fix: standalone writers take the repository lock

scripts/artifacts.py

@@ -208,11 +208,16 @@ def _strip_timestamps(text: str) -> str:
 class ArtifactLockBusy(RuntimeError):
     """Raised when another process already holds the artifact directory."""
 
-# A run that holds an artifact directory for its whole duration names it
-# here for the steps it spawns: they would otherwise be refused by it.
+# A run that holds artifact directories for its whole duration names them
+# here, os.pathsep apart, for the steps it spawns: they would otherwise be
+# refused by it. A pipeline holds the repository root and its --output-dir.
 HELD_LOCK_ENV = "RETROBIOS_HELD_ARTIFACT_DIR"
 
 
+def _held_directories() -> list[str]:
+    return [d for d in os.environ.get(HELD_LOCK_ENV, "").split(os.pathsep) if d]
+
+
 @contextlib.contextmanager
 def artifact_lock(directory: str, exclusive: bool = True) -> Iterator[None]:
     """Serialize access to a shared artifact directory across processes.
@@ -229,7 +234,7 @@ def artifact_lock(directory: str, exclusive: bool = True) -> Iterator[None]:
         return
 
     os.makedirs(directory, exist_ok=True)
-    if os.environ.get(HELD_LOCK_ENV) == os.path.realpath(directory):
+    if os.path.realpath(directory) in _held_directories():
         yield
         return
     lock_path = os.path.join(directory, ".lock")
@@ -256,16 +261,24 @@ def hold_artifact_lock(directory: str) -> Iterator[None]:
     run can purge the packs one step built before the next step checks
     them. The children this process spawns inherit the hold.
     """
-    with artifact_lock(directory):
-        previous = os.environ.get(HELD_LOCK_ENV)
-        os.environ[HELD_LOCK_ENV] = os.path.realpath(directory)
-        try:
-            yield
-        finally:
-            if previous is None:
-                os.environ.pop(HELD_LOCK_ENV, None)
-            else:
-                os.environ[HELD_LOCK_ENV] = previous
+    with artifact_lock(directory), publish_hold(directory):
+        yield
+
+
+@contextlib.contextmanager
+def publish_hold(directory: str) -> Iterator[None]:
+    """Name a directory this process holds to the steps it spawns."""
+    previous = os.environ.get(HELD_LOCK_ENV)
+    os.environ[HELD_LOCK_ENV] = os.pathsep.join(
+        [*_held_directories(), os.path.realpath(directory)]
+    )
+    try:
+        yield
+    finally:
+        if previous is None:
+            os.environ.pop(HELD_LOCK_ENV, None)
+        else:
+            os.environ[HELD_LOCK_ENV] = previous
 
 
 # What a pipeline run writes at fixed paths, whatever its --output-dir.
@@ -280,20 +293,40 @@ def hold_repository_outputs(root: str = ".") -> Iterator[None]:
     on dist/: both rewrote database.json, and the second to reach the
     manifest step found install/ held after minutes of work, or, when the
     steps did not overlap, rewrote the committed manifests from its own
-    state. The lock file is the repository root's .lock.
+    state. The lock file is the repository root's .lock. A pipeline passes
+    it to its steps with publish_hold.
     """
     stack = contextlib.ExitStack()
     try:
         stack.enter_context(artifact_lock(root))
     except ArtifactLockBusy as exc:
         raise ArtifactLockBusy(
-            f"another pipeline run is writing {', '.join(REPOSITORY_OUTPUTS)} "
+            f"another run is writing {', '.join(REPOSITORY_OUTPUTS)} "
             f"(lock: {os.path.join(root, '.lock')}). Wait for it to finish."
         ) from exc
     with stack:
         yield
 
