Files
neuron-tai/packages/node/meshnet_node/native_backend.py
Dobromir Popov 03e97ca31a fix: bind recipe identity to certified artifact bytes (DGR-025)
Append +artifact.<sha256> to the llama.cpp runtime axis, computed from the
exact bytes read by attest_loaded_runtime, so a differently-built shared
object with copied lock values can no longer forge a certified runtime
identity. Node/tracker parsers require the suffix; new test proves a
byte-identical-lock but different-binary artifact produces a different
recipe fingerprint. Regenerates conformance vectors accordingly.

105 passed in tests/test_native_identity_emission.py,
tests/test_runtime_pin_identity.py, tests/test_runtime_recipe_identity.py.
2026-07-21 13:22:02 +03:00

601 lines
24 KiB
Python

"""Authoritative identity boundary for a native GGUF Shard backend.
The native loader owns the facts about the mapped artifact. This module turns
that immutable report and separately pinned deployment inputs into the one
``ShardIdentity`` DGR-003 permits a native worker to emit. It is intentionally
not an adapter for the legacy Transformers backend: that backend has no
authoritative immutable GGUF artifact pin and must remain identity-free.
"""
from __future__ import annotations
import ctypes
import hashlib
import json
import re
from dataclasses import dataclass, field
from pathlib import Path
from .native_protocol import BUNDLE_VERSION, SCHEMA_VERSION, pb
from .runtime_pin import DEFAULT_LOCK_DIR, RuntimePin, load_runtime_pin
from .runtime_recipe import (
ArtifactIdentity,
DerivativeBinding,
RecipeIdentityError,
RuntimeRecipe,
ShardIdentity,
check_session_open,
handshake_error,
)
_HEX40 = re.compile(r"^[0-9a-f]{40}$")
_HEX64 = re.compile(r"^[0-9a-f]{64}$")
# The executing-runtime attestation contract.
#
# The repository lock is world-readable, so a Python object holding
# lock-shaped values proves nothing about the runtime that will execute:
# copying `load_runtime_pin()` into a self-report is exactly the forgery
# DGR-025 forbids. Attestation values are therefore accepted only when
# *extracted from the native artifact itself*, through two channels that must
# agree:
#
# 1. static — the artifact's bytes embed exactly one
# ``MESHNET-RUNTIME-ATTESTATION.v1:<canonical json>`` marker (NUL
# terminated). The DGR-027 CMake ABI-marker lane is where the native
# build bakes it in from the lock at configure time.
# 2. dynamic — the artifact must actually dlopen, and its exported
# ``llama_meshnet_runtime_attestation`` symbol must return that same
# marker. A marker pasted into a plain file is not an executing runtime.
#
# What this cannot prove: a cross-compiler bit-reproducible binary SHA, or
# that an adversary did not *build* a native artifact that embeds lock-true
# values while lying about its source. Manufacturing a lying native build is
# a categorically higher bar than authoring a Python dict, and real
# distributed certification (DGR-025's registered-but-dark ledger) remains
# the final backstop behind this boundary.
ATTESTATION_MARKER_PREFIX = b"MESHNET-RUNTIME-ATTESTATION.v1:"
ATTESTATION_SYMBOL = "llama_meshnet_runtime_attestation"
_ATTESTATION_STR_FIELDS = (
"runtime_name",
"upstream_commit",
"patched_tree",
"patch_stack_digest",
"build_recipe_digest",
)
_ATTESTATION_INT_FIELDS = ("boundary_schema_version", "protocol_schema_version")
# Module-private capability: evidence can only be minted where an artifact
# was actually read, scanned, loaded, and queried.
_EVIDENCE_TOKEN = object()
def attestation_payload(
*,
runtime_name: str,
upstream_commit: str,
patched_tree: str,
patch_stack_digest: str,
build_recipe_digest: str,
boundary_schema_version: int,
protocol_schema_version: int,
) -> bytes:
"""The canonical marker payload for one exact runtime.
This single encoding is shared by the build lane that embeds the marker,
the extractor that parses it, and the binding check that ties attestation
fields to the extracted evidence — so there is exactly one byte string a
given runtime identity can legitimately embed.
