src/bootstrap/download.rs

   1 use std::{
   2     env,
   3     ffi::{OsStr, OsString},
   4     fs::{self, File},
   5     io::{BufRead, BufReader, ErrorKind},
   6     path::{Path, PathBuf},
   7     process::{Command, Stdio},
   8 };
   9
  10 use once_cell::sync::OnceCell;
  11 use xz2::bufread::XzDecoder;
  12
  13 use crate::{
  14     config::RustfmtMetadata,
  15     native::detect_llvm_sha,
  16     t,
  17     util::{check_run, exe, program_out_of_date, try_run},
  18     Config,
  19 };
  20
  21 static SHOULD_FIX_BINS_AND_DYLIBS: OnceCell<bool> = OnceCell::new();
  22
  23 /// Generic helpers that are useful anywhere in bootstrap.
  24 impl Config {
  25     pub fn is_verbose(&self) -> bool {
  26         self.verbose > 0
  27     }
  28
  29     pub(crate) fn create(&self, path: &Path, s: &str) {
  30         if self.dry_run() {
  31             return;
  32         }
  33         t!(fs::write(path, s));
  34     }
  35
  36     pub(crate) fn remove(&self, f: &Path) {
  37         if self.dry_run() {
  38             return;
  39         }
  40         fs::remove_file(f).unwrap_or_else(|_| panic!("failed to remove {:?}", f));
  41     }
  42
  43     /// Create a temporary directory in `out` and return its path.
  44     ///
  45     /// NOTE: this temporary directory is shared between all steps;
  46     /// if you need an empty directory, create a new subdirectory inside it.
  47     pub(crate) fn tempdir(&self) -> PathBuf {
  48         let tmp = self.out.join("tmp");
  49         t!(fs::create_dir_all(&tmp));
  50         tmp
  51     }
  52
  53     /// Runs a command, printing out nice contextual information if it fails.
  54     /// Exits if the command failed to execute at all, otherwise returns its
  55     /// `status.success()`.
  56     pub(crate) fn try_run(&self, cmd: &mut Command) -> bool {
  57         if self.dry_run() {
  58             return true;
  59         }
  60         self.verbose(&format!("running: {:?}", cmd));
  61         try_run(cmd, self.is_verbose())
  62     }
  63
  64     /// Runs a command, printing out nice contextual information if it fails.
  65     /// Returns false if do not execute at all, otherwise returns its
  66     /// `status.success()`.
  67     pub(crate) fn check_run(&self, cmd: &mut Command) -> bool {
  68         if self.dry_run() {
  69             return true;
  70         }
  71         self.verbose(&format!("running: {:?}", cmd));
  72         check_run(cmd, self.is_verbose())
  73     }
  74
  75     /// Whether or not `fix_bin_or_dylib` needs to be run; can only be true
  76     /// on NixOS
  77     fn should_fix_bins_and_dylibs(&self) -> bool {
  78         let val = *SHOULD_FIX_BINS_AND_DYLIBS.get_or_init(|| {
  79             match Command::new("uname").arg("-s").stderr(Stdio::inherit()).output() {
  80                 Err(_) => return false,
  81                 Ok(output) if !output.status.success() => return false,
  82                 Ok(output) => {
  83                     let mut os_name = output.stdout;
  84                     if os_name.last() == Some(&b'\n') {
  85                         os_name.pop();
  86                     }
  87                     if os_name != b"Linux" {
  88                         return false;
  89                     }
  90                 }
  91             }
  92
  93             // If the user has asked binaries to be patched for Nix, then
  94             // don't check for NixOS or `/lib`.
  95             // NOTE: this intentionally comes after the Linux check:
  96             // - patchelf only works with ELF files, so no need to run it on Mac or Windows
  97             // - On other Unix systems, there is no stable syscall interface, so Nix doesn't manage the global libc.
  98             if self.patch_binaries_for_nix {
  99                 return true;
 100             }
 101
 102             // Use `/etc/os-release` instead of `/etc/NIXOS`.
 103             // The latter one does not exist on NixOS when using tmpfs as root.
