From c5d65aa58043cc1eac76838319009fce9de1b019 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jakub=20Ber=C3=A1nek?= Date: Sat, 4 Feb 2023 00:03:12 +0100 Subject: [PATCH 1/4] Apply BOLT optimizations without rebuilding LLVM --- src/bootstrap/bolt.rs | 21 ++++++++++----------- src/bootstrap/dist.rs | 23 ++++++++++++++++++++++- src/bootstrap/native.rs | 23 ----------------------- src/ci/stage-build.py | 1 - 4 files changed, 32 insertions(+), 36 deletions(-) diff --git a/src/bootstrap/bolt.rs b/src/bootstrap/bolt.rs index ea37cd47049..d424852b845 100644 --- a/src/bootstrap/bolt.rs +++ b/src/bootstrap/bolt.rs @@ -1,12 +1,13 @@ -use std::path::Path; +use std::path::{Path, PathBuf}; use std::process::Command; /// Uses the `llvm-bolt` binary to instrument the binary/library at the given `path` with BOLT. /// When the instrumented artifact is executed, it will generate BOLT profiles into /// `/tmp/prof.fdata..fdata`. -pub fn instrument_with_bolt_inplace(path: &Path) { +/// Returns a path to the instrumented artifact, created in a temporary directory. +pub fn instrument_with_bolt(path: &Path) -> PathBuf { let dir = std::env::temp_dir(); - let instrumented_path = dir.join("instrumented.so"); + let instrumented_path = dir.join(path.file_name().unwrap()); let status = Command::new("llvm-bolt") .arg("-instrument") @@ -21,9 +22,7 @@ pub fn instrument_with_bolt_inplace(path: &Path) { if !status.success() { panic!("Could not instrument {} with BOLT, exit code {:?}", path.display(), status.code()); } - - std::fs::copy(&instrumented_path, path).expect("Cannot copy instrumented artifact"); - std::fs::remove_file(instrumented_path).expect("Cannot delete instrumented artifact"); + instrumented_path } /// Uses the `llvm-bolt` binary to optimize the binary/library at the given `path` with BOLT, @@ -31,9 +30,11 @@ pub fn instrument_with_bolt_inplace(path: &Path) { /// /// The recorded profiles have to be merged using the `merge-fdata` tool from LLVM and the merged /// profile path should be then passed to this function. -pub fn optimize_library_with_bolt_inplace(path: &Path, profile_path: &Path) { +/// +/// Returns a path to the optimized artifact, created in a temporary directory. +pub fn optimize_with_bolt(path: &Path, profile_path: &Path) -> PathBuf { let dir = std::env::temp_dir(); - let optimized_path = dir.join("optimized.so"); + let optimized_path = dir.join(path.file_name().unwrap()); let status = Command::new("llvm-bolt") .arg(&path) @@ -65,7 +66,5 @@ pub fn optimize_library_with_bolt_inplace(path: &Path, profile_path: &Path) { if !status.success() { panic!("Could not optimize {} with BOLT, exit code {:?}", path.display(), status.code()); } - - std::fs::copy(&optimized_path, path).expect("Cannot copy optimized artifact"); - std::fs::remove_file(optimized_path).expect("Cannot delete optimized artifact"); + optimized_path } diff --git a/src/bootstrap/dist.rs b/src/bootstrap/dist.rs index 9b2b549612d..f6f6f4cc8ea 100644 --- a/src/bootstrap/dist.rs +++ b/src/bootstrap/dist.rs @@ -19,6 +19,7 @@ use std::process::Command; use object::read::archive::ArchiveFile; use object::BinaryFormat; +use crate::bolt::{instrument_with_bolt, optimize_with_bolt}; use crate::builder::{Builder, Kind, RunConfig, ShouldRun, Step}; use crate::cache::{Interned, INTERNER}; use crate::channel; @@ -1904,6 +1905,26 @@ fn add_env(builder: &Builder<'_>, cmd: &mut Command, target: TargetSelection) { } } +fn install_llvm_file(builder: &Builder<'_>, source: &Path, destination: &Path) { + if source.as_os_str().is_empty() { + return; + } + + // After LLVM is built, we modify (instrument or optimize) the libLLVM.so library file. + // This is not done in-place so that the built LLVM files are not "tainted" with BOLT. + // We perform the instrumentation/optimization here, on the fly, just before they are being + // packaged into some destination directory. + let postprocessed = if builder.config.llvm_bolt_profile_generate { + instrument_with_bolt(source) + } else if let Some(path) = &builder.config.llvm_bolt_profile_use { + optimize_with_bolt(source, &Path::new(&path)) + } else { + source.to_path_buf() + }; + + builder.install(&postprocessed, destination, 0o644); +} + /// Maybe add LLVM object files to the given destination lib-dir. Allows either static or dynamic linking. /// /// Returns whether the files were actually copied. @@ -1955,7 +1976,7 @@ fn maybe_install_llvm(builder: &Builder<'_>, target: TargetSelection, dst_libdir } else { PathBuf::from(file) }; - builder.install(&file, dst_libdir, 0o644); + install_llvm_file(builder, &file, dst_libdir); } !builder.config.dry_run() } else { diff --git a/src/bootstrap/native.rs b/src/bootstrap/native.rs index 5987b641b39..8c2bece1e62 100644 --- a/src/bootstrap/native.rs +++ b/src/bootstrap/native.rs @@ -16,7 +16,6 @@ use std::io; use std::path::{Path, PathBuf}; use std::process::Command; -use crate::bolt::{instrument_with_bolt_inplace, optimize_library_with_bolt_inplace}; use crate::builder::{Builder, RunConfig, ShouldRun, Step}; use crate::channel; use crate::config::{Config, TargetSelection}; @@ -523,34 +522,12 @@ impl Step for Llvm { } } - // After LLVM is built, we modify (instrument or optimize) the libLLVM.so library file - // in place. This is fine, because currently we do not support incrementally rebuilding - // LLVM after a configuration change, so to rebuild it the build files have to be removed, - // which will also remove these modified files. - if builder.config.llvm_bolt_profile_generate { - instrument_with_bolt_inplace(&get_built_llvm_lib_path(&res.llvm_config)); - } - if let Some(path) = &builder.config.llvm_bolt_profile_use { - optimize_library_with_bolt_inplace( - &get_built_llvm_lib_path(&res.llvm_config), - &Path::new(path), - ); - } - t!(stamp.write()); res } } -/// Returns path to a built LLVM library (libLLVM.so). -/// Assumes that we have built LLVM into a single library file. -fn get_built_llvm_lib_path(llvm_config_path: &Path) -> PathBuf { - let mut cmd = Command::new(llvm_config_path); - cmd.arg("--libfiles"); - PathBuf::from(output(&mut cmd).trim()) -} - fn check_llvm_version(builder: &Builder<'_>, llvm_config: &Path) { if !builder.config.llvm_version_check { return; diff --git a/src/ci/stage-build.py b/src/ci/stage-build.py index bd8fd524a26..ebd111e6453 100644 --- a/src/ci/stage-build.py +++ b/src/ci/stage-build.py @@ -805,7 +805,6 @@ def execute_build_pipeline(timer: Timer, pipeline: Pipeline, final_build_args: L gather_llvm_bolt_profiles(pipeline) print_free_disk_space(pipeline) - clear_llvm_files(pipeline) final_build_args += [ "--llvm-bolt-profile-use", pipeline.llvm_bolt_profile_merged_file() From bfc220a96e7b4198c2ae18c7fc6e36fd67448086 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jakub=20Ber=C3=A1nek?= Date: Tue, 7 Feb 2023 17:59:51 +0100 Subject: [PATCH 2/4] Create BOLT build steps to avoid running BOLT multiple times on the same file --- src/bootstrap/bolt.rs | 26 ++++----- src/bootstrap/dist.rs | 120 +++++++++++++++++++++++++++++++++++++++++- 2 files changed, 127 insertions(+), 19 deletions(-) diff --git a/src/bootstrap/bolt.rs b/src/bootstrap/bolt.rs index d424852b845..973dc4f602b 100644 --- a/src/bootstrap/bolt.rs +++ b/src/bootstrap/bolt.rs @@ -1,47 +1,40 @@ -use std::path::{Path, PathBuf}; +use std::path::Path; use std::process::Command; -/// Uses the `llvm-bolt` binary to instrument the binary/library at the given `path` with BOLT. +/// Uses the `llvm-bolt` binary to instrument the artifact at the given `path` with BOLT. /// When the instrumented artifact is executed, it will generate BOLT profiles into /// `/tmp/prof.fdata..fdata`. -/// Returns a path to the instrumented artifact, created in a temporary directory. -pub fn instrument_with_bolt(path: &Path) -> PathBuf { - let dir = std::env::temp_dir(); - let instrumented_path = dir.join(path.file_name().unwrap()); - +/// Creates the instrumented artifact at `output_path`. +pub fn instrument_with_bolt(path: &Path, output_path: &Path) { let status = Command::new("llvm-bolt") .arg("-instrument") .arg(&path) // Make sure that each process will write its profiles into a separate file .arg("--instrumentation-file-append-pid") .arg("-o") - .arg(&instrumented_path) + .arg(output_path) .status() .expect("Could not instrument artifact using BOLT"); if !status.success() { panic!