"""
return json.dumps(
{
"runtime_name": runtime_name,
"upstream_commit": upstream_commit,
"patched_tree": patched_tree,
"patch_stack_digest": patch_stack_digest,
"build_recipe_digest": build_recipe_digest,
"boundary_schema_version": boundary_schema_version,
"protocol_schema_version": protocol_schema_version,
},
sort_keys=True,
separators=(",", ":"),
ensure_ascii=False,
).encode("utf-8")
def expected_attestation_payload(
pin: RuntimePin,
*,
boundary_schema_version: int = BUNDLE_VERSION,
protocol_schema_version: int = int(SCHEMA_VERSION),
) -> bytes:
"""The marker payload a native build of this lock workspace must embed."""
return attestation_payload(
runtime_name=pin.runtime_name,
upstream_commit=pin.upstream_commit,
patched_tree=pin.patched_tree,
patch_stack_digest=pin.patch_stack_digest,
build_recipe_digest=pin.build_recipe_digest,
boundary_schema_version=boundary_schema_version,
protocol_schema_version=protocol_schema_version,
)
@dataclass(frozen=True)
class NativeArtifactEvidence:
"""Proof that attestation values came out of a loadable native artifact.
``binary_digest`` pins *which* artifact bytes were attested;
``payload_digest`` pins *what* those bytes attested, and is re-derived
from the attestation's own fields on construction so the values cannot be
edited after extraction (``dataclasses.replace`` laundering fails).
Only :func:`attest_loaded_runtime` can mint this object.
"""
artifact_path: str
binary_digest: str
payload_digest: str
_token: object = field(default=None, repr=False, compare=False)
def __post_init__(self) -> None:
if self._token is not _EVIDENCE_TOKEN:
raise RecipeIdentityError(
"native artifact evidence can only be minted by "
"attest_loaded_runtime() from an actually loaded native "
"artifact; it cannot be authored from repository lock values"
)
if not isinstance(self.artifact_path, str) or not self.artifact_path:
raise RecipeIdentityError("native artifact evidence must name the artifact")
for field_name in ("binary_digest", "payload_digest"):
value = getattr(self, field_name)
if not isinstance(value, str) or not _HEX64.fullmatch(value):
raise RecipeIdentityError(
f"native artifact evidence {field_name!r} must be a 64-hex sha256"
)
@dataclass(frozen=True)
class NativeRuntimeAttestation:
"""What the *executing* runtime reports about itself, at load time.
The repository lock says what the runtime is supposed to be; this says
what the loaded runtime *is* — the source tree it was built from, the
patch stack compiled into it, the numerically relevant build recipe, and
the boundary/protocol schema (ABI) it speaks. The values are never
accepted from a caller: they must arrive bound to
:class:`NativeArtifactEvidence`, which only
:func:`attest_loaded_runtime` can produce by reading, loading, and
querying the native artifact itself. Identity construction then compares
them to the lock/build-derived expectation and refuses on any difference,
so a worker cannot serve a lock it is not actually running — and cannot
fake one by copying the lock into a Python self-report.
Deliberately *not* attested: a compiler-specific binary SHA. Binding the
recorded build recipe is honest about what the manifest can prove;
bit-reproducible binary attestation is not claimed.
"""
runtime_name: str
upstream_commit: str
patched_tree: str
patch_stack_digest: str
build_recipe_digest: str
boundary_schema_version: int
protocol_schema_version: int
evidence: NativeArtifactEvidence
def __post_init__(self) -> None:
if not isinstance(self.evidence, NativeArtifactEvidence):
raise RecipeIdentityError(
"runtime attestation values must be extracted from the loaded "
"native artifact via attest_loaded_runtime(); a Python "
"self-report carrying copied lock values is not an attestation"
)
if not isinstance(self.runtime_name, str) or not self.runtime_name.strip():
raise RecipeIdentityError(
"runtime attestation must name the executing runtime"
)
for field_name, pattern, what in (
("upstream_commit", _HEX40, "40-hex upstream commit"),
("patched_tree", _HEX40, "40-hex patched source tree id"),
("patch_stack_digest", _HEX64, "64-hex patch-stack digest"),
("build_recipe_digest", _HEX64, "64-hex build-recipe digest"),
):
value = getattr(self, field_name)
if not isinstance(value, str) or not pattern.fullmatch(value):
raise RecipeIdentityError(
f"runtime attestation {field_name!r} must be an exact {what}"
)
for field_name in _ATTESTATION_INT_FIELDS:
value = getattr(self, field_name)
if isinstance(value, bool) or not isinstance(value, int) or value < 1:
raise RecipeIdentityError(
f"runtime attestation {field_name!r} must be a positive integer"
)
payload = attestation_payload(
**{name: getattr(self, name) for name in _ATTESTATION_STR_FIELDS},
**{name: getattr(self, name) for name in _ATTESTATION_INT_FIELDS},
)
if hashlib.sha256(payload).hexdigest() != self.evidence.payload_digest:
raise RecipeIdentityError(
"runtime attestation fields do not match the attestation "
"extracted from the native artifact; refusing values edited "
"after extraction"
)
def _parse_attestation_payload(payload: bytes) -> dict[str, object]:
"""Strictly parse one embedded marker payload, or refuse."""
