Home

8051 RTOS

1. PRIORITY 9 SCHEDULE FULL ACTIVATION 1 AUTOSTART FALSE MESSAGE sendHandlerl MESSAGE sendHandler2 MESSAGE sendHandler3 MESSAGE recCommand VSTACK 100 SSTACK 40 he ISR MonitorISR CATEGORY 2 LEVEL 4 ENBIT ES MESSAGE sendCommand he MESSAGE sendCommand MESSAGEPROPERTY SEND _ STATIC INTERNAL CDATATYPE myCommand RECEIVER recCommand hi he MESSAGE recCommand MESSAGEPROPERTY RECEIVE QUEUED INTERNAL SENDINGMESSAGE sendCommand QUEUESIZE 5 he NOTIFICATION ACTIVATETASK TASK monitor hi he APPMODE AUTOSTART APPMODE NONAUTOSTART EVENT internaldelay 3 9 8051 RTOS TASK bouncel hi PRIORITY SCHEDULE ACTIVATION AUTOSTART EVENT MESSAGE RESOURCE VSTACK SSTACK TASK bounce2 b PRIORITY SCHEDULE ACTIVATION AUTOSTART EVENT MESSAGE RESOURCE VSTACK SSTACK TASK bounce3 PRIORITY SCHEDULE ACTIVATION AUTOSTART EVENT MESSAGE RESOURCE VSTACK SSTACK 5 FULL 1 TRUE APPMODE AUTOSTART intervaldelay recHandler1l sem_printf 100 40 5 FULL 1 TRUE APPMODE AUTOSTART intervaldelay recHandler2 sem_printf 100 40 5 FULL 1 TRUE APPMODE AUTOSTART intervaldelay recHandler3 sem_printf 100 40 b ho b The OSEK VDX Implementation Language OIL TASK trsi PRIORITY 11 SCHEDULE FULL ACTIVATION 1 AUTOSTART FA
2. start the OS for current mode StartOS mode return 1 StartoOS is the one and only method defined by the OSEK VDX standard in which you can define the Application Mode for the current system environment To change modes the application must first shutdown and then the RTOS must restart using a the new mode 4 3 3 Changing Application Modes Application Reset Since sending operators across the whole ATM network is inefficient and costly a new system requirement could arise The ATM software shall be upgraded remotely at high speed and being service affecting Identically two mutually exclusive modes keep coexisting but no operator is needed on location How do we implement such a solution The ATM software can receive a command to change its mode the system shuts down and starts again with the other mode In order to download the following steps in order are necessary an operator in the bank headquarters issues remotely a command to enter download mode the ATM software undergoes an application reset and starts up again the link in download mode the operator sends download commands over the link and when the new image has been downloaded a command to enter normal mode is issued The ATM undergoes a second application reset and starts the new image in normal mode You should then implement a function like ChangeMode in order to change the mode void ChangeMode AppModeType mode if mode GetActiveApp
3. In fact the makefile calls the Tasking OIL Compiler TOC which preprocesses the application OIL file The TOC compiler outputs a number of configurational files written in C source code These files are used to build a dedicated RTOS library with the same name as the oi1 file This library is then built together with the rest of the application The compiler and assembly options of the project prevail while building the RTOS library only some compiler optimizations may change dd The RTOS library is only rebuilt upon changes in the OIL file since it constitutes its only dependency Changes in the application software will not affect the RTOS library 2 1 8051 RTOS 4 In your application source code files you must include the OSEK VDX standard OS and COM interfaces osek h to compile When you application use the Notification Flag mechanism you must also include the file lag h in your source code files 5 The RTOS system services used in the application software are extracted from the RTOS library during the linking phase The following table lists the files involved in an RTOS project Extension Description Application sou rce files c h asm mytypes h C source files header include files and optional hand coded assembly files are used to write the application code These files must be members of your DXP project and are used to build application objects You need to write myty
4. Description STATE Represents the state of a resource LOCKED UNLOCKED only if running in extended mode PRIORITY Ceiling priority of the resource NS property Property of the resource STANDARD INTERNAL or LINKED Table 12 3 Resource Debug Properties 12 4 Debugging an RTOS Application 12 5 How to Debug Alarms The debugger can display relevant information about all the alarms in the system 1 From the OSEK ORTI menu select MYOSEK and Alarms A window pops up showing a list with all the alarms in the system Every alarm is described with a set of properties The debugger displays the values of these properties The alarm properties are described in the next table Alarm Debug Property Description ALARMTIME CYCLETIME STATE COUNTER ACTION vs_task vs_event vs_callback Time left untill the alarm expires next Cycle time for cyclic alarms The value is O for non cyclic alarms Specifies whether the alarm is RUNNING STOPPED or INPROCESS the time has expired but the RTOS has not finished processing yet Counter on which the alarm is based Action ACTIVATETASK SETEVENT or ALARMCALLBACK that will be performed by the alarm when expiring Task that is activated by the alarm if ACTION is ACTIVATETASK or the task receiver of the event if ACTION is SETEVENT Event that is set when ACTION is SETEVENT Physical address of callback function when ACTION is ALARMCALLBACK Otherwi
5. e Basic tasks are supported e Extended tasks are supported 5 1 8051 RTOS 5 2 Defining a Task in the C Source db To configure a task you must declare a specific TASK object in the user OIL file of the project and assign values to all of its attributes Please refer to the OSEK VDX documentation for detailed information about all possible attributes of a task and how to use them With the macro TASK you can define a task in your application The name of the TASK object in the OIL file is passed as parameter to the macro You must always end the code of a task with the system service TerminateTask or ChainTask regardless of whether the RTOS defines a resistant default behaviour in case the task reaches a forbidden RET instruction For example to define the task mytask TASK mytask code for task mytask TerminateTask The OSEK VDX implementation uses this macro to encapsulate the implementation specific formatting of the task definition mytask is the identity of the task and is of the type TaskType During preprocessing the C name of the function that correspond to the task is created by adding the prefix _os_u_ to the task name You can then view the function with the debugger using the mingled name _os_u_mytask 5 2 Task management 5 3 The States of a Task A task goes through several different states during its lifetime A processor can only execute one instruction at a ti
6. vs LENO TR Indicates the length of the message data vs_DATAPTR Pointer to the receive message object for queued messages vs_DATAPTR points at the next available element of the queue if non empty vs_QueueUsed Number of non read messages already in the queue If vs_QueueUsed equals to 0 the queue is empty only for queue messages vs_QueueAvailable Number of messages that still might be received without suffering from overflow If vs_QueueAvailable equals to 0 the queue is full only for queue messages Table 12 6 Message Debug Properties 12 7 8051 RTOS 12 8 A Implementation Parameters Summary The implementation parameters provide detailed information concerning the functionality performance and memory demand From the implementation parameters you can obtain valuable information about the impact of the RTOS on your application 1 Introduction The OS OSEK VDX normative documents state that The operating system vendor provides a list of parameters specifying the implementation Detailed information is required concerning the functionality performance and memory demand From the implementation parameters you can obtain valuable information regarding the impact of the RTOS on your application There are three kinds of implementation parameters e Functionality Implementation Parameters They relate to the configuration of the system You should always take them into account
7. 10 RESOURCE R4 TASK T5 PRIORITY 9 RESOURCE R2 TASK T6 PRIORITY 8 RESOURCE R4 RESOURCE R3 hi The generated CEILING attributes are 7 13 8051 RTOS RESOURCE R1 CEILING 6 ko RESOURCE R2 CEILING 9 RESOURCE R3 CEILING 8 RESOURCE R4 CEILING 10 If we assume a fully preemptive policy TASK T priority 5 all tasks can preempt GetResource R1 priority 6 all tasks but T1 can preempt GetResource R2 priority 9 only T4 can preempt GetResource R3 ReleaseResource R3 ReleaseResource R2 priority 6 ReleaseResource R3 priority 5 TerminateTask In the example below the priorities become different TASK T 7 14 priority 5 all tasks can preempt GetResource R1 priority 6 all tasks but Tl can preempt GetResource R3 priority 8 T4 and T5 can preempt GetResource R2 priority 9 only T4 can preempt ReleaseResource R2 priority 8 ReleaseResource R3 priority 6 ReleaseResource R3 priority 5 TerminateTask Resource Management 7 4 Grouping Tasks What is a group of Tasks The RTOS allows tasks to combine aspects of preemptive and non preemptive scheduling by defining groups of tasks For tasks which have the same or lower priority as the highest priority within a group the tasks within the group behave like non preempta
8. 5 19 defining 5 2 grouping 7 15 idle 5 1 5 5 non standard attribute 5 15 scheduling 5 12 stack 5 15 state 5 3 terminating 5 8 Index U V unqueued message 10 3 10 10 virtual priority 5 5 initializing 10 11 virtual stack 5 17 updating makefile 2 7 run time 5 18 Index 3 8051 RTOS Index 4
9. The type of the LEVEL attribute is UINT32 and the possible values range from 0 to 31 LEVEL has no default value ENBIT The ENBIT attribute specifies the SFR register bit that enables disables the ISR The type of this attribute is STRING It has no default value The example below shows an OIL configuration example for a system with an external interrupt ISR isrExternal LEVEL ENBIT 0 EX0 hi The following situations are prohibited by the implementation e Define two ISRs objects with the same value for their LEVEL attribute The link phase shall fail since two routines become equal candidate for the same interrupt handler e Define two ISR objects with the same value for their ENBIT attribute This is a more dangerous situation since the program compiles links and builds e Define a name for the ENBIT attribute which is not defined in the sfr include file The compiler phase then already fails e Build your interrupt framework outside the RTOS scope You cannot use the function qualifier _interrupt to declare an interrupt since the RTOS would not have any control of it 9 3 8051 RTOS 9 3 Defining an Interrupt in the C Source To define an interrupt service routine you must use the macro ISR in your application software You must pass the name of the related ISR object in the OIL file as parameter to the macro ISR isrTimer code return The OSEK VDX implementation uses this macro to en
10. assembler linker locator absolute file debugger Figure 2 2 Altium RTOS integrated in DXP 2 4 Getting Started In the Edit part you make all your changes e Create and maintain a project and add a file user oil1 to it e Edit the source files in a project e Edit the user oil file e Set the options for each tool in the toolchain In the Build part you build your files e A makefile created by the Edit part is used to invoke the needed toolchain components resulting in an absolute object file The makefile rebuilds the RTOS library if the OIL file has changed In the Debug part you can debug your project e Call the TASKING debugger with the generated absolute object file The debugger uses a special ORTI file OQSEK Run time Interface to retrieve information This file is automatically generated by the TOC compiler This next sections will guide you step by step through the most important steps of building a simple RTOS application 2 5 8051 RTOS 2 3 Create a new Project Space for the MYRTOS Project Before you create your own RTOS application you need to create an embedded software project Create a new embedded software project 1 Start DXP DXP opens Look for the Pick a task section on your screen 2 From the Pick a task section select Embedded Software Development 3 Click on New Blank Embedded Software Project de Embedded Software Development Embedded Software Projects gt New Blank E
11. dd Please be aware that the costs of adding OIL objects are independent of whether your application accesses them or not with system services The measurements below have been obtained with a minimum application i e empty application labels and using only one system service StartoOs While obtaining the results below e All reference lists in OIL objects are empty bigger usage of both code and data area otherwise e VSTACK and SSTACK are both 10 and AUTOSTART is FALSE for TASK OIL objects e TASK objects have different PRIORITY values e EXTDATASIZE is 8 e CDATATYPE is char and QUEUESIZE is 5 for MESSAGE OIL objects e AUTOSTART is FALSE and ACTION is ACTIVATETASK for ALARM OIL objects e The CATEGORY value for the ISR objects is 2 Implementation Parameters RAM size ROM size RAM size ROM size OIL object bytes bytes bytes bytes EXTENDED EXTENDED STANDARD STANDARD MINIMUM 1 TASK 162 A44 162 9EB 15t TASK 5D 1C 5D 19 2st TASK 5D 1 5D 1 3st TASK 5D 1 5D 1 1St ISR 5 1DB 5 1D3 2St ISR 7 97 7 97 3st ISR 7 87 7 87 IS ALARM sys cntr 1 6B4 1 686 2st ALARM sys cntr 19 0 19 0 3st ALARM sys cnir 19 0 19 0 IS ALARM app cntr 17 0 17 0 2st ALARM app cntr 17 0 19 0 3st ALARM app cntr 17 0 19 0 1st EVENT 4 95 4 95 2st EVENT 10 0 10 0 3st EVENT 10 0 10 0 1st RESOURCE 9 64 9 64 2st RESOURCE A 0 A 0 3st RESOURCE A 0 A
12. never be removed or modified Table 2 1 Project files 2 2 Getting Started The next figure shows the relation between the files in an RTOS project and the development process OIL generator in DXP T Application hand coded OIL file oil mytypes h C source files c Tasking OIL Compiler toc gle Sees d e Se ar SG i Configurational files i g_conf c g_isrframe c g_conf h g_conf_types h flag h l RTOS Tiles bd der a der era C compiler C compiler S REN assembly files assembly files assembler assembler relocatable object files y ba archiver relocatable object files RTOS library linker script file linker lsl absolute object file y debugger y execution environment Figure 2 1 Development process 2 3 8051 RTOS 2 2 The Design Environment DARE Design Environment DXP is a Windows application that facilitates working with the tools in the toolchain and also offers project management and an integrated editor DXP has three main functions Edit Project management Build and Debug The figure below shows how these main functionalities relate to each other and how the RTOS system is integrated With DXP you can write compile assemble link locate and finally debug RTOS applications toolchain selection editor project management tool options makefile make compiler
13. you could over allocate memory to the stack of each task thus providing a safety margin However in embedded applications RAM is often the most precious resource and over allocating extra memory for all stacks could severely increase the final cost of the product AN Some optimizations can be achieved by sharing stack areas between tasks that can never preempt each other For instance basic tasks with the same priority level non preemptable basic tasks or tasks owning the same internal resource group of tasks With the help of a Stack Analyzer you could even predict beforehand the maximum size of the stack for a given task A Stack Analyzer inspects the call graph of a task and calculates the stack usage for a worst case scenario With these results you can set new sizes for the stacks in the OIL file and without any other change in the source code build a new image The next section presents mechanisms to detect stack overflows at run time B 2 Stack Overflow 2 Run time Stack Monitoring 2 1 IsStackinRange You can call the following service to detect possible stack overflows at run time StatusType IsStackInRange void This service compares current values of stack pointers with their absolute limits Checking the return value of this service helps you determine whether conditions of overflow exist at the moment of the call StatusType Description E OK No present conditions for stack overflow E_OS_SYS_VSTACK The V
14. 5 interrupts 12 6 messages 12 7 resources 12 4 system status 12 2 tasks 12 3 DXP build application 2 11 create a project space 2 6 E error handling 11 1 extended status 11 1 standard status 11 1 event 10 14 configuring 6 1 interface 6 6 using 6 2 F fatal error 11 2 file extensions 2 2 flag 10 14 functionality implementation parameters A 1 H hardware resources implementation parameters A 1 l implementation OIL file 2 1 2 2 implementation parameters A 1 hardware resources A 4 memory usage A 6 ROM usage A 5 internal resource 7 15 interrupt defining 9 4 enable disable all 9 10 interface 9 14 suspend resume all 9 12 suspend resume os interrupts 9 13 interrupt object 9 1 interrupt service routine 9 1 isr non standard attributes 9 3 isr object 9 1 L linked resource 7 10 makefile automatic creation of 2 7 updating 2 7 memory model 5 15 message 10 2 defining data type 10 7 interface 10 22 notification 10 14 queued 10 3 10 8 receiving 10 8 sending 10 6 unqueued 10 3 10 10 message notification 10 14 callback routine 10 14 event 10 14 flag 10 14 Index 1 8051 RTOS task 10 14 mutex 7 1 mutex lock 7 2 N non standard attributes 3 1 for the M16C 3 4 overview 3 2 O OIL system objects 3 1 OIL OSEK Implementation Language 2 1 OIL compiler 2 1 OIL file application part 3 7 3 8 implementation part 3 7 preprocessor comm
15. A separate queue is supported for each receiver and messages from these queues are consumed independently The OIL configuration below is erroneous since a second receiver cannot access the same queue receive message object TASK TaskD MESSAGE recMQTaskC hi TASK TaskC MESSAGE recMQTaskC hi Should our sender TaskA transmit the message to a new queue receiver TaskD a new queue receive message object must be added to the OIL file 10 9 8051 RTOS TASK TaskD MESSAGE recMQTaskD he MESSAGE recMQTaskD MESSAGEPROPERTY RECEIVE QUEUED INTERNAL SENDINGMESSAGE sendM QUEUESIZE 5 hi he Unqueued messages If the MESSAGEPROPERTY of msg is RECEIVE_UNQUEUVED_INTERNAL msq refers to a unqueue receive message object unqueued message Unqueued messages do not use the FIFO mechanism The application does not consume the message during reception of message data but a message can be read multiple times by an application once it has been received If no message has been received since the application started the application receives the message value set at initialization Unqueued messages are overwritten by new messages that arrive TASK TaskB myStruct st if E_OK ReceiveMessage recMU amp st an error occurred else message has been read processMsg amp st TerminateTask 10 10 Interprocess Communication Contrary to queue receive message obje
16. MESSAGE sendCommand hi MESSAGE sendCommand MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE myCommand he hi MESSAGE recCommand MESSAGEPROPERTY RECEIVE UNQUEUED INTERNAL SENDINGMESSAGE sendCommand he NOTIFICATION COMCALLBACK CALLBACKROUTINENAME myCallOut he hi 10 19 8051 RTOS 10 6 Starting and Ending the COM 10 6 1 Starting the COM The OSEK VDX COM standard provides you with the following service to start the communication component StatusType StartCOM COMApplicationModeType mode For internal communication this service performs little basically it sets some internal variables and if applicable it initializes all the unqueued receive message objects with the value of their standard attribute INITIALVALUE The StartCOM routine supports the possibility of starting the communication in different configurations with the parameter mode like the application modes and Startos You can define different modes with the multiple standard attribute COMAPPMODE in the COM OIL object The type of this multiple attribute is STRING The OSEK VDX implementation forces you to call the routine Startcom from task level Be carefull StartCOM must be called before any COM activity takes places in the system A good practice could be the use of an autostarting task performing some possible extra OS initialization plus calling Startcom This task becomes the only real autostartin
17. Priority Ceiling Protocol The principles of the ceiling priority protocol can be summarized as follows e At system generation the RTOS assigns to each resource a ceiling priority The ceiling priority is set at least to the highest priority of all tasks that access a resource or any of the resources linked to this resource The ceiling priority must be lower than the lowest priority of all tasks that do not access the resource and which have priorities higher than the highest priority of all tasks that access the resource e lf a task requires a resource and its current priority is lower than the ceiling priority of the resource the priority of the task will be raised to the ceiling priority of the resource e If the task releases the resource the priority of this task will be reset to its original rescheduling point The problem of priority inversion is eliminated since only one task is actually capable of locking a resource Refering to the example in the previous section 1 T1 gets the resource and the RTOS raises its priority to the ceiling priority of the resource R 2 T3 is activated and shall remain in the ready state at least while T1 locks resource R since its priority is never higher than the current priority of the system Remember that T1 can neither terminate nor wait for an event at this phase 3 T2 is also activated and remains in the ready state 7 11 8051 RTOS 4 T1 finally releases the resource The
18. RTOS to your project needs e be familiar with the most relevant RTOS concepts e know how to debug RTOS applications This manual assumes that you have already read the User s Manual of the toolchain documentation The manual leads you through the hottest topics of configuring and building RTOS applications overview of the functionality design hints debugging facilities and performance ed This manual expects you to have gone through the main topics of the online OSEK VDX standard documents These documents should be in fact a constant reference during the reading of this manual vii 8051 RTOS Short Table of Contents Chapter 1 Introduction to the RTOS Kernel Provides an introduction to the RTOS real time multitasking kernel It discusses the choice of making the RTOS compliant with the OSEK standard Additionally this chapter provides a high level introduction to real time concepts Chapter 2 Getting Started Overviews the files and their interrelations involved in every RTOS application and includes a self explanatory diagram of the development process as a whole Describes also how you can build your very first RTOS application guiding you step by step through the process Chapter 3 The OSEK VDX Implementation Language OIL Describes how you can configure your application with a file written in OIL Osek Implementation Language language which needs to be added to the project as a project member The chapt
19. USEGETSERVICEID must be set to TRUE In all cases below the standard attribute of the OS object USEPARAMETERACCESS must be set to TRUE OSError_GetResource_ResID Returns the value of parameter Res D of the failing system service GetResource OSError_ReleaseResource_ResID Returns the value of parameter Res D of the failing system service ReleaseResource OSError _StartOS_ Mode Returns the value of parameter Mode of the failing system service StartOS OSError_ActivateTask_TaskID Returns the value of parameter Task D of the failing system service Activate Task OSError_ChainTask_TaskID Returns the value of parameter TaskID of the failing system service ChainTask OSError_GetlaskState_TaskID Returns the value of parameter Task D of the failing system service GetTaskState OSError_GetTaskState_State Returns the value of parameter State of the failing system service GetTaskState OSError_GetAlarmBase_AlarmlD Returns the value of parameter AlarmID of the failing system service GetAlarmBase OSError_GetAlarmBase_Info Returns the value of parameter nfo of the failing system service GetAlarmBase 8051 RTOS Macro Service Description OSError_SetRelAlarm_AlarmID Returns the value of parameter AlarmID of the failing system service SetRelAlarm OSError_SetRelAlarm_increment Returns the value of parameter increment of the failing system service SetRelAlarm Returns the value of parameter cycle
20. a task from the StartupHook routine is less preferable than declaring the task as AUTOSTART in your OIL file 5 8 Task management A task is activated by the RTOS code when e Analarm expires with its attribute ACTION set to ACTIVATETASK e Amessage has been sent with its attribute NOTIFICATION set to ACTIVATETASK e You configured a task to be activated during the RTOS startup You must set the attribute AUTOSTART to TRUE and indicate under which application mode s the task must autostart APPMODE AppModel TASK autoT AUTOSTART TRUE APPMODE AppModel hi And the RTOS needs to be started in your C source DeclareAppMode AppModel int main int argc StartOS AppModel1 return After activation the task is ready to execute from the first statement The RTOS does not support C like parameter passing when starting a task Those parameters should be passed by message communication or by global variables d See Chapter 10 Interprocess Communication Since this implementation is ECC2 compliant a task can be activated once or multiple times maximum number of requests in parallel is defined at system generation with the attribute ACTIVATE of the TASK object If the task is in suspended mode the system service moves the task into the ready state If the task is not in the suspended mode and maximum number of multiple requests has not been reached yet the request is queued by the RTOS for lat
21. application OIL file You configure a send message object by defining a MESSAGE OIL object with the value for its MESSAGEPROPERTY set to SEND_STATIC_INTERNAL In this case and assuming that the type of the transmitted data is the built in type integer MESSAGE sendM MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE int hi b Configure the senders of the messages The senders of the message M are those TASK and or ISR OIL objects that can use the system service SendMessage to send the message M to its receivers if any The OSEK VDX standard offers you the standard attribute MESSAGE a multiple reference of type MESSAGE_TYPE in the TASK ISR OIL objects to add messages to the list of messages owned by the TASK or ISR In order to define the TASK TaskA as a sender of message M you only need to add sendM to the message list of the TaskA object TASK TaskA MESSAGE sendM ho Identify the Receive Message Objects for every message M has two receive message objects Let us call them recMU and recMQTaskC 10 4 Interprocess Communication 6 Configure the Receive Message Objects in the application OIL file You configure an unqueued receive message object by defining a MESSAGE OIL object with its value for its MESSAGEPROPERTY set to RECEIVE_UNQUEVED_INTERNAL You configure a queued receive message object by defining a MESSAGE OIL object with its value for its MESSAGEPROPERTY set to RECEIVE_QUEUED_INTERNA
