Upgrades the coverage map to Version 4

Changes the coverage map injected into binaries compiled with
`-Zinstrument-coverage` to LLVM Coverage Mapping Format, Version 4 (from
Version 3). Note, binaries compiled with this version will require LLVM
tools from at least LLVM Version 11.
This commit is contained in:
Rich Kadel 2020-11-23 12:56:07 -08:00
parent 40624dde6c
commit 1d8c381c01
8 changed files with 200 additions and 129 deletions

View File

@ -4,7 +4,7 @@ use crate::llvm;
use llvm::coverageinfo::CounterMappingRegion;
use rustc_codegen_ssa::coverageinfo::map::{Counter, CounterExpression};
use rustc_codegen_ssa::traits::{BaseTypeMethods, ConstMethods};
use rustc_codegen_ssa::traits::ConstMethods;
use rustc_data_structures::fx::FxIndexSet;
use rustc_llvm::RustString;
use rustc_middle::mir::coverage::CodeRegion;
@ -38,46 +38,50 @@ pub fn finalize<'ll, 'tcx>(cx: &CodegenCx<'ll, 'tcx>) {
let mut mapgen = CoverageMapGenerator::new();
// Encode coverage mappings and generate function records
let mut function_records = Vec::<&'ll llvm::Value>::new();
let coverage_mappings_buffer = llvm::build_byte_buffer(|coverage_mappings_buffer| {
for (instance, function_coverage) in function_coverage_map.into_iter() {
debug!("Generate coverage map for: {:?}", instance);
let mut function_data = Vec::new();
for (instance, function_coverage) in function_coverage_map.into_iter() {
debug!("Generate coverage map for: {:?}", instance);
let mangled_function_name = cx.tcx.symbol_name(instance).to_string();
let function_source_hash = function_coverage.source_hash();
let (expressions, counter_regions) =
function_coverage.get_expressions_and_counter_regions();
let mangled_function_name = cx.tcx.symbol_name(instance).to_string();
let function_source_hash = function_coverage.source_hash();
let (expressions, counter_regions) =
function_coverage.get_expressions_and_counter_regions();
let old_len = coverage_mappings_buffer.len();
mapgen.write_coverage_mappings(expressions, counter_regions, coverage_mappings_buffer);
let mapping_data_size = coverage_mappings_buffer.len() - old_len;
debug_assert!(
mapping_data_size > 0,
"Every `FunctionCoverage` should have at least one counter"
);
let coverage_mapping_buffer = llvm::build_byte_buffer(|coverage_mapping_buffer| {
mapgen.write_coverage_mapping(expressions, counter_regions, coverage_mapping_buffer);
});
debug_assert!(
coverage_mapping_buffer.len() > 0,
"Every `FunctionCoverage` should have at least one counter"
);
let function_record = mapgen.make_function_record(
cx,
mangled_function_name,
function_source_hash,
mapping_data_size,
);
function_records.push(function_record);
}
});
function_data.push((mangled_function_name, function_source_hash, coverage_mapping_buffer));
}
// Encode all filenames referenced by counters/expressions in this module
let filenames_buffer = llvm::build_byte_buffer(|filenames_buffer| {
coverageinfo::write_filenames_section_to_buffer(&mapgen.filenames, filenames_buffer);
});
let filenames_size = filenames_buffer.len();
let filenames_val = cx.const_bytes(&filenames_buffer[..]);
let filenames_ref = coverageinfo::hash_bytes(filenames_buffer);
// Generate the LLVM IR representation of the coverage map and store it in a well-known global
mapgen.save_generated_coverage_map(
cx,
function_records,
filenames_buffer,
coverage_mappings_buffer,
);
let cov_data_val = mapgen.generate_coverage_map(cx, filenames_size, filenames_val);