("Could not instrument {} with BOLT, exit code {:?}", path.display(), status.code()); } - instrumented_path } -/// Uses the `llvm-bolt` binary to optimize the binary/library at the given `path` with BOLT, +/// Uses the `llvm-bolt` binary to optimize the artifact at the given `path` with BOLT, /// using merged profiles from `profile_path`. /// /// The recorded profiles have to be merged using the `merge-fdata` tool from LLVM and the merged /// profile path should be then passed to this function. /// -/// Returns a path to the optimized artifact, created in a temporary directory. -pub fn optimize_with_bolt(path: &Path, profile_path: &Path) -> PathBuf { - let dir = std::env::temp_dir(); - let optimized_path = dir.join(path.file_name().unwrap()); - +/// Creates the optimized artifact at `output_path`. +pub fn optimize_with_bolt(path: &Path, profile_path: &Path, output_path: &Path) { let status = Command::new("llvm-bolt") .arg(&path) .arg("-data") .arg(&profile_path) .arg("-o") - .arg(&optimized_path) + .arg(output_path) // Reorder basic blocks within functions .arg("-reorder-blocks=ext-tsp") // Reorder functions within the binary @@ -66,5 +59,4 @@ pub fn optimize_with_bolt(path: &Path, profile_path: &Path) -> PathBuf { if !status.success() { panic!("Could not optimize {} with BOLT, exit code {:?}", path.display(), status.code()); } - optimized_path } diff --git a/src/bootstrap/dist.rs b/src/bootstrap/dist.rs index f6f6f4cc8ea..1c0957bc35e 100644 --- a/src/bootstrap/dist.rs +++ b/src/bootstrap/dist.rs @@ -18,6 +18,7 @@ use std::process::Command; use object::read::archive::ArchiveFile; use object::BinaryFormat; +use sha2::Digest; use crate::bolt::{instrument_with_bolt, optimize_with_bolt}; use crate::builder::{Builder, Kind, RunConfig, ShouldRun, Step}; @@ -1915,9 +1916,9 @@ fn install_llvm_file(builder: &Builder<'_>, source: &Path, destination: &Path) { // We perform the instrumentation/optimization here, on the fly, just before they are being // packaged into some destination directory. let postprocessed = if builder.config.llvm_bolt_profile_generate { - instrument_with_bolt(source) + builder.ensure(BoltInstrument::new(source.to_path_buf())) } else if let Some(path) = &builder.config.llvm_bolt_profile_use { - optimize_with_bolt(source, &Path::new(&path)) + builder.ensure(BoltOptimize::new(source.to_path_buf(), path.into())) } else { source.to_path_buf() }; @@ -2007,6 +2008,121 @@ pub fn maybe_install_llvm_runtime(builder: &Builder<'_>, target: TargetSelection } } +/// Creates an output path to a BOLT-manipulated artifact for the given `file`. +/// The hash of the file is used to make sure that we don't mix BOLT artifacts amongst different +/// files with the same name. +/// +/// We need to keep the file-name the same though, to make sure that copying the manipulated file +/// to a directory will not change the final file path. +fn create_bolt_output_path(builder: &Builder<'_>, file: &Path, hash: &str) -> PathBuf { + let directory = builder.out.join("bolt").join(hash); + t!(fs::create_dir_all(&directory)); + directory.join(file.file_name().unwrap()) +} + +/// Instrument the provided file with BOLT. +/// Returns a path to the instrumented artifact. +#[derive(Clone, Debug, Eq, Hash, PartialEq)] +pub struct BoltInstrument { + file: PathBuf, + hash: String, +} + +impl BoltInstrument { + fn new(file: PathBuf) -> Self { + let mut hasher = sha2::Sha256::new(); + hasher.update(t!(fs::read(&file))); + let hash = hex::encode(hasher.finalize().as_slice()); + + Self { file, hash } + } +} + +impl Step for BoltInstrument { + type Output = PathBuf; + + const ONLY_HOSTS: bool = false; + const DEFAULT: bool = false; + + fn should_run(run: ShouldRun<'_>) -> ShouldRun<'_> { + run.never() + } + + fn make_run(_run: RunConfig<'_>) {} + + fn run(self, builder: &Builder<'_>) -> PathBuf { + if builder.build.config.dry_run() { + return self.file.clone(); + } + + if builder.build.config.llvm_from_ci { + println!