 104             let is_nixos = match File::open("/etc/os-release") {
 105                 Err(e) if e.kind() == ErrorKind::NotFound => false,
 106                 Err(e) => panic!("failed to access /etc/os-release: {}", e),
 107                 Ok(os_release) => BufReader::new(os_release).lines().any(|l| {
 108                     let l = l.expect("reading /etc/os-release");
 109                     matches!(l.trim(), "ID=nixos" | "ID='nixos'" | "ID=\"nixos\"")
 110                 }),
 111             };
 112             is_nixos && !Path::new("/lib").exists()
 113         });
 114         if val {
 115             println!("info: You seem to be using Nix.");
 116         }
 117         val
 118     }
 119
 120     /// Modifies the interpreter section of 'fname' to fix the dynamic linker,
 121     /// or the RPATH section, to fix the dynamic library search path
 122     ///
 123     /// This is only required on NixOS and uses the PatchELF utility to
 124     /// change the interpreter/RPATH of ELF executables.
 125     ///
 126     /// Please see https://nixos.org/patchelf.html for more information
 127     fn fix_bin_or_dylib(&self, fname: &Path) {
 128         assert_eq!(SHOULD_FIX_BINS_AND_DYLIBS.get(), Some(&true));
 129         println!("attempting to patch {}", fname.display());
 130
 131         // Only build `.nix-deps` once.
 132         static NIX_DEPS_DIR: OnceCell<PathBuf> = OnceCell::new();
 133         let mut nix_build_succeeded = true;
 134         let nix_deps_dir = NIX_DEPS_DIR.get_or_init(|| {
 135             // Run `nix-build` to "build" each dependency (which will likely reuse
 136             // the existing `/nix/store` copy, or at most download a pre-built copy).
 137             //
 138             // Importantly, we create a gc-root called `.nix-deps` in the `build/`
 139             // directory, but still reference the actual `/nix/store` path in the rpath
 140             // as it makes it significantly more robust against changes to the location of
 141             // the `.nix-deps` location.
 142             //
 143             // bintools: Needed for the path of `ld-linux.so` (via `nix-support/dynamic-linker`).
 144             // zlib: Needed as a system dependency of `libLLVM-*.so`.
 145             // patchelf: Needed for patching ELF binaries (see doc comment above).
 146             let nix_deps_dir = self.out.join(".nix-deps");
 147             const NIX_EXPR: &str = "
 148             with (import <nixpkgs> {});
 149             symlinkJoin {
 150                 name = \"rust-stage0-dependencies\";
 151                 paths = [
 152                     zlib
 153                     patchelf
 154                     stdenv.cc.bintools
 155                 ];
 156             }
 157             ";
 158             nix_build_succeeded = self.try_run(Command::new("nix-build").args(&[
 159                 Path::new("-E"),
 160                 Path::new(NIX_EXPR),
 161                 Path::new("-o"),
 162                 &nix_deps_dir,
 163             ]));
 164             nix_deps_dir
 165         });
 166         if !nix_build_succeeded {
 167             return;
 168         }
 169
 170         let mut patchelf = Command::new(nix_deps_dir.join("bin/patchelf"));
 171         let rpath_entries = {
 172             // ORIGIN is a relative default, all binary and dynamic libraries we ship
 173             // appear to have this (even when `../lib` is redundant).
 174             // NOTE: there are only two paths here, delimited by a `:`
 175             let mut entries = OsString::from("$ORIGIN/../lib:");
 176             entries.push(t!(fs::canonicalize(nix_deps_dir)));
 177             entries.push("/lib");
 178             entries
 179         };
 180         patchelf.args(&[OsString::from("--set-rpath"), rpath_entries]);
 181         if !fname.extension().map_or(false, |ext| ext == "so") {
 182             // Finally, set the correct .interp for binaries
 183             let dynamic_linker_path = nix_deps_dir.join("nix-support/dynamic-linker");
 184             // FIXME: can we support utf8 here? `args` doesn't accept Vec<u8>, only OsString ...
 185             let dynamic_linker = t!(String::from_utf8(t!(fs::read(dynamic_linker_path))));
 186             patchelf.args(&["--set-interpreter", dynamic_linker.trim_end()]);
 187         }
 188
 189         self.try_run(patchelf.arg(fname));
 190     }
 191
 192     fn download_file(&self, url: &str, dest_path: &Path, help_on_error: &str) {
 193         self.verbose(&format!("download {url}"));
 194         // Use a temporary file in case we crash while downloading, to avoid a corrupt download in cache/.
 195         let tempfile = self.tempdir().join(dest_path.file_name().unwrap());
 196         // While bootstrap itself only supports http and https downloads, downstream forks might
 197         // need to download components from other protocols. The match allows them adding more
 198         // protocols without worrying about merge conflicts if we change the HTTP implementation.
 199         match url.split_once("://").map(|(proto, _)| proto) {
 200             Some("http") | Some("https") => {
 201                 self.download_http_with_retries(&tempfile, url, help_on_error)
 202             }
 203             Some(other) => panic!("unsupported protocol {other} in {url}"),
 204             None => panic!("no protocol in {url}"),
 205         }
 206         t!(std::fs::rename(&tempfile, dest_path));
 207     }
 208
 209     fn download_http_with_retries(&self, tempfile: &Path, url: &str, help_on_error: &str) {
 210         println!("downloading {}", url);
 211         // Try curl. If that fails and we are on windows, fallback to PowerShell.
 212         let mut curl = Command::new("curl");
 213         curl.args(&[
 214             "-#",
 215             "-y",
 216             "30",
 217             "-Y",
 218             "10", // timeout if speed is < 10 bytes/sec for > 30 seconds
 219             "--connect-timeout",
 220             "30", // timeout if cannot connect within 30 seconds
 221             "--retry",
 222             "3",
 223             "-Sf",
 224             "-o",
 225         ]);
 226         curl.arg(tempfile);
 227         curl.arg(url);
 228         if !self.check_run(&mut curl) {
 229             if self.build.contains("windows-msvc") {
 230                 println!("Fallback to PowerShell");
 231                 for _ in 0..3 {
 232                     if self.try_run(Command::new("PowerShell.exe").args(&[
 233                         "/nologo",
 234                         "-Command",
 235                         "[Net.ServicePointManager]::SecurityProtocol = [Net.SecurityProtocolType]::Tls12;",
 236                         &format!(
 237                             "(New-Object System.Net.WebClient).DownloadFile('{}', '{}')",
 238                             url, tempfile.to_str().expect("invalid UTF-8 not supported with powershell downloads"),
 239                         ),
 240                     ])) {
 241                         return;
 242                     }
 243                     println!("\nspurious failure, trying again");
 244                 }
 245             }
 246             if !help_on_error.is_empty() {
 247                 eprintln!("{}", help_on_error);
 248             }
 249             crate::detail_exit(1);
 250         }
 251     }
 252
 253     fn unpack(&self, tarball: &Path, dst: &Path, pattern: &str) {
 254         println!("extracting {} to {}", tarball.display(), dst.display());
 255         if !dst.exists() {
 256             t!(fs::create_dir_all(dst));
 257         }
 258
 259         // `tarball` ends with `.tar.xz`; strip that suffix
 260         // example: `rust-dev-nightly-x86_64-unknown-linux-gnu`
 261         let uncompressed_filename =
 262             Path::new(tarball.file_name().expect("missing tarball filename")).file_stem().unwrap();
 263         let directory_prefix = Path::new(Path::new(uncompressed_filename).file_stem().unwrap());
 264
 265         // decompress the file
 266         let data = t!(File::open(tarball));
 267         let decompressor = XzDecoder::new(BufReader::new(data));
 268
 269         let mut tar = tar::Archive::new(decompressor);
 270         for member in t!(tar.entries()) {
 271             let mut member = t!(member);
 272             let original_path = t!(member.path()).into_owned();
 273             // skip the top-level directory
 274             if original_path == directory_prefix {
 275                 continue;
 276             }
 277             let mut short_path = t!(original_path.strip_prefix(directory_prefix));
 278             if !short_path.starts_with(pattern) {
 279                 continue;
 280             }
 281             short_path = t!(short_path.strip_prefix(pattern));
 282             let dst_path = dst.join(short_path);
 283             self.verbose(&format!("extracting {} to {}", original_path.display(), dst.display()));
 284             if !t!(member.unpack_in(dst)) {
 285                 panic!("path traversal attack ??");
 286             }
 287             let src_path = dst.join(original_path);
 288             if src_path.is_dir() && dst_path.exists() {
 289                 continue;
 290             }
 291             t!(fs::rename(src_path, dst_path));
 292         }
 293         t!(fs::remove_dir_all(dst.join(directory_prefix)));
 294     }
 295
 296     /// Returns whether the SHA256 checksum of `path` matches `expected`.
