fix: fspy on intel mac (#54)

# fix fspy on intel mac

This PR adds:

- A test for shebang script execution in fspy to ensure proper tracking of file accesses
- Support for macOS x86_64 in CI workflows
- Updates the Oils for Unix binary URLs and checksums to version 0.37.0
- Updates uutils/coreutils to version 0.4.0
- Fixes scandir symbol linking on macOS x86_64 architecture

Author: branchseer

.github/workflows/ci.yml

@@ -37,6 +37,8 @@ jobs:
             target: x86_64-pc-windows-msvc
           - os: namespace-profile-mac-default
             target: aarch64-apple-darwin
+          - os: namespace-profile-mac-default
+            target: x86_64-apple-darwin
     runs-on: ${{ matrix.os }}
     steps:
       - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
@@ -49,17 +51,27 @@ jobs:
           save-cache: ${{ github.ref_name == 'main' }}
           cache-key: test
 
+      - run: rustup target add ${{ matrix.target }}
+
       - run: rustup target add x86_64-unknown-linux-musl
         if: ${{ matrix.os == 'ubuntu-latest' }}
 
       - run: pip install cargo-zigbuild
         if: ${{ matrix.os == 'ubuntu-latest' }}
 
-      - run: cargo check --all-targets --all-features
+      - run: cargo check --all-targets --all-features --target ${{ matrix.target }}
         env:
           RUSTFLAGS: '-D warnings --cfg tokio_unstable' # also update .cargo/config.toml
 
-      - run: cargo test
+      # fspy contains tests for Node.js fs accesses.
+      # use x86_64 node for x86_64-apple-darwin target.
+      - uses: actions/setup-node@395ad3262231945c25e8478fd5baf05154b1d79f # v6.1.0
+        with:
+          node-version-file: .node-version
+          architecture: x64
+        if: ${{ matrix.target == 'x86_64-apple-darwin' }}
+
+      - run: cargo test --target ${{ matrix.target }}
         if: ${{ matrix.os != 'ubuntu-latest' }}
 
       - run: cargo-zigbuild test --target x86_64-unknown-linux-gnu.2.17

crates/fspy/Cargo.toml

@@ -19,7 +19,7 @@ ouroboros = { workspace = true }
 rand = { workspace = true }
 tempfile = { workspace = true }
 thiserror = { workspace = true }
-tokio = { workspace = true, features = ["net", "process", "io-util", "sync"] }
+tokio = { workspace = true, features = ["net", "process", "io-util", "sync", "rt"] }
 which = { workspace = true, features = ["tracing"] }
 xxhash-rust = { workspace = true }
 

