compiler/rustc_symbol_mangling/src/legacy.rs

   1 use rustc_data_structures::stable_hasher::{HashStable, StableHasher};
   2 use rustc_hir::def_id::CrateNum;
   3 use rustc_hir::definitions::{DefPathData, DisambiguatedDefPathData};
   4 use rustc_middle::ty::print::{PrettyPrinter, Print, Printer};
   5 use rustc_middle::ty::subst::{GenericArg, GenericArgKind};
   6 use rustc_middle::ty::{self, Instance, Ty, TyCtxt, TypeVisitable};
   7 use rustc_middle::util::common::record_time;
   8
   9 use std::fmt::{self, Write};
  10 use std::mem::{self, discriminant};
  11
  12 pub(super) fn mangle<'tcx>(
  13     tcx: TyCtxt<'tcx>,
  14     instance: Instance<'tcx>,
  15     instantiating_crate: Option<CrateNum>,
  16 ) -> String {
  17     let def_id = instance.def_id();
  18
  19     // We want to compute the "type" of this item. Unfortunately, some
  20     // kinds of items (e.g., closures) don't have an entry in the
  21     // item-type array. So walk back up the find the closest parent
  22     // that DOES have an entry.
  23     let mut ty_def_id = def_id;
  24     let instance_ty;
  25     loop {
  26         let key = tcx.def_key(ty_def_id);
  27         match key.disambiguated_data.data {
  28             DefPathData::TypeNs(_) | DefPathData::ValueNs(_) => {
  29                 instance_ty = tcx.type_of(ty_def_id);
  30                 debug!(?instance_ty);
  31                 break;
  32             }
  33             _ => {
  34                 // if we're making a symbol for something, there ought
  35                 // to be a value or type-def or something in there
  36                 // *somewhere*
  37                 ty_def_id.index = key.parent.unwrap_or_else(|| {
  38                     bug!(
  39                         "finding type for {:?}, encountered def-id {:?} with no \
  40                          parent",
  41                         def_id,
  42                         ty_def_id
  43                     );
  44                 });
  45             }
  46         }
  47     }
  48
  49     // Erase regions because they may not be deterministic when hashed
  50     // and should not matter anyhow.
  51     let instance_ty = tcx.erase_regions(instance_ty);
  52
  53     let hash = get_symbol_hash(tcx, instance, instance_ty, instantiating_crate);
  54
  55     let mut printer = SymbolPrinter { tcx, path: SymbolPath::new(), keep_within_component: false };
  56     printer
  57         .print_def_path(
  58             def_id,
  59             if let ty::InstanceDef::DropGlue(_, _) = instance.def {
  60                 // Add the name of the dropped type to the symbol name
  61                 &*instance.substs
  62             } else {
  63                 &[]
  64             },
  65         )
  66         .unwrap();
  67
  68     if let ty::InstanceDef::VTableShim(..) = instance.def {
  69         let _ = printer.write_str("{{vtable-shim}}");
  70     }
  71
  72     if let ty::InstanceDef::ReifyShim(..) = instance.def {
  73         let _ = printer.write_str("{{reify-shim}}");
  74     }
  75
  76     printer.path.finish(hash)
  77 }
  78
  79 fn get_symbol_hash<'tcx>(
  80     tcx: TyCtxt<'tcx>,
  81
  82     // instance this name will be for
  83     instance: Instance<'tcx>,
  84
  85     // type of the item, without any generic
  86     // parameters substituted; this is
  87     // included in the hash as a kind of
  88     // safeguard.
  89     item_type: Ty<'tcx>,
  90
  91     instantiating_crate: Option<CrateNum>,
  92 ) -> u64 {
  93     let def_id = instance.def_id();
  94     let substs = instance.substs;
  95     debug!("get_symbol_hash(def_id={:?}, parameters={:?})", def_id, substs);
  96
  97     tcx.with_stable_hashing_context(|mut hcx| {
  98         let mut hasher = StableHasher::new();
  99
 100         record_time(&tcx.sess.perf_stats.symbol_hash_time, || {
 101             // the main symbol name is not necessarily unique; hash in the
 102             // compiler's internal def-path, guaranteeing each symbol has a
 103             // truly unique path
 104             tcx.def_path_hash(def_id).hash_stable(&mut hcx, &mut hasher);
 105
 106             // Include the main item-type. Note that, in this case, the
 107             // assertions about `needs_subst` may not hold, but this item-type
 108             // ought to be the same for every reference anyway.
 109             assert!(!item_type.has_erasable_regions());
 110             hcx.while_hashing_spans(false, |hcx| {
 111                 item_type.hash_stable(hcx, &mut hasher);
 112
 113                 // If this is a function, we hash the signature as well.
