src/librustc_symbol_mangling/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 log::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         assert!(!substs.needs_subst());
 120         substs.hash_stable(&mut hcx, &mut hasher);
 121
 122         if let Some(instantiating_crate) = instantiating_crate {
 123             tcx.original_crate_name(instantiating_crate)
 124                 .as_str()
 125                 .hash_stable(&mut hcx, &mut hasher);
 126             tcx.crate_disambiguator(instantiating_crate).hash_stable(&mut 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(&mut hcx, &mut hasher);
 133     });
 134
 135     // 64 bits should be enough to avoid collisions.
 136     hasher.finish::<u64>()
 137 }
 138
 139 // Follow C++ namespace-mangling style, see
 140 // http://en.wikipedia.org/wiki/Name_mangling for more info.
 141 //
 142 // It turns out that on macOS you can actually have arbitrary symbols in
 143 // function names (at least when given to LLVM), but this is not possible
 144 // when using unix's linker. Perhaps one day when we just use a linker from LLVM
 145 // we won't need to do this name mangling. The problem with name mangling is
 146 // that it seriously limits the available characters. For example we can't
 147 // have things like &T in symbol names when one would theoretically
 148 // want them for things like impls of traits on that type.
 149 //
 150 // To be able to work on all platforms and get *some* reasonable output, we
 151 // use C++ name-mangling.
 152 #[derive(Debug)]
 153 struct SymbolPath {
 154     result: String,
 155     temp_buf: String,
 156 }
 157
 158 impl SymbolPath {
 159     fn new() -> Self {
 160         let mut result =
 161             SymbolPath { result: String::with_capacity(64), temp_buf: String::with_capacity(16) };
 162         result.result.push_str("_ZN"); // _Z == Begin name-sequence, N == nested
 163         result
 164     }
 165
 166     fn finalize_pending_component(&mut self) {
 167         if !self.temp_buf.is_empty() {
 168             let _ = write!(self.result, "{}{}", self.temp_buf.len(), self.temp_buf);
 169             self.temp_buf.clear();
 170         }
 171     }
 172
 173     fn finish(mut self, hash: u64) -> String {
 174         self.finalize_pending_component();
 175         // E = end name-sequence
 176         let _ = write!(self.result, "17h{:016x}E", hash);
 177         self.result
 178     }
 179 }
 180
 181 struct SymbolPrinter<'tcx> {
 182     tcx: TyCtxt<'tcx>,
 183     path: SymbolPath,
 184
 185     // When `true`, `finalize_pending_component` isn't used.
 186     // This is needed when recursing into `path_qualified`,
 187     // or `path_generic_args`, as any nested paths are
 188     // logically within one component.
 189     keep_within_component: bool,
 190 }
 191
 192 // HACK(eddyb) this relies on using the `fmt` interface to get
 193 // `PrettyPrinter` aka pretty printing of e.g. types in paths,
 194 // symbol names should have their own printing machinery.
 195
 196 impl Printer<'tcx> for SymbolPrinter<'tcx> {
 197     type Error = fmt::Error;
 198
 199     type Path = Self;
 200     type Region = Self;
 201     type Type = Self;
 202     type DynExistential = Self;
 203     type Const = Self;
 204
 205     fn tcx(&self) -> TyCtxt<'tcx> {
 206         self.tcx
 207     }
 208
 209     fn print_region(self, _region: ty::Region<'_>) -> Result<Self::Region, Self::Error> {
 210         Ok(self)
 211     }
 212
 213     fn print_type(self, ty: Ty<'tcx>) -> Result<Self::Type, Self::Error> {
 214         match ty.kind {
 215             // Print all nominal types as paths (unlike `pretty_print_type`).
