coverage: Track used functions in a set instead of a map

This patch dismantles what was left of `FunctionCoverage` in `map_data.rs`,
replaces `function_coverage_map` with a set, and overhauls how we prepare
covfun records for unused functions.

compiler/rustc_codegen_llvm/src/coverageinfo/map_data.rs

@@ -1,26 +0,0 @@
-use rustc_middle::mir::coverage::{CoverageIdsInfo, FunctionCoverageInfo};
-
-pub(crate) struct FunctionCoverage<'tcx> {
-    #[expect(unused)] // This whole file gets deleted later in the same PR.
-    pub(crate) function_coverage_info: &'tcx FunctionCoverageInfo,
-    /// If `None`, the corresponding function is unused.
-    ids_info: Option<&'tcx CoverageIdsInfo>,
-}
-
-impl<'tcx> FunctionCoverage<'tcx> {
-    pub(crate) fn new_used(
-        function_coverage_info: &'tcx FunctionCoverageInfo,
-        ids_info: &'tcx CoverageIdsInfo,
-    ) -> Self {
-        Self { function_coverage_info, ids_info: Some(ids_info) }
-    }
-
-    pub(crate) fn new_unused(function_coverage_info: &'tcx FunctionCoverageInfo) -> Self {
-        Self { function_coverage_info, ids_info: None }
-    }
-
-    /// Returns true for a used (called) function, and false for an unused function.
-    pub(crate) fn is_used(&self) -> bool {
-        self.ids_info.is_some()
-    }
-}

compiler/rustc_codegen_llvm/src/coverageinfo/mapgen.rs

@@ -18,7 +18,6 @@ use tracing::debug;
 
 use crate::common::CodegenCx;
 use crate::coverageinfo::llvm_cov;
-use crate::coverageinfo::map_data::FunctionCoverage;
 use crate::coverageinfo::mapgen::covfun::prepare_covfun_record;
 use crate::llvm;
 
@@ -49,23 +48,11 @@ pub(crate) fn finalize(cx: &CodegenCx<'_, '_>) {
 
     debug!("Generating coverage map for CodegenUnit: `{}`", cx.codegen_unit.name());
 
-    // In order to show that unused functions have coverage counts of zero (0), LLVM requires the
-    // functions exist. Generate synthetic functions with a (required) single counter, and add the
-    // MIR `Coverage` code regions to the `function_coverage_map`, before calling
-    // `ctx.take_function_coverage_map()`.
-    if cx.codegen_unit.is_code_coverage_dead_code_cgu() {
-        add_unused_functions(cx);
-    }
-
     // FIXME(#132395): Can this be none even when coverage is enabled?
-    let function_coverage_map = match cx.coverage_cx {
+    let instances_used = match cx.coverage_cx {
-        Some(ref cx) => cx.take_function_coverage_map(),
+        Some(ref cx) => cx.instances_used.borrow(),
         None => return,
     };
-    if function_coverage_map.is_empty() {
-        // This CGU has no functions with coverage instrumentation.
-        return;
-    }
 
     // The order of entries in this global file table needs to be deterministic,
     // and ideally should also be independent of the details of stable-hashing,
@@ -74,18 +61,27 @@ pub(crate) fn finalize(cx: &CodegenCx<'_, '_>) {
     // are satisfied, the order can be arbitrary.
     let mut global_file_table = GlobalFileTable::new();
 
-    let covfun_records = function_coverage_map
+    let mut covfun_records = instances_used
-        .into_iter()
+        .iter()
+        .copied()
         // Sort by symbol name, so that the global file table is built in an
         // order that doesn't depend on the stable-hash-based order in which
         // instances were visited during codegen.
-        .sorted_by_cached_key(|&(instance, _)| tcx.symbol_name(instance).name)
+        .sorted_by_cached_key(|&instance| tcx.symbol_name(instance).name)
-        .filter_map(|(instance, function_coverage)| {
+        .filter_map(|instance| prepare_covfun_record(tcx, &mut global_file_table, instance, true))
-            let is_used = function_coverage.is_used();
-            prepare_covfun_record(tcx, &mut global_file_table, instance, is_used)
-        })
         .collect::<Vec<_>>();
 