 
+_REPO_ROOT = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
+
+
+def repository_output_lock(path: str) -> contextlib.AbstractContextManager[None]:
+    """The repository hold, when *path* is one of its fixed outputs.
+
+    Only a pipeline took it: generate_db.py run beside a pipeline replaced
+    database.json between its verify and its packs, and a manifest run on
+    install/ made the pipeline's own manifest step fail after minutes of
+    work. A path elsewhere (a scratch --output) takes nothing.
+    """
+    target = os.path.realpath(path)
+    for output in REPOSITORY_OUTPUTS:
+        fixed = os.path.join(_REPO_ROOT, output.rstrip("/"))
+        if target == fixed or target.startswith(fixed + os.sep):
+            return hold_repository_outputs(_REPO_ROOT)
+    return contextlib.nullcontext()
+
+
 def _build_timestamp(db: dict | None = None) -> str:
     """Timestamp for generated artifacts.
 

scripts/common.py

@@ -3547,6 +3547,8 @@ from artifacts import (  # noqa: E402,F401
     artifact_lock,
     hold_artifact_lock,
     hold_repository_outputs,
+    publish_hold,
+    repository_output_lock,
     _TIMESTAMP_PATTERNS,
     _strip_timestamps,
 )
@@ -3588,7 +3590,7 @@ def run_cli(main: Callable[[], int | None]) -> None:
     try:
         code = main()
         sys.stdout.flush()
-    except InputMissing as exc:
+    except (InputMissing, ArtifactLockBusy) as exc:
         print(f"ERROR: {exc}", file=sys.stderr)
         sys.exit(1)
     except BrokenPipeError:

scripts/generate_db.py

@@ -26,6 +26,7 @@ from common import (
     DEFAULT_PROVENANCE_DIR,
     list_registered_platforms,
     load_provenance_snapshots,
+    repository_output_lock,
     require_yaml,
     set_plain_mode,
     source_yaml_paths,
@@ -430,7 +431,11 @@ def main() -> int:
         help="Directory with dump-catalog snapshots",
     )
     args = parser.parse_args()
+    with repository_output_lock(args.output):
+        return _generate(args)
 
+
+def _generate(args: argparse.Namespace) -> int:
     bios_dir = Path(args.bios_dir)
     if not bios_dir.is_dir():
         print(f"Error: BIOS directory '{bios_dir}' not found", file=sys.stderr)
@@ -742,4 +747,6 @@ def _collect_all_aliases(
 
 
 if __name__ == "__main__":
-    sys.exit(main() or 0)
+    from common import run_cli
+
+    run_cli(main)

scripts/generate_games_index.py

@@ -49,6 +49,7 @@ from common import (  # noqa: E402
     load_emulator_profiles,
     load_platform_config,
     native_layer,
+    repository_output_lock,
     run_cli,
 )
 from region import normalize_declared  # noqa: E402
@@ -661,7 +662,13 @@ def main() -> int:
                         help="exit 1 when a file would change, naming what differs (implies --quiet)")
     parser.add_argument("--quiet", action="store_true", help="print only failures")
     args = parser.parse_args()
+    if args.check:
+        return _index(args)
+    with repository_output_lock(str(args.output_dir)):
+        return _index(args)
 
+
+def _index(args: argparse.Namespace) -> int:
     files, coverage = render(args.output_dir)
     if not (args.quiet or args.check):
         print_coverage(coverage, files)

scripts/generate_pack.py

@@ -48,6 +48,7 @@ from common import (
     profiles_for_systems,
     write_text_atomic,
     artifact_lock,
+    repository_output_lock,
     ArtifactLockBusy,
     build_target_cores_cache,
     build_zip_contents_index,
@@ -2809,9 +2810,11 @@ def _run_manifest_mode(
 