try:
doc = json.loads(payload.decode("utf-8"))
except (UnicodeDecodeError, json.JSONDecodeError) as exc:
raise RecipeIdentityError(
f"embedded runtime attestation marker is not valid JSON: {exc}"
) from exc
expected_keys = set(_ATTESTATION_STR_FIELDS) | set(_ATTESTATION_INT_FIELDS)
if not isinstance(doc, dict) or set(doc) != expected_keys:
raise RecipeIdentityError(
"embedded runtime attestation marker must record exactly the "
"attestation fields"
)
for name in _ATTESTATION_STR_FIELDS:
if not isinstance(doc[name], str):
raise RecipeIdentityError(
f"embedded runtime attestation field {name!r} must be a string"
)
for name in _ATTESTATION_INT_FIELDS:
if isinstance(doc[name], bool) or not isinstance(doc[name], int):
raise RecipeIdentityError(
f"embedded runtime attestation field {name!r} must be an integer"
)
if attestation_payload(**doc) != payload: # type: ignore[arg-type]
raise RecipeIdentityError(
"embedded runtime attestation marker is not in canonical form"
)
return doc
def attest_loaded_runtime(artifact_path: Path | str) -> NativeRuntimeAttestation:
"""Extract the executing runtime's attestation from its native artifact.
Fails closed when the artifact is missing or empty, embeds no attestation
marker (a runtime built without the attestation lane cannot prove what it
is), embeds conflicting markers, is not a loadable shared object, does
not export :data:`ATTESTATION_SYMBOL`, or reports through that symbol
anything other than the embedded marker.
The returned attestation is bound to the artifact by its byte digest and
to the extracted values by the payload digest. Loading the artifact does
execute its initializers — this is the same artifact the worker is about
to run inference with, so that adds no new execution. An OS-level swap
of the file between the byte read and the dlopen is a documented
residual race; distributed certification remains the final backstop.
"""
path = Path(artifact_path)
try:
data = path.read_bytes()
except FileNotFoundError:
raise RecipeIdentityError(
f"native runtime artifact not found at {path}; without the built "
"native runtime there is no executing identity to attest"
) from None
except OSError as exc:
raise RecipeIdentityError(
f"native runtime artifact at {path} is unreadable: {exc}"
) from exc
if not data:
raise RecipeIdentityError(f"native runtime artifact at {path} is empty")
payloads: list[bytes] = []
cursor = 0
while (start := data.find(ATTESTATION_MARKER_PREFIX, cursor)) >= 0:
end = data.find(b"\x00", start)
if end < 0:
raise RecipeIdentityError(
"embedded runtime attestation marker is not NUL-terminated"
)
payloads.append(data[start + len(ATTESTATION_MARKER_PREFIX) : end])
cursor = end
if not payloads:
raise RecipeIdentityError(
f"native runtime artifact at {path} embeds no runtime attestation "
"marker; a runtime built without the attestation lane cannot "
"prove what it is"
)
if len(set(payloads)) != 1:
raise RecipeIdentityError(
f"native runtime artifact at {path} embeds conflicting runtime "
"attestation markers"
)
payload = payloads[0]
doc = _parse_attestation_payload(payload)
try:
library = ctypes.CDLL(str(path), mode=ctypes.RTLD_LOCAL)
except OSError as exc:
raise RecipeIdentityError(
f"native runtime artifact at {path} is not a loadable native "
"artifact; an attestation marker copied into a plain file is not "
"an executing runtime"
) from exc
try:
symbol = getattr(library, ATTESTATION_SYMBOL)
except AttributeError:
raise RecipeIdentityError(
f"native runtime artifact at {path} does not export "
f"{ATTESTATION_SYMBOL}; the loaded runtime itself must report "
"its attestation"
) from None
symbol.restype = ctypes.c_char_p
symbol.argtypes = []
reported = symbol()
if reported != ATTESTATION_MARKER_PREFIX + payload:
raise RecipeIdentityError(
f"the runtime loaded from {path} reports a different attestation "
"than its artifact embeds; refusing an artifact that disagrees "
"with itself"
)
evidence = NativeArtifactEvidence(
artifact_path=str(path),
binary_digest=hashlib.sha256(data).hexdigest(),
payload_digest=hashlib.sha256(payload).hexdigest(),
_token=_EVIDENCE_TOKEN,
)
return NativeRuntimeAttestation(evidence=evidence, **doc) # type: ignore[arg-type]
@dataclass(frozen=True)
class NativeLoadedArtifactReport:
"""Immutable GGUF facts returned by the loaded native model.