 216             ty::FnDef(def_id, substs)
 217             | ty::Opaque(def_id, substs)
 218             | ty::Projection(ty::ProjectionTy { item_def_id: def_id, substs })
 219             | ty::UnnormalizedProjection(ty::ProjectionTy { item_def_id: def_id, substs })
 220             | ty::Closure(def_id, substs)
 221             | ty::Generator(def_id, substs, _) => self.print_def_path(def_id, substs),
 222             _ => self.pretty_print_type(ty),
 223         }
 224     }
 225
 226     fn print_dyn_existential(
 227         mut self,
 228         predicates: &'tcx ty::List<ty::ExistentialPredicate<'tcx>>,
 229     ) -> Result<Self::DynExistential, Self::Error> {
 230         let mut first = true;
 231         for p in predicates {
 232             if !first {
 233                 write!(self, "+")?;
 234             }
 235             first = false;
 236             self = p.print(self)?;
 237         }
 238         Ok(self)
 239     }
 240
 241     fn print_const(mut self, ct: &'tcx ty::Const<'tcx>) -> Result<Self::Const, Self::Error> {
 242         // only print integers
 243         if let ty::ConstKind::Value(ConstValue::Scalar(Scalar::Raw { .. })) = ct.val {
 244             if ct.ty.is_integral() {
 245                 return self.pretty_print_const(ct, true);
 246             }
 247         }
 248         self.write_str("_")?;
 249         Ok(self)
 250     }
 251
 252     fn path_crate(mut self, cnum: CrateNum) -> Result<Self::Path, Self::Error> {
 253         self.write_str(&self.tcx.original_crate_name(cnum).as_str())?;
 254         Ok(self)
 255     }
 256     fn path_qualified(
 257         self,
 258         self_ty: Ty<'tcx>,
 259         trait_ref: Option<ty::TraitRef<'tcx>>,
 260     ) -> Result<Self::Path, Self::Error> {
 261         // Similar to `pretty_path_qualified`, but for the other
 262         // types that are printed as paths (see `print_type` above).
 263         match self_ty.kind {
 264             ty::FnDef(..)
 265             | ty::Opaque(..)
 266             | ty::Projection(_)
 267             | ty::UnnormalizedProjection(_)
 268             | ty::Closure(..)
 269             | ty::Generator(..)
 270                 if trait_ref.is_none() =>
 271             {
 272                 self.print_type(self_ty)
 273             }
 274
 275             _ => self.pretty_path_qualified(self_ty, trait_ref),
 276         }
 277     }
 278
 279     fn path_append_impl(
 280         self,
 281         print_prefix: impl FnOnce(Self) -> Result<Self::Path, Self::Error>,
 282         _disambiguated_data: &DisambiguatedDefPathData,
 283         self_ty: Ty<'tcx>,
 284         trait_ref: Option<ty::TraitRef<'tcx>>,
 285     ) -> Result<Self::Path, Self::Error> {
 286         self.pretty_path_append_impl(
 287             |mut cx| {
 288                 cx = print_prefix(cx)?;
 289
 290                 if cx.keep_within_component {
 291                     // HACK(eddyb) print the path similarly to how `FmtPrinter` prints it.
 292                     cx.write_str("::")?;
 293                 } else {
 294                     cx.path.finalize_pending_component();
 295                 }
 296
 297                 Ok(cx)
 298             },
 299             self_ty,
 300             trait_ref,
 301         )
 302     }
 303     fn path_append(
 304         mut self,
 305         print_prefix: impl FnOnce(Self) -> Result<Self::Path, Self::Error>,
 306         disambiguated_data: &DisambiguatedDefPathData,
 307     ) -> Result<Self::Path, Self::Error> {
 308         self = print_prefix(self)?;
 309
 310         // Skip `::{{constructor}}` on tuple/unit structs.
 311         if let DefPathData::Ctor = disambiguated_data.data {
 312             return Ok(self);
 313         }
 314
 315         if self.keep_within_component {
 316             // HACK(eddyb) print the path similarly to how `FmtPrinter` prints it.