 114                 // This is not *strictly* needed, but it may help in some
 115                 // situations, see the `run-make/a-b-a-linker-guard` test.
 116                 if let ty::FnDef(..) = item_type.kind() {
 117                     item_type.fn_sig(tcx).hash_stable(hcx, &mut hasher);
 118                 }
 119
 120                 // also include any type parameters (for generic items)
 121                 substs.hash_stable(hcx, &mut hasher);
 122
 123                 if let Some(instantiating_crate) = instantiating_crate {
 124                     tcx.def_path_hash(instantiating_crate.as_def_id())
 125                         .stable_crate_id()
 126                         .hash_stable(hcx, &mut hasher);
 127                 }
 128
 129                 // We want to avoid accidental collision between different types of instances.
 130                 // Especially, `VTableShim`s and `ReifyShim`s may overlap with their original
 131                 // instances without this.
 132                 discriminant(&instance.def).hash_stable(hcx, &mut hasher);
 133             });
 134         });
 135
 136         // 64 bits should be enough to avoid collisions.
 137         hasher.finish::<u64>()
 138     })
 139 }
 140
 141 // Follow C++ namespace-mangling style, see
 142 // https://en.wikipedia.org/wiki/Name_mangling for more info.
 143 //
 144 // It turns out that on macOS you can actually have arbitrary symbols in
 145 // function names (at least when given to LLVM), but this is not possible
 146 // when using unix's linker. Perhaps one day when we just use a linker from LLVM
 147 // we won't need to do this name mangling. The problem with name mangling is
 148 // that it seriously limits the available characters. For example we can't
 149 // have things like &T in symbol names when one would theoretically
 150 // want them for things like impls of traits on that type.
 151 //
 152 // To be able to work on all platforms and get *some* reasonable output, we
 153 // use C++ name-mangling.
 154 #[derive(Debug)]
 155 struct SymbolPath {
 156     result: String,
 157     temp_buf: String,
 158 }
 159
 160 impl SymbolPath {
 161     fn new() -> Self {
 162         let mut result =
 163             SymbolPath { result: String::with_capacity(64), temp_buf: String::with_capacity(16) };
 164         result.result.push_str("_ZN"); // _Z == Begin name-sequence, N == nested
 165         result
 166     }
 167
 168     fn finalize_pending_component(&mut self) {
 169         if !self.temp_buf.is_empty() {
 170             let _ = write!(self.result, "{}{}", self.temp_buf.len(), self.temp_buf);
 171             self.temp_buf.clear();
 172         }
 173     }
 174
 175     fn finish(mut self, hash: u64) -> String {
 176         self.finalize_pending_component();
 177         // E = end name-sequence
 178         let _ = write!(self.result, "17h{:016x}E", hash);
 179         self.result
 180     }
 181 }
 182
 183 struct SymbolPrinter<'tcx> {
 184     tcx: TyCtxt<'tcx>,
 185     path: SymbolPath,
 186
 187     // When `true`, `finalize_pending_component` isn't used.
 188     // This is needed when recursing into `path_qualified`,
 189     // or `path_generic_args`, as any nested paths are
 190     // logically within one component.
 191     keep_within_component: bool,
 192 }
 193
 194 // HACK(eddyb) this relies on using the `fmt` interface to get
 195 // `PrettyPrinter` aka pretty printing of e.g. types in paths,
 196 // symbol names should have their own printing machinery.
 197
 198 impl<'tcx> Printer<'tcx> for &mut SymbolPrinter<'tcx> {
 199     type Error = fmt::Error;
 200
 201     type Path = Self;
 202     type Region = Self;
 203     type Type = Self;
 204     type DynExistential = Self;
 205     type Const = Self;
 206
 207     fn tcx(&self) -> TyCtxt<'tcx> {
 208         self.tcx
 209     }
 210
 211     fn print_region(self, _region: ty::Region<'_>) -> Result<Self::Region, Self::Error> {
 212         Ok(self)
 213     }
 214
 215     fn print_type(mut self, ty: Ty<'tcx>) -> Result<Self::Type, Self::Error> {
 216         match *ty.kind() {
 217             // Print all nominal types as paths (unlike `pretty_print_type`).
 218             ty::FnDef(def_id, substs)
 219             | ty::Opaque(def_id, substs)
 220             | ty::Projection(ty::ProjectionTy { item_def_id: def_id, substs })
 221             | ty::Closure(def_id, substs)
 222             | ty::Generator(def_id, substs, _) => self.print_def_path(def_id, substs),
 223
 224             // The `pretty_print_type` formatting of array size depends on
 225             // -Zverbose flag, so we cannot reuse it here.
