[PATCH 05/10] firewall: introduce stm32_firewall framework
Oleksii Moisieiev
Oleksii_Moisieiev at epam.com
Fri Jul 7 09:01:11 PDT 2023
Gatien CHEVALLIER <gatien.chevallier at foss.st.com> writes:
Hello Gatien,
> Hello Oleksii,
>
> On 7/7/23 15:50, Oleksii Moisieiev wrote:
>> Gatien Chevallier <gatien.chevallier at foss.st.com> writes:
>>
>>> Introduce a firewall framework that offers to firewall consumers different
>>> firewall services such as the ability to check their access rights against
>>> their firewall controller(s).
>>>
>>> The firewall framework offers a generic API that is defined in firewall
>>> controllers drivers to best fit the specificity of each firewall.
>>>
>>> There are various types of firewalls:
>>> -Peripheral firewalls that filter accesses to peripherals
>>> -Memory firewalls that filter accesses to memories or memory regions
>>> -Resource firewalls that filter accesses to internal resources such as
>>> reset and clock controllers
>>>
>>> A firewall controller must be probed at arch_initcall level and register
>>> to the framework so that consumers can use their services.
>>>
>>> Signed-off-by: Gatien Chevallier <gatien.chevallier at foss.st.com>
>>> ---
>>> MAINTAINERS | 5 +
>>> arch/arm64/Kconfig.platforms | 1 +
>>> drivers/bus/Kconfig | 10 +
>>> drivers/bus/Makefile | 1 +
>>> drivers/bus/stm32_firewall.c | 252 ++++++++++++++++++++++
>>> drivers/bus/stm32_firewall.h | 83 +++++++
>>> include/linux/bus/stm32_firewall_device.h | 134 ++++++++++++
>>> 7 files changed, 486 insertions(+)
>>> create mode 100644 drivers/bus/stm32_firewall.c
>>> create mode 100644 drivers/bus/stm32_firewall.h
>>> create mode 100644 include/linux/bus/stm32_firewall_device.h
>>>
>>> diff --git a/MAINTAINERS b/MAINTAINERS
>>> index 41385f01fa98..fabf95ba9b86 100644
>>> --- a/MAINTAINERS
>>> +++ b/MAINTAINERS
>>> @@ -20123,6 +20123,11 @@ T: git git://linuxtv.org/media_tree.git
>>> F: Documentation/devicetree/bindings/media/i2c/st,st-mipid02.yaml
>>> F: drivers/media/i2c/st-mipid02.c
>>> +ST STM32 FIREWALL
>>> +M: Gatien Chevallier <gatien.chevallier at foss.st.com>
>>> +S: Maintained
>>> +F: drivers/bus/stm32_firewall.c
>>> +
>>> ST STM32 I2C/SMBUS DRIVER
>>> M: Pierre-Yves MORDRET <pierre-yves.mordret at foss.st.com>
>>> M: Alain Volmat <alain.volmat at foss.st.com>
>>> diff --git a/arch/arm64/Kconfig.platforms b/arch/arm64/Kconfig.platforms
>>> index 6069120199bb..5a46e90f1e4e 100644
>>> --- a/arch/arm64/Kconfig.platforms
>>> +++ b/arch/arm64/Kconfig.platforms
>>> @@ -293,6 +293,7 @@ config ARCH_STM32
>>> select ARM_SMC_MBOX
>>> select ARM_SCMI_PROTOCOL
>>> select COMMON_CLK_SCMI
>>> + select STM32_FIREWALL
>>> help
>>> This enables support for ARMv8 based STMicroelectronics
>>> STM32 family, including:
>>> diff --git a/drivers/bus/Kconfig b/drivers/bus/Kconfig
>>> index fcfa280df98a..4d54a7ea52b2 100644
>>> --- a/drivers/bus/Kconfig
>>> +++ b/drivers/bus/Kconfig
>>> @@ -163,6 +163,16 @@ config QCOM_SSC_BLOCK_BUS
>>> i2c/spi/uart controllers, a hexagon core, and a clock controller
>>> which provides clocks for the above.
