Rollup merge of #133418 - Zalathar:spans, r=jieyouxu

coverage: Store coverage source regions as `Span` until codegen

Historically, coverage spans were converted into line/column coordinates during the MIR instrumentation pass.

This PR moves that conversion step into codegen, so that coverage spans spend most of their time stored as `Span` instead.

In addition to being conceptually nicer, this also reduces the size of coverage mappings in MIR, because `Span` is smaller than 4x u32.

---

There should be no changes to coverage output.

Author: matthiaskrgr

compiler/rustc_codegen_llvm/src/coverageinfo/ffi.rs

@@ -1,6 +1,4 @@
-use rustc_middle::mir::coverage::{CounterId, CovTerm, ExpressionId, SourceRegion};
-
-use crate::coverageinfo::mapgen::LocalFileId;
+use rustc_middle::mir::coverage::{CounterId, CovTerm, ExpressionId};
 
 /// Must match the layout of `LLVMRustCounterKind`.
 #[derive(Copy, Clone, Debug)]
@@ -126,45 +124,31 @@ pub(crate) struct CoverageSpan {
     /// Local index into the function's local-to-global file ID table.
     /// The value at that index is itself an index into the coverage filename
     /// table in the CGU's `__llvm_covmap` section.
-    file_id: u32,
+    pub(crate) file_id: u32,
 
     /// 1-based starting line of the source code span.
-    start_line: u32,
+    pub(crate) start_line: u32,
     /// 1-based starting column of the source code span.
-    start_col: u32,
+    pub(crate) start_col: u32,
     /// 1-based ending line of the source code span.
-    end_line: u32,
+    pub(crate) end_line: u32,
     /// 1-based ending column of the source code span. High bit must be unset.
-    end_col: u32,
-}
-
-impl CoverageSpan {
-    pub(crate) fn from_source_region(
-        local_file_id: LocalFileId,
-        code_region: &SourceRegion,
-    ) -> Self {
-        let file_id = local_file_id.as_u32();
-        let &SourceRegion { start_line, start_col, end_line, end_col } = code_region;
-        // Internally, LLVM uses the high bit of `end_col` to distinguish between
-        // code regions and gap regions, so it can't be used by the column number.
-        assert!(end_col & (1u32 << 31) == 0, "high bit of `end_col` must be unset: {end_col:#X}");
-        Self { file_id, start_line, start_col, end_line, end_col }
-    }
+    pub(crate) end_col: u32,
 }
 
 /// Must match the layout of `LLVMRustCoverageCodeRegion`.
 #[derive(Clone, Debug)]
 #[repr(C)]
 pub(crate) struct CodeRegion {
-    pub(crate) span: CoverageSpan,
+    pub(crate) cov_span: CoverageSpan,
     pub(crate) counter: Counter,
 }
 
 /// Must match the layout of `LLVMRustCoverageBranchRegion`.
 #[derive(Clone, Debug)]
 #[repr(C)]
 pub(crate) struct BranchRegion {
-    pub(crate) span: CoverageSpan,
+    pub(crate) cov_span: CoverageSpan,
     pub(crate) true_counter: Counter,
     pub(crate) false_counter: Counter,
 }
@@ -173,7 +157,7 @@ pub(crate) struct BranchRegion {
 #[derive(Clone, Debug)]
 #[repr(C)]
 pub(crate) struct MCDCBranchRegion {
-    pub(crate) span: CoverageSpan,
+    pub(crate) cov_span: CoverageSpan,
     pub(crate) true_counter: Counter,
     pub(crate) false_counter: Counter,
     pub(crate) mcdc_branch_params: mcdc::BranchParameters,
@@ -183,6 +167,6 @@ pub(crate) struct MCDCBranchRegion {
 #[derive(Clone, Debug)]
 #[repr(C)]
 pub(crate) struct MCDCDecisionRegion {
-    pub(crate) span: CoverageSpan,
+    pub(crate) cov_span: CoverageSpan,
     pub(crate) mcdc_decision_params: mcdc::DecisionParameters,
 }

compiler/rustc_codegen_llvm/src/coverageinfo/map_data.rs

@@ -3,9 +3,9 @@ use rustc_data_structures::fx::FxIndexSet;
 use rustc_index::bit_set::BitSet;
 use rustc_middle::mir::coverage::{
     CounterId, CovTerm, Expression, ExpressionId, FunctionCoverageInfo, Mapping, MappingKind, Op,
-    SourceRegion,
 };
 use rustc_middle::ty::Instance;
+use rustc_span::Span;
 use tracing::{debug, instrument};
 
 use crate::coverageinfo::ffi::{Counter, CounterExpression, ExprKind};
@@ -220,16 +220,16 @@ impl<'tcx> FunctionCoverage<'tcx> {
         })
     }
 
