mirror of https://github.com/eclipse-threadx/threadx.git synced 2026-10-06 06:59:08 +08:00

Files

T

Frédéric Desbiens 9a03838381 Stopped the test runners from reporting an incomplete build as test failures (#709 )

The cmake test runners drove Ninja with its default keep-going of 1, so the
first failing target ended the build. Every target scheduled after it was
simply absent, and ctest reports a missing binary as a failing test. A
single link error therefore produced a failure count that moved with build
scheduling order rather than with the code.

Measured on the RISC-V64 regression suite, where two targets genuinely
cannot link:

    before   79 of 95 test binaries built
    after    93 of 95 test binaries built

Fourteen perfectly good binaries were being skipped and counted as
failures. Passing -k 0 lets Ninja finish everything it can; the build still
exits non-zero when a target fails.

Three related problems in the same paths are fixed with it.

A failing configuration used to abort the loop over configurations, so
under set -e the ones after it went unbuilt or untested. The build loops
and the serial test loops now accumulate status and return it at the end,
which is what the parallel test branch already did with wait, and what
cmake_bootstrap.sh already documented for ctest.

Capturing that status removes the set -e protection inside the functions,
so two latent faults become reachable and are closed here. A failed pushd
would have let ctest run in the source tree, where it finds no tests and
reports success; the pushd is now guarded. And ctest's status was
discarded by the popd that follows it, so a configuration with failing
tests returned 0 and was reported as a pass; the status is now carried
past the popd and the summary steps.

Verified on the RISC-V32 suite, which has two genuine failures in each of
its five configurations. Both the serial and the parallel branch now test
all five and exit 8, where the serial branch previously stopped after the
first configuration.

The tx and smp runners are symlinks to scripts/cmake_bootstrap.sh, so they
are covered by the one change there.

Assisted-by: Claude Code (Opus 5) <noreply@anthropic.com>

2026-09-09 10:56:28 -04:00

cmake

Stopped the test runners from reporting an incomplete build as test failures (#709 )

2026-09-09 10:56:28 -04:00

fixtures

…

regression

Fixed the kernel object leaks on the static creation error paths (#584 )

2026-08-09 08:44:30 -04:00

readme.md

…

readme.md

FreeRTOS compatibility layer regression tests

Regression tests for utility/rtos_compatibility_layers/FreeRTOS/tx_freertos.c, built against the real ThreadX Linux port.

Running them

./scripts/build_freertos.sh
./scripts/test_freertos.sh

Or directly, which is the same thing:

test/freertos/cmake/run.sh build all
test/freertos/cmake/run.sh test all

CI runs both scripts through .github/workflows/regression_test.yml.

What they cover

Every function in the layer that creates an object takes one or two byte pool allocations for its bookkeeping and then creates one or more ThreadX kernel objects. When one of those kernel objects cannot be created, the function returns NULL, or pdFAIL for xTaskCreate(), and the caller is left with no handle and therefore no way to call the matching delete function. Anything the layer fails to release on the way out is lost until the system restarts.

That makes these error paths invisible from the outside: a leaking version and a correct version return exactly the same thing to the caller. The suite therefore counts the ThreadX primitives the layer reaches for, and checks that each error path gives back precisely what it took.

Test	Covers
`txfr_queue_create_test`	`xQueueCreate`, `xQueueCreateStatic`, `vQueueDelete`
`txfr_task_create_test`	`xTaskCreate`, `xTaskCreateStatic`
`txfr_sync_create_test`	semaphores, mutexes, event groups and timers

How the fault injection works

A test asks the harness to fail a chosen kernel creation call, then reads back how many allocations, releases, object creations and object deletions the layer performed. The interception is done with the linker's --wrap option, so tx_freertos.c is compiled exactly as it ships, with no test hooks in it.

Two things are worth knowing before adding tests:

tx_api.h maps the public API onto the error checking entry points, so the symbols that exist at link time are the _txe_ variants, and those are what the wrap list in cmake/regression/CMakeLists.txt names. A --wrap for a name that does not resolve is silently ignored, so a typo there produces a test that quietly never injects anything.
txfr_malloc() and txfr_free() cannot be wrapped, because they are defined in tx_freertos.c and called from within it, so the compiler resolves those calls internally. The byte pool counts stand in for them.

Because --wrap is a GNU ld and lld feature with no MSVC equivalent, this suite is Linux only. The CMake configuration stops with a clear message rather than failing later with confusing link errors.

Fixtures

fixtures/FreeRTOSConfig.h configures the layer for the tests. Two of its settings are not arbitrary:

configASSERT() and TX_FREERTOS_ASSERT_FAIL() are empty, since the tests drive error paths deliberately and neither may halt the run.
portDISABLE_INTERRUPTS() and portENABLE_INTERRUPTS() are defined up front. FreeRTOS.h picks those by compiler rather than by target, so a GNU build otherwise resolves them to the bare metal __disable_interrupts() intrinsic, which does not exist when the layer is hosted on Linux.

fixtures/tx_user.h supplies TX_THREAD_USER_EXTENSION, which the layer requires, as documented in the layer's own readme.md.

Why the build is 32 bit

The Linux port defines ULONG as unsigned int on x86_64, while the layer passes pointers through ULONG arguments, such as the task argument and the timer identifier. A 64 bit build truncates those pointers, which the compiler reports as -Wpointer-to-int-cast and which crashes the timer callback wrapper. The ThreadX and SMP suites build 32 bit for their own reasons; this suite has to.