22. b maximal size VISRSTACK nested interrupts gt end of ISR virtual stack other external data gt begin of virtual stack of a task maximal size is VSTACK gt __os_RTOS_VTSTACK task calls system services gt end of virtual stack of a task task virtual stack other external data Figure 5 6 Virtual stack Task management 5 8 The C Interface for Tasks You can use the following data types constants and system services in your C sources to deal with task related issues Element C Interface Data Types TaskType TaskRefType TaskStateType TaskStateRefType Constants RUNNING WAITING READY SUSPENDED INVALID_TASK System Services DeclareTask ActivateTask TerminateTask ChainTask Schedule GetTaskID GetTaskState Table 5 2 C Interface for Tasks d gt Please refer to the OSEK VDX documentation for an extensive description 5 19 8051 RTOS 5 20 6 Events Summary This chapter explains how the RTOS may synchronize tasks via events and describes how you can declare EVENT objects in the OIL file in order to optimize your event configuration 6 1 Introduction The OSEK VDX provides you with events as a synchronization method between tasks They differentiate basic from extended tasks Basically every extended task has a private array of binary semaphores The array index corresponds with a bit position in an event mask Thus waiting on an event mask
23. be incremented to 1 Mutex A software entity that prevents multiple tasks from entering the critical section Acquiring mutexes guarantees serialized access to critical regions of code and protects the calling task from being preempted in favor of another task which also attempts to access the same critical section until the mutex is dropped Priority inversion A lower priority task preempts a higher priority task while it has acquired a lock See also section 7 3 1 Priority Inversion 8051 RTOS Deadlock The impossibility of task execution due to infinite waiting for mutually locked resources See also section 7 3 2 Deadlocks Since OSEK VDX OS is meant to operate in a critical environment like the automobile industry both priority inversion and deadlocks are unacceptable 7 2 What is a Resource The OSEK VDX standard defines the use of resources to coordinate concurrent access of several tasks with different priorities to shared resources During these processes the RTOS guarantees that two tasks or interrupt routines do not occupy the same resource at the same time A resource is the OSEK VDX version of what the literature commonly refers to as semaphores AN or mutexes In OSEK context resources are an abstract mechanism for protecting data accesses by multiple tasks or ISRs A resource request effectively locks the corresponding data against concurrent access by another task This is usually called a mutex l
24. development phase 11 2 Error Handling 11 2 1 Standard Versus Extended Status Most of the system services have a StatusType return type All services return E_OK when no error occurs However the number of possible return values for each system service depends on the error status mode that you configure in the application OIL file STANDARD or EXTENDED In both cases a return value from a system service not equal to E_OK means that an error has occurred It is recommended to select extended status while your are developing the system In this mode the system services perform extra routinary integrity checks like e Service calls from legal location many services are forbidden at interrupt level or at hook routines e Integrity of objects they must be defined in the OIL file e Validity of ranges passed values might have limited ranges like you cannot set an alarm to expire after zero cycles e Consistency in configuration a task must own a resource if it attempts to take it or own a message if it attempts to send receive it All these extra return codes can be tested only in extended error mode To run in extended mode you must set the attribute STATUS of the OS object in the OIL file to EXTENDED 8051 RTOS When you finished debugging and the application is ready to be released you should enable the standard mode Since these tests will not be included in the program you will benefit from smaller images and faster programs
25. eee eee 5 5 5 4 1 Virtual versus Physical Priorities assez 5 5 5 4 2 Fast Scheduling are gra betas dace awe been eed ewe ewe wade 5 8 5 5 Activating and Terminating a Task A 5 8 5 6 Scheduling a EE 5 12 5 6 1 Full preemptive TGk 5 12 5 6 2 Non preemptive EEA 5 12 5 6 3 Scheduling Policy u u 5 13 5 7 The Stack Of Task 22asi000 rn nt Woden nea eee 5 15 5 7 1 The Memory Model asse 5 15 5 7 2 The System Stack a z 5 15 5 7 2 1 The Run Time System ETSk 5 16 5 7 2 2 Saving the System PISO 5 17 5 7 3 INIS 5 17 5 7 3 1 The Run Time Virtual GSK 5 18 5 8 The C Interface for Tasks ausa 5 19 Events 6 1 6 1 Juri aue KA 6 1 6 2 Configuring EVENS art erte dr ze rd eize Ee deure vedere eiads 6 1 6 3 The Usage of Events aurres zrureze rrez erga ertzak 6 2 6 4 The C Interface for ALS 6 6 Resource Management 7 1 7 1 Key Salala 7 1 7 2 What is a EUELA 7 2 7 3 The Ceiling Priority Protocol a z 7 10 7 3 1 Priority Inversion assez 7 10 7 3 2 Deadlocks iia neck otek delord nied ed eed lawl hanna enw 7 11 7 3 3 Description of The Priority Ceiling Protocol 7 11 7 3 4 Calculating the Ceiling Priority A ra 7 12 7 4 Grouping EEAO 7 15 7 5 The Scheduler as a Special Resource usura 7 17 7 6 The C Interface for Resources a z 7 18 Table of Contents Alarms 8 1 8 1 Jura iatea 8 1 8 2 HODIZ Suteen bios SIA noe e taia Er wee peza 8 1 8 2 1 What is a SS 8 1 8 2 2 The RTOS System Counter u u 8 3 8 3 What is an Alarm AA 8 6 8
26. eee eee ees 1 4 1 5 Why Using the Altium RTOS 0 0 e eee ee ees 1 4 Getting Started 2 1 2 1 What is an RTOS Project 0 eee ee eee 2 1 2 2 The Design Environment DXP a z 2 4 2 3 Create a new Project Space for the MYRTOS Project 2 6 2 4 Edit the Application Files 20 0 cee eee eee eee 2 8 2 5 Build Your Application 0 eee ees 2 11 2 6 Debug Your Application assez 2 11 The OSEK VDX Implementation Language OIL 3 1 3 1 Why an OIL La UEA 3 1 3 2 What are the OIL System Objects A ra 3 1 3 2 1 Standard and Non Standard Attributes a z 3 1 3 2 2 Overview of System Objects and Attributes 3 2 3 2 3 Non Standard Attributes for the 8051 assa 3 4 3 3 The Structure of an OIL File a z 3 7 3 3 1 Implementation SA 3 7 3 3 2 Asla BSA 3 8 3 4 Preprocessor eela 3 14 The startup process 4 1 4 1 Jurala etaa 4 1 4 2 SAGE 4 2 4 3 The Main MOZOS 4 3 4 3 1 What are Application Aaa 4 3 4 3 2 Defining Application MOS 4 3 4 3 3 Changing Application Modes Application Reset 4 5 4 3 4 Non mutually exclusive application modes ss 4 7 4 4 MISS KR UEU ESU 4 7 4 5 The Shut down Process 000 c eee eee eee eens 4 9 8051 RTOS Task management 5 1 5 1 What is a EEA 5 1 5 2 Defining a Task in the eae 5 2 5 3 The States ofa ERA 5 3 5 3 1 BaSIC TASKS AAA 5 3 5 3 2 Extended Tasks icjccacss aces eee ee dre ROS eee eee ee Reo 5 4 5 4 The Priority of a Task 1
27. is the internal container where the data is stored on the receiving side The message object is available for an arbitrary number of receivers if the message is unqueued or for only one receiver when queued In the example there will be two receive message objects one for TaskB and ISRA which size is the size of the transmitted data and a second one for TaskC which size is the size of the transmitted data times the size of the queue Symbolic Name A symbolic name is the application identification fora message object send or receive A symbolic name identifies either a send message object or a received message object for an unqueued receive message or a received message object for a queued receive message In fact the symbolic name becomes an alias for the container 10 3 8051 RTOS 10 3 Configuring Messages For every message you must define one symbolic name on the sending side On the receiver side you must define one symbolic name for all the receivers that do not queue the message Or you must define one extra symbolic name for every receiver that does queue the message The phases of message configuration are shown below with help of the example in the previous section 1 Isolate all the messages in the system M has been identified as the only messages for the system Identify the Send Message Object for every message M has one Send Message Object let us call it sendM Configure the Send Message Objects in the
28. it encounters unresolved externals During the lifetime of the StartupHook routine all the system interrupts are disabled and you have only restricted access to system services The only available services are GetActiveApplicationMode and Shutdown0S You should use this hook routine to add initialization code which strongly relies on the current selected Application Mode void StartupHook void AppModeType mode GetActiveApplicationMode switch mode case NORMAL InitLinkBankDriver break case DOWNLOAD InitLinkServerDriver InitFlashDriver break default ShutdownOS E_OK break return 4 The RTOS enables all system interrupts 5 The RTOS executes the highest priority task ready to execute AN If you define the AUTOSTART attribute as FALSE for all TASK objects in the OIL file the system enters directly into an RTOS defined idle state The system then waits for external events 4 8 The startup process 4 5 The Shut down Process The OSEK VDX standard defines a service to shut down the operating system void ShutdownOS StatusType 1 You can directly request the Shutdown0Os routine In this case you must define your own set of shut down reasons error codes define E_APP_ERROR1 64 define E_APP ERROR2 65 define E_APP_ERROR192 255 0 63 are reserved for the RTOS code As soon as your application encounters error N the ShutdownOs routine must be called with E_APP_ERRORN as p
29. must belong to the event list of the multiple attribute EVENT of the task The following OIL configuration specifies that task myTask can only wait for myEvent1 and myEveni2 OIL file EVENT myEventl EVENT myEvent2 EVENT myEvent3 TASK myTask EVENT EVENT myEvent1l myEvent2 If task myTask attempts to wait on another event than myEvent1 or myEveni2 the system service WaitEvent fails C source file TASK myTask if 6 2 TASK waits on an allowed event E_OK WaitEvent myEvent1 LogError WAITEVENT TerminateTask j CPU has been given to other tasks From another task ISR the event myEvent1 has been set for this task Now myTask can resume execution TASK attempts to wait on a forbidden event Events 10 11 if E_OK WaitEvent myEvent3 myEvent is not owned by myTask LogError WAITEVENT TerminateTask TerminateTask return When the scheduler moves the task from the ready to the running state the task resumes execution at the following immediate instruction A task can wait for several events at a time WaitEvent myEventl myEvent2 The task does not undergo the transition if just one of the events has occurred In this case the service immediately returns and the task remains in the running state Only when all the events are cleared for the calling task the invoking
30. of the failing system service SetRelAlarm OSError_SetRelAlarm_cycle OSError_SetAbsAlarm_AlarmID Returns the value of parameter A armI D of the failing system service SetAbsAlarm OSError_SetAbsAlarm_start Returns the value of parameter start of the failing system service SetAbsAlarm OSError_SetAbsAlarm_cycle Returns the value of parameter cycle of the failing system service SetAbsAlarm OSError_CancelAlarm_AlarmID Returns the value of parameter AlarmID of the failing system service CancelAlarm Returns the value of parameter AlarmI D of the failing system service GetAlarm OSError_GetAlarm_AlarmID OSError_GetAlarm_Tick Returns the value of parameter Tick of the failing system service GetAlarm OSError_SetEvent_TaskID Returns the value of parameter Task D of the failing system service SetEvent OSError_SetEvent_Mask Returns the value of parameter Mask of the failing system service SetEvent OSError_GetEvent_TaskID Returns the value of parameter Task D of the failing system service GetEvent OSError_GetEvent_Event Returns the value of parameter Event of the failing system service GetEvent OSError_WaitEvent_Mask Returns the value of parameter Mask of the failing system service WaitEvent OSError_ClearEvent_Mask Returns the value of parameter Mask of the failing system service ClearEvent OSError_IncrementCounter_CounterlD Returns the value of parameter CounterID of the failing system se
31. should always call this service with COM_SHUTDOWN_IMMEDIATE as parameter StopCoM sets the system ready for a new call to StartCOM 10 21 8051 RTOS 10 7 The C Interface for Messages You can use the following data types and system services in your C sources to deal with message related issues Element C Interface Data Types SymbolicName ApplicationDataRef FlagValue COMApplicationModeType COMShutdownModeType COMServiceldType LengthRef CalloutReturnType Constants COM_SHUTDOWN_IMMEDIATE E_COM_ID E_COM_LENGTH E_COM UNI E_COM_NOMSG System Services StartCOM StopCOM GetCOMApplicationMode InitMessage SendMessage ReceiveMessage GetMessageStatus ResetFlag_Flag ReadFlag_Flag COMErrorGetServiceld Table 10 1 The C Interface for Messages d gt Please refer to the OSEK VDX documentation for an extensive description 10 22 11 Error Handling Summary This chapter helps you to understand the available debug facilities and error checking posibilities Describes which services and mechanisms are available to handle errors in the system and how you can interfere with them by means of customizing certain Hook Routines 11 1 Introduction This chapter helps you to understand the available debug facilities and error checking posibilities of an OSEK VDX system All these techniques should constitute the debug module which is included in the application during the
32. the OS OIL object must have been set to TRUE OS StdOSs SHUTDOWNHOOK TRUE STACKMONITOR TRUE The ShutdownHook routine could look like follows void ShutdownHook StatusType Error switch Error case E_OK break case E_OS SYS VSTACK case E_OS SYS SSTACK case EOS SYS STACK case E_OS SYS _ISRSTACK Log STACK OVERFLOW n StackHandler Error default break return Stack Overflow And the a StackHandler routine could look like follows void StackHandler StatusType Error TaskType task GetTaskID amp task switch Error case E_OS_SYS_STACK Log T d Err S task break case E_OS SYS _ISRSTACK Log T d Err ISR task break case E_OS SYS VSTACK Log T d Err VT task break case E_OS_SYS_SSTACK Log T d Err ST task break default break Freeze everything The application needs to be rebuilt with right stack configuration DisableAllInterrupts while 1 return 8051 RTOS Index Index A alarm 8 1 declaring 8 10 expiring 8 1 interface 8 11 application OIL file 2 2 application oil file 2 1 applicatoin modes 4 3 changing 4 5 defining 4 3 B boot system 4 2 E callback routine 8 7 10 14 CCCB 10 1 ceiling priority calculating 7 12 ceiling priority protocol 7 10 conformonce class 10 1 counter 8 1 creating a makefile 2 7 critical code section 7 1 D deadlock 7 2 7 11 debug alarms 12
33. the TASK object in the OIL file to FALSE or TRUE Based on later events the basic task can flow through any of the four transitions 5 3 8051 RTOS Below is the OIL configuration for a basic non autostarting task TASK NotActTask PRIORITY 5 SCHEDULE FULL ACTIVATION 1 AUTOSTART FALSE he The OIL configuration for an auto starting task is TASK InitTask PRIORITY 9 SCHEDULE FULL ACTIVATION 1 AUTOSTART TRUE APPMODE NORMAL he 5 3 2 Extended Tasks Extended tasks have one additional state waiting In this state the extended task waits for an event to occur to resume execution While in this state a task cannot be scheduled since it is not ready to run Extended tasks can undergo two more transitions additional to the transitions of basic states e wait from running to waiting e trigger from waiting to ready Because extended tasks are intrinsically related to events it is very easy to define in the OIL file whether a task is basic or extended If the optional attribute EVENT exists for the TASK object the task will be an extended task Otherwise it is a basic task Below is the OIL configuration for an extended auto starting task EVENT MonitorReadEvent TASK Monitor PRIORITY 7 SCHEDULE FULL ACTIVATION 1 AUTOSTART TRUE APPMODE DOWNLOAD EVENT MonitorReadEvent hi 5 4 Task management 5 4 The Priority of a Task The scheduler deci
34. the alarm expires In that case you must specify which event is set and to which task ALARM sensorAE COUNTER sensorC ACTION SETEVENT TASK sensorT EVENT sensorE hi he When the alarm reaches the preset value the RTOS code sets event sensorE to task sensorT lf task sensorT waits for such an event normal case this becomes a point of rescheduling 4 You can configure an alarm to activate a certain task when it expires ALARM sensorAA COUNTER sensorC ACTION ACTIVATETASK TASK sensorT he b 8 6 Alarms When the alarm reaches the preset value the RTOS code will try to activate task sensorT If task sensorT is in the suspended state this becomes a point of rescheduling 5 You can configure an alarm to run a callback routine when it expires ALARM sensorAC COUNTER sensorC ACTION ALARMCALLBACK CALLBACK myCallBack b hi A callback must be defined in the application source like ALARMCALLBACK myCallBack application processing The processing level of a callback would be an Interrupt Service Routine Besides the OSEK VDX standard forbids preemption during the routine execution by means of suspending Category 2 interrupts Therefore this implementation expects very short code for the callback routines 6 You can use the system service SetRelAlarm to predefine a counter value for an alarm to expire relative to the actual counter value StatusT
35. the application software e Modifications and even additions of completely new tasks can be made in the application software without affecting critical system response requirements e Besides managing task execution most real time operating systems also provide facilities that include task communication task synchronization timers memory management etc e An RTOS hides the underlying hardware specific concerns to the user offering a runtime environment that is completely independent of the target processor e Easy migration to other targets provided that the RTOS vendor offers support for these other processor families 1 1 8051 RTOS 1 2 What is OSEK VDX In May 1993 OSEK was founded as a joint project in the German automotive industry aiming at an industry standard for an open ended architecture for distributed control units in vehicles OSEK is an abbreviation for the German term Offene Systeme und deren Schnittstellen f r die Elektronik im Kraftfahrzeug Open Systems and the Corresponding Interfaces for Automotive Electronics Meanwhile in France PSA and Renault were developing a similar system called VDX or Vehicle Distributed eXecutive The two projects merged in 1994 and a year later OSEK VDX was presented Although the OSEK VDxX standards were originally developed for the automotive industry the resulting specifications describe a small real time OS ideal for most embedded systems that are statically d
36. the resource if any which is currently locked by the running task RUNNINGTASKPRIORITY Current priority of the task referred by RUNNINGTASK it can be different from its static priority due to the Ceiling Priority Protocol RUNNINGISR2 Specifies which Category 2 ISR is running if current code executes in a Category 2 ISR RUNNINGISR2 is set to NO_ISR in case current code does not execute in a Category 2 ISR2 LASTERROR Last error code detected At start up the error code is initialised with E OK only if running with extended mode SYSLASTERROR System service where LASTERROR did occur only if running with extended mode CURRENTAPPMODE Current application mode Table 12 1 Status Debug Properties 12 2 Debugging an RTOS Application 12 3 How to Debug Tasks The debugger can display relevant information about all the tasks in the system 1 From the OSEK ORTI menu select MYOSEK and Tasks A window pops up showing a list with all the tasks in the system Every task is described with a set of properties The debugger displays the values of these properties The task properties are described in the next table Task Debug Property Description PRIORITY Current priority of the task it can be different from the static task priority due to the Ceiling Priority Protocol STATE Current state of the task SUSPENDED READY RUNNING or WAITING CURRACT Number of current activatio
37. the same functional call level Even when the function foo is corrected concerning the LIFO order of resource occupation like void foo void ReleaseResource R1 GetResource R2 some code accessing resource R2 ReleaseResource R2 there still can be a problem because ReleaseResource R1 is called from another level than GetResource R1 Calling the system services from different call levels can cause problems 10 You should not define RESOURCE objects to protect critical code which can only be accessed by tasks with the same priority The reason is simple If we join 6 with the idea that VDX OSEK systems do not allow round robbin scheduling of tasks at the same priority level we can conclude that two or more tasks with same priority accessing critical code will not suffer concurrency problems AN This is the basic idea behind the concept of the ceiling priority protocol 7 8 Resource Management 11 Be careful using nested resources since the occupation must be performed in strict last in first out LIFO order the resources have to be released in the reversed order of their occupation order OIL file TASK myTask RESOURCE RESOURCE myRes1 myRes2 bo C source file TASK myTask GetResource myRes1 GetResource myRes2 ReleaseResource myRes2 ReleaseResource myRes1 The below code sequence is incorrect because function foo is not allowed to release resource R1 TA
38. the usage of global data Although this mechanism might be extremely effective in some cases it fails to satisfy the most general case The communication services offer you a robust and reliable way to exchange data The conformance class CCCB The main purpose of conformance classes is to ensure that applications that were built for a particular conformance class are portable across different OSEK VDX COM implementations and CPUs featuring the same level of conformance class The CCCB conformance class e Does not offer support for external communication e Supports both unqueued queued messages 10 1 8051 RTOS e Supports SendMessage ReceiveMessage routines e Incorporates Notification Mechanisms only Class 1 e Supports GetMessageStatus API Within this conformance class only the interaction layer is implemented and not the network and or data link layers This implementation allows you the transfer of data between tasks and or interrupt service routines within the same CPU 10 2 Basic Concepts This section presents some basic concepts and definitions One sender sends a message to one or more receivers This is the leading principle of the communication mechanism Throughout the whole OSEK VDX documentation source code examples etc you will find sentences like the task TaskA sends message M to tasks TaskB and TaskC and to the interrupt service routine ISRA In this case for the message M TaskA is the sende
39. to the same priority level Smaller context switch latency times MIN_PRIO_LEVEL Lowest priority level used by the user No TASK OIL object can be defined with lower priority MAX_PRIO_LEVEL Highest priority level used by the user No limits Virtual priority values have no limits MAX_NO_ACTIVATIONS Upper limit for the number of task activations MAX_NO_EVENTS Maximum number of events objects per system per task Limits the number of EVENT OIL objects that can be defined in the appl OIL file MAX_NO_COUNTER Maximum number of counter objects per system per task Limits the number of COUNTER OIL objects that can be defined in the application OIL file Implementation Parameters Parameter Description Implementation MAX_NO_ALARM Maximum number of alarm objects per system per task Limits the number of ALARM OIL objects that can be defined in the appl OIL file MAX_NO_APPMODE Maximum number of application modes Limits the number of APPMODE objects that can be defined in the appl OIL file MAX_NO_RESOURCE Maximum number of resource objects per system per task Limits the number of RESOURCE OIL objects that can be defined in the application OIL file MAX_QUEUE_ SIZE Maximum size for the queues in QUEUE MESSAGE objects MAX_DATA_LENGTH Maximum length of the data ina message in bytes
40. way of data exchange between tasks and or interrupt service routines and how you can declare MESSAGE and COM objects in the application OIL file Chapter 11 Error Handling Helps you to understand the available debug facilities and error checking posibilities Describes which services and mechanisms are available to handle errors in the system and how you can interfere with them by means of customizing certain Hook Routines Chapter 12 Debugging an RTOS Application Explains how you can easily debug RTOS information with the Cross View Debugger and describes in detail all the information that you can obtain Appendix A Implementation Parameters The implementation parameters provide detailed information concerning the functionality performance and memory demand From the implementation parameters you can obtain valuable information about the impact of the RTOS on your application Appendix B Stack Overflow Describes how you can avoid problems caused by stack overflow 8051 RTOS 1 Introduction to the RTOS kernel Summary This chapter provides an introduction to the RTOS real time multitasking kernel It discusses the choice of making the RTOS compliant with the OSEK standard Additionally this chapter provides a high level introduction to real time concepts 1 1 Real time Applications A real time system is used when there are rigid time requirements on the operations of a processor to perform certain tasks There
41. 0 15t MESSAGE 1 0 1 0 2st MESSAGE 1E 0 1E 0 3st MESSAGE 1E 0 1E 0 System counter based ALARMS and application counter based ALARMS share a lot of code Therefor if an system counter based ALARM is already present it takes only little extra code when an application based ALARM is added The values for application counter based ALARMS are only valid if an system counter based ALARM is already present 3 3 Miscellaneous Timer units reserved for the OS None the operating system Interrupts traps and other hardware resources occupied by None 8051 RTOS B Stack Overflow Summary This appendix describes how you can avoid problems caused by stack overflow 1 Introduction For many years microprocessors have been included on chip memory management units MMU that enable individual tasks to run in hardware protected address spaces But many commercial real time operating systems never enable the MMU even if such hardware is present in the system This is the case with OSEK VDX systems all tasks share the same memory space It is easy to understand how a single errant pointer in one task can easily bring down the entire system or at least cause it to behave unexpectedly Apart from errant pointers the most common way to suffer from data corruption in an OSEK VDX system is due to a stack overflow Stack overflow is defined as an error condition which results from attempting to push more items on