-    /// Converts this function's coverage mappings into an intermediate form
-    /// that will be used by `mapgen` when preparing for FFI.
-    pub(crate) fn counter_regions(
+    /// Yields all this function's coverage mappings, after simplifying away
+    /// unused counters and counter expressions.
+    pub(crate) fn mapping_spans(
         &self,
-    ) -> impl Iterator<Item = (MappingKind, &SourceRegion)> + ExactSizeIterator {
+    ) -> impl Iterator<Item = (MappingKind, Span)> + ExactSizeIterator + Captures<'_> {
         self.function_coverage_info.mappings.iter().map(move |mapping| {
-            let Mapping { kind, source_region } = mapping;
+            let &Mapping { ref kind, span } = mapping;
             let kind =
                 kind.map_terms(|term| if self.is_zero_term(term) { CovTerm::Zero } else { term });
-            (kind, source_region)
+            (kind, span)
         })
     }
 

compiler/rustc_codegen_llvm/src/coverageinfo/mapgen.rs

@@ -1,12 +1,14 @@
+mod spans;
+
 use std::ffi::CString;
-use std::iter;
+use std::sync::Arc;
 
 use itertools::Itertools as _;
 use rustc_abi::Align;
 use rustc_codegen_ssa::traits::{
     BaseTypeCodegenMethods, ConstCodegenMethods, StaticCodegenMethods,
 };
-use rustc_data_structures::fx::{FxHashSet, FxIndexMap, FxIndexSet};
+use rustc_data_structures::fx::{FxHashSet, FxIndexMap};
 use rustc_hir::def_id::{DefId, LocalDefId};
 use rustc_index::IndexVec;
 use rustc_middle::mir::coverage::MappingKind;
@@ -15,7 +17,7 @@ use rustc_middle::{bug, mir};
 use rustc_session::RemapFileNameExt;
 use rustc_session::config::RemapPathScopeComponents;
 use rustc_span::def_id::DefIdSet;
-use rustc_span::{Span, Symbol};
+use rustc_span::{SourceFile, StableSourceFileId};
 use rustc_target::spec::HasTargetSpec;
 use tracing::debug;
 
@@ -72,11 +74,11 @@ pub(crate) fn finalize(cx: &CodegenCx<'_, '_>) {
         .map(|(instance, function_coverage)| (instance, function_coverage.into_finished()))
         .collect::<Vec<_>>();
 
-    let all_file_names = function_coverage_entries
+    let all_files = function_coverage_entries
         .iter()
         .map(|(_, fn_cov)| fn_cov.function_coverage_info.body_span)
-        .map(|span| span_file_name(tcx, span));
-    let global_file_table = GlobalFileTable::new(all_file_names);
+        .map(|span| tcx.sess.source_map().lookup_source_file(span.lo()));
+    let global_file_table = GlobalFileTable::new(all_files);
 
     // Encode all filenames referenced by coverage mappings in this CGU.
     let filenames_buffer = global_file_table.make_filenames_buffer(tcx);
@@ -103,15 +105,8 @@ pub(crate) fn finalize(cx: &CodegenCx<'_, '_>) {
             encode_mappings_for_function(tcx, &global_file_table, &function_coverage);
 
         if coverage_mapping_buffer.is_empty() {
-            if function_coverage.is_used() {
-                bug!(
-                    "A used function should have had coverage mapping data but did not: {}",
-                    mangled_function_name
-                );
-            } else {
-                debug!("unused function had no coverage mapping data: {}", mangled_function_name);
-                continue;
-            }
+            debug!("function has no mappings to embed; skipping");
+            continue;
         }
 
         if !is_used {
@@ -148,54 +143,62 @@ pub(crate) fn finalize(cx: &CodegenCx<'_, '_>) {
     }
 }
 
