compiler/rustc_metadata/src/rmeta/mod.rs

   1 use crate::creader::CrateMetadataRef;
   2 use decoder::Metadata;
   3 use def_path_hash_map::DefPathHashMapRef;
   4 use rustc_data_structures::fx::FxHashMap;
   5 use table::TableBuilder;
   6
   7 use rustc_ast as ast;
   8 use rustc_attr as attr;
   9 use rustc_data_structures::svh::Svh;
  10 use rustc_data_structures::sync::MetadataRef;
  11 use rustc_hir as hir;
  12 use rustc_hir::def::{CtorKind, DefKind};
  13 use rustc_hir::def_id::{CrateNum, DefId, DefIndex, DefPathHash, StableCrateId};
  14 use rustc_hir::definitions::DefKey;
  15 use rustc_hir::lang_items::LangItem;
  16 use rustc_index::bit_set::BitSet;
  17 use rustc_index::vec::IndexVec;
  18 use rustc_middle::metadata::ModChild;
  19 use rustc_middle::middle::codegen_fn_attrs::CodegenFnAttrs;
  20 use rustc_middle::middle::exported_symbols::{ExportedSymbol, SymbolExportInfo};
  21 use rustc_middle::middle::resolve_lifetime::ObjectLifetimeDefault;
  22 use rustc_middle::mir;
  23 use rustc_middle::ty::fast_reject::SimplifiedType;
  24 use rustc_middle::ty::query::Providers;
  25 use rustc_middle::ty::{self, ReprOptions, Ty, UnusedGenericParams};
  26 use rustc_middle::ty::{DeducedParamAttrs, GeneratorDiagnosticData, ParameterizedOverTcx, TyCtxt};
  27 use rustc_serialize::opaque::FileEncoder;
  28 use rustc_session::config::SymbolManglingVersion;
  29 use rustc_session::cstore::{CrateDepKind, ForeignModule, LinkagePreference, NativeLib};
  30 use rustc_span::edition::Edition;
  31 use rustc_span::hygiene::{ExpnIndex, MacroKind};
  32 use rustc_span::symbol::{Ident, Symbol};
  33 use rustc_span::{self, ExpnData, ExpnHash, ExpnId, Span};
  34 use rustc_target::spec::{PanicStrategy, TargetTriple};
  35
  36 use std::marker::PhantomData;
  37 use std::num::NonZeroUsize;
  38
  39 pub use decoder::provide_extern;
  40 use decoder::DecodeContext;
  41 pub(crate) use decoder::{CrateMetadata, CrateNumMap, MetadataBlob};
  42 use encoder::EncodeContext;
  43 pub use encoder::{encode_metadata, EncodedMetadata};
  44 use rustc_span::hygiene::SyntaxContextData;
  45
  46 mod decoder;
  47 mod def_path_hash_map;
  48 mod encoder;
  49 mod table;
  50
  51 pub(crate) fn rustc_version() -> String {
  52     format!("rustc {}", option_env!("CFG_VERSION").unwrap_or("unknown version"))
  53 }
  54
  55 /// Metadata encoding version.
  56 /// N.B., increment this if you change the format of metadata such that
  57 /// the rustc version can't be found to compare with `rustc_version()`.
  58 const METADATA_VERSION: u8 = 6;
  59
  60 /// Metadata header which includes `METADATA_VERSION`.
  61 ///
  62 /// This header is followed by the position of the `CrateRoot`,
  63 /// which is encoded as a 32-bit big-endian unsigned integer,
  64 /// and further followed by the rustc version string.
  65 pub const METADATA_HEADER: &[u8] = &[b'r', b'u', b's', b't', 0, 0, 0, METADATA_VERSION];
  66
  67 /// A value of type T referred to by its absolute position
  68 /// in the metadata, and which can be decoded lazily.
  69 ///
  70 /// Metadata is effective a tree, encoded in post-order,
  71 /// and with the root's position written next to the header.
  72 /// That means every single `LazyValue` points to some previous
  73 /// location in the metadata and is part of a larger node.
  74 ///
  75 /// The first `LazyValue` in a node is encoded as the backwards
  76 /// distance from the position where the containing node
  77 /// starts and where the `LazyValue` points to, while the rest
  78 /// use the forward distance from the previous `LazyValue`.
  79 /// Distances start at 1, as 0-byte nodes are invalid.
  80 /// Also invalid are nodes being referred in a different
  81 /// order than they were encoded in.