>>> +config STM32_FIREWALL
>>> + bool "STM32 Firewall framework"
>>> + depends on ARCH_STM32
>>> + default MACH_STM32MP157 || MACH_STM32MP13 || MACH_STM32MP25
>>> + help
>>> + Say y to enable firewall framework and its services. Firewall
>>> + controllers will be able to register to the framework. Firewall
>>> + controllers must be initialized and register to the firewall framework
>>> + at arch_initcall level.
>>> +
>>> config SUN50I_DE2_BUS
>>> bool "Allwinner A64 DE2 Bus Driver"
>>> default ARM64
>>> diff --git a/drivers/bus/Makefile b/drivers/bus/Makefile
>>> index d90eed189a65..fc0511450ec2 100644
>>> --- a/drivers/bus/Makefile
>>> +++ b/drivers/bus/Makefile
>>> @@ -26,6 +26,7 @@ obj-$(CONFIG_OMAP_INTERCONNECT) += omap_l3_smx.o omap_l3_noc.o
>>> obj-$(CONFIG_OMAP_OCP2SCP) += omap-ocp2scp.o
>>> obj-$(CONFIG_QCOM_EBI2) += qcom-ebi2.o
>>> obj-$(CONFIG_QCOM_SSC_BLOCK_BUS) += qcom-ssc-block-bus.o
>>> +obj-$(CONFIG_STM32_FIREWALL) += stm32_firewall.o
>>> obj-$(CONFIG_SUN50I_DE2_BUS) += sun50i-de2.o
>>> obj-$(CONFIG_SUNXI_RSB) += sunxi-rsb.o
>>> obj-$(CONFIG_OF) += simple-pm-bus.o
>>> diff --git a/drivers/bus/stm32_firewall.c b/drivers/bus/stm32_firewall.c
>>> new file mode 100644
>>> index 000000000000..510db5bc6eaf
>>> --- /dev/null
>>> +++ b/drivers/bus/stm32_firewall.c
>>> @@ -0,0 +1,252 @@
>>> +// SPDX-License-Identifier: GPL-2.0-or-later
>>> +/*
>>> + * Copyright (C) 2023, STMicroelectronics - All Rights Reserved
>>> + */
>>> +
>>> +#include <linux/bitfield.h>
>>> +#include <linux/bits.h>
>>> +#include <linux/bus/stm32_firewall_device.h>
>>> +#include <linux/device.h>
>>> +#include <linux/err.h>
>>> +#include <linux/init.h>
>>> +#include <linux/io.h>
>>> +#include <linux/kernel.h>
>>> +#include <linux/module.h>
>>> +#include <linux/of.h>
>>> +#include <linux/of_platform.h>
>>> +#include <linux/platform_device.h>
>>> +#include <linux/types.h>
>>> +
>>> +#include "stm32_firewall.h"
>>> +
>>> +/* Corresponds to STM32_FIREWALL_MAX_EXTRA_ARGS + firewall controller reference + firewall ID */
>>> +#define STM32_FIREWALL_MAX_ARGS (STM32_FIREWALL_MAX_EXTRA_ARGS + 2)
>>> +
>>> +static LIST_HEAD(firewall_controller_list);
>>> +static DEFINE_MUTEX(firewall_controller_list_lock);
>>> +
>>> +static int stm32_firewall_get_id(struct device_node *np, u32 *id)
>>> +{
>>> + u32 feature_domain_cell[2];
>>> +
>>> + /* Get property from device node */
>>> + if (of_property_read_u32_array(np, "feature-domains",
>>> + feature_domain_cell,
>>> + ARRAY_SIZE(feature_domain_cell))) {
>>> + pr_err("Unable to find get firewall ID property\n");
>>> + return -ENODEV;
>>> + }
>>> +
>>> + *id = feature_domain_cell[1];
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +/* Firewall device API */
>>> +
>>> +int stm32_firewall_get_firewall(struct device_node *np,
>>> + struct stm32_firewall *firewall)
>>> +{
>>> + struct stm32_firewall_controller *ctrl;
>>> + struct of_phandle_args args;
>>> + u32 controller_phandle;
>>> + bool match = false;
>>> + size_t i;
>>> + int err;
>>> +
>>> + if (!firewall)
>>> + return -EINVAL;
>>> +
>>> + /* The controller phandle is always the first argument of the feature-domains property. */
>>> + err = of_property_read_u32(np, "feature-domains", &controller_phandle);
>>> + if (err) {
>>> + pr_err("Unable to get feature-domains property for node %s\n", np->full_name);
>>> + return err;
>>> + }
>>> +
>>> + /* Parse property with phandle parsed out */
>>> + err = of_parse_phandle_with_args(np, "feature-domains", "#feature-domain-cells", 0, &args);
