upstream-seaweedfs/test/winfsp-conformance/known_failures.txt

# Known winfsp-tests failures for the SeaweedFS Windows mount.
#
# One test name per line, as winfsp-tests reports it. A trailing * matches a
# prefix, so "stream_*" covers a whole group. Listed tests are excluded from
# the run; a failure in anything NOT listed fails CI, which is what catches a
# regression.
#
# Every test runs on its own in a clean directory, so an entry here is a real
# defect rather than the wreckage of the test before it. The rdwr and flush
# group is gone from this list for exactly that reason: it passes in isolation
# and only failed as collateral. Every entry below is a
# gap in the mount rather than a quirk of the suite, and the list is meant to
# shrink. Keep a reason on each group — an entry with no reason cannot be told
# apart from one nobody has looked at.

# Cannot pass by construction
# ---------------------------
# Alternate data streams: the filer stores one byte range per entry, and the
# mount exposes no second stream.
stream_*
# Reparse points: Symlink is refused, so there is nothing to resolve.
reparse_*
# Byte-range locks live in the WinFsp driver, so they never reach the mount.
lock_*

# Security descriptors
# --------------------
# The mount carries unix mode bits and no ACL, and Chown accepts and discards,
# so a descriptor written here does not read back.
create_sd_test
getsecurity_test

# Memory-mapped IO
# ----------------
# rdwr-test.c:645 compares the mapped bytes against the pattern that was
# written and finds them different. A real data mismatch, not a cascade: it
# only surfaced once each test ran on its own and this one got far enough to
# check. Worth its own investigation.
rdwr_mmap_test

# Delete semantics
# ----------------
# WinFsp advertises POSIX unlink, so a file can be deleted while handles are
# still open; the mount does not implement the delete-pending state.
delete_access_test
delete_ex_test
delete_mmap_test
delete_pending_test

# Rename semantics
# ----------------
# Rename over an open file, and rename of a directory holding open handles,
# which WinFsp is stricter about than POSIX.
rename_ex_test
rename_mmap_test
rename_open_test
exec_rename_dir_test

# File information
# ----------------
# A regular file reports as archived now, but hidden, system and readonly are
# still not round-tripped: create_fileattr gets past the archive check and then
# asks for READONLY as well, which needs Chflags. Creation time is reported as
# ctime because the raw protocol's Attr carries no field for it outside darwin.
create_fileattr_test
# getfileinfo asserts a file created a moment ago has an access time within ten
# seconds of now, and it does not. Two causes have been fixed and neither was
# it: Utimens no longer stores a zero timestamp, and the in-memory overlays are
# dropped when an inode is released so a recycled number cannot inherit an old
# time. Still unexplained, so it stays listed rather than pretended about.
getfileinfo_test
create_readonlydir_test
setfileinfo_test

# Directory enumeration
# ---------------------
# querydir_namelen only: checkName caps at 255 bytes while WinFsp counts 255
# characters. The other two are expected to pass now that "." and ".." are
# reported for non-root directories, which is what WinFsp enumerates.
querydir_namelen_test

# Name length
# -----------
# checkName caps at 255 bytes while WinFsp allows 255 characters, so a CJK
# name that is legal to Windows is rejected here.
create_namelen_test

# Sharing modes
# -------------
# Windows share-access checking on create, which the mount does not enforce.
create_share_test
create_test