self-profiling: Add events for everything except trait selection.

src/librustc/dep_graph/graph.rs

@@ -845,6 +845,8 @@ impl DepGraph {
     // This method will only load queries that will end up in the disk cache.
     // Other queries will not be executed.
     pub fn exec_cache_promotions(&self, tcx: TyCtxt<'_>) {
+        let _prof_timer = tcx.prof.generic_activity("incr_comp_query_cache_promotion");
+
         let data = self.data.as_ref().unwrap();
         for prev_index in data.colors.values.indices() {
             match data.colors.get(prev_index) {

src/librustc/hir/lowering.rs

@@ -242,6 +242,8 @@ pub fn lower_crate(
     // incr. comp. yet.
     dep_graph.assert_ignored();
 
+    let _prof_timer = sess.prof.generic_activity("hir_lowering");
+
     LoweringContext {
         crate_root: sess.parse_sess.injected_crate_name.try_get().copied(),
         sess,

src/librustc/hir/map/mod.rs

@@ -1222,6 +1222,8 @@ pub fn map_crate<'hir>(sess: &crate::session::Session,
                        forest: &'hir Forest,
                        definitions: &'hir Definitions)
                        -> Map<'hir> {
+    let _prof_timer = sess.prof.generic_activity("build_hir_map");
+
     // Build the reverse mapping of `node_to_hir_id`.
     let hir_to_node_id = definitions.node_to_hir_id.iter_enumerated()
         .map(|(node_id, &hir_id)| (hir_id, node_id)).collect();

src/librustc_incremental/persist/load.rs

@@ -101,6 +101,7 @@ pub fn load_dep_graph(sess: &Session) -> DepGraphFuture {
     // before we fire the background thread.
 
     let time_passes = sess.time_passes();
+    let prof = sess.prof.clone();
 
     if sess.opts.incremental.is_none() {
         // No incremental compilation.
@@ -161,6 +162,8 @@ pub fn load_dep_graph(sess: &Session) -> DepGraphFuture {
 
     MaybeAsync::Async(std::thread::spawn(move || {
         time_ext(time_passes, "background load prev dep-graph", move || {
+            let _prof_timer = prof.generic_activity("incr_comp_load_dep_graph");
+
             match load_data(report_incremental_info, &path) {
                 LoadResult::DataOutOfDate => LoadResult::DataOutOfDate,
                 LoadResult::Error { message } => LoadResult::Error { message },
@@ -198,6 +201,8 @@ pub fn load_query_result_cache(sess: &Session) -> OnDiskCache<'_> {
         return OnDiskCache::new_empty(sess.source_map());
     }
 
+    let _prof_timer = sess.prof.generic_activity("incr_comp_load_query_result_cache");
+
     match load_data(sess.opts.debugging_opts.incremental_info, &query_cache_path(sess)) {
         LoadResult::Ok{ data: (bytes, start_pos) } => OnDiskCache::new(sess, bytes, start_pos),
         _ => OnDiskCache::new_empty(sess.source_map())

src/librustc_incremental/persist/save.rs

@@ -241,6 +241,8 @@ fn encode_work_product_index(work_products: &FxHashMap<WorkProductId, WorkProduc
 
 fn encode_query_cache(tcx: TyCtxt<'_>, encoder: &mut Encoder) {
     time(tcx.sess, "serialize query result cache", || {
+        let _timer = tcx.prof.generic_activity("incr_comp_serialize_result_cache");
+
         tcx.serialize_query_result_cache(encoder).unwrap();
     })
 }

src/librustc_interface/passes.rs

@@ -250,6 +250,8 @@ pub fn register_plugins<'a>(
 
     if sess.opts.incremental.is_some() {
         time(sess, "garbage-collect incremental cache directory", || {
+            let _prof_timer =
+                sess.prof.generic_activity("incr_comp_garbage_collect_session_directories");
             if let Err(e) = rustc_incremental::garbage_collect_session_directories(sess) {
                 warn!(
                     "Error while trying to garbage collect incremental \

