Stable hash

Cargo.toml

@@ -78,6 +78,7 @@ pulldown-cmark = { version = "0.11.0", default-features = false, features = ["ht
 rand = "0.8.5"
 regex = "1.10.4"
 rusqlite = { version = "0.31.0", features = ["bundled"] }
+rustc-stable-hash = { git = "https://github.com/rust-lang/rustc-stable-hash.git", rev = "3805516b78c7b2946ae2071d71ffc8235399652d" }
 rustfix = { version = "0.8.2", path = "crates/rustfix" }
 same-file = "1.0.6"
 security-framework = "2.10.0"
@@ -182,6 +183,7 @@ pathdiff.workspace = true
 rand.workspace = true
 regex.workspace = true
 rusqlite.workspace = true
+rustc-stable-hash.workspace = true
 rustfix.workspace = true
 same-file.workspace = true
 semver.workspace = true

src/cargo/core/compiler/build_runner/compilation_files.rs

@@ -650,7 +650,7 @@ fn compute_metadata(
     unit.is_std.hash(&mut hasher);
 
     MetaInfo {
-        meta_hash: Metadata(hasher.finish()),
+        meta_hash: Metadata(Hasher::finish(&hasher)),
         use_extra_filename: should_use_metadata(bcx, unit),
     }
 }

src/cargo/core/compiler/compile_kind.rs

@@ -195,6 +195,6 @@ impl CompileTarget {
                 self.name.hash(&mut hasher);
             }
         }
-        hasher.finish()
+        Hasher::finish(&hasher)
     }
 }

src/cargo/core/compiler/fingerprint/mod.rs

@@ -1482,7 +1482,7 @@ fn calculate_normal(
         local: Mutex::new(local),
         memoized_hash: Mutex::new(None),
         metadata,
-        config: config.finish(),
+        config: Hasher::finish(&config),
         compile_kind,
         rustflags: extra_flags,
         fs_status: FsStatus::Stale,

src/cargo/core/source_id.rs

@@ -786,70 +786,89 @@ mod tests {
     // Otherwise please just leave a comment in your PR as to why the hash value is
     // changing and why the old value can't be easily preserved.
     //
-    // The hash value depends on endianness and bit-width, so we only run this test on
-    // little-endian 64-bit CPUs (such as x86-64 and ARM64) where it matches the
-    // well-known value.
+    // The hash value should be stable across platforms, and doesn't depend on
+    // endianness and bit-width. One caveat is that absolute paths is inherently
+    // different on Windows than on Unix-like platforms. Unless we omit or strip
+    // the prefix components (e.g. `C:`), there is not way to have a
+    // cross-platform stable hash for absolute paths.
     #[test]
-    #[cfg(all(target_endian = "little", target_pointer_width = "64"))]
-    fn test_cratesio_hash() {
-        let gctx = GlobalContext::default().unwrap();
-        let crates_io = SourceId::crates_io(&gctx).unwrap();
-        assert_eq!(crate::util::hex::short_hash(&crates_io), "1ecc6299db9ec823");
-    }
-
-    // See the comment in `test_cratesio_hash`.
-    //
-    // Only test on non-Windows as paths on Windows will get different hashes.
-    #[test]
-    #[cfg(all(target_endian = "little", target_pointer_width = "64", not(windows)))]
     fn test_stable_hash() {
         use std::hash::Hasher;
         use std::path::Path;
 
+        use crate::util::StableHasher;
+
+        #[cfg(not(windows))]
+        let ws_root = Path::new("/tmp/ws");
+        #[cfg(windows)]
+        let ws_root = Path::new(r"C:\\tmp\ws");
+
         let gen_hash = |source_id: SourceId| {
-            let mut hasher = std::collections::hash_map::DefaultHasher::new();
-            source_id.stable_hash(Path::new("/tmp/ws"), &mut hasher);
-            hasher.finish()
+            let mut hasher = StableHasher::new();
+            source_id.stable_hash(ws_root, &mut hasher);
+            Hasher::finish(&hasher)
         };
 