 297     fn verify(&self, path: &Path, expected: &str) -> bool {
 298         use sha2::Digest;
 299
 300         self.verbose(&format!("verifying {}", path.display()));
 301         let mut hasher = sha2::Sha256::new();
 302         // FIXME: this is ok for rustfmt (4.1 MB large at time of writing), but it seems memory-intensive for rustc and larger components.
 303         // Consider using streaming IO instead?
 304         let contents = if self.dry_run() { vec![] } else { t!(fs::read(path)) };
 305         hasher.update(&contents);
 306         let found = hex::encode(hasher.finalize().as_slice());
 307         let verified = found == expected;
 308         if !verified && !self.dry_run() {
 309             println!(
 310                 "invalid checksum: \n\
 311                 found:    {found}\n\
 312                 expected: {expected}",
 313             );
 314         }
 315         return verified;
 316     }
 317 }
 318
 319 enum DownloadSource {
 320     CI,
 321     Dist,
 322 }
 323
 324 /// Functions that are only ever called once, but named for clarify and to avoid thousand-line functions.
 325 impl Config {
 326     pub(crate) fn maybe_download_rustfmt(&self) -> Option<PathBuf> {
 327         let RustfmtMetadata { date, version } = self.stage0_metadata.rustfmt.as_ref()?;
 328         let channel = format!("{version}-{date}");
 329
 330         let host = self.build;
 331         let bin_root = self.out.join(host.triple).join("rustfmt");
 332         let rustfmt_path = bin_root.join("bin").join(exe("rustfmt", host));
 333         let rustfmt_stamp = bin_root.join(".rustfmt-stamp");
 334         if rustfmt_path.exists() && !program_out_of_date(&rustfmt_stamp, &channel) {
 335             return Some(rustfmt_path);
 336         }
 337
 338         self.download_component(
 339             DownloadSource::Dist,
 340             format!("rustfmt-{version}-{build}.tar.xz", build = host.triple),
 341             "rustfmt-preview",
 342             &date,
 343             "rustfmt",
 344         );
 345         self.download_component(
 346             DownloadSource::Dist,
 347             format!("rustc-{version}-{build}.tar.xz", build = host.triple),
 348             "rustc",
 349             &date,
 350             "rustfmt",
 351         );
 352
 353         if self.should_fix_bins_and_dylibs() {
 354             self.fix_bin_or_dylib(&bin_root.join("bin").join("rustfmt"));
 355             self.fix_bin_or_dylib(&bin_root.join("bin").join("cargo-fmt"));
 356         }
 357
 358         self.create(&rustfmt_stamp, &channel);
 359         Some(rustfmt_path)
 360     }
 361
 362     pub(crate) fn download_ci_rustc(&self, commit: &str) {
 363         self.verbose(&format!("using downloaded stage2 artifacts from CI (commit {commit})"));
 364         let version = self.artifact_version_part(commit);
 365         let host = self.build.triple;
 366         let bin_root = self.out.join(host).join("ci-rustc");
 367         let rustc_stamp = bin_root.join(".rustc-stamp");
 368
 369         if !bin_root.join("bin").join("rustc").exists() || program_out_of_date(&rustc_stamp, commit)
 370         {
 371             if bin_root.exists() {
 372                 t!(fs::remove_dir_all(&bin_root));
 373             }
 374             let filename = format!("rust-std-{version}-{host}.tar.xz");
 375             let pattern = format!("rust-std-{host}");
 376             self.download_ci_component(filename, &pattern, commit);
 377             let filename = format!("rustc-{version}-{host}.tar.xz");
 378             self.download_ci_component(filename, "rustc", commit);
 379             // download-rustc doesn't need its own cargo, it can just use beta's.
