24. Configuring a System

24.1. Introduction

RTEMS must be configured for an application. This configuration encompasses a variety of information including the length of each clock tick, the maximum number of each information RTEMS object that can be created, the application initialization tasks, the task scheduling algorithm to be used, and the device drivers in the application.

Although this information is contained in data structures that are used by RTEMS at system initialization time, the data structures themselves should only rarely to be generated by hand. RTEMS provides a set of macros system which provides a simple standard mechanism to automate the generation of these structures.

The RTEMS header file <rtems/confdefs.h> is at the core of the automatic generation of system configuration. It is based on the idea of setting macros which define configuration parameters of interest to the application and defaulting or calculating all others. This variety of macros can automatically produce all of the configuration data required for an RTEMS application.

As a general rule, application developers only specify values for the configuration parameters of interest to them. They define what resources or features they require. In most cases, when a parameter is not specified, it defaults to zero (0) instances, a standards compliant value, or disabled as appropriate. For example, by default there will be 256 task priority levels but this can be lowered by the application. This number of priority levels is required to be compliant with the RTEID/ORKID standards upon which the Classic API is based. There are similar cases where the default is selected to be compliant with with the POSIX standard.

For each configuration parameter in the configuration tables, the macro corresponding to that field is discussed. The RTEMS Maintainers expect that all systems can be easily configured using the <rtems/confdefs.h> mechanism and that using this mechanism will avoid internal RTEMS configuration changes impacting applications.

24.2. Default Value Selection Philosophy

The user should be aware that the defaults are intentionally set as low as possible. By default, no application resources are configured. The <rtems/confdefs.h> file ensures that at least one application task or thread is configured and that at least one of the initialization task/thread tables is configured.

24.3. Sizing the RTEMS Workspace

The RTEMS Workspace is a user-specified block of memory reserved for use by RTEMS. The application should NOT modify this memory. This area consists primarily of the RTEMS data structures whose exact size depends upon the values specified in the Configuration Table. In addition, task stacks and floating point context areas are dynamically allocated from the RTEMS Workspace.

The <rtems/confdefs.h> mechanism calculates the size of the RTEMS Workspace automatically. It assumes that all tasks are floating point and that all will be allocated the minimum stack space. This calculation includes the amount of memory that will be allocated for internal use by RTEMS. The automatic calculation may underestimate the workspace size truly needed by the application, in which case one can use the CONFIGURE_MEMORY_OVERHEAD macro to add a value to the estimate. See Chapter 24 Section 15.1 - Specify Memory Overhead for more details.

The memory area for the RTEMS Workspace is determined by the BSP. In case the RTEMS Workspace is too large for the available memory, then a fatal run-time error occurs and the system terminates.

The file <rtems/confdefs.h> will calculate the value of the work_space_size parameter of the Configuration Table. There are many parameters the application developer can specify to help <rtems/confdefs.h> in its calculations. Correctly specifying the application requirements via parameters such as CONFIGURE_EXTRA_TASK_STACKS and CONFIGURE_MAXIMUM_TASKS is critical for production software.

For each class of objects, the allocation can operate in one of two ways. The default way has an ceiling on the maximum number of object instances which can concurrently exist in the system. Memory for all instances of that object class is reserved at system initialization. The second way allocates memory for an initial number of objects and increases the current allocation by a fixed increment when required. Both ways allocate space from inside the RTEMS Workspace.

See Chapter 24 Section 7 - Unlimited Objects for more details about the second way, which allows for dynamic allocation of objects and therefore does not provide determinism. This mode is useful mostly for when the number of objects cannot be determined ahead of time or when porting software for which you do not know the object requirements.

The space needed for stacks and for RTEMS objects will vary from one version of RTEMS and from one target processor to another. Therefore it is safest to use <rtems/confdefs.h> and specify your application’s requirements in terms of the numbers of objects and multiples of RTEMS_MINIMUM_STACK_SIZE, as far as is possible. The automatic estimates of space required will in general change when:

  • a configuration parameter is changed,
  • task or interrupt stack sizes change,
  • the floating point attribute of a task changes,
  • task floating point attribute is altered,
  • RTEMS is upgraded, or
  • the target processor is changed.

Failure to provide enough space in the RTEMS Workspace may result in fatal run-time errors terminating the system.

24.4. Potential Issues with RTEMS Workspace Size Estimation

The <rtems/confdefs.h> file estimates the amount of memory required for the RTEMS Workspace. This estimate is only as accurate as the information given to <rtems/confdefs.h> and may be either too high or too low for a variety of reasons. Some of the reasons that <rtems/confdefs.h> may reserve too much memory for RTEMS are:

  • All tasks/threads are assumed to be floating point.

Conversely, there are many more reasons that the resource estimate could be too low:

  • Task/thread stacks greater than minimum size must be accounted for explicitly by developer.
  • Memory for messages is not included.
  • Device driver requirements are not included.
  • Network stack requirements are not included.
  • Requirements for add-on libraries are not included.

In general, <rtems/confdefs.h> is very accurate when given enough information. However, it is quite easy to use a library and forget to account for its resources.

24.5. Format to be followed for making changes in this file

MACRO NAME:
Should be alphanumeric. Can have ‘_’ (underscore).
DATA TYPE:
Please refer to all existing formats.
RANGE:

The range depends on the Data Type of the macro.

  • If the data type is of type task priority, then its value should be an integer in the range of 1 to 255.
  • If the data type is an integer, then it can have numbers, characters (in case the value is defined using another macro) and arithmetic operations (+, -, *, /).
  • If the data type is a function pointer the first character should be an alphabet or an underscore. The rest of the string can be alphanumeric.
  • If the data type is RTEMS Attributes or RTEMS Mode then the string should be alphanumeric.
  • If the data type is RTEMS NAME then the value should be an integer>=0 or RTEMS_BUILD_NAME( 'U', 'I', '1', ' ' )
DEFAULT VALUE:

The default value should be in the following formats- Please note that the ‘.’ (full stop) is necessary.

  • In case the value is not defined then: This is not defined by default.
  • If we know the default value then: The default value is XXX.
  • If the default value is BSP Specific then: This option is BSP specific.
DESCRIPTION:
The description of the macro. (No specific format)
NOTES:
Any further notes. (No specific format)

24.6. Configuration Example

In the following example, the configuration information for a system with a single message queue, four (4) tasks, and a timeslice of fifty (50) milliseconds is as follows:

#include <bsp.h>
#define CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER
#define CONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER
#define CONFIGURE_MICROSECONDS_PER_TICK   1000 /* 1 millisecond */
#define CONFIGURE_TICKS_PER_TIMESLICE       50 /* 50 milliseconds */
#define CONFIGURE_RTEMS_INIT_TASKS_TABLE
#define CONFIGURE_MAXIMUM_TASKS 4
#define CONFIGURE_MAXIMUM_MESSAGE_QUEUES 1
#define CONFIGURE_MESSAGE_BUFFER_MEMORY \
           CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE(20, sizeof(struct USER_MESSAGE))
#define CONFIGURE_INIT
#include <rtems/confdefs.h>

In this example, only a few configuration parameters are specified. The impact of these are as follows:

  • The example specified CONFIGURE_RTEMS_INIT_TASK_TABLE but did not specify any additional parameters. This results in a configuration of an application which will begin execution of a single initialization task named Init which is non-preemptible and at priority one (1).
  • By specifying CONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER, this application is configured to have a clock tick device driver. Without a clock tick device driver, RTEMS has no way to know that time is passing and will be unable to support delays and wall time. Further configuration details about time are provided. Per CONFIGURE_MICROSECONDS_PER_TICK and CONFIGURE_TICKS_PER_TIMESLICE, the user specified they wanted a clock tick to occur each millisecond, and that the length of a timeslice would be fifty (50) milliseconds.
  • By specifying CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER, the application will include a console device driver. Although the console device driver may support a combination of multiple serial ports and display and keyboard combinations, it is only required to provide a single device named /dev/console. This device will be used for Standard Input, Output and Error I/O Streams. Thus when CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER is specified, implicitly three (3) file descriptors are reserved for the Standard I/O Streams and those file descriptors are associated with /dev/console during initialization. All console devices are expected to support the POSIX*termios* interface.
  • The example above specifies via CONFIGURE_MAXIMUM_TASKS that the application requires a maximum of four (4) simultaneously existing Classic API tasks. Similarly, by specifying CONFIGURE_MAXIMUM_MESSAGE_QUEUES, there may be a maximum of only one (1) concurrently existent Classic API message queues.
  • The most surprising configuration parameter in this example is the use of CONFIGURE_MESSAGE_BUFFER_MEMORY. Message buffer memory is allocated from the RTEMS Workspace and must be accounted for. In this example, the single message queue will have up to twenty (20) messages of type struct USER_MESSAGE.
  • The CONFIGURE_INIT constant must be defined in order to make <rtems/confdefs.h> instantiate the configuration data structures. This can only be defined in one source file per application that includes <rtems/confdefs.h> or the symbol table will be instantiated multiple times and linking errors produced.

This example illustrates that parameters have default values. Among other things, the application implicitly used the following defaults:

  • All unspecified types of communications and synchronization objects in the Classic and POSIX Threads API have maximums of zero (0).
  • The filesystem will be the default filesystem which is the In-Memory File System (IMFS).
  • The application will have the default number of priority levels.
  • The minimum task stack size will be that recommended by RTEMS for the target architecture.

24.7. Unlimited Objects

In real-time embedded systems the RAM is normally a limited, critical resource and dynamic allocation is avoided as much as possible to ensure predictable, deterministic execution times. For such cases, see Chapter 24 Section 3 - Sizing the RTEMS Workspace for an overview of how to tune the size of the workspace. Frequently when users are porting software to RTEMS the precise resource requirements of the software is unknown. In these situations users do not need to control the size of the workspace very tightly because they just want to get the new software to run; later they can tune the workspace size as needed.

The following API-independent object classes can be configured in unlimited mode:

  • POSIX Keys
  • POSIX Key Value Pairs

The following object classes in the Classic API can be configured in unlimited mode:

  • Tasks
  • Timers
  • Semaphores
  • Message Queues
  • Periods
  • Barriers
  • Partitions
  • Regions
  • Ports

Additionally, the following object classes from the POSIX API can be configured in unlimited mode:

  • Threads
  • Mutexes
  • Condition Variables
  • Timers
  • Message Queues
  • Message Queue Descriptors
  • Semaphores
  • Barriers
  • Read/Write Locks
  • Spinlocks

The following object classes can not be configured in unlimited mode:

  • Drivers
  • File Descriptors
  • User Extensions
  • POSIX Queued Signals

Due to the memory requirements of unlimited objects it is strongly recommended to use them only in combination with the unified work areas. See Chapter 24 Section 12.1 - Separate or Unified Work Areas for more information on unified work areas.

The following example demonstrates how the two simple configuration defines for unlimited objects and unified works areas can replace many seperate configuration defines for supported object classes:

#define CONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER
#define CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER
#define CONFIGURE_UNIFIED_WORK_AREAS
#define CONFIGURE_UNLIMITED_OBJECTS
#define CONFIGURE_RTEMS_INIT_TASKS_TABLE
#define CONFIGURE_INIT
#include <rtems/confdefs.h>

Users are cautioned that using unlimited objects is not recommended for production software unless the dynamic growth is absolutely required. It is generally considered a safer embedded systems programming practice to know the system limits rather than experience an out of memory error at an arbitrary and largely unpredictable time in the field.

24.7.1. Per Object Class Unlimited Object Instances

When the number of objects is not known ahead of time, RTEMS provides an auto-extending mode that can be enabled individually for each object type by using the macro rtems_resource_unlimited. This takes a value as a parameter, and is used to set the object maximum number field in an API Configuration table. The value is an allocation unit size. When RTEMS is required to grow the object table it is grown by this size. The kernel will return the object memory back to the RTEMS Workspace when an object is destroyed. The kernel will only return an allocated block of objects to the RTEMS Workspace if at least half the allocation size of free objects remain allocated. RTEMS always keeps one allocation block of objects allocated. Here is an example of using rtems_resource_unlimited:

#define CONFIGURE_MAXIMUM_TASKS rtems_resource_unlimited(5)

Object maximum specifications can be evaluated with the rtems_resource_is_unlimited and``rtems_resource_maximum_per_allocation`` macros.

24.7.2. Unlimited Object Instances

To ease the burden of developers who are porting new software RTEMS also provides the capability to make all object classes listed above operate in unlimited mode in a simple manner. The application developer is only responsible for enabling unlimited objects and specifying the allocation size.

24.7.3. Enable Unlimited Object Instances

CONSTANT:
CONFIGURE_UNLIMITED_OBJECTS
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_UNLIMITED_OBJECTS enables rtems_resource_unlimited mode for Classic API and POSIX API objects that do not already have a specific maximum limit defined.
NOTES:
When using unlimited objects, it is common practice to also specify CONFIGURE_UNIFIED_WORK_AREAS so the system operates with a single pool of memory for both RTEMS and application memory allocations.

24.7.4. Specify Unlimited Objects Allocation Size

CONSTANT:
CONFIGURE_UNLIMITED_ALLOCATION_SIZE
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
If not defined and CONFIGURE_UNLIMITED_OBJECTS is defined, the default value is eight (8).
DESCRIPTION:
CONFIGURE_UNLIMITED_ALLOCATION_SIZE provides an allocation size to use for rtems_resource_unlimited when using CONFIGURE_UNLIMITED_OBJECTS.
NOTES:
By allowing users to declare all resources as being unlimited the user can avoid identifying and limiting the resources used. CONFIGURE_UNLIMITED_OBJECTS does not support varying the allocation sizes for different objects; users who want that much control can define the rtems_resource_unlimited macros themselves.
#define CONFIGURE_UNLIMITED_OBJECTS
#define CONFIGURE_UNLIMITED_ALLOCATION_SIZE 5

24.8. Classic API Configuration

This section defines the Classic API related system configuration parameters supported by <rtems/confdefs.h>.

24.8.1. Specify Maximum Classic API Tasks

CONSTANT:
CONFIGURE_MAXIMUM_TASKS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_TASKS is the maximum number of Classic API Tasks that can be concurrently active.
NOTES:

This object class can be configured in unlimited allocation mode.

The calculations for the required memory in the RTEMS Workspace for tasks assume that each task has a minimum stack size and has floating point support enabled. The configuration parameter CONFIGURE_EXTRA_TASK_STACKS is used to specify task stack requirements ABOVE the minimum size required. See Chapter 24 Section 12.7 - Reserve Task/Thread Stack Memory Above Minimum for more information about CONFIGURE_EXTRA_TASK_STACKS.

The maximum number of POSIX threads is specified by CONFIGURE_MAXIMUM_POSIX_THREADS.

A future enhancement to <rtems/confdefs.h> could be to eliminate the assumption that all tasks have floating point enabled. This would require the addition of a new configuration parameter to specify the number of tasks which enable floating point support.

24.8.2. Specify Maximum Classic API Timers

CONSTANT:
CONFIGURE_ENABLE_CLASSIC_API_NOTEPADS
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default, and Classic API Notepads are not supported.
DESCRIPTION:
CONFIGURE_ENABLE_CLASSIC_API_NOTEPADS should be defined if the user wants to have support for Classic API Notepads in their application.
NOTES:
Disabling Classic API Notepads saves the allocation of sixteen (16) thirty-two bit integers. This saves sixty-four bytes per task/thread plus the allocation overhead. Notepads are rarely used in applications and this can save significant memory in a low RAM system. Classic API Notepads are deprecated, and this option has been removed from post 4.11 versions of RTEMS.

24.8.3. Specify Maximum Classic API Timers

CONSTANT:
CONFIGURE_MAXIMUM_TIMERS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_TIMERS is the maximum number of Classic API Timers that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.8.4. Specify Maximum Classic API Semaphores

CONSTANT:
CONFIGURE_MAXIMUM_SEMAPHORES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_SEMAPHORES is the maximum number of Classic API Semaphores that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.8.5. Specify Maximum Classic API Semaphores usable with MrsP

CONSTANT:
CONFIGURE_MAXIMUM_MRSP_SEMAPHORES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_MRSP_SEMAPHORES is the maximum number of Classic API Semaphores using the Multiprocessor Resource Sharing Protocol (MrsP) that can be concurrently active.
NOTES:
This configuration option is only used on SMP configurations. On uni-processor configurations the Priority Ceiling Protocol is used for MrsP semaphores and thus no extra memory is necessary.

24.8.6. Specify Maximum Classic API Message Queues

CONSTANT:
CONFIGURE_MAXIMUM_MESSAGE_QUEUES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_MESSAGE_QUEUES is the maximum number of Classic API Message Queues that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.8.7. Specify Maximum Classic API Barriers

CONSTANT:
CONFIGURE_MAXIMUM_BARRIERS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_BARRIERS is the maximum number of Classic API Barriers that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.8.8. Specify Maximum Classic API Periods

CONSTANT:
CONFIGURE_MAXIMUM_PERIODS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_PERIODS is the maximum number of Classic API Periods that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.8.9. Specify Maximum Classic API Partitions

CONSTANT:
CONFIGURE_MAXIMUM_PARTITIONS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_PARTITIONS is the maximum number of Classic API Partitions that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.8.10. Specify Maximum Classic API Regions

CONSTANT:
CONFIGURE_MAXIMUM_REGIONS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_REGIONS is the maximum number of Classic API Regions that can be concurrently active.
NOTES:
None.

24.8.11. Specify Maximum Classic API Ports

CONSTANT:
CONFIGURE_MAXIMUM_PORTS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_PORTS is the maximum number of Classic API Ports that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.8.12. Specify Maximum Classic API User Extensions

CONSTANT:
CONFIGURE_MAXIMUM_USER_EXTENSIONS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_USER_EXTENSIONS is the maximum number of Classic API User Extensions that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.9. Classic API Initialization Tasks Table Configuration

The <rtems/confdefs.h> configuration system can automatically generate an Initialization Tasks Table named Initialization_tasks with a single entry. The following parameters control the generation of that table.

24.9.1. Instantiate Classic API Initialization Task Table

CONSTANT:
CONFIGURE_RTEMS_INIT_TASKS_TABLE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:

CONFIGURE_RTEMS_INIT_TASKS_TABLE is defined if the user wishes to use a Classic RTEMS API Initialization Task Table. The table built by <rtems/confdefs.h> specifies the parameters for a single task. This is sufficient for applications which initialization the system from a single task.

By default, this field is not defined as the user MUST select their own API for initialization tasks.

NOTES:

The application may choose to use the initialization tasks or threads table from another API.

A compile time error will be generated if the user does not configure any initialization tasks or threads.

24.9.2. Specifying Classic API Initialization Task Entry Point

CONSTANT:
CONFIGURE_INIT_TASK_ENTRY_POINT
DATA TYPE:
Task entry function pointer (rtems_task_entry).
RANGE:
Valid task entry function pointer.
DEFAULT VALUE:
The default value is Init.
DESCRIPTION:
CONFIGURE_INIT_TASK_ENTRY_POINT is the entry point (a.k.a. function name) of the single initialization task defined by the Classic API Initialization Tasks Table.
NOTES:
The user must implement the function Init or the function name provided in this configuration parameter.

24.9.3. Specifying Classic API Initialization Task Name

CONSTANT:
CONFIGURE_INIT_TASK_NAME
DATA TYPE:
RTEMS Name (rtems_name).
RANGE:
Any value.
DEFAULT VALUE:
The default value is rtems_build_name( 'U', 'I', '1', ' ' ).
DESCRIPTION:
CONFIGURE_INIT_TASK_NAME is the name of the single initialization task defined by the Classic API Initialization Tasks Table.
NOTES:
None.

24.9.4. Specifying Classic API Initialization Task Stack Size

CONSTANT:
CONFIGURE_INIT_TASK_STACK_SIZE
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is RTEMS_MINIMUM_STACK_SIZE.
DESCRIPTION:
CONFIGURE_INIT_TASK_STACK_SIZE is the stack size of the single initialization task defined by the Classic API Initialization Tasks Table.
NOTES:
If the stack size specified is greater than the configured minimum, it must be accounted for in CONFIGURE_EXTRA_TASK_STACKS. See Chapter 24 Section 12.7 - Reserve Task/Thread Stack Memory Above Minimum for more information about CONFIGURE_EXTRA_TASK_STACKS.

24.9.5. Specifying Classic API Initialization Task Priority

CONSTANT:
CONFIGURE_INIT_TASK_PRIORITY
DATA TYPE:
RTEMS Task Priority (rtems_task_priority).
RANGE:
One (1) to CONFIGURE_MAXIMUM_PRIORITY.
DEFAULT VALUE:
The default value is 1, which is the highest priority in the Classic API.
DESCRIPTION:
CONFIGURE_INIT_TASK_PRIORITY is the initial priority of the single initialization task defined by the Classic API Initialization Tasks Table.
NOTES:
None.

24.9.6. Specifying Classic API Initialization Task Attributes

CONSTANT:
CONFIGURE_INIT_TASK_ATTRIBUTES
DATA TYPE:
RTEMS Attributes (rtems_attribute).
RANGE:
Valid task attribute sets.
DEFAULT VALUE:
The default value is RTEMS_DEFAULT_ATTRIBUTES.
DESCRIPTION:
CONFIGURE_INIT_TASK_ATTRIBUTES is the task attributes of the single initialization task defined by the Classic API Initialization Tasks Table.
NOTES:
None.

24.9.7. Specifying Classic API Initialization Task Modes

CONSTANT:
CONFIGURE_INIT_TASK_INITIAL_MODES
DATA TYPE:
RTEMS Mode (rtems_mode).
RANGE:
Valid task mode sets.
DEFAULT VALUE:
The default value is RTEMS_NO_PREEMPT.
DESCRIPTION:
CONFIGURE_INIT_TASK_INITIAL_MODES is the initial execution mode of the single initialization task defined by the Classic API Initialization Tasks Table.
NOTES:
None.

24.9.8. Specifying Classic API Initialization Task Arguments

CONSTANT:
CONFIGURE_INIT_TASK_ARGUMENTS
DATA TYPE:
RTEMS Task Argument (rtems_task_argument).
RANGE:
Complete range of the type.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_INIT_TASK_ARGUMENTS is the task argument of the single initialization task defined by the Classic API Initialization Tasks Table.
NOTES:
None.