+        let source_id = SourceId::crates_io(&GlobalContext::default().unwrap()).unwrap();
+        assert_eq!(gen_hash(source_id), 7062945687441624357);
+        assert_eq!(crate::util::hex::short_hash(&source_id), "25cdd57fae9f0462");
+
         let url = "https://my-crates.io".into_url().unwrap();
         let source_id = SourceId::for_registry(&url).unwrap();
-        assert_eq!(gen_hash(source_id), 18108075011063494626);
-        assert_eq!(crate::util::hex::short_hash(&source_id), "fb60813d6cb8df79");
+        assert_eq!(gen_hash(source_id), 8310250053664888498);
+        assert_eq!(crate::util::hex::short_hash(&source_id), "b2d65deb64f05373");
 
         let url = "https://your-crates.io".into_url().unwrap();
         let source_id = SourceId::for_alt_registry(&url, "alt").unwrap();
-        assert_eq!(gen_hash(source_id), 12862859764592646184);
-        assert_eq!(crate::util::hex::short_hash(&source_id), "09c10fd0cbd74bce");
+        assert_eq!(gen_hash(source_id), 14149534903000258933);
+        assert_eq!(crate::util::hex::short_hash(&source_id), "755952de063f5dc4");
 
         let url = "sparse+https://my-crates.io".into_url().unwrap();
         let source_id = SourceId::for_registry(&url).unwrap();
-        assert_eq!(gen_hash(source_id), 8763561830438022424);
-        assert_eq!(crate::util::hex::short_hash(&source_id), "d1ea0d96f6f759b5");
+        assert_eq!(gen_hash(source_id), 16249512552851930162);
+        assert_eq!(crate::util::hex::short_hash(&source_id), "327cfdbd92dd81e1");
 
         let url = "sparse+https://your-crates.io".into_url().unwrap();
         let source_id = SourceId::for_alt_registry(&url, "alt").unwrap();
-        assert_eq!(gen_hash(source_id), 5159702466575482972);
-        assert_eq!(crate::util::hex::short_hash(&source_id), "135d23074253cb78");
+        assert_eq!(gen_hash(source_id), 6156697384053352292);
+        assert_eq!(crate::util::hex::short_hash(&source_id), "64a713b6a6fb7055");
 
         let url = "file:///tmp/ws/crate".into_url().unwrap();
         let source_id = SourceId::for_git(&url, GitReference::DefaultBranch).unwrap();
-        assert_eq!(gen_hash(source_id), 15332537265078583985);
-        assert_eq!(crate::util::hex::short_hash(&source_id), "73a808694abda756");
-
-        let path = Path::new("/tmp/ws/crate");
+        assert_eq!(gen_hash(source_id), 473480029881867801);
+        assert_eq!(crate::util::hex::short_hash(&source_id), "199e591d94239206");
 
+        let path = &ws_root.join("crate");
         let source_id = SourceId::for_local_registry(path).unwrap();
-        assert_eq!(gen_hash(source_id), 18446533307730842837);
-        assert_eq!(crate::util::hex::short_hash(&source_id), "52a84cc73f6fd48b");
+        #[cfg(not(windows))]
+        {
+            assert_eq!(gen_hash(source_id), 11515846423845066584);
+            assert_eq!(crate::util::hex::short_hash(&source_id), "58d73c154f81d09f");
+        }
+        #[cfg(windows)]
+        {
+            assert_eq!(gen_hash(source_id), 6146331155906064276);
+            assert_eq!(crate::util::hex::short_hash(&source_id), "946fb2239f274c55");
+        }
 
         let source_id = SourceId::for_path(path).unwrap();
-        assert_eq!(gen_hash(source_id), 8764714075439899829);
-        assert_eq!(crate::util::hex::short_hash(&source_id), "e1ddd48578620fc1");
+        assert_eq!(gen_hash(source_id), 215644081443634269);
+        #[cfg(not(windows))]
+        assert_eq!(crate::util::hex::short_hash(&source_id), "64bace89c92b101f");
+        #[cfg(windows)]
+        assert_eq!(crate::util::hex::short_hash(&source_id), "01e1e6c391813fb6");
 
         let source_id = SourceId::for_directory(path).unwrap();
-        assert_eq!(gen_hash(source_id), 17459999773908528552);
-        assert_eq!(crate::util::hex::short_hash(&source_id), "6568fe2c2fab5bfe");
+        #[cfg(not(windows))]
+        {
+            assert_eq!(gen_hash(source_id), 6127590343904940368);
+            assert_eq!(crate::util::hex::short_hash(&source_id), "505191d1f3920955");
+        }
+        #[cfg(windows)]
+        {
+            assert_eq!(gen_hash(source_id), 10423446877655960172);
+            assert_eq!(crate::util::hex::short_hash(&source_id), "6c8ad69db585a790");
+        }
     }
 
