perf(library-select): #236 parallel BFS walker + scan memoization + tracing spans

Cargo.toml

@@ -66,6 +66,8 @@ rusqlite = { version = "0.31", features = ["bundled"] }
 shell-words = "1"
 bincode = "1"
 zccache-artifact = "1.4.0"
+rayon = "1"
+tracing-test = "0.2"
 
 # Process containment: all subprocess spawns the daemon performs (compilers,
 # esptool, qemu, simavr, node, npm, …) and any grandchildren they fork must

crates/fbuild-header-scan/Cargo.toml

@@ -7,6 +7,8 @@ rust-version.workspace = true
 license.workspace = true
 
 [dependencies]
+rayon = { workspace = true }
+tracing = { workspace = true }
 
 [dev-dependencies]
 criterion = { workspace = true }

crates/fbuild-header-scan/src/lib.rs

@@ -10,7 +10,7 @@ mod scanner;
 mod walker;
 
 pub use scanner::{scan, IncludeKind, IncludeRef, Span};
-pub use walker::{walk, WalkResult};
+pub use walker::{walk, walk_with_state, WalkResult, WalkState};
 
 /// Bumped whenever the scanner output shape changes. Mixed into cache keys so a
 /// scanner change invalidates memoized library-selection results.

crates/fbuild-header-scan/src/walker.rs

@@ -5,56 +5,169 @@
 //! plus the set of include strings that could not be resolved. The walker is
 //! BFS over a visited set so cycles, diamonds, and arbitrary depth all
 //! terminate correctly.
+//!
+//! Two public entry points:
+//! * [`walk`] -- one-shot convenience wrapper that allocates a fresh
+//!   [`WalkState`] internally. `WalkResult::reached` is the full set of files
+//!   reached from `seeds`.
+//! * [`walk_with_state`] -- accepts a caller-owned [`WalkState`] so multiple
+//!   walks can share a scan cache and a `visited` set across calls (used by
+//!   `fbuild-library-select` to avoid re-reading files between LDF passes).
+//!   `WalkResult::reached` is the *delta* of canonical paths newly discovered
+//!   in this call; the union of deltas across calls equals the full set.
 
-use std::collections::{BTreeSet, HashSet, VecDeque};
+use std::collections::{BTreeSet, HashMap, HashSet, VecDeque};
 use std::path::{Path, PathBuf};
 
+use rayon::prelude::*;
+
 use crate::scanner::{scan, IncludeKind, IncludeRef};
 
 /// Result of a walk. `reached` and `unresolved` are sorted for deterministic
 /// cache keys.
+///
+/// For [`walk`] (fresh-state wrapper) `reached` is the full set of files
+/// transitively reached from the seeds. For [`walk_with_state`] the same
+/// fields contain only the *delta* added in this call -- files already
+/// present in the shared `WalkState::visited` set are not re-emitted.
 #[derive(Debug, Clone, Default, PartialEq, Eq)]
 pub struct WalkResult {
     pub reached: Vec<PathBuf>,
     pub unresolved: Vec<String>,
 }
 
+/// State that can be shared across multiple [`walk_with_state`] calls so the
+/// include-scan results are memoized and each on-disk file is read at most
+/// once for the lifetime of the state.
+///
+/// Used by `fbuild-library-select::resolve_with_stats` to share scan results
+/// across LDF passes -- pass 1 reads every file once, pass 2 re-seeds with
+/// library `.cpp` files but reuses the cached scans for everything already
+/// reached.
+#[derive(Debug, Default)]
+pub struct WalkState {
+    /// Canonical paths the walker has already enqueued/visited.
+    visited: HashSet<PathBuf>,
+    /// Canonical path -> parsed include list. Populated lazily on first read.
+    /// Missing entries mean either "not yet read" or "read failed" -- they are
+    /// indistinguishable here, matching the existing `let Ok(...) else
+    /// { continue }` semantics of the original walker.
+    scan_cache: HashMap<PathBuf, Vec<IncludeRef>>,
+    /// Number of successful `std::fs::read_to_string` invocations across the
+    /// lifetime of this state. Each unique file is counted exactly once
+    /// because subsequent walks hit `scan_cache` instead.
+    files_read: usize,
+}
+
+impl WalkState {
+    /// Create an empty state. No files have been scanned, nothing is visited.
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    /// Number of files physically read from disk so far. Used by
+    /// `resolve_with_stats` to assert the no-re-read contract in tests.
+    pub fn files_read(&self) -> usize {
+        self.files_read
+    }
+}
+
 /// Walk the include graph starting from `seeds` over `search_paths`.
 ///
 /// `search_paths` is consulted in order for `<...>` includes and as a
 /// secondary lookup for `"..."` includes (after the same-directory check).
 /// A file is added to `reached` exactly once. Files outside `search_paths`
 /// are still reached if they are seeds or `"..."`-resolved relative to a
 /// seed/visited file.
+///
+/// Allocates a fresh [`WalkState`] internally, so `WalkResult::reached`
+/// contains every file transitively reached from `seeds`.
 pub fn walk(seeds: &[PathBuf], search_paths: &[PathBuf]) -> WalkResult {
+    let mut state = WalkState::new();
+    walk_with_state(seeds, search_paths, &mut state)
+}
+
+/// Walk the include graph using a caller-owned [`WalkState`] so the scan cache
+/// and visited set persist across calls.
+///
+/// `WalkResult::reached` contains only the *delta* of canonical paths newly
+/// reached in this call. Files already in `state.visited` from a previous
+/// call are not re-emitted (and not re-read).
+///
+/// The BFS proceeds in waves: each wave reads all not-yet-cached files in
+/// parallel via rayon, then resolves every `#include` in every cached scan
+/// result to enqueue the next wave.
+#[tracing::instrument(
+    name = "ldf_walk",
+    skip_all,
+    fields(seeds = seeds.len(), search_paths = search_paths.len())
+)]
+pub fn walk_with_state(
+    seeds: &[PathBuf],
+    search_paths: &[PathBuf],
+    state: &mut WalkState,
+) -> WalkResult {
+    tracing::debug!(
+        seeds = seeds.len(),
+        search_paths = search_paths.len(),
+        "ldf_walk"
+    );
     let mut reached: BTreeSet<PathBuf> = BTreeSet::new();
     let mut unresolved: BTreeSet<String> = BTreeSet::new();
-    let mut visited: HashSet<PathBuf> = HashSet::new();
-    let mut queue: VecDeque<PathBuf> = VecDeque::new();
+    let mut frontier: VecDeque<PathBuf> = VecDeque::new();
 
     for seed in seeds {
         let canon = canon(seed);
-        if visited.insert(canon.clone()) {
-            queue.push_back(canon.clone());
+        if state.visited.insert(canon.clone()) {
+            frontier.push_back(canon.clone());
             reached.insert(canon);
         }
     }
 
-    while let Some(file) = queue.pop_front() {
-        let Ok(text) = std::fs::read_to_string(&file) else {
-            continue;
-        };
-        for inc in scan(&text) {
-            match resolve(&inc, &file, search_paths) {
-                Some(resolved) => {
-                    let canon = canon(&resolved);
-                    if visited.insert(canon.clone()) {
-                        reached.insert(canon.clone());
-                        queue.push_back(canon);
+    while !frontier.is_empty() {
+        // Read all not-yet-cached files in the current wave in parallel.
+        let to_read: Vec<PathBuf> = frontier
+            .iter()
+            .filter(|p| !state.scan_cache.contains_key(*p))
+            .cloned()
+            .collect();
+
+        if !to_read.is_empty() {
+            let scanned: Vec<(PathBuf, Vec<IncludeRef>)> = to_read
+                .par_iter()
+                .filter_map(|p| {
+                    let text = std::fs::read_to_string(p).ok()?;
+                    Some((p.clone(), scan(&text)))
+                })
+                .collect();
+
+            for (path, includes) in scanned {
+                state.scan_cache.insert(path, includes);
+                state.files_read += 1;
+            }
+        }
+
+        // Resolve includes for every file in the frontier and build the next
+        // wave from any newly discovered canonical paths.
+        let current: Vec<PathBuf> = frontier.drain(..).collect();
+        for file in &current {
+            let Some(includes) = state.scan_cache.get(file).cloned() else {
+                // Read failed (file is a directory, permission denied, etc.).
+                // Match the existing behavior: silently skip.
+                continue;
+            };
+            for inc in &includes {
+                match resolve_include(inc, file, search_paths) {
+                    Some(resolved) => {
+                        let canon = canon(&resolved);
+                        if state.visited.insert(canon.clone()) {
+                            reached.insert(canon.clone());
+                            frontier.push_back(canon);
+                        }
+                    }
+                    None => {
+                        unresolved.insert(inc.path.clone());
                     }
-                }
-                None => {
-                    unresolved.insert(inc.path.clone());
                 }
             }
         }
@@ -66,7 +179,7 @@ pub fn walk(seeds: &[PathBuf], search_paths: &[PathBuf]) -> WalkResult {
     }
 }
 
-fn resolve(inc: &IncludeRef, from: &Path, search_paths: &[PathBuf]) -> Option<PathBuf> {
+fn resolve_include(inc: &IncludeRef, from: &Path, search_paths: &[PathBuf]) -> Option<PathBuf> {
     if inc.kind == IncludeKind::Quoted {
         if let Some(parent) = from.parent() {
             let candidate = parent.join(&inc.path);

crates/fbuild-library-select/Cargo.toml

@@ -20,6 +20,7 @@ zccache-artifact = { workspace = true }
 tempfile = { workspace = true }
 criterion = { workspace = true }
 fbuild-test-support = { path = "../fbuild-test-support" }
+tracing-test = { workspace = true, features = ["no-env-filter"] }
 
 [[bench]]
 name = "resolve_cold"