src/librustc_mir/monomorphize/collector.rs

@@ -285,7 +285,11 @@ pub fn collect_crate_mono_items(
     tcx: TyCtxt<'_>,
     mode: MonoItemCollectionMode,
 ) -> (FxHashSet<MonoItem<'_>>, InliningMap<'_>) {
+    let _prof_timer = tcx.prof.generic_activity("monomorphization_collector");
+
     let roots = time(tcx.sess, "collecting roots", || {
+        let _prof_timer = tcx.prof
+            .generic_activity("monomorphization_collector_root_collections");
         collect_roots(tcx, mode)
     });
 
@@ -295,6 +299,9 @@ pub fn collect_crate_mono_items(
     let mut inlining_map = MTLock::new(InliningMap::new());
 
     {
+        let _prof_timer = tcx.prof
+            .generic_activity("monomorphization_collector_graph_walk");
+
         let visited: MTRef<'_, _> = &mut visited;
         let inlining_map: MTRef<'_, _> = &mut inlining_map;
 

src/librustc_mir/monomorphize/partitioning.rs

@@ -134,10 +134,15 @@ pub fn partition<'tcx, I>(
 where
     I: Iterator<Item = MonoItem<'tcx>>,
 {
+    let _prof_timer = tcx.prof.generic_activity("cgu_partitioning");
+
     // In the first step, we place all regular monomorphizations into their
     // respective 'home' codegen unit. Regular monomorphizations are all
     // functions and statics defined in the local crate.
-    let mut initial_partitioning = place_root_mono_items(tcx, mono_items);
+    let mut initial_partitioning = {
+        let _prof_timer = tcx.prof.generic_activity("cgu_partitioning_place_roots");
+        place_root_mono_items(tcx, mono_items)
+    };
 
     initial_partitioning.codegen_units.iter_mut().for_each(|cgu| cgu.estimate_size(tcx));
 
@@ -146,8 +151,8 @@ where
     // If the partitioning should produce a fixed count of codegen units, merge
     // until that count is reached.
     if let PartitioningStrategy::FixedUnitCount(count) = strategy {
+        let _prof_timer = tcx.prof.generic_activity("cgu_partitioning_merge_cgus");
         merge_codegen_units(tcx, &mut initial_partitioning, count);
-
         debug_dump(tcx, "POST MERGING:", initial_partitioning.codegen_units.iter());
     }
 
@@ -155,8 +160,11 @@ where
     // monomorphizations have to go into each codegen unit. These additional
     // monomorphizations can be drop-glue, functions from external crates, and
     // local functions the definition of which is marked with `#[inline]`.
-    let mut post_inlining = place_inlined_mono_items(initial_partitioning,
-                                                            inlining_map);
+    let mut post_inlining = {
+        let _prof_timer =
+            tcx.prof.generic_activity("cgu_partitioning_place_inline_items");
+        place_inlined_mono_items(initial_partitioning, inlining_map)
+    };
 
     post_inlining.codegen_units.iter_mut().for_each(|cgu| cgu.estimate_size(tcx));
 
@@ -165,6 +173,8 @@ where
     // Next we try to make as many symbols "internal" as possible, so LLVM has
     // more freedom to optimize.
     if !tcx.sess.opts.cg.link_dead_code {
+        let _prof_timer =
+            tcx.prof.generic_activity("cgu_partitioning_internalize_symbols");
         internalize_symbols(tcx, &mut post_inlining, inlining_map);
     }
 

src/librustc_resolve/lib.rs

@@ -1255,6 +1255,9 @@ impl<'a> Resolver<'a> {
 
     /// Entry point to crate resolution.
     pub fn resolve_crate(&mut self, krate: &Crate) {
+        let _prof_timer =
+            self.session.prof.generic_activity("resolve_crate");
+
         ImportResolver { r: self }.finalize_imports();
         self.finalize_macro_resolutions();