compiler/rustc_codegen_llvm/src/back/archive.rs

   1 //! A helper class for dealing with static archives
   2
   3 use std::env;
   4 use std::ffi::{c_char, c_void, CStr, CString, OsString};
   5 use std::io;
   6 use std::mem;
   7 use std::path::{Path, PathBuf};
   8 use std::ptr;
   9 use std::str;
  10
  11 use crate::common;
  12 use crate::errors::{
  13     DlltoolFailImportLibrary, ErrorCallingDllTool, ErrorCreatingImportLibrary, ErrorWritingDEFFile,
  14 };
  15 use crate::llvm::archive_ro::{ArchiveRO, Child};
  16 use crate::llvm::{self, ArchiveKind, LLVMMachineType, LLVMRustCOFFShortExport};
  17 use rustc_codegen_ssa::back::archive::{
  18     get_native_object_symbols, try_extract_macho_fat_archive, ArArchiveBuilder,
  19     ArchiveBuildFailure, ArchiveBuilder, ArchiveBuilderBuilder, UnknownArchiveKind,
  20 };
  21
  22 use rustc_session::cstore::DllImport;
  23 use rustc_session::Session;
  24
  25 /// Helper for adding many files to an archive.
  26 #[must_use = "must call build() to finish building the archive"]
  27 pub(crate) struct LlvmArchiveBuilder<'a> {
  28     sess: &'a Session,
  29     additions: Vec<Addition>,
  30 }
  31
  32 enum Addition {
  33     File { path: PathBuf, name_in_archive: String },
  34     Archive { path: PathBuf, archive: ArchiveRO, skip: Box<dyn FnMut(&str) -> bool> },
  35 }
  36
  37 impl Addition {
  38     fn path(&self) -> &Path {
  39         match self {
  40             Addition::File { path, .. } | Addition::Archive { path, .. } => path,
  41         }
  42     }
  43 }
  44
  45 fn is_relevant_child(c: &Child<'_>) -> bool {
  46     match c.name() {
  47         Some(name) => !name.contains("SYMDEF"),
  48         None => false,
  49     }
  50 }
  51
  52 /// Map machine type strings to values of LLVM's MachineTypes enum.
  53 fn llvm_machine_type(cpu: &str) -> LLVMMachineType {
  54     match cpu {
  55         "x86_64" => LLVMMachineType::AMD64,
  56         "x86" => LLVMMachineType::I386,
  57         "aarch64" => LLVMMachineType::ARM64,
  58         "arm" => LLVMMachineType::ARM,
  59         _ => panic!("unsupported cpu type {}", cpu),
  60     }
  61 }
  62
  63 impl<'a> ArchiveBuilder<'a> for LlvmArchiveBuilder<'a> {
  64     fn add_archive(
  65         &mut self,
  66         archive: &Path,
  67         skip: Box<dyn FnMut(&str) -> bool + 'static>,
  68     ) -> io::Result<()> {
  69         let mut archive = archive.to_path_buf();
  70         if self.sess.target.llvm_target.contains("-apple-macosx") {
  71             if let Some(new_archive) = try_extract_macho_fat_archive(&self.sess, &archive)? {
  72                 archive = new_archive
  73             }
  74         }
  75         let archive_ro = match ArchiveRO::open(&archive) {
  76             Ok(ar) => ar,
  77             Err(e) => return Err(io::Error::new(io::ErrorKind::Other, e)),
  78         };
  79         if self.additions.iter().any(|ar| ar.path() == archive) {
  80             return Ok(());
  81         }
  82         self.additions.push(Addition::Archive {
  83             path: archive,
  84             archive: archive_ro,
  85             skip: Box::new(skip),
  86         });
  87         Ok(())
  88     }
  89
  90     /// Adds an arbitrary file to this archive
  91     fn add_file(&mut self, file: &Path) {
  92         let name = file.file_name().unwrap().to_str().unwrap();
  93         self.additions
  94             .push(Addition::File { path: file.to_path_buf(), name_in_archive: name.to_owned() });
  95     }
  96
  97     /// Combine the provided files, rlibs, and native libraries into a single
  98     /// `Archive`.