for (mangled_function_name, function_source_hash, coverage_mapping_buffer) in function_data {
save_function_record(
cx,
mangled_function_name,
function_source_hash,
filenames_ref,
coverage_mapping_buffer,
);
}
// Save the coverage data value to LLVM IR
coverageinfo::save_cov_data_to_mod(cx, cov_data_val);
}
struct CoverageMapGenerator {
@ -92,12 +96,12 @@ impl CoverageMapGenerator {
/// Using the `expressions` and `counter_regions` collected for the current function, generate
/// the `mapping_regions` and `virtual_file_mapping`, and capture any new filenames. Then use
/// LLVM APIs to encode the `virtual_file_mapping`, `expressions`, and `mapping_regions` into
/// the given `coverage_mappings` byte buffer, compliant with the LLVM Coverage Mapping format.
fn write_coverage_mappings(
/// the given `coverage_mapping` byte buffer, compliant with the LLVM Coverage Mapping format.
fn write_coverage_mapping(
&mut self,
expressions: Vec<CounterExpression>,
counter_regions: impl Iterator<Item = (Counter, &'a CodeRegion)>,
coverage_mappings_buffer: &RustString,
coverage_mapping_buffer: &RustString,
) {
let mut counter_regions = counter_regions.collect::<Vec<_>>();
if counter_regions.is_empty() {
@ -145,89 +149,78 @@ impl CoverageMapGenerator {
virtual_file_mapping,
expressions,
mapping_regions,
coverage_mappings_buffer,
coverage_mapping_buffer,
);
}
/// Generate and return the function record `Value`
fn make_function_record(
&mut self,
cx: &CodegenCx<'ll, 'tcx>,
mangled_function_name: String,
function_source_hash: u64,
mapping_data_size: usize,
) -> &'ll llvm::Value {
let name_ref = coverageinfo::compute_hash(&mangled_function_name);
let name_ref_val = cx.const_u64(name_ref);
let mapping_data_size_val = cx.const_u32(mapping_data_size as u32);
let func_hash_val = cx.const_u64(function_source_hash);
cx.const_struct(
&[name_ref_val, mapping_data_size_val, func_hash_val],
/*packed=*/ true,
)
}
/// Combine the filenames and coverage mappings buffers, construct coverage map header and the
/// array of function records, and combine everything into the complete coverage map. Save the
/// coverage map data into the LLVM IR as a static global using a specific, well-known section
/// and name.
fn save_generated_coverage_map(
/// Construct coverage map header and the array of function records, and combine them into the
/// coverage map. Save the coverage map data into the LLVM IR as a static global using a
/// specific, well-known section and name.
fn generate_coverage_map(
self,
cx: &CodegenCx<'ll, 'tcx>,
function_records: Vec<&'ll llvm::Value>,
filenames_buffer: Vec<u8>,
mut coverage_mappings_buffer: Vec<u8>,
) {
// Concatenate the encoded filenames and encoded coverage mappings, and add additional zero
// bytes as-needed to ensure 8-byte alignment.
let mut coverage_size = coverage_mappings_buffer.len();
let filenames_size = filenames_buffer.len();
let remaining_bytes =
(filenames_size + coverage_size) % coverageinfo::COVMAP_VAR_ALIGN_BYTES;
if remaining_bytes > 0 {
let pad = coverageinfo::COVMAP_VAR_ALIGN_BYTES - remaining_bytes;
coverage_mappings_buffer.append(&mut [0].repeat(pad));
coverage_size += pad;
}
let filenames_and_coverage_mappings = [filenames_buffer, coverage_mappings_buffer].concat();
let filenames_and_coverage_mappings_val =
cx.const_bytes(&filenames_and_coverage_mappings[..]);
filenames_size: usize,
filenames_val: &'ll llvm::Value,
) -> &'ll llvm::Value {
debug!(
"cov map: n_records = {}, filenames_size = {}, coverage_size = {}, 0-based version = {}",
function_records.len(),
"cov map: filenames_size = {}, 0-based version = {}",
filenames_size,
coverage_size,
coverageinfo::mapping_version()
);
// Create the coverage data header
let n_records_val = cx.const_u32(function_records.len() as u32);
// Create the coverage data header (Note, fields 0 and 2 are now always zero,
// as of `llvm::coverage::CovMapVersion::Version4`.
let zero_was_n_records_val = cx.const_u32(0);
let filenames_size_val = cx.const_u32(filenames_size as u32);
let coverage_size_val = cx.const_u32(coverage_size as u32);
let zero_was_coverage_size_val = cx.const_u32(0 as u32);
let version_val = cx.const_u32(coverageinfo::mapping_version());
let cov_data_header_val = cx.const_struct(
&[n_records_val, filenames_size_val, coverage_size_val, version_val],
&[zero_was_n_records_val, filenames_size_val, zero_was_coverage_size_val, version_val],
/*packed=*/ false,
);
// Create the function records array
let name_ref_from_u64 = cx.type_i64();
let mapping_data_size_from_u32 = cx.type_i32();
let func_hash_from_u64 = cx.type_i64();
let function_record_ty = cx.type_struct(
&[name_ref_from_u64, mapping_data_size_from_u32, func_hash_from_u64],
/*packed=*/ true,
);
let function_records_val = cx.const_array(function_record_ty, &function_records[..]);
// Create the complete LLVM coverage data value to add to the LLVM IR
let cov_data_val = cx.const_struct(
&[cov_data_header_val, function_records_val, filenames_and_coverage_mappings_val],
/*packed=*/ false,
);
// Save the coverage data value to LLVM IR
coverageinfo::save_map_to_mod(cx, cov_data_val);
cx.const_struct(&[cov_data_header_val, filenames_val], /*packed=*/ false)
}
}
/// Construct a function record and combine it with the function's coverage mapping data.
/// Save the function record into the LLVM IR as a static global using a
/// specific, well-known section and name.
fn save_function_record(
cx: &CodegenCx<'ll, 'tcx>,
mangled_function_name: String,
function_source_hash: u64,
filenames_ref: u64,
coverage_mapping_buffer: Vec<u8>,
) {
// Concatenate the encoded coverage mappings
let coverage_mapping_size = coverage_mapping_buffer.len();
let coverage_mapping_val = cx.const_bytes(&coverage_mapping_buffer[..]);
let func_name_hash = coverageinfo::hash_str(&mangled_function_name);
let func_name_hash_val = cx.const_u64(func_name_hash);
let coverage_mapping_size_val = cx.const_u32(coverage_mapping_size as u32);
let func_hash_val = cx.const_u64(function_source_hash);
let filenames_ref_val = cx.const_u64(filenames_ref);
let func_record_val = cx.const_struct(
&[
func_name_hash_val,
coverage_mapping_size_val,
func_hash_val,
filenames_ref_val,
coverage_mapping_val,
],
/*packed=*/ true,
);
// At the present time, the coverage map for Rust assumes every instrumented function `is_used`.
// Note that Clang marks functions as "unused" in `CodeGenPGO::emitEmptyCounterMapping`. (See:
// https://github.com/rust-lang/llvm-project/blob/de02a75e398415bad4df27b4547c25b896c8bf3b/clang%2Flib%2FCodeGen%2FCodeGenPGO.cpp#L877-L878
// for example.)
//
// It's not yet clear if or how this may be applied to Rust in the future, but the `is_used`
// argument is available and handled similarly.
let is_used = true;
coverageinfo::save_func_record_to_mod(cx, func_name_hash, func_record_val, is_used);
}

