[llvm-cov] Add support for baseline coverage (#117910)

fmeum · web-flow · commit dc9e300f12f3 · 2025-06-13T08:49:30.000-07:00
When no profile is provided, but the new --empty-profile option is
specifed, the export/report/show commands now emit coverage data
equivalent to that obtained from a profile with all zero counters
("baseline coverage").

This is useful for build systems (e.g. Bazel) that can track coverage
information for each build target, even those that are never linked into
tests and thus don't have runtime coverage data recorded. By merging in
baseline coverage, lines in files that aren't linked into tests are
correctly reported as uncovered.
diff --git a/llvm/docs/CommandGuide/llvm-cov.rst b/llvm/docs/CommandGuide/llvm-cov.rst
@@ -380,6 +380,11 @@ OPTIONS
  Fail if an object file cannot be found for a binary ID present in the profile,
  neither on the command line nor via binary ID lookup.
 
+.. option:: -empty-profile
+
+ Display the baseline coverage of the binaries with all zero execution counts.
+ Mutually exclusive with -instr-profile.
+
 .. program:: llvm-cov report
 
 .. _llvm-cov-report:
@@ -470,6 +475,11 @@ OPTIONS
  Fail if an object file cannot be found for a binary ID present in the profile,
  neither on the command line nor via binary ID lookup.
 
+.. option:: -empty-profile
+
+ Display the baseline coverage of the binaries with all zero execution counts.
+ Mutually exclusive with -instr-profile.
+
 .. program:: llvm-cov export
 
 .. _llvm-cov-export:
@@ -562,6 +572,11 @@ OPTIONS
  Fail if an object file cannot be found for a binary ID present in the profile,
  neither on the command line nor via binary ID lookup.
 
+.. option:: -empty-profile
+
+ Export the baseline coverage of the binaries with all zero execution counts.
+ Mutually exclusive with -instr-profile.
+
 CONVERT-FOR-TESTING COMMAND
 ---------------------------
 
diff --git a/llvm/include/llvm/ProfileData/Coverage/CoverageMapping.h b/llvm/include/llvm/ProfileData/Coverage/CoverageMapping.h
@@ -991,18 +991,23 @@ class CoverageMapping {
   // Load coverage records from readers.
   static Error loadFromReaders(
       ArrayRef<std::unique_ptr<CoverageMappingReader>> CoverageReaders,
-      IndexedInstrProfReader &ProfileReader, CoverageMapping &Coverage);
+      std::optional<std::reference_wrapper<IndexedInstrProfReader>>
+          &ProfileReader,
+      CoverageMapping &Coverage);
 
   // Load coverage records from file.
   static Error
   loadFromFile(StringRef Filename, StringRef Arch, StringRef CompilationDir,
-               IndexedInstrProfReader &ProfileReader, CoverageMapping &Coverage,
-               bool &DataFound,
+               std::optional<std::reference_wrapper<IndexedInstrProfReader>>
+                   &ProfileReader,
+               CoverageMapping &Coverage, bool &DataFound,
                SmallVectorImpl<object::BuildID> *FoundBinaryIDs = nullptr);
 
   /// Add a function record corresponding to \p Record.
-  Error loadFunctionRecord(const CoverageMappingRecord &Record,
-                           IndexedInstrProfReader &ProfileReader);
+  Error loadFunctionRecord(
+      const CoverageMappingRecord &Record,
+      const std::optional<std::reference_wrapper<IndexedInstrProfReader>>
+          &ProfileReader);
 
   /// Look up the indices for function records which are at least partially
   /// defined in the specified file. This is guaranteed to return a superset of
@@ -1018,15 +1023,16 @@ class CoverageMapping {
   /// Load the coverage mapping using the given readers.
   LLVM_ABI static Expected<std::unique_ptr<CoverageMapping>>
   load(ArrayRef<std::unique_ptr<CoverageMappingReader>> CoverageReaders,
-       IndexedInstrProfReader &ProfileReader);
+       std::optional<std::reference_wrapper<IndexedInstrProfReader>>
+           &ProfileReader);
 
   /// Load the coverage mapping from the given object files and profile. If
   /// \p Arches is non-empty, it must specify an architecture for each object.
   /// Ignores non-instrumented object files unless all are not instrumented.
   LLVM_ABI static Expected<std::unique_ptr<CoverageMapping>>
-  load(ArrayRef<StringRef> ObjectFilenames, StringRef ProfileFilename,
-       vfs::FileSystem &FS, ArrayRef<StringRef> Arches = {},
-       StringRef CompilationDir = "",
+  load(ArrayRef<StringRef> ObjectFilenames,
+       std::optional<StringRef> ProfileFilename, vfs::FileSystem &FS,
+       ArrayRef<StringRef> Arches = {}, StringRef CompilationDir = "",
        const object::BuildIDFetcher *BIDFetcher = nullptr,
        bool CheckBinaryIDs = false);
 
diff --git a/llvm/lib/ProfileData/Coverage/CoverageMapping.cpp b/llvm/lib/ProfileData/Coverage/CoverageMapping.cpp
@@ -823,7 +823,8 @@ class MCDCDecisionRecorder {
 
