Fix split post-processing of LLVM-based coverage

`--experimental_split_coverage_postprocessing` now works in combination
with `--experimental_generate_llvm_lcov`, which required adding the
runtime objects as inputs to the separate coverage post-processing
spawn. Previously, they were only added to the runfiles of the test
action, which aren't (and shouldn't) be added to the coverage spawn.

As a side effect, this provides a more natural fix for bazelbuild#15121 then
bb6f1a7, which added an additional
check to only add those runtime objects to the `llvm-cov` command line
that were present in runfiles. Now, since all runtime objects are staged
as inputs, they are known to be available. This improves coverage
accuracy when e.g. runtime objects are packaged into jars and loaded at
runtime.

src/main/starlark/builtins_bzl/common/cc/cc_binary.bzl

@@ -182,7 +182,7 @@ def _add(ctx, linking_statically):
 def _get_file_content(objects):
     result = []
     for obj in objects:
-        result.append(obj.short_path)
+        result.append(obj.path)
         result.append("\n")
     return "".join(result)
 
@@ -192,7 +192,7 @@ def _add_transitive_info_providers(ctx, cc_toolchain, cpp_config, feature_config
         runtime_objects_list = ctx.actions.declare_file(ctx.label.name + "runtime_objects_list.txt")
         file_content = _get_file_content(runtime_objects_for_coverage)
         ctx.actions.write(output = runtime_objects_list, content = file_content, is_executable = False)
-        additional_meta_data = [runtime_objects_list]
+        additional_meta_data = [runtime_objects_list] + runtime_objects_for_coverage
 
     instrumented_files_provider = cc_helper.create_cc_instrumented_files_info(
         ctx = ctx,

src/test/shell/bazel/bazel_coverage_cc_test_llvm.sh

@@ -154,6 +154,55 @@ end_of_record"
   assert_equals "$expected_result" "$(cat $(get_coverage_file_path_from_test_log))"
 }
 
+function test_cc_test_llvm_coverage_produces_lcov_report_with_split_postprocessing() {
+  local -r clang="/usr/bin/clang"
+  if [[ ! -x ${clang} ]]; then
+    return
+  fi
+  local -r clang_version=$(clang --version | grep -o "clang version [0-9]*" | cut -d " " -f 3)
+  if [ "$clang_version" -lt 9 ] || [ "$clang_version" -eq 10 ] || [ "$clang_version" -eq 11 ]; then
+    # No lcov produced with <9.0, no branch coverage with 10.0 and 11.0.
+    echo "clang versions <9.0 as well as 10.0 and 11.0 are not supported." && return
+  fi
+
+  local -r llvm_profdata="/usr/bin/llvm-profdata"
+  if [[ ! -x ${llvm_profdata} ]]; then
+    return
+  fi
+
+  local -r llvm_cov="/usr/bin/llvm-cov"
+  if [[ ! -x ${llvm_cov} ]]; then
+    return
+  fi
+
+  setup_a_cc_lib_and_t_cc_test
+
+  BAZEL_USE_LLVM_NATIVE_COVERAGE=1 GCOV=$llvm_profdata CC=$clang \
+    BAZEL_LLVM_COV=$llvm_cov bazel coverage --experimental_generate_llvm_lcov \
+      --experimental_split_coverage_postprocessing --experimental_fetch_all_coverage_outputs \
+        --test_env=VERBOSE_COVERAGE=1 --test_output=all //:t &>$TEST_log || fail "Coverage for //:t failed"
+
+
+
+  local expected_result="SF:a.cc
+FN:3,_Z1ab
+FNDA:1,_Z1ab
+FNF:1
+FNH:1
+DA:3,1
+DA:4,1
+DA:5,1
+DA:6,1
+DA:7,0
+DA:8,0
+DA:9,1
+LH:5
+LF:7
+end_of_record"
+
+  assert_equals "$expected_result" "$(cat $(get_coverage_file_path_from_test_log))"
+}
+
 function test_cc_test_with_runtime_objects_not_in_runfiles() {
   local -r clang="/usr/bin/clang"
   if [[ ! -x ${clang} ]]; then

tools/test/collect_cc_coverage.sh

@@ -85,9 +85,7 @@ function llvm_coverage_lcov() {
   while read -r line; do
     if [[ ${line: -24} == "runtime_objects_list.txt" ]]; then
       while read -r line_runtime_object; do
-        if [[ -e "${RUNFILES_DIR}/${TEST_WORKSPACE}/${line_runtime_object}" ]]; then
-          object_param+=" -object ${RUNFILES_DIR}/${TEST_WORKSPACE}/${line_runtime_object}"
-        fi
+        object_param+=" -object ${line_runtime_object}"
       done < "${line}"
     fi
   done < "${COVERAGE_MANIFEST}"