-/// Maps "global" (per-CGU) file ID numbers to their underlying filenames.
+/// Maps "global" (per-CGU) file ID numbers to their underlying source files.
 struct GlobalFileTable {
-    /// This "raw" table doesn't include the working dir, so a filename's
+    /// This "raw" table doesn't include the working dir, so a file's
     /// global ID is its index in this set **plus one**.
-    raw_file_table: FxIndexSet<Symbol>,
+    raw_file_table: FxIndexMap<StableSourceFileId, Arc<SourceFile>>,
 }
 
 impl GlobalFileTable {
-    fn new(all_file_names: impl IntoIterator<Item = Symbol>) -> Self {
-        // Collect all of the filenames into a set. Filenames usually come in
-        // contiguous runs, so we can dedup adjacent ones to save work.
-        let mut raw_file_table = all_file_names.into_iter().dedup().collect::<FxIndexSet<Symbol>>();
+    fn new(all_files: impl IntoIterator<Item = Arc<SourceFile>>) -> Self {
+        // Collect all of the files into a set. Files usually come in contiguous
+        // runs, so we can dedup adjacent ones to save work.
+        let mut raw_file_table = all_files
+            .into_iter()
+            .dedup_by(|a, b| a.stable_id == b.stable_id)
+            .map(|f| (f.stable_id, f))
+            .collect::<FxIndexMap<StableSourceFileId, Arc<SourceFile>>>();
 
-        // Sort the file table by its actual string values, not the arbitrary
-        // ordering of its symbols.
-        raw_file_table.sort_unstable_by(|a, b| a.as_str().cmp(b.as_str()));
+        // Sort the file table by its underlying filenames.
+        raw_file_table.sort_unstable_by(|_, a, _, b| {
+            Ord::cmp(&a.name, &b.name).then_with(|| Ord::cmp(&a.stable_id, &b.stable_id))
+        });
 
         Self { raw_file_table }
     }
 
-    fn global_file_id_for_file_name(&self, file_name: Symbol) -> GlobalFileId {
-        let raw_id = self.raw_file_table.get_index_of(&file_name).unwrap_or_else(|| {
-            bug!("file name not found in prepared global file table: {file_name}");
+    fn global_file_id_for_file(&self, file: &SourceFile) -> GlobalFileId {
+        let raw_id = self.raw_file_table.get_index_of(&file.stable_id).unwrap_or_else(|| {
+            bug!("file not found in prepared global file table: {:?}", file.name);
         });
         // The raw file table doesn't include an entry for the working dir
         // (which has ID 0), so add 1 to get the correct ID.
         GlobalFileId::from_usize(raw_id + 1)
     }
 
     fn make_filenames_buffer(&self, tcx: TyCtxt<'_>) -> Vec<u8> {
+        let mut table = Vec::with_capacity(self.raw_file_table.len() + 1);
+
         // LLVM Coverage Mapping Format version 6 (zero-based encoded as 5)
         // requires setting the first filename to the compilation directory.
         // Since rustc generates coverage maps with relative paths, the
         // compilation directory can be combined with the relative paths
         // to get absolute paths, if needed.
-        use rustc_session::RemapFileNameExt;
-        use rustc_session::config::RemapPathScopeComponents;
-        let working_dir: &str = &tcx
-            .sess
-            .opts
-            .working_dir
-            .for_scope(tcx.sess, RemapPathScopeComponents::MACRO)
-            .to_string_lossy();
-
-        // Insert the working dir at index 0, before the other filenames.
-        let filenames =
-            iter::once(working_dir).chain(self.raw_file_table.iter().map(Symbol::as_str));
-        llvm_cov::write_filenames_to_buffer(filenames)
+        table.push(
+            tcx.sess
+                .opts
+                .working_dir
+                .for_scope(tcx.sess, RemapPathScopeComponents::MACRO)
+                .to_string_lossy(),
+        );
+
+        // Add the regular entries after the base directory.
+        table.extend(self.raw_file_table.values().map(|file| {
+            file.name.for_scope(tcx.sess, RemapPathScopeComponents::MACRO).to_string_lossy()
+        }));
+
+        llvm_cov::write_filenames_to_buffer(table.iter().map(|f| f.as_ref()))
     }
 }
 
@@ -208,7 +211,7 @@ rustc_index::newtype_index! {
     /// An index into a function's list of global file IDs. That underlying list
     /// of local-to-global mappings will be embedded in the function's record in
     /// the `__llvm_covfun` linker section.
-    pub(crate) struct LocalFileId {}
+    struct LocalFileId {}
 }
 
