Chet Powers

This commit **adds support for the PRBS31 polynomial** within the **`ChenabAsic`** hardware abstraction layer. It enhances the `getSupportedPrbsPolynomials` function in `fboss/agent/hw/switch_asics/ChenabAsic.cpp` to include this new option. This is a **new capability** that expands the range of pseudo-random binary sequence (PRBS) tests available for the Chenab ASIC, improving its diagnostic and testing functionalities.

This commit introduces a **new `PagingSupport` enum** and integrates it into the **`TcvrState` struct** within the `transceiver.thrift` interface definition, enhancing the **transceiver information model**. This **new feature** allows explicit indication of whether a module supports memory paging. The `QsfpModule.cpp` implementation is updated to populate this new `pagingSupport` field during `updateCachedTransceiverInfoLocked`, reflecting the module's actual capabilities. This addition provides crucial information that will be leveraged **later in the stack** to conditionally validate `portToMediaLane` mappings, preventing incorrect assertions on non-paged modules.

This commit **fixes a testing infrastructure issue** by **relocating the SDK state dumping logic** within the **AgentEnsembleTest** framework. Previously, the SDK debug dump, triggered by `FLAGS_enable_sdk_dump`, was performed in `AgentEnsembleTest::TearDown`, which was bypassed during **warmboot test scenarios** where the process exits early. By moving this logic to the `tearDownAgentEnsemble` method, the SDK state will now be consistently dumped, significantly improving **debuggability and testability** for warmboot-enabled tests. This ensures critical diagnostic information is always captured, regardless of the test exit path.

This commit **enhances the debuggability of FBOSS agent link tests** by introducing logging of all environment variables at the very beginning of test execution. Specifically, the `agentEnsembleLinkTestMain` function within the `fboss/agent/test/link_tests` module now prints the environment, providing immediate context for test failures or unexpected behavior. This **testing utility improvement** also includes adding a `//folly/system:env_util` dependency to the `BUCK` file to support the environment variable utilities. The change primarily impacts the **FBOSS agent's testing infrastructure**, making it easier to diagnose issues related to test environment configurations and improving overall **test observability**.

This commit **refactors** the **transceiver snapshot management logic**, relocating its responsibility from `QsfpModule` to the **`TransceiverManager`**. This **feature enhancement** enables the system to **generate and store snapshots for transceivers that are absent or experiencing I2C communication errors**, a critical capability previously unavailable as `QsfpModule` only instantiated for present and readable devices. The **`TransceiverManager`** now implements new snapshot management logic, including methods like `updateSnapshots`, and refines error handling to update `timeCollected` on communication failures. This change directly supports **Next-Gen Transceiver (NGT)** requirements by providing more comprehensive monitoring of transceiver states, regardless of their operational status.

This commit **improves debuggability** within the **FBOSS agent's port state management** by enhancing logging. Specifically, it **modifies the `getInterfaceID` function** in `fboss/agent/state/Port.cpp` to include the **port name** in the log message. This **maintenance change** provides crucial context when the `intf.size()` check fails, making it significantly easier to diagnose issues. The added detail will streamline troubleshooting during Network Provisioning and Installation (NPI) events, helping to quickly pinpoint the cause of interface ID retrieval failures.

This commit introduces a **bug fix** within the **QSFP service** specifically for **Wedge platforms**, ensuring robust handling of transceiver management interfaces. It addresses an issue where the `mgmtInterface` variable could remain uninitialized if an exception occurred during the `getTransceiverManagementInterface` call. By explicitly initializing `mgmtInterface` to `NONE` in `WedgeQsfp.cpp`, the system now guarantees a defined value, even in error scenarios. This **maintenance fix** prevents potential crashes or undefined behavior that could arise from attempting to use an uninitialized variable, thereby improving the stability of transceiver operations.

This commit **adjusts build system dependencies** within the **`fboss/qsfp_service` module** by adding `@manual` annotations to specific targets across its `BUCK` files. This **maintenance fix** prevents automatic dependency resolution tools, like `arc lint`, from incorrectly re-linking existing `qsfp_service` components from the **Credo SDK** to the newly introduced PAI SDK. By explicitly marking these dependencies as manual, the commit ensures that platform-specific and test targets continue to consistently link against the Credo SDK, avoiding potential build issues or runtime mismatches. Additionally, `boost_assign` is now explicitly exported as an external dependency for the `fboss/qsfp_service/module`.

This commit **conditionally skips** the `opticsVdmPerformanceMonitoring` test within the `AgentEnsembleOpticsTest.cpp` suite for **Open Source Software (OSS) builds**. This **maintenance change** addresses a known incompatibility where the underlying `StatsPublisher` functionality, essential for collecting the required performance statistics, is currently stubbed out in OSS environments. By preventing this test from executing in unsupported configurations, it **reduces noise and false failures**, thereby improving the stability and clarity of debugging efforts for NPI link tests within the **FBOSS agent's test infrastructure**.

This commit **standardizes the generation of platform mapping files** within `fboss/lib/platform_mapping_v2` to consistently include a newline character at the end of each file. This **maintenance fix** addresses a conflict where pre-commit hooks required newlines, but the `test_generated_files_match` test failed if they were present. By modifying the `generate_platform_mappings` function in `fboss/lib/platform_mapping_v2/gen.py` to always append a newline, this change resolves CI failures. It ensures all **generated platform mapping JSON files** conform to a consistent style, improving overall code hygiene and preventing future formatting conflicts.

