Testing Framework

Relevant source files

• src/tests.rs

The testing framework provides a unified, macro-based testing infrastructure that ensures consistent validation and performance measurement across all scheduler implementations in the crate. The framework automatically generates comprehensive test suites for each scheduler type, including basic functionality verification and performance benchmarks.

For information about the specific scheduler implementations being tested, see Scheduler Implementations. For details about the development workflow and CI pipeline that executes these tests, see Development Guide.

Macro-Based Test Generation

The testing framework is built around the def_test_sched macro, which generates a complete test module for any given scheduler and task type combination. This approach ensures that all scheduler implementations are tested consistently with identical test logic.

Test Generation Architecture

flowchart TD
MacroDefinition["def_test_sched macro"]
MacroParams["Parameters: name, scheduler, task"]
ModuleGeneration["Generated Test Module"]
BasicTest["test_sched()"]
YieldBench["bench_yield()"]
RemoveBench["bench_remove()"]
MacroInvocations["Macro Invocations"]
FifoTests["fifo module"]
RRTests["rr module"]
CFSTests["cfs module"]
FifoScheduler["FifoScheduler"]
FifoTask["FifoTask"]
RRScheduler["RRScheduler"]
RRTask["RRTask"]
CFScheduler["CFScheduler"]
CFSTask["CFSTask"]

CFSTests --> CFSTask
CFSTests --> CFScheduler
FifoTests --> FifoScheduler
FifoTests --> FifoTask
MacroDefinition --> MacroParams
MacroInvocations --> CFSTests
MacroInvocations --> FifoTests
MacroInvocations --> RRTests
MacroParams --> ModuleGeneration
ModuleGeneration --> BasicTest
ModuleGeneration --> RemoveBench
ModuleGeneration --> YieldBench
RRTests --> RRScheduler
RRTests --> RRTask

Sources: src/tests.rs(L1 - L84) 

The macro accepts three parameters: a module name identifier, a scheduler type, and a task wrapper type. It then generates a complete test module containing three distinct test functions that exercise different aspects of scheduler behavior.

Test Types

The framework provides three categories of tests, each targeting specific scheduler capabilities and performance characteristics.

Basic Functionality Tests

The test_sched function verifies core scheduler operations using a controlled set of 11 tasks. This test validates the fundamental scheduler contract defined by the BaseScheduler trait.

Test Execution Flow

flowchart TD
TestStart["test_sched() begins"]
CreateScheduler["scheduler = Scheduler::new()"]
AddTasks["Add 11 tasks (i = 0..10)"]
SchedulingLoop["Execute 109 scheduling cycles"]
PickNext["pick_next_task()"]
ValidateTask["assert_eq!(*next.inner(), i % 11)"]
TaskTick["task_tick(&next)"]
PutPrev["put_prev_task(next, false)"]
NextIteration["i < 109?"]
DrainScheduler["Drain remaining tasks"]
CountTasks["Count remaining tasks"]
ValidateCount["assert_eq!(n, 11)"]
TestComplete["Test Complete"]

AddTasks --> SchedulingLoop
CountTasks --> ValidateCount
CreateScheduler --> AddTasks
DrainScheduler --> CountTasks
NextIteration --> DrainScheduler
NextIteration --> PickNext
PickNext --> ValidateTask
PutPrev --> NextIteration
SchedulingLoop --> PickNext
TaskTick --> PutPrev
TestStart --> CreateScheduler
ValidateCount --> TestComplete
ValidateTask --> TaskTick

Sources: src/tests.rs(L8 - L29) 

The test performs multiple scheduling rounds to verify that tasks are selected in the expected order based on each scheduler's algorithm. The task_tick call ensures proper time accounting for time-based schedulers like Round Robin and CFS.

Performance Benchmarks

The framework includes two performance benchmarks that measure different scheduler operations under load.

Yield Benchmark

The bench_yield function measures task switching performance by repeatedly picking and yielding tasks in a large task set.

flowchart TD
BenchSetup["Setup 1M tasks"]
BenchLoop["3M yield cycles"]
PickTask["pick_next_task()"]
PutTask["put_prev_task(task, false)"]
Measure["Measure elapsed time"]
Result["Time per yield operation"]

BenchLoop --> PickTask
BenchSetup --> BenchLoop
Measure --> Result
PickTask --> PutTask
PutTask --> Measure

Sources: src/tests.rs(L32 - L52) 

Remove Benchmark

The bench_remove function measures task removal performance by removing tasks from the scheduler in reverse order.

Sources: src/tests.rs(L54 - L77) 

Test Application Across Schedulers

The framework applies identical test logic to all three scheduler implementations through macro invocations, ensuring consistent validation while accommodating scheduler-specific type parameters.

Scheduler Test Instantiation

flowchart TD
MacroExpansion["def_test_sched macro expansion"]
TestModules["Generated Test Modules"]
FifoMod["mod fifo"]
RRMod["mod rr"]
CFSMod["mod cfs"]
FifoTypes["FifoScheduler, FifoTask"]
RRTypes["RRScheduler, RRTask"]
CFSTypes["CFScheduler, CFSTask"]
FifoTests["fifo::test_sched(), fifo::bench_yield(), fifo::bench_remove()"]
RRTests["rr::test_sched(), rr::bench_yield(), rr::bench_remove()"]
CFSTests["cfs::test_sched(), cfs::bench_yield(), cfs::bench_remove()"]

CFSMod --> CFSTypes
CFSTypes --> CFSTests
FifoMod --> FifoTypes
FifoTypes --> FifoTests
MacroExpansion --> TestModules
RRMod --> RRTypes
RRTypes --> RRTests
TestModules --> CFSMod
TestModules --> FifoMod
TestModules --> RRMod

Sources: src/tests.rs(L82 - L84) 

Each scheduler implementation receives type-appropriate parameters:

• FIFO: Simple usize task payload with FifoScheduler and FifoTask
• Round Robin: usize payload with time slice constant 5 for RRScheduler<usize, 5> and RRTask<usize, 5>
• CFS: usize payload with CFScheduler and CFSTask for virtual runtime scheduling

Performance Measurement

The benchmarks provide quantitative performance data for comparing scheduler implementations. The framework measures operation latency using std::time::Instant and reports results in time per operation.

Benchmark Output Format

The benchmarks output performance metrics in a standardized format:

• Yield Benchmark: "{scheduler_name}: task yield speed: {duration}/task"
• Remove Benchmark: "{scheduler_name}: task remove speed: {duration}/task"

This consistent reporting enables direct performance comparison between scheduler algorithms and helps identify performance characteristics specific to each implementation.

Sources: src/tests.rs(L47 - L51)  src/tests.rs(L72 - L76) 

The testing framework ensures that all scheduler implementations meet the same functional requirements while providing objective performance data to guide scheduler selection for specific use cases.