>>> + if (err) {
>>> + pr_err("Unable to read feature-domains arguments for node %s\n", np->full_name);
>>> + return err;
>>> + }
>>> +
>>> + /* The phandle is parsed out */
>>> + if (args.args_count > STM32_FIREWALL_MAX_ARGS - 1)
>>> + return -EINVAL;
>>> +
>>> + of_node_put(np);
>>> +
>>> + /* Check if the parsed phandle corresponds to a registered firewall controller */
>>> + mutex_lock(&firewall_controller_list_lock);
>>> + list_for_each_entry(ctrl, &firewall_controller_list, entry) {
>>> + if (ctrl->dev->of_node->phandle == controller_phandle) {
>>> + match = true;
>>> + firewall->firewall_ctrl = ctrl;
>>> + break;
>>> + }
>>> + }
>>> + mutex_unlock(&firewall_controller_list_lock);
>>> + if (!match) {
>>> + firewall->firewall_ctrl = NULL;
>>> + pr_err("No firewall controller registered for %s\n", np->full_name);
>>> + return -ENODEV;
>>> + }
>>> +
>>> + /*
>>> + * The firewall ID is always the second argument of the feature-domains property.
>>> + * The first argument is already parsed out, so args.args[0] is the second argument.
>>> + */
>>> + firewall->firewall_id = args.args[0];
>>> +
>>> + /* Extra args start at the third argument */
>>> + for (i = 0; i < args.args_count; i++)
>>> + firewall->extra_args[i] = args.args[i + 1];
>>> +
>>> + /* Remove the firewall ID arg that is not an extra argument */
>>> + if (args.args_count >= 1)
>>> + firewall->extra_args_size = args.args_count - 1;
>> As I can see you support some extra args, that can be provided in
>> the
>> feature-domain property. But in the binding description I see maxItems:
>> 2. I beliewe this should be highlighted in bindings description.
>>
>
> Good point, maybe it would be better to define a high maxItem value in
> the binding description that is aligned with STM32_FIREWALL_MAX_ARGS.
>
>>> +
>>> + return 0;
>>> +}
>>> +EXPORT_SYMBOL_GPL(stm32_firewall_get_firewall);
>>> +
>>> +int stm32_firewall_grant_access(struct stm32_firewall *firewall)
>>> +{
>>> + struct stm32_firewall_controller *firewall_controller;
>>> +
>>> + if (!firewall || firewall->firewall_id == U32_MAX)
>>> + return -EINVAL;
>>> +
>>> + firewall_controller = firewall->firewall_ctrl;
>>> +
>>> + if (!firewall_controller)
>>> + return -ENODEV;
>>> +
>>> + return firewall_controller->grant_access(firewall_controller, firewall->firewall_id);
>>> +}
>>> +EXPORT_SYMBOL_GPL(stm32_firewall_grant_access);
>>> +
>>> +int stm32_firewall_grant_access_by_id(struct stm32_firewall *firewall, u32 subsystem_id)
>>> +{
>>> + struct stm32_firewall_controller *firewall_controller;
>>> +
>>> + if (!firewall || subsystem_id == U32_MAX || firewall->firewall_id == U32_MAX)
>>> + return -EINVAL;
>>> +
>>> + firewall_controller = firewall->firewall_ctrl;
>>> +
>>> + if (!firewall_controller)
>>> + return -ENODEV;
>>> +
>>> + return firewall_controller->grant_access(firewall_controller, subsystem_id);
>>> +}
>>> +EXPORT_SYMBOL_GPL(stm32_firewall_grant_access_by_id);
>>> +
>>> +void stm32_firewall_release_access(struct stm32_firewall *firewall)
>>> +{
>>> + struct stm32_firewall_controller *firewall_controller;
>>> +
>>> + if (!firewall || firewall->firewall_id == U32_MAX) {
>>> + pr_err("Incorrect arguments when releasing a firewall access");
>>> + return;
>>> + }
>>> +
>>> + firewall_controller = firewall->firewall_ctrl;
>>> +
>>> + if (!firewall_controller) {
>>> + pr_debug("No firewall controller to release");
>>> + return;
>>> + }
>>> +
>>> + firewall_controller->release_access(firewall_controller, firewall->firewall_id);