42. 4 The C Interface ARAETA 8 11 Interrupts 9 1 9 1 urei a ala 9 1 9 2 The ISR Object A ra 9 1 9 2 1 The ISR Non Standard Attributes u z 9 3 9 3 Defining an Interrupt in the C Source u u 9 4 9 4 The Category of an ISR ORIS 9 4 9 5 Nested SEE 9 6 9 6 ISRs and Resources 0 teeta 9 7 9 7 ISRs and Messages 00 cece eect e ee eee 9 8 9 8 Fast Disable Enable API Services u z 9 10 9 8 1 Disable Enable All Interrupts s ra 9 10 9 8 2 Suspend Resume All Interrupts s z 9 12 9 8 3 Suspend Resume OS Interrupts assa 9 13 9 9 The C Interface for Interrupts 0 cece eee eee 9 14 Interprocess Communication 10 1 10 1 Iual eau ala 10 1 10 2 BasSiG eela AA 10 2 10 3 Configuring LA 10 4 10 4 Message Transmission 0 eee e eee eee 10 6 10 4 1 Sending a LAE 10 6 10 4 2 How to Define the Data Type of a Message 10 7 10 4 3 Receiving a AA 10 8 10 4 4 Initializing Unqueued Messages cee eee eee eee 10 11 10 4 5 Long versus Short Messages u z 10 13 10 5 Message dauie ala 10 14 10 5 1 Notification Example Activate Sek 10 15 10 5 2 Notification Example Set ESTA 10 16 10 5 3 Notification Example ESO 10 17 10 5 4 Notification Example Callback u z 10 19 10 6 Starting and Ending the GON 10 20 10 6 1 Starting the COM a z 10 20 10 6 2 Starting the COM Extension usura 10 21 10 6 3 Stopping the Sa 10 21 10 7 The C Interface for Messages assa 10 22 8051 RTOS vi
43. 8051 RTOS GU0102 v1 2 January 20 2004 Software hardware documentation and related materials Copyright 2004 Altium Limited All rights reserved You are permitted to print this document provided that 1 the use of such is for personal use only and will not be copied or posted on any network computer or broadcast in any media and 2 no modifications of the document is made Unauthorized duplication in whole or part of this document by any means mechanical or electronic including translation into another language except for brief excerpts in published reviews is prohibited without the express written permission of Altium Limited Unauthorized duplication of this work may also be prohibited by local statute Violators may be subject to both criminal and civil penalties including fines and or imprisonment Altium CAMtastic Design Explorer DXP LiveDesign NanoBoard NanoTalk Nexar nVisage CircuitStudio P CAD Protel Situs TASKING and Topological Autorouting and their respective logos are trademarks or registered trademarks of Altium Limited or its subsidiaries All other registered or unregistered trademarks referenced herein are the property of their respective owners and no trademark rights to the same are claimed Table of Contents Table of Contents Introduction to the RTOS kernel 1 1 1 1 Real time Alea a EA 1 1 1 2 MADERAS BIAR LN 1 2 1 3 The OSEK VDX Documentation Assur 1 3 1 4 The Altium RTOS 0 0 cece
44. ASK T1 PRIORITY 3 TASK T2 PRIORITY 1 b TASK T3 PRIORITY 1 b System B TASK T1 PRIORITY 17 TASK T2 PRIORITY 9 TASK T3 PRIORITY 9 The equivalent ready to run arrays of such systems would then be 4 3 lt T1 2 System A 1 lt T2 lt T3 0 lt Idle Figure 5 2 Virtual ready to run array for System A 5 6 Task management 18 17 lt T1 16 11 15 10 EE System B 8 2 7 1 0 lt Idle Figure 5 3 Virtual ready to run array for System B There are no functional differences As soon as T1 undergoes the wait or terminate transition T2 is scheduled T2 can only be preempted by T1 T3 only runs after T2 undergoes a wait or terminate transition However it is easy to infer from the diagrams that system A has the better run time response of the system In system B for instance there are 15 useless priority levels defined in the system Besides these levels can never hold a ready task the scheduler also wastes CPU cycles in checking them And RAM area has been allocated for them In a hard real time system these unnecessary checks must be avoided Since all this information can be interpreted beforehand by the RTOS code all these configurations will end up in the same physical ready to run array 2 lt T1 1 lt T2 lt T3 0 lt Idle Figure 5 4 Virtual rea
45. EK VDxX standard provides you with the following ways to initialize unqueued messages 1 Assign a value to the INITIALVALUE subattribute in the MESSAGE OIL object MESSAGE sendM MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE int b b 10 11 8051 RTOS MESSAGE recMU MESSAGEPROPERTY RECEIVE _UNQUEUED_INTERNAL SENDINGMESSAGE sendM INITIALVALUE 3 b ho To guarantee that recMU cannot be received before its container has been initialized with the value 3 this initialization is performed in the StartCoM routine db See section 10 6 1 Starting the COM for more information regarding the com hook routine StartCOM However note that OIL only allows the specification of a limited range of unsigned integer initialization values This means that OIL can only be used to initialize messages that correspond to unsigned integer types within OIL s range of values Thus the OIL configuration below makes no sense MESSAGE sendM MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE myStruct b b MESSAGE recMU MESSAGEPROPERTY RECEIVE _UNQUEUED_INTERNAL SENDINGMESSAGE sendM INITIALVALUE 3 b 2 Letthe StartComM routine initialize the unqueued message with the default value zero Again this applies only when the messages correspond to unsigned integer types within OIL s range of values 3 You can always use the following system service to initialize messages that are too large o
46. Error Handling 11 1 Introduction 11 2 Error Handling 11 2 1 Standard Versus Extended Status 11 2 2 The ErrorHook Routine 11 2 3 11 3 Debug Routines 11 4 OIL Examples Debugging an RTOS Application 12 1 Introduction 12 2 How to Debug the System Status 12 3 How to Debug Tasks 12 4 How to Debug Resources 12 5 How to Debug Alarms 12 6 How to Debug ISRs 12 7 How to Debug Messages Implementation Parameters The COMErrorHook Routine a a 1 Jaurla elek GA 2 Functionality Implementation Parameters 3 Hardware Resource Implementation Parameters 3 1 The ROM Usage by System Services 3 2 The ROM RAM Usage of OIL Objects 3 3 Miscellaneous 5 Stack Overflow 1 Jauzia a ie ici acoegch crc reei ea in dnc ee dea tae Soe a aoe ant ad 2 Run time Stack Monitoring 0 0 cece eee 2 1 ESAS AAE EUA 2 2 Stack Monitor wise ie cee da kinei a A ai a a Index 11 1 11 1 11 1 11 1 11 2 11 6 11 8 11 9 12 1 12 1 12 2 12 3 12 4 12 5 12 6 12 7 A 1 Manual Purpose and Structure Manual Purpose This manual aims to provide you with the necessary information to build real time applications using the RTOS Real Time Operating System micro kernel delivered with the toolchain After reading the document you should e know how to build real time RTOS applications using understand the benefits of using the RTOS e be able to customize the
47. Handler2 SEND_STATIC_INTERNAL GORE cHandler3 RECEIVE UNQUEUED INTERNAL sendHandlerl 3 RECEIVE UNQUEUED INTERNAL sendHandler2 3 RECEIVE UNQUEUED INTERNAL sendHandler3 3 3 13 8051 RTOS COM Com COMERRORHOOK TRUE COMUSEGETSERVICEID TRUE COMUSEPARAMETERACCESS TRUE COMSTARTCOMEXTENSION TRUE COMSTATUS COMEXTENDED b CPU Sample application 3 4 Preprocessor Commands The OIL preprocessor allows the following preprocessor commands e include file or include lt file gt e ifdef A e ifndef A e else e endif d It is highly recommended to divide the functionality of the OIL file into separated files The syntax is the same as in ISO ANSI C The OIL preprocessor accepts C style comments and C rules apply 3 14 4 The startup process Summary This chapter explains what happens inside the system from application reset until the first application task is scheduled and describes how you can interfere with the start up process by customizing certain Hook Routines 4 1 Introduction This chapter details the various phases the system undergoes from CPU reset until the first application task is scheduled You can intervene in this process via the Hook Routines and the Application Modes d Please refer to the OSEK VDX OIL documentation for details about the Hook Routines and Application Modes The startup process includes the followin
48. L The subattribute QUEUESIZE defines the length of the receive queue In both cases the subattribute SENDINGMESSAGE defines which is the related Send Message Object In our example MESSAGE recMU MESSAGEPROPERTY RECEIVE UNQUEUED_ INTERNAL SENDINGMESSAGE sendM b hi MESSAGE recMQTaskC MESSAGEPROPERTY RECEIVE QUEUED INTERNAL SENDINGMESSAGE sendM QUEUESIZE 5 b hi 7 Configure the receivers of the messages The receivers of the message M are those TASK and or ISR OIL objects which can use the system service ReceiveMessage to read the transmitted data of message M The OSEK VDX standard offers you the standard attribute MESSAGE a multiple reference of type MESSAGE_TYPE in the TASK ISR OIL objects to add messages to the list of messages owned by the TASK or ISR To define the TASK TaskB and the ISR SRA as unqueud receivers for message M you need to add recMU to the message list of the TaskB and ISRA objects And to define the TASK TaskC as a queued receiver for message M you need to add recMQTaskC to the message list of the TaskC object TASK TaskB MESSAGE recMU 10 5 8051 RTOS TASK ISRA MESSAGE recMU b TASK TaskC MESSAGE recMQTaskC hi 10 4 Message Transmission 10 4 1 Sending a Message Sending a message requires the transfer of the application message data to all the receiving message objects This process is done automatically for internal communication Yo
49. LSE RESOURCE sem_out VSTACK 80 SSTACK 40 he COUNTER crsi MAXALLOWEDVALUE 500 TICKSPERBASE 1 MINCYCLE 3 he ALARM arsi COUNTER crsi ACTION ACTIVATETASK TASK trsi AUTOSTART TRUE CYCLETIME 200 APPMODE AUTOSTART ALARMTIME 200 he he ALARM aintervall COUNTER SYS_TIMER ACTION SETEVENT TASK bouncel EVENT intervaldelay hi AUTOSTART FALSE he 3 11 8051 RTOS ALARM ainterval2 COUNTER SYS_TIMER ACTION SETEVENT TASK bounce2 EVENT intervaldelay hi AUTOSTART FALSE he ALARM ainterval3 COUNTER SYS_TIMER ACTION SETEVENT TASK bounce3 EVENT intervaldelay hi AUTOSTART FALSE hi MESSAGE sendHandlerl MESSAGEPROPERTY SEND_STATIC_INTERNAL CDATATYPE int RECEIVER recHandler1 hi hi RESOURCE sem_printf RESOURCEPROPERTY STANDARD hi RESOURCE sem_out RESOURCEPROPERTY STANDARD hi 3 12 The OSEK VDX Implementation Language OIL MESSAGE sendHandler2 MESSAGEPROPERTY CDATATYPE i RECEIVER re he b MESSAGE sendHandler3 MESSAGEPROPERTY CDATATYPE i RECEIVER re he b MESSAGE recHandlerl MESSAGEPROPERTY SENDINGMESSAGE INITIALVALUE MESSAGE recHandler2 MESSAGEPROPERTY SENDINGMESSAGE INITIALVALUE b he MESSAGE recHandler3 MESSAGEPROPERTY SENDINGMESSAGE INITIALVALUE b SEND_STATIC_INTERNAL nt c
50. ORGETSERVICEID properties for the interaction layer COMUSEPARAMETERACCESS COMSTARTCOMEXTENSION COMAPPMODE COMSTATUS NM The network management subsystem Table 3 1 OIL objects and their standard and non standard attributes 3 3 8051 RTOS 3 2 3 Non Standard Attributes for the 8051 This section describes the non standard attributes which are specific for the 8051 db Please refer to the OSEK VDX OIL documentation for the semantics of all standard attributes OS object CORE The CORE attribute specifies the Processor Definition of the project The type of this attribute is ENUM and has one of the following values TSK51A TSK52A TSK52B The default value is TSK51A EXTDATASIZE The EXTDATASIZE attribute indicates the maximum size in bytes of the extended data section The extended data section resides in the internal data area and it is used for extra register allocation You define the maximum size of this area with the compiler option x default is 4 bytes Since this area is shared by all tasks the RTOS needs to save restore it during context switch it is part of the context of the task The type of this attribute us UINT32 and the default value is 4 bytes dd You need to update this attribute everytime you change the option x of the compiler LONGMSG The LONGMSG boolean attribute determines whether Category 2 ISRS are suspended during the copy of messages from the RTOS buffers to the application or vi
51. OS code contributes to the system stack depth with _os_RTOS_SISRI1STACK bytes for an interrupt service routine where no system services are used Category 1 and with _os_RTOS_SISR2STACK bytes for an interrupt where system services are called Category 2 See section 9 4 The Category of an ISR object in Chapter Interrupts to learn what Category 1 and Category 2 interrupts are A final contribution for the system stack comes from the processor context which is pushed on this stack before context switching The size of the context is given in bytes by os CONTEXTSIZE d See also Appendix A Implementation Parameters In the next figure you can see the worst case scenario of the usage of the system stack for a system where m Category 2 ISRs and n Category 1 ISRs can be nested Task management other internal data gt end of system stack size gt m _os_ RTOS_SISR2STACK system stack of n _os RTOS SISR1STACK nested interrupts CPU context size gt _os_CONTEXTSIZE maximal size is SSTACK gt pe RTOS STSTACK task calls system services gt begin of system stack Task system stack other internal data Figure 5 5 System stack use worst case You configure the maximum run time size for the system stack in the linker options of the project 5 7 2 2 Saving the System Stack The RTOS saves the system stack of the preempted task in a dedicated location in external memory The memory reserved to
52. OSTICKDURATIONINMSCS Time in ms between two consecutive A 10 ticks of the hardware system clock OSTICKDURATION Time in nano seconds between two consecutive ticks of the hardware 10 000 000 system clock OSMINCYCLE Absolute minimum log expiring time 1 in timer units OSTICKSPERBASE Number of clock ticks per timer unit 1 OSMAXALLOWEDVALUE This value determines the upper limit for A 65535 the timer value unit MAX_TIMEOUT Maximum timeout for an alarm based on the system counter 8051 RTOS 3 Hardware Resource Implementation Parameters This section tries to help you understand the impact of the RTOS on the total size of the application The contributions of the RTOS to the total size fall roughly into two categories e ROM usage by system services The main idea is that there is a direct relation between the ROM size and the number of different system services your application uses the more system services your application uses the larger the code area The implementation reserves one module object per system service in the RTOS library The linker will extract only the modules for the system services that are used in the application code All these modules contribute mainly to code area e ROM RAM usage by OIL objects Each added OIL object takes both data area external and code area The next subsections present some experimental results regarding memory usage While obtaining these results
53. RITY 4 TASK T2 SCHEDULE FULL PRIORITY 2 TASK nP1 SCHEDULE NONE PRIORITY 3 EVENT eT2 nP1 AUTOSTART TRUE APPMODE default TASK nP2 SCHEDULE FULL PRIORITY 4 b 5 13 8051 RTOS C source file Like all other OIL objects you need to declare the tasks before you can use them in your code DeclareTask T1 DeclareTask T2 DeclareTask nP1 DeclareTask nP2 TASK T1 TerminateTask TASK T2 SetEvent nP1 eT2 nP1 TerminateTask TASK nP2 TerminateTask TASK nP1 T1 T2 nP2 are activated but they cannot preempt the running task ActivateTask T1l ActivateTask T2 ActivateTask nP2 E eae E This call allows CPU scheduling to tasks with higher priority Schedule 1 Tl runs first and terminates 2 nP2 runs and terminates 3 nPl resumes execution lt An ISR activates T1 dE erez EU WaitEvent eT2 nP1 1 Tl runs next and terminates 2 Finally T2 runs It sets eT2 nP1 to trigger again nP1 JE 12 EA TerminateTask T2 terminates 5 14 Task management 5 7 The Stack of Task The memory usage becomes a crucial discussion point for small embedded applications Thus you need to know in great detail how the RTOS allocates memory for each of its tasks You also need to know as far as possible how you can customize the process t
54. RTOS revers its priority to its normal static level This is a point of rescheduling T3 starts running 5 T3 terminates and T2 starts running 6 T2 terminates and T1 resumes running The only drawback is that T2 is inhibited by a lower priority task T1 But this occurs only during the locking time which can be calculated and or minimized The latency time in this scenario for T2 is far less than for T3 in the previous case Deadlock is also easily eliminated because T2 cannot preempt T1 T1 must occupy and release LIFO R1 and R2 before T2 attempts to take R2 Ceiling Priority Protocol at Interrupt levels The extension of the ceiling priority protocol to interrupt levels is also simple in this implementation Suppose that a resource R is owned by the interrupt service routines ISR4 ISRy and tasks T4 Ty Let P be the maximum interrupt priority level of these ISRs When task Tj occupies R Tj behaves as a non preemptable task and all the ISR interrupts with priority P and lower are temporarily disabled all R owners included Thus while Tj owns the resource it can only be preempted by interrupts with a priority higher than P Since Tj runs as non preemptable even if high priority tasks are activated from an ISRo interrupt with higher priority than P they will not be scheduled until Tj releases resource R When the task Tj releases R the priority of this task is reset to its original rescheduling point Possible pending interr
55. S H define MYTYPES H define PAYLOADSIZE 20 typedef struct mystruct myStruct struct mystruct unsigned char header unsigned char payload PAYLOADSIZE unsigned char crc endif If you configure a MESSAGE OIL object like MESSAGE sendM MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE myStruct he hi but you fail to provide the type definition for myStruct in the mytypes h file the RTOS code will not compile At system generation time the attribute LENGTH of a MESSAGE OIL object stores the size in bytes of the indicated CDATATYPE attribute sizeof myStruct This is done automatically by the RTOS tools Thus since the LENGTH of the message is known by the RTOS a call to SendMessage copies LENGTH number of bytes into the receive objects starting at DataRef In our example include mytypes h DeclareMessage sendM 10 7 8051 RTOS TASK TaskA myStruct st prepare the message st header 0x12 for i 0 i lt PAYLOADSIZE i st payload i 0 st cre OxFF send message to receive objects SendMessage sendM amp st TerminateTask 10 4 3 Receiving a Message To receive a message use the following system service StatusType ReceiveMessage SymbolicName msg ApplicationDataRef DataRef msg a Receive Message Object in your file which MESSAGEPROPERTY attribute has either RECEIVE_UNQUEUED_INTERNAL or RECEIVE_QUEUED_INTERNAL as value msg must belon
56. S2 and 1 if S3 If we select S2 the OIL file would look as follows COUNTER sensorC MAXALLOWEDVALUE 100 TICKSPERBASE 2 MINCYCLE 5 i After processing the OIL file the OSEK VDX implementation defines the following constants for you OSMAXALLOWEDVALUE_sensorcC 100 OSTICKSPERBASE sensorC E 42 OSMINCYCLE_sensorC 2 5 If the counter demands hardware or software initialization you can use the StartUpHook routine to place the initialization code You are responsible for detecting the recurring events of your own counters and as a follow up notifying the RTOS You must inform the RTOS about the arrival of a new sensor tick with the system service StatusType IncrementCounter CounterType 8 2 Alarms In this example you must increment the counter in the handler of the IO hardware interrupt connected to the sensor ISR sensorHandler IncrementCounter sensorC Gb See section 9 3 Defining an Interrupt in the C Source in Chapter Interrupts Some other OSEK VDX implementations might offer you extra API services to set the counter internal value to reset it and so on This implementation exclusively adds the necessary IncrementCounter API to the standard 8 2 2 The RTOS System Counter The implementation always offers one counter that is derived from a hardware timer This counter is known as the system counter The unit of time is the system tick as the interval between two consec
57. SK incorrect GetResource R1 some code accessing resource R1 foo ReleaseResource R2 void foo GetResource R2 code accessing resource res_2 ReleaseResource R1 7 9 8051 RTOS 12 The RTOS forbids nested access to the same resource In the rare cases where you need nested access to the very same resource it is recommended to use a second resource with the same behaviour so called linked resources You can configure linked resources in your OIL file like RESOURCE myResS i RESOURCEPROPERTY STANDARD hi RESOURCE myResL RESOURCEPROPERTY LINKED LINKEDRESOURCE myResS hi hi 7 3 The Ceiling Priority Protocol The OSEK VDxX standard defines the ceiling priority protocol in order to eliminate priority inversion and deadlocks 7 3 1 Priority Inversion A typical problem of common synchronization mechanisms is priority inversion This means that a lower priority task delays the execution of higher priority task Let us assume three tasks T1 T2 and T3 with increasing priorities T1 has the lowest priority and T3 the highest T1 is running and the otthers are suspended All tasks are fully preemptable Let us assume that T1 and T3 share resource R Theoretically the longest time that T3 can be delayed should be the maximum time that T1 locks the resource R However let us see what happens in the following time sequence 1 T1 occupies the resource R T2 and T3 are
58. SK myTask1l RESOURCE myResource hi TASK myTask2 RESOURCE myResource hi ISR myIsr RESOURCE myResource hi The following source code makes sure that the counter update does not suffer from concurrency problems C source file TASK myTask1l GetResource myResource no_counter ReleaseResource myResource TASK myTask2 GetResource myResource no_counter ReleaseResource myResource ISR myIsr GetResource myResource no_counter ReleaseResource myResource 7 3 8051 RTOS 4 Try to avoid superfluous resource definitions A resource that is owned by only one task is useless It decreases the performance of the system because e Memory is allocated with the configuration data for a useless resource e Execution speed decreases because of useless system services around not even real critical code e Longer internal searches of the RTOS So a resource should be owned by at least two tasks by a task and an interrupt service routine or by two interrupt service routines 5 You should define only one RESOURCE object for all the critical sections accessed by the same occupiers Let us assume that a second counter so_counter needs to be updated globally by these three actors There is no need yet to change the OIL file TASK myTaskl GetResource myResource so_counter ReleaseResource myResource GetResource myResource no_cou
59. SP pointer is currently out of range task level You must increase the value of the non standard VSTACK attribute in the corresponding TASK OIL object E DS SYS STACK The SP pointer is currently out of range You must increase the value of the stack size in the linker locator options E_OS_SYS_ISRSTACK The VSP pointer is currently out of range interrupt level You must increase the value of the non standard VISRSTACK attribute of the OS OIL object AN You must be aware that a E_OK return value does not imply that no data corruption exists in the system A E_OK return value only means that the stack pointers are all in range at the moment of the call it tells you nothing about previous out of range situations 2 2 Stack Monitor You can request the RTOS to continuously monitor possible stack overflows Although expensive in run time performance the RTOS will inform you as soon as possible with the precise cause of the stack overflow A non standard attribute STACKMONITOR is offered in the OS OIL object to request these tests When this attribute is set to TRUE the RTOS code performs internal stack overflow tests after every task run thus in every context switch If reasons for stack overflow have been encountered during the last task run the RTOS code calls the service ShutdownOS The value of the call parameter will help you determining the precise cause of the overflow The possible values are listed in the ta
60. You must then set the attribute STATUS of the OS object in the OIL file to STANDARD Be aware that you cannot test on extended error codes while running in standard mode So far we have referred to OSEK VDX OS but in an equivalent manner the OSEK VDX COM defines also standard and extended error checking modes for the system services of the OSEK VDX COM module You can define this error checking mode for the COM routines by setting the COMSTATUS attribute of the COM object to COMSTANDARD or COMEXTENDED Fatal Errors So far we have been speaking about application errors the RTOS cannot perform the service request correctly but assumes the correctness of its internal data If the RTOS cannot assume the correctness of its internal data it will not return from the system services The RTOS will call the ShutdownHook instead provided that the standard attribute SHUTDOWNHOOK of the OS OIL object is TRUE These are Fatal Errors db See also section 4 5 The Shutdown Process 11 2 2 The ErrorHook Routine In both standard and extended modes when a system service returns a StatusType value not equal to E_OK the RTOS calls the ErrorHook routine provided that you set the ERRORHOOK attribute of the OS object to TRUE void ErrorHook StatusType dd If ERRORHOOK is set but you fail to define the ErrorHook routine in your code the linking phase will fail due to unresolved externals ErrorHook is called at the end of the system servi
61. activated T3 preempts T1 and inmediatly requests R As Ris occupied T3 enters the waiting state T1 waits till T2 terminates or enters the waiting state 2 3 4 5 The scheduler selects T2 as the running task 6 7 T1 runs again and releases R 8 T3 immediately preempts T1 and runs again 7 10 Resource Management Although T2 does not use resource R it is in fact delaying T3 during its lifetime So a task with high priority sharing a resource with a task with low priority can be delayed by tasks with intermediate priorities That should not happen 7 3 2 Deadlocks Deadlocks are even more serious when locking resources causes a conflict between two tasks Each task can lock a resource that the other task needs and neither of the two is allowed to complete Imagine what would happen in the following scenario 1 Task T1 occupies the resource R1 2 Task T2 preempts T1 3 Task T2 occupies resource R2 4 Task T2 attempts to occupy resource R1 and enters the waiting state 5 Task T1 resumes attempts to occupy resource R2 and enters the waiting state 6 This results in a deadlock T1 and T2 wait forever With a properly designed application you can avoid deadlocks but this requires strict programming techniques The only real safe method of eliminating deadlocks is inherent to the RTOS itself This RTOS offers the priority ceiling protocol to avoid priority inversion and deadlocks 7 3 3 Description of The
62. ands 3 14 restrictions 3 7 structure 3 7 ORTI 12 1 ORTI OSEK VDX Run time Interface 1 3 os interrupt suspend resume 9 13 OSEK Implementation Language OIL 1 3 2 1 OSEK Run Time Interface ORTI 12 1 OSEK VDX 1 2 application 1 2 communication system 1 3 implementation 1 2 operating system 1 2 OSEK VDX Run time Interface ORTI 1 3 P performance implementation parameters A 1 physical priority 5 5 priority physical 5 5 virtual 5 5 priority ceiling protocol 7 2 priority inversion 7 1 7 10 project add new files 2 7 project space create 2 6 Q queued message 10 3 10 8 R real time applications 1 1 hard real time system 1 1 soft real time system 1 1 Index 2 Real Time Operating System 1 1 receive message object 10 3 receiver 10 2 resource 7 2 C interface 7 18 internal 7 15 linked resource 7 10 scheduler 7 17 scheduler 7 17 RTOS 1 1 rtos initialization 4 7 run time system stack 5 16 run time virtual stack 5 18 S scheduling tasks full preemptable 5 12 non preemptable 5 12 policy 5 13 send message object 10 3 sender 10 2 shut down 4 9 stack system 5 15 virtual 5 17 stack of a task 5 15 standard attributes 3 1 startup code 4 2 system boot 4 2 system counter 8 3 system objects 3 1 system services 1 2 5 19 system stack 5 15 run time 5 16 saving 5 17 T task 5 1 10 14 activating 5 8 basic 5 3 extended 5 4 system services
63. arameter ShutdownOS E_APP_ERRORN 2 The Shutdownos routine can also be reached internally by the operating system in case the RTOS encounters an internal fatal error The possible RTOS error codes then are E_OS SYS ERROR In case of a fatal internal error E OS SYS _SSTACK _ Incase ofa stack overflows E OE SYS_VSTACK E OE ENE STACK E OE ENE _ISRSTACK AN The RTOS error codes are defined in types h In the body of Shutdownos the RTOS calls the hook routine ShutdownHook provided that you have assigned the value TRUE to the SHUTDOWNHOOK attribute of the OS object in the OIL file void ShutdownHook StatusType AN If the SHUTDOWNHOOK attribute in the OIL file is set but you do not define the ShutdownHook routine in your source the linking phase fails because it encounters unresolved externals During the lifetime of the ShutdownHook routine all system interrupts are disabled and the only available service is GetActiveApplicationMode 4 9 8051 RTOS The OSEK VDxX standard allows you to define any system behaviour in this routine including no return from the routine Typical actions would be e Use a logging mechanism to study the reason for shutdown you must always check both application and RTOS error codes e Just return if E_OK See also section 4 3 3 Changing Application Modes Application Reset e Ifasevere error is encountered you can force the system to shut down In case you decide to
64. are two flavors of real time systems A hard real time system must guarantee that critical tasks complete on time Processing must be done within the defined constraints or the system will fail A soft real time system is less restrictive a critical task gets priority over other tasks and retains that priority until a point of rescheduling In a soft real time system failure to produce the correct response at the correct time is also undesirable but not fatal In reality most applications consist of tasks with both hard and soft real time constraints If these tasks are single purposed they could be implemented as semi independent program segments Still the programmer needs to embed the processor allocation logic inside the application tasks Implementations of this kind typically take the form of a control loop that continually checks for tasks to execute Such techniques suffer from numerous problems and represent no solutions for regular applications Besides they complicate the maintenance and reusability of the software An RTOS Real Time Operating System is a dedicated operating system fully compromised to overcome the time constraints of a real time system An RTOS provides like any other operating system an environment in which a user can execute programs in a convenient and structured manner but at no risk of failing with the real time constraints The benefits of using an RTOS are e An RTOS eliminates the need for processor allocation in
65. at fit within minimum resources 8 to 512 kB of ROM and 1 to 32 kB of RAM 1 4 2 Getting Started Summary This chapter gives an overview of the files and their interrelations involved in an RTOS application and includes a self explanatory diagram of the development process as a whole It also guides you through the process of building your very first RTOS application 2 1 What is an RTOS Project This chapter first discusses in detail the files that are involved in an RTOS project In the remaining sections an example project is created The basic ideas of an RTOS project are listed below 1 An RTOS project is a normal DXP project where you must add a file written in the OSEK Implementation Language OIL to the project members This file has the extension oil and contains the specific details of the system configuration We refer to it as the application OIL file Gb See Chapter 3 OIL Language 2 One and only one of the project members can have the extension oil The application OIL file first includes a target specific implementation OIL file include lt osek osek oil gt which is delivered with the product followed by a user defined part with all details for the configuration of your RTOS application 3 By adding an OIL file to the list of project members an extra project dependency is created in the project makefile The makefile now contains the rules to generate a special RTOS library from the OIL file
66. ble below The default value of STACKMONITOR is TRUE less run time overhead This method leaves almost no chance to the system to either hang or crash since the corrupted data would normally belong to another task which has had still no chance to run LAN It is recommended to run every application at least once with STACKMONITOR set to TRUE during the debugging phase B 3 8051 RTOS StatusType Description E OK No conditions for stack overflow have been detected E OS_SYS_VSTACK Conditions for a stack overflow in the virtual stack buffer of the exiting task have been detected You must increase the value of the non standard VSTACK attribute in the corresponding TASK OIL object E OS_SYS_SSTACK Conditions for a stack overflow in the system stack buffer of the exiting task have been detected You must increase the value of the non standard SSTACK attribute in the corresponding TASK OIL object E DS GKE SIAK Conditions for a stack overflow in the system stack have been detected You must increase the value of the stack size in the linker locator options E OS_SYS_ISRSTACK Conditions for a stack overflow in the virtual stack of the interrupt frames have been detected You must increase the value of the non standard VISRSTACK attribute of the OS OIL object You can now study the body of the hook routine ShutdownHook to find the reason for the overflow Obviously the SHUTDOWNHOOK standard attribute of
67. ble tasks For tasks with a higher priority than the highest priority within the group tasks within the group behave like preemptable tasks How to group Tasks You can use an internal resource to define explicitely a group of tasks or equivalently a group of tasks is defined as the set of tasks which own the same internal resource RESOURCE resInt RESOURCEPROPERTY INTERNAL hi The RTOS automatically takes the internal resource when an owner task enters the running state not when the task is activated As a result the priority of the task is automatically changed to the ceiling priority of the resource At the points of rescheduling the internal resource is automatically released the priority of the task is set back to the original In the following configuration all the tasks are initially suspended An event triggers the task infirst to activate The idle task is preempted and the task infirst will start execution Because it owns the internal resource resint the task starts running with the ceiling priority of resint 3 instead of with its static priority 1 This is the starting point in next example Example OIL file TASK infirst PRIORITY RESOURCE 1 resInt TASK insecond PRIORITY 2 RESOURCE resInt TASK inthird PRIORITY 3 RESOURCE resInt TASK outfirst PRIORITY 2 TASK outsecond PRIORITY 4 7 15 8051 RTOS C source file TASK outfirst Term
68. can not use the system services TerminateTask ChainTask Schedule and or WaitEvent while a resource is being occupied The task will not leave the running state under that condition ISR myTask GetResource myRes no_counter This is forbidden even if myEvent is owned by myTask WaitEvent myEvent ReleaseResource myRes 7 The RTOS assures you that an interrupt service routine is only processed under the condition that all resources that might be needed by that interrupt service routine are released TASK myTask2 GetResource myResN myIsr and myOtherIsr disabled no_counter ReleaseResource myResN GetResource myResS myIsr disabled so_counter ReleaseResource myResS 8 Make sure that resources are not still ocuppied at task termination or interrupt completion since this scenario can lead the system to undefined behaviour You should always encapsulate the access of a resource by the calls GetResource and ReleaseResource Avoid code like 7 7 8051 RTOS GetResource R1 switch condition case CASE 1 do_somethingl ReleaseResource R1 break case CASE 2 WRONG no release here do_something2 break default do_something3 ReleaseResource R1 The resource can be locked forever rejecting all further attempts to access the resource 9 You should use the system services GetResource and ReleaseResource from
69. capsulate the implementation specific formatting of the routine definition isrTimer is the identity of the ISR related object and has srType type The C name of the function that correspond to the interrupt service routine is created by prepending the tag _os_u_ The function can then be viewed with the debugger using the mingled name _os_u_isrTimer The OSEK VDxX standard focusing on portability internally hides all the intrinsic details of the target Migration to another platform should have almost no impact on your application source code You remain unaware of how internally the interrupt framework is built i e how the RTOS dispatches the execution flow to_os_u_isrTimer when a timer interrupt is generated 9 4 The Category of an ISR Object OSEK VDX defines two types of interrupts by ISR category 1 and 2 Category 1 These ISRs cannot use system services The internal status of the RTOS before and after the interrupt always prevails After the ISR has finished processing continues exactly at the same instruction where the interrupt occurred ISRs of this category have the least overhead and since they can always run concurrently with the RTOS code they are hardly disabled They execute normally at high priorities and have small handlers An example could be a serial interrupt that provides printf functionality ISR isrSerial CATEGORY 1 LEVEL 4 ENBIT ES he 9 4 Interrupts Category 2 These ISRs can
70. cated in the indirect addressable internal RAM segment together with the IDATA segments and grows upwards Since every task must save its return addresses history at context switch the RTOS saves and restores the system stack to a dedicated area in external RAM You configure the size of this dedicated area per task with the non standard attribute of the TASK OIL object SSTACK Thus you need to estimate the highest possible number of nested calls for every task 5 15 8051 RTOS Since tasks normally use RTOS system services the contribution of the RTOS code to the system stack growth must be considered The value of the maximum penetration depth of the RTOS code in the system stack in bytes is defined as_os RTOS STSTACK bytes As a result the value of the attribute SSTACK of a task that uses system services must be always higher than _os_RTOS_STSTACK How much higher depends on the usage of the system stack by the code of that particular task db See also Appendix A Implementation Parameters amp The dynamic size of the system stack influences the context switch time To minimize this effect you must avoid excessive use of nested function calls in your code 5 7 2 1 The Run Time System Stack Since in this architecture an interrupt service routine also stores the return addresses in the same system stack you must consider this contribution to avoid run time system stack overflows thus corrupting other internal data The RT
71. ce and immediately before returning to the calling function The RTOS does not call ErrorHook if the failing system service is called from the ErrorHook itself recursive calls never occur Therefore you can only detect possible errors in OS system services in the ErrorHook itself by evaluating directly their return value Macro services inside ErrorHook Once inside the ErrorHook routine the RTOS provides you with mechanisms to access valuable information With these mechanisms you can check which system service has failed and what its parameters were You can use the macro services listed in the next table for this purpose Error Handling Macro Service Description OSErrorGetServiceld Provides the system service identifier where the error occurred The return value of the macro is a service identifier of type OSServicelIdType and its possible values are OSServicelD_GetResource OSServicelD_ReleaseResource OSServicelD_GetTaskID OSServicelD_StartOS OSServicelD_ActivateTask OSServicelD_TerminateTask OSServicelD_GetTaskState OSServicelD_ Schedule OSServicelD_GetActiveApplicationMode OSServicelD_GetSystemTime OSServicelD_GetAlarmBase OSServicelD_GetAlarm OSServicelD_SetRelAlarm OSServicelD_SetAbsAlarm OSServicelD_CancelAlarm OSServicelD_SetEvent OSServicelD_GetEvent OSServicelD_WaitEvent OSServicelD_ClearEvent OSServicelD_Shutdown0S OSServicelD_IncrementCounter The value of the standard attribute of the OS object
72. ck you must measure this contribution to avoid run time virtual stack overflows thus corrupting other external data areas The RTOS changes the value of the virtual stack pointer register upon entering an interrupt at first nesting level so that the automatics and parameters of the interrupt routines are pushed and popped in a dedicated area The value is restored upon leaving an interrupt at first nesting level You configure the maximum contribution in bytes to the run time virtual stack of the interrupts with the non standard attribute VISRSTACK of the OS object When you give values to this non standard attribute be aware that the RTOS code contributes to the system stack depth with _os_RTOS_VISR1STACK bytes for an interrupt service routine where no system services are used Category 1 and with os RTOS VISR2STACK bytes for an interrupt where system services are called Category 2 See section 9 4 The Category of an ISR object in Chapter Interrupts to learn what Category 1 and Category 2 interrupts are d See also Appendix A Implementation Parameters d The RTOS allocates then VISRSTACK bytes for the interrupt routines and VSTACK bytes for every task These areas are allocated statically at compile time If your application does not use ISR objects you would define the VISRSTACK attribute as zero The figure below shows the run time virtual stack other external data gt begin of ISR virtual stack virtual stack for
73. crementCounter 0x48 0x40 InitMessage 0x96 0x64 ReceiveMessage 0x1C3 0x176 ReleaseResource 0x132 Ox8C Schedule 0x1C0 Ox6C SendMessage 0x47C 0x3D1 SetAbsAlarm 0x23A 0x1AB SetEvent 0x102 0xD7 SetRelAlarm 0x159 Ox8E ShutdownOS 0x5 0x5 StartCOM 0xD2 OxA6 StartOS 0x14C8 0x1438 StopCOM DOA 0x12 TerminateTask OxCC 0x76 WaitEvent 0x12A OxCE 8051 RTOS The smallest possible application uses at least the StartOs system service 0x2b bytes in code area Each system service you add to your application increases code size with at least the value as given in the table For example given an application with the following system services ActivateTask TerminateTask SetEvent WaitEvent ClearEvent The contribution of the system services modules to the final code size of the application is approximately 14C8 A9 CC 102 12A A5 1438 92 76 d7 CE 6D 190E EXTENDED mode 1752 STANDARD mode d The OIL file used to obtain these results is very similar to the OIL file example in section 3 3 2 Application Part in Chapter O L Language 3 2 The ROM RAM Usage of OIL Objects Every time you add an OIL object to you application system this takes a certain amount of memory The table below helps you to estimate the memory costs of adding new OIL objects to the application The first row shows data for the most basic RTOS configuration just one TASK object the remaining rows give the increase in memory per added OIL objects
74. cts the addition of new receivers for an unqueued message is straightforward You only need to add the recMU message object to the MESSAGE list of the new receivers in the OIL file TASK TaskD MESSAGE recMU hi However a receiver can never own more than one symbolic name per message a maximum of one queue and or unqueue receive message objects per message per receiver The configuration below is therefore erroneous TASK TaskC MESSAGE recMQTaskC MESSAGE recMU hi 10 4 4 Initializing Unqueued Messages Since the OSEK VDX COM standard allows only the senders to send messages the initial value of messages theoretically could only be set by a TASK ISR sender of the message Consequently every attempt to receive a message before the application actually sent the message will contain indeterminated data For a queued message this is not a problem since a returned value of ReceiveMessage equal to E_COM_NOMSG indicates that no message has arrived yet to the queue You should consistently check the returned status when receiving queued messages For an unqueued message none of the possible returned values indicate this situation and the application probably continues with erroneous data LAN There is a workaround for it which uses Notification Mechanisms See section 10 5 3 Notification Example Flag A better practice is to initialize the unqueued receive message object before any receiver tries to read it The OS
75. d to the project as a project member The chapter ends with a working example of an OIL file 3 1 Why an OIL Language Purpose of the OIL language The OSEK VDX Implementation Language OIL language is used to configure the RTOS library An OIL configuration file contains the definition of the application The usage of OIL to configure OSEK VDX systems enhances the portability of RTOS applications among different target processors Hand coded or generated Depending on the OSEK VDX implementation you must either write the OIL file manually or you can use a graphical user interface which helps you create the OIL file 3 2 What are the OIL System Objects Every version of OIL language defines syntactically and semantically a set of O L system objects These objects are defined in the OSEK standard One of the system objects is CPU This serves as a container for all other objects Objects are defined by their attributes b Refer to http www osek vdx org mirror oil241 OG for detailed information 3 2 1 Standard and Non Standard Attributes Every OIL system object has attributes that can hold values According to the OIL standard each object has at least a minimum mandatory set of attributes called the standard attributes Besides the standard attributes an OSEK VDX implementation may define additional attributes non standard attributes for any OIL system object To configure a system for an specific OSEK VDX implementation yo
76. dards do not explicitly cover I O 1 2 Introduction to the RTOS kernel OSEK VDX communication COM The communication specification provides interfaces for the transfer of data within vehicle networks systems This communication takes place between and within network stations CPUs This specification defines an interaction layer and requirements to the underlying network layer and or data link layer The interaction layer provides the application programming interface API of OSEK VDX COM to support the transfer of messages within and between network stations For network communication the interaction layer uses services provided by the lower layers CPU internal communication is handled by the interaction layer only OSEK VDX Implementation Language OIL To reach the original goal of the OSEK VDxX project in having portable software a way of describing an OSEK VDxX system is defined This is the motivation for the definition of a standardised OSEK VDX Implementation Language abbreviated OIL OSEK VDX Run Time Interface ORTI To provide debugging support on the level of OSEK objects it is necessary to have debuggers that are capable of displaying and debugging OSEK components The ORTI specification provides an interface for debugging and monitoring tools to access OSEK objects in target memory Tools can evaluate internal data structures of OSEK objects and their location in memory ORTI consists of a language to describe kerne
77. des on the basis of the task priority precedence which is the next of the ready tasks to be transferred into the running state The value 0 is defined as the lowest priority of a task and it is reserved for the idle task To enhance efficiency a dynamic priority management is not supported Accordingly the priority of a task is defined statically you cannot change it during execution AN In special cases the operating system can treat tasks with a lower priority as tasks with a higher priority See Section 7 3 The Ceiling Priority Protocol in Chapter Resource Management Since this OSEK VDX implementation is ECC2 compliant more than one task with the same priority can execute The implementation uses a first in first out FIFO queue for each priority level containing all the ready tasks within that priority Some facts about the ready queues are listed below e Every ready queue corresponds to a priority level e Tasks are queued in activation order in the ready queue that corresponds to their static priority e All the tasks that are queued must be in the ready state e Since the waiting tasks are not in any ready queue they do not block the start of subsequent tasks with identical priority e The system priority corresponds to the highest priority among all the non empty ready queues e The running task is the first task in the ready queue with the system priority e A task being released from the waiting state is treated like the
78. dy to run array for System A and System B Internally the RTOS code deals always with physical priorities The maximum size of the ready to run array determines the upper limit for the number of physical priorities See Appendix A mplementation Parameters to find out the maximum possible number of physical priority levels in the system 5 7 8051 RTOS 5 4 2 Fast Scheduling Every physical priority level holds a ready queue You can define multiple tasks with the same priority However if you define only one task per priority level the scheduler becomes faster In this situation the RTOS software does not have to deal with ready queues but merely with pointers to the task control blocks Whenever possible you should try to define only one TASK OIL object with the same value for its PRIORITY standard attribute You will benefit not only from better run time responses but also from smaller RAM and ROM sizes 5 5 Activating and Terminating a Task Tasks must be properly activated and terminated You can activate tasks directly from a task or interrupt level or from the StartupHook routine To activate a task use the system service StatusType ActivateTask TaskType task Example DeclareTask myTask ISR myISR ActivateTask myTask TASK myOtherTask code ActivateTask myTask TerminateTask j void StartupHook void ActivateTask myTask d Although allowed activating
79. e The Application Part is slaved to the Implementation Part In object oriented terminology we would say that the Implementation Part contains the class definitions of all OIL objects for all projects Ina specific project the classes are instantiated in the Application Part of the OIL configuration file For example In the Implementation Part an OIL object TASK exists which defines PRIORITY as one of its attributes In the Application Part you must now instantiate classes of the kind TASK and give values to their PRIORITY attributes Restrictions e Atleast one CPU object must be defined in the Application Part and it must be defined first since it is the container object for all other objects defined in the configuration All the other objects are defined inside the CPU object e One and only one OS object can reside in each CPU container since it defines global characteristics for the system CPU speed RTOS hardware resources etc are typical OSEK VDX implementations usually add many non standard attributes in this object e One and only one COM object can reside in each CPU container With an OIL generator tool a friendly GUI interface will guide you in the process of configuring the Application Part of the OIL file and it will output a syntactically correct OIL configuration file for you Without such a tool you must hand code the Application Part taking care of its grammatical correctness of the OIL file 3 3 1 Imple
80. e STARTUPHOOK ERRORHOOK SHUTDOWNHOOK PRETASKHOOK POSTTASKHOOK USEGETSERVICEID USEPARAMETERACCESS USERESSCHEDULER STACKMONITOR COMERRORHOOK COMUSEGETSERVICEID COMUSEPARAMETERACCESS and COMSTARTCOMEXTENSION were set to FALSE e Default compiler options were used The tables distinguishes the following cases e EXTENDED The STATUS standard attribute of OS OIL object is set as EXTENDED and the COMSTATUS standard attribute of COM OIL object is set as COMEXTENDED e STANDARD The STATUS standard attribute of OS OIL object is set as STANDARD and the COMSTATUS standard attribute of COM OIL object is set as COMEXTENDED These results have been taken with a specific configuration and built with specific options They will differ for each application and configuration This table is provided to give inside in how the usage of system services and OIL objects affect the total size of your application Implementation Parameters 3 1 The ROM Usage by System Services System Service ROM size bytes EXTENDED ROM size bytes STANDARD ActivateTask OxA9 0x92 CancelAlarm OxC6 OxBF ChainTask 0x18A OxF2 ClearEvent OxA5 Ox6D GetAlarm 0x82 0x67 GetAlarmBase OxAE 0x8B GetApplicationMode 0x8 0x8 GetCOMApplicationMode 0x8 0x8 GetEvent 0x71 Ox4E GetMessageStatus Ox9E 0x79 GetResource 0x15B OxBC GetTaskID 0x52 Ox3E GetTaskState 0x5C Ox4A In
81. e below the task commandHandler checks every POLLMSCS the flag associated with the receive object recCommanad If the flag is set the task receives the message otherwise it enters again the waiting state for the next POLLMSCS You can check the status of the Flag with the following API FlagValue ReadFlag_FlgComm void If the returned value is COM_TRUE the Flag was set a new message has been received 10 17 8051 RTOS include flag h include mytypes h DeclareMessage recCommand TASK commandHandler myStruct st while 1 if COM_TRUE ReadFlag_FlgComm ReceiveMessage recCommand amp St processMsg amp st else SetRelAlarm alarm POLLMSCS OSTICKDURATIONINMSCS 0 WaitEvent event ClearEvent event TerminateTask return You must always include the file flag h when using Flag Notification Mechanisms Otherwise your application will not compile This solution is even less safe It assumes that the minimum average interarrival time between two consecutive commands is at least greater than POLLMSCS plus the code execution overhead in the while loop 10 18 Interprocess Communication 10 5 4 Notification Example Callback A COM Callback must be defined in your application source as follows COMCallOut myCallOut return COM TRUE The return type for a COMCallout routine is CalloutReturnType In the previous example ISR SerialRx
82. efined i e with no dynamic run time allocation of memory The OSEK VDX specification consists of five normative documents e OS operating system e COM communication e NM network monitoring not discussed in this manual e OIL osek implementation language e ORTI osek vdx real time interface An OSEK VDX implementation refers to a particular implementation of one or more of the standards These standards tend to define the minimum requirements for a compliant system but individual implementations can vary because of different processor requirements and or capabilities An OSEK VDX application refers to an application that was developed using a particular OSEK VDX implementation OSEK VDX operating system OS The specification of the OSEK VDX OS covers a pool of services and processing mechanisms The operating system controls the real time execution in concurrent executing applications and provides you with a dedicated programming environment The architecture of the OSEK VDX OS distinguishes three processing levels an interrupt level a logical level for operating system activities and a task level The interrupt level is assigned higher priorities than the task level In addition to the management of the processing levels the operating system offers also system services to manage tasks events resources counters alarms and to handle errors You can consider system services as library functions in C d The OSEK VDX stan