View File

@ -23,7 +23,7 @@ use tracing::debug;
pub mod mapgen;
const COVMAP_VAR_ALIGN_BYTES: usize = 8;
const VAR_ALIGN_BYTES: usize = 8;
/// A context object for maintaining all state needed by the coverageinfo module.
pub struct CrateCoverageContext<'tcx> {
@ -177,17 +177,20 @@ pub(crate) fn write_mapping_to_buffer(
);
}
}
pub(crate) fn hash_str(strval: &str) -> u64 {
let strval = CString::new(strval).expect("null error converting hashable str to C string");
unsafe { llvm::LLVMRustCoverageHashCString(strval.as_ptr()) }
}
pub(crate) fn compute_hash(name: &str) -> u64 {
let name = CString::new(name).expect("null error converting hashable name to C string");
unsafe { llvm::LLVMRustCoverageComputeHash(name.as_ptr()) }
pub(crate) fn hash_bytes(bytes: Vec<u8>) -> u64 {
unsafe { llvm::LLVMRustCoverageHashByteArray(bytes.as_ptr().cast(), bytes.len()) }
}
pub(crate) fn mapping_version() -> u32 {
unsafe { llvm::LLVMRustCoverageMappingVersion() }
}
pub(crate) fn save_map_to_mod<'ll, 'tcx>(
pub(crate) fn save_cov_data_to_mod<'ll, 'tcx>(
cx: &CodegenCx<'ll, 'tcx>,
cov_data_val: &'ll llvm::Value,
) {
@ -198,7 +201,7 @@ pub(crate) fn save_map_to_mod<'ll, 'tcx>(
debug!("covmap var name: {:?}", covmap_var_name);
let covmap_section_name = llvm::build_string(|s| unsafe {
llvm::LLVMRustCoverageWriteSectionNameToString(cx.llmod, s);
llvm::LLVMRustCoverageWriteMapSectionNameToString(cx.llmod, s);
})
.expect("Rust Coverage section name failed UTF-8 conversion");
debug!("covmap section name: {:?}", covmap_section_name);
@ -206,8 +209,43 @@ pub(crate) fn save_map_to_mod<'ll, 'tcx>(
let llglobal = llvm::add_global(cx.llmod, cx.val_ty(cov_data_val), &covmap_var_name);
llvm::set_initializer(llglobal, cov_data_val);
llvm::set_global_constant(llglobal, true);
llvm::set_linkage(llglobal, llvm::Linkage::InternalLinkage);
llvm::set_linkage(llglobal, llvm::Linkage::PrivateLinkage);
llvm::set_section(llglobal, &covmap_section_name);
llvm::set_alignment(llglobal, COVMAP_VAR_ALIGN_BYTES);
llvm::set_alignment(llglobal, VAR_ALIGN_BYTES);
cx.add_used_global(llglobal);
}
pub(crate) fn save_func_record_to_mod<'ll, 'tcx>(
cx: &CodegenCx<'ll, 'tcx>,
func_name_hash: u64,
func_record_val: &'ll llvm::Value,
is_used: bool,
) {
// Assign a name to the function record. This is used to merge duplicates.
//
// In LLVM, a "translation unit" (effectively, a `Crate` in Rust) can describe functions that
// are included-but-not-used. If (or when) Rust generates functions that are
// included-but-not-used, note that a dummy description for a function included-but-not-used
// in a Crate can be replaced by full description provided by a different Crate. The two kinds
// of descriptions play distinct roles in LLVM IR; therefore, assign them different names (by
// appending "u" to the end of the function record var name, to prevent `linkonce_odr` merging.
let func_record_var_name =
format!("__covrec_{:X}{}", func_name_hash, if is_used { "u" } else { "" });
debug!("function record var name: {:?}", func_record_var_name);
let func_record_section_name = llvm::build_string(|s| unsafe {
llvm::LLVMRustCoverageWriteFuncSectionNameToString(cx.llmod, s);
})
.expect("Rust Coverage function record section name failed UTF-8 conversion");
debug!("function record section name: {:?}", func_record_section_name);
let llglobal = llvm::add_global(cx.llmod, cx.val_ty(func_record_val), &func_record_var_name);
llvm::set_initializer(llglobal, func_record_val);
llvm::set_global_constant(llglobal, true);
llvm::set_linkage(llglobal, llvm::Linkage::LinkOnceODRLinkage);
llvm::set_visibility(llglobal, llvm::Visibility::Hidden);
llvm::set_section(llglobal, &func_record_section_name);
llvm::set_alignment(llglobal, VAR_ALIGN_BYTES);
llvm::set_comdat(cx.llmod, llglobal, &func_record_var_name);
cx.add_used_global(llglobal);
}

View File

@ -1791,10 +1791,14 @@ extern "C" {
pub fn LLVMRustCoverageCreatePGOFuncNameVar(F: &'a Value, FuncName: *const c_char)
-> &'a Value;
pub fn LLVMRustCoverageComputeHash(Name: *const c_char) -> u64;
pub fn LLVMRustCoverageHashCString(StrVal: *const c_char) -> u64;
pub fn LLVMRustCoverageHashByteArray(Bytes: *const c_char, NumBytes: size_t) -> u64;
#[allow(improper_ctypes)]
pub fn LLVMRustCoverageWriteSectionNameToString(M: &Module, Str: &RustString);
pub fn LLVMRustCoverageWriteMapSectionNameToString(M: &Module, Str: &RustString);
#[allow(improper_ctypes)]
pub fn LLVMRustCoverageWriteFuncSectionNameToString(M: &Module, Str: &RustString);
#[allow(improper_ctypes)]
pub fn LLVMRustCoverageWriteMappingVarNameToString(Str: &RustString);