>>> +}
>>> +EXPORT_SYMBOL_GPL(stm32_firewall_release_access);
>>> +
>>> +void stm32_firewall_release_access_by_id(struct stm32_firewall *firewall, u32 subsystem_id)
>>> +{
>>> + struct stm32_firewall_controller *firewall_controller;
>>> +
>>> + if (!firewall || subsystem_id == U32_MAX || firewall->firewall_id == U32_MAX) {
>>> + pr_err("Incorrect arguments when releasing a firewall access");
>>> + return;
>>> + }
>>> +
>>> + firewall_controller = firewall->firewall_ctrl;
>>> +
>>> + if (!firewall_controller) {
>>> + pr_debug("No firewall controller to release");
>>> + return;
>>> + }
>>> +
>>> + firewall_controller->release_access(firewall_controller, subsystem_id);
>>> +}
>>> +EXPORT_SYMBOL_GPL(stm32_firewall_release_access_by_id);
>>> +
>>> +/* Firewall controller API */
>>> +
>>> +int stm32_firewall_controller_register(struct stm32_firewall_controller *firewall_controller)
>>> +{
>>> + pr_info("Registering firewall controller %s", dev_name(firewall_controller->dev));
>>> +
>>> + if (!firewall_controller)
>>> + return -ENODEV;
>>> +
>>> + mutex_lock(&firewall_controller_list_lock);
>>> + list_add_tail(&firewall_controller->entry, &firewall_controller_list);
>>> + mutex_unlock(&firewall_controller_list_lock);
>>> +
>>> + return 0;
>>> +
>>> +}
>>> +
>>> +void stm32_firewall_controller_unregister(struct stm32_firewall_controller *firewall_controller)
>>> +{
>>> + struct stm32_firewall_controller *ctrl, *tmp;
>>> + bool controller_removed = false;
>>> +
>>> + if (!firewall_controller) {
>>> + pr_debug("Null reference while unregistering firewall controller");
>>> + return;
>>> + }
>>> +
>>> + mutex_lock(&firewall_controller_list_lock);
>>> + list_for_each_entry_safe(ctrl, tmp, &firewall_controller_list, entry) {
>>> + if (ctrl == firewall_controller) {
>>> + controller_removed = true;
>>> + list_del_init(&ctrl->entry);
>>> + break;
>>> + }
>>> + }
>> IIUC list_for_each_entry_safe protects you from removing node during
>> loop. But all list operations are done under
>> firewall_controller_list_lock mutex. I beliewe there is no need for
>> _safe call under the mutex because removing node during loop is
>> impossible. I think it worth investigation if it's safe to use
>> list_for_each_entry_safe without lock.
>
> AFAICT, list_for_each_entry_safe() does not protect against concurrent
> accesses, so I'll keep the mutex lock.
>
> Deleting while iterating will cause trouble if not using the safe
> version. I think it is fine as it is.
>
Ok, in that case I whould rather drop _safe, because for your case it is
impossible to remove entry while iterating the list.
This will save a couple of CPU instructions.
>>> + mutex_unlock(&firewall_controller_list_lock);
>>> +
>>> + if (!controller_removed)
>>> + pr_debug("There was no firewall controller named %s to unregister",
>>> + dev_name(firewall_controller->dev));
>>> +}
>>> +
>>> +void stm32_firewall_populate_bus(struct stm32_firewall_controller *firewall_controller)
>>> +{
>>> + struct device_node *child;
>>> + struct device *parent;
>>> + u32 firewall_id;
>>> + int err;
>>> +
>>> + parent = firewall_controller->dev;
>>> +
>>> + dev_dbg(parent, "Populating %s system bus\n", dev_name(firewall_controller->dev));
>>> +
>>> + for_each_available_child_of_node(dev_of_node(parent), child) {
>>> + err = stm32_firewall_get_id(child, &firewall_id);
>>> + if (err < 0 ||
>>> + firewall_controller->grant_access(firewall_controller, firewall_id)) {
>>> + /*
>>> + * Peripheral access not allowed or not defined.