83. efore the interrupt occurred Defining an ISR object in the OIL file Should your application make use for example of an external interrupt you define one ISR OIL object in your OIL file ISR isrExternal he The RTOS software continously needs to enable disable particular interrupts to avoid fatal preemptions Each of the 8051 interrupts has its own enable disable bit in a Special Function Register Check beforehand the manual of your core and or the sfr files in the PRODDIR c51 include directory The RTOS enables the interrupt by setting the corresponding bit db You provide the RTOS with this information as a non standard attribute of the ISR OIL object See also the next section 9 2 1 The ISR Non Standard Attributes Depending on the used core different levels of interrupt priority levels are offered The RTOS does not modify the priority level of any interrupt at run time except from possible extensions of the Ceiling Priority Protocol to interrupt level You normally place the initialization code for the ISR objects in the StartupHook routine where they are also normally enabled However in specific applications you might want to do this if at all later 9 2 Interrupts 9 2 1 The ISR Non Standard Attributes This implementation defines two non standard attributes for the ISR OIL objects LEVEL ENBIT They help the RTOS to configure the interrupt LEVEL Determines the entry in the vector table
84. er ends with a working example of an OIL file Chapter 4 The Startup Process Opens the black box of what happens in the system since application reset until the first application task is scheduled and describes how you can interfere with the start up process by customizing certain Hook Routines Chapter 5 Task Management Explains how the RTOS manages tasks scheduling policies tasks states and describes how you can declare TASK objects in the OIL file in order to optimize your task configuration Chapter 6 Events Explains how the RTOS may synchronize tasks via events and describes how you can declare EVENT objects in the OIL file in order to optimize your event configuration Chapter 7 Resource Management Explains how the RTOS performs resource management resource occupation ceiling priority protocol internal resources and describes how you can declare RESOURCE objects in the OIL file in order to optimize your resource configuration Chapter 8 Alarms Describes how the RTOS offers alarms mechanisms based on counting specific recurring events and describes how you can declare these objects in the OIL file in order to optimize your alarm configuration viii Chapter 9 Interrupts Describes how you can declare ISR objects in the application OIL file in order to optimize the interrupt configuration Chapter 10 Interprocess Communication Describes why the communication services offer you a robust and reliable
85. er processing Example of Activating a Task OIL file TASK Init PRIORITY 2 TASK activate PRIORITY ACTIVATION 1 b 5 9 8051 RTOS C source file TASK activate TerminateTask TASK Init activate task is set to ready ActivateTask activate log new requests for i 1 i lt 5 itt if E_OK ActivateTask activate never here always E OE while 1 for i1 0 i lt 5 itt if E_OS LIMIT ActivateTask activate never here always E_OS LIMIT while 1 TerminateTask Now activate task will run five times return Terminating a task You must explicitly terminate a task with one of the system services void TerminateTask void or void ChainTask TaskType The OSEK VDX standard has an undefined behaviour if the return instruction is encountered at task level Task management Situations like demonstrated in the example should be avoided TASK myTask unsigned char var readPulse switch var case READY sendSignal TerminateTask break case NONREADY TerminateTask break default break return Although apparently innocuous the behaviour of the whole system is completely undefined if var does not equal to READY or NONREADY In that case the switch reaches default where the function is not properly terminated Be aware that calling TerminateTask from in
86. er the alarm has already expired or not With the CancelAlarm routine you actually cancel a running alarm The next example shows how to set a timeout associated with an action TASK sensorT TickType tick start alarm setRelAlarm sensorAE 90 0 indicate I am waiting Action WaitAction wait for the event It is set when the action has completed or by the alarm WaitEvent sensorE the event has been set GetAlarm sensoraAE amp tick if tick Action was completed CancelAlarm sensorAE else Timeout TerminateTask 10 You can use the system counter as a generator for software timers A software timer basically guarantees the application that a certain action will occur after a certain period If 1 the action is short enough to be the body of a callback function and 2 the time period in which the action must occur is 500 milisecs and 3 assuming that an alarm SoftwareTimer has been already declared the next system call will take care of executing the callback code in exactly 500 miliseconds SetRelAlarm SoftwareTimer 500 OSTICKDURATIONINMSCS 0 8 9 8051 RTOS 11 You can use SetRelAlarm WaitEvent and ClearEvent to workaround an OSEK VDX version of an standard delay service OIL file EVENT delayE TASK myTask i EVENT delayE hi ALARM delayA COUNTER SYSTEM TIMER ACTION SETEVENT TASK myTask EVENT delayE h
87. ese objects in the OIL file in order to optimize your alarm configuration 8 1 Introduction The RTOS provides services for processing recurring events A recurring event is an abstract entity which has been defined in the scope of a particular application Each application monitors its events therefore differently A wheel has rotated five more degrees or routine has been called could be examples of recurring events Each of these recurring events must be registered in a dedicated counter a COUNTER OIL object You must increment this counter each time the associated event occurs In the wheel example you probably increment the counter in an interrupt service routine In the routine example you increment the counter in the body of f Based on these counters you can install alarms Activation of a specific task every time the wheel rotates 180 degrees or setting an event when routine f has been called ten times are examples of such alarms We say that an alarm expires when the counter reaches a preset value You can bind specific actions to alarms and the RTOS execute these actions upon expiration of the alarms 8 2 Counters 8 2 1 What is a Counter A counter is an abstract entity directly associated with a recurring event The counter registers happenings of its event ticks in a counter value which must be incremented every time the event takes place The OSEK VDX standard does not provide you
88. et watchdog e Addition of some power on tests e Call the label main to start the application 4 2 The startup process 4 3 The Main Module At the moment of entering main only minimal controller initialization has occurred At this point the application can run extra application specific initialization routines before the RTOS starts This code cannot call an RTOS system service The OSEK VDX standard defines a service to start the operating system void StartOS AppModeType 4 3 1 What are Application Modes Application Modes allow you as a matter of speaking to have multiple applications in one single image Application Modes allow application images to structure the software running in the processor depending on external conditions These conditions must be tested by the application software upon system reset The difference in functionality between applications that start in different modes is determined by e Which tasks and which alarms automatically start after the RTOS initialization e Mode specific code the mode can be detected at run time by using the system service GetActiveApplicationMode You define and set all these dependencies in the OIL file of the project The OSEK VDX does not set a limit for the number of Application Mode objects See Appendix A Implementation Paramaters for the maximum number of application modes in this implementation 4 3 2 Defining Application Modes T
89. ether a new message has arrived or whether it is still the old one There must exist ways to synchronize senders and receivers The OSEK VDxX standard defines standard notification mechanisms as a follow up of the transmition and or reception of a message For internal communication only Notification Class 1 is supported which means that as soon as the message has been stored in the receive message object a notification mechanism is invoked automatically aN Check the OSEK VDX COM documentation for more information about other Notification Classes They do not apply however to internal communication since internal transmission is always performed before returning from the SendMessage system service and without loss of data The following notification mechanisms are provided Callback routine A callback routine provided by the application is called Flag A flag is set The application can check the flag with the ReadFlag API service Resetting the flag is performed by the application with the ResetFlag API service Additionally calls to ReceiveMessage reset also the flag Task An application task is activated Event An event for an application task is set The notification mechanism can be defined only for a receiver message object With these mechanisms you can synchronize the copy of data into the receive message object with the receiver Since Notifications occur before returning from SendMessage this system service become
90. ever allowed Avoid situations like below static void f void DisableAllInterrupts otherThings EnableAlliInterrupts return 9 11 8051 RTOS TASK myTask DisableAllInterrupts someThings EnableAlliInterrupts return As a rule of thumb you should try to avoid function calls while in the critical section 9 8 2 Suspend Resume All Interrupts You can use the following system services to suspend resume all maskable interrupts void SuspendAllInterrupts void void ResumeAllInterrupts void They enhance the previous pair Disable EnableAllInterrupts in order to allow nesting In case of nesting pairs of calls the interrupt recognition status saved by the first call of SuspendAlliInterrupts is restored by the last call of the ResumeAllInterrupts service static void f void nothing happens with the status SuspendAllInterrupts otherThings status is not yet restored ResumeAllInterrupts return TASK myTask Ga status is saved now SuspendAllInterrupts someThings GARE status is restored now ResumeAllInterrupts return d The considerations for the pair DisableAllInterrupts EnableAllInterrupts apply here too 9 12 Interrupts 9 8 3 Suspend Resume OS Interrupts The previous pairs disabled all maskable interrupts including your Category 1 ISRs while in the critical code section However theoreticall
91. external or internal event When an interrupt occurs the processor suspends the current path of execution and transfers control to the appropriate Interrupt Service Routine ISR The exact operation of an interrupt is so inherently processor specific that they may constitute a major bottleneck when porting applications to other targets In most of the embedded applications the interrupts constitute a critical interface with external events where the response time often is crucial 9 2 The ISR Object Among the typical events that cause interrupts are e Overflow of hardware timers e Reception Transmission of Serial Character e External Events e Reset E Please check the documentation of your core to find out the interrupt sources of the core you are using Every interrupt has to be associated with a piece of code called Interrupt Service Routine or ISR The architecture defines a specific code address for each ISR which is stored in the nterrupt vector Table IVT 9 1 8051 RTOS The interrupt execution steps are listed below 1 CPU finishes the instruction it is currently executing and stores the PC on the system stack 2 CPU saves the current status of all interrupts internally 3 Fetches the ISR address for the interrupt from IVT and jumps to that address 4 Executes the ISR until it reaches the RETI instruction 5 Upon RETI the CPU pops back the old PC from the stack and continues with whatever it was doing b
92. forms the following actions during the initialization process 1 The RTOS initializes some internal data structures on the basis of what is stated in the OIL file In particular it prepares autostarting tasks and alarms to start running 2 The RTOS hardware timer is initialized Not all applications need a system counter only those with ALARM OIL objects based on the system counter You determine this with the non standard attribute USERTOSTIMER When USERTOSTIMER is set to TRUE e The RTOS builds the framework for the system counter interrupt e Due to the amount of derivatives systems it must still be you who actually initializes the hardware clock In fact you must provide a definition for the Interface void InitRTOSTimer void If the USERTOSTIMER attribute in the OIL file is set but you do not define the InitRTOSTimer routine in your source the linking phase fails because it encounters unresolved externals db See section 8 4 Initialization of the RTOS Timer E You can always set alarms to be based on the SYSTEM_TIMER counter which has this hardware timer as a source 3 The RTOS calls the hook routine StartupHook provided that you have assigned the value TRUE to the STARTUPHOOK attribute of the OS object in the OIL file 4 7 8051 RTOS void StartupHook void dd If the STARTUPHOOK attribute in the OIL file is set but you do not define the StartupHook routine in your source the linking phase fails because
93. g phases 1 System boot 2 C entry point main 3 StartOS 4 RTOS initialization phase Hook Routines After the startup process the first task is scheduled 4 1 8051 RTOS 4 2 System Boot When the processor first starts up it always looks at the same place in the system ROM memory area for the start of the system boot program The boot code runs sequentially until it reaches the point where it jumps to the label main This code runs of course in the absence of the operating system In general embedded systems differ so much from each other that the boot code tends to be almost unique for each system You can create the system boot in two ways 1 Reuse the standard startup code provided by the toolchain and enhance it if necessary to suit the specific needs of your hardware The standard startup code merely contains what is necessary to get the system running It is easy configurable via Project Options in the Project menu 2 You may decide to create the system boot code if some board specific actions need to be taken ata very early stage Some of the most common actions are e Initialization of critical microprocessor registers including standard pointers like the stack pointer e Initialization of specific peripheral registers for your unique hardware design e Initialization of all static variables using the delivered init routine e Distinguish the source of processor reset hard or soft reset power on res
94. g task in the configuration the old other autostarting tasks will be activated directly from this task which runs non preemptable OIL file TASK Init SCHEDULE NON ACTIVATION 1 AUTOSTART TRUE APPMODE validMode he COM myCOM COMAPPMODE COMMODE he 10 20 Interprocess Communication C source file DeclareComAppMode COMMODE TASK Init Whatever is left to initialize InitOS if other tasks need to run at startup ActivateTask autoT1 ActivateTask autoTN StartCOM COMMODE TerminateTask 10 6 2 Starting the COM Extension If in the OIL file the standard attribute COMSTARTCOMEXTENSION of the COM OIL object is set to TRUE the StartcoM routine calls a user supplied function called StartCOMExtension to interfere with the start up process See section 10 4 4 Initializing Unqueued Messages to learn about how to use this com hook routine in order to initialize messages that are too large or too complex for their initial value to be specified in the OIL file 10 6 3 Stopping the COM The OSEK VDX COM standard provides you with the following service to stop the communication component StatusType StopCOM COMShutdownModeType mode If the given parameter mode equals to COM_SHUTDOWN_IMMEDIATE the service shuts down the communication component immediately This implementation does not define any other additional shutdown mode for the COM component Thus you
95. g to the MESSAGE list of the receiver DataRef points to the application data to be transmitted the type of ApplicationDataRef isa pointer to void When the application calls the ReceiveMessage system service the message object s data are copied to the application buffer pointed to by DataRef Queued messages If the MESSAGEPROPERTY of msg is RECEIVE_QUEUED_INTERNAL msg refers to a queue receive message object queued message A queued message behaves like a FIFO first in first out queue When the queue is empty no message data will be provided to the application When the queue is not empty and the application receives the message the application is provided with the oldest message data and removes this message data from the queue If new message data arrives and the queue is not full this new message is stored in the queue If new message data arrives and queue is full this message is lost and the next ReceiveMessage call on this message object returns the information that a message has been lost 10 8 Interprocess Communication TASK TaskC myStruct st StatusType ret ret ReceiveMessage recMQTaskC amp st if ret E_COM_NOMSG Queue is empty no new messages st contains no valid data return st contains valid data if ret E_COM LIMIT A message has been lost else Everything is okey processMsg amp st TerminateTask
96. he EnableAlliInterrupts service does the opposite it restores the saved state in the previous routine You can call these services from Category 1 ISR and Category 2 ISR and from the task level but not from hook routines These services are intended to encapsulate a critical section of the code no maskable interrupts means no scheduling which in its turn means no concurrency It is a much faster and costless alternative to the GetResource ReleaseResource routines TASK myTask DisableAllInterrupts critical code section EnableAlliInterrupts The critical area should be extremely short though since the system as a whole is on hold even Category 1 ISRs are disabled DisableAllInterrupts must always precede the critical section and immediately after the section EnableAllInterrupts must follow Interrupts Avoid situations situations like below ISR isr DisableAllInterrupts if A EnableAlliInterrupts doSth else return This causes the system to be outbalanced when returning from isr if A is zero Also no API service calls are allowed within this critical section You must avoid code like below TASK myTask DisableAllInterrupts critical code section SetEvent task event not allowed EnableAlliInterrupts You should be careful when building library functions which are potential users of of these services since nested calls are n
97. he application modes are strongly related to the decissions you take at system architecture level The following example of an ATM Automated Teller Machine illustrates this dependency Let us assume an ATM which must meet the following system requirement The ATM software shall be upgraded locally at high speed and being service affecting How could we implement a solution for such a scenario The software can be designed in such a way it has two mutually exclusive modes a download mode and mode for normal operation In this example an external pin must be read before the RTOS starts after reset to distinguish between the modes if the pin is not asserted the image starts in normal mode if asserted the image starts in download mode The download process takes place locally to download a new image an operator must go to the ATM location The operator must in order 1 open the ATM box 2 assert the pin 3 reset the machine to start it in download mode 4 download the image from a notebook 5 deassert the pin 6 reset the machine to start it in normal operation mode 7 close the ATM box Obviously while downloading a new image the ATM is out of service In the OIL file you would define two APPMODE objects like APPMODE NORMAL APPMODE DOWNLOAD 4 3 8051 RTOS In NORMAL operation mode the ATM machine is ready to address customer queries A task is started that offers for example a tactile grap
98. hical interface waiting for customer queries another task maintains a link with a bank database to update the customer latest data a third task interfaces with the paper money repository In DOWNLOAD operation mode the ATM machine starts a task to process download messages over a fast serial line Another task flashes the image to the target read only memory The system is entirely dedicated to download purposes Code example OIL file You define application modes in the OIL file like APPMODE NORMAL APPMODE DOWNLOAD C source file In the application source code like with all other OIL objects you need to declare the application mode before using it The example assumes that the system boot code passes the type of reset that has occurred as argc The application is designed so that the execution flow returns to the boot code when the RTOS shuts down DeclareAppMode mode AppModeType GetAppMode void AppModeType mode check if the pin has been asserted if PinIsAsserted StartOS shall start in DOWNLOAD mode mode DOWNLOAD else StartOS shall start in NORMAL mode mode NORMAL return mode 4 4 The startup process int main int argc ResetType reset argc AppModeType mode common initialization for all modes InitInCLevel different action depending on reset type InitSystem reset find out the run time current mode mode GetAppMode
99. hus Category 2 ISRs must be always disabled during the copy process As a simple mental exercise imagine what would happen if a SendMessage call is preempted by another SendMessage After the two calls the buffer data are corrupted a mix of both messages While copying to from queue receive message objects concurrency problems can be avoided without real need to disable Category 2 ISRs implementing locks in every member of the message queue So keeping interrupts enabled while copying the messages is certainly possible Extra software is needed to handle locks in the queue If the messages to be copied are very long you simply cannot allow interrupts to be disabled during the whole copy process However if messages are so short that interrupts can be easily disabled during the copy process this extra handling should be best avoided If you set the non standard attribute LONGMSG to TRUE Category 2 ISRs are disabled while the copy process of queued messages and extra software i e run time performance is needed If set to FALSE Category 2 ISRs are enabled but no extra software is needed 10 13 8051 RTOS 10 5 Message Notification So far you know how to send and receive messages The question that still remains is how does the receiver know that the sender has just sent a new message For queued messages the receiver can survive by constantly checking the receive queue For unqueued messages the receivers have no means to know wh
100. i hi C source file TASK myTask delay the system one sc SetRelAlarm delayA 1000 OSTICKDURATIONINMSCS 0 WaitEvent delayE ClearEvent delayE TerminateTask Declaring an alarm Like all other OIL objects you need to declare the task before using it in your source in order to compile your module DeclareAlarm myAlarm TASK myTask CancelAlarm myAlarm Alarms 8 4 The C Interface for Alarms You can use the following data types and system services in your C sources to deal with alarm related issues Element C Interface Data Types TickType TickRefType AlarmBaseType AlarmBaseRefType AlarmType Constants OSMAXALLOWEDVALUE_x Xx is a counter name OSTICKSPERBASE x x is a counter name OSMINCYCLE x Xx is a counter name OSMAXALLOWEDVALUE OSTICKSPERBASE OSMINCYCLE OSTICKDURATION OSTICKDURATIONINMSCS SYS_TIMER System Services DeclareAlarm DeclareCounter GetAlarmBase GetAlarm IncrementCounter SetRelAlarm SetAbsAlarm CancelAlarm Table 8 1 The C Interface for Alarms d gt Please refer to the OSEK VDX documentation for an extensive description 8 11 8051 RTOS 8 12 9 Interrupts Summary This chapter describes how you can declare ISR objects in the application OIL file in order to optimize the interrupt configuration 9 1 Introduction An interrupt is a mechanism for providing immediate response to an
101. id concurrency problems when several tasks and or isrs have access to the same critical code section If your ISR demands manipulation of a certain critical section which access is controlled by resource R you need to add R to the list of resources owned by the ISR The OSEK VDxX standard offers you the standard attribute RESOURCE a multiple reference of type RESOURCE_TYPE to add resources to the list of resources owned by the ISR dd Category 1 ISRs cannot own resources If isrUART1 and isrUARTO are ISR OIL objects that update the same counter increased by one unit in isrUART1 and decreased by one unit in isrUARTO in their handlers they must own the same resource R A task printNetto can also be activated to output the value OIL file ISR isrUARTO CATEGORY 2 RESOURCE R b ISR isrUART1 CATEGORY 2 RESOURCE R b TASK printNetto RESOURCE R C source file DeclareResource R ISR isrUartTx GetResource R netto_counter ReleaseResource R 9 7 8051 RTOS ISR isrUartRx GetResource R netto_counter ReleaseResource R TASK printNetto int netto GetResource R netto netto_counter ReleaseResource R printf d netto d 9 7 ISRs and Messages In Chapter 10 Interprocess Communication it will be described how an ISR object might be defined as a sender and or a receiver of messages In both cases the ISR object must own the message Messages are
102. implies a wait operation on multiple semaphores at a time Most hook routines and interrupt service routines can set and check events however only an extended task owning the event can wait for it and or clear it AN Since this implementation does not deal with external communication the scope of the events limits to one application 6 2 Configuring Events To configure an EVENT you must declare a specific EVENT object in the application OIL file of the project EVENT myEvent You do not need to define values for the standard attribute MASK of the EVENT OIL object The RTOS will always calculate this value internally In this example myEvent becomes the identity of the event a specific bit value in a mask b In Appendix A mplementation Parameters you can find the maximum number of events supported by this implementation If you define more EVENT OIL objects in your application OIL file than the maximum number supported the system behaviour would be at least unpredictable 6 1 8051 RTOS 6 3 The Usage of Events 1 An extended task might enter the waiting state and allow other tasks the CPU time until a certain event occurs Thus events form the criteria for the transition of tasks from the running state into the waiting state The transition is performed with the system service StatusType WaitEvent EventMaskType 2 A task can undergo such a transition only if it owns that specific event Or equivalently the event