View File

@ -220,12 +220,24 @@ pub fn set_linkage(llglobal: &Value, linkage: Linkage) {
}
}
pub fn set_visibility(llglobal: &Value, visibility: Visibility) {
unsafe {
LLVMRustSetVisibility(llglobal, visibility);
}
}
pub fn set_alignment(llglobal: &Value, bytes: usize) {
unsafe {
ffi::LLVMSetAlignment(llglobal, bytes as c_uint);
}
}
pub fn set_comdat(llmod: &Module, llglobal: &Value, name: &str) {
unsafe {
LLVMRustSetComdat(llmod, llglobal, name.as_ptr().cast(), name.len());
}
}
/// Safe wrapper around `LLVMGetParam`, because segfaults are no fun.
pub fn get_param(llfn: &Value, index: c_uint) -> &Value {
unsafe {

View File

@ -3,7 +3,6 @@
#include "llvm/ProfileData/Coverage/CoverageMappingWriter.h"
#include "llvm/ProfileData/InstrProf.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/Support/LEB128.h"
#include <iostream>
@ -13,15 +12,14 @@ extern "C" void LLVMRustCoverageWriteFilenamesSectionToBuffer(
const char* const Filenames[],
size_t FilenamesLen,
RustStringRef BufferOut) {
// LLVM 11's CoverageFilenamesSectionWriter uses its new `Version4` format,
// so we're manually writing the `Version3` format ourselves.
RawRustStringOstream OS(BufferOut);
encodeULEB128(FilenamesLen, OS);
SmallVector<StringRef,32> FilenameRefs;
for (size_t i = 0; i < FilenamesLen; i++) {
StringRef Filename(Filenames[i]);
encodeULEB128(Filename.size(), OS);
OS << Filename;
FilenameRefs.push_back(StringRef(Filenames[i]));
}
auto FilenamesWriter = coverage::CoverageFilenamesSectionWriter(
makeArrayRef(FilenameRefs));
RawRustStringOstream OS(BufferOut);
FilenamesWriter.write(OS);
}
extern "C" void LLVMRustCoverageWriteMappingToBuffer(
@ -45,20 +43,37 @@ extern "C" LLVMValueRef LLVMRustCoverageCreatePGOFuncNameVar(LLVMValueRef F, con
return wrap(createPGOFuncNameVar(*cast<Function>(unwrap(F)), FuncNameRef));
}
extern "C" uint64_t LLVMRustCoverageComputeHash(const char *Name) {
StringRef NameRef(Name);
return IndexedInstrProf::ComputeHash(NameRef);
extern "C" uint64_t LLVMRustCoverageHashCString(const char *StrVal) {
StringRef StrRef(StrVal);
return IndexedInstrProf::ComputeHash(StrRef);
}
extern "C" void LLVMRustCoverageWriteSectionNameToString(LLVMModuleRef M,
RustStringRef Str) {
extern "C" uint64_t LLVMRustCoverageHashByteArray(
const char *Bytes,
unsigned NumBytes) {
StringRef StrRef(Bytes, NumBytes);
return IndexedInstrProf::ComputeHash(StrRef);
}
static void WriteSectionNameToString(LLVMModuleRef M,
InstrProfSectKind SK,
RustStringRef Str) {
Triple TargetTriple(unwrap(M)->getTargetTriple());
auto name = getInstrProfSectionName(IPSK_covmap,
TargetTriple.getObjectFormat());
auto name = getInstrProfSectionName(SK, TargetTriple.getObjectFormat());
RawRustStringOstream OS(Str);
OS << name;
}
extern "C" void LLVMRustCoverageWriteMapSectionNameToString(LLVMModuleRef M,
RustStringRef Str) {
WriteSectionNameToString(M, IPSK_covmap, Str);
}
extern "C" void LLVMRustCoverageWriteFuncSectionNameToString(LLVMModuleRef M,
RustStringRef Str) {
WriteSectionNameToString(M, IPSK_covfun, Str);
}
extern "C" void LLVMRustCoverageWriteMappingVarNameToString(RustStringRef Str) {
auto name = getCoverageMappingVarName();
RawRustStringOstream OS(Str);
@ -66,5 +81,5 @@ extern "C" void LLVMRustCoverageWriteMappingVarNameToString(RustStringRef Str) {
}
extern "C" uint32_t LLVMRustCoverageMappingVersion() {
return coverage::CovMapVersion::Version3;
return coverage::CovMapVersion::Version4;
}