24.9.9. Not Using Generated Initialization Tasks Table

CONSTANT:
CONFIGURE_HAS_OWN_INIT_TASK_TABLE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_HAS_OWN_INIT_TASK_TABLE is defined if the user wishes to define their own Classic API Initialization Tasks Table. This table should be named Initialization_tasks.
NOTES:
This is a seldom used configuration parameter. The most likely use case is when an application desires to have more than one initialization task.

24.10. POSIX API Configuration

The parameters in this section are used to configure resources for the RTEMS POSIX API. They are only relevant if the POSIX API is enabled at configure time using the --enable-posix option.

24.10.1. Specify Maximum POSIX API Threads

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_THREADS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_THREADS is the maximum number of POSIX API Threads that can be concurrently active.
NOTES:

This object class can be configured in unlimited allocation mode.

This calculations for the required memory in the RTEMS Workspace for threads assume that each thread has a minimum stack size and has floating point support enabled. The configuration parameter CONFIGURE_EXTRA_TASK_STACKS is used to specify thread stack requirements ABOVE the minimum size required. See Chapter 24 Section 12.7 - Reserve Task/Thread Stack Memory Above Minimum for more information about CONFIGURE_EXTRA_TASK_STACKS.

The maximum number of Classic API Tasks is specified by CONFIGURE_MAXIMUM_TASKS.

All POSIX threads have floating point enabled.

24.10.2. Specify Maximum POSIX API Mutexes

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_MUTEXES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_MUTEXES is the maximum number of POSIX API Mutexes that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.10.3. Specify Maximum POSIX API Condition Variables

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_CONDITION_VARIABLES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_CONDITION_VARIABLES is the maximum number of POSIX API Condition Variables that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.10.4. Specify Maximum POSIX API Keys

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_KEYS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_KEYS is the maximum number of POSIX API Keys that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.10.5. Specify Maximum POSIX API Timers

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_TIMERS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_TIMERS is the maximum number of POSIX API Timers that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.10.6. Specify Maximum POSIX API Queued Signals

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_QUEUED_SIGNALS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_QUEUED_SIGNALS is the maximum number of POSIX API Queued Signals that can be concurrently active.
NOTES:
None.

24.10.7. Specify Maximum POSIX API Message Queues

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_MESSAGE_QUEUES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_MESSAGE_QUEUES is the maximum number of POSIX API Message Queues that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.10.8. Specify Maximum POSIX API Message Queue Descriptors

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_MESSAGE_QUEUE_DESCRIPTORS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
greater than or equal to CONFIGURE_MAXIMUM_POSIX_MESSAGES_QUEUES
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_MESSAGE_QUEUE_DESCRIPTORS is the maximum number of POSIX API Message Queue Descriptors that can be concurrently active.
NOTES:

This object class can be configured in unlimited allocation mode.

CONFIGURE_MAXIMUM_POSIX_MESSAGE_QUEUE_DESCRIPTORS should be greater than or equal to CONFIGURE_MAXIMUM_POSIX_MESSAGE_QUEUES.

24.10.9. Specify Maximum POSIX API Semaphores

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_SEMAPHORES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_SEMAPHORES is the maximum number of POSIX API Semaphores that can be concurrently active.
NOTES:
None.

24.10.10. Specify Maximum POSIX API Barriers

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_BARRIERS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_BARRIERS is the maximum number of POSIX API Barriers that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.10.11. Specify Maximum POSIX API Spinlocks

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_SPINLOCKS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_SPINLOCKS is the maximum number of POSIX API Spinlocks that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.10.12. Specify Maximum POSIX API Read/Write Locks

CONSTANT:
CONFIGURE_MAXIMUM_POSIX_RWLOCKS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_POSIX_RWLOCKS is the maximum number of POSIX API Read/Write Locks that can be concurrently active.
NOTES:
This object class can be configured in unlimited allocation mode.

24.11. POSIX Initialization Threads Table Configuration

The <rtems/confdefs.h> configuration system can automatically generate a POSIX Initialization Threads Table named POSIX_Initialization_threads with a single entry. The following parameters control the generation of that table.

24.11.1. Instantiate POSIX API Initialization Thread Table

CONSTANT:

CONFIGURE_POSIX_INIT_THREAD_TABLE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This field is not defined by default, as the user MUST select their own API for initialization tasks.
DESCRIPTION:

CONFIGURE_POSIX_INIT_THREAD_TABLE is defined if the user wishes to use a POSIX API Initialization Threads Table. The table built by <rtems/confdefs.h> specifies the parameters for a single thread. This is sufficient for applications which initialization the system from a single task.

By default, this field is not defined as the user MUST select their own API for initialization tasks.

NOTES:

The application may choose to use the initialization tasks or threads table from another API.

A compile time error will be generated if the user does not configure any initialization tasks or threads.

24.11.2. Specifying POSIX API Initialization Thread Entry Point

CONSTANT:
CONFIGURE_POSIX_INIT_THREAD_ENTRY_POINT
DATA TYPE:
POSIX thread function pointer (void *(*entry_point)(void *)).
RANGE:
Undefined or a valid POSIX thread function pointer.
DEFAULT VALUE:
The default value is POSIX_Init.
DESCRIPTION:
CONFIGURE_POSIX_INIT_THREAD_ENTRY_POINT is the entry point (a.k.a. function name) of the single initialization thread defined by the POSIX API Initialization Threads Table.
NOTES:
The user must implement the function POSIX_Init or the function name provided in this configuration parameter.

24.11.3. Specifying POSIX API Initialization Thread Stack Size

CONSTANT:
CONFIGURE_POSIX_INIT_THREAD_STACK_SIZE
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 2 * RTEMS_MINIMUM_STACK_SIZE.
DESCRIPTION:
CONFIGURE_POSIX_INIT_THREAD_STACK_SIZE is the stack size of the single initialization thread defined by the POSIX API Initialization Threads Table.
NOTES:
If the stack size specified is greater than the configured minimum, it must be accounted for in CONFIGURE_EXTRA_TASK_STACKS. See Chapter 24 Section 12.7 - Reserve Task/Thread Stack Memory Above Minimum for more information about CONFIGURE_EXTRA_TASK_STACKS.

24.11.4. Not Using Generated POSIX Initialization Threads Table

CONSTANT:
CONFIGURE_POSIX_HAS_OWN_INIT_THREAD_TABLE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_POSIX_HAS_OWN_INIT_THREAD_TABLE is defined if the user wishes to define their own POSIX API Initialization Threads Table. This table should be named POSIX_Initialization_threads.
NOTES:
This is a seldom used configuration parameter. The most likely use case is when an application desires to have more than one initialization task.

24.12. Basic System Information

This section defines the general system configuration parameters supported by <rtems/confdefs.h>.

24.12.1. Separate or Unified Work Areas

CONSTANT:
CONFIGURE_UNIFIED_WORK_AREAS
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default, which specifies that the C Program Heap and the RTEMS Workspace will be separate.
DESCRIPTION:

When defined, the C Program Heap and the RTEMS Workspace will be one pool of memory.

When not defined, there will be separate memory pools for the RTEMS Workspace and C Program Heap.

NOTES:

Having separate pools does have some advantages in the event a task blows a stack or writes outside its memory area. However, in low memory systems the overhead of the two pools plus the potential for unused memory in either pool is very undesirable.

In high memory environments, this is desirable when you want to use the RTEMS “unlimited” objects option. You will be able to create objects until you run out of all available memory rather then just until you run out of RTEMS Workspace.

24.12.2. Length of Each Clock Tick

CONSTANT:
CONFIGURE_MICROSECONDS_PER_TICK
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
This is not defined by default. When not defined, the clock tick quantum is configured to be 10,000 microseconds which is ten (10) milliseconds.
DESCRIPTION:

This constant is used to specify the length of time between clock ticks.

When the clock tick quantum value is too low, the system will spend so much time processing clock ticks that it does not have processing time available to perform application work. In this case, the system will become unresponsive.

The lowest practical time quantum varies widely based upon the speed of the target hardware and the architectural overhead associated with interrupts. In general terms, you do not want to configure it lower than is needed for the application.

The clock tick quantum should be selected such that it all blocking and delay times in the application are evenly divisible by it. Otherwise, rounding errors will be introduced which may negatively impact the application.

NOTES:

This configuration parameter has no impact if the Clock Tick Device driver is not configured.

There may be BSP specific limits on the resolution or maximum value of a clock tick quantum.

24.12.3. Specifying Timeslicing Quantum

CONSTANT:
CONFIGURE_TICKS_PER_TIMESLICE
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 50.
DESCRIPTION:
This configuration parameter specifies the length of the timeslice quantum in ticks for each task.
NOTES:
This configuration parameter has no impact if the Clock Tick Device driver is not configured.

24.12.4. Specifying the Number of Thread Priority Levels

CONSTANT:
CONFIGURE_MAXIMUM_PRIORITY
DATA TYPE:
Unsigned integer (uint8_t).
RANGE:
Valid values for this configuration parameter must be one (1) less than than a power of two (2) between 4 and 256 inclusively. In other words, valid values are 3, 7, 31, 63, 127, and 255.
DEFAULT VALUE:
The default value is 255, because RTEMS must support 256 priority levels to be compliant with various standards. These priorities range from zero (0) to 255.
DESCRIPTION:

This configuration parameter specified the maximum numeric priority of any task in the system and one less that the number of priority levels in the system.

Reducing the number of priorities in the system reduces the amount of memory allocated from the RTEMS Workspace.

NOTES:

The numerically greatest priority is the logically lowest priority in the system and will thus be used by the IDLE task.

Priority zero (0) is reserved for internal use by RTEMS and is not available to applications.

With some schedulers, reducing the number of priorities can reduce the amount of memory used by the scheduler. For example, the Deterministic Priority Scheduler (DPS) used by default uses three pointers of storage per priority level. Reducing the number of priorities from 256 levels to sixteen (16) can reduce memory usage by about three (3) kilobytes.

24.12.5. Specifying the Minimum Task Size

CONSTANT:
CONFIGURE_MINIMUM_TASK_STACK_SIZE
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
This is not defined by default, which sets the executive to the recommended minimum stack size for this processor.
DESCRIPTION:

The configuration parameter is set to the number of bytes the application wants the minimum stack size to be for every task or thread in the system.

Adjusting this parameter should be done with caution. Examining the actual usage using the Stack Checker Usage Reporting facility is recommended.

NOTES:

This parameter can be used to lower the minimum from that recommended. This can be used in low memory systems to reduce memory consumption for stacks. However, this must be done with caution as it could increase the possibility of a blown task stack.

This parameter can be used to increase the minimum from that recommended. This can be used in higher memory systems to reduce the risk of stack overflow without performing analysis on actual consumption.

24.12.6. Configuring the Size of the Interrupt Stack

CONSTANT:
CONFIGURE_INTERRUPT_STACK_SIZE
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is CONFIGURE_MINIMUM_TASK_STACK_SIZE, which is the minimum interrupt stack size.
DESCRIPTION:
CONFIGURE_INTERRUPT_STACK_SIZE is set to the size of the interrupt stack. The interrupt stack size is often set by the BSP but since this memory may be allocated from the RTEMS Workspace, it must be accounted for.
NOTES:

In some BSPs, changing this constant does NOT change the size of the interrupt stack, only the amount of memory reserved for it.

