Hash the sysroot contents

Author: saethlin

Cargo.lock

@@ -12,3 +12,4 @@ description = "Build a rustc sysroot with custom flags"
 tempfile = "3"
 rustc_version = "0.4"
 anyhow = "1.0"
+walkdir = "2.4"

Cargo.toml

@@ -7,14 +7,16 @@
 use std::collections::hash_map::DefaultHasher;
 use std::env;
 use std::ffi::{OsStr, OsString};
-use std::fs;
+use std::fs::{self, File};
 use std::hash::{Hash, Hasher};
+use std::io::{BufRead, BufReader};
 use std::ops::Not;
 use std::path::{Path, PathBuf};
 use std::process::Command;
 
 use anyhow::{bail, Context, Result};
 use tempfile::TempDir;
+use walkdir::WalkDir;
 
 /// Returns where the given rustc stores its sysroot source code.
 pub fn rustc_sysroot_src(mut rustc: Command) -> Result<PathBuf> {
@@ -84,6 +86,30 @@ fn make_writeable(p: &Path) -> Result<()> {
     Ok(())
 }
 
+/// Hash the contents of every file in a directory, recursively.
+pub fn hash_recursive(path: &Path, hasher: &mut DefaultHasher) -> Result<()> {
+    for entry in WalkDir::new(path).follow_links(true).into_iter() {
+        let entry = entry?;
+        // WalkDir yields the directories as well, and File::open will succeed on them. The
+        // reliable way to distinguish directories here is to check explicitly.
+        if entry.file_type().is_dir() {
+            continue;
+        }
+        let file = File::open(entry.path())?;
+        let mut reader = BufReader::new(file);
+        while let Ok(buf) = reader.fill_buf() {
+            if buf.is_empty() {
+                // EOF is not an error, we just get an empty buffer.
+                break;
+            }
+            hasher.write(buf);
+            let len = buf.len();
+            reader.consume(len);
+        }
+    }
+    Ok(())
+}
+
 /// The build mode to use for this sysroot.
 #[derive(Copy, Clone, Debug, PartialEq, Eq, Hash)]
 pub enum BuildMode {
@@ -220,18 +246,16 @@ impl SysrootBuilder {
         &self,
         src_dir: &Path,
         rustc_version: &rustc_version::VersionMeta,
-    ) -> u64 {
+    ) -> Result<u64> {
         let mut hasher = DefaultHasher::new();
 
-        // For now, we just hash in the information we have in `self`.
-        // Ideally we'd recursively hash the entire folder but that sounds slow?
-        src_dir.hash(&mut hasher);
+        hash_recursive(src_dir, &mut hasher)?;
         self.config.hash(&mut hasher);
         self.mode.hash(&mut hasher);
         self.rustflags.hash(&mut hasher);
         rustc_version.hash(&mut hasher);
 
-        hasher.finish()
+        Ok(hasher.finish())
     }
 
     fn sysroot_read_hash(&self) -> Option<u64> {
@@ -372,7 +396,7 @@ path = "lib.rs"
         }
 
         // Check if we even need to do anything.
-        let cur_hash = self.sysroot_compute_hash(src_dir, &rustc_version);
+        let cur_hash = self.sysroot_compute_hash(src_dir, &rustc_version)?;
         if self.sysroot_read_hash() == Some(cur_hash) {
             // Already done!
             return Ok(());