+    // In a single designated CGU, also prepare covfun records for functions
+    // in this crate that were instrumented for coverage, but are unused.
+    if cx.codegen_unit.is_code_coverage_dead_code_cgu() {
+        let mut unused_instances = gather_unused_function_instances(cx);
+        // Sort the unused instances by symbol name, for the same reason as the used ones.
+        unused_instances.sort_by_cached_key(|&instance| tcx.symbol_name(instance).name);
+        covfun_records.extend(unused_instances.into_iter().filter_map(|instance| {
+            prepare_covfun_record(tcx, &mut global_file_table, instance, false)
+        }));
+    }
+
     // If there are no covfun records for this CGU, don't generate a covmap record.
     // Emitting a covmap record without any covfun records causes `llvm-cov` to
     // fail when generating coverage reports, and if there are no covfun records
@@ -261,7 +257,7 @@ fn generate_covmap_record<'ll>(cx: &CodegenCx<'ll, '_>, version: u32, filenames_
 /// coverage map (in a single designated CGU) so that we still emit coverage mappings for them.
 /// We also end up adding their symbol names to a special global array that LLVM will include in
 /// its embedded coverage data.
-fn add_unused_functions(cx: &CodegenCx<'_, '_>) {
+fn gather_unused_function_instances<'tcx>(cx: &CodegenCx<'_, 'tcx>) -> Vec<ty::Instance<'tcx>> {
     assert!(cx.codegen_unit.is_code_coverage_dead_code_cgu());
 
     let tcx = cx.tcx;
@@ -279,20 +275,17 @@ fn add_unused_functions(cx: &CodegenCx<'_, '_>) {
             && !usage.used_via_inlining.contains(&d)
     };
 
-    // Scan for unused functions that were instrumented for coverage.
+    // FIXME(#79651): Consider trying to filter out dummy instantiations of
-    for def_id in tcx.mir_keys(()).iter().copied().filter(|&def_id| is_unused_fn(def_id)) {
+    // unused generic functions from library crates, because they can produce
-        // Get the coverage info from MIR, skipping functions that were never instrumented.
+    // "unused instantiation" in coverage reports even when they are actually
-        let body = tcx.optimized_mir(def_id);
+    // used by some downstream crate in the same binary.
-        let Some(function_coverage_info) = body.function_coverage_info.as_deref() else { continue };
 
-        // FIXME(79651): Consider trying to filter out dummy instantiations of
+    tcx.mir_keys(())
-        // unused generic functions from library crates, because they can produce
+        .iter()
-        // "unused instantiation" in coverage reports even when they are actually
+        .copied()
-        // used by some downstream crate in the same binary.
+        .filter(|&def_id| is_unused_fn(def_id))
-
+        .map(|def_id| make_dummy_instance(tcx, def_id))
-        debug!("generating unused fn: {def_id:?}");
+        .collect::<Vec<_>>()
-        add_unused_function_coverage(cx, def_id, function_coverage_info);
-    }
 }
 
 struct UsageSets<'tcx> {
@@ -357,16 +350,11 @@ fn prepare_usage_sets<'tcx>(tcx: TyCtxt<'tcx>) -> UsageSets<'tcx> {
     UsageSets { all_mono_items, used_via_inlining, missing_own_coverage }
 }
 
-fn add_unused_function_coverage<'tcx>(
+fn make_dummy_instance<'tcx>(tcx: TyCtxt<'tcx>, local_def_id: LocalDefId) -> ty::Instance<'tcx> {
-    cx: &CodegenCx<'_, 'tcx>,
+    let def_id = local_def_id.to_def_id();
-    def_id: LocalDefId,
-    function_coverage_info: &'tcx mir::coverage::FunctionCoverageInfo,
-) {
-    let tcx = cx.tcx;
-    let def_id = def_id.to_def_id();
 
