enable PGO on x86_64-apple-darwin

Author: csmoe

.github/workflows/ci.yml

@@ -317,7 +317,7 @@ jobs:
             os: ubuntu-20.04-8core-32gb
           - name: dist-x86_64-apple
             env:
-              SCRIPT: "./x.py dist bootstrap --include-default-paths --host=x86_64-apple-darwin --target=x86_64-apple-darwin"
+              SCRIPT: PGO_HOST=x86_64-apple-darwin python3 src/ci/stage-build.py python3 ./x.py dist bootstrap --include-default-paths --host=x86_64-apple-darwin --target=x86_64-apple-darwin
               RUST_CONFIGURE_ARGS: "--enable-full-tools --enable-sanitizers --enable-profiler --set rust.jemalloc --set llvm.ninja=false --set rust.lto=thin"
               RUSTC_RETRY_LINKER_ON_SEGFAULT: 1
               MACOSX_DEPLOYMENT_TARGET: 10.7

src/bootstrap/compile.rs

@@ -950,6 +950,7 @@ fn rustc_llvm_env(builder: &Builder<'_>, cargo: &mut Cargo, target: TargetSelect
             } else {
                 panic!("clang has no clang_rt.profile library for {target}");
             };
+
             let clang = builder.cc(target);
             let clang_rt_dir = get_clang_rt_dir(clang, false);
             let clang_rt_profile_lib = format!("libclang_rt.profile_{clang_rt_profile_lib_suffix}");

src/bootstrap/llvm.rs

@@ -331,7 +331,10 @@ impl Step for Llvm {
         // This flag makes sure `FileCheck` is copied in the final binaries directory.
         cfg.define("LLVM_INSTALL_UTILS", "ON");
 
+        let mut enabled_llvm_projects = Vec::new();
+
         if builder.config.llvm_profile_generate {
+            enabled_llvm_projects.push("compiler-rt");
             cfg.define("LLVM_BUILD_INSTRUMENTED", "IR");
             if let Ok(llvm_profile_dir) = std::env::var("LLVM_PROFILE_DIR") {
                 cfg.define("LLVM_PROFILE_DATA_DIR", llvm_profile_dir);
@@ -344,6 +347,7 @@ impl Step for Llvm {
         if builder.config.llvm_bolt_profile_generate
             || builder.config.llvm_bolt_profile_use.is_some()
         {
+            enabled_llvm_projects.push("bolt");
             // Relocations are required for BOLT to work.
             ldflags.push_all("-Wl,-q");
         }
@@ -403,8 +407,6 @@ impl Step for Llvm {
             cfg.define("LLVM_BUILD_32_BITS", "ON");
         }
 
-        let mut enabled_llvm_projects = Vec::new();
-
         if util::forcing_clang_based_tests() {
             enabled_llvm_projects.push("clang");
             enabled_llvm_projects.push("compiler-rt");
@@ -828,7 +830,7 @@ impl Step for Lld {
             if let Some(clang_cl_path) = builder.config.llvm_clang_cl.as_ref() {
                 // Find clang's runtime library directory and push that as a search path to the
                 // cmake linker flags.
-                let clang_rt_dir = get_clang_rt_dir(clang_cl_path);
+                let clang_rt_dir = get_clang_rt_dir(clang_cl_path, true);
                 ldflags.push_all(&format!("/libpath:{}", clang_rt_dir.display()));
             }
         }

src/bootstrap/util.rs

@@ -488,32 +488,22 @@ fn absolute_windows(path: &std::path::Path) -> std::io::Result<std::path::PathBu
     }
 }
 
-/// Adapted from https://github.com/llvm/llvm-project/blob/782e91224601e461c019e0a4573bbccc6094fbcd/llvm/cmake/modules/HandleLLVMOptions.cmake#L1058-L1079
-///
-/// When `clang-cl` is used with instrumentation, we need to add clang's runtime library resource
-/// directory to the linker flags, otherwise there will be linker errors about the profiler runtime
-/// missing. This function returns the path to that directory.
 pub fn get_clang_rt_dir(clang: impl AsRef<Path>, is_msvc: bool) -> PathBuf {
-    // Similar to how LLVM does it, to find clang's library runtime directory:
-    // - we ask `clang-cl` to locate the `clang_rt.builtins` lib.
-    let mut builtins_locator = Command::new(clang.as_ref());
+    let mut cmd = Command::new(clang.as_ref());
     if is_msvc {
-        builtins_locator.args(&["/clang:-print-libgcc-file-name", "/clang:--rtlib=compiler-rt"]);
+        cmd.args(&["/clang:-print-runtime-dir"]);
     } else {
-        builtins_locator.args(&["-print-libgcc-file-name", "-rtlib=compiler-rt"]);
+        cmd.args(&["-print-runtime-dir"]);
     };
 
