Skip to content

patina-v15.0.0

Choose a tag to compare

View all tags
@github-actions github-actions released this 19 Nov 02:13
· 482 commits to refs/heads/main since this release
patina-v15.0.0
729365b
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

What's Changed

⚠️ Breaking Changes

  • Refactor Patina Perf MM Code and Use the MM Comm Service [Rebase \& FF] @makubacki (#746)
    Change Details
      ## Description

    Closes #444

    Refactors performance code interacting with MM to be more ergonomic and use the MM Communication service (instead of the MM Communication protocol). Use zerocopy in some places to reduce overhead from scroll. General other improvements for error handling and adding more testing.

    perf: Update dependencies for MM communication support

    Add patina_mm dependency to workspace and performance component to
    enable MM communication. Replace scroll with zerocopy for improved
    serialization performance and reduced dependency weight.

    • Add patina_mm dependency to workspace Cargo.toml
    • Add patina_mm and zerocopy dependencies to performance component
    • Remove scroll dependency in favor of zerocopy for serialization
    • Add zerocopy with derive features to SDK

    This establishes the foundation for modernizing MM communication in the
    performance component by replacing HOB-based communication with a
    service-based approach.

    patina_performance: Add MM module

    Add a dedicated MM communication module for performance related
    MM definitions and functionality such as record fetching.

    • SmmCommHeader is zerocopy wire format compatible
    • Adds GetRecordSize for querying total performance record size
    • Adds GetRecordDataByOffset for chunked record retrieval wit
      configurable buffer sizes

    patina_sdk/performance: Improve perf record serialization infra

    • Adds the PerformanceRecordHeader struct to represent and manage
      performance record headers.
    • Replaces scroll::Error with Error::Serialization so scroll errror
      types are not leaking into performance code.
    • Implement direct little-endian serialization helpers for certain
      primitives
    • Favors zerocopy instead of scroll in relevant places.
    • Adds improved bounds checking for serialization operations.

    patina_performance: Update configuration with explicit defaults

    Enhance PerfConfig with explicit default values and improved
    documentation to provide clearer configuration semantics.

    • Add explicit DEFAULT_ENABLE_COMPONENT and
      DEFAULT_ENABLED_MEASUREMENTS constants
    • Implement custom Default trait instead of derive to better
      control defaults
    • Improve documentation
    • Improve PerfConfig usage ergonomics

    patina_performance: Use the MM Communication service

    Drop HOB-based MM comm buffers and the Communicate protocol being
    used in the performance component to instead use the Communicate
    service.

    Implements more robust error handling, general code cleanup, and
    additional tests.

    • Uses the MmCommunication service

    • Adds better error handling using the MmPerformanceError enum

    • Implements chunked MM record fetching with configurable buffer
      sizes

    • Adds PerformanceRecordIterator for more memory-efficient perf
      record processing

    • Adds unit tests covering different scenarios such as zero records,
      single record, and multi-chunk scenarios

    • Mocks the MmCommunication service implementations to better
      isolate testing

    patina_sdk: Refactor perf measurement module to drop MM comm protocol

    Remove protocol-based MM communication in the patina_sdk perf code
    and clean up the event_callback namespace.

    • Remove Vec import as it's no longer needed for temporary buffers
    • Remove fetch_and_add_mm_performance_records from event_callback
      module
    • Clean up unused imports and simplify module dependencies

    patina_sdk/performance: Remove temp SMM module

    Remove the temporary _smm module that provided scroll-based MM
    communication.

    The new MM communication approach in the mm.rs module provides
    the same functionality with improved error handling, better
    resource management, and integration with the Patina service
    architecture.

    patina_performance: Add records module

    Adds structures that represent performance records so they can
    be parsed and printed in debug messages.

    patina_mm: Add MM communication buffer update protocol

    Implements the MM communication buffer update protocol to support`
    dynamic updates to MM communication buffers during boot. This is
    an option that can be enabled in the MM Comm buffer config. The
    protocol is currently published the Mu MM Supervisor Standalone MM
    implementation.