Patches which result in this constant only being used in memory calculations when the interrupt stack is intended to be allocated from the RTEMS Workspace would be welcomed by the RTEMS Project.

24.12.7. Reserve Task/Thread Stack Memory Above Minimum

CONSTANT:
CONFIGURE_EXTRA_TASK_STACKS
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Undefined or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
This configuration parameter is set to the number of bytes the applications wishes to add to the task stack requirements calculated by <rtems/confdefs.h>.
NOTES:
This parameter is very important. If the application creates tasks with stacks larger then the minimum, then that memory is NOT accounted for by <rtems/confdefs.h>.

24.12.8. Automatically Zeroing the RTEMS Workspace and C Program Heap

CONSTANT:
CONFIGURE_ZERO_WORKSPACE_AUTOMATICALLY
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default, unless overridden by the BSP. The default is NOT to zero out the RTEMS Workspace or C Program Heap.
DESCRIPTION:
This macro indicates whether RTEMS should zero the RTEMS Workspace and C Program Heap as part of its initialization. If defined, the memory regions are zeroed. Otherwise, they are not.
NOTES:
Zeroing memory can add significantly to system boot time. It is not necessary for RTEMS but is often assumed by support libraries.

24.12.9. Enable The Task Stack Usage Checker

CONSTANT:
CONFIGURE_STACK_CHECKER_ENABLED
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default, and thus stack checking is disabled.
DESCRIPTION:
This configuration parameter is defined when the application wishes to enable run-time stack bounds checking.
NOTES:

In 4.9 and older, this configuration parameter was named STACK_CHECKER_ON.

This increases the time required to create tasks as well as adding overhead to each context switch.

24.12.10. Specify Application Specific User Extensions

CONSTANT:
CONFIGURE_INITIAL_EXTENSIONS
DATA TYPE:
List of user extension initializers (rtems_extensions_table).
RANGE:
Undefined or a list of one or more user extensions.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
If CONFIGURE_INITIAL_EXTENSIONS is defined by the application, then this application specific set of initial extensions will be placed in the initial extension table.
NOTES:
None.

24.13. Configuring Custom Task Stack Allocation

RTEMS allows the application or BSP to define its own allocation and deallocation methods for task stacks. This can be used to place task stacks in special areas of memory or to utilize a Memory Management Unit so that stack overflows are detected in hardware.

24.13.1. Custom Task Stack Allocator Initialization

CONSTANT:
CONFIGURE_TASK_STACK_ALLOCATOR_INIT
DATA TYPE:
Function pointer.
RANGE:
Undefined, NULL or valid function pointer.
DEFAULT VALUE:
The default value is NULL, which indicates that task stacks will be allocated from the RTEMS Workspace.
DESCRIPTION:
CONFIGURE_TASK_STACK_ALLOCATOR_INIT configures the initialization method for an application or BSP specific task stack allocation implementation.
NOTES:
A correctly configured system must configure the following to be consistent:
  • CONFIGURE_TASK_STACK_ALLOCATOR_INIT
  • CONFIGURE_TASK_STACK_ALLOCATOR
  • CONFIGURE_TASK_STACK_DEALLOCATOR

24.13.2. Custom Task Stack Allocator

CONSTANT:
CONFIGURE_TASK_STACK_ALLOCATOR
DATA TYPE:
Function pointer.
RANGE:
Undefined or valid function pointer.
DEFAULT VALUE:
The default value is _Workspace_Allocate, which indicates that task stacks will be allocated from the RTEMS Workspace.
DESCRIPTION:
CONFIGURE_TASK_STACK_ALLOCATOR may point to a user provided routine to allocate task stacks.
NOTES:
A correctly configured system must configure the following to be consistent:
  • CONFIGURE_TASK_STACK_ALLOCATOR_INIT
  • CONFIGURE_TASK_STACK_ALLOCATOR
  • CONFIGURE_TASK_STACK_DEALLOCATOR

24.13.3. Custom Task Stack Deallocator

CONSTANT:
CONFIGURE_TASK_STACK_DEALLOCATOR
DATA TYPE:
Function pointer.
RANGE:
Undefined or valid function pointer.
DEFAULT VALUE:
The default value is _Workspace_Free, which indicates that task stacks will be allocated from the RTEMS Workspace.
DESCRIPTION:
CONFIGURE_TASK_STACK_DEALLOCATOR may point to a user provided routine to free task stacks.
NOTES:
A correctly configured system must configure the following to be consistent:
  • CONFIGURE_TASK_STACK_ALLOCATOR_INIT
  • CONFIGURE_TASK_STACK_ALLOCATOR
  • CONFIGURE_TASK_STACK_DEALLOCATOR

24.14. Configuring Memory for Classic API Message Buffers

This section describes the configuration parameters related to specifying the amount of memory reserved for Classic API Message Buffers.

24.14.1. Calculate Memory for a Single Classic Message API Message Queue

CONSTANT:
CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE(max_messages, size_per)
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is None.
DESCRIPTION:

This is a helper macro which is used to assist in computing the total amount of memory required for message buffers. Each message queue will have its own configuration with maximum message size and maximum number of pending messages.

The interface for this macro is as follows:

CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE(max_messages, size_per)

Where max_messages is the maximum number of pending messages and size_per is the size in bytes of the user message.

NOTES:

This macro is only used in support of CONFIGURE_MESSAGE_BUFFER_MEMORY.

24.14.2. Reserve Memory for All Classic Message API Message Queues

CONSTANT:
CONFIGURE_MESSAGE_BUFFER_MEMORY
DATA TYPE:
integer summation macro
RANGE:
undefined (zero) or calculation resulting in a positive integer
DEFAULT VALUE:
This is not defined by default, and zero (0) memory is reserved.
DESCRIPTION:
This macro is set to the number of bytes the application requires to be reserved for pending Classic API Message Queue buffers.
NOTES:

The following illustrates how the help macro CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE can be used to assist in calculating the message buffer memory required. In this example, there are two message queues used in this application. The first message queue has maximum of 24 pending messages with the message structure defined by the type one_message_type. The other message queue has maximum of 500 pending messages with the message structure defined by the type other_message_type.

#define CONFIGURE_MESSAGE_BUFFER_MEMORY \
            (CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE( \
                 24, sizeof(one_message_type) \
             ) + \
             CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE( \
                 500, sizeof(other_message_type) \
             )

24.15. Seldom Used Configuration Parameters

This section describes configuration parameters supported by <rtems/confdefs.h> which are seldom used by applications. These parameters tend to be oriented to debugging system configurations and providing work-arounds when the memory estimated by <rtems/confdefs.h> is incorrect.

24.15.1. Specify Memory Overhead

CONSTANT:
CONFIGURE_MEMORY_OVERHEAD
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
This parameter is set to the number of kilobytes the application wishes to add to the requirements calculated by <rtems/confdefs.h>.
NOTES:
This configuration parameter should only be used when it is suspected that a bug in <rtems/confdefs.h> has resulted in an underestimation. Typically the memory allocation will be too low when an application does not account for all message queue buffers or task stacks.

24.15.2. Do Not Generate Configuration Information

CONSTANT:
CONFIGURE_HAS_OWN_CONFIGURATION_TABLE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
This configuration parameter should only be defined if the application is providing their own complete set of configuration tables.
NOTES:
None.

24.16. C Library Support Configuration

This section defines the file system and IO library related configuration parameters supported by <rtems/confdefs.h>.

24.16.1. Specify Maximum Number of File Descriptors

CONSTANT:
CONFIGURE_LIBIO_MAXIMUM_FILE_DESCRIPTORS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
If CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER is defined, then the default value is 3, otherwise the default value is 0. Three file descriptors allows RTEMS to support standard input, output, and error I/O streams on /dev/console.
DESCRIPTION:
This configuration parameter is set to the maximum number of file like objects that can be concurrently open.
NOTES:
None.

24.16.2. Disable POSIX Termios Support

CONSTANT:
CONFIGURE_TERMIOS_DISABLED
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default, and resources are reserved for the termios functionality.
DESCRIPTION:
This configuration parameter is defined if the software implementing POSIX termios functionality is not going to be used by this application.
NOTES:
The termios support library should not be included in an application executable unless it is directly referenced by the application or a device driver.

24.16.3. Specify Maximum Termios Ports

CONSTANT:
CONFIGURE_NUMBER_OF_TERMIOS_PORTS
DATA TYPE:
Unsigned integer.
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 1, so a console port can be used.
DESCRIPTION:
This configuration parameter is set to the number of ports using the termios functionality. Each concurrently active termios port requires resources.
NOTES:
If the application will be using serial ports including, but not limited to, the Console Device (e.g. CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER), then it is highly likely that this configuration parameter should NOT be is defined.

24.17. File System Configuration Parameters

This section defines File System related configuration parameters.

24.17.1. Providing Application Specific Mount Table

CONSTANT:
CONFIGURE_HAS_OWN_MOUNT_TABLE
DATA TYPE:
Undefined or an array of type rtems_filesystem_mount_table_t.
RANGE:
Undefined or an array of type rtems_filesystem_mount_table_t.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
This configuration parameter is defined when the application provides their own filesystem mount table. The mount table is an array of rtems_filesystem_mount_table_t entries pointed to by the global variable rtems_filesystem_mount_table. The number of entries in this table is in an integer variable named rtems_filesystem_mount_table_t.
NOTES:
None.

24.17.2. Configure devFS as Root File System

CONSTANT:
CONFIGURE_USE_DEVFS_AS_BASE_FILESYSTEM
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default. If no other root file system configuration parameters are specified, the IMFS will be used as the root file system.
DESCRIPTION:
This configuration parameter is defined if the application wishes to use the device-only filesytem as the root file system.
NOTES:

The device-only filesystem supports only device nodes and is smaller in executable code size than the full IMFS and miniIMFS.

The devFS is comparable in functionality to the pseudo-filesystem name space provided before RTEMS release 4.5.0.

24.17.3. Specifying Maximum Devices for devFS

CONSTANT:
CONFIGURE_MAXIMUM_DEVICES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
If BSP_MAXIMUM_DEVICES is defined, then the default value is BSP_MAXIMUM_DEVICES, otherwise the default value is 4.
DESCRIPTION:
CONFIGURE_MAXIMUM_DEVICES is defined to the number of individual devices that may be registered in the device file system (devFS).
NOTES:
This option is specific to the device file system (devFS) and should not be confused with the CONFIGURE_MAXIMUM_DRIVERS option. This parameter only impacts the devFS and thus is only used by <rtems/confdefs.h> when CONFIGURE_USE_DEVFS_AS_BASE_FILESYSTEM is specified.

24.17.4. Disable File System Support

CONSTANT:
CONFIGURE_APPLICATION_DISABLE_FILESYSTEM
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default. If no other root file system configuration parameters are specified, the IMFS will be used as the root file system.
DESCRIPTION:
This configuration parameter is defined if the application dose not intend to use any kind of filesystem support. This include the device infrastructure necessary to support printf().
NOTES:
None.