 380             let filename = format!("rustc-dev-{version}-{host}.tar.xz");
 381             self.download_ci_component(filename, "rustc-dev", commit);
 382             let filename = format!("rust-src-{version}.tar.xz");
 383             self.download_ci_component(filename, "rust-src", commit);
 384
 385             if self.should_fix_bins_and_dylibs() {
 386                 self.fix_bin_or_dylib(&bin_root.join("bin").join("rustc"));
 387                 self.fix_bin_or_dylib(&bin_root.join("bin").join("rustdoc"));
 388                 self.fix_bin_or_dylib(
 389                     &bin_root.join("libexec").join("rust-analyzer-proc-macro-srv"),
 390                 );
 391                 let lib_dir = bin_root.join("lib");
 392                 for lib in t!(fs::read_dir(&lib_dir), lib_dir.display().to_string()) {
 393                     let lib = t!(lib);
 394                     if lib.path().extension() == Some(OsStr::new("so")) {
 395                         self.fix_bin_or_dylib(&lib.path());
 396                     }
 397                 }
 398             }
 399
 400             t!(fs::write(rustc_stamp, commit));
 401         }
 402     }
 403
 404     /// Download a single component of a CI-built toolchain (not necessarily a published nightly).
 405     // NOTE: intentionally takes an owned string to avoid downloading multiple times by accident
 406     fn download_ci_component(&self, filename: String, prefix: &str, commit: &str) {
 407         Self::download_component(self, DownloadSource::CI, filename, prefix, commit, "ci-rustc")
 408     }
 409
 410     fn download_component(
 411         &self,
 412         mode: DownloadSource,
 413         filename: String,
 414         prefix: &str,
 415         key: &str,
 416         destination: &str,
 417     ) {
 418         let cache_dst = self.out.join("cache");
 419         let cache_dir = cache_dst.join(key);
 420         if !cache_dir.exists() {
 421             t!(fs::create_dir_all(&cache_dir));
 422         }
 423
 424         let bin_root = self.out.join(self.build.triple).join(destination);
 425         let tarball = cache_dir.join(&filename);
 426         let (base_url, url, should_verify) = match mode {
 427             DownloadSource::CI => (
 428                 self.stage0_metadata.config.artifacts_server.clone(),
 429                 format!("{key}/{filename}"),
 430                 false,
 431             ),
 432             DownloadSource::Dist => {
 433                 let dist_server = env::var("RUSTUP_DIST_SERVER")
 434                     .unwrap_or(self.stage0_metadata.config.dist_server.to_string());
 435                 // NOTE: make `dist` part of the URL because that's how it's stored in src/stage0.json
 436                 (dist_server, format!("dist/{key}/{filename}"), true)
 437             }
 438         };
 439
 440         // For the beta compiler, put special effort into ensuring the checksums are valid.
 441         // FIXME: maybe we should do this for download-rustc as well? but it would be a pain to update
 442         // this on each and every nightly ...
 443         let checksum = if should_verify {
 444             let error = format!(
 445                 "src/stage0.json doesn't contain a checksum for {url}. \
 446                 Pre-built artifacts might not be available for this \
 447                 target at this time, see https://doc.rust-lang.org/nightly\
 448                 /rustc/platform-support.html for more information."