>>> + * Mark the node as populated so platform bus won't probe it
>>> + */
>>> + of_node_set_flag(child, OF_POPULATED);
>>> + dev_err(parent, "%s: Device driver will not be probed\n",
>>> + child->full_name);
>>> + }
>>> + }
>>> +}
>>> diff --git a/drivers/bus/stm32_firewall.h b/drivers/bus/stm32_firewall.h
>>> new file mode 100644
>>> index 000000000000..8d92e8c1ab77
>>> --- /dev/null
>>> +++ b/drivers/bus/stm32_firewall.h
>>> @@ -0,0 +1,83 @@
>>> +/* SPDX-License-Identifier: GPL-2.0-or-later */
>>> +/*
>>> + * Copyright (C) 2023, STMicroelectronics - All Rights Reserved
>>> + */
>>> +
>>> +#ifndef _STM32_FIREWALL_H
>>> +#define _STM32_FIREWALL_H
>>> +
>>> +#include <linux/kernel.h>
>>> +#include <linux/list.h>
>>> +#include <linux/of.h>
>>> +#include <linux/platform_device.h>
>>> +#include <linux/types.h>
>>> +
>>> +/**
>>> + * STM32_PERIPHERAL_FIREWALL: This type of firewall protects peripherals
>> Intendation
>>
>>> + * STM32_MEMORY_FIREWALL: This type of firewall protects memories/subsets of memory
>>> + * zones
>>> + * STM32_RESOURCE_FIREWALL: This type of firewall protects internal resources
>>> + * STM32_NOTYPE_FIREWALL: Undefined firewall type
>>> + */
>>> +
>>> +#define STM32_PERIPHERAL_FIREWALL BIT(1)
>> Intendation
>>
>
> I'm sorry but I don't understand. I indent with tabs, maybe this is why
> it appears strange here. The indentation is fine, same for others below.
>
Nevermind, Just looks strange in patch.
>>> +#define STM32_MEMORY_FIREWALL BIT(2)
>>> +#define STM32_RESOURCE_FIREWALL BIT(3)
>>> +#define STM32_NOTYPE_FIREWALL BIT(4)
>>> +
>>> +/**
>>> + * struct stm32_firewall_controller - Information on firewall controller supplying services
>>> + *
>>> + * @name Name of the firewall controller
>>> + * @dev Device reference of the firewall controller
>> Intendation
>>
>>> + * @mmio Base address of the firewall controller
>>> + * @entry List entry of the firewall controller list
>>> + * @type Type of firewall
>>> + * @max_entries Number of entries covered by the firewall
>> Intendation
>>
>>> + * @grant_access Callback used to grant access for a device access against a
>>> + * firewall controller
>>> + * @release_access Callback used to release resources taken by a device when access was
>>> + * granted
>>> + * @grant_memory_range_access Callback used to grant access for a device to a given memory region
>>> + */
>>> +struct stm32_firewall_controller {
>>> + const char *name;
>>> + struct device *dev;
>>> + void __iomem *mmio;
>>> + struct list_head entry;
>>> + unsigned int type;
>>> + unsigned int max_entries;
>>> +
>>> + int (*grant_access)(struct stm32_firewall_controller *ctrl, u32 id);
>>> + void (*release_access)(struct stm32_firewall_controller *ctrl, u32 id);
>>> + int (*grant_memory_range_access)(struct stm32_firewall_controller *ctrl, phys_addr_t paddr,
>>> + size_t size);
>>> +};
>>> +
>>> +/**
>>> + * int stm32_firewall_controller_register - Register a firewall controller to the STM32 firewall
>>> + * framework
>>> + * @firewall_controller Firewall controller to register
>>> + *
>>> + * Returns 0 in case of success or -ENODEV if no controller was given.
