def check_materialized_are_accessible(
tracker: ProblemTracker,
instance: EdenInstance,
checkout: EdenCheckout,
) -> None:
mismatched_mode = []
inaccessible_inodes = []
with instance.get_thrift_client_legacy() as client:
materialized = client.debugInodeStatus(
bytes(checkout.path),
b"",
flags=DIS_REQUIRE_MATERIALIZED,
sync=SyncBehavior(),
)
for materialized_dir in materialized:
path = Path(os.fsdecode(materialized_dir.path))
try:
st = os.lstat(checkout.path / path)
except OSError:
inaccessible_inodes += [path]
continue
if not stat.S_ISDIR(st.st_mode):
mismatched_mode += [(path, stat.S_IFDIR, st.st_mode)]
for dirent in materialized_dir.entries:
if dirent.materialized:
dirent_path = path / Path(os.fsdecode(dirent.name))
try:
dirent_stat = os.lstat(checkout.path / dirent_path)
except OSError:
inaccessible_inodes += [dirent_path]
continue
# TODO(xavierd): Symlinks are for now recognized as files.
dirent_mode = (stat.S_IFREG if stat.S_ISLNK(
dirent_stat.st_mode) else stat.S_IFMT(dirent_stat.st_mode))
if dirent_mode != stat.S_IFMT(dirent.mode):
mismatched_mode += [(dirent_path, dirent_stat.st_mode,
dirent.mode)]
if inaccessible_inodes != []:
tracker.add_problem(
MaterializedInodesAreInaccessible(inaccessible_inodes))
if mismatched_mode != []:
tracker.add_problem(
MaterializedInodesHaveDifferentModeOnDisk(mismatched_mode))
def check_loaded_content(
tracker: ProblemTracker,
instance: EdenInstance,
checkout: EdenCheckout,
query_prjfs_file: Callable[[Path], PRJ_FILE_STATE],
) -> None:
with instance.get_thrift_client_legacy() as client:
loaded = client.debugInodeStatus(
bytes(checkout.path),
b"",
flags=DIS_REQUIRE_LOADED,
sync=SyncBehavior(),
)
errors = []
for loaded_dir in loaded:
path = Path(os.fsdecode(loaded_dir.path))
for dirent in loaded_dir.entries:
if not stat.S_ISREG(dirent.mode) or dirent.materialized:
continue
dirent_path = path / Path(os.fsdecode(dirent.name))
filestate = query_prjfs_file(checkout.path / dirent_path)
if (filestate & PRJ_FILE_STATE.HydratedPlaceholder !=
PRJ_FILE_STATE.HydratedPlaceholder):
# We should only compute the sha1 of files that have been read.
continue
def compute_file_sha1(file: Path) -> bytes:
hasher = hashlib.sha1()
with open(checkout.path / dirent_path, "rb") as f:
while True:
buf = f.read(1024 * 1024)
if buf == b"":
break
hasher.update(buf)
return hasher.digest()
sha1 = client.getSHA1(bytes(checkout.path),
[bytes(dirent_path)],
sync=SyncBehavior())[0].get_sha1()
on_disk_sha1 = compute_file_sha1(checkout.path / dirent_path)
if sha1 != on_disk_sha1:
errors += [(dirent_path, sha1, on_disk_sha1)]
if errors != []:
tracker.add_problem(LoadedFileHasDifferentContentOnDisk(errors))