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