v0.6.0

Added

• CpuInfo.cpu_stepping_int (defaults to 0). resolve() takes it as an optional arg and detect() fills it in on all three platforms
• zenmaster.init() re-exported at top level.
• smu.read_pm_table_full(family). Every backend was running the full mailbox sequence twice per read_pm_sensors() call: once for read_pm_table(), once again for read_pm_table_version(), because they were written as two separate functions that each redid the same version/address/transfer round trip. Now there's one fetch, and both older functions call into it.
• Linux ModuleStatus.reason can be not_installed or unsigned now, not just not_loaded. Previously "never installed," "installed but unsigned," and "installed but idle" all collapsed into the same reason, so anyone consuming this had to shell out to modinfo themselves to tell them apart.
• smu.close() (also at top level as zenmaster.close()) releases whatever the active backend is holding: the Windows PawnIO handle, macOS's DirectHW/IOKit connection, or the Linux SMU file descriptor. Nothing closed these before; a long-running process (a daemon that wants to drop and reacquire access) had no way to let go of them.
• More of smu's surface is re-exported at the top level: active_backend, pm_table_supported, send_mp1, send_rsmu, query_mp1, query_rsmu, read_pm_table, read_pm_table_version, read_pm_table_full. The CLI already needed all of these; now import zenmaster gets you the same surface without dropping down to zenmaster.smu.
• BackendUnavailable now appends a link to the wiki's Installation guide to whatever message it's raised with. Every "PawnIO isn't installed" / "ryzen_smu is too old" / "DirectHW isn't loaded" error now points somewhere with the actual setup steps, instead of leaving you to go find them.

Removed

• AMDFamily0F.bin, AMDFamily10.bin, AMDFamily17.bin, AMDReset.bin.

Changed

• send_arg() raises UnsupportedCPU for a family with no SMU support.
• send_arg() no longer swallows every exception as a hardware rejection. Only OSError, struct.error, and ctypes.ArgumentError count as SMU_FAILED now. A real bug in a backend used to look identical to the SMU just saying no.
• CLI shows help instead of doing nothing when you run it with no action flag and no tuning args. zenmaster --json by itself used to just exit 0 silently.
• hardware.detect() caches its result for the life of the process. It used to re-parse /proc/cpuinfo, PROCESSOR_IDENTIFIER, or sysctl and recompute the codename on every single call, which is why callers (U4L included) kept building their own memoization around it. resolve() is untouched and still takes whatever you pass it every time, since that's the one meant for feeding in values you didn't read off the real CPU.
• The mailbox send/query/poll/retry logic that macOS and Windows each had their own near-identical copy of now lives once in zenmaster/mailbox.py. Linux already shared its two backends this way; the other two platforms didn't.
• All three backends now check "has init() run yet" the same way: a bare _require_init() guard called at the top of send_mp1/send_rsmu/query_mp1/query_rsmu, raising SMUNotInitialized before any hardware I/O happens.

Fixed

• Windows: the PawnIO registry lookup (_pawnio_info) was a separate registry read for module_version(), module_version_ok(), module_status(), and is_available() each. It's read once and cached now.
• Windows: read_pm_table_full didn't check for a zero table version, unlike Linux and macOS, so a rejected version query could fall through and read from a garbage address instead of returning None.
• apply() now counts an unparseable value (--stapm-limit=notanumber) as a rejection. It already returned an error result for this case, but had_rejection stayed False, so a caller only checking that flag would think the preset applied cleanly.

v0.5.0

Added

• macOS backend (AMD Hackintosh): CPU detection via sysctl, and SMU access through DirectHW.kext, with a kext-free fallback over IOPCIBridge (tuning only, no PM table)
• CLI --iopci (macOS only) forces the IOPCIBridge path instead of DirectHW
• zenmaster.hardware.detect() reads machdep.cpu.family / machdep.cpu.model / machdep.cpu.brand_string on macOS instead of /proc/cpuinfo

Changed

• --info (text and --json) omits the Backend/Driver fields entirely when the SMU hasn't been initialized, instead of printing "not initialized"
• --table, --dump-table, and --sensors point to DirectHW.kext specifically when they fail on the IOPCIBridge path, instead of a generic "not available" message

v0.4.0

Added

• CLI --sensors (and --sensors --json): compact live readout of temp, load, socket power, iGPU clock/temp, and memory clock
• CLI --info now shows a Driver line with the driver name, version, and status (or PCI direct access); same data under --info --json
• smu.read_pm_sensors(family) reads and decodes the PM table in one call, returning a PmSensors (or None). table.read_sensors(data, ver) decodes raw bytes you already have
• smu.module_status() returns one ModuleStatus verdict for the driver (ok, version, min_version, reason). Helpers: module_version(), module_version_ok(), is_available(), secure_boot_enabled(), driver_name()
• smu.ensure_backend(): like init() but returns the backend or None instead of raising
• smu.unavailable_reason(): the message explaining why the SMU can't be used, or None when it can
• smu.send_arg(family, name, value) sends one value to every mailbox the arg maps to, returning [(mailbox, opcode, status), ...]
• resolve(name, cpu_family, cpu_model) builds a CpuInfo from explicit values without reading /proc/cpuinfo
• runner.is_supported(family)
• All of the above are re-exported at the top level, so apps don't import from zenmaster.linux