crates/fspy/build.rs

@@ -55,29 +55,29 @@ const MACOS_BINARY_DOWNLOADS: &[(&str, &[(&str, &str, u128)])] = &[
         "aarch64",
         &[
             (
-                "https://github.com/branchseer/oils-for-unix-binaries/releases/download/0.29.0-manual/oils-for-unix-0.29.0-aarch64-apple-darwin.tar.gz",
+                "https://github.com/branchseer/oils-for-unix-build/releases/download/oils-for-unix-0.37.0/oils-for-unix-0.37.0-darwin-arm64.tar.gz",
                 "oils-for-unix",
-                149_945_237_112_824_769_531_360_595_981_178_091_193,
+                282073174065923237490435663309538399576,
             ),
             (
-                "https://github.com/uutils/coreutils/releases/download/0.1.0/coreutils-0.1.0-aarch64-apple-darwin.tar.gz",
-                "coreutils-0.1.0-aarch64-apple-darwin/coreutils",
-                255_656_813_290_649_147_736_009_964_224_176_006_890,
+                "https://github.com/uutils/coreutils/releases/download/0.4.0/coreutils-0.4.0-aarch64-apple-darwin.tar.gz",
+                "coreutils-0.4.0-aarch64-apple-darwin/coreutils",
+                35998406686137668997937014088186935383,
             ),
         ],
     ),
     (
         "x86_64",
         &[
             (
-                "https://github.com/branchseer/oils-for-unix-binaries/releases/download/0.29.0-manual/oils-for-unix-0.29.0-x86_64-apple-darwin.tar.gz",
+                "https://github.com/branchseer/oils-for-unix-build/releases/download/oils-for-unix-0.37.0/oils-for-unix-0.37.0-darwin-x86_64.tar.gz",
                 "oils-for-unix",
-                286_203_014_616_009_968_685_843_701_528_129_413_859,
+                142673558272427867831039361796426010330,
             ),
             (
-                "https://github.com/uutils/coreutils/releases/download/0.1.0/coreutils-0.1.0-x86_64-apple-darwin.tar.gz",
-                "coreutils-0.1.0-x86_64-apple-darwin/coreutils",
-                75_344_743_234_387_926_348_628_744_659_874_018_387,
+                "https://github.com/uutils/coreutils/releases/download/0.4.0/coreutils-0.4.0-x86_64-apple-darwin.tar.gz",
+                "coreutils-0.4.0-x86_64-apple-darwin/coreutils",
+                120898281113671104995723556995187526689,
             ),
         ],
     ),
@@ -87,6 +87,7 @@ fn fetch_macos_binaries() -> anyhow::Result<()> {
     if env::var("CARGO_CFG_TARGET_OS").unwrap() != "macos" {
         return Ok(());
     }
+
     let out_dir = current_dir().unwrap().join(Path::new(&std::env::var_os("OUT_DIR").unwrap()));
 
     let target_arch = env::var("CARGO_CFG_TARGET_ARCH").unwrap();

crates/fspy/tests/shebang.rs

@@ -0,0 +1,44 @@
+#![cfg(unix)]
+
+mod test_utils;
+
+use std::{
+    os::unix::fs::PermissionsExt,
+    path::Path,
+    process::{Command, Stdio},
+};
+
+use fspy::AccessMode;
+use test_log::test;
+use test_utils::assert_contains;
+use tokio::fs;
+
+#[test(tokio::test)]
+async fn spawn_sh_shebang() -> anyhow::Result<()> {
+    let tmp_dir = tempfile::TempDir::new()?;
+
+    let shebang_script_path = tmp_dir.path().join("fspy_test_shebang_script.sh");
+    let shebang_script_path = shebang_script_path.into_os_string().into_string().unwrap();
+
+    fs::write(&shebang_script_path, "#!/bin/sh\ncat hello\n").await?;
+
+    let mut perms = fs::metadata(&shebang_script_path).await?.permissions();
+    perms.set_mode(0o755);
+    fs::set_permissions(&shebang_script_path, perms).await?;
+
+    let accesses = track_child!(shebang_script_path.clone(), |shebang_script_path: String| {
+        let _ignored = Command::new(&shebang_script_path)
+            .current_dir("/")
+            .stdin(Stdio::null())
+            .stdout(Stdio::piped())
+            .stderr(Stdio::piped())
+            .status()
+            .expect("Failed to execute shebang script");
+    })
+    .await?;
+
+    assert_contains(&accesses, Path::new(&shebang_script_path), AccessMode::READ);
+    assert_contains(&accesses, Path::new("/hello"), AccessMode::READ);
+
+    Ok(())
+}

crates/fspy_preload_unix/src/libc.rs

@@ -1,6 +1,10 @@
 pub use libc::*;
 
 unsafe extern "C" {
+    // On macOS x86_64, directory functions use $INODE64 symbol suffix for 64-bit inode support.
+    // On arm64, 64-bit inodes are the only option so no suffix is needed.
+    // https://github.com/apple-open-source-mirror/Libc/blob/5e566be7a7047360adfb35ffc44c6a019a854bea/include/dirent.h#L198
+    #[cfg_attr(all(target_os = "macos", target_arch = "x86_64"), link_name = "scandir$INODE64")]
     pub unsafe fn scandir(
         dirname: *const c_char,
         namelist: *mut c_void,
@@ -9,6 +13,7 @@ unsafe extern "C" {
     ) -> c_int;
 