  82 #[must_use]
  83 struct LazyValue<T> {
  84     position: NonZeroUsize,
  85     _marker: PhantomData<fn() -> T>,
  86 }
  87
  88 impl<T: ParameterizedOverTcx> ParameterizedOverTcx for LazyValue<T> {
  89     type Value<'tcx> = LazyValue<T::Value<'tcx>>;
  90 }
  91
  92 impl<T> LazyValue<T> {
  93     fn from_position(position: NonZeroUsize) -> LazyValue<T> {
  94         LazyValue { position, _marker: PhantomData }
  95     }
  96 }
  97
  98 /// A list of lazily-decoded values.
  99 ///
 100 /// Unlike `LazyValue<Vec<T>>`, the length is encoded next to the
 101 /// position, not at the position, which means that the length
 102 /// doesn't need to be known before encoding all the elements.
 103 ///
 104 /// If the length is 0, no position is encoded, but otherwise,
 105 /// the encoding is that of `LazyArray`, with the distinction that
 106 /// the minimal distance the length of the sequence, i.e.
 107 /// it's assumed there's no 0-byte element in the sequence.
 108 struct LazyArray<T> {
 109     position: NonZeroUsize,
 110     num_elems: usize,
 111     _marker: PhantomData<fn() -> T>,
 112 }
 113
 114 impl<T: ParameterizedOverTcx> ParameterizedOverTcx for LazyArray<T> {
 115     type Value<'tcx> = LazyArray<T::Value<'tcx>>;
 116 }
 117
 118 impl<T> LazyArray<T> {
 119     fn from_position_and_num_elems(position: NonZeroUsize, num_elems: usize) -> LazyArray<T> {
 120         LazyArray { position, num_elems, _marker: PhantomData }
 121     }
 122
 123     fn empty() -> LazyArray<T> {
 124         LazyArray::from_position_and_num_elems(NonZeroUsize::new(1).unwrap(), 0)
 125     }
 126 }
 127
 128 /// A list of lazily-decoded values, with the added capability of random access.
 129 ///
 130 /// Random-access table (i.e. offering constant-time `get`/`set`), similar to
 131 /// `LazyArray<T>`, but without requiring encoding or decoding all the values
 132 /// eagerly and in-order.
 133 struct LazyTable<I, T> {
 134     position: NonZeroUsize,
 135     encoded_size: usize,
 136     _marker: PhantomData<fn(I) -> T>,
 137 }
 138
 139 impl<I: 'static, T: ParameterizedOverTcx> ParameterizedOverTcx for LazyTable<I, T> {
 140     type Value<'tcx> = LazyTable<I, T::Value<'tcx>>;
 141 }
 142
 143 impl<I, T> LazyTable<I, T> {
 144     fn from_position_and_encoded_size(
 145         position: NonZeroUsize,
 146         encoded_size: usize,
 147     ) -> LazyTable<I, T> {
 148         LazyTable { position, encoded_size, _marker: PhantomData }
 149     }
 150 }
 151
 152 impl<T> Copy for LazyValue<T> {}
 153 impl<T> Clone for LazyValue<T> {
 154     fn clone(&self) -> Self {
 155         *self
 156     }
 157 }
 158
 159 impl<T> Copy for LazyArray<T> {}
 160 impl<T> Clone for LazyArray<T> {
 161     fn clone(&self) -> Self {
 162         *self
 163     }
 164 }
 165
 166 impl<I, T> Copy for LazyTable<I, T> {}
 167 impl<I, T> Clone for LazyTable<I, T> {
 168     fn clone(&self) -> Self {
 169         *self
 170     }
 171 }
 172
 173 /// Encoding / decoding state for `Lazy`s (`LazyValue`, `LazyArray`, and `LazyTable`).
 174 #[derive(Copy, Clone, PartialEq, Eq, Debug)]
 175 enum LazyState {
 176     /// Outside of a metadata node.
 177     NoNode,
 178
 179     /// Inside a metadata node, and before any `Lazy`s.
 180     /// The position is that of the node itself.
 181     NodeStart(NonZeroUsize),
 182
 183     /// Inside a metadata node, with a previous `Lazy`s.
 184     /// The position is where that previous `Lazy` would start.
 185     Previous(NonZeroUsize),
 186 }
 187
 188 type SyntaxContextTable = LazyTable<u32, LazyValue<SyntaxContextData>>;
 189 type ExpnDataTable = LazyTable<ExpnIndex, LazyValue<ExpnData>>;
 190 type ExpnHashTable = LazyTable<ExpnIndex, LazyValue<ExpnHash>>;
 191
 192 #[derive(MetadataEncodable, MetadataDecodable)]
 193 pub(crate) struct ProcMacroData {
 194     proc_macro_decls_static: DefIndex,
 195     stability: Option<attr::Stability>,
 196     macros: LazyArray<DefIndex>,
 197 }
 198
 199 /// Serialized metadata for a crate.