     // Make a dummy instance that fills in all generics with placeholders.
-    let instance = ty::Instance::new(
+    ty::Instance::new(
         def_id,
         ty::GenericArgs::for_item(tcx, def_id, |param, _| {
             if let ty::GenericParamDefKind::Lifetime = param.kind {
@@ -375,9 +363,5 @@ fn add_unused_function_coverage<'tcx>(
                 tcx.mk_param_from_def(param)
             }
         }),
-    );
+    )
-
-    // An unused function's mappings will all be rewritten to map to zero.
-    let function_coverage = FunctionCoverage::new_unused(function_coverage_info);
-    cx.coverage_cx().function_coverage_map.borrow_mut().insert(instance, function_coverage);
 }

compiler/rustc_codegen_llvm/src/coverageinfo/mod.rs

@@ -5,7 +5,7 @@ use rustc_abi::Size;
 use rustc_codegen_ssa::traits::{
     BuilderMethods, ConstCodegenMethods, CoverageInfoBuilderMethods, MiscCodegenMethods,
 };
-use rustc_data_structures::fx::{FxHashMap, FxIndexMap};
+use rustc_data_structures::fx::{FxHashMap, FxIndexSet};
 use rustc_middle::mir::coverage::CoverageKind;
 use rustc_middle::ty::Instance;
 use rustc_middle::ty::layout::HasTyCtxt;
@@ -13,18 +13,16 @@ use tracing::{debug, instrument};
 
 use crate::builder::Builder;
 use crate::common::CodegenCx;
-use crate::coverageinfo::map_data::FunctionCoverage;
 use crate::llvm;
 
 pub(crate) mod ffi;
 mod llvm_cov;
-pub(crate) mod map_data;
 mod mapgen;
 
 /// Extra per-CGU context/state needed for coverage instrumentation.
 pub(crate) struct CguCoverageContext<'ll, 'tcx> {
     /// Coverage data for each instrumented function identified by DefId.
-    pub(crate) function_coverage_map: RefCell<FxIndexMap<Instance<'tcx>, FunctionCoverage<'tcx>>>,
+    pub(crate) instances_used: RefCell<FxIndexSet<Instance<'tcx>>>,
     pub(crate) pgo_func_name_var_map: RefCell<FxHashMap<Instance<'tcx>, &'ll llvm::Value>>,
     pub(crate) mcdc_condition_bitmap_map: RefCell<FxHashMap<Instance<'tcx>, Vec<&'ll llvm::Value>>>,
 
@@ -34,17 +32,13 @@ pub(crate) struct CguCoverageContext<'ll, 'tcx> {
 impl<'ll, 'tcx> CguCoverageContext<'ll, 'tcx> {
     pub(crate) fn new() -> Self {
         Self {
-            function_coverage_map: Default::default(),
+            instances_used: RefCell::<FxIndexSet<_>>::default(),
             pgo_func_name_var_map: Default::default(),
             mcdc_condition_bitmap_map: Default::default(),
             covfun_section_name: Default::default(),
         }
     }
 
-    fn take_function_coverage_map(&self) -> FxIndexMap<Instance<'tcx>, FunctionCoverage<'tcx>> {
-        self.function_coverage_map.replace(FxIndexMap::default())
-    }
-
     /// LLVM use a temp value to record evaluated mcdc test vector of each decision, which is
     /// called condition bitmap. In order to handle nested decisions, several condition bitmaps can
     /// be allocated for a function body. These values are named `mcdc.addr.{i}` and are a 32-bit
@@ -157,12 +151,7 @@ impl<'tcx> CoverageInfoBuilderMethods<'tcx> for Builder<'_, '_, 'tcx> {
         // Mark the instance as used in this CGU, for coverage purposes.
         // This includes functions that were not partitioned into this CGU,
         // but were MIR-inlined into one of this CGU's functions.
-        coverage_cx.function_coverage_map.borrow_mut().entry(instance).or_insert_with(|| {
+        coverage_cx.instances_used.borrow_mut().insert(instance);
-            FunctionCoverage::new_used(
-                function_coverage_info,
-                bx.tcx.coverage_ids_info(instance.def),
-            )
-        });
 
         match *kind {
             CoverageKind::SpanMarker | CoverageKind::BlockMarker { .. } => unreachable!(