refactor: use proc macro for wasi ignore reasons

.github/skills/code-review/SKILL.md

@@ -87,7 +87,10 @@ patches: assistants can reproduce GPL sources verbatim.
 - Tests must be original (see the provenance rule above) and pick values that are
   clearly distinct from GNU's fixtures.
 - Use `.no_output()` rather than `.no_stdout().no_stderr()`.
-- Anything that can't run under WASI needs `#[cfg_attr(wasi_runner, ignore)]`.
+- Anything that can't run under WASI needs `#[uutests::wasi_ignore(ReasonKey)]`,
+  reusing an existing key from `tests/uutests_procs/src/lib.rs` when one fits and
+  adding a new key only when no existing reason matches the root cause. Don't use
+  a raw `#[cfg_attr(wasi_runner, ignore = "...")]`.
 - If a GNU test was failing while the Rust suite passed, the PR should add a Rust
   test so the gap can't reopen.
 

.github/workflows/wasi.yml

@@ -56,7 +56,9 @@ jobs:
         cargo build --target ${{ matrix.job.target }} --no-default-features --features feat_wasm
         # Run host-compiled integration tests against the WASI binary.
         # Tests incompatible with WASI are annotated with
-        # #[cfg_attr(wasi_runner, ignore)] in the test source files.
+        # #[uutests::wasi_ignore(ReasonKey)] in the test source files, which
+        # expands to #[cfg_attr(wasi_runner, ignore = "...")]; the reason keys
+        # live in tests/uutests_procs/src/lib.rs.
         # The binary above is built without default features, so it has no
         # caret diagnostics; the host test crate does have them, hence the
         # not(wasi_runner) in the cfg guarding every `mod diagnostics`.

AGENTS.md

@@ -36,6 +36,19 @@ add a Rust test so it cannot silently regress.
 
 No tests, no merge.
 
+### Tests that can't run under WASI
+
+A test that can't run under the WASI runner is skipped with
+`#[uutests::wasi_ignore(ReasonKey)]`, not a raw
+`#[cfg_attr(wasi_runner, ignore = "...")]`. Every reason is defined once in
+`tests/uutests_procs/src/lib.rs`.
+
+Reuse an existing key when one fits the root cause. Add a new key only when no
+existing reason matches, and keep the list small - prefer widening an existing
+reason over introducing a near-duplicate. When a skip's real cause differs from
+its key, switch the key rather than leaving it mislabeled. See
+`docs/src/wasi-test-gaps.md`.
+
 ## Keep the PR description short
 
 Describe the problem being solved and what changed. That is all.
@@ -55,4 +68,5 @@ rules, this file is only a pointer:
 - `DEVELOPMENT.md` - build, test and environment setup
 - `README.md`, `SECURITY.md`, `CODE_OF_CONDUCT.md`
 - `docs/src/*.md` - platforms, performance, l10n, multicall, test coverage
+- `docs/src/wasi-test-gaps.md` - WASI test skips and the `wasi_ignore` reason keys
 - the per-utility `README.md` / `locales/*.ftl` files when touching a utility

Cargo.toml

@@ -27,6 +27,7 @@ members = [
   "src/uucore",
   "src/uucore_procs",
   "tests/uutests",
+  "tests/uutests_procs",
   # "fuzz",  # TODO
 ]
 
@@ -119,6 +120,7 @@ rustc-hash = "2.1.1"
 rustix = { version = "1.1.4", default-features = false }
 self_cell = "1.0.4"
 selinux = "0.6"
+syn = { version = "2.0", features = ["full"] }
 tempfile = "3.15.0"
 terminal_size = "0.4.0"
 textwrap = { version = "0.16.1", features = ["terminal_size"] }
@@ -161,6 +163,7 @@ uu_ls = { version = "0.13.0", path = "src/uu/ls" }
 uu_base32 = { version = "0.13.0", path = "src/uu/base32" }
 uu_checksum_common = { version = "0.13.0", path = "src/uu/checksum_common" }
 uutests = { version = "0.13.0", path = "tests/uutests" }
+uutests_procs = { version = "0.13.0", path = "tests/uutests_procs" }
 
 # This is the linting configuration for all crates.
 # In order to use these, all crates have `[lints] workspace = true` section.

docs/src/wasi-test-gaps.md

@@ -1,8 +1,10 @@
 # WASI integration test gaps
 
-Tests annotated with `#[cfg_attr(wasi_runner, ignore = "...")]` are skipped when running integration tests against a WASI binary via wasmtime. This document tracks the reasons so that gaps in WASI support are visible in one place.
+Tests annotated with `#[uutests::wasi_ignore(ReasonKey)]` are skipped when running integration tests against a WASI binary via wasmtime. The macro expands to `#[cfg_attr(wasi_runner, ignore = "...")]`, resolving `ReasonKey` to its reason text. Every reason lives once, alphabetically, in `tests/uutests_procs/src/lib.rs`; this document gives the prose overview so that gaps in WASI support are visible in one place.
 