The report is copied from ``llama_model_meshnet_range_report`` plus the
parsed GGUF metadata while the model is live. Byte counts are operational
evidence rather than compatibility axes, but keeping them beside the range
prevents a caller from substituting an unverified range declaration.
``runtime_attestation`` must be the evidence-bound attestation extracted
from the loaded native artifact by :func:`attest_loaded_runtime`; a
report without one cannot be turned into an identity at all, and one
cannot exist without an actual native artifact to extract it from.
"""
owned_start_layer: int
owned_end_layer: int
mapped_bytes: int
resident_bytes: int
registered_bytes: int
architecture: str
architecture_digest: str
layer_count: int
runtime_attestation: NativeRuntimeAttestation
def __post_init__(self) -> None:
if self.owned_start_layer < 0 or self.owned_end_layer <= self.owned_start_layer:
raise RecipeIdentityError("native report has an invalid owned layer range")
if self.layer_count < 1 or self.owned_end_layer > self.layer_count:
raise RecipeIdentityError("native report range is outside GGUF layer metadata")
if min(self.mapped_bytes, self.resident_bytes, self.registered_bytes) < 0:
raise RecipeIdentityError("native report byte counts must be non-negative")
if not isinstance(self.runtime_attestation, NativeRuntimeAttestation):
raise RecipeIdentityError(
"native report must carry the executing runtime's attestation"
)
@dataclass(frozen=True)
class ImmutableArtifactPin:
"""Deployment-supplied immutable bytes pin, never inferred from a model name."""
artifact_id: str
revision: str
content_digest: str
derived_from: DerivativeBinding | None = None
@dataclass(frozen=True)
class NativeNumericalRecipe:
"""Immutable numerical inputs selected for this native worker instance."""
weight_quantization: str
activation_dtype: str
compute_dtype: str
kv_dtype: str
kv_layout: str
architecture_adapter: str
backend_id: str
recipe_id: str
recipe_version: str
catalogue_version: str
boundary_schema_version: int = BUNDLE_VERSION
protocol_schema_version: int = int(SCHEMA_VERSION)
@dataclass(frozen=True)
class NativeIdentityInputs:
"""Everything a native backend needs to emit one exact identity.
``tokenizer_revision`` must be the content-addressed identity computed by
:func:`meshnet_node.runtime_recipe.tokenizer_identity` over the loaded
tokenizer/config bytes; identity construction rejects anything else.
"""
loaded_artifact: NativeLoadedArtifactReport
artifact_pin: ImmutableArtifactPin
tokenizer_revision: str
numerical_recipe: NativeNumericalRecipe
def _require_attested_runtime(
attested: NativeRuntimeAttestation,
expected: RuntimePin,
recipe: NativeNumericalRecipe,
) -> None:
"""Fail closed unless the executing runtime is the locked, built runtime.
Every comparison is exact and every difference is separately fatal: an
attestation that agrees on the commit but not the patch stack (or the
patched tree, or the build recipe, or the ABI) is a different runtime
wearing the lock's name, and letting it emit the lock's identity is
exactly the substitution DGR-025 exists to prevent.
"""
for what, got, want in (
("runtime name", attested.runtime_name, expected.runtime_name),
("upstream commit", attested.upstream_commit, expected.upstream_commit),
("patched source tree", attested.patched_tree, expected.patched_tree),
(
"ordered patch stack",
attested.patch_stack_digest,
expected.patch_stack_digest,
),
(
"build recipe",
attested.build_recipe_digest,
expected.build_recipe_digest,
),
):
if got != want:
raise RecipeIdentityError(
f"the executing runtime's attested {what} does not match the "
"lock/build-derived expectation; refusing to emit an identity "
"for a runtime this node is not provably running"
)
for what, got, want in (
(
"boundary schema",
attested.boundary_schema_version,
recipe.boundary_schema_version,
),
(
"protocol schema",
attested.protocol_schema_version,
recipe.protocol_schema_version,
),
):
if got != want:
raise RecipeIdentityError(
f"the executing runtime's attested {what} version ({got}) does "
f"not match the recipe's ({want}); an ABI the runtime does not "
"actually speak cannot be part of its identity"
)
def shard_identity_from_native_report(
inputs: NativeIdentityInputs,
*,
lock_dir: Path = DEFAULT_LOCK_DIR,
) -> ShardIdentity:
"""Derive identity only from the native report and immutable pinned inputs.