     #[cfg(target_os = "macos")]
+    #[cfg_attr(target_arch = "x86_64", link_name = "scandir_b$INODE64")]
     pub unsafe fn scandir_b(
         dirname: *const c_char,
         namelist: *mut c_void,

crates/fspy_shared_unix/src/spawn/macos.rs

@@ -75,17 +75,18 @@ pub fn handle_exec(
 }
 
 pub static COREUTILS_FUNCTIONS: Set<&'static [u8]> = phf_set! {
-    b"[", b"arch", b"b2sum", b"b3sum", b"base32", b"base64", b"basename", b"basenc",
-    b"cat", b"chgrp", b"chmod", b"chown", b"chroot", b"cksum", b"comm", b"cp", b"csplit",
-    b"cut", b"date", b"dd", b"df", b"dir", b"dircolors", b"dirname", b"du", b"echo", b"env",
-    b"expand", b"expr", b"factor", b"false", b"fmt", b"fold", b"groups", b"hashsum", b"head",
-    b"hostid", b"hostname", b"id", b"install", b"join", b"kill", b"link", b"ln", b"logname",
-    b"ls", b"md5sum", b"mkdir", b"mkfifo", b"mknod", b"mktemp", b"more", b"mv", b"nice", b"nl",
-    b"nohup", b"nproc", b"numfmt", b"od", b"paste", b"pathchk", b"pinky", b"pr", b"printenv",
-    b"printf", b"ptx", b"pwd", b"readlink", b"realpath", b"rm", b"rmdir", b"seq", b"sha1sum",
-    b"sha224sum", b"sha256sum", b"sha3-224sum", b"sha3-256sum", b"sha3-384sum", b"sha3-512sum",
-    b"sha384sum", b"sha3sum", b"sha512sum", b"shake128sum", b"shake256sum", b"shred", b"shuf",
-    b"sleep", b"sort", b"split", b"stat", b"stdbuf", b"stty", b"sum", b"sync", b"tac", b"tail",
-    b"tee", b"test", b"timeout", b"touch", b"tr", b"true", b"truncate", b"tsort", b"tty", b"uname",
-    b"unexpand", b"uniq", b"unlink", b"uptime", b"users", b"vdir", b"wc", b"who", b"whoami", b"yes",
+    b"[", b"arch", b"b2sum", b"base32", b"base64", b"basename", b"basenc", b"cat",
+    b"chgrp", b"chmod", b"chown", b"chroot", b"cksum", b"comm", b"cp", b"csplit",
+    b"cut", b"date", b"dd", b"df", b"dir", b"dircolors", b"dirname", b"du", b"echo",
+    b"env", b"expand", b"expr", b"factor", b"false", b"fmt", b"fold", b"groups",
+    b"hashsum", b"head", b"hostid", b"hostname", b"id", b"install", b"join", b"kill",
+    b"link", b"ln", b"logname", b"ls", b"md5sum", b"mkdir", b"mkfifo", b"mknod",
+    b"mktemp", b"more", b"mv", b"nice", b"nl", b"nohup", b"nproc", b"numfmt", b"od",
+    b"paste", b"pathchk", b"pinky", b"pr", b"printenv", b"printf", b"ptx", b"pwd",
+    b"readlink", b"realpath", b"rm", b"rmdir", b"seq", b"sha1sum", b"sha224sum",
+    b"sha256sum", b"sha384sum", b"sha512sum", b"shred", b"shuf", b"sleep", b"sort",
+    b"split", b"stat", b"stdbuf", b"stty", b"sum", b"sync", b"tac", b"tail", b"tee",
+    b"test", b"timeout", b"touch", b"tr", b"true", b"truncate", b"tsort", b"tty",
+    b"uname", b"unexpand", b"uniq", b"unlink", b"uptime", b"users", b"vdir", b"wc",
+    b"who", b"whoami", b"yes"
 };