View File

@ -1462,7 +1462,7 @@ extern "C" void LLVMRustSetComdat(LLVMModuleRef M, LLVMValueRef V,
const char *Name, size_t NameLen) {
Triple TargetTriple(unwrap(M)->getTargetTriple());
GlobalObject *GV = unwrap<GlobalObject>(V);
if (!TargetTriple.isOSBinFormatMachO()) {
if (TargetTriple.supportsCOMDAT()) {
StringRef NameRef(Name, NameLen);
GV->setComdat(unwrap(M)->getOrInsertComdat(NameRef));
}

View File

@ -12,10 +12,12 @@ ifeq ($(UNAME),Darwin)
INSTR_PROF_DATA_SUFFIX=,regular,live_support
DATA_SECTION_PREFIX=__DATA,
LLVM_COV_SECTION_PREFIX=__LLVM_COV,
COMDAT_IF_SUPPORTED=
else
INSTR_PROF_DATA_SUFFIX=
DATA_SECTION_PREFIX=
LLVM_COV_SECTION_PREFIX=
COMDAT_IF_SUPPORTED=, comdat
endif
ifeq ($(LINK_DEAD_CODE),yes)
@ -29,24 +31,28 @@ ifdef IS_WINDOWS
-check-prefixes=CHECK,WINDOWS \
-DPRIVATE_GLOBAL='internal global' \
-DDEFINE_INTERNAL='$(DEFINE_INTERNAL)' \
-DCOMDAT_IF_SUPPORTED='$(COMDAT_IF_SUPPORTED)' \
-DINSTR_PROF_DATA='.lprfd$$M' \
-DINSTR_PROF_NAME='.lprfn$$M' \
-DINSTR_PROF_CNTS='.lprfc$$M' \
-DINSTR_PROF_VALS='.lprfv$$M' \
-DINSTR_PROF_VNODES='.lprfnd$$M' \
-DINSTR_PROF_COVMAP='.lcovmap$$M' \
-DINSTR_PROF_COVFUN='.lcovfun$$M' \
-DINSTR_PROF_ORDERFILE='.lorderfile$$M'
else
LLVM_FILECHECK_OPTIONS=\
-check-prefixes=CHECK \
-DPRIVATE_GLOBAL='private global' \
-DDEFINE_INTERNAL='$(DEFINE_INTERNAL)' \
-DCOMDAT_IF_SUPPORTED='$(COMDAT_IF_SUPPORTED)' \
-DINSTR_PROF_DATA='$(DATA_SECTION_PREFIX)__llvm_prf_data$(INSTR_PROF_DATA_SUFFIX)' \
-DINSTR_PROF_NAME='$(DATA_SECTION_PREFIX)__llvm_prf_names' \
-DINSTR_PROF_CNTS='$(DATA_SECTION_PREFIX)__llvm_prf_cnts' \
-DINSTR_PROF_VALS='$(DATA_SECTION_PREFIX)__llvm_prf_vals' \
-DINSTR_PROF_VNODES='$(DATA_SECTION_PREFIX)__llvm_prf_vnds' \
-DINSTR_PROF_COVMAP='$(LLVM_COV_SECTION_PREFIX)__llvm_covmap' \
-DINSTR_PROF_COVFUN='$(LLVM_COV_SECTION_PREFIX)__llvm_covfun' \
-DINSTR_PROF_ORDERFILE='$(DATA_SECTION_PREFIX)__llvm_orderfile'
endif

View File

@ -3,7 +3,10 @@
WINDOWS: $__llvm_profile_runtime_user = comdat any
CHECK: @__llvm_coverage_mapping = internal constant
CHECK: @__covrec_{{[A-F0-9]+}}u = linkonce_odr hidden constant
CHECK-SAME: section "[[INSTR_PROF_COVFUN]]"[[COMDAT_IF_SUPPORTED]], align 8
CHECK: @__llvm_coverage_mapping = private constant
CHECK-SAME: section "[[INSTR_PROF_COVMAP]]", align 8
WINDOWS: @__llvm_profile_runtime = external global i32