This commit introduces a **new capability** to accurately report **transceiver communication errors** within the `qsfp_service` subsystem. Previously, the `TransceiverInfo` could misrepresent module presence or hide ongoing `I2c` communication failures, making it difficult to diagnose issues. Now, the `TransceiverManager` and `QsfpModule` components track communication success, setting a `communicationError` flag in `TransceiverInfo` if issues are detected since the last successful refresh. This significantly improves the **diagnosability** and **reliability** of transceiver status reporting, ensuring that users can correctly identify modules experiencing communication problems.

This commit **fixes a service management issue** by **enhancing the `restartQsfpService` utility function** within `fboss/agent/test/link_tests/LinkTestUtils.cpp`. Previously, the utility failed to correctly restart the **QSFP service** in **Open Source Software (OSS) builds**, leading to test failures. The update adds logic to properly identify and restart the `qsfp_service_oss` daemon, ensuring robust and reliable testing for **FBOSS agent link tests** across different build environments. This **maintenance improvement** prevents future test failures related to the QSFP service not restarting correctly in OSS environments.

This commit delivers a **bug fix** for the **`fboss2` CLI's `show transceiver` command**, resolving an issue where multi-port transceivers were incorrectly displaying inactive interfaces as bypass modules. The logic within `queryClient` is refined to accurately identify a transceiver as a **bypass module** only if *none* of its interfaces have an associated agent port, and to skip any unconfigured interfaces on non-bypass multi-port optics. This ensures that network operators receive correct and clear information about **transceiver configurations**, preventing misinterpretation of module status, and is validated by a new test case.

This commit introduces a **new capability** to explicitly track communication failures with optical transceiver modules. It adds a `communicationError` boolean field to the `TcvrState` Thrift struct within the **QSFP service**'s data model, allowing the system to indicate when a module is unreachable. The change also updates the **FSDB** generated Thrift path model (`fsdb_model_thriftpath.h`) to include definitions for this new field. This enhancement improves the **diagnosability** of optical hardware by providing clear visibility into transceiver communication issues for monitoring and reporting.

This commit **fixes critical bugs** in the **`fboss/cli/fboss2` `show transceiver` command** related to **port filtering** and **bypass module detection**. Previously, filtering by a specific port would incorrectly display all entries for a transceiver and could misclassify multi-port modules as bypass modules if their associated ports were filtered out. The **refactoring of transceiver data querying and filtering logic** within `fboss/cli/fboss2/commands/show/transceiver/CmdShowTransceiver.h` now ensures that `show transceiver` accurately filters output and correctly identifies bypass modules. **New unit tests** have been added to `fboss/cli/fboss2/test/CmdShowTransceiverTest.cpp` to validate these fixes and prevent future regressions, ensuring reliable transceiver status reporting.

This commit **enhances the `fboss2` CLI's `show transceiver` command** to correctly display information for **bypass modules**, which previously were not fully represented due to their lack of an associated ASIC port. The command now iterates directly over transceiver entries instead of relying on port entries, and introduces a 'Bypass' status for transceivers without a direct port association. This **feature enhancement** and **data model update** (making `isUp` optional in `TransceiverDetail`) improves **network visibility** by ensuring all transceiver types, including bypass modules, are accurately reported in the CLI.

This commit **fixes a test failure** in the **`AgentEnsemblePhyInfoTest`** that occurred on mixed Elbert platforms. Previously, the test incorrectly asserted that all ports would have external PHY (xphy) information, leading to failures when some PIMs lacked xphy. The **test logic is now refined** to only check for xphy information on ports that actually possess an xphy, aligning with the approach used in FSDB tests. This is achieved by introducing a new helper function `getXphyCabledPorts` in `fboss/agent/test/link_tests/AgentEnsembleLinkTest.h` and updating `xPhyInfoTest` in `fboss/agent/test/link_tests/AgentEnsemblePhyInfoTest.cpp` to use it, along with an assertion to ensure at least one xphy port is present. This makes the **xphy information test more robust and accurate** for diverse hardware configurations.

This commit introduces a **new capability** to the `fboss/agent` **test infrastructure** by adding an `enable_sdk_dump` flag. This flag, integrated into the `TearDown` method of `AgentEnsembleTest.cpp`, allows for the generation of **SDK debug dumps** even when tests pass. The primary purpose is to facilitate **debugging warmboot test failures** by collecting both coldboot and warmboot state, even if the coldboot version of the test succeeds. This diagnostic feature is only active when the `--enable_debug_dumps` flag is passed to the `netcastle CLI`, providing more robust **diagnostic data collection** for complex test scenarios.

This commit **fixes a platform configuration error** within the **FBOSS agent's SAI platform layer** by correcting the `supportsTransceiver` method in `fboss/agent/platforms/sai/SaiMinipack3NPlatformPort.cpp` to return `true`. Previously, this method incorrectly returned `false`, leading to issues where the `SaiPlatformPort::getTransceiverMapping` would return an empty index. This **correction** ensures that the **Minipack3N platform** is properly recognized as supporting transceivers, resolving a failing `getTransceivers` test case and enabling accurate transceiver detection and mapping for this hardware.

This commit **refactors** the codebase by **relocating FEC-related utility functions**, including `get_fec_mode` and its dependencies, from the `phy_checks` module to the more general `snapshot_util` module. This **maintenance** change enables **code reuse**, allowing these functions to be leveraged by other modules like `optics_checks` for verifying FEC tails. The move centralizes common functionality, improving modularity and preventing duplication across different diagnostic checks within the **QSFP service**. Additionally, minor documentation updates were made to the `VdmDiagsStats` struct in `transceiver.thrift`.