 317             self.write_str("::")?;
 318         } else {
 319             self.path.finalize_pending_component();
 320         }
 321
 322         self.write_str(&disambiguated_data.data.as_symbol().as_str())?;
 323         Ok(self)
 324     }
 325     fn path_generic_args(
 326         mut self,
 327         print_prefix: impl FnOnce(Self) -> Result<Self::Path, Self::Error>,
 328         args: &[GenericArg<'tcx>],
 329     ) -> Result<Self::Path, Self::Error> {
 330         self = print_prefix(self)?;
 331
 332         let args = args.iter().cloned().filter(|arg| match arg.unpack() {
 333             GenericArgKind::Lifetime(_) => false,
 334             _ => true,
 335         });
 336
 337         if args.clone().next().is_some() {
 338             self.generic_delimiters(|cx| cx.comma_sep(args))
 339         } else {
 340             Ok(self)
 341         }
 342     }
 343 }
 344
 345 impl PrettyPrinter<'tcx> for SymbolPrinter<'tcx> {
 346     fn region_should_not_be_omitted(&self, _region: ty::Region<'_>) -> bool {
 347         false
 348     }
 349     fn comma_sep<T>(mut self, mut elems: impl Iterator<Item = T>) -> Result<Self, Self::Error>
 350     where
 351         T: Print<'tcx, Self, Output = Self, Error = Self::Error>,
 352     {
 353         if let Some(first) = elems.next() {
 354             self = first.print(self)?;
 355             for elem in elems {
 356                 self.write_str(",")?;
 357                 self = elem.print(self)?;
 358             }
 359         }
 360         Ok(self)
 361     }
 362
 363     fn generic_delimiters(
 364         mut self,
 365         f: impl FnOnce(Self) -> Result<Self, Self::Error>,
 366     ) -> Result<Self, Self::Error> {
 367         write!(self, "<")?;
 368
 369         let kept_within_component = mem::replace(&mut self.keep_within_component, true);
 370         self = f(self)?;
 371         self.keep_within_component = kept_within_component;
 372
 373         write!(self, ">")?;
 374
 375         Ok(self)
 376     }
 377 }
 378
 379 impl fmt::Write for SymbolPrinter<'_> {
 380     fn write_str(&mut self, s: &str) -> fmt::Result {
 381         // Name sanitation. LLVM will happily accept identifiers with weird names, but
 382         // gas doesn't!
 383         // gas accepts the following characters in symbols: a-z, A-Z, 0-9, ., _, $
 384         // NVPTX assembly has more strict naming rules than gas, so additionally, dots
 385         // are replaced with '$' there.
 386
 387         for c in s.chars() {
 388             if self.path.temp_buf.is_empty() {
 389                 match c {
 390                     'a'..='z' | 'A'..='Z' | '_' => {}
 391                     _ => {
 392                         // Underscore-qualify anything that didn't start as an ident.
 393                         self.path.temp_buf.push('_');
 394                     }
 395                 }
 396             }
 397             match c {
 398                 // Escape these with $ sequences
 399                 '@' => self.path.temp_buf.push_str("$SP$"),
 400                 '*' => self.path.temp_buf.push_str("$BP$"),
 401                 '&' => self.path.temp_buf.push_str("$RF$"),
 402                 '<' => self.path.temp_buf.push_str("$LT$"),
 403                 '>' => self.path.temp_buf.push_str("$GT$"),
 404                 '(' => self.path.temp_buf.push_str("$LP$"),
 405                 ')' => self.path.temp_buf.push_str("$RP$"),
 406                 ',' => self.path.temp_buf.push_str("$C$"),
 407
 408                 '-' | ':' | '.' if self.tcx.has_strict_asm_symbol_naming() => {
 409                     // NVPTX doesn't support these characters in symbol names.
 410                     self.path.temp_buf.push('$')
 411                 }
 412
 413                 // '.' doesn't occur in types and functions, so reuse it
 414                 // for ':' and '-'
 415                 '-' | ':' => self.path.temp_buf.push('.'),
 416
 417                 // Avoid crashing LLVM in certain (LTO-related) situations, see #60925.
 418                 'm' if self.path.temp_buf.ends_with(".llv") => self.path.temp_buf.push_str("$u6d$"),
 419
 420                 // These are legal symbols
 421                 'a'..='z' | 'A'..='Z' | '0'..='9' | '_' | '.' | '$' => self.path.temp_buf.push(c),
 422
 423                 _ => {
 424                     self.path.temp_buf.push('$');
 425                     for c in c.escape_unicode().skip(1) {
 426                         match c {
 427                             '{' => {}
 428                             '}' => self.path.temp_buf.push('$'),
 429                             c => self.path.temp_buf.push(c),
 430                         }
 431                     }
 432                 }
 433             }
 434         }
 435
 436         Ok(())
 437     }
 438 }