103. inateTask TASK inthird TerminateTask TASK outsecond TerminateTask TASK infirst Activate outfirst infirst runs outfirst is ready ActivateTask inthird infirst runs inthird is ready ActivateTask outsecond outsecond has run infirst resumes execution Schedule inthird and outfirst have run TerminateTask Features of internal resources are e A task can belong exclusively to a one group of tasks therefore owning a maximum of one internal resource e Internal resources cannot be occupied and or released in the standard way by the software application but they are managed strictly internally within a clearly defined set of system functions Determining the most appropiate range for the priorities of tasks owning an internal resource becomes a key factor in the design In most cases this range of priorities should be reserved exlusively for the members of the group Otherwise low priority tasks in the group could delay tasks outside the group with higher priority This is exactly what has happened in the previous example outfirst was delayed by infirst Resource Management 7 5 The Scheduler as a Special Resource The scheduler can be considered as a special resource that can be locked by the running task As a result while the running task has locked the scheduler it behaves like a non preemptive task with the same re scheduling points If you plan
104. ing this configuration The real time responses of the system are also enhanced since many run time checks are not performed 8051 RTOS 11 10 12 Debugging an RTOS Application Summary This chapter explains how you can easily debug RTOS information with the Cross View Debugger and describes in detail all the information that you can obtain 12 1 Introduction This chapter describes how the debugger can help you to debug your OSEK VDX application Often while debugging you will find situations where having access to certain RTOS information becomes crucial For instance e if you are running code that is shared by many tasks you may need to know which task is executing at that moment e you may need to know the state of your tasks e you may need to know which event a task is waiting for e you may need to know what the priority of the system is once the task has got a resource To provide debug information the OSEK VDX standard proposes a universal interface for development tools the OSEK Run Time Interface ORTI An ORTI file clearly specifies to the debugger what information to access how to access it and how to present it The toolchain generates at system generation time an ORTI file with rules for the debugger to display valuable kernel information at run time Every time you make changes in the OIL file a new ORTI file is generated You need to start the debugger every time a new ORTI file is created The info
105. it rescheduling points for non preemptive tasks are e Successful termination of a task e Successful termination of a task with explicit activation of a successor task e Explicit call of the scheduler e A transition into the waiting state If the tasks in the system are all non preemptive the scheduling policy of the system as whole is said to be non preemptive Be aware of the special constraints that non preemptive scheduling imposes on possible timing requirements while designing your TASK objects A non preemptable task prevents all other tasks from CPU time so their execution time should be extremely short 5 12 Task management 5 6 3 Scheduling Policy In the most general case the system runs with the so called mixed preemptive scheduling policy full preemptable and non preemptable tasks are mixed The current scheduling policy depends on the preemption properties of the running task non preemptable or full preemptive If the running task has its SCHEDULE attribute set to FULL in the OIL file the scheduling policy is fully preemptive Otherwise the scheduling policy will be non preemptive Typically an application will operate in mixed preemptive mode where most of the tasks can be safely preempted while the non preemptable tasks constitute only a small subset among all tasks The code belows shows the behaviour of the system with a mixed preemptive policy OIL file EVENT eT2 nP1 TASK T1 SCHEDULE FULL PRIO
106. ject the task commandHandler is also activated Upon return from the ISR code the commandHandler task is running This tasks receives the oldest message of the recCommand queue message object and interpretates it This method is safe The only problem that could arise would be an overrun in the ISR receive buffer in case the execution time in the ISR code SendMessage exceeds the minimal interarrival time between two consecutive interrupts 10 15 8051 RTOS 10 5 2 Notification Example Set Event Consider the OIL configuration below ISR SerialRx MESSAGE sendCommand he EVENT commandEvent TASK commandHandler MESSAGE recCommand ACTIVATION 3 EVENT commandEvent he MESSAGE sendCommand MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE myCommand ko he MESSAGE recCommand MESSAGEPROPERTY RECEIVE QUEUED INTERNAL SENDINGMESSAGE sendCommand QUEUESIZE 3 hi NOTIFICATION SETEVENT TASK commandHandler EVENT commandEvent hi he In this solution the commandHandler task is never in the suspended state it remains most of the time waiting for event commandEvent The Notification mechanism for recCommand sets now commandEvent immediately after that the command message was copied to the receive object in the ISR code Upon return from the ISR the task commandHandler resumes execution clears the event commandEvent receives and interpretates the message before waiting again for the even
107. l Container of all the other objects OS The OS that runs on the CPU STATUS STARTUPHOOK All system objects are controlled by OS ERRORHOOK SHUTDOWNHOOK PRETASKHOOK POSTTASKHOOK USEGETSERVICEID USEPARAMETERACCESS USERESSCHEDULER CORE EXTDATASIZE LONGMSG MAXNESTEDISR MULTISTART SMAINSTACK STACKMONITOR USERTOSTIMER VISRSTACK APPMODE Defines different modes of operation for the application 3 2 The OSEK VDX Implementation Language OIL OIL system Description Standard Attributes object Non Standard Attributes ISR Interrupt service routines supported by OS CATEGORY RESOURCE MESSAGE ENBIT LEVEL RESOURCE The resource that can be occupied by a task RESOURCEPROPERTY TASK The task handled by the OS PRIORITY SCHEDULE ACTIVATION AUTOSTART RESOURCE EVENT MESSAGE SSTACK VSTACK COUNTER The counter represents hardware software tick MAXALLOWEDVALUE source for alarms TICKSPERBASE MINCYCLE EVENT The event on which tasks may react MASK ALARM The alarm is based on a counter and can either COUNTER activate a task or set an event or activate an ACTION alarm callback routine AUTOSTART MESSAGE The message is defined in OSEK COM and MESSAGEPROPERTY defines a mechanism for data exchange NOTIFICATION between different entities tasks or ISRs COM The communication subsystem The COM object COMERRORHOOK has standard attributes to define general COMERR
108. l objects KOIL Kernel Object Interface Language and a description of OSEK specific objects and attributes 1 3 The OSEK VDX Documentation Information about the OSEK VDX organization including all the standards is available online at www osek vdx org Currently the Altium RTOS is implemented to follow e Operating System OS Version 2 2 1 e Communication COM Version 3 0 1 e OSEK Implementation Language OIL Version 2 4 1 e OSEK VDX Run Time Interface ORTI Version 2 1 1 3 8051 RTOS 1 4 The Altium RTOS The Altium RTOS is a real time preemptive multitasking kernel designed for time critical embedded applications and is developed by Altium The future plans of Altium in parallel to OSEK plans aim to certify the Altium RTOS according to the Specification Binding SB5 following the certification plans in the Certification Binding CB5 of the OSEK industry standard In the first release of the RTOS only a subset internal communication of COMS3 0 is supported This subset of standards is included as online documentation The standards are copyright protected The RTOS is written in ANSI C and assembly and delivered as source code For every RTOS application the RTOS source code is compiled after some mandatory configurational input from the application developer to generate a customized RTOS library The RTOS image is actually this generated library The application source code must be linked with this RTOS library i
109. level but not from hook routine level Below you find an example of how to install an alarm that sets event sensorE to task sensorT exactly when the wheel angle is 0 or 180 degrees SetAbsAlarm sensorAE 0 180 8 You can configure an alarm to run at startup Normally these are periodic alarms carrying out periodic actions required by the application Even before the first tasks have been scheduled the alarm is already running to expiration You must set the attribute AUTOSTART to TRUE The subattributes ALARMTIME and CYCLETIME behave the same as increment and cycle for the SetRelAlarm system service You must indicate also under which application modes the alarm should autostart The next example counts the total number of turns OIL file APPMODE AppModeA ALARM sensorAC COUNTER sensorC ACTION ALARMCALLBACK CALLBACK myCallBack hi AUTOSTART TRUE CYCLETIME 359 APPMODE AppModeA ALARMTIME 359 hi hi The alarm will autostart if the environment is correct C source file DeclareAppMode AppModeA int no_turns 0 int main int argc StartOS AppModeA return 8 8 Alarms ALARMCALLBACK myCallBack no_turns 9 You can use a combination of the system services GetAlarm and CancelAlarm to set timeouts associated to actions StatusType GetAlarm AlarmType TickRefType StatusType CancelAlarm AlarmType With the GetAlarm routine you can check wheth
110. licationMode GetActiveApplicationMode returns the current application mode CurrentMode mode ShutdownOS E_OK return 4 5 8051 RTOS In NORMAL mode the application calls this routine when it has been signaled remotely to start downloading an image First a global variable is updated to indicate the desired next mode and then the system shuts down with the system service ShutdownOS The implementation makes sure that after the execution of this system service the system effectively returns from the API that started the RTOS thus from StartOS The main code should now be changed to static AppModeType CurrentMode NORMAL int main int argc ResetType reset argc InitInCLevel InitSystem reset while 1 StartOS CurrentMode return 1 ed The non standard attribute MULTISTART of the OS object must be defined as TRUE in the OIL file because StartOS can be called more than once Only in systems where application resets will never occur the value of the MULTISTART attribute can be set as FALSE This saves code and data size Dependency on the System Requirements An ATM machine could enter temporarily the download mode out of service without causing great disaster at most some customers will be bothered to try other ATMs nearby But could a telecommunication router stop performing its normal traffic operation because it is upgrading remotely its software This depends str
111. located The resulting file is myrtos abs After you application has been built you can check the following for yourself e in your project folder a dedicated folder with the name myrtos rtos has been created e the TOC tool has placed the generational files g conf c g_isrframe c g conf_types h and g_conf h in this directory e the TOC tool has placed the generational files orti txt and flag h in the project folder e alibrary osek 1ib has been created in the folder myrtos rtos and it has been copied to your project folder with the name myrtos 1ib You can compare the contents of this directory to the files shown in figure 2 1 2 6 Debug Your Application The application myrtos abs is the final result ready for execution and or debugging Since the RTOS environmennt supports ORTI files you can easily gain access to RTOS information during the simulation of your application Make sure that main c is the active file 1 From the Debug menu select Simulate 2 From the View menu select Workspace Panels Embedded RTOS The RTOS panel opens In this panel you can easily obtain RTOS information 3 From the Debug menu select RTOS System Status repeat this step for Tasks and or Resources 2 11 8051 RTOS 2 12 3 The OSEK VDX Implementation Language OIL Summary This chapter describes how you can configure your application with a file written in OIL Osek Implementation Language language which needs to be adde
112. mbedded Software Project The Projects panel opens The new project file Embedded Project1 PrjEmb is shown No documents are added to the project yet e Workspacel Dsnwrk X OF SUE Ost See Embedded Project Pit EO No Documents Added Files go gure Navigator A Getting Started Now save your project You are free to choose a name and a location for the project but you can also follow this example 4 5 From the File menu select Save Project As The Save Embedded Project1 PrjEmb As dialog appears Inthis dialog browse to the folder Altium2004 Examples Embedded Create a new folder with the name firstrtos Browse to this folder Enter the file name for your project myrtos PrjEmb and make sure it is saved as type Embedded Software Project PrjEmb Click on the Save button Add new files to the project Now you can add files you want to be part of your project You can either add existing files or create and add new files In this example two new files are needed main c and myrtos oil 6 In the Projects panel right click on your project myrtos PrjEmb and select Add New to Project C File A new empty file with the name Source1 C is added to your project and opened From the File menu select Save As The Save Source1 C As dialog appears Save your file as main c Repeat steps 6 and 7 for the file myrtos oil Add this file as type text docume
113. me so even when several tasks are competing for the processor at the same time only one task is actually running The RTOS is responsible of saving and restoring the context of the tasks when they undergo such state transitions A task can be in one of the following states Task state Description running The CPU is now assigned to the task and it is executing the tesk Only one task can be in this state at any point in time ready All functional prerequisites for a transition into the running state exist and the task only waits for allocation of the processor The scheduler decides which ready task is executed next waiting A task cannot continue execution because it has to wait for at least one event extended tasks only suspended In the suspended state the task is passive and can be activated Table 5 1 Task States 5 3 1 Basic Tasks A basic task runs to completion unless preempted by a higher priority task Basic tasks can only exist in one of three states e suspended e ready e runnig So a basic state cannot be in the waiting state A basic task can undergo only these next four transitions e activation from suspended to ready e start from ready to running e preempted from running to ready e terminate from running to suspended When the RTOS is started with Startos a basic task is either in the suspended or ready state depending on whether you set the AUTOSTART attribute of
114. mentation Part The Implementation part of the OIL file is delivered with the product in the file osek oil You can find this file in the general include directory of the toolchain This file the Implementation OIL file represents a mandatory interface for the Application OIL part in the OIL file of all projects 3 7 8051 RTOS 3 3 2 Application Part The Application OIL part contains all instances of OIL objects for a given application Below you will find an example of how an OIL file may look like You must select proper names for the OIL objects since they become variables with global scope and with type ObjectType For instance if you define in the OIL file an EVENT object download you cannot define a function with such a name in your source code the least you can expect is link errors include lt osek osek oil gt CPU Sample CPU1 OS Stdos STATUS EXTENDED STARTUPHOOK TRUE ERRORHOOK TRUE SHUTDOWNHOOK TRUE PRETASKHOOK TRUE POSTTASKHOOK TRUE USEGETSERVICEID TRUE USEPARAMETERACCESS TRUE USERESSCHEDULER TRUE CORE TSK51A USERTOSTIMER TRUE RTOSTIMERLEVEL 1 MULTISTART TRUE EXTDATASIZE 12 hi EVENT intervaldelay TASK init PRIORITY 7 SCHEDULE FULL ACTIVATION 1 RESOURCE sem_out AUTOSTART TRUE APPMODE AUTOSTART APPMODE NONAUTOSTART b VSTACK 100 SSTACK 40 3 8 The OSEK VDX Implementation Language OIL TASK monitor
115. n order to build the final application object 1 5 Why Using the Altium RTOS The benefits of using the RTOS to build embedded applications a natural consequence of its future conformance with OSEK VDX products are listed below e High degree of modularity and ability for flexible configurations e Focusing on the time critical aspects the dynamic generation of system objects is left out Instead generation of system objects is done in the system generation phase The user statically specifies the number of tasks resources and services required statically e Error inquiries within the operating system are obviated to a large extent in order to not affect the speed of the overall system unnecessarily A system version with extended error inquiries has been defined It is intended for the test phase and or for less time critical applications e The interface between the application software and the operating system is defined by system services in an ISO ANSI C like syntax with well defined functionality The interface is identical for all implementations of the operating system on various processor families e For better portability of application software the OSEK standard defines a language for a standardised configuration information This language OIL OSEK Implementation Language supports a portable description of all OSEK specific objects such as tasks and alarms Ideal applications are compact real time system th
116. newest task in the ready queue of its priority e The following fundamental steps are necessary to determine the next task to be processed 1 The scheduler searches for all tasks in the ready running state 2 From the set of tasks in the ready running state the scheduler determines the set of tasks with the highest priority 3 Within the set of tasks in the ready running state and of highest priority the scheduler finds the oldest task 5 4 1 Virtual versus Physical Priorities We define virtual priority of a task as the priority of a task as it is given in the application OIL file We define physical priority of a task as the real run time priority of the task Let us think of an application OIL file with three TASK OIL objects defined such that TASK T1 PRIORITY 6 TASK T2 PRIORITY 4 TASK T3 PRIORITY 4 The ready to run array comprises all the ready queues of the system In such a system there will be two ready queues in the ready to run array one per priority level 5 5 8051 RTOS The next figure shows the ready to run array where T1 is running and tasks T2 and T3 are ready T2 being the oldest 7 6 lt T1 5 4 lt T2 lt T3 3 2 1 0 lt Idle Figure 5 1 Virtual ready to run array Now what would in terms of functionality be the differences between this configuration and the next systems System A T
117. ns for the task vs_Sched Scheduling policy of the task NON or FULL vs_Ev_Wait wait mask of the task a mask of all the events that the task is waiting for if any vs_Ev_Set set mask for the TASK object a mask of all the events that are already set for the task vs_Resource Resource locked by the task if running through critical code vs_Group Internal Resource identifying the group to which this task belongs if any vs_StackUsed Indicates the number of bytes currently in use for the Virtual Stack of the task vs_StackAvailable Indicates the number of bytes still available for the Virtual Stack of the task vs_SystemStackPtr Points at the last saved byte in the system stack at task level From here you can do stack tracing System stack stores return addresses Only for ready and waiting tasks vs_VirtualStackPtr Points at the last saved byte in the virtual stack at task level Only for ready and waiting tasks Table 12 2 Task Debug Properties 12 3 8051 RTOS 12 4 Howto Debug Resources The debugger can display relevant information about all the resources in the system 1 From the OSEK ORTI menu select MYOSEK and Resources A window pops up showing a list with all the resources in the system Every resource is described with a set of properties The debugger displays the values of these properties The resource properties are described in the next table Resource Debug Property
118. nt The new project is now ready to be edited DXP automatically creates a makefile for the project in this case myrtos mak This file contains the rules to build your application DXP updates the makefile every time you modify your project settings amp Note that because DXP detected the presence of an oi1 file the makefile contains rules to generate also the RTOS library from the file myrtos oil You can check this for yourself by opening the file myrtos mak 2 7 8051 RTOS 2 4 Edit the Application Files In order to get a working rtos project you must edit main c and myrtos oil It is not necessary to pay attention to the exact contents of the files at this moment Edit the user source code 1 As an example type the following C source in the main c document window include lt osek osek h gt DeclareTask task0 DeclareTask taskl DeclareTask task2 DeclareEvent E1 DeclareEvent E2 DeclareAppMode AP1 int main int argc void argc StartOS AP1 return 1 TASK task0 EventMaskType event ActivateTask taskl while 1 WaitEvent E1 E2 GetEvent task0 amp event if event amp El ActivateTask task2 else if event amp E2 ActivateTask taskl ClearEvent El E2 2 8 Getting Started TASK task1 SetEvent task0 E1 TerminateTask TASK task2 2 Click on the Save Active Document lt Ctrl S gt button to save thi
119. ntMaskType event Wait for any of three events WaitEvent myEvent1 myEvent2 myEvent3 Which one occurred GetEvent myTask amp event take actions if event myEvent1 Actionl if event myEvent2 Action2 if event myEvent3 Action3 TerminateTask return Every time myTask gets activated events are all cleared it waits for one of the three events When running again it uses GetEvent to find out which events have triggered the task 12 You can clear events yourself with the system service StatusType ClearEvent EventMaskType Events 13 You can call this service only from an extended task level never from a hook routine a basic task or an interrupt Adding this service to the previous example you can build a simplified version of an event handler task TASK eventHandler EventMaskType event while 1 Wait for any of three events WaitEvent myEventl myEvent2 myEvent3 Which one occurred GetEvent eventHandler amp event Clear events ClearEvents myEventl myEvent2 myEvent3 take actions if event myEvent1l Actionl if event myEvent2 Action2 if event myEvent3 Action3 TerminateTask return 14 Like all other OIL objects you need to declare the event before using it in your source in order to compile your module DeclareEvent m
120. nter ReleaseResource myResource TASK myTask2 GetResource myResource no_counter so_counter ReleaseResource myResource ISR myIsr GetResource myResource no_counter so_counter t ReleaseResource myResource 7 4 Resource Management But in case this second global counter needs to be updated by a third task myTask3 and a second ISR myOtherlsr then OIL file TASK myTaskl RESOURCE RESOURCE b TASK myTask2 RESOURCE RESOURCE hi ISR myIsr RESOURCE RESOURCE hi TASK myTask3 RESOURCE b ISR myOtherIsr RESOURCE b C source file TASK myTask1 myResS myResN myResS myResN myResS myResN myResN myResN GetResource myResS so_counter ReleaseResource myResS GetResource myResN no_counter ReleaseResource myResN 8051 RTOS TASK myTask2 pen GetResource myResN no_counter ReleaseResource myResN GetResource myResS so_counter ReleaseResource myResS TASK myTask3 eea GetResource myResN so_counter ReleaseResource myResN ISR myIsr ae GetResource myResN no counter ReleaseResource myResN GetResource myResS so_counter ReleaseResource myResS ISR myOtherIsr eea GetResource myResN so_counter ReleaseResource myResN 7 6 Resource Management 6 You
121. o the needs of a particular application In particular since this architecture offers no stack overflow protection mechanisms you should take special care to avoid run time stack overflows d See Appendix B Stack Overflow for an extensive discussion 5 7 1 The Memory Model The RTOS code is compiled with the reentrant model This model favours context switching techniques since almost no copying needs to be done while saving restoring the task AN You can still compile some parts of the application with the large memory model or the mixed memory model e You can compile a task with this model if none of its code and or data is shared with other tasks e You can in very exceptional cases define an interrupt handler with the function qualifier __interrupt and compile it with the large model Since this piece of code runs beyond the scope of the RTOS you must make sure that it executes always with the highest priority to prevent it from being preempted by an RTOS interrupt Obviously this handler cannot share neither code nor data with other routines or call system services These must be considered as extremely rare situations under normal and desirable circumstances you will compile all code with the reentrant model With the reentrant model you make use of two stacks the system stack and the virtual stack 5 7 2 The System Stack The system stack is used using direct internal RAM for return addresses only The stack is allo
122. ocess Communication Unqueued Receive Message On the receiving side an unqueued message is not consumed by reading it returns the last received value each time it is read If in the example the message M is unqueued for TaskB and ISRA both will read from the same buffer on the receive side when using ReceiveMessage The buffer is overwritten for both TaskB and ISRA only when TaskA sends new data ReceiveMessage keeps reading the same value meanwhile More than one receiver can read from the same unqueued buffer Queued Receive Message On the receiving side a queued message can only be read once the read operation removes the oldest message from the queue If in the example the message M is queued for TaskC there will be a dedicated receive queue buffering the messages for TaskC every queued receiver has its own receive queue The queue size for a particular message is specified per receiver If a receive queue is full and a new message arrives the message is lost for this receiver Obviously to prevent this situation the average processing time of a message should be shorter than the average message interarrival time Send Message Object The send message object is the internal container where the data is stored on the sending side every time that a sender of the message attempts to send the message In the example there will be only one send message object for TaskA Receive Message Object The receive message object