    High-level changes:

    • Add support for the MmCommBufferUpdate protocol
    • Update the patina_mm component to support MM Comm buffer updates
    • Update patina_mm config to support MM Comm buffer updates
    • Integration with SW MMI manager for buffer update handling
    • Component integration tests for buffer update functionality

    An example of a pending MM Communication buffer being applied when
    the perf component triggers a SW MMI:

    DEBUG - Starting MM communication: buffer_id=1, data_size=40, recipient=C095791A-3001-47B2-80C9-EAC7319F2FA4
INFO - Pending buffer update detected, applying now
INFO - Adding new comm buffer 1: addr=0x7ddd9000, size=0x10000
INFO - Successfully applied pending comm buffer 1 update
TRACE - patina_mm\src\component\communicator.rs:466: Buffer validation: buffer_len=65536, required_len=64
TRACE - patina_mm\src\component\communicator.rs:473: Resetting the comm buffer and internal tracking state
TRACE - patina_mm\src\component\communicator.rs:476: Setting up communication buffer for MM request
TRACE - patina_mm\src\config.rs:569: Setting message info for buffer 1: recipient=C095791A-3001-47B2-80C9-EAC7319F2FA4
TRACE - patina_mm\src\config.rs:540: Validating capacity for buffer 1: buffer_size=65536, message_size=0
TRACE - patina_mm\src\config.rs:558: Buffer 1 capacity validation passed
TRACE - patina_mm\src\config.rs:527: Buffer 1 state consistency was verified successfully
TRACE - patina_mm\src\config.rs:586: Message info set successfully for buffer 1
TRACE - patina_mm\src\config.rs:598: Setting message for buffer 1: message_size=40
TRACE - patina_mm\src\config.rs:540: Validating capacity for buffer 1: buffer_size=65536, message_size=40
TRACE - patina_mm\src\config.rs:558: Buffer 1 capacity validation passed
TRACE - patina_mm\src\config.rs:610: Buffer 1: writing header and message data
TRACE - patina_mm\src\config.rs:527: Buffer 1 state consistency was verified successfully
DEBUG - Buffer 1 message set successfully: header_size=24, message_size=40
DEBUG - Outgoing MM communication request: buffer_id=1, data_size=40, recipient=C095791A-3001-47B2-80C9-EAC7319F2FA4
DEBUG - Request Data (hex): [01, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00]
TRACE - patina_mm\src\component\communicator.rs:488: Comm buffer before request: CommunicateBuffer(id: 0x1. len: 0x10000)
. . .

    Add additional MM Communicator unit tests

    Coverage for:

    • Debug output
    • Default trait implementation
    • Using an MM executor

    Improve MM perf record debuggability

    Prints additional information about MM performance records retrieved,
    including record-specific information. That allows better visibility
    into the records themselves and MM communication between the Patina
    Performance component the MM performance code.

    An example of two MM performance records received from the MM UEFI
    variable driver:

    DEBUG - MM communication response received: size=156
INFO - Performance: Processing 116 bytes of MM performance data
DEBUG - Performance: MM record #1 - type: 0x1011 (Dynamic String Event), length: 58, revision: 1, data_len: 54
DEBUG -   Record #1: progress_id: 0x0030, apic_id: 0, timestamp: 4105960673, guid: 7EE2C0C1-C21A-4113-A53A-66824A95696F
DEBUG -     String: "MmVariableServiceInitia"
DEBUG - Performance: MM record #2 - type: 0x1011 (Dynamic String Event), length: 58, revision: 1, data_len: 54
DEBUG -   Record #2: progress_id: 0x0031, apic_id: 0, timestamp: 4157443753, guid: 7EE2C0C1-C21A-4113-A53A-66824A95696F
DEBUG -     String: "MmVariableServiceInitia"
INFO - Performance: MM record summary - total: 2, added: 2, failed: 0
    • Impacts functionality?
    • Impacts security?
    • Breaking change?
    • Includes tests?
    • Includes documentation?