 /// Holds a mapping from "local" (per-function) file IDs to "global" (per-CGU)
@@ -234,13 +237,6 @@ impl VirtualFileMapping {
     }
 }
 
-fn span_file_name(tcx: TyCtxt<'_>, span: Span) -> Symbol {
-    let source_file = tcx.sess.source_map().lookup_source_file(span.lo());
-    let name =
-        source_file.name.for_scope(tcx.sess, RemapPathScopeComponents::MACRO).to_string_lossy();
-    Symbol::intern(&name)
-}
-
 /// Using the expressions and counter regions collected for a single function,
 /// generate the variable-sized payload of its corresponding `__llvm_covfun`
 /// entry. The payload is returned as a vector of bytes.
@@ -251,11 +247,13 @@ fn encode_mappings_for_function(
     global_file_table: &GlobalFileTable,
     function_coverage: &FunctionCoverage<'_>,
 ) -> Vec<u8> {
-    let counter_regions = function_coverage.counter_regions();
-    if counter_regions.is_empty() {
+    let mapping_spans = function_coverage.mapping_spans();
+    if mapping_spans.is_empty() {
         return Vec::new();
     }
 
+    let fn_cov_info = function_coverage.function_coverage_info;
+
     let expressions = function_coverage.counter_expressions().collect::<Vec<_>>();
 
     let mut virtual_file_mapping = VirtualFileMapping::default();
@@ -265,42 +263,47 @@ fn encode_mappings_for_function(
     let mut mcdc_decision_regions = vec![];
 
     // Currently a function's mappings must all be in the same file as its body span.
-    let file_name = span_file_name(tcx, function_coverage.function_coverage_info.body_span);
+    let source_map = tcx.sess.source_map();
+    let source_file = source_map.lookup_source_file(fn_cov_info.body_span.lo());
 
-    // Look up the global file ID for that filename.
-    let global_file_id = global_file_table.global_file_id_for_file_name(file_name);
+    // Look up the global file ID for that file.
+    let global_file_id = global_file_table.global_file_id_for_file(&source_file);
 
     // Associate that global file ID with a local file ID for this function.
     let local_file_id = virtual_file_mapping.local_id_for_global(global_file_id);
-    debug!("  file id: {local_file_id:?} => {global_file_id:?} = '{file_name:?}'");
 
-    // For each counter/region pair in this function+file, convert it to a
+    let make_cov_span = |span| {
+        spans::make_coverage_span(local_file_id, source_map, fn_cov_info, &source_file, span)
+    };
+
+    // For each coverage mapping span in this function+file, convert it to a
     // form suitable for FFI.
-    for (mapping_kind, region) in counter_regions {
-        debug!("Adding counter {mapping_kind:?} to map for {region:?}");
-        let span = ffi::CoverageSpan::from_source_region(local_file_id, region);
+    for (mapping_kind, span) in mapping_spans {
+        debug!("Adding counter {mapping_kind:?} to map for {span:?}");
+        let Some(cov_span) = make_cov_span(span) else { continue };
         match mapping_kind {
             MappingKind::Code(term) => {
-                code_regions.push(ffi::CodeRegion { span, counter: ffi::Counter::from_term(term) });
+                code_regions
+                    .push(ffi::CodeRegion { cov_span, counter: ffi::Counter::from_term(term) });
             }
             MappingKind::Branch { true_term, false_term } => {
                 branch_regions.push(ffi::BranchRegion {
-                    span,
+                    cov_span,
                     true_counter: ffi::Counter::from_term(true_term),
                     false_counter: ffi::Counter::from_term(false_term),
                 });
             }
             MappingKind::MCDCBranch { true_term, false_term, mcdc_params } => {
                 mcdc_branch_regions.push(ffi::MCDCBranchRegion {
-                    span,
+                    cov_span,
                     true_counter: ffi::Counter::from_term(true_term),
                     false_counter: ffi::Counter::from_term(false_term),
                     mcdc_branch_params: ffi::mcdc::BranchParameters::from(mcdc_params),
                 });
             }
             MappingKind::MCDCDecision(mcdc_decision_params) => {
                 mcdc_decision_regions.push(ffi::MCDCDecisionRegion {
-                    span,
+                    cov_span,
                     mcdc_decision_params: ffi::mcdc::DecisionParameters::from(mcdc_decision_params),
                 });
             }