Драйверы для linux kernel

Device Drivers¶

See the kerneldoc for the struct device_driver.

Allocation¶

Device drivers are statically allocated structures. Though there may be multiple devices in a system that a driver supports, struct device_driver represents the driver as a whole (not a particular device instance).

Initialization¶

The driver must initialize at least the name and bus fields. It should also initialize the devclass field (when it arrives), so it may obtain the proper linkage internally. It should also initialize as many of the callbacks as possible, though each is optional.

Declaration¶

As stated above, struct device_driver objects are statically allocated. Below is an example declaration of the eepro100 driver. This declaration is hypothetical only; it relies on the driver being converted completely to the new model:

Most drivers will not be able to be converted completely to the new model because the bus they belong to has a bus-specific structure with bus-specific fields that cannot be generalized.

The most common example of this are device ID structures. A driver typically defines an array of device IDs that it supports. The format of these structures and the semantics for comparing device IDs are completely bus-specific. Defining them as bus-specific entities would sacrifice type-safety, so we keep bus-specific structures around.

Bus-specific drivers should include a generic struct device_driver in the definition of the bus-specific driver. Like this:

A definition that included bus-specific fields would look like (using the eepro100 driver again):

Some may find the syntax of embedded struct initialization awkward or even a bit ugly. So far, it’s the best way we’ve found to do what we want…

Registration¶

The driver registers the structure on startup. For drivers that have no bus-specific fields (i.e. don’t have a bus-specific driver structure), they would use driver_register and pass a pointer to their struct device_driver object.

Most drivers, however, will have a bus-specific structure and will need to register with the bus using something like pci_driver_register.

It is important that drivers register their driver structure as early as possible. Registration with the core initializes several fields in the struct device_driver object, including the reference count and the lock. These fields are assumed to be valid at all times and may be used by the device model core or the bus driver.

Transition Bus Drivers¶

By defining wrapper functions, the transition to the new model can be made easier. Drivers can ignore the generic structure altogether and let the bus wrapper fill in the fields. For the callbacks, the bus can define generic callbacks that forward the call to the bus-specific callbacks of the drivers.

This solution is intended to be only temporary. In order to get class information in the driver, the drivers must be modified anyway. Since converting drivers to the new model should reduce some infrastructural complexity and code size, it is recommended that they are converted as class information is added.

Access¶

Once the object has been registered, it may access the common fields of the object, like the lock and the list of devices:

The devices field is a list of all the devices that have been bound to the driver. The LDM core provides a helper function to operate on all the devices a driver controls. This helper locks the driver on each node access, and does proper reference counting on each device as it accesses it.

sysfs¶

When a driver is registered, a sysfs directory is created in its bus’s directory. In this directory, the driver can export an interface to userspace to control operation of the driver on a global basis; e.g. toggling debugging output in the driver.

A future feature of this directory will be a ‘devices’ directory. This directory will contain symlinks to the directories of devices it supports.

Callbacks¶

The probe() entry is called in task context, with the bus’s rwsem locked and the driver partially bound to the device. Drivers commonly use container_of() to convert “dev” to a bus-specific type, both in probe() and other routines. That type often provides device resource data, such as pci_dev.resource[] or platform_device.resources, which is used in addition to dev->platform_data to initialize the driver.

This callback holds the driver-specific logic to bind the driver to a given device. That includes verifying that the device is present, that it’s a version the driver can handle, that driver data structures can be allocated and initialized, and that any hardware can be initialized. Drivers often store a pointer to their state with dev_set_drvdata(). When the driver has successfully bound itself to that device, then probe() returns zero and the driver model code will finish its part of binding the driver to that device.

A driver’s probe() may return a negative errno value to indicate that the driver did not bind to this device, in which case it should have released all resources it allocated.

Optionally, probe() may return -EPROBE_DEFER if the driver depends on resources that are not yet available (e.g., supplied by a driver that hasn’t initialized yet). The driver core will put the device onto the deferred probe list and will try to call it again later. If a driver must defer, it should return -EPROBE_DEFER as early as possible to reduce the amount of time spent on setup work that will need to be unwound and reexecuted at a later time.

-EPROBE_DEFER must not be returned if probe() has already created child devices, even if those child devices are removed again in a cleanup path. If -EPROBE_DEFER is returned after a child device has been registered, it may result in an infinite loop of .probe() calls to the same driver.

sync_state is called only once for a device. It’s called when all the consumer devices of the device have successfully probed. The list of consumers of the device is obtained by looking at the device links connecting that device to its consumer devices.

