Re: [PATCH 2/3] irqchip: xilinx: Enable generic irq multi handler

[Marc Zyngier]

Michal,

On 2020-02-12 08:39, Michal Simek wrote:
> Register default arch handler via driver instead of directly pointing 
> to
> xilinx intc controller. This patch makes architecture code more 
> generic.
>
> Signed-off-by: Michal Simek <michal.simek@xxxxxxxxxx>
> Reviewed-by: Stefan Asserhall <stefan.asserhall@xxxxxxxxxx>
> ---
>
>  arch/microblaze/Kconfig           |  1 +
>  arch/microblaze/include/asm/irq.h |  3 ---
>  arch/microblaze/kernel/irq.c      | 16 +---------------
>  drivers/irqchip/irq-xilinx-intc.c | 22 +++++++++++++++++++++-
>  4 files changed, 23 insertions(+), 19 deletions(-)
>
> diff --git a/arch/microblaze/Kconfig b/arch/microblaze/Kconfig
> index 6a331bd57ea8..3a314aa2efa1 100644
> --- a/arch/microblaze/Kconfig
> +++ b/arch/microblaze/Kconfig
> @@ -47,6 +47,7 @@ config MICROBLAZE
>  	select CPU_NO_EFFICIENT_FFS
>  	select MMU_GATHER_NO_RANGE if MMU
>  	select SPARSE_IRQ
> +	select GENERIC_IRQ_MULTI_HANDLER
>
>  # Endianness selection
>  choice
> diff --git a/arch/microblaze/include/asm/irq.h
> b/arch/microblaze/include/asm/irq.h
> index eac2fb4b3fb9..5166f0893e2b 100644
> --- a/arch/microblaze/include/asm/irq.h
> +++ b/arch/microblaze/include/asm/irq.h
> @@ -14,7 +14,4 @@
>  struct pt_regs;
>  extern void do_IRQ(struct pt_regs *regs);
>
> -/* should be defined in each interrupt controller driver */
> -extern unsigned int xintc_get_irq(void);
> -
>  #endif /* _ASM_MICROBLAZE_IRQ_H */
> diff --git a/arch/microblaze/kernel/irq.c 
> b/arch/microblaze/kernel/irq.c
> index 903dad822fad..1f8cb4c4f74f 100644
> --- a/arch/microblaze/kernel/irq.c
> +++ b/arch/microblaze/kernel/irq.c
> @@ -20,27 +20,13 @@
>  #include <linux/irqchip.h>
>  #include <linux/of_irq.h>
>
> -static u32 concurrent_irq;
> -
>  void __irq_entry do_IRQ(struct pt_regs *regs)
>  {
> -	unsigned int irq;
>  	struct pt_regs *old_regs = set_irq_regs(regs);
>  	trace_hardirqs_off();
>
>  	irq_enter();
> -	irq = xintc_get_irq();
> -next_irq:
> -	BUG_ON(!irq);
> -	generic_handle_irq(irq);
> -
> -	irq = xintc_get_irq();
> -	if (irq != -1U) {
> -		pr_debug("next irq: %d\n", irq);
> -		++concurrent_irq;
> -		goto next_irq;
> -	}
> -
> +	handle_arch_irq(regs);
>  	irq_exit();
>  	set_irq_regs(old_regs);

If you're going to embrace common code, maybe you should do it fully,
see below.

>  	trace_hardirqs_on();
> diff --git a/drivers/irqchip/irq-xilinx-intc.c
> b/drivers/irqchip/irq-xilinx-intc.c
> index cf1bb470d7b5..ad9e678c24ac 100644
> --- a/drivers/irqchip/irq-xilinx-intc.c
> +++ b/drivers/irqchip/irq-xilinx-intc.c
> @@ -125,7 +125,7 @@ static unsigned int xintc_get_irq_local(struct
> xintc_irq_chip *irqc)
>  	return irq;
>  }
>
> -unsigned int xintc_get_irq(void)
> +static unsigned int xintc_get_irq(void)
>  {
>  	u32 hwirq;
>  	unsigned int irq = -1;
> @@ -178,6 +178,25 @@ static void xil_intc_irq_handler(struct irq_desc 
> *desc)
>  	chained_irq_exit(chip, desc);
>  }
>
> +static u32 concurrent_irq;
> +
> +static void xil_intc_handle_irq(struct pt_regs *regs)
> +{
> +	unsigned int irq;
> +
> +	irq = xintc_get_irq();
> +next_irq:
> +	BUG_ON(!irq);

Don't BUG_ON() for something that is probably a spurious interrupt.
Handle it gracefully instead.

> +	generic_handle_irq(irq);
> +
> +	irq = xintc_get_irq();
> +	if (irq != -1U) {
> +		pr_debug("next irq: %d\n", irq);

You already have debug information in xintc_get_irq(). Do you need more?

> +		++concurrent_irq;

What is the purpose of this "concurrent_irq"? It's not concurrent at all
(it is actually broken in an SMP context), and nothing reads it.

> +		goto next_irq;
> +	}

Overall, this could be written in a much more elegant way:

        irq_hw_number_t hwirq;
        do {
                hwirq = xintc_read(IVR);
                if (likely(irq != -1UL))
                        handle_domain_irq(xintc_irqc->root_domain,
                                          hwirq, regs);
        } while (hwirq != -1UL);

and you can get rid of most of do_IRQ() (all the set_irq_regs,
irq_enter logic is taken care of) as well as xintc_get_irq().

> +}
> +
>  static int __init xilinx_intc_of_init(struct device_node *intc,
>  					     struct device_node *parent)
>  {
> @@ -248,6 +267,7 @@ static int __init xilinx_intc_of_init(struct
> device_node *intc,
>  	} else {
>  		primary_intc = irqc;
>  		irq_set_default_host(primary_intc->root_domain);
> +		set_handle_irq(xil_intc_handle_irq);
>  	}
>
>  	return 0;

Thanks,

        M.
--
Jazz is not dead. It just smells funny...