 200 /// When compiling a proc-macro crate, we encode many of
 201 /// the `LazyArray<T>` fields as `Lazy::empty()`. This serves two purposes:
 202 ///
 203 /// 1. We avoid performing unnecessary work. Proc-macro crates can only
 204 /// export proc-macros functions, which are compiled into a shared library.
 205 /// As a result, a large amount of the information we normally store
 206 /// (e.g. optimized MIR) is unneeded by downstream crates.
 207 /// 2. We avoid serializing invalid `CrateNum`s. When we deserialize
 208 /// a proc-macro crate, we don't load any of its dependencies (since we
 209 /// just need to invoke a native function from the shared library).
 210 /// This means that any foreign `CrateNum`s that we serialize cannot be
 211 /// deserialized, since we will not know how to map them into the current
 212 /// compilation session. If we were to serialize a proc-macro crate like
 213 /// a normal crate, much of what we serialized would be unusable in addition
 214 /// to being unused.
 215 #[derive(MetadataEncodable, MetadataDecodable)]
 216 pub(crate) struct CrateRoot {
 217     name: Symbol,
 218     triple: TargetTriple,
 219     extra_filename: String,
 220     hash: Svh,
 221     stable_crate_id: StableCrateId,
 222     required_panic_strategy: Option<PanicStrategy>,
 223     panic_in_drop_strategy: PanicStrategy,
 224     edition: Edition,
 225     has_global_allocator: bool,
 226     has_alloc_error_handler: bool,
 227     has_panic_handler: bool,
 228     has_default_lib_allocator: bool,
 229
 230     crate_deps: LazyArray<CrateDep>,
 231     dylib_dependency_formats: LazyArray<Option<LinkagePreference>>,
 232     lib_features: LazyArray<(Symbol, Option<Symbol>)>,
 233     stability_implications: LazyArray<(Symbol, Symbol)>,
 234     lang_items: LazyArray<(DefIndex, LangItem)>,
 235     lang_items_missing: LazyArray<LangItem>,
 236     diagnostic_items: LazyArray<(Symbol, DefIndex)>,
 237     native_libraries: LazyArray<NativeLib>,
 238     foreign_modules: LazyArray<ForeignModule>,
 239     traits: LazyArray<DefIndex>,
 240     impls: LazyArray<TraitImpls>,
 241     incoherent_impls: LazyArray<IncoherentImpls>,
 242     interpret_alloc_index: LazyArray<u32>,
 243     proc_macro_data: Option<ProcMacroData>,
 244
 245     tables: LazyTables,
 246     debugger_visualizers: LazyArray<rustc_span::DebuggerVisualizerFile>,
 247
 248     exported_symbols: LazyArray<(ExportedSymbol<'static>, SymbolExportInfo)>,
 249
 250     syntax_contexts: SyntaxContextTable,
 251     expn_data: ExpnDataTable,
 252     expn_hashes: ExpnHashTable,
 253
 254     def_path_hash_map: LazyValue<DefPathHashMapRef<'static>>,
 255
 256     source_map: LazyTable<u32, LazyValue<rustc_span::SourceFile>>,
 257
 258     compiler_builtins: bool,
 259     needs_allocator: bool,
 260     needs_panic_runtime: bool,
 261     no_builtins: bool,
 262     panic_runtime: bool,
 263     profiler_runtime: bool,
 264     symbol_mangling_version: SymbolManglingVersion,
 265 }
 266
 267 /// On-disk representation of `DefId`.
 268 /// This creates a type-safe way to enforce that we remap the CrateNum between the on-disk
 269 /// representation and the compilation session.