The first attempt to call sync_state() is made during late_initcall_sync() to give firmware and drivers time to link devices to each other. During the first attempt at calling sync_state(), if all the consumers of the device at that point in time have already probed successfully, sync_state() is called right away. If there are no consumers of the device during the first attempt, that too is considered as “all consumers of the device have probed” and sync_state() is called right away.

If during the first attempt at calling sync_state() for a device, there are still consumers that haven’t probed successfully, the sync_state() call is postponed and reattempted in the future only when one or more consumers of the device probe successfully. If during the reattempt, the driver core finds that there are one or more consumers of the device that haven’t probed yet, then sync_state() call is postponed again.

A typical use case for sync_state() is to have the kernel cleanly take over management of devices from the bootloader. For example, if a device is left on and at a particular hardware configuration by the bootloader, the device’s driver might need to keep the device in the boot configuration until all the consumers of the device have probed. Once all the consumers of the device have probed, the device’s driver can synchronize the hardware state of the device to match the aggregated software state requested by all the consumers. Hence the name sync_state().

While obvious examples of resources that can benefit from sync_state() include resources such as regulator, sync_state() can also be useful for complex resources like IOMMUs. For example, IOMMUs with multiple consumers (devices whose addresses are remapped by the IOMMU) might need to keep their mappings fixed at (or additive to) the boot configuration until all its consumers have probed.

While the typical use case for sync_state() is to have the kernel cleanly take over management of devices from the bootloader, the usage of sync_state() is not restricted to that. Use it whenever it makes sense to take an action after all the consumers of a device have probed:

remove is called to unbind a driver from a device. This may be called if a device is physically removed from the system, if the driver module is being unloaded, during a reboot sequence, or in other cases.

It is up to the driver to determine if the device is present or not. It should free any resources allocated specifically for the device; i.e. anything in the device’s driver_data field.

If the device is still present, it should quiesce the device and place it into a supported low-power state.

suspend is called to put the device in a low power state.

Источник

Platform Devices and Drivers¶

See
for the driver model interface to the platform bus: platform_device, and platform_driver. This pseudo-bus is used to connect devices on busses with minimal infrastructure, like those used to integrate peripherals on many system-on-chip processors, or some “legacy” PC interconnects; as opposed to large formally specified ones like PCI or USB.

Platform devices¶

Platform devices are devices that typically appear as autonomous entities in the system. This includes legacy port-based devices and host bridges to peripheral buses, and most controllers integrated into system-on-chip platforms. What they usually have in common is direct addressing from a CPU bus. Rarely, a platform_device will be connected through a segment of some other kind of bus; but its registers will still be directly addressable.

Platform devices are given a name, used in driver binding, and a list of resources such as addresses and IRQs:

Platform drivers¶

Platform drivers follow the standard driver model convention, where discovery/enumeration is handled outside the drivers, and drivers provide probe() and remove() methods. They support power management and shutdown notifications using the standard conventions:

Note that probe() should in general verify that the specified device hardware actually exists; sometimes platform setup code can’t be sure. The probing can use device resources, including clocks, and device platform_data.

Platform drivers register themselves the normal way:

Or, in common situations where the device is known not to be hot-pluggable, the probe() routine can live in an init section to reduce the driver’s runtime memory footprint:

Kernel modules can be composed of several platform drivers. The platform core provides helpers to register and unregister an array of drivers:

If one of the drivers fails to register, all drivers registered up to that point will be unregistered in reverse order. Note that there is a convenience macro that passes THIS_MODULE as owner parameter:

Device Enumeration¶

As a rule, platform specific (and often board-specific) setup code will register platform devices:

The general rule is to register only those devices that actually exist, but in some cases extra devices might be registered. For example, a kernel might be configured to work with an external network adapter that might not be populated on all boards, or likewise to work with an integrated controller that some boards might not hook up to any peripherals.

In some cases, boot firmware will export tables describing the devices that are populated on a given board. Without such tables, often the only way for system setup code to set up the correct devices is to build a kernel for a specific target board. Such board-specific kernels are common with embedded and custom systems development.

In many cases, the memory and IRQ resources associated with the platform device are not enough to let the device’s driver work. Board setup code will often provide additional information using the device’s platform_data field to hold additional information.

Embedded systems frequently need one or more clocks for platform devices, which are normally kept off until they’re actively needed (to save power). System setup also associates those clocks with the device, so that calls to clk_get(&pdev->dev, clock_name) return them as needed.