24.17.5. Use a Root IMFS with a Minimalistic Feature Set

CONSTANT:
CONFIGURE_USE_MINIIMFS_AS_BASE_FILESYSTEM
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:

In case this configuration option is defined, then the following configuration options will be defined as well

  • CONFIGURE_IMFS_DISABLE_CHMOD,
  • CONFIGURE_IMFS_DISABLE_CHOWN,
  • CONFIGURE_IMFS_DISABLE_UTIME,
  • CONFIGURE_IMFS_DISABLE_LINK,
  • CONFIGURE_IMFS_DISABLE_SYMLINK,
  • CONFIGURE_IMFS_DISABLE_READLINK,
  • CONFIGURE_IMFS_DISABLE_RENAME, and
  • CONFIGURE_IMFS_DISABLE_UNMOUNT.

24.17.6. Specify Block Size for IMFS

CONSTANT:
CONFIGURE_IMFS_MEMFILE_BYTES_PER_BLOCK
DATA TYPE:
Boolean feature macro.
RANGE:
Valid values for this configuration parameter are a power of two (2) between 16 and 512 inclusive. In other words, valid values are 16, 32, 64, 128, 256,and 512.
DEFAULT VALUE:
The default IMFS block size is 128 bytes.
DESCRIPTION:

This configuration parameter specifies the block size for in-memory files managed by the IMFS. The configured block size has two impacts. The first is the average amount of unused memory in the last block of each file. For example, when the block size is 512, on average one-half of the last block of each file will remain unused and the memory is wasted. In contrast, when the block size is 16, the average unused memory per file is only 8 bytes. However, it requires more allocations for the same size file and thus more overhead per block for the dynamic memory management.

Second, the block size has an impact on the maximum size file that can be stored in the IMFS. With smaller block size, the maximum file size is correspondingly smaller. The following shows the maximum file size possible based on the configured block size:

  • when the block size is 16 bytes, the maximum file size is 1,328 bytes.
  • when the block size is 32 bytes, the maximum file size is 18,656 bytes.
  • when the block size is 64 bytes, the maximum file size is 279,488 bytes.
  • when the block size is 128 bytes, the maximum file size is 4,329,344 bytes.
  • when the block size is 256 bytes, the maximum file size is 68,173,568 bytes.
  • when the block size is 512 bytes, the maximum file size is 1,082,195,456 bytes.

24.17.7. Disable Change Owner Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_CHOWN
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to change the owner is disabled in the root IMFS.

24.17.8. Disable Change Mode Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_CHMOD
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to change the mode is disabled in the root IMFS.

24.17.9. Disable Change Times Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_UTIME
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to change times is disabled in the root IMFS.

24.17.13. Disable Rename Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_RENAME
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to rename nodes is disabled in the root IMFS.

24.17.14. Disable Directory Read Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_READDIR
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to read a directory is disabled in the root IMFS. It is still possible to open nodes in a directory.

24.17.15. Disable Mount Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_MOUNT
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to mount other file systems is disabled in the root IMFS.

24.17.16. Disable Unmount Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_UNMOUNT
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to unmount file systems is disabled in the root IMFS.

24.17.17. Disable Make Nodes Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_MKNOD
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to make directories, devices, regular files and FIFOs is disabled in the root IMFS.

24.17.18. Disable Make Files Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_MKNOD_FILE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to make regular files is disabled in the root IMFS.

24.17.19. Disable Remove Nodes Support of Root IMFS

CONSTANT:
CONFIGURE_IMFS_DISABLE_RMNOD
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
In case this configuration option is defined, then the support to remove nodes is disabled in the root IMFS.

24.18. Block Device Cache Configuration

This section defines Block Device Cache (bdbuf) related configuration parameters.

24.18.1. Enable Block Device Cache

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_LIBBLOCK
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
Provides a Block Device Cache configuration.
NOTES:
Each option of the Block Device Cache configuration can be explicitly set by the user with the configuration options below. The Block Device Cache is used for example by the RFS and DOSFS file systems.

24.18.2. Size of the Cache Memory

CONSTANT:
CONFIGURE_BDBUF_CACHE_MEMORY_SIZE
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 32768 bytes.
DESCRIPTION:
Size of the cache memory in bytes.
NOTES:
None.

24.18.3. Minimum Size of a Buffer

CONSTANT:
CONFIGURE_BDBUF_BUFFER_MIN_SIZE
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 512 bytes.
DESCRIPTION:
Defines the minimum size of a buffer in bytes.
NOTES:
None.

24.18.4. Maximum Size of a Buffer

CONSTANT:
CONFIGURE_BDBUF_BUFFER_MAX_SIZE
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
It must be positive and an integral multiple of the buffer minimum size.
DEFAULT VALUE:
The default value is 4096 bytes.
DESCRIPTION:
Defines the maximum size of a buffer in bytes.
NOTES:
None.

24.18.5. Swapout Task Swap Period

CONSTANT:
CONFIGURE_SWAPOUT_SWAP_PERIOD
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 250 milliseconds.
DESCRIPTION:
Defines the swapout task swap period in milliseconds.
NOTES:
None.

24.18.6. Swapout Task Maximum Block Hold Time

CONSTANT:
CONFIGURE_SWAPOUT_BLOCK_HOLD
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 1000 milliseconds.
DESCRIPTION:
Defines the swapout task maximum block hold time in milliseconds.
NOTES:
None.

24.18.7. Swapout Task Priority

CONSTANT:
CONFIGURE_SWAPOUT_TASK_PRIORITY
DATA TYPE:
Task priority (rtems_task_priority).
RANGE:
Valid task priority.
DEFAULT VALUE:
The default value is 15.
DESCRIPTION:
Defines the swapout task priority.
NOTES:
None.

24.18.8. Maximum Blocks per Read-Ahead Request

CONSTANT:
CONFIGURE_BDBUF_MAX_READ_AHEAD_BLOCKS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
Defines the maximum blocks per read-ahead request.
NOTES:
A value of 0 disables the read-ahead task (default). The read-ahead task will issue speculative read transfers if a sequential access pattern is detected. This can improve the performance on some systems.

24.18.9. Maximum Blocks per Write Request

CONSTANT:
CONFIGURE_BDBUF_MAX_WRITE_BLOCKS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 16.
DESCRIPTION:
Defines the maximum blocks per write request.
NOTES:
None.

24.18.10. Task Stack Size of the Block Device Cache Tasks

CONSTANT:
CONFIGURE_BDBUF_TASK_STACK_SIZE
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is RTEMS_MINIMUM_STACK_SIZE.
DESCRIPTION:
Defines the task stack size of the Block Device Cache tasks in bytes.
NOTES:
None.

24.18.11. Read-Ahead Task Priority

CONSTANT:
CONFIGURE_BDBUF_READ_AHEAD_TASK_PRIORITY
DATA TYPE:
Task priority (rtems_task_priority).
RANGE:
Valid task priority.
DEFAULT VALUE:
The default value is 15.
DESCRIPTION:
Defines the read-ahead task priority.
NOTES:
None.

24.18.12. Swapout Worker Task Count

CONSTANT:
CONFIGURE_SWAPOUT_WORKER_TASKS
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
Defines the swapout worker task count.
NOTES:
None.

24.18.13. Swapout Worker Task Priority

CONSTANT:
CONFIGURE_SWAPOUT_WORKER_TASK_PRIORITY
DATA TYPE:
Task priority (rtems_task_priority).
RANGE:
Valid task priority.
DEFAULT VALUE:
The default value is 15.
DESCRIPTION:
Defines the swapout worker task priority.
NOTES:
None.

24.19. BSP Specific Settings

This section describes BSP specific configuration settings used by <rtems/confdefs.h>. The BSP specific configuration settings are defined in <bsp.h>.

24.19.1. Disable BSP Configuration Settings

CONSTANT:
CONFIGURE_DISABLE_BSP_SETTINGS
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
All BSP specific configuration settings can be disabled by the application with the CONFIGURE_DISABLE_BSP_SETTINGS option.
NOTES:
None.

24.19.2. Specify BSP Supports sbrk()

CONSTANT:
CONFIGURE_MALLOC_BSP_SUPPORTS_SBRK
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This option is BSP specific.
DESCRIPTION:

This configuration parameter is defined by a BSP to indicate that it does not allocate all available memory to the C Program Heap used by the Malloc Family of routines.

If defined, when malloc() is unable to allocate memory, it will call the BSP supplied sbrk() to obtain more memory.

NOTES:
This parameter should not be defined by the application. Only the BSP knows how it allocates memory to the C Program Heap.

24.19.3. Specify BSP Specific Idle Task

CONSTANT:
BSP_IDLE_TASK_BODY
DATA TYPE:
Function pointer.
RANGE:
Undefined or valid function pointer.
DEFAULT VALUE:
This option is BSP specific.
DESCRIPTION:
If BSP_IDLE_TASK_BODY is defined by the BSP and CONFIGURE_IDLE_TASK_BODY is not defined by the application, then this BSP specific idle task body will be used.
NOTES:
As it has knowledge of the specific CPU model, system controller logic, and peripheral buses, a BSP specific IDLE task may be capable of turning components off to save power during extended periods of no task activity

24.19.4. Specify BSP Suggested Value for IDLE Task Stack Size

CONSTANT:
BSP_IDLE_TASK_STACK_SIZE
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Undefined or positive.
DEFAULT VALUE:
This option is BSP specific.
DESCRIPTION:
If BSP_IDLE_TASK_STACK_SIZE is defined by the BSP and CONFIGURE_IDLE_TASK_STACK_SIZE is not defined by the application, then this BSP suggested idle task stack size will be used.
NOTES:

The order of precedence for configuring the IDLE task stack size is:

  • RTEMS default minimum stack size.
  • If defined, then CONFIGURE_MINIMUM_TASK_STACK_SIZE.
  • If defined, then the BSP specific BSP_IDLE_TASK_SIZE.
  • If defined, then the application specified CONFIGURE_IDLE_TASK_SIZE.

24.19.5. Specify BSP Specific User Extensions

CONSTANT:
BSP_INITIAL_EXTENSION
DATA TYPE:
List of user extension initializers (rtems_extensions_table).
RANGE:
Undefined or a list of user extension initializers.
DEFAULT VALUE:
This option is BSP specific.
DESCRIPTION:
If BSP_INITIAL_EXTENSION is defined by the BSP, then this BSP specific initial extension will be placed as the last entry in the initial extension table.
NOTES:
None.

24.19.6. Specifying BSP Specific Interrupt Stack Size

CONSTANT:
BSP_INTERRUPT_STACK_SIZE
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Undefined or positive.
DEFAULT VALUE:
This option is BSP specific.
DESCRIPTION:
If BSP_INTERRUPT_STACK_SIZE is defined by the BSP and CONFIGURE_INTERRUPT_STACK_SIZE is not defined by the application, then this BSP specific interrupt stack size will be used.
NOTES:
None.

24.19.7. Specifying BSP Specific Maximum Devices

CONSTANT:
BSP_MAXIMUM_DEVICES
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Undefined or positive.
DEFAULT VALUE:
This option is BSP specific.
DESCRIPTION:
If BSP_MAXIMUM_DEVICES is defined by the BSP and CONFIGURE_MAXIMUM_DEVICES is not defined by the application, then this BSP specific maximum device count will be used.
NOTES:
This option is specific to the device file system (devFS) and should not be confused with the CONFIGURE_MAXIMUM_DRIVERS option. This parameter only impacts the devFS and thus is only used by <rtems/confdefs.h> when CONFIGURE_USE_DEVFS_AS_BASE_FILESYSTEM is specified.