     #[test]

src/cargo/ops/cargo_compile/mod.rs

@@ -648,7 +648,7 @@ fn traverse_and_share(
         .collect();
     // Here, we have recursively traversed this unit's dependencies, and hashed them: we can
     // finalize the dep hash.
-    let new_dep_hash = dep_hash.finish();
+    let new_dep_hash = Hasher::finish(&dep_hash);
 
     // This is the key part of the sharing process: if the unit is a runtime dependency, whose
     // target is the same as the host, we canonicalize the compile kind to `CompileKind::Host`.

src/cargo/sources/registry/mod.rs

@@ -456,7 +456,7 @@ fn short_name(id: SourceId, is_shallow: bool) -> String {
     // CAUTION: This should not change between versions. If you change how
     // this is computed, it will orphan previously cached data, forcing the
     // cache to be rebuilt and potentially wasting significant disk space. If
-    // you change it, be cautious of the impact. See `test_cratesio_hash` for
+    // you change it, be cautious of the impact. See `test_stable_hash` for
     // a similar discussion.
     let hash = hex::short_hash(&id);
     let ident = id.url().host_str().unwrap_or("").to_string();

src/cargo/util/hasher.rs

@@ -1,24 +1,5 @@
-//! Implementation of a hasher that produces the same values across releases.
+//! A hasher that produces the same values across releases and platforms.
 //!
-//! The hasher should be fast and have a low chance of collisions (but is not
-//! sufficient for cryptographic purposes).
-#![allow(deprecated)]
+//! This is a wrapper around [`rustc_stable_hash::StableHasher`].
 
-use std::hash::{Hasher, SipHasher};
-
-pub struct StableHasher(SipHasher);
-
-impl StableHasher {
-    pub fn new() -> StableHasher {
-        StableHasher(SipHasher::new())
-    }
-}
-
-impl Hasher for StableHasher {
-    fn finish(&self) -> u64 {
-        self.0.finish()
-    }
-    fn write(&mut self, bytes: &[u8]) {
-        self.0.write(bytes)
-    }
-}
+pub use rustc_stable_hash::StableSipHasher128 as StableHasher;

src/cargo/util/hex.rs

@@ -10,7 +10,7 @@ pub fn to_hex(num: u64) -> String {
 pub fn hash_u64<H: Hash>(hashable: H) -> u64 {
     let mut hasher = StableHasher::new();
     hashable.hash(&mut hasher);
-    hasher.finish()
+    Hasher::finish(&hasher)
 }
 
 pub fn hash_u64_file(mut file: &File) -> std::io::Result<u64> {
@@ -23,7 +23,7 @@ pub fn hash_u64_file(mut file: &File) -> std::io::Result<u64> {
         }
         hasher.write(&buf[..n]);
     }
-    Ok(hasher.finish())
+    Ok(Hasher::finish(&hasher))
 }
 
 pub fn short_hash<H: Hash>(hashable: &H) -> String {

src/cargo/util/rustc.rs

@@ -374,7 +374,7 @@ fn rustc_fingerprint(
         _ => (),
     }
 
-    Ok(hasher.finish())
+    Ok(Hasher::finish(&hasher))
 }
 
 fn process_fingerprint(cmd: &ProcessBuilder, extra_fingerprint: u64) -> u64 {
@@ -384,5 +384,5 @@ fn process_fingerprint(cmd: &ProcessBuilder, extra_fingerprint: u64) -> u64 {
     let mut env = cmd.get_envs().iter().collect::<Vec<_>>();
     env.sort_unstable();
     env.hash(&mut hasher);
-    hasher.finish()
+    Hasher::finish(&hasher)
 }