Changed

• Linux now checks the ryzen_smu version (drv_version) and requires >= 0.1.7, with the installed version in the error
• Backend selection follows Secure Boot: off uses PCI direct access, on uses ryzen_smu. A loaded module is no longer preferred over PCI when Secure Boot is off
• On Windows, the driver is PawnIO with no version requirement, and Secure Boot does not apply

v0.3.0

Added

• get-* query commands now return the value the SMU reports, shown in the CLI and in --json as a returned field; smu.query_mp1() / smu.query_rsmu() expose the raw read-back for library users
• --version flag, which prints the installed version and checks PyPI for a newer release; zenmaster.check_update() exposes the same check to library users
• Exception hierarchy for library users: ZenMasterError (base) with BackendUnavailable, SMUNotInitialized, UnsupportedCPU, all still subclass RuntimeError, so existing except RuntimeError keeps working
• smu.SmuStatus IntEnum for SMU status codes (SMU_OK etc. now alias it)
• ApplyResult TypedDict documenting the apply() result contract
• py.typed marker so downstream type-checkers use the bundled type hints
• PM-table support for the Raven/Picasso-era APUs (RavenRidge, Picasso, Dali, Pollock), which use a distinct table opcode set
• Linux can now read the PM table over PCI direct access (via /dev/mem) when ryzen_smu is not loaded

Changed

• tdc-limit and edc-limit now have descriptions and appear under "VRM & Currents" in --help instead of "Unknown command"
• Moved the MP1/RSMU mailbox tables, SMU status codes, and PM-table metadata into shared modules (mailbox.py, pmtable.py)
• Unified the Linux register send/poll path into one routine shared by the ryzen_smu and PCI backends; simplified apply() token parsing
• Removed unused runner helpers (get_socket_short, has_smu_support, get_commands)

Fixed

• Windows --table / --dump-table now work on VanGogh (Steam Deck) and Mendocino, which were missing from the PM-table family list

v0.2.0

Added

• Public library API: from zenmaster import detect, apply, runner, smu, CpuInfo
• runner.is_flag_arg() to query args that take no value
• vrmcvip-current (VanGogh) for full RyzenAdj arg parity

Changed

• Linux PCI backend now probes writability with a 0x47 register round-trip
  (like UXTU4Linux) instead of only checking that the config file exists, so a
  non-writable bus (no root, lockdown) fails at init with a clear message
• Unified project description to "Adjust power management settings for Ryzen
  CPUs on Linux and Windows"
• Rewrote README with a fuller overview, compatibility table, and an accurate
  RyzenAdj comparison
• Args that require a value now report an error (or show help on the CLI)
  when passed without =value, instead of silently sending 0
• apply() results always include a consistent set of keys (error is
  None on success), making the return shape stable for consumers
• Argument names are normalised to lowercase/hyphen form in apply() results
• Modernised packaging to PEP 639 SPDX license metadata

Fixed

• Windows PM table read now aborts if the final table-transfer retry is still
  rejected, instead of reading stale/garbage physical memory
• Windows SMU polling sleeps briefly after a fast initial spin, avoiding a busy
  loop that pinned a core while waiting for the mailbox
• --skin-temp-limit (a power limit in mW) is no longer multiplied by 256;
  only the temperature args apu-skin-temp and dgpu-skin-temp are scaled,
  matching RyzenAdj
• Negative values (e.g. Curve Optimiser --set-coall=-20) now wrap to
  unsigned 32-bit like RyzenAdj, instead of being clamped to 0
• Flag args (--max-performance, --enable-oc, …) always send 0, ignoring any
  value passed, matching RyzenAdj's boolean handling
• smu.send_* now raise a clear error if called before smu.init()
• smu.init() is idempotent and no longer leaks the Windows PawnIO handle
• CLI shows help for a malformed argument before requiring root
• apply() handles malformed preset strings (unclosed quotes) gracefully
• --table no longer shows duplicate TDC values in EDC rows on Zen 5 tables

v0.1.1

Fixed

• Workflow and packaging fixes for initial PyPI release