coverage-progress.md

Coverage Progress

Authoritative flow:

1. dbus-unit-test.py
2. unit-test.py -g
3. meson test -C build/coverage --print-errorlogs
4. ninja -C build/coverage coverage-html

Format: lines / functions / branches / decisions

Run Log

2026-03-25 baseline after fixing unit-test.py coverage reuse

• common: 95.3 / 98.1 / 82.3 / 76.2
• fw-update: 95.8 / 96.1 / 82.9 / 88.0
• libpldmresponder: 96.5 / 99.4 / 86.1 / 77.5
• oem: 98.6 / 100.0 / 88.0 / 90.8
• platform-mc: 99.2 / 90.9 / 88.1 / 94.7
• pldmd: 100.0 / 75.0 / 90.0 / -
• pldmtool: 97.4 / 100.0 / 90.1 / 81.4
• requester: 94.7 / 96.0 / 86.3 / 80.0
• Whole repo: 96.9 / 96.8 / 86.3 / 86.1

2026-03-25 after requester coverage batch 1

• common: 95.3 / 98.1 / 82.3 / 76.2
• fw-update: 95.8 / 96.1 / 82.8 / 88.0
• libpldmresponder: 96.5 / 99.4 / 86.1 / 77.5
• oem: 98.6 / 100.0 / 88.0 / 90.8
• platform-mc: 99.2 / 90.9 / 88.1 / 94.7
• pldmd: 100.0 / 75.0 / 90.0 / -
• pldmtool: 97.4 / 100.0 / 90.1 / 81.4
• requester: 94.9 / 96.1 / 86.6 / 80.4
• Whole repo: 96.9 / 96.8 / 86.3 / 86.2

Batch summary:

• Added requester-only coverage tests in requester/test/handler_coverage_test.cpp for:
  • queue-empty and already-active pollEndpointQueue() short-circuits
  • missing-endpoint unregisterRequest() for uncovered handler instantiations
  • active-request unregisterRequest() promoting the next queued request
  • queued-request search paths for MockRequest and FailingSendRequest
  • RX-direction storeTransportError() selection of the affected EID

2026-03-25 after requester coverage batch 2

• common: 95.3 / 98.1 / 82.4 / 76.2
• fw-update: 95.8 / 96.1 / 82.8 / 88.0
• libpldmresponder: 96.5 / 99.4 / 86.1 / 77.5
• oem: 98.6 / 100.0 / 88.2 / 91.2
• platform-mc: 99.2 / 90.9 / 88.1 / 94.7
• pldmd: 100.0 / 75.0 / 90.0 / -
• pldmtool: 97.1 / 100.0 / 89.7 / 81.4
• requester: 95.9 / 96.2 / 88.2 / 80.0
• Whole repo: 97.0 / 96.8 / 86.4 / 86.2

Batch summary:

• Added requester transport-error filter tests in requester/test/handler_coverage_test.cpp for explicit PLDM-type matches and 0xFF wildcard matches across NiceMock<MockRequest>, Request, MockRequest, and FailingSendRequest.
• Added the remaining unknown-key instanceIdExpiryCallBack() assertion coverage for Request and NiceMock<MockRequest> in requester/test/handler_coverage_test.cpp.
• Fixed the pldmtool CI D-Bus-path tests in pldmtool_cmd_helper_test.cpp so they use local mapper/object-manager services instead of poisoning the shared session bus with missing-service error paths.
• Reduced runtime in the slow requester timeout tests in handler_test.cpp so handler_test stays below the Meson 30s timeout in the authoritative CI run.