>>> + */
>>> +int stm32_firewall_controller_register(struct stm32_firewall_controller *firewall_controller);
>>> +
>>> +/**
>>> + * int stm32_firewall_controller_unregister - Unregister a firewall controller from the STM32
>>> + * firewall framework
>>> + * @firewall_controller Firewall controller to unregister
>>> + */
>>> +void stm32_firewall_controller_unregister(struct stm32_firewall_controller *firewall_controller);
>>> +
>>> +/**
>>> + * stm32_firewall_populate_bus - Populate device tree nodes that have a correct firewall
>>> + * configuration. This is used at boot-time only, as a sanity check
>>> + * between device tree and firewalls hardware configurations to
>>> + * prevent a kernel crash when a device driver is not granted access
>>> + *
>>> + * @firewall_controller Firewall controller which nodes will be populated or not
>>> + */
>>> +void stm32_firewall_populate_bus(struct stm32_firewall_controller *firewall_controller);
>>> +
>>> +#endif /* _STM32_FIREWALL_H */
>>> diff --git a/include/linux/bus/stm32_firewall_device.h b/include/linux/bus/stm32_firewall_device.h
>>> new file mode 100644
>>> index 000000000000..ccaecea7fc6c
>>> --- /dev/null
>>> +++ b/include/linux/bus/stm32_firewall_device.h
>>> @@ -0,0 +1,134 @@
>>> +/* SPDX-License-Identifier: GPL-2.0-or-later */
>>> +/*
>>> + * Copyright (C) 2023, STMicroelectronics - All Rights Reserved
>>> + */
>>> +
>>> +#ifndef STM32_FIREWALL_DEVICE_H
>>> +#define STM32_FIREWALL_DEVICE_H
>>> +
>>> +#include <linux/of.h>
>>> +#include <linux/platform_device.h>
>>> +#include <linux/types.h>
>>> +
>>> +#define STM32_FIREWALL_MAX_EXTRA_ARGS 5
>> It's not clear to me why it's 5. Comment above sais this is
>> implementation defined. Maybe this parameter can be configurable?
>>
>
> Maybe it should be part of compat_data of firewall controllers :)
>
>>> +
>>> +/* Opaque reference to stm32_firewall_controller */
>>> +struct stm32_firewall_controller;
>>> +
>>> +/**
>>> + * stm32_firewall - Information on a device's firewall. Each device can have more than one firewall.
>>> + *
>>> + * @firewall_ctrl Pointer referencing a firewall controller of the device. It is
>>> + * opaque so a device cannot manipulate the controller's ops or access
>>> + * the controller's data
>>> + * @extra_args Extra arguments that are implementation dependent
>>> + * @extra_args_size Number of extra arguments
>>> + * @firewall_id Firewall ID associated the device for this firewall controller
>>> + */
>>> +struct stm32_firewall {
>>> + struct stm32_firewall_controller *firewall_ctrl;
>>> + u32 extra_args[STM32_FIREWALL_MAX_EXTRA_ARGS];
>>> + size_t extra_args_size;
>>> + u32 firewall_id;
>>> +};
>>> +
>>> +#if IS_ENABLED(CONFIG_STM32_FIREWALL)
>>> +/**
>>> + * stm32_firewall_get_firewall - Get the firewall(s) associated to given device.
>>> + * The firewall controller reference is always the first argument
>>> + * of the feature-domains property.
>>> + * The firewall ID is always the second argument of the
>>> + * feature-domains property.
>>> + *
>>> + * @np Device node to parse
>> Intendation
>>
>>> + * @firewall Resulting firewall reference(s)
>>> + *
>>> + * Returns 0 on success, -ENODEV if there's no match with a firewall controller or appropriate errno
>>> + * code if error occurred.
>>> + */
>>> +int stm32_firewall_get_firewall(struct device_node *np, struct stm32_firewall *firewall);
>>> +
>>> +/**
>>> + * stm32_firewall_grant_access - Request firewall access rights and grant access.