 270 #[derive(Copy, Clone)]
 271 pub(crate) struct RawDefId {
 272     krate: u32,
 273     index: u32,
 274 }
 275
 276 impl Into<RawDefId> for DefId {
 277     fn into(self) -> RawDefId {
 278         RawDefId { krate: self.krate.as_u32(), index: self.index.as_u32() }
 279     }
 280 }
 281
 282 impl RawDefId {
 283     /// This exists so that `provide_one!` is happy
 284     fn decode(self, meta: (CrateMetadataRef<'_>, TyCtxt<'_>)) -> DefId {
 285         self.decode_from_cdata(meta.0)
 286     }
 287
 288     fn decode_from_cdata(self, cdata: CrateMetadataRef<'_>) -> DefId {
 289         let krate = CrateNum::from_u32(self.krate);
 290         let krate = cdata.map_encoded_cnum_to_current(krate);
 291         DefId { krate, index: DefIndex::from_u32(self.index) }
 292     }
 293 }
 294
 295 #[derive(Encodable, Decodable)]
 296 pub(crate) struct CrateDep {
 297     pub name: Symbol,
 298     pub hash: Svh,
 299     pub host_hash: Option<Svh>,
 300     pub kind: CrateDepKind,
 301     pub extra_filename: String,
 302 }
 303
 304 #[derive(MetadataEncodable, MetadataDecodable)]
 305 pub(crate) struct TraitImpls {
 306     trait_id: (u32, DefIndex),
 307     impls: LazyArray<(DefIndex, Option<SimplifiedType>)>,
 308 }
 309
 310 #[derive(MetadataEncodable, MetadataDecodable)]
 311 pub(crate) struct IncoherentImpls {
 312     self_ty: SimplifiedType,
 313     impls: LazyArray<DefIndex>,
 314 }
 315
 316 /// Define `LazyTables` and `TableBuilders` at the same time.
 317 macro_rules! define_tables {
 318     ($($name:ident: Table<$IDX:ty, $T:ty>),+ $(,)?) => {
 319         #[derive(MetadataEncodable, MetadataDecodable)]
 320         pub(crate) struct LazyTables {
 321             $($name: LazyTable<$IDX, $T>),+
 322         }
 323
 324         #[derive(Default)]
 325         struct TableBuilders {
 326             $($name: TableBuilder<$IDX, $T>),+
 327         }
 328
 329         impl TableBuilders {
 330             fn encode(&self, buf: &mut FileEncoder) -> LazyTables {
 331                 LazyTables {
 332                     $($name: self.$name.encode(buf)),+
 333                 }
 334             }
 335         }
 336     }
 337 }
 338
 339 define_tables! {
 340     attributes: Table<DefIndex, LazyArray<ast::Attribute>>,
 341     children: Table<DefIndex, LazyArray<DefIndex>>,
 342
 343     opt_def_kind: Table<DefIndex, DefKind>,
 344     visibility: Table<DefIndex, LazyValue<ty::Visibility<DefIndex>>>,
 345     def_span: Table<DefIndex, LazyValue<Span>>,
 346     def_ident_span: Table<DefIndex, LazyValue<Span>>,
 347     lookup_stability: Table<DefIndex, LazyValue<attr::Stability>>,
 348     lookup_const_stability: Table<DefIndex, LazyValue<attr::ConstStability>>,
 349     lookup_default_body_stability: Table<DefIndex, LazyValue<attr::DefaultBodyStability>>,
 350     lookup_deprecation_entry: Table<DefIndex, LazyValue<attr::Deprecation>>,
 351     // As an optimization, a missing entry indicates an empty `&[]`.
 352     explicit_item_bounds: Table<DefIndex, LazyArray<(ty::Predicate<'static>, Span)>>,
 353     explicit_predicates_of: Table<DefIndex, LazyValue<ty::GenericPredicates<'static>>>,
 354     generics_of: Table<DefIndex, LazyValue<ty::Generics>>,
 355     // As an optimization, a missing entry indicates an empty `&[]`.
 356     inferred_outlives_of: Table<DefIndex, LazyArray<(ty::Clause<'static>, Span)>>,
 357     super_predicates_of: Table<DefIndex, LazyValue<ty::GenericPredicates<'static>>>,
 358     type_of: Table<DefIndex, LazyValue<Ty<'static>>>,
 359     variances_of: Table<DefIndex, LazyArray<ty::Variance>>,
 360     fn_sig: Table<DefIndex, LazyValue<ty::PolyFnSig<'static>>>,
 361     codegen_fn_attrs: Table<DefIndex, LazyValue<CodegenFnAttrs>>,
 362     impl_trait_ref: Table<DefIndex, LazyValue<ty::EarlyBinder<ty::TraitRef<'static>>>>,
 363     const_param_default: Table<DefIndex, LazyValue<ty::EarlyBinder<rustc_middle::ty::Const<'static>>>>,
 364     object_lifetime_default: Table<DefIndex, LazyValue<ObjectLifetimeDefault>>,
 365     optimized_mir: Table<DefIndex, LazyValue<mir::Body<'static>>>,
 366     mir_for_ctfe: Table<DefIndex, LazyValue<mir::Body<'static>>>,
 367     promoted_mir: Table<DefIndex, LazyValue<IndexVec<mir::Promoted, mir::Body<'static>>>>,
 368     // FIXME(compiler-errors): Why isn't this a LazyArray?
