get number of interrupts per chip

From: Boris Brezillon
Date: Mon Jan 04 2016 - 03:34:02 EST


On Mon, 4 Jan 2016 13:28:29 +0900
Milo Kim <milo.kim@xxxxxx> wrote:

> Number of interrupts per each chip is determined when IRQ controller
> allocates IRQ chip by calling irq_alloc_domain_generic_chips().
> This number is fixed by atmel-aic-common part. The value is 32.
> So each AIC driver can use this value directly in IRQ chip operation.

Sorry, but I don't like the idea of hardcoding the number of irqs per
chip just to optimize some functions that are rarely called.
What if atmel creates a chip using the AIC but exposing less than 32
irqs? You'll have to change all those places.

Do you have a strong reason to do this kind of optimization?

>
> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> Cc: Jason Cooper <jason@xxxxxxxxxxxxxx>
> Cc: Marc Zyngier <marc.zyngier@xxxxxxx>
> Cc: Alexandre Belloni <alexandre.belloni@xxxxxxxxxxxxxxxxxx>
> Cc: Boris BREZILLON <boris.brezillon@xxxxxxxxxxxxxxxxxx>
> Cc: Ludovic Desroches <ludovic.desroches@xxxxxxxxx>
> Cc: Nicolas Ferre <nicolas.ferre@xxxxxxxxx>
> Cc: linux-kernel@xxxxxxxxxxxxxxx
> Signed-off-by: Milo Kim <milo.kim@xxxxxx>
> ---
> drivers/irqchip/irq-atmel-aic.c | 2 +-
> drivers/irqchip/irq-atmel-aic5.c | 9 +++------
> 2 files changed, 4 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/irqchip/irq-atmel-aic.c b/drivers/irqchip/irq-atmel-aic.c
> index c499949..f2c0fd9 100644
> --- a/drivers/irqchip/irq-atmel-aic.c
> +++ b/drivers/irqchip/irq-atmel-aic.c
> @@ -188,7 +188,7 @@ static int aic_irq_domain_xlate(struct irq_domain *d,
> if (ret)
> return ret;
>
> - idx = intspec[0] / dgc->irqs_per_chip;
> + idx = intspec[0] / AIC_IRQS_PER_CHIP;
> if (idx >= dgc->num_chips)
> return -EINVAL;
>
> diff --git a/drivers/irqchip/irq-atmel-aic5.c b/drivers/irqchip/irq-atmel-aic5.c
> index f5848c8..50d540b 100644
> --- a/drivers/irqchip/irq-atmel-aic5.c
> +++ b/drivers/irqchip/irq-atmel-aic5.c
> @@ -153,14 +153,13 @@ static int aic5_set_type(struct irq_data *d, unsigned type)
> static void aic5_suspend(struct irq_data *d)
> {
> struct irq_domain *domain = d->domain;
> - struct irq_domain_chip_generic *dgc = domain->gc;
> struct irq_chip_generic *bgc = irq_get_domain_generic_chip(domain, 0);
> struct irq_chip_generic *gc = irq_data_get_irq_chip_data(d);
> int i;
> u32 mask;
>
> irq_gc_lock(bgc);
> - for (i = 0; i < dgc->irqs_per_chip; i++) {
> + for (i = 0; i < AIC_IRQS_PER_CHIP; i++) {
> mask = 1 << i;
> if ((mask & gc->mask_cache) == (mask & gc->wake_active))
> continue;
> @@ -177,14 +176,13 @@ static void aic5_suspend(struct irq_data *d)
> static void aic5_resume(struct irq_data *d)
> {
> struct irq_domain *domain = d->domain;
> - struct irq_domain_chip_generic *dgc = domain->gc;
> struct irq_chip_generic *bgc = irq_get_domain_generic_chip(domain, 0);
> struct irq_chip_generic *gc = irq_data_get_irq_chip_data(d);
> int i;
> u32 mask;
>
> irq_gc_lock(bgc);
> - for (i = 0; i < dgc->irqs_per_chip; i++) {
> + for (i = 0; i < AIC_IRQS_PER_CHIP; i++) {
> mask = 1 << i;
> if ((mask & gc->mask_cache) == (mask & gc->wake_active))
> continue;
> @@ -201,13 +199,12 @@ static void aic5_resume(struct irq_data *d)
> static void aic5_pm_shutdown(struct irq_data *d)
> {
> struct irq_domain *domain = d->domain;
> - struct irq_domain_chip_generic *dgc = domain->gc;
> struct irq_chip_generic *bgc = irq_get_domain_generic_chip(domain, 0);
> struct irq_chip_generic *gc = irq_data_get_irq_chip_data(d);
> int i;
>
> irq_gc_lock(bgc);
> - for (i = 0; i < dgc->irqs_per_chip; i++) {
> + for (i = 0; i < AIC_IRQS_PER_CHIP; i++) {
> irq_reg_writel(bgc, i + gc->irq_base, AT91_AIC5_SSR);
> irq_reg_writel(bgc, 1, AT91_AIC5_IDCR);
> irq_reg_writel(bgc, 1, AT91_AIC5_ICCR);



--
Boris Brezillon, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/