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