 @contextlib.contextmanager
 def _pack_output_lock(output_dir: str, exclusive: bool = True) -> Iterator[None]:
-    """Hold the output directory for the duration of a pack run."""
+    """Hold the output directory for the duration of a pack run, and the
+    repository's fixed outputs when it is one of them (install/)."""
     try:
-        with artifact_lock(output_dir, exclusive=exclusive):
+        with artifact_lock(output_dir, exclusive=exclusive), \
+                repository_output_lock(output_dir) if exclusive else contextlib.nullcontext():
             yield
     except ArtifactLockBusy as exc:
         print(f"ERROR: {exc}", file=sys.stderr)

scripts/generate_readme.py

@@ -30,6 +30,7 @@ from common import (
     load_emulator_profiles,
     load_platform_config,
     load_platform_registry,
+    repository_output_lock,
     resolve_platform_cores,
     system_index,
     unique_emulator_profiles,
@@ -1034,7 +1035,11 @@ def main() -> None:
         help="platform YAML directory (default: %(default)s)",
     )
     args = parser.parse_args()
+    with repository_output_lock("README.md"):
+        _write(args)
 
+
+def _write(args: argparse.Namespace) -> None:
     import contextlib
 
     from common import load_data_dir_registry

scripts/generate_site.py

@@ -12080,11 +12080,14 @@ def main() -> None:
         # files in the sweep, so the second is refused instead.
         import contextlib
 
-        from common import load_data_dir_registry
+        from common import load_data_dir_registry, repository_output_lock
         from refresh_data_dirs import hold_data_dirs
 
         # The coverage pages resolve against data/, held while they are read.
-        with artifact_lock(args.docs_dir), contextlib.ExitStack() as held:
+        # Whatever --docs-dir says, website/src/data and wiki/ are written.
+        with artifact_lock(args.docs_dir), \
+                repository_output_lock(str(PUBLISHED_PARTS_FILE)), \
+                contextlib.ExitStack() as held:
             hold_data_dirs(held, load_data_dir_registry(args.platforms_dir))
             _generate(args)
     except ArtifactLockBusy as exc:

scripts/pipeline.py

@@ -73,6 +73,7 @@ from common import (
     hold_artifact_lock,
     hold_repository_outputs,
     load_data_dir_registry,
+    publish_hold,
 )
 from refresh_data_dirs import hold_data_dirs
 
@@ -1002,9 +1003,11 @@ def main() -> None:
     # rewrote them beside the run on dist/.
     held = contextlib.ExitStack()
     try:
-        held.enter_context(
-            hold_repository_outputs(str(Path(__file__).resolve().parent.parent))
-        )
+        root = str(Path(__file__).resolve().parent.parent)
+        held.enter_context(hold_repository_outputs(root))
+        # Its steps write database.json, install/ and README.md themselves,
+        # and take the same lock to do so.
+        held.enter_context(publish_hold(root))
         if not args.skip_packs:
             held.enter_context(hold_artifact_lock(args.output_dir))
     except ArtifactLockBusy as exc:

scripts/sitesources.py

@@ -735,13 +735,18 @@ def main() -> int:
     if not args.fetch and not any(path.is_dir() for path in cache_dirs):
         print(f"Error: no tree cache at {', '.join(map(str, cache_dirs))}", file=sys.stderr)
         return 1
-    from common import load_emulator_profiles, write_if_changed  # noqa: PLC0415
+    from common import (  # noqa: PLC0415
+        load_emulator_profiles,
+        repository_output_lock,
+        write_if_changed,
+    )
 
     profiles = load_emulator_profiles(args.emulators_dir, skip_aliases=False)
-    held, unread = build_source_paths(profiles, cache_dirs, args.fetch)
-    held = keep_unread(held, unread, read_record(args.output))
-    args.output.parent.mkdir(parents=True, exist_ok=True)
-    write_if_changed(str(args.output), json.dumps(held, indent=1, sort_keys=True) + "\n")
+    with repository_output_lock(str(args.output)):
+        held, unread = build_source_paths(profiles, cache_dirs, args.fetch)
+        held = keep_unread(held, unread, read_record(args.output))
+        args.output.parent.mkdir(parents=True, exist_ok=True)
+        write_if_changed(str(args.output), json.dumps(held, indent=1, sort_keys=True) + "\n")
     print(f"{sum(map(len, held.values()))} paths held in {len(held)} trees, "
           f"{len(unread)} trees not cached: {args.output}")
     return 0