123. ock In OSEK it is implemented by temporarily raising the priority of the calling process so that possible other contending lock requests cannot actually happen during the duration of the lock This mechanism is also known as the Priority Ceiling Protocol see section 7 3 Priority Ceiling Protocol The important aspect of this particular mutex implementation is that the resource request never waits This makes it specially suitable for ISRs In this regard the priority levels are internally expanded to include all maskable interrupt levels on top of the highest real priority level Some general ideas are listed below 1 You must define resources for critical sections that you encounter in the system which are liable to concurrency problems 2 You configure resources in your OIL file like RESOURCE myResource RESOURCEPROPERTY STANDARD b 3 It is useless to define a resource without defining its occupiers In the OIL file you can configure a task or an interrupt service routine to own a resource a task or an interrupt service routine owning a resource means that they can occupy and release the resource 7 2 Resource Management Example Let us assume that myTask1 myTask2 and an asynchronous interrupt service routine named my SR need to update the same global counter no_counter In your OIL file you must configure a new RESOURCE object and define myTask1 myTask2 and myISR as owners of the resource OIL file TA
124. ongly on the system requirements Let us assume then a system requirement The ATM software shall be upgraded remotely at low speed and being non service affecting Downloading is performed in the background mingled with the other components in normal operation In such system there is no download mode 4 6 The startup process 4 3 4 Non mutually exclusive application modes So far we think of application modes as performing mutually exclusive tasks Although this describes well the general case it should not be taken as a dogma Let us think what happens in the following mental example Imagine a rack system composed of three identical boards where functionality depends on their position in the rack If the functionality is very different we would possibly maintain three software images But what happens if their functionality is almost identical It would be then convenient to distinguish three modes for each position and determine at run time which code to execute the system service GetActiveApplicationMode will specify the current position in the rack The benefits are obvious it eases the factory programming process only one software image and the boards will become interchangeable 4 4 The RTOS Initialization This section shows what happens inside the system from the moment that you call Startos until the first application task is scheduled and explains how you can intervene in this process The RTOS per
125. or SendMessage DataRef define SENDMESSAGE 2 sys SENDMESSAGE break default all other cases break log errors in a circular buffer containing the last ten errors errorLog gt Paraml Paraml errorLog gt Paran2 Param2 errorLog gt Error Error errorLog gt sys sys errorLogt if errorLog gt startLog sizeof startLog errorLog startLog return 8051 RTOS 11 3 Debug Routines The RTOS calls two other hook routines PreTaskHook and PostTaskHook to enhance your debugging facilities if in the OIL file you set the attributes PRETASKHOOK and POSTTASKHOOK of the OS object to TRUE void PreTaskHook void void PostTaskHook void dd If PRETASKHOOK POSTTASKHOOK is set but you fail to define the PreTaskHook or PostTaskHook routine in your code the linking phase will fail due to unresolved externals PreTaskHook is called by the RTOS after it has switched tasks but before passing control of the CPU to the new task This allows you to determine which task is about to run PostTaskHook is called after the RTOS determines that a switch is to occur but always before the switch actually occurs This allows you to determine which task has just completed or has been preempted You can use PreTaskHook and PostTaskHook to perform for instance some time measurements CPU task usage etc aN In both cases there is still already a task in the running state so that y
126. ou can determine which task is about to be preempted or scheduled with the OS system service Get TaskId The body of the PreTaskHook routine could look like void PreTaskHook void TaskType task TaskStateType state GetTaskID amp task if task INVALID TASK i cannot be here while 1l if RUNNING GetTaskState task amp state i cannot be here while 1l debug code return Error Handling 11 4 OIL Examples To enjoy maximum debug facilities you must configure your OIL file as follows OS Stdos STATUS EXTENDED STARTUPHOOK TRUE ERRORHOOK TRUE SHUTDOWNHOOK TRUE PRETASKHOOK TRUE POSTTASKHOOK TRUE USEGETSERVICEID TRUE USEPARAMETERACCESS TRUE COM Com COMERRORHOOK TRUE COMUSEGETSERVICEID TRUE COMUSEPARAMETERACCESS TRUE COMSTARTCOMEXTENSION TRUE COMSTATUS COMEXTENDED hi To cut out all debug facilities when releasing the product you must configure your OIL file as follows OS Stdos STATUS STANDARD STARTUPHOOK FALSE ERRORHOOK FALSE SHUTDOWNHOOK FALSE PRETASKHOOK FALSE POSTTASKHOOK FALSE USEGETSERVICEID FALSE USEPARAMETERACCESS FALSE COM Com COMERRORHOOK FALSE COMUSEGETSERVICEID FALSE COMUSEPARAMETERACCESS FALSE COMSTARTCOMEXTENSION FALSE COMSTATUS COMSTANDARD b If the size of your image becomes critical you can notably reduce the ROM area size of your application by choos
127. pes h when using messages with non basic CDATATYPE attributes The application OIL file and the configurational files user oil g_conf c g_conf h g_conf_types h g_isrframe c flag h orti txt You must write exactly one application OIL file to configure the RTOS library It is the only oil member of the project and contains the input for the Tasking OIL Compiler TOC These configurational files are intermediate files ANSI C generated by the TOC compiler after processing the OIL file The files g are compiled together with the RTOS source files to build the RTOS library of the project The file flag h is an extra interface for the application software The file orti txt is the runtime debug interface They are rebuilt when you change your OIL file RTOS source files C76 dd t c ON osek h The source code files of the RTOS are located in PRODDIR c51 osek They are used by all the RTOS projects to build their RTOS libraries They should never be removed or modified The RTOS application interface osek h is located in PRODDIR c51 include osek and constitutes the only interface for your code as an RTOS user Implementation OIL file osek oil The implementation OIL file which is located in PRODDIR c51 include osek must be included from the OIL files of all RTOS applications It imposes how and what can be configured in this current RTOS release It should
128. r and TaskB TaskC and ISRA are the receivers This situation is taken as an example for the rest of this section Message The message is the physical application data that is exchanged between a sender and its receivers A message can have zero or more senders and zero or more receivers Messages are equal to data i e bits It has no direct OIL representation The set sender s of a message message data and receiver s of a message represents a closed unit of communication On both the sender and receiver sides there must be an agreement about the type of the exchange message data Sender of a message A TASK or an ISR OIL object can be allowed to send a particular message The sender qualifier relates to a specific message A TASK or an ISR OIL object for example can be a sender for message M1 a receiver for message M2 and none of both for message M3 In the example TaskA prepares the data and uses the system service SendMessage to starts the transmission to the receivers Receiver of a message A TASK or an ISR OIL object can be allowed to receive a particular message Like the sender the receiver qualifier relates to a specific message A TASK or an ISR OIL object for example can be a receiver for message M7 a sender for message M2 and none of both for message M3 In the example TaskB TaskC and ISRA use the system service ReceiveMessage to receive the data sent by TaskA 10 2 Interpr
129. r too complex for their initial value to be specified in the OIL file StatusType InitMessage SymbolicName msg ApplicationDataRef DataRef Although you can call the InitMessage routine at any point in the application s execution the safest practice is to initialize all unqueued messages in the hook routine StartCOMExtension 10 12 Interprocess Communication See section 10 6 2 Starting the COM Extension for more information regarding the com hook routine StartCOMExtension StatusType StartCOMExtension void myStruct st prepare the default message st header 0x12 for i 0 i lt PAYLOADSIZE i d st payload i 0 st cre OxFF initialize the receive message object recMU InitMessage recMU amp st TerminateTask j You can also use InitMessage to reset your unqueued messages at any moment after you have called StartCOM and before you call StopCcoM 10 4 5 Long versus Short Messages Sending a message involves the copying of data from an application buffer into at least one receive message objects Reversely receiving a message involves the copying of data from a receive message object into an application buffer Concurrency problems to protect critical data are resolved inside the RTOS code by means of temporarily suspending all Category 2 ISRs and the system timer While copying to from unqueue receive message objects we need always protection against concurrency t
130. return from ShutdownHook the RTOS cleans up all opened objects and returns from the previous call to Startos The control is given back to the application in main Example of a Shut Down The next example illustrates how a typical ShutdownOs hook routine can look like counts number of resets static int no_resets 0 switch off the system extern void SwitchOff void handles error N extern void HandlerErrorN void void ShutdownHook StatusType Error TaskType task switch Error case E_OK Example of good application reset break case E_OS SYS_VSTACK case E_OS SYS_SSTACK case E OS EEE STAGE case E OS SYS_ISRSTACK case E OS ERE ERROR The startup process RTOS has detected a stack overflow or a fatal internal error We allow the system three errors like these before switching the system off if no_resets 3 SwitchOff break case E_APP_ERRORN application handler for Error N HandlerErrorN break default no idea what has happened SwitchOff break After returning the RTOS will clean up all opened objects and return from the previous call to StartOS The control will be normally given back to the application in main return 4 11 8051 RTOS 4 12 5 Task management Summary This chapter explains how the RTOS manages tasks scheduling policies tasks states and de
131. rmation provided by the ORTI file and displayed by the debugger must be extremely precise and well documented This information can be both dynamic stacks current task last error and static added to help you comprehend the run time environment 12 1 8051 RTOS 12 2 How to Debug the System Status Before you start debugging make sure that e The following options are set in the debugger options of your project orti orti txt radm osek_radm Otherwise the debugger does not load the ORTI file e The system runs in extended mode the attribute STATUS of the OS object must be set to EXTENDED and the COMSTATUS of the COM object must be set to COMEXTENDED if you aim for a maximum of information If you need to debug your application in standard mode you must be aware of the fact that some of the information will be indeterminated Every time you stop the debugger you can have a first look at the current status of the system via some general information This information intends to provide a global and fast description of the system 1 From the OSEK ORTI menu select MYOSEK and System Status A window pops up showing values for some global status properties The status properties are described in the next table Status Debug Property Description RUNNINGTASK Specifies which task is currently running RUNNINGTASK is set to IDLE when none of the application tasks is in running state RUNNINGRESOURCE Specifies
132. rorHook is called at the end of the system service and immediately before returning to the calling function The RTOS does not call COMErrorHook if the failing system service is called from the COMErrorHook itself recursive calls never occur Therefore you can only detect possible error in COM system services in the COMErrorHook itself by evaluating directly their return value Once inside the COMErrorHook routine the RTOS provides you with mechanisms to access valuable information With these mechanisms you can check which COM system service has failed and what its parameters were You can use the macro services listed in the next table for this purpose Macro services inside ComErrorHook Macro Service Description COMErrorGetServiceld Provides the system service identifier where the error has been arisen The return value of the macro is a service identifier of type COMServiceldType and its possible values are COMServicelD_StartCOM COMServicelD_StopCOM COMServicelD_GetCOMApplicationMode COMServicelD_InitMessage COMServicelD_SendMessage COMServicelD_ReceiveMessage COMServicelD_GetMessageStatus The value of the standard attribute of the COM object COMERRORGETSERVICEID must be set to TRUE In all cases below the standard attribute of the COM object COMUSEPARAMETERACCESS must be set to TRUE Returns the value of parameter Mode of the failing system service StartCOM Returns the value of parameter Mode of the failing
133. rvice IncrementCounter Table 11 1 Error Management macro services Error Handling Example of ErrorHook definition The body of the ErrorHook routine could look like void ErrorHook StatusType Error int Paraml Param2 Param3 sys switch OSErrorGetServiceld case OSServiceID_ SetRelAlarm Paraml OSError _SetRelAlarm AlarmID Param2 OSError_SetRelAlarm_increment Param3 OSError _SetRelAlarm_cycle define SETRELALARM 1 sys SETRELALARM break case OSServiceID_GetEvent Paraml OSError _GetEvent_TaskID Param2 OSError _GetEvent_Event define GETEVENT 2 sys GETEVENT break all other cases default break log error in a circular buffer with the last 10 errors errorLog gt Paraml Paraml errorLog gt Paran2 Param2 errorLog gt Param3 Param3 errorLog gt Error Error errorLog gt sys sys errorLogt if errorLog gt startLog sizeof startLog errorLog startLog return 8051 RTOS 11 2 3 The COMErrorHook Routine In both comstandard and comextended modes when a system service returns a StatusType value not equal to E OK the RTOS calls the COMErrorHook routine provided that you set the COMERRORHOOK attribute of the COM object to TRUE in your OIL file void COMErrorHook StatusType ed If COMERRORHOOK is set but you fail to define the COMErrorHook routine in your code the linking phase will fail due to unresolved externals COMEr
134. s a rescheduling point for the RTOS Thus the application may or may not return immediately from the system service imagine what happens if a higher priority task is activated 10 14 Interprocess Communication 10 5 1 Notification Example Activate Task Imagine an application with a serial line command handler The race condition for the reception of the command is the arrival of a line feed At that moment the serial ISR object seria Rx must send a message to a TASK object called commandHandler with the new command The task commandHan ler interpretates the given commands and has the highest priority A possible OIL configuration for this system is shown below ISR SerialRx MESSAGE sendCommand TASK commandHandler MESSAGE recCommand ACTIVATION 3 hi MESSAGE sendCommand MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE myCommand ko hi MESSAGE recCommand MESSAGEPROPERTY RECEIVE QUEUED INTERNAL SENDINGMESSAGE sendCommand QUEUESIZE 3 he NOTIFICATION ACTIVATETASK TASK commandHandler hi The sender message object of the command message is sendCommand The only owner of the sender object is SerialRx it is the only sender When the race condition is met the command message is sent with the last buffered command from the ISR code The command is copied to all the receiver objects in this case only recCommand And since there is a Notification mechanism defined for this receive ob
135. s file SetEvent task0 E2 TerminateTask Edit the user OIL file 1 Edit the myrtos oil file with the following text include lt osek osek oil gt CPU myRTOS Os Stdos STATUS STARTUPHOOK ERRORHOOK SHUTDOWNHOOK PRETASKHOOK POSTTASKHOOK USEGETSERVICEID USEPARAMETERACCESS USERESSCHEDULER CORE USERTOSTIMER b EVENT El EVENT E2 APPMODE AP1 EXTENDED FALSE FALSE FALSE FALSE TSK51A FALSE FALSE FALSE FALSE SO FALSE 2 9 8051 RTOS TASK task0 PRIORITY 5 SCHEDULE FULL ACTIVATION 1 AUTOSTART TRUE APPMODE AP1 EVENT El EVENT E2 hi TASK taskl PRIORITY 5 SCHEDULE FULL ACTIVATION 1 AUTOSTART FALSE he TASK task2 PRIORITY 5 SCHEDULE FULL ACTIVATION 1 AUTOSTART FALSE hi i 2 Click on the Save Active Document lt Ctrl S gt button to save this file Getting Started 2 5 Build Your Application If you have modified and saved the project files you can actually build your first RTOS application This results in an absolute object file which is ready to be debugged You can build this project with default project options Build your Application To build the currently active project e Right click on your project myrtos PrjEmb and select Compile Embedded Project myrtos PrjEmb The OIL file is compiled and the RTOS library is built All files together are compiled assembled linked and
136. save the system stack for every task is SSTACK OS CONTEXTSIZE This memory is allocated statically at compile time 5 7 3 The Virtual Stack The RTOS code is compiled in reentrant model In this model automatics and parameters are all accessed using a virtual stack pointer register allocated as a 16 bit pointer in direct addressable internal RAM label SE The virtual stack must be located in external RAM and grows downwards A task tracks the history of its automatics and parameters in a dedicated area in external data reserved by the RTOS therefore having a virtual stack on its own Context switching for automatics and parameters is as easy as changing the value of the virtual stack pointer You can configure the size of these dedicated areas per task with the attribute VSTACK of the TASK OIL object AN Since tasks normally use RTOS system services the contribution of the RTOS code to the virtual stack growth must be considered The value of the maximum penetration depth of the RTOS code in the virtual stack in bytes is defined as oe RTOS VTSTACK bytes As a result the value of the attribute VSTACK of a task that uses system services must be always higher than _os_RTOS_VTSTACK How much higher depends on the usage of the virtual stack by that particular task d See also Appendix A Implementation Parameters 5 17 8051 RTOS 5 7 3 1 The Run Time Virtual Stack Since an interrupt service routine also uses the virtual sta
137. scribes how you can declare TASK objects in the OIL file in order to optimize your task configuration 5 1 What is a Task A task is a semi independent program segment with a dedicated purpose Most modern real time applications require multiple tasks A task provides the framework for the execution of functions The RTOS provides concurrent and asynchronous execution of tasks The scheduler organises the sequence of task execution including a mechanism which is active when no other system or application function is active the idle mechanism ECC2 OSEK VDX Conformance CLASS The concept of task is obviously the most important concept in the OSEK VDX OS standard In the standard a task is either basic or extended has a well defined static priority is or not preemptable can or cannot enter the waiting state is or not the only owner of a priority level and so on All possible definitions of these attributes and their interrelation define four different conformance class levels BCC1 BCC2 ECC1 ECC2 This OSEK VDX implementation is ECC2 compliant The reason to choose this conformance class is simple it includes all the others any task developed for a BCCx level can be used in a ECCx level and any task written for an xCC1 level can be used in a xCC2 level As a consequence of being ECC2 compliant e The number of task activations can be larger than one e The number of tasks occupying a particular priority level can be larger than one
138. se a 0 value is given Table 12 4 Alarm Debug Properties 12 5 8051 RTOS 12 6 Howto Debug ISRs The debugger can display information about all the ISRs in the system 1 From the OSEK ORTI menu select MYOSEK and ISRs A window pops up showing a list with all the ISRs in the system Every ISR is described with a set of properties The debugger displays the values of these properties The ISR properties are described in the next table ISR Debug Property Description vs_Level Entry in the Vector Interrupt Table Category of the ISR Resource locked by the isr if running through critical code vs_Category vs_Resource vs_ Status Indicates whether the ISR is disabled enabled ON OFF Table 12 5 ISR Debug Properties 12 6 Debugging an RTOS Application 12 7 Howto Debug Messages The debugger can display information only about the receive message objects in the system since the send messages are routed directly to the receiving side 1 From the OSEK ORTI menu select MYOSEK and Messages A window pops up showing a list with all the messages in the system Every message is described with a set of properties The debugger displays the values of these properties The message properties are described in the next table Message Debug Property Description TYPE Indicates the message type RECEIVE_UNQUEUED_INTERNAL or RECEIVE_QUEUED_INTERNAL SENDINGMESSAGE Symbolic name of the sender
139. se versa If set to TRUE the RTOS expects long messages so the interrupts will not be suspended This is at the costs of extra handling The default value is FALSE Gb See section 5 4 The Priority of a Task in Chapter Task Management MULTISTART The MULTISTART boolean attribute specifies whether the system is allowed to start the RTOS more than once undergoing application resets via the usage of ShutdownOS It has a default value of TRUE db See section 4 5 The Shut Down Process in chapter Startup Process 3 4 The OSEK VDX Implementation Language OIL MAXNESTEDISR The MAXNESTEDISR attribute specifies the maximum number of nested ISRs The type of this attribute is UINT32 and the possible values range from 1 to 8 The default value is 2 SMAINSTACK The SMAINSTACK attribute specifies the maximum usage in bytes of the system stack before the application starts the RTOS with the system service StartOS The RTOS allocates a dedicated buffer to save these bytes You can easily find the best value by comparing the value of the system stack pointer after main and before the call to StartOS The type of this attribute is UINT32 It has no default value STACKMONITOR With the STACKMONITOR attribute you can request the RTOS to monitor continuously possible stack overflows for you Although expensive in run time performance the RTOS will inform you as soon as possible with the precise cause of the stack overflow The defaul
140. system service StopCOM Returns the value of parameter Message of the failing system service InitMessage COMError_StartCOM_Mode COMError_StopCOM_Mode COMError_InitMessage_Message COMError_InitMessage_DataRef Returns the value of parameter DataRef of the failing system service InitMessage COMError_SendMessage_Message Returns the value of parameter Message of the failing system service SendMessage 11 6 Error Handling Macro Service Description COMError_ReceiveMessage_Message COMError_ReceiveMessage_DataRef COMError_SendMessage_DataRef Returns the value of parameter DataRef of the failing system service SendMessage Returns the value of parameter Message of the failing system service ReceiveMessage Returns the value of parameter DataRef of the failing system service ReceiveMessage COMError_GetMessageStatus Message Returns the value of parameter Message of the failing system service GetMessageSiatus Table 11 2 Error Management COM macro services Example of COMErrorHook definition The body of the COMErrorHook routine could look like void COMErrorHook StatusType Error int Paraml Param2 sys switch COMErrorGetServicelId case COMServiceID_InitMessage Paraml COMError InitMessage_ Message define INITMESSAGE 1 sys INITMESSAGE break case COMServiceID_SendMessage Paraml COMError SendMessage Message Param2 COMErr
141. t This method is far less safe than the previous one Apart from the previous problem there is a second drawback If the cycle clear event receive message interpretate message wait event is longer than the minimum time between the arrival of two consecutive serial commands events could sometimes get lost and some commands would not be processed 10 16 Interprocess Communication 10 5 3 Notification Example Flag Associated with a receive message object a flag will be set when a new message overwrites the container It remains set until the application explicitely resets the flag or calls ReceiveMessage AN Although theoretically available for all messages the Notification Flag mechanism normally applies only to unqueued messages The drawback is that when the flag is set all you know is that at least one message arrived since the ReceiveMessage call But you never can tell how many messages you might have lost in between But it does solve the problem of uninitialized unqueued messages Next you will find another configuration for the previous problem ISR SerialRx MESSAGE sendCommand hi TASK commandHandler MESSAGE recCommand hi MESSAGE sendCommand MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE myCommand TE hi MESSAGE recCommand MESSAGEPROPERTY RECEIVE UNQUEUED INTERNAL SENDINGMESSAGE sendCommand NOTIFICATION FLAG FLAGNAME FlgComm hi In the C sourc
142. t value is FALSE d See Appendix B Stack Overflow for an extensive description of this attribute USERTOSTIMER The USERTOSTIMER is a parametrized boolean attribute which determines whether ALARM OIL objects based on the system counter have been configured in the system If set to TRUE the RTOS provides the interrupt framework for the timer unit and the application provides its initialization In this case you must set the parameter RTOSTIMERLEVEL with the entry of the timer unit in the Interrupt Vector Table The type of RTOSTIMERLEVEL is UINT32 It has no default value The default value for USERTOSTIMER is FALSE VISRSTACK The RTOS allocates a dedicated buffer for the run time virtual stacks of the interrupts The VISRSTACK attribute specifies the size of this buffer in bytes You should consider the worst case scenario maximum nested number and the RTOS contribution see Appendix A Implementation Parameters The type of this attribute is UINT32 The default value is 100 bytes ISR object ENBIT The ENBIT attribute specifies the SFR register bit that enables disables the ISR The type of this attribute is STRING It has no default value Check the sfr files in PRODDIR c51 include to find the appropiate sfr file name 3 5 8051 RTOS LEVEL The LEVEL attrribute specifies the entry in the Vector Interrupt Table The RTOS source code uses this value as argument for the _ _ interrupt function qualifier The t