24.19.8. BSP Recommends RTEMS Workspace be Cleared

CONSTANT:
BSP_ZERO_WORKSPACE_AUTOMATICALLY
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This option is BSP specific.
DESCRIPTION:
If BSP_ZERO_WORKSPACE_AUTOMATICALLY is defined by the BSP and CONFIGURE_ZERO_WORKSPACE_AUTOMATICALLY is not defined by the application, then the workspace will be zeroed automatically.
NOTES:
Zeroing memory can add significantly to system boot time. It is not necessary for RTEMS but is often assumed by support libraries.

24.19.9. Specify BSP Prerequisite Drivers

CONSTANT:
CONFIGURE_BSP_PREREQUISITE_DRIVERS
DATA TYPE:
List of device driver initializers (rtems_driver_address_table).
RANGE:
Undefined or array of device drivers.
DEFAULT VALUE:
This option is BSP specific.
DESCRIPTION:
CONFIGURE_BSP_PREREQUISITE_DRIVERS is defined if the BSP has device drivers it needs to include in the Device Driver Table. This should be defined to the set of device driver entries that will be placed in the table at the FRONT of the Device Driver Table and initialized before any other drivers INCLUDING any application prerequisite drivers.
NOTES:
CONFIGURE_BSP_PREREQUISITE_DRIVERS is typically used by BSPs to configure common infrastructure such as bus controllers or probe for devices.

24.20. Idle Task Configuration

This section defines the IDLE task related configuration parameters supported by <rtems/confdefs.h>.

24.20.1. Specify Application Specific Idle Task Body

CONSTANT:
CONFIGURE_IDLE_TASK_BODY
DATA TYPE:
Function pointer.
RANGE:
Undefined or valid function pointer.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_IDLE_TASK_BODY is set to the function name corresponding to the application specific IDLE thread body. If not specified, the BSP or RTEMS default IDLE thread body will be used.
NOTES:
None.

24.20.2. Specify Idle Task Stack Size

CONSTANT:
CONFIGURE_IDLE_TASK_STACK_SIZE
DATA TYPE:
Unsigned integer (size_t).
RANGE:
Undefined or positive.
DEFAULT VALUE:
The default value is RTEMS_MINIMUM_STACK_SIZE.
DESCRIPTION:
CONFIGURE_IDLE_TASK_STACK_SIZE is set to the desired stack size for the IDLE task.
NOTES:
None.

24.20.3. Specify Idle Task Performs Application Initialization

CONSTANT:
CONFIGURE_IDLE_TASK_INITIALIZES_APPLICATION
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default, the user is assumed to provide one or more initialization tasks.
DESCRIPTION:
CONFIGURE_IDLE_TASK_INITIALIZES_APPLICATION is set to indicate that the user has configured NO user initialization tasks or threads and that the user provided IDLE task will perform application initialization and then transform itself into an IDLE task.
NOTES:

If you use this option be careful, the user IDLE task CANNOT block at all during the initialization sequence. Further, once application initialization is complete, it must make itself preemptible and enter an IDLE body loop.

The IDLE task must run at the lowest priority of all tasks in the system.

24.21. Scheduler Algorithm Configuration

This section defines the configuration parameters related to selecting a scheduling algorithm for an application. For the schedulers built into RTEMS, the configuration is straightforward. All that is required is to define the configuration macro which specifies which scheduler you want for in your application. The currently available schedulers are:

The pluggable scheduler interface also enables the user to provide their own scheduling algorithm. If you choose to do this, you must define multiple configuration macros.

24.21.1. Use Deterministic Priority Scheduler

CONSTANT:
CONFIGURE_SCHEDULER_PRIORITY
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is defined by default. This is the default scheduler and specifying this configuration parameter is redundant.
DESCRIPTION:
The Deterministic Priority Scheduler is the default scheduler in RTEMS for uni-processor applications and is designed for predictable performance under the highest loads. It can block or unblock a thread in a constant amount of time. This scheduler requires a variable amount of memory based upon the number of priorities configured in the system.
NOTES:
This scheduler may be explicitly selected by defining CONFIGURE_SCHEDULER_PRIORITY although this is equivalent to the default behavior.

24.21.2. Use Simple Priority Scheduler

CONSTANT:
CONFIGURE_SCHEDULER_SIMPLE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:

When defined, the Simple Priority Scheduler is used at the thread scheduling algorithm. This is an alternative scheduler in RTEMS. It is designed to provide the same task scheduling behaviour as the Deterministic Priority Scheduler while being simpler in implementation and uses less memory for data management. It maintains a single sorted list of all ready threads. Thus blocking or unblocking a thread is not a constant time operation with this scheduler.

This scheduler may be explicitly selected by defining CONFIGURE_SCHEDULER_SIMPLE.

NOTES:
This scheduler is appropriate for use in small systems where RAM is limited.

24.21.3. Use Earliest Deadline First Scheduler

CONSTANT:
CONFIGURE_SCHEDULER_EDF
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:

The Earliest Deadline First Scheduler (EDF) is an alternative scheduler in RTEMS for uni-processor applications. The EDF schedules tasks with dynamic priorities equal to deadlines. The deadlines are declared using only Rate Monotonic manager which handles periodic behavior. Period is always equal to deadline. If a task does not have any deadline declared or the deadline is cancelled, the task is considered a background task which is scheduled in case no deadline-driven tasks are ready to run. Moreover, multiple background tasks are scheduled according their priority assigned upon initialization. All ready tasks reside in a single ready queue.

This scheduler may be explicitly selected by defining CONFIGURE_SCHEDULER_EDF.

NOTES:
None.

24.21.4. Use Constant Bandwidth Server Scheduler

CONSTANT:
CONFIGURE_SCHEDULER_CBS
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:

The Constant Bandwidth Server Scheduler (CBS) is an alternative scheduler in RTEMS for uni-processor applications. The CBS is a budget aware extension of EDF scheduler. The goal of this scheduler is to ensure temporal isolation of tasks. The CBS is equipped with a set of additional rules and provides with an extensive API.

This scheduler may be explicitly selected by defining CONFIGURE_SCHEDULER_CBS.

NOTES:
None.

24.21.5. Use Deterministic Priority SMP Scheduler

CONSTANT:
CONFIGURE_SCHEDULER_PRIORITY_SMP
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:

The Deterministic Priority SMP Scheduler is derived from the Deterministic Priority Scheduler but is capable of scheduling threads across multiple processors.

In a configuration with SMP enabled at configure time, it may be explicitly selected by defining CONFIGURE_SCHEDULER_PRIORITY_SMP.

NOTES:

This scheduler is only available when RTEMS is configured with SMP support enabled.

This scheduler is currently the default in SMP configurations and is only selected when CONFIGURE_SMP_APPLICATION is defined.

24.21.6. Use Simple SMP Priority Scheduler

CONSTANT:
CONFIGURE_SCHEDULER_SIMPLE_SMP
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:

The Simple SMP Priority Scheduler is derived from the Simple Priority Scheduler but is capable of scheduling threads across multiple processors. It is designed to provide the same task scheduling behaviour as the Deterministic Priority Scheduler while distributing threads across multiple processors. Being based upon the Simple Priority Scheduler, it also maintains a single sorted list of all ready threads. Thus blocking or unblocking a thread is not a constant time operation with this scheduler.

In addition, when allocating threads to processors, the algorithm is not constant time. This algorithm was not designed with efficiency as a primary design goal. Its primary design goal was to provide an SMP-aware scheduling algorithm that is simple to understand.

In a configuration with SMP enabled at configure time, it may be explicitly selected by defining CONFIGURE_SCHEDULER_SIMPLE_SMP.

NOTES:
This scheduler is only available when RTEMS is configured with SMP support enabled.

24.21.7. Configuring a Scheduler Name

CONSTANT:
CONFIGURE_SCHEDULER_NAME
DATA TYPE:
RTEMS Name (rtems_name).
RANGE:
Any value.
DEFAULT VALUE:
The default name is
  • "UCBS" for the Uni-Processor CBS scheduler,
  • "UEDF" for the Uni-Processor EDF scheduler,
  • "UPD " for the Uni-Processor Deterministic Priority scheduler,
  • "UPS " for the Uni-Processor Simple Priority scheduler,
  • "MPA " for the Multi-Processor Priority Affinity scheduler, and
  • "MPD " for the Multi-Processor Deterministic Priority scheduler, and
  • "MPS " for the Multi-Processor Simple Priority scheduler.
DESCRIPTION:
Schedulers can be identified via rtems_scheduler_ident. The name of the scheduler is determined by the configuration.
NOTES:
None.

24.21.8. Configuring a User Provided Scheduler

CONSTANT:
CONFIGURE_SCHEDULER_USER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:

RTEMS allows the application to provide its own task/thread scheduling algorithm. In order to do this, one must define CONFIGURE_SCHEDULER_USER to indicate the application provides its own scheduling algorithm. If CONFIGURE_SCHEDULER_USER is defined then the following additional macros must be defined:

  • CONFIGURE_SCHEDULER_CONTEXT must be defined to a static definition of the scheduler context of the user scheduler.
  • CONFIGURE_SCHEDULER_CONTROLS must be defined to a scheduler control initializer for the user scheduler.
  • CONFIGURE_SCHEDULER_USER_PER_THREAD must be defined to the type of the per-thread information of the user scheduler.
NOTES:
At this time, the mechanics and requirements for writing a new scheduler are evolving and not fully documented. It is recommended that you look at the existing Deterministic Priority Scheduler in cpukit/score/src/schedulerpriority*.c for guidance. For guidance on the configuration macros, please examine cpukit/sapi/include/confdefs.h for how these are defined for the Deterministic Priority Scheduler.

24.21.9. Configuring Clustered Schedulers

Clustered scheduling helps to control the worst-case latencies in a multi-processor system. The goal is to reduce the amount of shared state in the system and thus prevention of lock contention. Modern multi-processor systems tend to have several layers of data and instruction caches. With clustered scheduling it is possible to honour the cache topology of a system and thus avoid expensive cache synchronization traffic.

We have clustered scheduling in case the set of processors of a system is partitioned into non-empty pairwise-disjoint subsets. These subsets are called clusters. Clusters with a cardinality of one are partitions. Each cluster is owned by exactly one scheduler instance. In order to use clustered scheduling the application designer has to answer two questions.

  1. How is the set of processors partitioned into clusters?
  2. Which scheduler is used for which cluster?
CONFIGURATION:

The schedulers in an SMP system are statically configured on RTEMS. Firstly the application must select which scheduling algorithms are available with the following defines

  • CONFIGURE_SCHEDULER_PRIORITY_SMP,
  • CONFIGURE_SCHEDULER_SIMPLE_SMP, and
  • CONFIGURE_SCHEDULER_PRIORITY_AFFINITY_SMP.

This is necessary to calculate the per-thread overhead introduced by the schedulers. After these definitions the configuration file must #include <rtems/scheduler.h> to have access to scheduler specific configuration macros. Each scheduler needs a context to store state information at run-time. To provide a context for each scheduler is the next step. Use the following macros to create scheduler contexts

  • RTEMS_SCHEDULER_CONTEXT_PRIORITY_SMP(name, prio_count),
  • RTEMS_SCHEDULER_CONTEXT_SIMPLE_SMP(name), and
  • RTEMS_SCHEDULER_CONTEXT_PRIORITY_AFFINITY_SMP(name, prio_count).

