mptyl/ThothII
1
0
Fork 0
Code Issues 25 Pull Requests Actions Packages Projects Releases 1 Wiki Activity

fix: harden runtime config snapshot publication

Browse Source
This commit is contained in:
marcopan
2026-08-11 10:50:21 +02:00
parent 971315aa80
commit e9613767c5
5 changed files with 280 additions and 101 deletions
Download Patch File Download Diff File Expand all files Collapse all files
harness/tht/config.py
+92 -11
View File
@@ -3,6 +3,7 @@ import json
import os
import re
import stat
import sys
import warnings
from ipaddress import ip_address
from pathlib import Path
@@ -596,8 +597,9 @@ def _read_runtime_fd(fd: int, label: str, expected_mode: int = 0o400) -> tuple[b
def _strict_runtime_manifest(raw: object) -> dict[str, object]:
required = {
"version", "workspace_id", "workspace_revision", "descriptor_git_blob",
"descriptor_sha256", "config_sha256", "config_dwh_binding", "config_dev",
"config_ino", "config_size", "config_mode", "config_uid", "config_nlink",
"descriptor_sha256", "descriptor_dev", "descriptor_ino", "config_sha256",
"config_dwh_binding", "config_dev", "config_ino", "config_size", "config_mode",
"config_uid", "config_nlink",
}
if not isinstance(raw, dict) or set(raw) != required or raw.get("version") != 1:
raise ConfigError("Manifest runtime non valido")
@@ -609,7 +611,7 @@ def _strict_runtime_manifest(raw: object) -> dict[str, object]:
for key in ("descriptor_sha256", "config_sha256"):
if not isinstance(raw[key], str) or not re.fullmatch(r"[0-9a-f]{64}", raw[key]):
raise ConfigError("Manifest runtime non valido")
for key in ("config_dev", "config_ino", "config_size", "config_uid", "config_nlink"):
for key in ("descriptor_dev", "descriptor_ino", "config_dev", "config_ino", "config_size", "config_uid", "config_nlink"):
if not isinstance(raw[key], str) or not raw[key].isdigit():
raise ConfigError("Manifest runtime non valido")
if raw["config_mode"] != "400":
@@ -617,22 +619,100 @@ def _strict_runtime_manifest(raw: object) -> dict[str, object]:
return raw
def _open_runtime_component(parent: int, name: str) -> int:
before = os.stat(name, dir_fd=parent, follow_symlinks=False)
if stat.S_ISLNK(before.st_mode):
raise OSError("runtime config path contains a symlink")
flags = os.O_RDONLY | getattr(os, "O_DIRECTORY", 0) | os.O_NOFOLLOW
fd = os.open(name, flags, dir_fd=parent)
after = os.fstat(fd)
if (before.st_dev != after.st_dev or before.st_ino != after.st_ino
or not stat.S_ISDIR(after.st_mode)):
os.close(fd)
raise OSError("runtime config path changed during open")
return fd
def _open_runtime_file(path: Path, expected_mode: int) -> int:
if not path.is_absolute():
raise OSError("runtime config path must be absolute")
parts = list(path.parts)
if sys.platform == "darwin" and len(parts) > 1 and parts[1] in ("var", "tmp"):
parts = ["/", "private", *parts[1:]]
if not parts or parts[0] != "/" or any(part in ("", ".", "..") or "/" in part for part in parts[1:]):
raise OSError("runtime config path is invalid")
current = os.open("/", os.O_RDONLY | getattr(os, "O_DIRECTORY", 0))
try:
for component in parts[1:-1]:
nxt = _open_runtime_component(current, component)
os.close(current)
current = nxt
fd = os.open(parts[-1], os.O_RDONLY | os.O_NOFOLLOW, dir_fd=current)
info = os.fstat(fd)
if (not stat.S_ISREG(info.st_mode) or info.st_nlink != 1
or stat.S_IMODE(info.st_mode) != expected_mode or info.st_uid != os.getuid()):
os.close(fd)
raise OSError("unsafe runtime file")
return fd
finally:
os.close(current)
def _runtime_manifest_path(config_path: Path) -> Path:
if config_path.name == "" or config_path.suffix != ".yaml" or config_path.parent.name != "runtime-config":
raise OSError("runtime config path is not canonical")
if not re.fullmatch(r"[0-9a-f]{40}", config_path.stem):
raise OSError("runtime config path is not canonical")
return config_path.parent.parent / "runtime-config-manifests" / f"{config_path.stem}.json"
def load_config(path: Path) -> Config:
# Backend runtime leases pass the verified canonical config as fd 3 while retaining
# the ordinary absolute -c argument for diagnostics and source identity. Never reopen
# that pathname: an ancestor or leaf replacement after spawn must not alter bytes used
# by the harness.
# Registry leases authenticate the canonical pathname through a durable manifest
# digest. The config and manifest are opened component-by-component; FD 3/4 are
# reserved for the maintenance writer/root ABI.
runtime_manifest: dict[str, object] | None = None
expected_manifest = os.environ.get("THT_RUNTIME_CONFIG_MANIFEST_SHA256")
runtime_fd = os.environ.get("THT_CONFIG_FD")
manifest_fd = os.environ.get("THT_CONFIG_MANIFEST_FD")
expected_manifest = os.environ.get("THT_CONFIG_MANIFEST_SHA256")
if runtime_fd is not None or manifest_fd is not None or expected_manifest is not None:
if runtime_fd is None or manifest_fd is None or expected_manifest is None or not re.fullmatch(r"[0-9a-f]{64}", expected_manifest):
legacy_expected = os.environ.get("THT_CONFIG_MANIFEST_SHA256")
if expected_manifest is not None:
if not re.fullmatch(r"[0-9a-f]{64}", expected_manifest):
raise ConfigError("Handoff runtime incompleto")
config_fd = manifest_fd_local = None
try:
config_fd = _open_runtime_file(path, 0o400)
manifest_fd_local = _open_runtime_file(_runtime_manifest_path(path), 0o600)
config_bytes, config_info = _read_runtime_fd(config_fd, "config")
manifest_bytes, manifest_info = _read_runtime_fd(manifest_fd_local, "manifest", 0o600)
if hashlib.sha256(manifest_bytes).hexdigest() != expected_manifest:
raise ConfigError("Manifest runtime modificato")
runtime_manifest = _strict_runtime_manifest(json.loads(manifest_bytes.decode("utf-8")))
if (runtime_manifest["config_sha256"] != hashlib.sha256(config_bytes).hexdigest()
or int(runtime_manifest["config_dev"]) != config_info.st_dev
or int(runtime_manifest["config_ino"]) != config_info.st_ino
or int(runtime_manifest["config_size"]) != config_info.st_size
or int(runtime_manifest["config_uid"]) != config_info.st_uid
or int(runtime_manifest["config_nlink"]) != config_info.st_nlink
or config_info.st_dev == manifest_info.st_dev and config_info.st_ino == manifest_info.st_ino):
raise ConfigError("Identità config runtime non valida")
source_text = config_bytes.decode("utf-8")
except (OSError, UnicodeError, ValueError, json.JSONDecodeError) as exc:
if isinstance(exc, ConfigError):
raise
raise ConfigError("File di configurazione runtime non attendibile") from exc
finally:
if config_fd is not None:
os.close(config_fd)
if manifest_fd_local is not None:
os.close(manifest_fd_local)
elif runtime_fd is not None or manifest_fd is not None or legacy_expected is not None:
# Compatibility for direct /dev/fd callers. New backend leases never use it.
if runtime_fd is None or manifest_fd is None or legacy_expected is None or not re.fullmatch(r"[0-9a-f]{64}", legacy_expected):
raise ConfigError("Handoff runtime incompleto")
try:
config_bytes, config_info = _read_runtime_fd(int(runtime_fd), "config")
manifest_bytes, manifest_info = _read_runtime_fd(int(manifest_fd), "manifest", 0o600)
if hashlib.sha256(manifest_bytes).hexdigest() != expected_manifest:
if hashlib.sha256(manifest_bytes).hexdigest() != legacy_expected:
raise ConfigError("Manifest runtime modificato")
runtime_manifest = _strict_runtime_manifest(json.loads(manifest_bytes.decode("utf-8")))
if (runtime_manifest["config_sha256"] != hashlib.sha256(config_bytes).hexdigest()
@@ -655,6 +735,7 @@ def load_config(path: Path) -> Config:
source_text = path.read_text()
except OSError as exc:
raise ConfigError(f"File di configurazione non trovato: {path}") from exc
try:
raw = yaml.safe_load(source_text)
except yaml.YAMLError as exc: