Create the previous dep graph index on a background thread

Author: Zoxc

compiler/rustc_incremental/src/persist/load.rs

@@ -11,6 +11,7 @@ use rustc_session::config::IncrementalStateAssertion;
 use rustc_session::{Session, StableCrateId};
 use rustc_span::{ErrorGuaranteed, Symbol};
 use std::path::{Path, PathBuf};
+use std::sync::Arc;
 
 use super::data::*;
 use super::file_format;
@@ -95,7 +96,7 @@ fn delete_dirty_work_product(sess: &Session, swp: SerializedWorkProduct) {
     work_product::delete_workproduct_files(sess, &swp.work_product);
 }
 
-fn load_dep_graph(sess: &Session) -> LoadResult<(SerializedDepGraph, WorkProductMap)> {
+fn load_dep_graph(sess: &Session) -> LoadResult<(Arc<SerializedDepGraph>, WorkProductMap)> {
     let prof = sess.prof.clone();
 
     if sess.opts.incremental.is_none() {
@@ -169,7 +170,7 @@ fn load_dep_graph(sess: &Session) -> LoadResult<(SerializedDepGraph, WorkProduct
                 return LoadResult::DataOutOfDate;
             }
 
-            let dep_graph = SerializedDepGraph::decode::<DepsType>(&mut decoder);
+            let dep_graph = SerializedDepGraph::decode::<DepsType>(&mut decoder, sess);
 
             LoadResult::Ok { data: (dep_graph, prev_work_products) }
         }

compiler/rustc_incremental/src/persist/save.rs

@@ -9,6 +9,7 @@ use rustc_serialize::opaque::{FileEncodeResult, FileEncoder};
 use rustc_serialize::Encodable as RustcEncodable;
 use rustc_session::Session;
 use std::fs;
+use std::sync::Arc;
 
 use super::data::*;
 use super::dirty_clean;
@@ -149,7 +150,7 @@ fn encode_query_cache(tcx: TyCtxt<'_>, encoder: FileEncoder) -> FileEncodeResult
 /// and moves it to the permanent dep-graph path
 pub(crate) fn build_dep_graph(
     sess: &Session,
-    prev_graph: SerializedDepGraph,
+    prev_graph: Arc<SerializedDepGraph>,
     prev_work_products: WorkProductMap,
 ) -> Option<DepGraph> {
     if sess.opts.incremental.is_none() {

compiler/rustc_query_system/src/dep_graph/graph.rs

@@ -14,6 +14,7 @@ use std::fmt::Debug;
 use std::hash::Hash;
 use std::marker::PhantomData;
 use std::sync::atomic::Ordering::Relaxed;
+use std::sync::Arc;
 
 use super::query::DepGraphQuery;
 use super::serialized::{GraphEncoder, SerializedDepGraph, SerializedDepNodeIndex};
@@ -83,7 +84,7 @@ pub struct DepGraphData<D: Deps> {
 
     /// The dep-graph from the previous compilation session. It contains all
     /// nodes and edges as well as all fingerprints of nodes that have them.
-    previous: SerializedDepGraph,
+    previous: Arc<SerializedDepGraph>,
 
     colors: DepNodeColorMap,
 
@@ -115,7 +116,7 @@ where
 impl<D: Deps> DepGraph<D> {
     pub fn new(
         profiler: &SelfProfilerRef,
-        prev_graph: SerializedDepGraph,
+        prev_graph: Arc<SerializedDepGraph>,
         prev_work_products: WorkProductMap,
         encoder: FileEncoder,
         record_graph: bool,

compiler/rustc_query_system/src/dep_graph/serialized.rs

@@ -41,14 +41,17 @@ use crate::dep_graph::EdgesVec;
 use rustc_data_structures::fingerprint::Fingerprint;
 use rustc_data_structures::fingerprint::PackedFingerprint;
 use rustc_data_structures::fx::FxHashMap;
+use rustc_data_structures::outline;
 use rustc_data_structures::profiling::SelfProfilerRef;
 use rustc_data_structures::sync::Lock;
 use rustc_data_structures::unhash::UnhashMap;
 use rustc_index::{Idx, IndexVec};
 use rustc_serialize::opaque::{FileEncodeResult, FileEncoder, IntEncodedWithFixedSize, MemDecoder};
 use rustc_serialize::{Decodable, Decoder, Encodable, Encoder};
-use std::iter;
+use rustc_session::Session;
 use std::marker::PhantomData;
+use std::sync::{Arc, OnceLock};
+use std::{iter, thread};
 
 // The maximum value of `SerializedDepNodeIndex` leaves the upper two bits
 // unused so that we can store multiple index types in `CompressedHybridIndex`,
@@ -69,23 +72,33 @@ const DEP_NODE_PAD: usize = DEP_NODE_SIZE - 1;
 const DEP_NODE_WIDTH_BITS: usize = DEP_NODE_SIZE / 2;
 
 /// Data for use when recompiling the **current crate**.
-#[derive(Debug)]
 pub struct SerializedDepGraph {
     /// The set of all DepNodes in the graph
     nodes: IndexVec<SerializedDepNodeIndex, DepNode>,
+
     /// The set of all Fingerprints in the graph. Each Fingerprint corresponds to
     /// the DepNode at the same index in the nodes vector.
     fingerprints: IndexVec<SerializedDepNodeIndex, Fingerprint>,
+
     /// For each DepNode, stores the list of edges originating from that
     /// DepNode. Encoded as a [start, end) pair indexing into edge_list_data,
     /// which holds the actual DepNodeIndices of the target nodes.
     edge_list_indices: IndexVec<SerializedDepNodeIndex, EdgeHeader>,
+
     /// A flattened list of all edge targets in the graph, stored in the same
     /// varint encoding that we use on disk. Edge sources are implicit in edge_list_indices.
     edge_list_data: Vec<u8>,
+
     /// Stores a map from fingerprints to nodes per dep node kind.
-    /// This is the reciprocal of `nodes`.
-    index: Vec<UnhashMap<PackedFingerprint, SerializedDepNodeIndex>>,
+    /// This is the reciprocal of `nodes`. This is computed on demand for each dep kind.
+    /// The entire index is also computed in a background thread.
+    index: Vec<OnceLock<UnhashMap<PackedFingerprint, SerializedDepNodeIndex>>>,
+
+    /// Stores the number of node for each dep node kind.
+    index_sizes: Vec<usize>,
+
+    /// A profiler reference for used in the index prefetching thread.
+    prof: SelfProfilerRef,
 }
 
 impl Default for SerializedDepGraph {
@@ -96,6 +109,8 @@ impl Default for SerializedDepGraph {
             edge_list_indices: Default::default(),
             edge_list_data: Default::default(),
             index: Default::default(),
+            index_sizes: Default::default(),
+            prof: SelfProfilerRef::new(None, None),
         }
     }
 }
@@ -138,9 +153,35 @@ impl SerializedDepGraph {
         self.nodes[dep_node_index]
     }
 
+    /// This computes and sets up the index for just the specified `DepKind`.
+    fn setup_index(&self, dep_kind: DepKind) {
+        let _timer = self.prof.generic_activity("incr_comp_dep_graph_setup_index");
+
+        let mut index = UnhashMap::with_capacity_and_hasher(
+            self.index_sizes[dep_kind.as_usize()],
+            Default::default(),
+        );
+
+        for (idx, node) in self.nodes.iter_enumerated() {
+            if node.kind == dep_kind {
+                index.insert(node.hash, idx);
+            }
+        }
+
+        // This may race with the prefetching thread, but that will set the same value.
+        self.index[dep_kind.as_usize()].set(index).ok();
+    }
+
     #[inline]
     pub fn node_to_index_opt(&self, dep_node: &DepNode) -> Option<SerializedDepNodeIndex> {
-        self.index.get(dep_node.kind.as_usize())?.get(&dep_node.hash).cloned()
+        let index = self.index.get(dep_node.kind.as_usize())?;
+        let index = index.get().unwrap_or_else(|| {
+            outline(|| {
+                self.setup_index(dep_node.kind);
+                self.index[dep_node.kind.as_usize()].get().unwrap()
+            })
+        });
+        index.get(&dep_node.hash).cloned()
     }
 
     #[inline]
@@ -152,6 +193,31 @@ impl SerializedDepGraph {
     pub fn node_count(&self) -> usize {
         self.nodes.len()
     }
+
+    /// This spawns a thread that prefetches the index.
+    fn prefetch(self: &Arc<Self>) {
+        if !self.index.is_empty() {
+            let this = self.clone();
+            thread::spawn(move || {
+                let _timer = this.prof.generic_activity("incr_comp_prefetch_dep_graph_index");
+
+                let mut index: Vec<_> = this
+                    .index_sizes
+                    .iter()
+                    .map(|&n| UnhashMap::with_capacity_and_hasher(n, Default::default()))
+                    .collect();
+
+                for (idx, node) in this.nodes.iter_enumerated() {
+                    index[node.kind.as_usize()].insert(node.hash, idx);
+                }
+
+                for (i, index) in index.into_iter().enumerate() {
+                    // This may race with `setup_index`, but that will set the same value.
+                    this.index[i].set(index).ok();
+                }
+            });
+        }
+    }
 }
 
 /// A packed representation of an edge's start index and byte width.
@@ -185,8 +251,8 @@ fn mask(bits: usize) -> usize {
 }
 
 impl SerializedDepGraph {
-    #[instrument(level = "debug", skip(d))]
-    pub fn decode<D: Deps>(d: &mut MemDecoder<'_>) -> SerializedDepGraph {
+    #[instrument(level = "debug", skip(d, sess))]
+    pub fn decode<D: Deps>(d: &mut MemDecoder<'_>, sess: &Session) -> Arc<SerializedDepGraph> {
         // The last 16 bytes are the node count and edge count.
         debug!("position: {:?}", d.position());
         let (node_count, edge_count) =
@@ -253,16 +319,21 @@ impl SerializedDepGraph {
         // end of the array. This padding ensure it doesn't.
         edge_list_data.extend(&[0u8; DEP_NODE_PAD]);
 
-        // Read the number of each dep kind and use it to create an hash map with a suitable size.
-        let mut index: Vec<_> = (0..(D::DEP_KIND_MAX + 1))
-            .map(|_| UnhashMap::with_capacity_and_hasher(d.read_u32() as usize, Default::default()))
-            .collect();
-
-        for (idx, node) in nodes.iter_enumerated() {
-            index[node.kind.as_usize()].insert(node.hash, idx);
-        }
-
-        SerializedDepGraph { nodes, fingerprints, edge_list_indices, edge_list_data, index }
+        // Read the number of nodes for each dep kind.
+        let index_sizes: Vec<_> =
+            (0..(D::DEP_KIND_MAX + 1)).map(|_| d.read_u32() as usize).collect();
+
+        let result = Arc::new(SerializedDepGraph {
+            nodes,
+            fingerprints,
+            edge_list_indices,
+            edge_list_data,
+            index: (0..index_sizes.len()).map(|_| OnceLock::new()).collect(),
+            index_sizes,
+            prof: sess.prof.clone(),
+        });
+        result.prefetch();
+        result
     }
 }