 Error CoverageMapping::loadFunctionRecord(
     const CoverageMappingRecord &Record,
-    IndexedInstrProfReader &ProfileReader) {
+    const std::optional<std::reference_wrapper<IndexedInstrProfReader>>
+        &ProfileReader) {
   StringRef OrigFuncName = Record.FunctionName;
   if (OrigFuncName.empty())
     return make_error<CoverageMapError>(coveragemap_error::malformed,
@@ -837,35 +838,44 @@ Error CoverageMapping::loadFunctionRecord(
   CounterMappingContext Ctx(Record.Expressions);
 
   std::vector<uint64_t> Counts;
-  if (Error E = ProfileReader.getFunctionCounts(Record.FunctionName,
-                                                Record.FunctionHash, Counts)) {
-    instrprof_error IPE = std::get<0>(InstrProfError::take(std::move(E)));
-    if (IPE == instrprof_error::hash_mismatch) {
-      FuncHashMismatches.emplace_back(std::string(Record.FunctionName),
-                                      Record.FunctionHash);
-      return Error::success();
+  if (ProfileReader) {
+    if (Error E = ProfileReader.value().get().getFunctionCounts(
+            Record.FunctionName, Record.FunctionHash, Counts)) {
+      instrprof_error IPE = std::get<0>(InstrProfError::take(std::move(E)));
+      if (IPE == instrprof_error::hash_mismatch) {
+        FuncHashMismatches.emplace_back(std::string(Record.FunctionName),
+                                        Record.FunctionHash);
+        return Error::success();
+      }
+      if (IPE != instrprof_error::unknown_function)
+        return make_error<InstrProfError>(IPE);
+      Counts.assign(getMaxCounterID(Ctx, Record) + 1, 0);
     }
-    if (IPE != instrprof_error::unknown_function)
-      return make_error<InstrProfError>(IPE);
+  } else {
     Counts.assign(getMaxCounterID(Ctx, Record) + 1, 0);
   }
   Ctx.setCounts(Counts);
 
   bool IsVersion11 =
-      ProfileReader.getVersion() < IndexedInstrProf::ProfVersion::Version12;
+      ProfileReader && ProfileReader.value().get().getVersion() <
+                           IndexedInstrProf::ProfVersion::Version12;
 
   BitVector Bitmap;
-  if (Error E = ProfileReader.getFunctionBitmap(Record.FunctionName,
-                                                Record.FunctionHash, Bitmap)) {
-    instrprof_error IPE = std::get<0>(InstrProfError::take(std::move(E)));
-    if (IPE == instrprof_error::hash_mismatch) {
-      FuncHashMismatches.emplace_back(std::string(Record.FunctionName),
-                                      Record.FunctionHash);
-      return Error::success();
+  if (ProfileReader) {
+    if (Error E = ProfileReader.value().get().getFunctionBitmap(
+            Record.FunctionName, Record.FunctionHash, Bitmap)) {
+      instrprof_error IPE = std::get<0>(InstrProfError::take(std::move(E)));
+      if (IPE == instrprof_error::hash_mismatch) {
+        FuncHashMismatches.emplace_back(std::string(Record.FunctionName),
+                                        Record.FunctionHash);
+        return Error::success();
+      }
+      if (IPE != instrprof_error::unknown_function)
+        return make_error<InstrProfError>(IPE);
+      Bitmap = BitVector(getMaxBitmapSize(Record, IsVersion11));
     }
-    if (IPE != instrprof_error::unknown_function)
-      return make_error<InstrProfError>(IPE);
-    Bitmap = BitVector(getMaxBitmapSize(Record, IsVersion11));
+  } else {
+    Bitmap = BitVector(getMaxBitmapSize(Record, false));
   }
   Ctx.setBitmap(std::move(Bitmap));
 
@@ -959,10 +969,14 @@ Error CoverageMapping::loadFunctionRecord(
 // of CoverageMappingReader instances.
 Error CoverageMapping::loadFromReaders(
     ArrayRef<std::unique_ptr<CoverageMappingReader>> CoverageReaders,
-    IndexedInstrProfReader &ProfileReader, CoverageMapping &Coverage) {
-  assert(!Coverage.SingleByteCoverage ||
-         *Coverage.SingleByteCoverage == ProfileReader.hasSingleByteCoverage());
-  Coverage.SingleByteCoverage = ProfileReader.hasSingleByteCoverage();
+    std::optional<std::reference_wrapper<IndexedInstrProfReader>>
+        &ProfileReader,
+    CoverageMapping &Coverage) {
+  assert(!Coverage.SingleByteCoverage || !ProfileReader ||
+         *Coverage.SingleByteCoverage ==
+             ProfileReader.value().get().hasSingleByteCoverage());
+  Coverage.SingleByteCoverage =
+      !ProfileReader || ProfileReader.value().get().hasSingleByteCoverage();
   for (const auto &CoverageReader : CoverageReaders) {
     for (auto RecordOrErr : *CoverageReader) {
       if (Error E = RecordOrErr.takeError())
@@ -977,7 +991,8 @@ Error CoverageMapping::loadFromReaders(
 
