Re: [RFC PATCH] of: platform: Skip mapping of interrupts in of_device_alloc()

[Marc Zyngier <maz@kernel.org>]

On 2021-12-09 00:10, Lad Prabhakar wrote:
> of_device_alloc() in early boot stage creates a interrupt mapping if
> there exists a "interrupts" property in the node.
> 
> For hierarchical interrupt domains using "interrupts" property in the 
> node
> bypassed the hierarchical setup and messed up the irq setup.
> 
> This patch adds a check in of_device_alloc() to skip interrupt mapping 
> if
> "not-interrupt-producer" property is present in the node. This allows
> nodes to describe the interrupts using "interrupts" property.
> 
> Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
> ---
> Hi All,
> 
> Spawning from discussion [1], here is simple patch (not the ideal 
> probably
> welcome for suggestions) from stopping the OF code from creating a map 
> for
> the interrupts when using "interrupts" property.
> 
> [1] https://lore.kernel.org/lkml/87pmqrck2m.wl-maz@kernel.org/
>     T/#mbd1e47c1981082aded4b32a52e2c04291e515508
> 
> Cheers,
> Prabhakar
> ---
>  drivers/of/platform.c | 13 ++++++++++---
>  1 file changed, 10 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/of/platform.c b/drivers/of/platform.c
> index b3faf89744aa..629776ca1721 100644
> --- a/drivers/of/platform.c
> +++ b/drivers/of/platform.c
> @@ -114,7 +114,7 @@ struct platform_device *of_device_alloc(struct
> device_node *np,
>  				  struct device *parent)
>  {
>  	struct platform_device *dev;
> -	int rc, i, num_reg = 0, num_irq;
> +	int rc, i, num_reg = 0, num_irq = 0;
>  	struct resource *res, temp_res;
> 
>  	dev = platform_device_alloc("", PLATFORM_DEVID_NONE);
> @@ -124,7 +124,14 @@ struct platform_device *of_device_alloc(struct
> device_node *np,
>  	/* count the io and irq resources */
>  	while (of_address_to_resource(np, num_reg, &temp_res) == 0)
>  		num_reg++;
> -	num_irq = of_irq_count(np);
> +
> +	/*
> +	 * we don't want to map the interrupts of hierarchical interrupt 
> domain
> +	 * into the parent domain yet. This will be the job of the 
> hierarchical
> +	 * interrupt driver code to map the interrupts as and when needed.
> +	 */
> +	if (!of_property_read_bool(np, "not-interrupt-producer"))

I don't think this is right. If anything, the expected behaviour should 
be
indicated by the driver and not the device node.

> +		num_irq = of_irq_count(np);
> 
>  	/* Populate the resource table */
>  	if (num_irq || num_reg) {
> @@ -140,7 +147,7 @@ struct platform_device *of_device_alloc(struct
> device_node *np,
>  			rc = of_address_to_resource(np, i, res);
>  			WARN_ON(rc);
>  		}
> -		if (of_irq_to_resource_table(np, res, num_irq) != num_irq)
> +		if (num_irq && of_irq_to_resource_table(np, res, num_irq) != 
> num_irq)
>  			pr_debug("not all legacy IRQ resources mapped for %pOFn\n",
>  				 np);
>  	}

The root of the issue is that all the resource allocation is done 
upfront,
way before we even have a driver that could potentially deal with this
device. This is a potential waste of resource, and it triggers the
issue you noticed.

If you delay the resource allocation until there is an actual match with 
a
driver, you could have a per-driver flag telling you whether the IRQ
allocation should be performed before the probe() function is called.

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