Message-ID: <1a98dc26-fc3f-988a-28b1-0452a465a364@arm.com>
Date:   Thu, 13 Dec 2018 14:22:35 +0000
From:   Marc Zyngier <marc.zyngier@....com>
To:     Robert Richter <rrichter@...ium.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Jason Cooper <jason@...edaemon.net>
Cc:     "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Stuart Yoder <stuyoder@...il.com>,
        Laurentiu Tudor <laurentiu.tudor@....com>,
        Matthias Brugger <matthias.bgg@...il.com>,
        Will Deacon <will.deacon@....com>,
        Lorenzo Pieralisi <Lorenzo.Pieralisi@....com>,
        "Richter, Robert" <Robert.Richter@...ium.com>,
        Julien Thierry <julien.thierry@....com>
Subject: Re: [PATCH v2 01/10] irqdomain: Add interface to request an irq
 domain

Hi Robert,

On 28/11/2018 14:43, Robert Richter wrote:
> This patch introduces a new interface to allow irq domain
> initialization regardless of order dependencies. This is done by
> requesting a domain and registering a callback function that is called
> as soon as a domain becomes available.
> 
> A typical irq domain initialization code is the following:
> 
> Parent initialization:
> 
> 	...
> 	domain = msi_create_irq_domain(fwnode, info, parent);
> 	if (domain)
> 		irq_domain_update_bus_token(domain, bus_token);
> 	...
> 
> Child initialization:
> 
> 	...
> 	parent = irq_find_matching_fwnode(fwnode, bus_token);
> 	if (!parent)
> 		...
> 	create_irq_domain(parent, ...);
> 	...
> 
> In case the parent is not yet available, the child initialization
> fails. Thus, the current implementation requires the parent domain
> being initialized before the child domain. With a complex irq domain
> hierarchy it becomes more and more difficult to grant that order as
> irq domains are enabled in separate subsystems. Care must be taken
> when initializing parent and child domains in the same initcall
> level. E.g. Arm's gic-v3-its implementation might have the following
> tree and dependencies:
> 
> 	gic-v3
> 	├── its-node-0
> 	│   ├── pci-host-0
> 	│   ├── platform-bus
> 	│   ...
> 	├── its-node-1
> 	│   ├── pci-host-1
> 	...
> 
> All domains must be initialized in top-down order of the tree.
> 
> This patch introduces an interface that allows domain initialization
> without any order requirements, e.g. to be able to initialize the irq
> domains in the same initcall level. The following functions have been
> introduced to allow the registration of a callback handler:
> 
> 	irq_domain_request_fwnode()
> 	irq_domain_request_host()
> 
> Instead of using the irq_find_matching_fwnode() function and it's
> variants the child code replaces them with the new functions and
> looks e.g. like the following:
> 
> 	...
> 	irq_domain_request_fwnode(fwnode, bus_token,
> 				create_irq_domain, name, priv);
> 	...
> 
> Here, the callback function create_irq_domain() is called as soon as
> the parent becomes available. All registered handlers are stored in a
> list. With each update of the bus token using irq_domain_update_bus_
> token(), the list is checked if that domain is requested by a handler
> and if that is the case it's callback function is called and the
> request removed from the list.
> 
> With a late_initcall all requests from the list should already have
> been handled, otherwise all remaining requests are removed with an
> error reported.
> 
> Signed-off-by: Robert Richter <rrichter@...ium.com>
> ---
>  include/linux/irqdomain.h |  15 +++++
>  kernel/irq/irqdomain.c    | 163 ++++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 178 insertions(+)
> 
> diff --git a/include/linux/irqdomain.h b/include/linux/irqdomain.h
> index 068aa46f0d55..27e83803627d 100644
> --- a/include/linux/irqdomain.h
> +++ b/include/linux/irqdomain.h
> @@ -311,6 +311,21 @@ static inline struct irq_domain *irq_find_host(struct device_node *node)
>  	return d;
>  }
>  
> +typedef int (*irq_domain_callback_t)(struct irq_domain *, void *);
> +int irq_domain_request_fwnode(struct fwnode_handle *fwnode,
> +			enum irq_domain_bus_token bus_token,
> +			irq_domain_callback_t callback,
> +			const char *name, void *priv);
> +
> +static inline int irq_domain_request_host(struct device_node *node,
> +					enum irq_domain_bus_token bus_token,
> +					irq_domain_callback_t callback,
> +					void *priv)
> +{
> +	return irq_domain_request_fwnode(of_node_to_fwnode(node), bus_token,
> +					callback, node->full_name, priv);
> +}

I don't think you need this helper. There is no backward compat to
preserve in this case, and I believe all the potential users would be
using the fwnode variant.