-To find all annotated tests: `grep -rn 'wasi_runner, ignore' tests/`
+When a test cannot run under WASI, annotate it with `#[uutests::wasi_ignore(ReasonKey)]` using the key that matches the root cause. Reuse an existing key whenever one fits rather than adding a near-duplicate; add a new key to `tests/uutests_procs/src/lib.rs` only when no existing reason describes the cause, and keep the key list and this document in sync. If the real cause of an existing skip turns out to differ from its key, switch the key (and refine the reason text) rather than leaving it mislabeled.
+
+To find all annotated tests: `grep -rn 'wasi_ignore' tests/`. For the full, authoritative list of reason keys and their text, see `tests/uutests_procs/src/lib.rs`.
 
 ## Tools not yet covered by integration tests
 

tests/by-util/test_base64.rs

@@ -18,9 +18,9 @@ fn test_version() {
         .stdout_is(format!("base64 {}\n", uucore::crate_version!()));
 }
 
+#[uutests::wasi_ignore(NonUtf8ArgsUnsupported)]
 #[test]
 #[cfg(target_os = "linux")]
-#[cfg_attr(wasi_runner, ignore = "WASI: argv/filenames must be valid UTF-8")]
 fn test_base64_non_utf8_paths() {
     use std::os::unix::ffi::OsStringExt;
     let (at, mut ucmd) = at_and_ucmd!();
@@ -288,9 +288,9 @@ cyBvdmVyIHRoZSBsYXp5IGRvZy4=
         );
 }
 
+#[uutests::wasi_ignore(HostPathsNotVisible)]
 #[test]
 #[cfg(all(target_os = "linux", not(target_env = "musl")))]
-#[cfg_attr(wasi_runner, ignore = "WASI sandbox: host paths not visible")]
 fn test_read_error() {
     new_ucmd!()
         .arg("/proc/self/mem")

tests/by-util/test_basename.rs

@@ -163,8 +163,8 @@ fn test_too_many_args_output() {
 }
 
 #[cfg(unix)]
+#[uutests::wasi_ignore(NonUtf8ArgsUnsupported)]
 #[test]
-#[cfg_attr(wasi_runner, ignore = "WASI: argv/filenames must be valid UTF-8")]
 fn test_invalid_utf8_args() {
     let param = uucore::os_str_from_bytes(b"/tmp/some-\xc0-file.k\xf3")
         .expect("Only unix platforms can test non-unicode names");

tests/by-util/test_basenc.rs

@@ -189,9 +189,9 @@ fn test_base16() {
 }
 
 // This test fails if the Base16 output buffer is not explicitly flushed.
+#[uutests::wasi_ignore(HostPathsNotVisible)]
 #[test]
 #[cfg(target_os = "linux")]
-#[cfg_attr(wasi_runner, ignore = "WASI sandbox: host paths not visible")]
 fn test_base16_write_error_is_reported() {
     new_ucmd!()
         .arg("--base16")
@@ -407,9 +407,9 @@ fn test_file() {
         .stdout_is("Zm9v\n");
 }
 
+#[uutests::wasi_ignore(NonUtf8ArgsUnsupported)]
 #[test]
 #[cfg(target_os = "linux")]
-#[cfg_attr(wasi_runner, ignore = "WASI: argv/filenames must be valid UTF-8")]
 fn test_file_with_non_utf8_name() {
     use std::os::unix::ffi::OsStringExt;
     let (at, mut ucmd) = at_and_ucmd!();

tests/by-util/test_cat.rs

@@ -86,9 +86,9 @@ fn test_no_options_big_input() {
     }
 }
 
+#[uutests::wasi_ignore(NoFifo)]
 #[test]
 #[cfg(unix)]
-#[cfg_attr(wasi_runner, ignore = "WASI: no FIFO/mkfifo support")]
 fn test_fifo_symlink() {
     use std::io::Write;
     use std::thread;
@@ -119,11 +119,11 @@ fn test_fifo_symlink() {
     thread.join().unwrap();
 }
 