2026-03-26 after common/fw-update/pldmd coverage batch

• common: 99.5 / 100.0 / 83.8 / 82.1
• fw-update: 97.0 / 96.9 / 84.2 / 88.9
• libpldmresponder: 96.5 / 99.4 / 86.4 / 77.5
• oem: 99.4 / 100.0 / 89.1 / 91.2
• platform-mc: 99.2 / 91.2 / 86.7 / 94.4
• pldmd: 94.0 / 95.2 / 79.5 / 83.8
• pldmtool: 97.4 / 100.0 / 90.0 / 79.5
• requester: 98.6 / 100.0 / 90.5 / 85.0
• Whole repo: 97.8 / 98.0 / 86.5 / 86.7

Batch summary:

• Added more common coverage in common/test/pldm_utils_test.cpp for:
  • DBusHandler::getDbusProperty<T>() association success coverage
  • remaining scalar and association mismatch branches for both mocked and direct template calls
  • direct checkForFruPresence() success-path D-Bus decoding coverage
• Added mock-mode async D-Bus constructor coverage in common/test/dbus_async_utils_mock_coverage_test.cpp across the remaining stored property types with custom services.
• Added fw-update timer/deferred-response coverage in fw-update/test/component_updater_test.cpp for:
  • onResponseSendComplete() no-op and send-failure paths
  • createRequestFwDataTimer() pending-discovery early return
  • createCompleteCommandsTimeoutTimer() RequestFirmwareData state and pending-discovery early return
• Focused validation passed for pldm_utils_test, dbus_async_utils_mock_coverage_test, component_updater_test, and pldmd_internal_test.
• Authoritative run passed 68/68 tests through unit-test.py -g, but the directory branch totals moved only slightly, so the next batch needs to chase larger uncovered file clusters rather than more low-level template branches.

2026-03-26 latest authoritative baseline

• common: 99.5 / 100.0 / 83.3 / 82.1
• fw-update: 97.5 / 96.9 / 85.1 / 89.7
• libpldmresponder: 96.6 / 99.4 / 86.5 / 77.5
• oem: 99.4 / 100.0 / 89.1 / 91.2
• platform-mc: 99.2 / 91.2 / 86.6 / 94.4
• pldmd: 94.4 / 95.2 / 79.9 / 85.4
• pldmtool: 97.4 / 100.0 / 90.0 / 79.5
• requester: 98.6 / 100.0 / 90.5 / 85.0
• Whole repo: 97.9 / 98.0 / 86.7 / 86.9

Batch summary:

• Added more common async D-Bus coverage in common/test/dbus_async_utils_test.cpp, then removed invalid exception-path attempts that were terminating because the exercised helpers are noexcept or only hold references.
• Added a real D-Bus positive-path inventory status test in fw-update/test/inventory_manager_internal_test.cpp for InventoryManager::logDeviceStatusErrors().
• Focused validation passed for dbus_async_utils_test, inventory_manager_internal_test, and manager_internal_test.
• Authoritative run passed 68/68 tests through unit-test.py -g, and the next batch is targeting the highest-ROI remaining branch clusters rather than continuing on low-yield template branches.

2026-03-26 authoritative baseline after workspace-temp cleanup

• common: 99.5 / 100.0 / 84.2 / 82.1
• fw-update: 97.6 / 96.9 / 85.3 / 90.3
• libpldmresponder: 96.6 / 99.4 / 86.5 / 77.5
• oem: 99.4 / 100.0 / 90.1 / 90.8
• platform-mc: 99.3 / 91.2 / 86.2 / 94.4
• pldmd: 94.6 / 95.2 / 79.9 / 85.4
• pldmtool: 97.4 / 100.0 / 90.0 / 79.5
• requester: 98.6 / 100.0 / 90.5 / 85.0
• Whole repo: 97.9 / 98.0 / 86.7 / 87.0

Batch summary:

• Updated openbmc-build-scripts/scripts/unit-test.py so Meson coverage subprocesses use an absolute temp root under build/coverage/.tmp, plus TEST_TMPDIR and GTEST_TMP_DIR, instead of the broken container /tmp.
• Added shared test temp helpers in test/test_tmp_utils.hpp and migrated the affected test scaffolding and wrappers away from hardcoded /tmp paths in:
  • test/test_instance_id.hpp
  • test/pldmd_internal_test.cpp
  • test/pldmd_main_coverage_helper.cpp
  • pldmtool/test/meson.build
  • pldmtool/test/pldmtool_base_cmd_test.cpp
  • pldmtool/test/pldmtool_cmd_helper_test.cpp
  • pldmtool/test/pldmtool_fw_update_cmd_test.cpp
  • pldmtool/test/pldmtool_main_test.cpp
  • pldmtool/test/pldmtool_platform_cmd_test.cpp
  • libpldmresponder/test/libpldmresponder_fru_test.cpp
  • libpldmresponder/test/libpldmresponder_pdr_effecter_test.cpp
  • libpldmresponder/test/libpldmresponder_bios_config_test.cpp
  • libpldmresponder/test/libpldmresponder_platform_test.cpp
  • libpldmresponder/test/bios_handler_test_config.hpp
  • libpldmresponder/test/libpldmresponder_bios_handler_test.cpp
  • platform-mc/test/oem_events_private_test.cpp
  • fw-update/test/other_device_update_manager_test.cpp
  • common/test/readHostEID_test.cpp
  • common/test/pldm_utils_test.cpp
• Focused validation passed for the previously failing target set, then the authoritative unit-test.py -g run passed 68/68 and regenerated the report successfully.

Notes

• pldmtool_cmd_helper_test was marked non-parallel in pldmtool/test/meson.build because the test shares D-Bus state and was flaky under parallel Meson execution during the coverage run.
• unit-test.py was updated to reconfigure and clean the existing build/coverage tree before the coverage run so stale outputs from removed sources do not poison gcovr input.

2026-03-26 authoritative run after common/libpldmresponder/pldmd helper batch

• common: 99.5 / 100.0 / 84.5 / 82.1
• fw-update: 97.6 / 96.9 / 85.3 / 90.6
• libpldmresponder: 96.9 / 99.4 / 87.0 / 78.7
• oem: 99.4 / 100.0 / 90.1 / 90.8
• platform-mc: 99.3 / 93.4 / 86.2 / 94.4
• pldmd: 94.6 / 95.2 / 80.0 / 85.4
• pldmtool: 97.4 / 100.0 / 90.0 / 79.5
• requester: 98.6 / 100.0 / 90.5 / 85.0
• Whole repo: 98.0 / 98.4 / 86.9 / 87.3

Batch summary:

• Added common async D-Bus coverage in:
  • common/test/dbus_async_utils_test.cpp
  • common/test/dbus_async_utils_mock_coverage_test.cpp
• Added pldmd subprocess and failure-hook coverage in:
  • test/pldmd_internal_test.cpp
  • test/pldmd_main_coverage_helper.cpp
• Added libpldmresponder coverage in:
  • libpldmresponder/test/libpldmresponder_base_test.cpp
  • libpldmresponder/test/libpldmresponder_fru_test.cpp
  • libpldmresponder/test/libpldmresponder_platform_test.cpp
• Removed three libpldmresponder_platform_test assertions that depended on the known unreachable OEM getStateSensorReadings() wrapper path documented below in the production-bug notes. The authoritative CI run passed 68/68 after that correction.

Production Bugs To Revisit

libpldmresponder/platform.cpp request length mismatch

• Function: Handler::getStateSensorReadings()
• Current behavior:
  • outer guard checks payloadLength != PLDM_GET_SENSOR_READING_REQ_BYTES
  • decode path calls decode_get_state_sensor_readings_req()
  • decoder expects PLDM_GET_STATE_SENSOR_READINGS_REQ_BYTES
• Result:
  • success path is unreachable for valid GetStateSensorReadings decoding
  • tests can only hit invalid-length behavior unless production code is corrected
• This was not changed during the coverage work because you requested no production-code edits.