 226             ty::Array(ty, size) => {
 227                 self.write_str("[")?;
 228                 self = self.print_type(ty)?;
 229                 self.write_str("; ")?;
 230                 if let Some(size) = size.kind().try_to_bits(self.tcx().data_layout.pointer_size) {
 231                     write!(self, "{}", size)?
 232                 } else if let ty::ConstKind::Param(param) = size.kind() {
 233                     self = param.print(self)?
 234                 } else {
 235                     self.write_str("_")?
 236                 }
 237                 self.write_str("]")?;
 238                 Ok(self)
 239             }
 240
 241             _ => self.pretty_print_type(ty),
 242         }
 243     }
 244
 245     fn print_dyn_existential(
 246         mut self,
 247         predicates: &'tcx ty::List<ty::Binder<'tcx, ty::ExistentialPredicate<'tcx>>>,
 248     ) -> Result<Self::DynExistential, Self::Error> {
 249         let mut first = true;
 250         for p in predicates {
 251             if !first {
 252                 write!(self, "+")?;
 253             }
 254             first = false;
 255             self = p.print(self)?;
 256         }
 257         Ok(self)
 258     }
 259
 260     fn print_const(self, ct: ty::Const<'tcx>) -> Result<Self::Const, Self::Error> {
 261         // only print integers
 262         match (ct.kind(), ct.ty().kind()) {
 263             (ty::ConstKind::Value(ty::ValTree::Leaf(scalar)), ty::Int(_) | ty::Uint(_)) => {
 264                 // The `pretty_print_const` formatting depends on -Zverbose
 265                 // flag, so we cannot reuse it here.
 266                 let signed = matches!(ct.ty().kind(), ty::Int(_));
 267                 write!(
 268                     self,
 269                     "{:#?}",
 270                     ty::ConstInt::new(scalar, signed, ct.ty().is_ptr_sized_integral())
 271                 )?;
 272             }
 273             _ => self.write_str("_")?,
 274         }
 275         Ok(self)
 276     }
 277
 278     fn path_crate(self, cnum: CrateNum) -> Result<Self::Path, Self::Error> {
 279         self.write_str(self.tcx.crate_name(cnum).as_str())?;
 280         Ok(self)
 281     }
 282     fn path_qualified(
 283         self,
 284         self_ty: Ty<'tcx>,
 285         trait_ref: Option<ty::TraitRef<'tcx>>,
 286     ) -> Result<Self::Path, Self::Error> {
 287         // Similar to `pretty_path_qualified`, but for the other
 288         // types that are printed as paths (see `print_type` above).
 289         match self_ty.kind() {
 290             ty::FnDef(..)
 291             | ty::Opaque(..)
 292             | ty::Projection(_)
 293             | ty::Closure(..)
 294             | ty::Generator(..)
 295                 if trait_ref.is_none() =>
 296             {
 297                 self.print_type(self_ty)
 298             }
 299
 300             _ => self.pretty_path_qualified(self_ty, trait_ref),
 301         }
 302     }
 303
 304     fn path_append_impl(
 305         self,
 306         print_prefix: impl FnOnce(Self) -> Result<Self::Path, Self::Error>,
 307         _disambiguated_data: &DisambiguatedDefPathData,
 308         self_ty: Ty<'tcx>,
 309         trait_ref: Option<ty::TraitRef<'tcx>>,
 310     ) -> Result<Self::Path, Self::Error> {
 311         self.pretty_path_append_impl(
 312             |mut cx| {
 313                 cx = print_prefix(cx)?;
 314
 315                 if cx.keep_within_component {
 316                     // HACK(eddyb) print the path similarly to how `FmtPrinter` prints it.
 317                     cx.write_str("::")?;
 318                 } else {
 319                     cx.path.finalize_pending_component();
 320                 }
 321
 322                 Ok(cx)
 323             },
 324             self_ty,
 325             trait_ref,
 326         )
 327     }
 328     fn path_append(
 329         mut self,
 330         print_prefix: impl FnOnce(Self) -> Result<Self::Path, Self::Error>,
 331         disambiguated_data: &DisambiguatedDefPathData,
 332     ) -> Result<Self::Path, Self::Error> {
 333         self = print_prefix(self)?;
 334
 335         // Skip `::{{extern}}` blocks and `::{{constructor}}` on tuple/unit structs.
 336         if let DefPathData::ForeignMod | DefPathData::Ctor = disambiguated_data.data {
 337             return Ok(self);
 338         }
 339
 340         if self.keep_within_component {
 341             // HACK(eddyb) print the path similarly to how `FmtPrinter` prints it.