The ``runtime_version`` axis is never accepted from a caller: it is derived
from the committed lock workspace, and the loaded runtime's attestation
must match that lock/build-derived expectation exactly — otherwise this
raises and no identity exists to register, admit, or certify.
"""
report = inputs.loaded_artifact
pin = inputs.artifact_pin
recipe = inputs.numerical_recipe
if recipe.backend_id.strip().lower() not in {"llama.cpp", "llama-cpp"}:
raise RecipeIdentityError(
"native llama.cpp identity requires backend_id 'llama.cpp' or 'llama-cpp'"
)
runtime_pin = load_runtime_pin(lock_dir)
_require_attested_runtime(report.runtime_attestation, runtime_pin, recipe)
# The lock-derived prefix identifies the intended source/patch/build
# recipe. The executing artifact digest identifies the bytes that actually
# supplied the attestation. Without this suffix, any independently built
# shared object could copy the public lock values into its marker and claim
# the exact same compatibility identity as the certified artifact.
runtime_version = (
f"{runtime_pin.runtime_version}"
f"+artifact.{report.runtime_attestation.evidence.binary_digest}"
)
artifact = ArtifactIdentity(
artifact_id=pin.artifact_id,
revision=pin.revision,
content_digest=pin.content_digest,
architecture=report.architecture,
architecture_digest=report.architecture_digest,
layer_count=report.layer_count,
derived_from=pin.derived_from,
)
return ShardIdentity(
artifact=artifact,
recipe=RuntimeRecipe(
weight_quantization=recipe.weight_quantization,
activation_dtype=recipe.activation_dtype,
compute_dtype=recipe.compute_dtype,
kv_dtype=recipe.kv_dtype,
kv_layout=recipe.kv_layout,
tokenizer_revision=inputs.tokenizer_revision,
architecture_adapter=recipe.architecture_adapter,
backend_id=recipe.backend_id,
runtime_version=runtime_version,
boundary_schema_version=recipe.boundary_schema_version,
protocol_schema_version=recipe.protocol_schema_version,
recipe_id=recipe.recipe_id,
recipe_version=recipe.recipe_version,
catalogue_version=recipe.catalogue_version,
),
shard_start=report.owned_start_layer,
shard_end=report.owned_end_layer,
)
class NativeSessionRejected(RecipeIdentityError):
"""A native worker refused a ``SessionOpen`` before allocating session state."""
def __init__(self, error: "pb.ShardError") -> None:
super().__init__(f"native SessionOpen rejected: {error.code}")
self.error = error
class NativeWorkerBackendAdapter:
"""Small backend-facing adapter around the native loaded-artifact seam."""
def __init__(self, identity_inputs: NativeIdentityInputs) -> None:
self.identity_inputs = identity_inputs
self.identity = shard_identity_from_native_report(identity_inputs)
@property
def loaded_artifact_report(self) -> NativeLoadedArtifactReport:
return self.identity_inputs.loaded_artifact
def check_session_open(
self,
opened: "pb.SessionOpen",
*,
expected_route_session_id: str | None = None,
expected_route_epoch: int | None = None,
) -> None:
"""Reject incompatible/stale opens at the native worker boundary."""
mismatches = check_session_open(
self.identity,
opened,
expected_route_session_id=expected_route_session_id,
expected_route_epoch=expected_route_epoch,
)
error = handshake_error(mismatches)
if error is not None:
raise NativeSessionRejected(error)
def on_session_open(
self,
opened: "pb.SessionOpen",
*,
expected_route_session_id: str | None = None,
expected_route_epoch: int | None = None,
) -> "pb.SessionAccepted":
"""The native worker's SessionOpen boundary, before session allocation."""
self.check_session_open(
opened,
expected_route_session_id=expected_route_session_id,
expected_route_epoch=expected_route_epoch,
)
return pb.SessionAccepted(
schema_version=SCHEMA_VERSION,
route_session_id=opened.route_session_id,
route_epoch=opened.route_epoch,
fingerprint=self.identity.fingerprint.to_proto(),
)