Hands-On: AUTOSAR Stack Debugging

// Exercise 1: OsTask_10ms is taking > 10ms; system shows timing violations
// Goal: find which runnable is causing the overrun

// Step 1: Set breakpoints at task entry and exit (OS scheduler hooks)
Break.Set OsTask_10ms_Entry /Program    // first instruction of 10ms task body
Break.Set OsTask_10ms_Exit  /Program    // last instruction before ChainTask/TerminateTask

// Step 2: Use STM timer to measure wall-clock task duration
// Aurix STM0 counts at CPU clock (300 MHz); increment = 3.33 ns
LOCAL &t_start &t_end &duration_us
Go
WAIT !STATE.RUN() 1s      // halt at entry
&t_start=Data.Long(SFR:STM0.TIM0)
Go
WAIT !STATE.RUN() 1s      // halt at exit
&t_end=Data.Long(SFR:STM0.TIM0)
&duration_us=(&t_end-&t_start)/300.    // 300 ticks per microsecond
PRINT "OsTask_10ms duration: " FORMAT.FLOAT(2.,1.,&duration_us) " µs"
IF &duration_us>10000.
    PRINT %ERROR "OVERRUN: task took " FORMAT.FLOAT(2.,0.,&duration_us) " µs (budget: 10000)"

// Step 3: Profile which function is using the most time
// Run with ETM trace enabled; use runtime-measurement-profiling for full profile
// Quick method: set breakpoints inside each runnable; measure STM between calls
// Expected finding: Can_MainFunction has unexpected loop iterating all 128 mailboxes
//                   instead of only active ones — linear scan O(n) vs O(1) expected

// Exercise 2: System appears frozen; OsTask_10ms and OsTask_100ms both waiting
// Classic OSEK resource priority ceiling deadlock scenario

// Step 1: Check all task states
TASK.List    // expected: both tasks in WAITING state simultaneously
             // impossible in non-extended OSEK unless there is a deadlock

// Step 2: Check which resources are held
// OSEK resource occupation stored in task control block
TASK.select "OsTask_10ms"
Var.View p_TaskControlBlock_10ms.resource_mask    // which resources held?

TASK.select "OsTask_100ms"
Var.View p_TaskControlBlock_100ms.resource_mask

// Step 3: Trace back to the offending GetResource call
TASK.select "OsTask_10ms"
Frame.view /Caller    // which function called GetResource?

// Expected finding:
// OsTask_10ms holds OsResource_ComBuffer and waits for OsResource_CalTable
// OsTask_100ms holds OsResource_CalTable and waits for OsResource_ComBuffer
// -> circular wait -> deadlock
// Fix: establish consistent resource acquisition order across all tasks

// Exercise 3: ProtectionHook fires with E_OS_STACKFAULT — find the root cause

// Step 1: Break at ProtectionHook entry
Break.Set Os_ProtectionHook /Program

// Step 2: Run to hook
Go
WAIT !STATE.RUN() 5s

// Step 3: Inspect error parameters
Var.View OSError_GetServiceId()      // which OS service was being called?
Var.View OSError_GetParam1()         // first parameter
TASK.List                            // which task triggered the protection?

// Step 4: Switch to the faulting task context
TASK.select "OsTask_1ms"             // assume this is the faulting task
Frame.view /Caller                   // what was the task executing?

// Step 5: Inspect stack state
LOCAL &base &used
&base=Var.VALUE(OsTask_1ms_Stack)
&used=&base+Var.VALUE(sizeof_OsTask_1ms_Stack)-Register(SP)
PRINT "Stack used: " &used " bytes"

// Expected: recursive CRC function with no recursion depth limit
// Input from unchecked NVM causes 500+ recursion levels -> stack exhaustion
// Fix: convert recursive CRC to iterative; or add depth limit guard