("warning: trying to use BOLT with LLVM from CI, this will probably not work"); + } + + println!("Instrumenting {} with BOLT", self.file.display()); + + let output_path = create_bolt_output_path(builder, &self.file, &self.hash); + if !output_path.is_file() { + instrument_with_bolt(&self.file, &output_path); + } + output_path + } +} + +/// Optimize the provided file with BOLT. +/// Returns a path to the optimized artifact. +/// +/// The hash is stored in the step to make sure that we don't optimize the same file +/// twice (even under different file paths). +#[derive(Clone, Debug, Eq, Hash, PartialEq)] +pub struct BoltOptimize { + file: PathBuf, + profile: PathBuf, + hash: String, +} + +impl BoltOptimize { + fn new(file: PathBuf, profile: PathBuf) -> Self { + let mut hasher = sha2::Sha256::new(); + hasher.update(t!(fs::read(&file))); + hasher.update(t!(fs::read(&profile))); + let hash = hex::encode(hasher.finalize().as_slice()); + + Self { file, profile, hash } + } +} + +impl Step for BoltOptimize { + type Output = PathBuf; + + const ONLY_HOSTS: bool = false; + const DEFAULT: bool = false; + + fn should_run(run: ShouldRun<'_>) -> ShouldRun<'_> { + run.never() + } + + fn make_run(_run: RunConfig<'_>) {} + + fn run(self, builder: &Builder<'_>) -> PathBuf { + if builder.build.config.dry_run() { + return self.file.clone(); + } + + if builder.build.config.llvm_from_ci { + println!("warning: trying to use BOLT with LLVM from CI, this will probably not work"); + } + + println!("Optimizing {} with BOLT", self.file.display()); + + let output_path = create_bolt_output_path(builder, &self.file, &self.hash); + if !output_path.is_file() { + optimize_with_bolt(&self.file, &self.profile, &output_path); + } + output_path + } +} + #[derive(Clone, Debug, Eq, Hash, PartialEq)] pub struct LlvmTools { pub target: TargetSelection, From 91bb563e13a20f84b2f2027a33f4855f19720c67 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jakub=20Ber=C3=A1nek?= Date: Wed, 8 Feb 2023 08:49:51 +0100 Subject: [PATCH 3/4] Try to avoid the last rustc rebuild --- src/ci/stage-build.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/ci/stage-build.py b/src/ci/stage-build.py index ebd111e6453..fe3083dc31e 100644 --- a/src/ci/stage-build.py +++ b/src/ci/stage-build.py @@ -798,12 +798,15 @@ def execute_build_pipeline(timer: Timer, pipeline: Pipeline, final_build_args: L "--llvm-profile-use", pipeline.llvm_profile_merged_file(), "--llvm-bolt-profile-generate", + "--rust-profile-use", + pipeline.rustc_profile_merged_file() ]) record_metrics(pipeline, rustc_build) with stage3.section("Gather profiles"): gather_llvm_bolt_profiles(pipeline) + # LLVM is not being cleared here, we want to reuse the previous build print_free_disk_space(pipeline) final_build_args += [ "--llvm-bolt-profile-use", From 9aad2ad3615468463cfa036801525bc2f7e3553a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jakub=20Ber=C3=A1nek?= Date: Sat, 4 Mar 2023 17:36:00 +0100 Subject: [PATCH 4/4] Add check for dry run --- src/bootstrap/dist.rs | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/src/bootstrap/dist.rs b/src/bootstrap/dist.rs index 1c0957bc35e..d7008df4179 100644 --- a/src/bootstrap/dist.rs +++ b/src/bootstrap/dist.rs @@ -1907,7 +1907,7 @@ fn add_env(builder: &Builder<'_>, cmd: &mut Command, target: TargetSelection) { } fn install_llvm_file(builder: &Builder<'_>, source: &Path, destination: &Path) { - if source.as_os_str().is_empty() { + if builder.config.dry_run() { return; } @@ -2048,8 +2048,6 @@ impl Step for BoltInstrument { run.never() } - fn make_run(_run: RunConfig<'_>) {} - fn run(self, builder: &Builder<'_>) -> PathBuf { if builder.build.config.dry_run() { return self.file.clone(); @@ -2102,8 +2100,6 @@ impl Step for BoltOptimize { run.never() } - fn make_run(_run: RunConfig<'_>) {} - fn run(self, builder: &Builder<'_>) -> PathBuf { if builder.build.config.dry_run() { return self.file.clone();