Legacy Drivers: Device Probing¶

Some drivers are not fully converted to the driver model, because they take on a non-driver role: the driver registers its platform device, rather than leaving that for system infrastructure. Such drivers can’t be hotplugged or coldplugged, since those mechanisms require device creation to be in a different system component than the driver.

The only “good” reason for this is to handle older system designs which, like original IBM PCs, rely on error-prone “probe-the-hardware” models for hardware configuration. Newer systems have largely abandoned that model, in favor of bus-level support for dynamic configuration (PCI, USB), or device tables provided by the boot firmware (e.g. PNPACPI on x86). There are too many conflicting options about what might be where, and even educated guesses by an operating system will be wrong often enough to make trouble.

This style of driver is discouraged. If you’re updating such a driver, please try to move the device enumeration to a more appropriate location, outside the driver. This will usually be cleanup, since such drivers tend to already have “normal” modes, such as ones using device nodes that were created by PNP or by platform device setup.

None the less, there are some APIs to support such legacy drivers. Avoid using these calls except with such hotplug-deficient drivers:

You can use platform_device_alloc() to dynamically allocate a device, which you will then initialize with resources and platform_device_register() . A better solution is usually:

You can use platform_device_register_simple() as a one-step call to allocate and register a device.

Device Naming and Driver Binding¶

The platform_device.dev.bus_id is the canonical name for the devices. It’s built from two components:

platform_device.name … which is also used to for driver matching.

platform_device.id … the device instance number, or else “-1” to indicate there’s only one.

These are concatenated, so name/id “serial”/0 indicates bus_id “serial.0”, and “serial/3” indicates bus_id “serial.3”; both would use the platform_driver named “serial”. While “my_rtc”/-1 would be bus_id “my_rtc” (no instance id) and use the platform_driver called “my_rtc”.

Driver binding is performed automatically by the driver core, invoking driver probe() after finding a match between device and driver. If the probe() succeeds, the driver and device are bound as usual. There are three different ways to find such a match:

Whenever a device is registered, the drivers for that bus are checked for matches. Platform devices should be registered very early during system boot.

When a driver is registered using platform_driver_register(), all unbound devices on that bus are checked for matches. Drivers usually register later during booting, or by module loading.

Registering a driver using platform_driver_probe() works just like using platform_driver_register(), except that the driver won’t be probed later if another device registers. (Which is OK, since this interface is only for use with non-hotpluggable devices.)

Early Platform Devices and Drivers¶

The early platform interfaces provide platform data to platform device drivers early on during the system boot. The code is built on top of the early_param() command line parsing and can be executed very early on.

Example: “earlyprintk” class early serial console in 6 steps

1. Registering early platform device data¶

The architecture code registers platform device data using the function early_platform_add_devices(). In the case of early serial console this should be hardware configuration for the serial port. Devices registered at this point will later on be matched against early platform drivers.

2. Parsing kernel command line¶

The architecture code calls parse_early_param() to parse the kernel command line. This will execute all matching early_param() callbacks. User specified early platform devices will be registered at this point. For the early serial console case the user can specify port on the kernel command line as “earlyprintk=serial.0” where “earlyprintk” is the class string, “serial” is the name of the platform driver and 0 is the platform device id. If the id is -1 then the dot and the id can be omitted.

3. Installing early platform drivers belonging to a certain class¶

The architecture code may optionally force registration of all early platform drivers belonging to a certain class using the function early_platform_driver_register_all(). User specified devices from step 2 have priority over these. This step is omitted by the serial driver example since the early serial driver code should be disabled unless the user has specified port on the kernel command line.

4. Early platform driver registration¶

Compiled-in platform drivers making use of early_platform_init() are automatically registered during step 2 or 3. The serial driver example should use early_platform_init(“earlyprintk”, &platform_driver).

5. Probing of early platform drivers belonging to a certain class¶

The architecture code calls early_platform_driver_probe() to match registered early platform devices associated with a certain class with registered early platform drivers. Matched devices will get probed(). This step can be executed at any point during the early boot. As soon as possible may be good for the serial port case.

6. Inside the early platform driver probe()В¶

The driver code needs to take special care during early boot, especially when it comes to memory allocation and interrupt registration. The code in the probe() function can use is_early_platform_device() to check if it is called at early platform device or at the regular platform device time. The early serial driver performs register_console() at this point.

For further information, see
.

© Copyright The kernel development community.

Источник