+#[uutests::wasi_ignore(NoRlimitSetrlimitSupport)]
 #[test]
 // TODO(#7542): Re-enable on Android once we figure out why setting limit is broken.
 // #[cfg(any(target_os = "linux", target_os = "android"))]
 #[cfg(target_os = "linux")]
-#[cfg_attr(wasi_runner, ignore = "WASI: rlimit/setrlimit not supported")]
 fn test_closes_file_descriptors() {
     // Each file creates a pipe, which has two file descriptors.
     // If they are not closed then five is certainly too many.
@@ -139,9 +139,9 @@ fn test_closes_file_descriptors() {
         .succeeds();
 }
 
+#[uutests::wasi_ignore(NoPipeSignalSupport)]
 #[test]
 #[cfg(unix)]
-#[cfg_attr(wasi_runner, ignore = "WASI: no pipe/signal support")]
 fn test_broken_pipe() {
     let mut cmd = new_ucmd!();
     let mut child = cmd
@@ -516,9 +516,9 @@ fn test_squeeze_blank_before_numbering() {
 }
 
 /// This tests reading from Unix character devices
+#[uutests::wasi_ignore(HostPathsNotVisible)]
 #[test]
 #[cfg(unix)]
-#[cfg_attr(wasi_runner, ignore = "WASI sandbox: host paths not visible")]
 fn test_dev_random() {
     #[cfg(any(target_os = "linux", target_os = "android"))]
     const DEV_RANDOM: &str = "/dev/urandom";
@@ -547,9 +547,9 @@ fn test_dev_random() {
 
 /// Reading from /dev/full should return an infinite amount of zero bytes.
 /// Wikipedia says there is support on Linux, FreeBSD, and `NetBSD`.
+#[uutests::wasi_ignore(HostPathsNotVisible)]
 #[test]
 #[cfg(any(target_os = "linux", target_os = "freebsd", target_os = "netbsd"))]
-#[cfg_attr(wasi_runner, ignore = "WASI sandbox: host paths not visible")]
 fn test_dev_full() {
     let mut proc = new_ucmd!()
         .set_stdout(Stdio::piped())
@@ -563,9 +563,9 @@ fn test_dev_full() {
     proc.kill();
 }
 
+#[uutests::wasi_ignore(HostPathsNotVisible)]
 #[test]
 #[cfg(any(target_os = "linux", target_os = "freebsd", target_os = "netbsd"))]
-#[cfg_attr(wasi_runner, ignore = "WASI sandbox: host paths not visible")]
 fn test_dev_full_show_all() {
     let buf_len = 2048;
     let mut proc = new_ucmd!()
@@ -586,9 +586,9 @@ fn test_dev_full_show_all() {
 // For some reason splice() on first of those files fails, resulting in
 // fallback inside `write_fast`, the other splice succeeds, in effect
 // without additional flush output gets reversed.
+#[uutests::wasi_ignore(HostPathsNotVisible)]
 #[test]
 #[cfg(target_os = "linux")]
-#[cfg_attr(wasi_runner, ignore = "WASI sandbox: host paths not visible")]
 fn test_write_fast_fallthrough_uses_flush() {
     const PROC_INIT_CMDLINE: &str = "/proc/1/cmdline";
     let cmdline = read_to_string(PROC_INIT_CMDLINE).unwrap();
@@ -599,9 +599,9 @@ fn test_write_fast_fallthrough_uses_flush() {
         .stdout_only(format!("{cmdline}abcde\nfghij\nklmno\npqrst\nuvwxyz\n")); // spell-checker:disable-line
 }
 
+#[uutests::wasi_ignore(NoUnixDomainSockets)]
 #[test]
 #[cfg(unix)]
-#[cfg_attr(wasi_runner, ignore = "WASI: no Unix domain socket support")]
 fn test_domain_socket() {
     use std::os::unix::net::UnixListener;
 
@@ -630,8 +630,8 @@ fn test_write_to_self_empty() {
     s.ucmd().set_stdout(file).arg("file.txt").succeeds();
 }
 
+#[uutests::wasi_ignore(CannotDetectUnsafeOverwrite)]
 #[test]
-#[cfg_attr(wasi_runner, ignore = "WASI: cannot detect unsafe overwrite")]
 fn test_write_to_self() {
     let s = TestScenario::new(util_name!());
     let file_path = s.fixtures.plus("first_file");
@@ -689,8 +689,8 @@ fn test_cat_rw_self_succeeds() {
 /// Test that cat handles self-referential input gracefully.
 ///
 /// `cat fx fx3 1<>fx3`
+#[uutests::wasi_ignore(CannotDetectUnsafeOverwrite)]
 #[test]
-#[cfg_attr(wasi_runner, ignore = "WASI: cannot detect unsafe overwrite")]
 fn test_cat_rw_self_conflict_fails() {
     let (at, mut ucmd) = at_and_ucmd!();
     at.write("source", "a");
@@ -713,13 +713,10 @@ fn test_cat_rw_self_conflict_fails() {
     assert_eq!(dest_contents, "acde");
 }
 