    How This Was Tested

    • cargo make all
    • Boot on QEMU Q35 with default Q35 build and run settings
    • Boot on QEMU Q35 with PERF_TRACE_ENABLE
    • Add two MM perf records for testing and verify they are found using MM communication
    • Verify the MM communicate buffer can be updated successfully when the MM Communication Buffer update protocol is published and the following performance configuration is used:
      enable_comm_buffer_updates: true,
  updatable_buffer_id: Some(1),

    Integration Instructions

    Update performance configuration to include the following fields:

      enable_comm_buffer_updates: false,
  updatable_buffer_id: None,

    If the MM Communication Buffer Update protocol is installed on the platform (protocol with GUID 2a22e38f-9d1c-49d0-bdce-7ddac16da45d), then use the following configration:

      enable_comm_buffer_updates: true,
  updatable_buffer_id: Some(1),

    Where updatable_buffer_id is the ID of the comm buffer that should be updated with the information in the protocol.



🚀 Features & ✨ Enhancements

  • Refactor TplMutex @joschock (#1056)
    Change Details
      ## Description

    Renames the tpl_lock.rs module to tpl_mutex and refactors the logic. Key improvements:

    • Remove the dependency on BootServices for TPL manipulation. TplMutex now directly calls the raise_tpl/restore_tpl APIs directly rather than through the BootServices table. This greatly simplifies the usage model and testing of TplMutex.

    Also updates Synchronization chapter of the docbook.

    This change pursuant to RFC 0021 #1036

    • Impacts functionality?
    • Impacts security?
    • Breaking change?
    • Includes tests?
    • Includes documentation?

    How This Was Tested

    Boot testing on hardware platform.

    Integration Instructions

    N/A

      </blockquote>
  <hr>
</details>
  • Always enable GICv4 support. @joschock (#1074)
    Change Details
      ## Description

    This PR always enables GICv4 support in the arm-gic crate. GicV4 support is backwards compatible with GicV3, and is necessary for properly calculating the GIC Redistributor register offsets.

    • Impacts functionality?
    • Impacts security?
    • Breaking change?
    • Includes tests?
    • Includes documentation?

    How This Was Tested

    Booted on hardware with GicV4, observed that GIC redistributor accesses are no longer incorrect.

    Integration Instructions

    N/A - changes are backward compat with GicV3, though the expectation is that most systems are GicV4.

      </blockquote>
  <hr>
</details>

📖 Documentation Updates

  • Refactor TplMutex @joschock (#1056)
    Change Details
      ## Description

    Renames the tpl_lock.rs module to tpl_mutex and refactors the logic. Key improvements:

    • Remove the dependency on BootServices for TPL manipulation. TplMutex now directly calls the raise_tpl/restore_tpl APIs directly rather than through the BootServices table. This greatly simplifies the usage model and testing of TplMutex.

    Also updates Synchronization chapter of the docbook.

    This change pursuant to RFC 0021 #1036

    • Impacts functionality?
    • Impacts security?
    • Breaking change?
    • Includes tests?
    • Includes documentation?

    How This Was Tested

    Boot testing on hardware platform.

    Integration Instructions

    N/A

      </blockquote>
  <hr>
</details>
  • Refactor Patina Perf MM Code and Use the MM Comm Service [Rebase \& FF] @makubacki (#746)
    Change Details
      ## Description

    Closes #444

    Refactors performance code interacting with MM to be more ergonomic and use the MM Communication service (instead of the MM Communication protocol). Use zerocopy in some places to reduce overhead from scroll. General other improvements for error handling and adding more testing.

    perf: Update dependencies for MM communication support

    Add patina_mm dependency to workspace and performance component to
    enable MM communication. Replace scroll with zerocopy for improved
    serialization performance and reduced dependency weight.