>>> + *
>>> + * @firewall Firewall reference containing the ID to check against its firewall
>>> + * controller
>>> + *
>>> + * Returns 0 if access is granted, -EACCES if access is denied, -ENODEV if firewall is null or
>>> + * appropriate errno code if error occurred
>>> + */
>>> +int stm32_firewall_grant_access(struct stm32_firewall *firewall);
>>> +
>>> +/**
>>> + * stm32_firewall_release_access - Release access granted from a call to
>>> + * stm32_firewall_grant_access().
>>> + *
>>> + * @firewall Firewall reference containing the ID to check against its firewall
>>> + * controller
>>> + */
>>> +void stm32_firewall_release_access(struct stm32_firewall *firewall);
>>> +
>>> +/**
>>> + * stm32_firewall_grant_access_by_id - Request firewall access rights of a given device
>>> + * based on a specific firewall ID
>>> + *
>>> + * Warnings:
>>> + * There is no way to ensure that the given ID will correspond to the firewall referenced in the
>>> + * device node if the ID did not come from stm32_firewall_get_firewall(). In that case, this
>>> + * function must be used with caution.
>>> + * This function should be used for subsystem resources that do not have the same firewall ID
>>> + * as their parent.
>>> + * U32_MAX is an invalid ID.
>>> + *
>>> + * @firewall Firewall reference containing the firewall controller
>>> + * @subsystem_id Firewall ID of the subsystem resource
>>> + *
>>> + * Returns 0 if access is granted, -EACCES if access is denied, -ENODEV if firewall is null or
>>> + * appropriate errno code if error occurred
>>> + */
>>> +int stm32_firewall_grant_access_by_id(struct stm32_firewall *firewall, u32 subsystem_id);
>> Can we store all registered IDs that were found by populate_bus
>> functio, or is it expected when this function was called before
>> populate_bus call?
>>
>
> This function is intended to be used by devices that needs to check
> specific ID. Maybe IDs of peripherals that do not have a driver in
> Linux / driver isn't probed / ...
>
> About storing all IDs found when populating the bus, do you have some
> use in mind? :) Else I'd prefer to save some memory.
>
> Best regards,
> Gatien
>
Then I think it worth mentioning in Warnings comment. What bothered me
is that you've said that there is no way to check ID while I was seeing
possible implementation.
>>> +
>>> +/**
>>> + * stm32_firewall_release_access_by_id - Release access granted from a call to
>>> + * stm32_firewall_grant_access_by_id().
>>> + *
>>> + * Warnings:
>>> + * There is no way to ensure that the given ID will correspond to the firewall referenced in the
>>> + * device node if the ID did not come from stm32_firewall_get_firewall(). In that case, this
>>> + * function must be used with caution.
>>> + * This function should be used for subsystem resources that do not have the same firewall ID
>>> + * as their parent.
>>> + * U32_MAX is an invalid ID.
>>> + *
>>> + * @firewall Firewall reference containing the firewall controller
>>> + * @subsystem_id Firewall ID of the subsystem resource
>>> + */
>>> +void stm32_firewall_release_access_by_id(struct stm32_firewall *firewall, u32 subsystem_id);
>>> +
>>> +#else /* CONFIG_STM32_FIREWALL */
>>> +
>>> +int stm32_firewall_get_firewall(struct device_node *np, struct stm32_firewall *firewall)
>>> +{
>>> + return -ENODEV;
>>> +}
>>> +
>>> +int stm32_firewall_grant_access(struct stm32_firewall *firewall)
>>> +{
>>> + return -ENODEV;
>>> +}
>>> +
>>> +void stm32_firewall_release_access(struct stm32_firewall *firewall)
>>> +{
>>> +}
>>> +
>>> +int stm32_firewall_grant_access_by_id(struct stm32_firewall *firewall, u32 subsystem_id)
>>> +{
>>> + return -ENODEV;
>>> +}
>>> +
>>> +void stm32_firewall_release_access_by_id(struct stm32_firewall *firewall, u32 subsystem_id)
>>> +{
>>> +}
>>> +
>>> +#endif /* CONFIG_STM32_FIREWALL */
>>> +#endif /* STM32_FIREWALL_DEVICE_H */
>>
--
Thanks,
Oleksii
More information about the linux-arm-kernel
mailing list