 Expected<std::unique_ptr<CoverageMapping>> CoverageMapping::load(
     ArrayRef<std::unique_ptr<CoverageMappingReader>> CoverageReaders,
-    IndexedInstrProfReader &ProfileReader) {
+    std::optional<std::reference_wrapper<IndexedInstrProfReader>>
+        &ProfileReader) {
   auto Coverage = std::unique_ptr<CoverageMapping>(new CoverageMapping());
   if (Error E = loadFromReaders(CoverageReaders, ProfileReader, *Coverage))
     return std::move(E);
@@ -986,18 +1001,19 @@ Expected<std::unique_ptr<CoverageMapping>> CoverageMapping::load(
 
 // If E is a no_data_found error, returns success. Otherwise returns E.
 static Error handleMaybeNoDataFoundError(Error E) {
-  return handleErrors(
-      std::move(E), [](const CoverageMapError &CME) {
-        if (CME.get() == coveragemap_error::no_data_found)
-          return static_cast<Error>(Error::success());
-        return make_error<CoverageMapError>(CME.get(), CME.getMessage());
-      });
+  return handleErrors(std::move(E), [](const CoverageMapError &CME) {
+    if (CME.get() == coveragemap_error::no_data_found)
+      return static_cast<Error>(Error::success());
+    return make_error<CoverageMapError>(CME.get(), CME.getMessage());
+  });
 }
 
 Error CoverageMapping::loadFromFile(
     StringRef Filename, StringRef Arch, StringRef CompilationDir,
-    IndexedInstrProfReader &ProfileReader, CoverageMapping &Coverage,
-    bool &DataFound, SmallVectorImpl<object::BuildID> *FoundBinaryIDs) {
+    std::optional<std::reference_wrapper<IndexedInstrProfReader>>
+        &ProfileReader,
+    CoverageMapping &Coverage, bool &DataFound,
+    SmallVectorImpl<object::BuildID> *FoundBinaryIDs) {
   auto CovMappingBufOrErr = MemoryBuffer::getFileOrSTDIN(
       Filename, /*IsText=*/false, /*RequiresNullTerminator=*/false);
   if (std::error_code EC = CovMappingBufOrErr.getError())
@@ -1033,13 +1049,23 @@ Error CoverageMapping::loadFromFile(
 }
 
 Expected<std::unique_ptr<CoverageMapping>> CoverageMapping::load(
-    ArrayRef<StringRef> ObjectFilenames, StringRef ProfileFilename,
-    vfs::FileSystem &FS, ArrayRef<StringRef> Arches, StringRef CompilationDir,
+    ArrayRef<StringRef> ObjectFilenames,
+    std::optional<StringRef> ProfileFilename, vfs::FileSystem &FS,
+    ArrayRef<StringRef> Arches, StringRef CompilationDir,
     const object::BuildIDFetcher *BIDFetcher, bool CheckBinaryIDs) {
-  auto ProfileReaderOrErr = IndexedInstrProfReader::create(ProfileFilename, FS);
-  if (Error E = ProfileReaderOrErr.takeError())
-    return createFileError(ProfileFilename, std::move(E));
-  auto ProfileReader = std::move(ProfileReaderOrErr.get());
+  std::unique_ptr<IndexedInstrProfReader> ProfileReader;
+  if (ProfileFilename) {
+    auto ProfileReaderOrErr =
+        IndexedInstrProfReader::create(ProfileFilename.value(), FS);
+    if (Error E = ProfileReaderOrErr.takeError())
+      return createFileError(ProfileFilename.value(), std::move(E));
+    ProfileReader = std::move(ProfileReaderOrErr.get());
+  }
+  auto ProfileReaderRef =
+      ProfileReader
+          ? std::optional<std::reference_wrapper<IndexedInstrProfReader>>(
+                *ProfileReader)
+          : std::nullopt;
   auto Coverage = std::unique_ptr<CoverageMapping>(new CoverageMapping());
   bool DataFound = false;
 
@@ -1053,16 +1079,17 @@ Expected<std::unique_ptr<CoverageMapping>> CoverageMapping::load(
 