I'm not keen at all on the name thing though. An OF node already has
one, and an irqchip fwnode caries one as part of the irqchip_fwid structure.

> +
>  /**
>   * irq_domain_add_linear() - Allocate and register a linear revmap irq_domain.
>   * @of_node: pointer to interrupt controller's device tree node.
> diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c
> index 3366d11c3e02..ebe628dad568 100644
> --- a/kernel/irq/irqdomain.c
> +++ b/kernel/irq/irqdomain.c
> @@ -21,6 +21,7 @@
>  #include <linux/fs.h>
>  
>  static LIST_HEAD(irq_domain_list);
> +static LIST_HEAD(irq_domain_requests);
>  static DEFINE_MUTEX(irq_domain_mutex);
>  
>  static struct irq_domain *irq_default_domain;
> @@ -45,6 +46,111 @@ static inline void debugfs_remove_domain_dir(struct irq_domain *d) { }
>  const struct fwnode_operations irqchip_fwnode_ops;
>  EXPORT_SYMBOL_GPL(irqchip_fwnode_ops);
>  
> +struct irq_domain_request {
> +	struct list_head		list;
> +	struct fwnode_handle		*fwnode;
> +	enum irq_domain_bus_token	bus_token;
> +	irq_domain_callback_t		callback;
> +	char				*name;
> +	void				*priv;
> +};
> +
> +static void irq_domain_call_handler(struct irq_domain *domain,
> +		irq_domain_callback_t callback, const char *name, void *priv)
> +{
> +	int ret;
> +
> +	ret = callback(domain, priv);
> +	if (ret)
> +		pr_err("%s: Domain request handler failed: %d\n",
> +			name, ret);
> +
> +	of_node_put(irq_domain_get_of_node(domain));
> +}
> +
> +static void irq_domain_free_request(struct irq_domain_request *request)
> +{
> +	kfree(request->name);
> +	kfree(request);
> +}
> +
> +static void irq_domain_handle_requests(struct fwnode_handle *fwnode,
> +                        enum irq_domain_bus_token bus_token)
> +{
> +	struct irq_domain *domain;
> +	struct irq_domain_request *request;
> +
> +	if (!fwnode)
> +		return;
> +redo:
> +	domain = irq_find_matching_fwnode(fwnode, bus_token);
> +	if (!domain)
> +		return;
> +
> +	mutex_lock(&irq_domain_mutex);
> +
> +	/*
> +	 * For serialization of irq domain updates make sure to handle
> +	 * (and remove) the request only if the domain still matches
> +	 * the request.
> +	 */
> +	if ((domain->fwnode != fwnode) || (domain->bus_token != bus_token)) {
> +		mutex_unlock(&irq_domain_mutex);
> +		goto redo;
> +	}
> +
> +	list_for_each_entry(request, &irq_domain_requests, list) {
> +		if (request->fwnode != fwnode ||
> +		    request->bus_token != bus_token)
> +			continue;
> +
> +		list_del(&request->list);
> +		mutex_unlock(&irq_domain_mutex);
> +
> +		irq_domain_call_handler(domain, request->callback,
> +					request->name, request->priv);
> +		irq_domain_free_request(request);
> +
> +		goto redo;
> +	}

This feels a bit convoluted. Pulling the various calls out of the loop
would be better.