  99     fn build(mut self: Box<Self>, output: &Path) -> bool {
 100         match self.build_with_llvm(output) {
 101             Ok(any_members) => any_members,
 102             Err(e) => self.sess.emit_fatal(ArchiveBuildFailure { error: e }),
 103         }
 104     }
 105 }
 106
 107 pub struct LlvmArchiveBuilderBuilder;
 108
 109 impl ArchiveBuilderBuilder for LlvmArchiveBuilderBuilder {
 110     fn new_archive_builder<'a>(&self, sess: &'a Session) -> Box<dyn ArchiveBuilder<'a> + 'a> {
 111         // FIXME use ArArchiveBuilder on most targets again once reading thin archives is
 112         // implemented
 113         if true || sess.target.arch == "wasm32" || sess.target.arch == "wasm64" {
 114             Box::new(LlvmArchiveBuilder { sess, additions: Vec::new() })
 115         } else {
 116             Box::new(ArArchiveBuilder::new(sess, get_llvm_object_symbols))
 117         }
 118     }
 119
 120     fn create_dll_import_lib(
 121         &self,
 122         sess: &Session,
 123         lib_name: &str,
 124         dll_imports: &[DllImport],
 125         tmpdir: &Path,
 126         is_direct_dependency: bool,
 127     ) -> PathBuf {
 128         let name_suffix = if is_direct_dependency { "_imports" } else { "_imports_indirect" };
 129         let output_path = {
 130             let mut output_path: PathBuf = tmpdir.to_path_buf();
 131             output_path.push(format!("{}{}", lib_name, name_suffix));
 132             output_path.with_extension("lib")
 133         };
 134
 135         let target = &sess.target;
 136         let mingw_gnu_toolchain = common::is_mingw_gnu_toolchain(target);
 137
 138         let import_name_and_ordinal_vector: Vec<(String, Option<u16>)> = dll_imports
 139             .iter()
 140             .map(|import: &DllImport| {
 141                 if sess.target.arch == "x86" {
 142                     (
 143                         common::i686_decorated_name(import, mingw_gnu_toolchain, false),
 144                         import.ordinal(),
 145                     )
 146                 } else {
 147                     (import.name.to_string(), import.ordinal())
 148                 }
 149             })
 150             .collect();
 151
 152         if mingw_gnu_toolchain {
 153             // The binutils linker used on -windows-gnu targets cannot read the import
 154             // libraries generated by LLVM: in our attempts, the linker produced an .EXE
 155             // that loaded but crashed with an AV upon calling one of the imported
 156             // functions. Therefore, use binutils to create the import library instead,
 157             // by writing a .DEF file to the temp dir and calling binutils's dlltool.
 158             let def_file_path =
 159                 tmpdir.join(format!("{}{}", lib_name, name_suffix)).with_extension("def");
 160
 161             let def_file_content = format!(
 162                 "EXPORTS\n{}",
 163                 import_name_and_ordinal_vector
 164                     .into_iter()
 165                     .map(|(name, ordinal)| {
 166                         match ordinal {
 167                             Some(n) => format!("{} @{} NONAME", name, n),
 168                             None => name,
 169                         }
 170                     })
 171                     .collect::<Vec<String>>()
 172                     .join("\n")
 173             );
 174
 175             match std::fs::write(&def_file_path, def_file_content) {
 176                 Ok(_) => {}
 177                 Err(e) => {
 178                     sess.emit_fatal(ErrorWritingDEFFile { error: e });
 179                 }
 180             };
 181
 182             // --no-leading-underscore: For the `import_name_type` feature to work, we need to be
 183             // able to control the *exact* spelling of each of the symbols that are being imported:
 184             // hence we don't want `dlltool` adding leading underscores automatically.
 185             let dlltool = find_binutils_dlltool(sess);
 186             let temp_prefix = {
 187                 let mut path = PathBuf::from(&output_path);
 188                 path.pop();
 189                 path.push(lib_name);
 190                 path
 191             };
 192             let result = std::process::Command::new(dlltool)
 193                 .args([
 194                     "-d",
 195                     def_file_path.to_str().unwrap(),
 196                     "-D",
 197                     lib_name,
 198                     "-l",
 199                     output_path.to_str().unwrap(),
 200                     "--no-leading-underscore",
 201                     "--temp-prefix",
 202                     temp_prefix.to_str().unwrap(),
 203                 ])
 204                 .output();
 205
 206             match result {
 207                 Err(e) => {
 208                     sess.emit_fatal(ErrorCallingDllTool { error: e });
 209                 }
 210                 Ok(output) if !output.status.success() => {
 211                     sess.emit_fatal(DlltoolFailImportLibrary {
 212                         stdout: String::from_utf8_lossy(&output.stdout),
 213                         stderr: String::from_utf8_lossy(&output.stderr),
 214                     })
 215                 }
 216                 _ => {}
 217             }
 218         } else {
 219             // we've checked for \0 characters in the library name already
 220             let dll_name_z = CString::new(lib_name).unwrap();
 221
 222             let output_path_z = rustc_fs_util::path_to_c_string(&output_path);
 223
 224             trace!("invoking LLVMRustWriteImportLibrary");
 225             trace!("  dll_name {:#?}", dll_name_z);
 226             trace!("  output_path {}", output_path.display());
 227             trace!(
 228                 "  import names: {}",
 229                 dll_imports
 230                     .iter()
 231                     .map(|import| import.name.to_string())
 232                     .collect::<Vec<_>>()
 233                     .join(", "),
 234             );
 235
 236             // All import names are Rust identifiers and therefore cannot contain \0 characters.
 237             // FIXME: when support for #[link_name] is implemented, ensure that the import names
 238             // still don't contain any \0 characters. Also need to check that the names don't
 239             // contain substrings like " @" or "NONAME" that are keywords or otherwise reserved
 240             // in definition files.
 241             let cstring_import_name_and_ordinal_vector: Vec<(CString, Option<u16>)> =
 242                 import_name_and_ordinal_vector
 243                     .into_iter()
 244                     .map(|(name, ordinal)| (CString::new(name).unwrap(), ordinal))
 245                     .collect();
 246
 247             let ffi_exports: Vec<LLVMRustCOFFShortExport> = cstring_import_name_and_ordinal_vector
 248                 .iter()
 249                 .map(|(name_z, ordinal)| LLVMRustCOFFShortExport::new(name_z.as_ptr(), *ordinal))
 250                 .collect();
 251             let result = unsafe {
 252                 crate::llvm::LLVMRustWriteImportLibrary(
 253                     dll_name_z.as_ptr(),
 254                     output_path_z.as_ptr(),
 255                     ffi_exports.as_ptr(),
 256                     ffi_exports.len(),
 257                     llvm_machine_type(&sess.target.arch) as u16,
 258                     !sess.target.is_like_msvc,
 259                 )
 260             };
 261
 262             if result == crate::llvm::LLVMRustResult::Failure {
 263                 sess.emit_fatal(ErrorCreatingImportLibrary {
 264                     lib_name,
 265                     error: llvm::last_error().unwrap_or("unknown LLVM error".to_string()),
 266                 });
 267             }
 268         };
 269
 270         output_path
 271     }
 272 }
 273
 274 // The object crate doesn't know how to get symbols for LLVM bitcode and COFF bigobj files.
 275 // As such we need to use LLVM for them.
 276 #[deny(unsafe_op_in_unsafe_fn)]
 277 fn get_llvm_object_symbols(
 278     buf: &[u8],
 279     f: &mut dyn FnMut(&[u8]) -> io::Result<()>,
 280 ) -> io::Result<bool> {
 281     let is_bitcode = unsafe { llvm::LLVMRustIsBitcode(buf.as_ptr(), buf.len()) };
 282
 283     // COFF bigobj file, msvc LTO file or import library. See
 284     // https://github.com/llvm/llvm-project/blob/453f27bc9/llvm/lib/BinaryFormat/Magic.cpp#L38-L51
 285     let is_unsupported_windows_obj_file = buf.get(0..4) == Some(b"\0\0\xFF\xFF");
 286
 287     if is_bitcode || is_unsupported_windows_obj_file {
 288         let mut state = Box::new(f);
 289
 290         let err = unsafe {
 291             llvm::LLVMRustGetSymbols(
 292                 buf.as_ptr(),
 293                 buf.len(),
 294                 &mut *state as *mut &mut _ as *mut c_void,
 295                 callback,
 296                 error_callback,
 297             )
 298         };
 299
 300         if err.is_null() {
 301             return Ok(true);
 302         } else {
 303             return Err(unsafe { *Box::from_raw(err as *mut io::Error) });
 304         }
 305
 306         unsafe extern "C" fn callback(
 307             state: *mut c_void,
 308             symbol_name: *const c_char,
 309         ) -> *mut c_void {
 310             let f = unsafe { &mut *(state as *mut &mut dyn FnMut(&[u8]) -> io::Result<()>) };
 311             match f(unsafe { CStr::from_ptr(symbol_name) }.to_bytes()) {
 312                 Ok(()) => std::ptr::null_mut(),
 313                 Err(err) => Box::into_raw(Box::new(err)) as *mut c_void,
 314             }
 315         }
 316
 317         unsafe extern "C" fn error_callback(error: *const c_char) -> *mut c_void {
 318             let error = unsafe { CStr::from_ptr(error) };
 319             Box::into_raw(Box::new(io::Error::new(
 320                 io::ErrorKind::Other,
 321                 format!("LLVM error: {}", error.to_string_lossy()),
 322             ))) as *mut c_void
 323         }
 324     } else {
 325         get_native_object_symbols(buf, f)
 326     }
 327 }
 328
 329 impl<'a> LlvmArchiveBuilder<'a> {
 330     fn build_with_llvm(&mut self, output: &Path) -> io::Result<bool> {
 331         let kind = &*self.sess.target.archive_format;
 332         let kind = kind
 333             .parse::<ArchiveKind>()
 334             .map_err(|_| kind)
 335             .unwrap_or_else(|kind| self.sess.emit_fatal(UnknownArchiveKind { kind }));
 336
 337         let mut additions = mem::take(&mut self.additions);
 338         let mut strings = Vec::new();
 339         let mut members = Vec::new();
 340
 341         let dst = CString::new(output.to_str().unwrap())?;
 342
 343         unsafe {
 344             for addition in &mut additions {
 345                 match addition {
 346                     Addition::File { path, name_in_archive } => {
 347                         let path = CString::new(path.to_str().unwrap())?;
