[PATCH 00/27] landlock: Selftest fixes and cleanups
Günther Noack
gnoack3000 at gmail.com
Thu Oct 8 14:25:29 UTC 2026
A collection of Landlock Selftest fixes
=======================================
This patch set is bundled together to be slightly less spammy,
but the patches are mostly independent. The commits are ordered
such that the simpler and more orthogonal patches come earlier.
The patch set is LLM-assisted (and that turned out to be very useful
for spotting errors). Needless to say, I have verified all commits by
hand, and reworked them where necessary.
The last few patches are a bit more cross-cutting and probably more
prone to merge conflicts, as they correct cases where we have been
using EXPECT and ASSERT in the wrong way (details in the commit
messages). I believe that the rationale for each of these is
reasonable, but I am happy to discuss.
Motivation
==========
LLMs (and humans ;-)) have an unfortunate tendency to pick up and
replicate bad coding style patterns, so I hope that a broader cleanup
of the test style will avoid this spreading further.
When applying this patch set, please base it on a branch that has Wang
Yan's snprintf() patch in it [1], which I also support as a selftest
cleanup.
–Günther
[1] https://lore.kernel.org/all/20260626070016.61049-1-wangyan01@kylinos.cn/
Günther Noack (27):
selftests/landlock: Propagate grandchild failures in unix_scoping
selftests/landlock: Propagate EXPECT failures from forked children
selftests/landlock: Check the thread result in audit.thread
selftests/landlock: Zero-initialize the buffer in inconsistent_attr
selftests/landlock: Fix the openat() success check in fs_bench
selftests/landlock: Fix the reported clock tick rate in fs_bench
selftests/landlock: Fix the ruleset_fd check in empty_or_same_ruleset
selftests/landlock: Handle a missing control message in recv_fd()
selftests/landlock: Assert tracefs_extract_field() results that are
used
selftests/landlock: Assert audit initialization in fixture setups
selftests/landlock: Assert helper results that are used afterwards
selftests/landlock: Assert audit_count_records() before using its
result
selftests/landlock: Fix memory leak in audit_init_filter_exe()
selftests/landlock: Make audit_message large enough for any exe filter
selftests/landlock: Close leaked file descriptors
selftests/landlock: Check errno for combined VERSION and ERRATA flags
selftests/landlock: Fix SIGURG handler setup in scoped_signal_test
selftests/landlock: Fix regex_escape() error handling in
matches_log_fs_extra()
selftests/landlock: Fix off-by-one in regex_escape()
selftests/landlock: Fix the snprintf() error check in
matches_log_fs_extra()
selftests/landlock: Use _exit() in the named_pipe_ioctl child
selftests/landlock: Fix messages and comments in helper programs
selftests/landlock: Fix stale comments in common.h and audit.h
selftests/landlock: Use ASSERT for syscall failures before errno
checks
selftests/landlock: Assert WIFEXITED() before checking the exit status
selftests/landlock: Use EXPECT for closing FDs and stopping threads
selftests/landlock: Use EXPECT for checks that later steps do not need
tools/testing/selftests/landlock/audit.h | 20 +-
tools/testing/selftests/landlock/audit_test.c | 56 +-
tools/testing/selftests/landlock/base_test.c | 190 +--
tools/testing/selftests/landlock/common.h | 32 +-
tools/testing/selftests/landlock/fs_bench.c | 5 +-
tools/testing/selftests/landlock/fs_test.c | 1300 +++++++++--------
tools/testing/selftests/landlock/net_test.c | 64 +-
.../testing/selftests/landlock/ptrace_test.c | 44 +-
.../selftests/landlock/sandbox-and-launch.c | 6 +-
.../landlock/scoped_abstract_unix_test.c | 45 +-
.../selftests/landlock/scoped_signal_test.c | 23 +-
.../testing/selftests/landlock/scoped_test.c | 2 +-
.../selftests/landlock/trace_fs_test.c | 8 +-
tools/testing/selftests/landlock/trace_test.c | 26 +-
tools/testing/selftests/landlock/tsync_test.c | 22 +-
.../selftests/landlock/wait-pipe-sandbox.c | 16 +-
tools/testing/selftests/landlock/wait-pipe.c | 2 +-
17 files changed, 952 insertions(+), 909 deletions(-)
--
2.56.0
More information about the Linux-security-module-archive
mailing list