The name parameter is used as part of a designator for a global variable, so the usual C/C++ designator rules apply. Additional parameters are scheduler specific. The schedulers are registered in the system via the scheduler table. To create the scheduler table define CONFIGURE_SCHEDULER_CONTROLS to a list of the following scheduler control initializers

  • RTEMS_SCHEDULER_CONTROL_PRIORITY_SMP(name, obj_name),
  • RTEMS_SCHEDULER_CONTROL_SIMPLE_SMP(name, obj_name), and
  • RTEMS_SCHEDULER_CONTROL_PRIORITY_AFFINITY_SMP(name, obj_name).

The name parameter must correspond to the parameter defining the scheduler context. The obj_name determines the scheduler object name and can be used in rtems_scheduler_ident() to get the scheduler object identifier.

The last step is to define which processor uses which scheduler. For this purpose a scheduler assignment table must be defined. The entry count of this table must be equal to the configured maximum processors (CONFIGURE_SMP_MAXIMUM_PROCESSORS). A processor assignment to a scheduler can be optional or mandatory. The boot processor must have a scheduler assigned. In case the system needs more mandatory processors than available then a fatal run-time error will occur. To specify the scheduler assignments define CONFIGURE_SMP_SCHEDULER_ASSIGNMENTS to a list of RTEMS_SCHEDULER_ASSIGN(index, attr) and RTEMS_SCHEDULER_ASSIGN_NO_SCHEDULER macros. The index parameter must be a valid index into the scheduler table. The attr parameter defines the scheduler assignment attributes. By default a scheduler assignment to a processor is optional. For the scheduler assignment attribute use one of the mutually exclusive variants

  • RTEMS_SCHEDULER_ASSIGN_DEFAULT,
  • RTEMS_SCHEDULER_ASSIGN_PROCESSOR_MANDATORY, and
  • RTEMS_SCHEDULER_ASSIGN_PROCESSOR_OPTIONAL.
ERRORS:

In case one of the scheduler indices in``CONFIGURE_SMP_SCHEDULER_ASSIGNMENTS`` is invalid a link-time error will occur with an undefined reference to RTEMS_SCHEDULER_INVALID_INDEX.

Some fatal errors may occur in case of scheduler configuration inconsistencies or a lack of processors on the system. The fatal source is RTEMS_FATAL_SOURCE_SMP. None of the errors is internal.

  • SMP_FATAL_BOOT_PROCESSOR_NOT_ASSIGNED_TO_SCHEDULER - the boot processor must have a scheduler assigned.
  • SMP_FATAL_MANDATORY_PROCESSOR_NOT_PRESENT - there exists a mandatory processor beyond the range of physically or virtually available processors. The processor demand must be reduced for this system.
  • SMP_FATAL_START_OF_MANDATORY_PROCESSOR_FAILED - the start of a mandatory processor failed during system initialization. The system may not have this processor at all or it could be a problem with a boot loader for example. Check the CONFIGURE_SMP_SCHEDULER_ASSIGNMENTS definition.
  • SMP_FATAL_MULTITASKING_START_ON_UNASSIGNED_PROCESSOR - it is not allowed to start multitasking on a processor with no scheduler assigned.
EXAMPLE:

The following example shows a scheduler configuration for a hypothetical product using two chip variants. One variant has four processors which is used for the normal product line and another provides eight processors for the high-performance product line. The first processor performs hard-real time control of actuators and sensors. The second processor is not used by RTEMS at all and runs a Linux instance to provide a graphical user interface. The additional processors are used for a worker thread pool to perform data processing operations.

The processors managed by RTEMS use two Deterministic Priority scheduler instances capable of dealing with 256 priority levels. The scheduler with index zero has the name "IO ". The scheduler with index one has the name "WORK". The scheduler assignments of the first, third and fourth processor are mandatory, so the system must have at least four processors, otherwise a fatal run-time error will occur during system startup. The processor assignments for the fifth up to the eighth processor are optional so that the same application can be used for the normal and high-performance product lines. The second processor has no scheduler assigned and runs Linux. A hypervisor will ensure that the two systems cannot interfere in an undesirable way.

#define CONFIGURE_SMP_MAXIMUM_PROCESSORS 8
#define CONFIGURE_MAXIMUM_PRIORITY 255
/* Make the scheduler algorithm available */
#define CONFIGURE_SCHEDULER_PRIORITY_SMP
#include <rtems/scheduler.h>
/* Create contexts for the two scheduler instances */
RTEMS_SCHEDULER_CONTEXT_PRIORITY_SMP(io, CONFIGURE_MAXIMUM_PRIORITY + 1);
RTEMS_SCHEDULER_CONTEXT_PRIORITY_SMP(work, CONFIGURE_MAXIMUM_PRIORITY + 1);
/* Define the scheduler table */
#define CONFIGURE_SCHEDULER_CONTROLS \\
            RTEMS_SCHEDULER_CONTROL_PRIORITY_SMP( \
                io, \
                rtems_build_name('I', 'O', ' ', ' ') \
            ), \
            RTEMS_SCHEDULER_CONTROL_PRIORITY_SMP( \
                work, \
                rtems_build_name('W', 'O', 'R', 'K') \
            )
 /* Define the processor to scheduler assignments */
#define CONFIGURE_SMP_SCHEDULER_ASSIGNMENTS \
            RTEMS_SCHEDULER_ASSIGN(0, RTEMS_SCHEDULER_ASSIGN_PROCESSOR_MANDATORY), \
            RTEMS_SCHEDULER_ASSIGN_NO_SCHEDULER, \
            RTEMS_SCHEDULER_ASSIGN(1, RTEMS_SCHEDULER_ASSIGN_PROCESSOR_MANDATORY), \
            RTEMS_SCHEDULER_ASSIGN(1, RTEMS_SCHEDULER_ASSIGN_PROCESSOR_MANDATORY), \
            RTEMS_SCHEDULER_ASSIGN(1, RTEMS_SCHEDULER_ASSIGN_PROCESSOR_OPTIONAL), \
            RTEMS_SCHEDULER_ASSIGN(1, RTEMS_SCHEDULER_ASSIGN_PROCESSOR_OPTIONAL), \
            RTEMS_SCHEDULER_ASSIGN(1, RTEMS_SCHEDULER_ASSIGN_PROCESSOR_OPTIONAL), \
            RTEMS_SCHEDULER_ASSIGN(1, RTEMS_SCHEDULER_ASSIGN_PROCESSOR_OPTIONAL)

24.22. SMP Specific Configuration Parameters

When RTEMS is configured to support SMP target systems, there are other configuration parameters which apply.

24.22.1. Enable SMP Support for Applications

CONSTANT:
CONFIGURE_SMP_APPLICATION
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_SMP_APPLICATION must be defined to enable SMP support for the application.
NOTES:
This define may go away in the future in case all RTEMS components are SMP ready. This configuration define is ignored on uni-processor configurations.

24.22.2. Specify Maximum Processors in SMP System

CONSTANT:
CONFIGURE_SMP_MAXIMUM_PROCESSORS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Defined or undefined.
DEFAULT VALUE:
The default value is 1, (if CONFIGURE_SMP_APPLICATION is defined).
DESCRIPTION:
CONFIGURE_SMP_MAXIMUM_PROCESSORS must be set to the number of processors in the SMP configuration.
NOTES:
If there are more processors available than configured, the rest will be ignored. This configuration define is ignored on uni-processor configurations.

24.23. Device Driver Table

This section defines the configuration parameters related to the automatic generation of a Device Driver Table. As <rtems/confdefs.h> only is aware of a small set of standard device drivers, the generated Device Driver Table is suitable for simple applications with no custom device drivers.

Note that network device drivers are not configured in the Device Driver Table.

24.23.1. Specifying the Maximum Number of Device Drivers

CONSTANT:

CONFIGURE_MAXIMUM_DRIVERS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
This is computed by default, and is set to the number of device drivers configured using the CONFIGURE_APPLICATIONS_NEEDS_XXX_DRIVER configuration parameters.
DESCRIPTION:
CONFIGURE_MAXIMUM_DRIVERS is defined as the number of device drivers per node.
NOTES:
If the application will dynamically install device drivers, then this configuration parameter must be larger than the number of statically configured device drivers. Drivers configured using the CONFIGURE_APPLICATIONS_NEEDS_XXX_DRIVER configuration parameters are statically installed.

24.23.2. Enable Console Device Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER is defined if the application wishes to include the Console Device Driver.
NOTES:

This device driver is responsible for providing standard input and output using /dev/console.

BSPs should be constructed in a manner that allows printk() to work properly without the need for the console driver to be configured.

24.23.3. Enable Clock Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER is defined if the application wishes to include the Clock Device Driver.
NOTES:

This device driver is responsible for providing a regular interrupt which invokes a clock tick directive.

If neither the Clock Driver not Benchmark Timer is enabled and the configuration parameter CONFIGURE_APPLICATION_DOES_NOT_NEED_CLOCK_DRIVER is not defined, then a compile time error will occur.

24.23.4. Enable the Benchmark Timer Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_TIMER_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_NEEDS_TIMER_DRIVER is defined if the application wishes to include the Timer Driver. This device driver is used to benchmark execution times by the RTEMS Timing Test Suites.
NOTES:
If neither the Clock Driver not Benchmark Timer is enabled and the configuration parameter CONFIGURE_APPLICATION_DOES_NOT_NEED_CLOCK_DRIVER is not defined, then a compile time error will occur.

24.23.5. Specify Clock and Benchmark Timer Drivers Are Not Needed

CONSTANT:
CONFIGURE_APPLICATION_DOES_NOT_NEED_CLOCK_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_DOES_NOT_NEED_CLOCK_DRIVER is defined when the application does NOT want the Clock Device Driver and is NOT using the Timer Driver. The inclusion or exclusion of the Clock Driver must be explicit in user applications.
NOTES:
This configuration parameter is intended to prevent the common user error of using the Hello World example as the baseline for an application and leaving out a clock tick source.

24.23.6. Enable Real-Time Clock Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_RTC_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_NEEDS_RTC_DRIVER is defined if the application wishes to include the Real-Time Clock Driver.
NOTES:

Most BSPs do not include support for a real-time clock. This is because many boards do not include the required hardware.

If this is defined and the BSP does not have this device driver, then the user will get a link time error for an undefined symbol.

24.23.7. Enable the Watchdog Device Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_WATCHDOG_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_NEEDS_WATCHDOG_DRIVER is defined if the application wishes to include the Watchdog Driver.
NOTES:

Most BSPs do not include support for a watchdog device driver. This is because many boards do not include the required hardware.

If this is defined and the BSP does not have this device driver, then the user will get a link time error for an undefined symbol.

24.23.8. Enable the Graphics Frame Buffer Device Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_FRAME_BUFFER_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_NEEDS_FRAME_BUFFER_DRIVER is defined if the application wishes to include the BSP’s Frame Buffer Device Driver.
NOTES:

Most BSPs do not include support for a Frame Buffer Device Driver. This is because many boards do not include the required hardware.

If this is defined and the BSP does not have this device driver, then the user will get a link time error for an undefined symbol.

24.23.9. Enable Stub Device Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_STUB_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_NEEDS_STUB_DRIVER is defined if the application wishes to include the Stub Device Driver.
NOTES:
This device driver simply provides entry points that return successful and is primarily a test fixture. It is supported by all BSPs.

24.23.10. Specify Application Prerequisite Device Drivers

CONSTANT:
CONFIGURE_APPLICATION_PREREQUISITE_DRIVERS
DATA TYPE:
device driver entry structures
RANGE:
Undefined or set of device driver entry structures
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_PREREQUISITE_DRIVERS is defined if the application has device drivers it needs to include in the Device Driver Table. This should be defined to the set of device driver entries that will be placed in the table at the FRONT of the Device Driver Table and initialized before any other drivers EXCEPT any BSP prerequisite drivers.
NOTES:
In some cases, it is used by System On Chip BSPs to support peripheral buses beyond those normally found on the System On Chip. For example, this is used by one RTEMS system which has implemented a SPARC/ERC32 based board with VMEBus. The VMEBus Controller initialization is performed by a device driver configured via this configuration parameter.

24.23.11. Specify Extra Application Device Drivers

CONSTANT:
CONFIGURE_APPLICATION_EXTRA_DRIVERS
DATA TYPE:
device driver entry structures
RANGE:
Undefined or set of device driver entry structures
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_APPLICATION_EXTRA_DRIVERS is defined if the application has device drivers it needs to include in the Device Driver Table. This should be defined to the set of device driver entries that will be placed in the table at the END of the Device Driver Table.
NOTES:
None.

24.23.12. Enable /dev/null Device Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_NULL_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
This configuration variable is specified to enable /dev/null device driver.
NOTES:
This device driver is supported by all BSPs.

24.23.13. Enable /dev/zero Device Driver

CONSTANT:
CONFIGURE_APPLICATION_NEEDS_ZERO_DRIVER
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
This configuration variable is specified to enable /dev/zero device driver.
NOTES:
This device driver is supported by all BSPs.

24.23.14. Specifying Application Defined Device Driver Table

CONSTANT:
CONFIGURE_HAS_OWN_DEVICE_DRIVER_TABLE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default, indicating the <rtems/confdefs.h> is providing the device driver table.
DESCRIPTION:

CONFIGURE_HAS_OWN_DEVICE_DRIVER_TABLE is defined if the application wishes to provide their own Device Driver Table.

The table must be an array of rtems_driver_address_table entries named`` _IO_Driver_address_table``. The application must also provide a const variable _IO_Number_of_drivers of type size_t indicating the number of entries in the _IO_Driver_address_table.

NOTES:
It is expected that there the application would only rarely need to do this.

24.24. Multiprocessing Configuration

This section defines the multiprocessing related system configuration parameters supported by <rtems/confdefs.h>. They are only used if the Multiprocessing Support (distinct from the SMP support) is enabled at configure time using the --enable-multiprocessing option.

Additionally, this class of Configuration Constants are only applicable if CONFIGURE_MP_APPLICATION is defined.

24.24.1. Specify Application Will Use Multiprocessing

CONSTANT:
CONFIGURE_MP_APPLICATION
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
This configuration parameter must be defined to indicate that the application intends to be part of a multiprocessing configuration. Additional configuration parameters are assumed to be provided.
NOTES:
This has no impact unless RTEMS was configured and built using the --enable-multiprocessing option.

24.24.2. Configure Node Number in Multiprocessor Configuration

CONSTANT:
CONFIGURE_MP_NODE_NUMBER
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is NODE_NUMBER, which is assumed to be set by the compilation environment.
DESCRIPTION:
CONFIGURE_MP_NODE_NUMBER is the node number of this node in a multiprocessor system.
NOTES:
In the RTEMS Multiprocessing Test Suite, the node number is derived from the Makefile variable NODE_NUMBER. The same code is compiled with the NODE_NUMBER set to different values. The test programs behave differently based upon their node number.

24.24.3. Configure Maximum Node in Multiprocessor Configuration

CONSTANT:
CONFIGURE_MP_MAXIMUM_NODES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 2.
DESCRIPTION:
CONFIGURE_MP_MAXIMUM_NODES is the maximum number of nodes in a multiprocessor system.
NOTES:
None.

24.24.4. Configure Maximum Global Objects in Multiprocessor Configuration

CONSTANT:
CONFIGURE_MP_MAXIMUM_GLOBAL_OBJECTS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Positive.
DEFAULT VALUE:
The default value is 32.
DESCRIPTION:
CONFIGURE_MP_MAXIMUM_GLOBAL_OBJECTS is the maximum number of concurrently active global objects in a multiprocessor system.
NOTES:
This value corresponds to the total number of objects which can be created with the RTEMS_GLOBAL attribute.

24.24.5. Configure Maximum Proxies in Multiprocessor Configuration

CONSTANT:
CONFIGURE_MP_MAXIMUM_PROXIES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Undefined or positive.
DEFAULT VALUE:
The default value is 32.
DESCRIPTION:
CONFIGURE_MP_MAXIMUM_PROXIES is the maximum number of concurrently active thread/task proxies on this node in a multiprocessor system.
NOTES:
Since a proxy is used to represent a remote task/thread which is blocking on this node. This configuration parameter reflects the maximum number of remote tasks/threads which can be blocked on objects on this node.

24.24.6. Configure MPCI in Multiprocessor Configuration

CONSTANT:
CONFIGURE_MP_MPCI_TABLE_POINTER
DATA TYPE:
pointer to rtems_mpci_table
RANGE:
undefined or valid pointer
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_MP_MPCI_TABLE_POINTER is the pointer to the MPCI Configuration Table. The default value of this field is``&MPCI_table``.
NOTES:
RTEMS provides a Shared Memory MPCI Device Driver which can be used on any Multiprocessor System assuming the BSP provides the proper set of supporting methods.

24.24.7. Do Not Generate Multiprocessor Configuration Table

CONSTANT:
CONFIGURE_HAS_OWN_MULTIPROCESSING_TABLE
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_HAS_OWN_MULTIPROCESSING_TABLE is defined if the application wishes to provide their own Multiprocessing Configuration Table. The generated table is named Multiprocessing_configuration.
NOTES:
This is a configuration parameter which is very unlikely to be used by an application. If you find yourself wanting to use it in an application, please reconsider and discuss this on the RTEMS Users mailing list.

24.25. Ada Tasks

This section defines the system configuration parameters supported by <rtems/confdefs.h> related to configuring RTEMS to support a task using Ada tasking with GNAT/RTEMS.

These configuration parameters are only available when RTEMS is built with the --enable-ada configure option and the application specifies CONFIGURE_GNAT_RTEMS.

Additionally RTEMS includes an Ada language binding to the Classic API which has a test suite. This test suite is enabled only when``–enable-tests`` and --enable-expada are specified on the configure command.

24.25.1. Specify Application Includes Ada Code

CONSTANT:
CONFIGURE_GNAT_RTEMS
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_GNAT_RTEMS is defined to inform RTEMS that the GNAT Ada run-time is to be used by the application.
NOTES:
This configuration parameter is critical as it makes``<rtems/confdefs.h>`` configure the resources (POSIX API Threads, Mutexes, Condition Variables, and Keys) used implicitly by the GNAT run-time.

24.25.2. Specify the Maximum Number of Ada Tasks.

CONSTANT:
CONFIGURE_MAXIMUM_ADA_TASKS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Undefined or positive.
DEFAULT VALUE:
If CONFIGURE_GNAT_RTEMS is defined, then the default value is 20, otherwise the default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_ADA_TASKS is the number of Ada tasks that can be concurrently active in the system.
NOTES:
None.

24.25.3. Specify the Maximum Fake Ada Tasks

CONSTANT:
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 0.
DESCRIPTION:
CONFIGURE_MAXIMUM_FAKE_ADA_TASKS is the number of fake Ada tasks that can be concurrently active in the system. A fake Ada task is a non-Ada task that makes calls back into Ada code and thus implicitly uses the Ada run-time.
NOTES:
None.

24.26. PCI Library

This section defines the system configuration parameters supported by rtems/confdefs.h related to configuring the PCI Library for RTEMS.

The PCI Library startup behaviour can be configured in four different ways depending on how CONFIGURE_PCI_CONFIG_LIB is defined:

PCI_LIB_AUTO
Used to enable the PCI auto configuration software. PCI will be automatically probed, PCI buses enumerated, all devices and bridges will be initialized using Plug & Play software routines. The PCI device tree will be populated based on the PCI devices found in the system, PCI devices will be configured by allocating address region resources automatically in PCI space according to the BSP or host bridge driver set up.
PCI_LIB_READ
Used to enable the PCI read configuration software. The current PCI configuration is read to create the RAM representation (the PCI device tree) of the PCI devices present. PCI devices are assumed to already have been initialized and PCI buses enumerated, it is therefore required that a BIOS or a boot loader has set up configuration space prior to booting into RTEMS.
PCI_LIB_STATIC
Used to enable the PCI static configuration software. The user provides a PCI tree with information how all PCI devices are to be configured at compile time by linking in a custom struct pci_bus pci_hb tree. The static PCI library will not probe PCI for devices, instead it will assume that all devices defined by the user are present, it will enumerate the PCI buses and configure all PCI devices in static configuration accordingly. Since probe and allocation software is not needed the startup is faster, has smaller footprint and does not require dynamic memory allocation.
PCI_LIB_PERIPHERAL
Used to enable the PCI peripheral configuration. It is similar to PCI_LIB_STATIC, but it will never write the configuration to the PCI devices since PCI peripherals are not allowed to access PCI configuration space.

Note that selecting PCI_LIB_STATIC or PCI_LIB_PERIPHERAL but not defining pci_hb will reuslt in link errors. Note also that in these modes Plug & Play is not performed.

24.27. Go Tasks

24.27.1. Specify Application Includes Go Code

CONSTANT:
CONFIGURE_ENABLE_GO
DATA TYPE:
Boolean feature macro.
RANGE:
Defined or undefined.
DEFAULT VALUE:
This is not defined by default.
DESCRIPTION:
CONFIGURE_ENABLE_GO is defined to inform RTEMS that the Go run-time is to be used by the application.
NOTES:
The Go language support is experimental

24.27.2. Specify the maximum number of Go routines

CONSTANT:
CONFIGURE_MAXIMUM_GOROUTINES
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 400
DESCRIPTION:
CONFIGURE_MAXIMUM_GOROUTINES is defined to specify the maximum number of Go routines.
NOTES:
The Go language support is experimental

24.27.3. Specify the maximum number of Go Channels

CONSTANT:
CONFIGURE_MAXIMUM_GO_CHANNELS
DATA TYPE:
Unsigned integer (uint32_t).
RANGE:
Zero or positive.
DEFAULT VALUE:
The default value is 500
DESCRIPTION:
CONFIGURE_MAXIMUM_GO_CHANNELS is defined to specify the maximum number of Go channels.
NOTES:
The Go language support is experimental

24.28. Configuration Data Structures

It is recommended that applications be configured using <rtems/confdefs.h> as it is simpler and insulates applications from changes in the underlying data structures. However, it is sometimes important to understand the data structures that are automatically filled in by the configuration parameters. This section describes the primary configuration data structures.

If the user wishes to see the details of a particular data structure, they are are advised to look at the source code. After all, that is one of the advantages of RTEMS.