 342             self.write_str("::")?;
 343         } else {
 344             self.path.finalize_pending_component();
 345         }
 346
 347         write!(self, "{}", disambiguated_data.data)?;
 348
 349         Ok(self)
 350     }
 351     fn path_generic_args(
 352         mut self,
 353         print_prefix: impl FnOnce(Self) -> Result<Self::Path, Self::Error>,
 354         args: &[GenericArg<'tcx>],
 355     ) -> Result<Self::Path, Self::Error> {
 356         self = print_prefix(self)?;
 357
 358         let args =
 359             args.iter().cloned().filter(|arg| !matches!(arg.unpack(), GenericArgKind::Lifetime(_)));
 360
 361         if args.clone().next().is_some() {
 362             self.generic_delimiters(|cx| cx.comma_sep(args))
 363         } else {
 364             Ok(self)
 365         }
 366     }
 367 }
 368
 369 impl<'tcx> PrettyPrinter<'tcx> for &mut SymbolPrinter<'tcx> {
 370     fn should_print_region(&self, _region: ty::Region<'_>) -> bool {
 371         false
 372     }
 373     fn comma_sep<T>(mut self, mut elems: impl Iterator<Item = T>) -> Result<Self, Self::Error>
 374     where
 375         T: Print<'tcx, Self, Output = Self, Error = Self::Error>,
 376     {
 377         if let Some(first) = elems.next() {
 378             self = first.print(self)?;
 379             for elem in elems {
 380                 self.write_str(",")?;
 381                 self = elem.print(self)?;
 382             }
 383         }
 384         Ok(self)
 385     }
 386
 387     fn generic_delimiters(
 388         mut self,
 389         f: impl FnOnce(Self) -> Result<Self, Self::Error>,
 390     ) -> Result<Self, Self::Error> {
 391         write!(self, "<")?;
 392
 393         let kept_within_component = mem::replace(&mut self.keep_within_component, true);
 394         self = f(self)?;
 395         self.keep_within_component = kept_within_component;
 396
 397         write!(self, ">")?;
 398
 399         Ok(self)
 400     }
 401 }
 402
 403 impl fmt::Write for SymbolPrinter<'_> {
 404     fn write_str(&mut self, s: &str) -> fmt::Result {
 405         // Name sanitation. LLVM will happily accept identifiers with weird names, but
 406         // gas doesn't!
 407         // gas accepts the following characters in symbols: a-z, A-Z, 0-9, ., _, $
 408         // NVPTX assembly has more strict naming rules than gas, so additionally, dots
 409         // are replaced with '$' there.
 410
 411         for c in s.chars() {
 412             if self.path.temp_buf.is_empty() {
 413                 match c {
 414                     'a'..='z' | 'A'..='Z' | '_' => {}
 415                     _ => {
 416                         // Underscore-qualify anything that didn't start as an ident.
 417                         self.path.temp_buf.push('_');
 418                     }
 419                 }
 420             }
 421             match c {
 422                 // Escape these with $ sequences
 423                 '@' => self.path.temp_buf.push_str("$SP$"),
 424                 '*' => self.path.temp_buf.push_str("$BP$"),
 425                 '&' => self.path.temp_buf.push_str("$RF$"),
 426                 '<' => self.path.temp_buf.push_str("$LT$"),
 427                 '>' => self.path.temp_buf.push_str("$GT$"),
 428                 '(' => self.path.temp_buf.push_str("$LP$"),
 429                 ')' => self.path.temp_buf.push_str("$RP$"),
 430                 ',' => self.path.temp_buf.push_str("$C$"),
 431
 432                 '-' | ':' | '.' if self.tcx.has_strict_asm_symbol_naming() => {
 433                     // NVPTX doesn't support these characters in symbol names.
 434                     self.path.temp_buf.push('$')
 435                 }
 436
 437                 // '.' doesn't occur in types and functions, so reuse it
 438                 // for ':' and '-'
 439                 '-' | ':' => self.path.temp_buf.push('.'),
 440
 441                 // Avoid crashing LLVM in certain (LTO-related) situations, see #60925.
 442                 'm' if self.path.temp_buf.ends_with(".llv") => self.path.temp_buf.push_str("$u6d$"),
 443
 444                 // These are legal symbols
 445                 'a'..='z' | 'A'..='Z' | '0'..='9' | '_' | '.' | '$' => self.path.temp_buf.push(c),
 446
 447                 _ => {
 448                     self.path.temp_buf.push('$');
 449                     for c in c.escape_unicode().skip(1) {
 450                         match c {
 451                             '{' => {}
 452                             '}' => self.path.temp_buf.push('$'),
 453                             c => self.path.temp_buf.push(c),
 454                         }
 455                     }
 456                 }
 457             }
 458         }
 459
 460         Ok(())
 461     }
 462 }