 348                         let name = CString::new(name_in_archive.clone())?;
 349                         members.push(llvm::LLVMRustArchiveMemberNew(
 350                             path.as_ptr(),
 351                             name.as_ptr(),
 352                             None,
 353                         ));
 354                         strings.push(path);
 355                         strings.push(name);
 356                     }
 357                     Addition::Archive { archive, skip, .. } => {
 358                         for child in archive.iter() {
 359                             let child = child.map_err(string_to_io_error)?;
 360                             if !is_relevant_child(&child) {
 361                                 continue;
 362                             }
 363                             let child_name = child.name().unwrap();
 364                             if skip(child_name) {
 365                                 continue;
 366                             }
 367
 368                             // It appears that LLVM's archive writer is a little
 369                             // buggy if the name we pass down isn't just the
 370                             // filename component, so chop that off here and
 371                             // pass it in.
 372                             //
 373                             // See LLVM bug 25877 for more info.
 374                             let child_name =
 375                                 Path::new(child_name).file_name().unwrap().to_str().unwrap();
 376                             let name = CString::new(child_name)?;
 377                             let m = llvm::LLVMRustArchiveMemberNew(
 378                                 ptr::null(),
 379                                 name.as_ptr(),
 380                                 Some(child.raw),
 381                             );
 382                             members.push(m);
 383                             strings.push(name);
 384                         }
 385                     }
 386                 }
 387             }
 388
 389             let r = llvm::LLVMRustWriteArchive(
 390                 dst.as_ptr(),
 391                 members.len() as libc::size_t,
 392                 members.as_ptr() as *const &_,
 393                 true,
 394                 kind,
 395             );
 396             let ret = if r.into_result().is_err() {
 397                 let err = llvm::LLVMRustGetLastError();
 398                 let msg = if err.is_null() {
 399                     "failed to write archive".into()
 400                 } else {
 401                     String::from_utf8_lossy(CStr::from_ptr(err).to_bytes())
 402                 };
 403                 Err(io::Error::new(io::ErrorKind::Other, msg))
 404             } else {
 405                 Ok(!members.is_empty())
 406             };
 407             for member in members {
 408                 llvm::LLVMRustArchiveMemberFree(member);
 409             }
 410             ret
 411         }
 412     }
 413 }
 414
 415 fn string_to_io_error(s: String) -> io::Error {
 416     io::Error::new(io::ErrorKind::Other, format!("bad archive: {}", s))
 417 }
 418
 419 fn find_binutils_dlltool(sess: &Session) -> OsString {
 420     assert!(sess.target.options.is_like_windows && !sess.target.options.is_like_msvc);
 421     if let Some(dlltool_path) = &sess.opts.unstable_opts.dlltool {
 422         return dlltool_path.clone().into_os_string();
 423     }
 424
 425     let mut tool_name: OsString = if sess.host.arch != sess.target.arch {
 426         // We are cross-compiling, so we need the tool with the prefix matching our target
 427         if sess.target.arch == "x86" {
 428             "i686-w64-mingw32-dlltool"
 429         } else {
 430             "x86_64-w64-mingw32-dlltool"
 431         }
 432     } else {
 433         // We are not cross-compiling, so we just want `dlltool`
 434         "dlltool"
 435     }
 436     .into();
 437
 438     if sess.host.options.is_like_windows {
 439         // If we're compiling on Windows, add the .exe suffix
 440         tool_name.push(".exe");
 441     }
 442
 443     // NOTE: it's not clear how useful it is to explicitly search PATH.
 444     for dir in env::split_paths(&env::var_os("PATH").unwrap_or_default()) {
 445         let full_path = dir.join(&tool_name);
 446         if full_path.is_file() {
 447             return full_path.into_os_string();
 448         }
 449     }
 450
 451     // The user didn't specify the location of the dlltool binary, and we weren't able
 452     // to find the appropriate one on the PATH. Just return the name of the tool
 453     // and let the invocation fail with a hopefully useful error message.
 454     tool_name
 455 }