> +
> +	mutex_unlock(&irq_domain_mutex);
> +}
> +
> +static int __init irq_domain_drain_requests(void)
> +{
> +	struct irq_domain_request *request;
> +	struct irq_domain *domain;
> +	int ret = 0;
> +redo:
> +	mutex_lock(&irq_domain_mutex);
> +
> +	list_for_each_entry(request, &irq_domain_requests, list) {
> +		list_del(&request->list);
> +		mutex_unlock(&irq_domain_mutex);
> +
> +		domain = irq_find_matching_fwnode(request->fwnode,
> +						request->bus_token);
> +		if (domain) {
> +			irq_domain_call_handler(domain, request->callback,
> +						request->name, request->priv);
> +		} else {
> +			ret = -ENODEV;
> +			pr_err("%s-%d: Unhandled domain request\n",
> +				request->name, request->bus_token);
> +		}
> +
> +		irq_domain_free_request(request);
> +
> +		goto redo;
> +	}

Why isn't this written in terms of irq_domain_handle_requests instead?

> +
> +	mutex_unlock(&irq_domain_mutex);
> +
> +	return ret;
> +}
> +late_initcall(irq_domain_drain_requests);

Why do you have to do all matching in this late initcall? I'd have hoped
for requests to be satisfied as domains get added, not as a last ditch
effort.

> +
>  /**
>   * irq_domain_alloc_fwnode - Allocate a fwnode_handle suitable for
>   *                           identifying an irq domain
> @@ -293,6 +399,8 @@ void irq_domain_update_bus_token(struct irq_domain *domain,
>  	debugfs_add_domain_dir(domain);
>  
>  	mutex_unlock(&irq_domain_mutex);
> +
> +	irq_domain_handle_requests(domain->fwnode, bus_token);
>  }
>  
>  /**
> @@ -417,6 +525,61 @@ struct irq_domain *irq_find_matching_fwspec(struct irq_fwspec *fwspec,
>  EXPORT_SYMBOL_GPL(irq_find_matching_fwspec);
>  
>  /**
> + * irq_domain_request_fwnode() - Requests a domain for a given fwspec
> + * @fwspec: FW specifier for an interrupt
> + * @bus_token: domain-specific data
> + * @callback: function to be called once domain becomes available
> + * @name: name to be used for fwnode
> + * @priv: private data to be passed to callback
> + *
> + * The callback function is called as soon as the domain is available.
> + */
> +int irq_domain_request_fwnode(struct fwnode_handle *fwnode,
> +			enum irq_domain_bus_token bus_token,
> +			irq_domain_callback_t callback,
> +			const char *name, void *priv)
> +{
> +	struct irq_domain *parent;
> +	struct irq_domain_request *request;
> +
> +	if (!fwnode || bus_token == DOMAIN_BUS_ANY || !callback || !name)

Why are you refusing DOMAIN_BUS_ANY? If that's what the caller wants, we
should satisfy it.

> +		return -EINVAL;
> +
> +	parent = irq_find_matching_fwnode(fwnode, bus_token);
> +	if (parent) {
> +		irq_domain_call_handler(parent, callback, name, priv);
> +		return 0;
> +	}
> +
> +	request = kzalloc(sizeof(*request), GFP_KERNEL);
> +	if (!request)
> +		return -ENOMEM;
> +
> +	request->fwnode = fwnode;
> +	request->bus_token = bus_token;
> +	request->callback = callback;
> +	request->name = kstrdup(name, GFP_KERNEL);
> +	request->priv = priv;
> +	INIT_LIST_HEAD(&request->list);
> +
> +	if (!request->name) {
> +		kfree(request);
> +		return -ENOMEM;
> +	}
> +
> +	of_node_get(to_of_node(fwnode));
> +
> +	mutex_lock(&irq_domain_mutex);
> +	list_add_tail(&request->list, &irq_domain_requests);
> +	mutex_unlock(&irq_domain_mutex);
> +
> +	/* recheck in case list changed */
> +	irq_domain_handle_requests(fwnode, bus_token);

Why do you need this? The whole thing is racy anyway, and you already
have a catch-all as a late initcall...

> +
> +	return 0;
> +}
> +
> +/**
>   * irq_domain_check_msi_remap - Check whether all MSI irq domains implement
>   * IRQ remapping
>   *
> 

Overall, this is a bit hackish. Why don't you resolve the deferred
requests as part of a domain being added instead? It would make things
much more straightforward.

Thanks,

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

Powered by blists - more mailing lists