    • Add patina_mm dependency to workspace Cargo.toml
    • Add patina_mm and zerocopy dependencies to performance component
    • Remove scroll dependency in favor of zerocopy for serialization
    • Add zerocopy with derive features to SDK

    This establishes the foundation for modernizing MM communication in the
    performance component by replacing HOB-based communication with a
    service-based approach.

    patina_performance: Add MM module

    Add a dedicated MM communication module for performance related
    MM definitions and functionality such as record fetching.

    • SmmCommHeader is zerocopy wire format compatible
    • Adds GetRecordSize for querying total performance record size
    • Adds GetRecordDataByOffset for chunked record retrieval wit
      configurable buffer sizes

    patina_sdk/performance: Improve perf record serialization infra

    • Adds the PerformanceRecordHeader struct to represent and manage
      performance record headers.
    • Replaces scroll::Error with Error::Serialization so scroll errror
      types are not leaking into performance code.
    • Implement direct little-endian serialization helpers for certain
      primitives
    • Favors zerocopy instead of scroll in relevant places.
    • Adds improved bounds checking for serialization operations.

    patina_performance: Update configuration with explicit defaults

    Enhance PerfConfig with explicit default values and improved
    documentation to provide clearer configuration semantics.

    • Add explicit DEFAULT_ENABLE_COMPONENT and
      DEFAULT_ENABLED_MEASUREMENTS constants
    • Implement custom Default trait instead of derive to better
      control defaults
    • Improve documentation
    • Improve PerfConfig usage ergonomics

    patina_performance: Use the MM Communication service

    Drop HOB-based MM comm buffers and the Communicate protocol being
    used in the performance component to instead use the Communicate
    service.

    Implements more robust error handling, general code cleanup, and
    additional tests.

    • Uses the MmCommunication service

    • Adds better error handling using the MmPerformanceError enum

    • Implements chunked MM record fetching with configurable buffer
      sizes

    • Adds PerformanceRecordIterator for more memory-efficient perf
      record processing

    • Adds unit tests covering different scenarios such as zero records,
      single record, and multi-chunk scenarios

    • Mocks the MmCommunication service implementations to better
      isolate testing

    patina_sdk: Refactor perf measurement module to drop MM comm protocol

    Remove protocol-based MM communication in the patina_sdk perf code
    and clean up the event_callback namespace.

    • Remove Vec import as it's no longer needed for temporary buffers
    • Remove fetch_and_add_mm_performance_records from event_callback
      module
    • Clean up unused imports and simplify module dependencies

    patina_sdk/performance: Remove temp SMM module

    Remove the temporary _smm module that provided scroll-based MM
    communication.

    The new MM communication approach in the mm.rs module provides
    the same functionality with improved error handling, better
    resource management, and integration with the Patina service
    architecture.

    patina_performance: Add records module

    Adds structures that represent performance records so they can
    be parsed and printed in debug messages.

    patina_mm: Add MM communication buffer update protocol

    Implements the MM communication buffer update protocol to support`
    dynamic updates to MM communication buffers during boot. This is
    an option that can be enabled in the MM Comm buffer config. The
    protocol is currently published the Mu MM Supervisor Standalone MM
    implementation.

    High-level changes:

    • Add support for the MmCommBufferUpdate protocol
    • Update the patina_mm component to support MM Comm buffer updates
    • Update patina_mm config to support MM Comm buffer updates
    • Integration with SW MMI manager for buffer update handling
    • Component integration tests for buffer update functionality

    An example of a pending MM Communication buffer being applied when
    the perf component triggers a SW MMI:

    DEBUG - Starting MM communication: buffer_id=1, data_size=40, recipient=C095791A-3001-47B2-80C9-EAC7319F2FA4
INFO - Pending buffer update detected, applying now
INFO - Adding new comm buffer 1: addr=0x7ddd9000, size=0x10000
INFO - Successfully applied pending comm buffer 1 update
TRACE - patina_mm\src\component\communicator.rs:466: Buffer validation: buffer_len=65536, required_len=64
TRACE - patina_mm\src\component\communicator.rs:473: Resetting the comm buffer and internal tracking state
TRACE - patina_mm\src\component\communicator.rs:476: Setting up communication buffer for MM request
TRACE - patina_mm\src\config.rs:569: Setting message info for buffer 1: recipient=C095791A-3001-47B2-80C9-EAC7319F2FA4
TRACE - patina_mm\src\config.rs:540: Validating capacity for buffer 1: buffer_size=65536, message_size=0
TRACE - patina_mm\src\config.rs:558: Buffer 1 capacity validation passed
TRACE - patina_mm\src\config.rs:527: Buffer 1 state consistency was verified successfully
TRACE - patina_mm\src\config.rs:586: Message info set successfully for buffer 1
TRACE - patina_mm\src\config.rs:598: Setting message for buffer 1: message_size=40
TRACE - patina_mm\src\config.rs:540: Validating capacity for buffer 1: buffer_size=65536, message_size=40
TRACE - patina_mm\src\config.rs:558: Buffer 1 capacity validation passed
TRACE - patina_mm\src\config.rs:610: Buffer 1: writing header and message data
TRACE - patina_mm\src\config.rs:527: Buffer 1 state consistency was verified successfully
DEBUG - Buffer 1 message set successfully: header_size=24, message_size=40
DEBUG - Outgoing MM communication request: buffer_id=1, data_size=40, recipient=C095791A-3001-47B2-80C9-EAC7319F2FA4
DEBUG - Request Data (hex): [01, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00]
TRACE - patina_mm\src\component\communicator.rs:488: Comm buffer before request: CommunicateBuffer(id: 0x1. len: 0x10000)
. . .

    Add additional MM Communicator unit tests

    Coverage for:

    • Debug output
    • Default trait implementation
    • Using an MM executor

    Improve MM perf record debuggability

    Prints additional information about MM performance records retrieved,
    including record-specific information. That allows better visibility
    into the records themselves and MM communication between the Patina
    Performance component the MM performance code.

    An example of two MM performance records received from the MM UEFI
    variable driver:

    DEBUG - MM communication response received: size=156
INFO - Performance: Processing 116 bytes of MM performance data
DEBUG - Performance: MM record #1 - type: 0x1011 (Dynamic String Event), length: 58, revision: 1, data_len: 54
DEBUG -   Record #1: progress_id: 0x0030, apic_id: 0, timestamp: 4105960673, guid: 7EE2C0C1-C21A-4113-A53A-66824A95696F
DEBUG -     String: "MmVariableServiceInitia"
DEBUG - Performance: MM record #2 - type: 0x1011 (Dynamic String Event), length: 58, revision: 1, data_len: 54
DEBUG -   Record #2: progress_id: 0x0031, apic_id: 0, timestamp: 4157443753, guid: 7EE2C0C1-C21A-4113-A53A-66824A95696F
DEBUG -     String: "MmVariableServiceInitia"
INFO - Performance: MM record summary - total: 2, added: 2, failed: 0
    • Impacts functionality?
    • Impacts security?
    • Breaking change?
    • Includes tests?
    • Includes documentation?

    How This Was Tested

    • cargo make all
    • Boot on QEMU Q35 with default Q35 build and run settings
    • Boot on QEMU Q35 with PERF_TRACE_ENABLE
    • Add two MM perf records for testing and verify they are found using MM communication
    • Verify the MM communicate buffer can be updated successfully when the MM Communication Buffer update protocol is published and the following performance configuration is used:
      enable_comm_buffer_updates: true,
  updatable_buffer_id: Some(1),

    Integration Instructions

    Update performance configuration to include the following fields:

      enable_comm_buffer_updates: false,
  updatable_buffer_id: None,

    If the MM Communication Buffer Update protocol is installed on the platform (protocol with GUID 2a22e38f-9d1c-49d0-bdce-7ddac16da45d), then use the following configration:

      enable_comm_buffer_updates: true,
  updatable_buffer_id: Some(1),

    Where updatable_buffer_id is the ID of the comm buffer that should be updated with the information in the protocol.



Full Changelog: patina-v14.5.0...v15.0.0

Contributors

makubacki and joschock
Assets 2
Loading