143. task is put into the waiting state WaitEvent can only be invoked from the task level of an extended task never from interrupt level or a hook routine You can set an event equivalent to an event occurs to a specific task directly with the system service StatusType SetEvent TaskType EventMaskType If you need to trigger an event for more than one task you need to call SetEvent for each combination of task and event An event can be set indirectly by the RTOS code upon expiration of an alarm provided that its ACTION attribute is set to SETEVENT in the OIL file or when a message is transmitted provided that its NOTIFICATION attribute is set to SETEVENT An event can be triggered from interrupt level common situation You cannot set events to suspended tasks You can set several events at a time for a waiting task or ready task SetEvent myTask myEventl myEvent2 If myTask is waiting for any of the triggered events it will enter the ready state If the task is ready the events shall remain set waiting for evaluation in the next call to WaitEvent You can check which events have been set for a given task with the system service StatusType GetEvent TaskType EventMaskRefType You can call this service to distinguish which event triggered the task and then take the corresponding action This service can be called from a hook routine or from interrupt level 6 3 8051 RTOS TASK myTask Eve
144. tensive description 9 14 10 Interprocess Communication Summary This chapter describes why the communication services offer you a robust and reliable way of data exchange between tasks and or interrupt service routines and how you can declare MESSAGE and COM objects in the application OIL file 10 1 Introduction The OSEK VDX COM normative document provides interfaces for the transfer of data within networks systems E Although the COM and OS standards could be mutually exclusive this implementation combines them both The implementation OIL file see section 3 3 1 mplementation Part in Chapter OIL Language already defines two COM OIL objects the MESSAGE and COM objects So in the application OIL file you configure both the OS and the COM Both the OS and the COM APIs are included in the system header file osek h You must include this header file in your C source to compile your application As was stated in Chapter 1 Introduction to the RTOS Kernel this implementation supports only a subset internal communication of COM3 0 In internal communication the interprocess communication is limited to a single microcontroller where a physical network does not exist amp This COM implementation provides all the features defined by the OSEK DVX standard for the Communication Conformance Class CCCB Without the benefit of communication services the only possibility for tasks and interrupt service routines to share data is
145. terrupts or from hook routines can bring the system to a complete undefined state You should terminate tasks only from task level 5 11 8051 RTOS 5 6 Scheduling a Task A task can be scheduled with one of the following scheduling policies full preemtable and non preemptable scheduling You must assign a scheduling policy to every task in your application OIL file setting the attribute SCHEDULE to either FULL or NON 5 6 1 Full preemptive Tasks Full preemptive scheduling means that the running task can be rescheduled at any moment by the occurrence of trigger conditions pre set by the operating system The running task enters the ready state as soon as a higher priority task becomes ready The rescheduling points for full preemptive scheduling are e Successful termination of a task e Successful termination of a task with activation of a successor task e Activating a task at task level e Explicit wait call if a transition into the waiting state takes place e Setting an event to a waiting task at task level e Release of resource at task level e Return from interrupt level to task level If the tasks in the system are all full preemptive the scheduling policy of the system as whole is fully preemptive During interrupt service routines no rescheduling is performed 5 6 2 Non preemptive Tasks Non preemptive scheduling means that task switching is only performed via an explicitly defined system services The explic
146. the timer value unit OSTICKSPERBASE This value determines how many clock ticks constitute a timer unit OSMINCYCLE This value represents an absolute minimum for the possible expiring times of system alarms You can set alarms which actions are meant to happen after OSMINCYCLE time units OSTICKDURATION The timer duration in nanoseconds OSTICKDURATIONINMSCS The timer tick duration in miliseconds The values for these parameters depend on the RTOS implementation You can look them up in Appendix A Implementation Parameters 8 5 8051 RTOS 8 3 What is an Alarm An alarm is a counter based mechanism to provide services to activate tasks set events or call specific routines when the alarm expires When you set an alarm you basically state two things 1 a preset value for the counter and 2 the action to be triggered when the counter reaches that value 1 You must attach an alarm to only one counter although it is possible to attach multiple alarms to the same counter In the OIL file this looks like follows COUNTER sensorC MAXALLOWEDVALUE 100 TICKSPERBASE 2 MINCYCLE 1 ALARM sensorAE COUNTER sensorC ALARM sensorAA COUNTER sensorC 2 Inthe OIL file you must configure the action to be performed when the alarm expires This information is permanent it is not possible to change the associated action to an alarm at runtime 3 You can configure an alarm to set a specific event when
147. to a stack than space has been allocated for Often stack overflow will simply overwrite the adjacent memory locations causing bugs that are hard to trace A task could unintentionally corrupt the data or stack of another task A misbehaved task could even corrupt the RTOS s own code or internal data structures In our OSEK VDX implementation a static stack area is allocated for every task The user configures the size of the stack in the OIL file The stack area of a task normally tracks the history of return addresses automatics and parameters of every task The RTOS statically allocates buffers for all the stacks at compile time in the generated file g_conf c These buffers occupy contiguous positions in data area If you set the size of stack for task T to 40 bytes the following code for task T would invariantly lead to data corruption TASK T char local 41 This stack overflow example would cause serious problems Since the whole memory area is accessible there has been data corruption The corruption will affect the RTOS s internal data to the stack of another task or to application data Both are possible Later when this corrupted data is used by the application the system will fail to perform correctly Finding the cause of the problem can be extremely difficult since the failure might occur at a time after the moment when the stack actually suffered the overflow 8051 RTOS In order to avoid stack overflow
148. to use the scheduler as a resource you must first set the attribute USERESSCHEDULER to TRUE in the OS object of your OIL file C source code then can look as follows TASK myTask Ea preemptable GetResource RES SCHEDULER I am non preemptable ReleaseResource RES SCHEDULER preemptable e You can neither define nor configure this resource in the user OIL file It is a system resource generated by the RTOS and already present in this OSEK VDX implementation e You do not need to add it to the resource list of any task e Interrupts are received and processed irrespective of the state of the resource Declaring a resource Like all other OIL objects you need to declare the resource before using it in your C source DeclareResource myResource TASK myTask GetResource myResource 7 17 8051 RTOS 7 6 The C Interface for Resources You can use the following data types constants and system services in your C sources to deal with resource related issues Element C Interface Data Types ResourceType Constants RES SCHEDULER System Services DeclareResource GetResource ReleaseResource Table 7 1 C Interface for Resources Gs Please refer to the OSEK VDX documentation for an extensive description 8 Alarms Summary This chapter describes how the RTOS offers alarms mechanisms based on counting specific recurring events and describes how you can declare th
149. u need to instantiate and or define OIL objects and assign values to their attributes 3 1 8051 RTOS IN An OSEK VDX implementation can limit the given set of values for object attributes Since the non standard attributes are OSEK VDX implementation specific they are not portable However there are two reasons to justify non standard attributes e they can address platform specific features e they can provide extra configuration possibilities for a certain target 3 2 2 Overview of System Objects and Attributes The next table shows the list of system objects with their standard attributes as defined by OIL2 4 1 and the non standard attributes for the 8051 The non standard attributes are marked italic AN Because the Altium RTOS supports only internal communication the IPDU object and some standard attributes of other objects are not included in the list d In addition to the attributes listed in the table below there are a number of non standard attributes which are not included in this table These extra attributes all start with the keyword WITH_AUTO and take AUTO as their default value you can search for them in the file osek oil This subset of attributes can be considered as internals of the implementation and are not user configurable Instead the TOC tool initializes them OIL system Description Standard Attributes object Non Standard Atiributes CPU The CPU on which the application runs under the RTOS contro
150. u now can use the following system service to send a message StatusType sendMessage SymbolicName msg msg DataRef ApplicationDataRef DataRef a Send Message Object in your file with value SEND_STATIC_INTERNAL for its MESSAGEPROPERTY attribute msg must belongs to the MESSAGE list of the sender points to the application data to be transmitted the type of ApplicationDataRef isa pointer to void In our example this could lead to the following C source code DeclareMessage sendM TASK TaskA int data getData data getData SendMessage sendM amp data TerminateTask When you return from the SendMessage system service the data has already been transmitted i e copied into the receive message objects 10 6 Interprocess Communication 10 4 2 How to Define the Data Type of a Message When the value of the subattribute CDATATYPE for a SEND_STATIC_INTERNAL message does not correspond to a basic type you need to add an extra header file in the project that contains the type definition the RTOS software needs this information to build its internal buffers The name of the file is hard coded in the RTOS code as mytypes h and its location must be the project folder Let us assume that you want to send a message whose layout can be divided into a header first byte payload next 20 bytes and crc last byte You must edit the file myt ypes h with the type definitions ifndef MYTYPE
151. unter e You decide how these resources are used e A general method for all hardware systems and derivatives is extremely hard to define And the constraints e The interarrival time of the clock hardware interrupt must be OSTICKDURATIONINMSCS mscs See Appendix A Implementation Parameters e You cannot handle application code in the clock interrupt e You cannot define an interrupt handler at entry RTOSTIMERLEVEL in Vector Table Besides you need to define three extra routines 1 void DisableRTOSTimer void Disables the clock maskable interrupt associated with the system counter 2 void EnableRTOSTimer void Enables the clock maskable interrupt associated with the system counter 3 void ReloadRTOSTimer void Reloads if not auto reload the associated timer registers otherwise empty To be called from the interrupt framework A failure to define any of these three routines will be detected at linking phase As the RES_SCHEDULER object you can neither define nor configure this counter the SYSTEM_TIMER object in the OIL file It is a counter already generated by the RTOS and already present in this OSEK VDX implementation Following the same discussion you cannot call the IncrementCounter system service with SYSTEM_TIMER as parameter 8 4 Alarms The OSEK VDX implementation defines the following system constants related to the system timer OSMAXALLOWEDVALUE This value determines the upper limit for
152. upts They can be preempted if e the incoming ISR has a higher PRIORITY value e the incoming ISR has the same PRIORITY value but higher harware priority check your chipset manual The attribute MAXNESTEDISR of the OS object configures the maximum run time depth of nested interrupt levels You may encounter problems when interrupts of different categories are nested The premises are e All interrupts must be processed before returning to task level e Rescheduling takes place on termination of an ISR of category 2 but only at first nesting level What happens then if a Category 2 ISR2 interrupts a category 1 ISR1 No rescheduling takes place in SR2 because it is a nested interrupt but also not when SA7 is falling back to task level Therefore having high priority tasks activated or events set from interrupt level in ISR2 has caused no rescheduling and the system has returned to task level at the very same point Thus any activities corresponding to the calls of the operating system in the interrupting SR2 are unbounded delayed until the next rescheduling point You can solve this situation by configuring all the Category 1 ISRs so that they can never be preempted by any Category 2 ISR Thus the highest priority among all Category 2 ISRs should be lower than the lowest priority among all category 1 ISRs 9 6 Interrupts 9 6 ISRs and Resources In Chapter 7 Resource Management it was described how we can use resources to avo
153. upts with priority P or lower and or higher priority ready tasks activated by interrupts with priority higher than P are now allowed to execute When the interrupt service routine ISR gets resource R all other ISR interrupts with priority P and lower are temporarily disabled this includes all other R owners until R is released The RTOS must handle possible nested accesses to resources at different priority levels 7 3 4 Calculating the Ceiling Priority This RTOS implementation adds a non standard attribute to the RESOURCE object UINT32 WITH_AUTO 0 255 CEILING AUTO where the ceiling priority is calculated for every resource Although you can assign your own value for this attribute overwriting the generated RTOS value in the OIL file you should always let the RTOS generate the value for you AN See Chapter 12 Debugging an RTOS Application for a description about how to check that value with the debugger 7 12 Resource Management Consider the following OIL file configuration RESOURCE R1 RESOURCEPROPERTY STANDARD RESOURCE R2 RESOURCEPROPERTY STANDARD RESOURCE R3 RESOURCEPROPERTY STANDARD RESOURCE R4 RESOURCEPROPERTY STANDARD TASK T PRIORITY 5 RESOURCE R1 RESOURCE R2 RESOURCE R3 hi TASK T1 PRIORITY 6 RESOURCE R1 TASK T2 PRIORITY 7 RESOURCE R2 RESOURCE R4 hi TASK T3 PRIORITY 8 RESOURCE R3 b TASK T4 PRIORITY
154. use a subset of system services Thus the internal status of the RTOS might have been altered after the ISR has been served Now after the ISR s handler processing may or may not continue exactly at the same instruction where the interrupt did occur If no other interrupt is active and the preempted task does not have the highest priority among the tasks in the ready to run array anymore rescheduling will take place instead ISRs of this category have the most overhead and because they cannot always run concurrently with the RTOS code they access internals of the RTOS via their system services they are constantly enabled disabled An example could be a serial interrupt receiving characters and storing them in a buffer When a end of frame character is received a message is sent to a task in order to process the new frame ISR isrSerial CATEGORY 2 LEVEL 4 ENBIT ES hi These interrupts typically require a task to be activated an event to be set or a message to be sent The list of available system services follows ActivateTask GetTaskID GetTaskState GetResource ReleaseResource SetEvent GetEvent GetAlarmBase GetAlarm SetRelAlarm SetAbsAlarm CancelAlarm GetActiveApplicationMode ShutdowOSs Category 2 ISRs can establish for instance run time differences in functionality with the GetApplicationMode system service 9 5 8051 RTOS 9 5 Nested ISRs The peripheral IO interrupts are maskable interr
155. used to pass information between interrupt service routines like you pass arguments to a function The OSEK VDX standard offers you the standard attribute MESSAGE a multiple reference of type MESSAGE_TYPE to add messages to the list of messages owned by the ISR Let us suppose that an ISR object isrSender sends a message to another ISR object isrRec Your application OIL file and C source file now look like follows OIL file MESSAGE recMsg MESSAGEPROPERTY RECEIVE UNQUEUED_INTERNAL SENDINGMESSAGE sendMsg INITIALVALUE 1 hi hi MESSAGE sendMsg MESSAGEPROPERTY SEND STATIC INTERNAL CDATATYPE int hi hi 9 8 Interrupts C source file DeclareMessage sendMsg DeclareMessage recMsg ISR isrSender int data GetData SendMessage sendMsg amp data return ISR isrRec int data ReceiveMessage recMsg amp data ProcessData data return 8051 RTOS 9 8 Fast Disable Enable API Services The OSEK VDX standard provides you with a number of fast disable enable API functions The implementation hides all the internal target details for you These services come always in pairs 9 8 1 Disable Enable All Interrupts You can use the following system services to disable enable all maskable interrupts void DisableAllInterrupts void void EnableAlliInterrupts void The DisableAlliInterrupts service clears the global interrupt enable bit and saves the current state T
156. utive hardware clock interrupts The implementation parameter OSTICKDURATIONINMSCS defines the length in miliseconds of the system tick The OSEK VDX standard does not provide you with the concept of a timer interface If you want to time certain actions in your application you must declare an alarm with the system timer as a counter ALARM myAlarm COUNTER SYSTEM TIMER hi If you plan to use the system counter i e if you define ALARM OIL objects that are based on the system_counter you first need to set the non standard attribute of the OS OIL object USERTOSTIMER to TRUE OS myOS USERTOSTIMER TRUE RTOSTIMERLEVEL 1 hi The RTOSTIMERLEVEL sub attribute declares the entry of the timer interrupt in the Interrupt Vector Table The RTOS needs this information to build the interrupt framework The timer interrupt behaves as a Category 2 ISR See section 9 4 The Category of an ISR object in Chapter Interrupts to learn what Category 1 and Category 2 interrupts are 8 3 8051 RTOS When the USERTOSTIMER attribute is set to TRUE the RTOS code calls the following routine to initialize the system counter during the start up process extern void InitRTOSTimer void You are responsible for providing a definition for the routine Otherwise the linking phase will fail due to unresolved externals The advantages of this method are listed below e You decide exactly which hardware resources are taken by the system co
157. when writing your application OIL file e Hardware Resources Implementation Parameters They evaluate the impact of having a RTOS on the hardware resources of the system RAM ROM interrupts times etc e Performance Implementation Parameters They measure the real time response of the RTOS The basic conditions to reproduce the measurement of those parameters are mentioned 8051 RTOS 2 Functionality Implementation Parameters Parameter Description Implementation MAX_NO_TASK Maximum number of tasks Limits the total number of TASK OIL objects in the OIL file MAX_NO_ACTIVE_TASK Maximum number of active tasks i e not suspended in the system The most general scenario is when all tasks can be active at any given time thus all having a stack of their own If this Parameter equals to 1 only one task can be active at a time and the same stack can be shared by all tasks MAX_NO_PRIO_LEVEL Maximum number of physical priority levels Limits the total number of TASK OIL objects with different PRIORITY value The total number of physical priority levels is calculated by the TOC tool after processing the application OIL file MAX_TASK_PER_LEVEL Maximum number of tasks per priority level The implementation supports the general case thus allowing many ready task in the same priority level However better performance is achieved when no more than one task is assigned statically
158. with means to interact with a counter object at run time Let us assume that your application needs to monitor the rotation of a wheel Your application software project agreement works with one degree as the atomic unit in the system You must define a COUNTER object in your OIL file to represent the rotated angle 8 1 8051 RTOS COUNTER sensorC MAXALLOWEDVALUE 359 MINCYCLE 5 ho MAXALLOWEDVALUE is set to 359 since this corresponds to one complete full turn 860 degrees is equivalent to 0 degrees MINCYCLE depends on the application sensibility and or the hardware itself In this example your application cannot log any action that happens before the wheel has rotated five degrees You build the application regardless of the hardware interface that shall eventually monitor the wheel rotation Ideally the dependency on such a device should be minimized Suppose three different sensors S1 S2 S3 are candidates for the hardware interface They all interface equally with your chip the pulse is converted into an external IO hardware interrupt But they send the pulses at different rates S1 every quarter of a degree S2 every half a degree and S3 every degree The impact of this on your application is minimal You only need to modify the TICKSPERBASE attribute of your OIL file This attribute defines how many ticks are requested to increase the counter unit by one Hence the value for the attribute must be 4 if S1 2 if
159. y there is no need to disable the Category 1 ISRs in order to prevent concurrency problems they are never a rescheduling point If you do not want to disable a Category 1 ISR while executing certain critical code you can use the following pair to encapsulate the section instead void SuspendOSInterrupts void void ResumeOSInterrupts void This pair saves and restores the status of only the Category 2 ISRs The overhead in code is bigger since they must go through the interrupt enable bits for all Category 2 ISRs to save restore their priority levels OIL file ISR myISR1 CATEGORY 1 ISR myISR2 CATEGORY 2 C source file TASK myTask SuspendOSInterrupts critical code myISR1 is enabled myISR2 is disabled ResumeOSInterrupts return The considerations for the pair DisableAllInterrupts EnableAllInterrupts apply here too Like the pair SuspendAllInterrupts ResumeAllInterrupts nesting is allowed 9 13 8051 RTOS 9 9 The C Interface for Interrupts You can use the following data types and system services in your C sources to deal with interrupt related issues Element C Interface Data Types Constants System Services EnableAlllnterrupts DisableAlllnterrupts ResumeAllInterrupts SuspendAlllnterrupts ResumeOSinterrupts SuspendOSInterrupts Table 9 1 The C Interface for Interrupts d Please refer to the OSEK VDX documentation for an ex
160. yEvent TASK task WaitEvent myEvent 6 5 8051 RTOS 6 4 The C Interface for Events You can use the following data types and system services in your C sources to deal with event related issues Element C Interface Data Types EventMaskType EventMaskRefType Constants System Services DeclareEvent SetEvent ClearEvent GetEvent WaitEvent Table 6 1 The C Interface for Events d Please refer to the OSEK VDX documentation for an extensive description 6 6 7 Resource Management Summary This chapter explains how the RTOS performs resource management and describes how you can declare RESOURCE objects in the OIL file in order to optimize your resource configuration 7 1 Key Concepts Below a number of key concepts are explained Critical code A critical code section is a piece of software that must be executed atomically to preserve data integrity and hardware integrity Critical code sections handle for example e access to shared variables e most manipulations of linked lists e code that increment counters An example of critical code is Ls ai Li If i is initially set to zero and two processes both execute this code as a result the value of i should be incremented to 2 If Process A executes the code and then Process B does the result will be correct However if A executes and during the increment instruction process B also executes the same code i may only
161. ype of this attribute is UINT32 It has no default value TASK object SSTACK The SSTACK attribute specifies the contribution of the task to the system stack in bytes The type of this attribute is UINT32 The default value is 30 db Section 5 7 2 The System Stack in Chapter Task Management VSTACK The VSTACK attribute specifies the contribution of the task to the virtual stack in bytes The type of this attribute is UINT32 The default value is 30 d gt Section 5 7 3 The Virtual Stack in Chapter Task Management 3 6 The OSEK VDX Implementation Language OIL 3 3 The Structure of an OIL File The complete OIL configuration consists of two parts files e Implementation Part Definition of the OIL system objects with their standard and implementation specific features the standard and non standard attributes The implementation part is delivered with the product as a separate system OIL file which you must include before the application part The file is named osek oil and should never be modified It is located in the general include directory of the toolchain e Application Part Defines the structure of the application located on the particular CPU For every RTOS application you must write an application part or user OIL file In this file you instantiate objects that are defined in the implementation part The user OIL file must therefore include the system OIL file like you include header files in a C sourc
162. ype SetRelAlarm AlarmType alarm TickType increment TickType cycle When this sytem service is invoked the alarm is set to expire at the current value of the counter plus an increment if the alarm is not in use obviously The increment value must be in all cases at least equal to the MINCYCLE attribute of the associated counter The cycle parameter determines whether the alarm is periodic or not If not zero the alarm is restarted upon expiration with cycle as a new timeout interval at least equal to the MINCYCLE attribute of the associated counter If cycle equals to zero the alarm will not be restarted This service can be called at task or interrupt level but not from hook routine level Below you will find an example of how to install an alarm that will activate task sensorT every 90 degrees SetRelAlarm sensorAA 90 90 7 You can use the system service SetAbsAlarm to predefine a counter value for an alarm to expire in absolute terms Type SetAbsAlarm AlarmType alarm TickType increment TickType cycle When this sytem service is invoked the alarm is set to expire at an specific absolute value of the counter if the alarm is not in use obviously The cycle determines whether the alarm is periodic or not If not zero the alarm is restarted upon expiration with cycle as a new timeout interval If cycle equals to zero k the alarm will not be restarted 8 7 8051 RTOS This service can be called at task or interrupt

8051 RTOS

Contents

Download Pdf Manuals

Related Search

Related Contents