   SmallVector<object::BuildID> FoundBinaryIDs;
   for (const auto &File : llvm::enumerate(ObjectFilenames)) {
-    if (Error E =
-            loadFromFile(File.value(), GetArch(File.index()), CompilationDir,
-                         *ProfileReader, *Coverage, DataFound, &FoundBinaryIDs))
+    if (Error E = loadFromFile(File.value(), GetArch(File.index()),
+                               CompilationDir, ProfileReaderRef, *Coverage,
+                               DataFound, &FoundBinaryIDs))
       return std::move(E);
   }
 
   if (BIDFetcher) {
     std::vector<object::BuildID> ProfileBinaryIDs;
-    if (Error E = ProfileReader->readBinaryIds(ProfileBinaryIDs))
-      return createFileError(ProfileFilename, std::move(E));
+    if (ProfileReader)
+      if (Error E = ProfileReader->readBinaryIds(ProfileBinaryIDs))
+        return createFileError(ProfileFilename.value(), std::move(E));
 
     SmallVector<object::BuildIDRef> BinaryIDsToFetch;
     if (!ProfileBinaryIDs.empty()) {
@@ -1082,12 +1109,12 @@ Expected<std::unique_ptr<CoverageMapping>> CoverageMapping::load(
       if (PathOpt) {
         std::string Path = std::move(*PathOpt);
         StringRef Arch = Arches.size() == 1 ? Arches.front() : StringRef();
-        if (Error E = loadFromFile(Path, Arch, CompilationDir, *ProfileReader,
-                                  *Coverage, DataFound))
+        if (Error E = loadFromFile(Path, Arch, CompilationDir, ProfileReaderRef,
+                                   *Coverage, DataFound))
           return std::move(E);
       } else if (CheckBinaryIDs) {
         return createFileError(
-            ProfileFilename,
+            ProfileFilename.value(),
             createStringError(errc::no_such_file_or_directory,
                               "Missing binary ID: " +
                                   llvm::toHex(BinaryID, /*LowerCase=*/true)));
diff --git a/llvm/test/tools/llvm-cov/showLineExecutionCounts-lcov-baseline.test b/llvm/test/tools/llvm-cov/showLineExecutionCounts-lcov-baseline.test
@@ -0,0 +1,37 @@
+// FULL: SF:{{.*}}showLineExecutionCounts.cpp
+// FULL: FN:6,main
+// FULL: FNDA:0,main
+// FULL: FNF:1
+// FULL: FNH:0
+int main() {                              // FULL: DA:[[@LINE]],0
+  int x = 0;                              // FULL: DA:[[@LINE]],0
+                                          // FULL: DA:[[@LINE]],0
+  if (x) {                                // FULL: DA:[[@LINE]],0
+    x = 0;                                // FULL: DA:[[@LINE]],0
+  } else {                                // FULL: DA:[[@LINE]],0
+    x = 1;                                // FULL: DA:[[@LINE]],0
+  }                                       // FULL: DA:[[@LINE]],0
+                                          // FULL: DA:[[@LINE]],0
+  for (int i = 0; i < 100; ++i) {         // FULL: DA:[[@LINE]],0
+    x = 1;                                // FULL: DA:[[@LINE]],0
+  }                                       // FULL: DA:[[@LINE]],0
+                                          // FULL: DA:[[@LINE]],0
+  x = x < 10 ? x + 1 : x - 1;             // FULL: DA:[[@LINE]],0
+  x = x > 10 ?                            // FULL: DA:[[@LINE]],0
+        x - 1:                            // FULL: DA:[[@LINE]],0
+        x + 1;                            // FULL: DA:[[@LINE]],0
+                                          // FULL: DA:[[@LINE]],0
+  return 0;                               // FULL: DA:[[@LINE]],0
+}                                         // FULL: DA:[[@LINE]],0
+// FULL: LF:20
+// FULL: LH:0
+// FULL: end_of_record
+// RUN: llvm-cov export -format=lcov %S/Inputs/lineExecutionCounts.covmapping -empty-profile %s | FileCheck -check-prefixes=FULL %s
+
+// RUN: llvm-cov export -format=lcov -summary-only %S/Inputs/lineExecutionCounts.covmapping -empty-profile %s | FileCheck -check-prefixes=SUMMARYONLY %s
+// SUMMARYONLY: SF:{{.*}}showLineExecutionCounts.cpp
+// SUMMARYONLY: FNF:1
+// SUMMARYONLY: FNH:0
+// SUMMARYONLY: LF:20
+// SUMMARYONLY: LH:0
+// SUMMARYONLY: end_of_record
diff --git a/llvm/tools/llvm-cov/CodeCoverage.cpp b/llvm/tools/llvm-cov/CodeCoverage.cpp
diff --git a/llvm/unittests/ProfileData/CoverageMappingTest.cpp b/llvm/unittests/ProfileData/CoverageMappingTest.cpp