-    let clang_rt_builtins = output(&mut builtins_locator);
-    let clang_rt_builtins = Path::new(clang_rt_builtins.trim());
+    let dir = output(&mut cmd);
+    let dir = PathBuf::from(dir.trim());
     assert!(
-        clang_rt_builtins.exists(),
-        "`clang-cl` must correctly locate the library runtime directory"
+        dir.exists(),
+        "`{}` must correctly locate the library runtime directory",
+        clang.as_ref().display()
     );
-
-    // - the profiler runtime will be located in the same directory as the builtins lib, like
-    // `$LLVM_DISTRO_ROOT/lib/clang/$LLVM_VERSION/lib/windows`.
-    let clang_rt_dir = clang_rt_builtins.parent().expect("The clang lib folder should exist");
-    clang_rt_dir.to_path_buf()
+    dir
 }
 
 pub fn lld_flag_no_threads(is_windows: bool) -> &'static str {

src/ci/github-actions/ci.yml

@@ -491,7 +491,7 @@ jobs:
 
           - name: dist-x86_64-apple
             env:
-              SCRIPT: ./x.py dist bootstrap --include-default-paths --host=x86_64-apple-darwin --target=x86_64-apple-darwin
+              SCRIPT: PGO_HOST=x86_64-apple-darwin python3 src/ci/stage-build.py python3 ./x.py dist bootstrap --include-default-paths --host=x86_64-apple-darwin --target=x86_64-apple-darwin
               RUST_CONFIGURE_ARGS: --enable-full-tools --enable-sanitizers --enable-profiler --set rust.jemalloc --set llvm.ninja=false --set rust.lto=thin
               RUSTC_RETRY_LINKER_ON_SEGFAULT: 1
               MACOSX_DEPLOYMENT_TARGET: 10.7

src/ci/scripts/install-clang.sh

@@ -18,10 +18,15 @@ if isMacOS; then
     if [[ ${USE_XCODE_CLANG-0} -eq 1 ]]; then
         bindir="$(xcode-select --print-path)/Toolchains/XcodeDefault.xctoolchain/usr/bin"
     else
-        file="${MIRRORS_BASE}/clang%2Bllvm-${LLVM_VERSION}-x86_64-apple-darwin.tar.xz"
-        retry curl -f "${file}" -o "clang+llvm-${LLVM_VERSION}-x86_64-apple-darwin.tar.xz"
-        tar xJf "clang+llvm-${LLVM_VERSION}-x86_64-apple-darwin.tar.xz"
-        bindir="$(pwd)/clang+llvm-${LLVM_VERSION}-x86_64-apple-darwin/bin"
+        OSX_ARCH=$(uname -m)
+        file="${MIRRORS_BASE}/clang%2Bllvm-${LLVM_VERSION}-${OSX_ARCH}-apple-darwin.tar.xz"
+        retry curl -f "${file}" -o "clang+llvm-${LLVM_VERSION}-${OSX_ARCH}-apple-darwin.tar.xz"
+
+        mkdir -p citools && cd citools
+
+        tar xJf "clang+llvm-${LLVM_VERSION}-${OSX_ARCH}-apple-darwin.tar.xz"
+        mv "clang+llvm-${LLVM_VERSION}-${OSX_ARCH}-apple-darwin" clang-rust
+        bindir="$(pwd)/clang-rust/bin"
     fi
 
     ciCommandSetEnv CC "${bindir}/clang"

src/ci/stage-build.py

@@ -207,6 +207,47 @@ def rustc_profile_template_path(self) -> Path:
     def supports_bolt(self) -> bool:
         return False
 
+class DarwinPipeline(Pipeline):
+    def __init__(self):
+        self.checkout_dir = Path(os.getcwd())
+
+    def checkout_path(self) -> Path:
+        return self.checkout_dir
+
+    def downloaded_llvm_dir(self) -> Path:
+        return self.checkout_path() / "citools" / "clang-rust"
+
+    def build_root(self) -> Path:
+        return self.checkout_path()
+
+    def opt_artifacts(self) -> Path:
+        return Path("/tmp/tmp-multistage/opt-artifacts")
+
+    def build_rustc_perf(self):
+        # rustc-perf version from 2022-07-22
+        perf_commit = "9dfaa35193154b690922347ee1141a06ec87a199"
+        rustc_perf_zip_path = self.opt_artifacts() / "perf.zip"
+
+        def download_rustc_perf():
+            download_file(
+                f"https://github.com/rust-lang/rustc-perf/archive/{perf_commit}.zip",
+                rustc_perf_zip_path
+            )
+            with change_cwd(self.opt_artifacts()):
+                unpack_archive(rustc_perf_zip_path)
+                move_path(Path(f"rustc-perf-{perf_commit}"), self.rustc_perf_dir())
+                delete_file(rustc_perf_zip_path)
+
+        retry_action(download_rustc_perf, "Download rustc-perf")
+
+        with change_cwd(self.rustc_perf_dir()):
+            cmd([self.cargo_stage_0(), "build", "-p", "collector"], env=dict(
+                RUSTC=str(self.rustc_stage_0()),
+                RUSTC_BOOTSTRAP="1"
+            ))
+
+    def supports_bolt(self) -> bool:
+        return False
 
 def get_timestamp() -> float:
     return time.time()
@@ -576,6 +617,8 @@ def create_pipeline() -> Pipeline:
         return LinuxPipeline()
     elif sys.platform in ("cygwin", "win32"):
         return WindowsPipeline()
+    elif sys.platform == "darwin":
+        return DarwinPipeline()
     else:
         raise Exception(f"Optimized build is not supported for platform {sys.platform}")