 449             );
 450             let sha256 = self.stage0_metadata.checksums_sha256.get(&url).expect(&error);
 451             if tarball.exists() {
 452                 if self.verify(&tarball, sha256) {
 453                     self.unpack(&tarball, &bin_root, prefix);
 454                     return;
 455                 } else {
 456                     self.verbose(&format!(
 457                         "ignoring cached file {} due to failed verification",
 458                         tarball.display()
 459                     ));
 460                     self.remove(&tarball);
 461                 }
 462             }
 463             Some(sha256)
 464         } else if tarball.exists() {
 465             self.unpack(&tarball, &bin_root, prefix);
 466             return;
 467         } else {
 468             None
 469         };
 470
 471         self.download_file(&format!("{base_url}/{url}"), &tarball, "");
 472         if let Some(sha256) = checksum {
 473             if !self.verify(&tarball, sha256) {
 474                 panic!("failed to verify {}", tarball.display());
 475             }
 476         }
 477
 478         self.unpack(&tarball, &bin_root, prefix);
 479     }
 480
 481     pub(crate) fn maybe_download_ci_llvm(&self) {
 482         if !self.llvm_from_ci {
 483             return;
 484         }
 485         let llvm_root = self.ci_llvm_root();
 486         let llvm_stamp = llvm_root.join(".llvm-stamp");
 487         let llvm_sha = detect_llvm_sha(&self, self.rust_info.is_managed_git_subrepository());
 488         let key = format!("{}{}", llvm_sha, self.llvm_assertions);
 489         if program_out_of_date(&llvm_stamp, &key) && !self.dry_run() {
 490             self.download_ci_llvm(&llvm_sha);
 491             if self.should_fix_bins_and_dylibs() {
 492                 for entry in t!(fs::read_dir(llvm_root.join("bin"))) {
 493                     self.fix_bin_or_dylib(&t!(entry).path());
 494                 }
 495             }
 496
 497             // Update the timestamp of llvm-config to force rustc_llvm to be
 498             // rebuilt. This is a hacky workaround for a deficiency in Cargo where
 499             // the rerun-if-changed directive doesn't handle changes very well.
 500             // https://github.com/rust-lang/cargo/issues/10791
 501             // Cargo only compares the timestamp of the file relative to the last
 502             // time `rustc_llvm` build script ran. However, the timestamps of the
 503             // files in the tarball are in the past, so it doesn't trigger a
 504             // rebuild.
 505             let now = filetime::FileTime::from_system_time(std::time::SystemTime::now());
 506             let llvm_config = llvm_root.join("bin").join(exe("llvm-config", self.build));
 507             t!(filetime::set_file_times(&llvm_config, now, now));
 508
 509             if self.should_fix_bins_and_dylibs() {
 510                 let llvm_lib = llvm_root.join("lib");
 511                 for entry in t!(fs::read_dir(&llvm_lib)) {
 512                     let lib = t!(entry).path();
 513                     if lib.extension().map_or(false, |ext| ext == "so") {
 514                         self.fix_bin_or_dylib(&lib);
 515                     }
 516                 }
 517             }
 518
 519             t!(fs::write(llvm_stamp, key));
 520         }
 521     }
 522
 523     fn download_ci_llvm(&self, llvm_sha: &str) {
 524         let llvm_assertions = self.llvm_assertions;
 525
 526         let cache_prefix = format!("llvm-{}-{}", llvm_sha, llvm_assertions);
 527         let cache_dst = self.out.join("cache");
 528         let rustc_cache = cache_dst.join(cache_prefix);
 529         if !rustc_cache.exists() {
 530             t!(fs::create_dir_all(&rustc_cache));
 531         }
 532         let base = if llvm_assertions {
 533             &self.stage0_metadata.config.artifacts_with_llvm_assertions_server
 534         } else {
 535             &self.stage0_metadata.config.artifacts_server
 536         };
 537         let version = self.artifact_version_part(llvm_sha);
 538         let filename = format!("rust-dev-{}-{}.tar.xz", version, self.build.triple);
 539         let tarball = rustc_cache.join(&filename);
 540         if !tarball.exists() {
 541             let help_on_error = "error: failed to download llvm from ci
 542
 543     help: old builds get deleted after a certain time
 544     help: if trying to compile an old commit of rustc, disable `download-ci-llvm` in config.toml:
 545
 546     [llvm]
 547     download-ci-llvm = false
 548     ";
 549             self.download_file(&format!("{base}/{llvm_sha}/{filename}"), &tarball, help_on_error);
 550         }
 551         let llvm_root = self.ci_llvm_root();
 552         self.unpack(&tarball, &llvm_root, "rust-dev");
 553     }
 554 }