+#[uutests::wasi_ignore(SymlinkLoopNoEloop)]
 #[test]
 #[cfg(unix)]
 #[cfg(not(target_os = "openbsd"))]
-#[cfg_attr(
-    wasi_runner,
-    ignore = "WASI: symlink loop traversal does not surface ELOOP ('Too many levels of symbolic links')"
-)]
 fn test_error_loop() {
     let (at, mut ucmd) = at_and_ucmd!();
     at.symlink_file("2", "1");
@@ -750,9 +747,9 @@ fn test_u_ignored() {
     }
 }
 
+#[uutests::wasi_ignore(ErrnoMessageMismatch)]
 #[test]
 #[cfg(unix)]
-#[cfg_attr(wasi_runner, ignore = "WASI: errno/error-message mismatches")]
 fn test_write_fast_read_error() {
     use std::os::unix::fs::PermissionsExt;
 
@@ -771,9 +768,9 @@ fn test_write_fast_read_error() {
     ucmd.arg("foo").fails().stderr_contains("Permission denied");
 }
 
+#[uutests::wasi_ignore(NonUtf8ArgsUnsupported)]
 #[test]
 #[cfg(target_os = "linux")]
-#[cfg_attr(wasi_runner, ignore = "WASI: argv/filenames must be valid UTF-8")]
 fn test_cat_non_utf8_paths() {
     use std::ffi::OsStr;
     use std::os::unix::ffi::OsStrExt;
@@ -796,9 +793,9 @@ fn test_cat_non_utf8_paths() {
     assert_eq!(output, "Hello, non-UTF-8 world!\n");
 }
 
+#[uutests::wasi_ignore(CannotDetectUnsafeOverwrite)]
 #[test]
 #[cfg(unix)]
-#[cfg_attr(wasi_runner, ignore = "WASI: cannot detect unsafe overwrite")]
 fn test_appending_same_input_output() {
     let (at, mut ucmd) = at_and_ucmd!();
 

tests/by-util/test_comm.rs

@@ -677,8 +677,8 @@ fn test_comm_eintr_handling() {
         .stdout_contains("line3");
 }
 
+#[uutests::wasi_ignore(NonUtf8ArgsUnsupported)]
 #[test]
-#[cfg_attr(wasi_runner, ignore = "WASI: argv/filenames must be valid UTF-8")]
 fn test_output_lossy_utf8() {
     let scene = TestScenario::new(util_name!());
     let at = &scene.fixtures;
@@ -702,9 +702,9 @@ fn test_output_lossy_utf8() {
         .stdout_is_bytes(b"\xfe\n\t\t\xff\n\t\xfe\n");
 }
 
+#[uutests::wasi_ignore(HostPathsNotVisible)]
 #[test]
 #[cfg(any(target_os = "linux", target_os = "android"))]
-#[cfg_attr(wasi_runner, ignore = "WASI sandbox: host paths not visible")]
 fn test_comm_anonymous_pipes() {
     use std::fmt::Write as _;
     use std::{io::Write, os::fd::AsRawFd, process};
@@ -742,9 +742,9 @@ fn test_comm_anonymous_pipes() {
         .stdout_is("99999\n");
 }
 
+#[uutests::wasi_ignore(HostPathsNotVisible)]
 #[test]
 #[cfg(all(target_os = "linux", not(target_env = "musl")))]
-#[cfg_attr(wasi_runner, ignore = "WASI sandbox: host paths not visible")]
 fn test_read_error() {
     new_ucmd!()
         .arg("/proc/self/mem")
@@ -759,9 +759,9 @@ fn test_read_error() {
         .stderr_contains("comm: /proc/self/mem: Input/output error");
 }
 
+#[uutests::wasi_ignore(NoLocaleData)]
 #[test]
 #[cfg(unix)]
-#[cfg_attr(wasi_runner, ignore = "WASI: the guest does not inherit LC_ALL")]
 fn test_locale_collation() {
     // In a UTF-8 locale the collation puts `a1` before `a-b` and the byte order
     // puts them the other way round. Reading a file `sort` produced with a byte