 369     thir_abstract_const: Table<DefIndex, LazyValue<ty::Const<'static>>>,
 370     impl_parent: Table<DefIndex, RawDefId>,
 371     impl_polarity: Table<DefIndex, ty::ImplPolarity>,
 372     constness: Table<DefIndex, hir::Constness>,
 373     is_intrinsic: Table<DefIndex, ()>,
 374     impl_defaultness: Table<DefIndex, hir::Defaultness>,
 375     // FIXME(eddyb) perhaps compute this on the fly if cheap enough?
 376     coerce_unsized_info: Table<DefIndex, LazyValue<ty::adjustment::CoerceUnsizedInfo>>,
 377     mir_const_qualif: Table<DefIndex, LazyValue<mir::ConstQualifs>>,
 378     rendered_const: Table<DefIndex, LazyValue<String>>,
 379     asyncness: Table<DefIndex, hir::IsAsync>,
 380     fn_arg_names: Table<DefIndex, LazyArray<Ident>>,
 381     generator_kind: Table<DefIndex, LazyValue<hir::GeneratorKind>>,
 382     trait_def: Table<DefIndex, LazyValue<ty::TraitDef>>,
 383
 384     trait_item_def_id: Table<DefIndex, RawDefId>,
 385     inherent_impls: Table<DefIndex, LazyArray<DefIndex>>,
 386     expn_that_defined: Table<DefIndex, LazyValue<ExpnId>>,
 387     unused_generic_params: Table<DefIndex, LazyValue<UnusedGenericParams>>,
 388     params_in_repr: Table<DefIndex, LazyValue<BitSet<u32>>>,
 389     repr_options: Table<DefIndex, LazyValue<ReprOptions>>,
 390     // `def_keys` and `def_path_hashes` represent a lazy version of a
 391     // `DefPathTable`. This allows us to avoid deserializing an entire
 392     // `DefPathTable` up front, since we may only ever use a few
 393     // definitions from any given crate.
 394     def_keys: Table<DefIndex, LazyValue<DefKey>>,
 395     def_path_hashes: Table<DefIndex, DefPathHash>,
 396     proc_macro_quoted_spans: Table<usize, LazyValue<Span>>,
 397     generator_diagnostic_data: Table<DefIndex, LazyValue<GeneratorDiagnosticData<'static>>>,
 398     may_have_doc_links: Table<DefIndex, ()>,
 399     variant_data: Table<DefIndex, LazyValue<VariantData>>,
 400     assoc_container: Table<DefIndex, ty::AssocItemContainer>,
 401     // Slot is full when macro is macro_rules.
 402     macro_rules: Table<DefIndex, ()>,
 403     macro_definition: Table<DefIndex, LazyValue<ast::DelimArgs>>,
 404     proc_macro: Table<DefIndex, MacroKind>,
 405     module_reexports: Table<DefIndex, LazyArray<ModChild>>,
 406     deduced_param_attrs: Table<DefIndex, LazyArray<DeducedParamAttrs>>,
 407     // Slot is full when opaque is TAIT.
 408     is_type_alias_impl_trait: Table<DefIndex, ()>,
 409
 410     trait_impl_trait_tys: Table<DefIndex, LazyValue<FxHashMap<DefId, Ty<'static>>>>,
 411 }
 412
 413 #[derive(TyEncodable, TyDecodable)]
 414 struct VariantData {
 415     discr: ty::VariantDiscr,
 416     /// If this is unit or tuple-variant/struct, then this is the index of the ctor id.
 417     ctor: Option<(CtorKind, DefIndex)>,
 418     is_non_exhaustive: bool,
 419 }
 420
 421 // Tags used for encoding Spans:
 422 const TAG_VALID_SPAN_LOCAL: u8 = 0;
 423 const TAG_VALID_SPAN_FOREIGN: u8 = 1;
 424 const TAG_PARTIAL_SPAN: u8 = 2;
 425
 426 // Tags for encoding Symbol's
 427 const SYMBOL_STR: u8 = 0;
 428 const SYMBOL_OFFSET: u8 = 1;
 429 const SYMBOL_PREINTERNED: u8 = 2;
 430
 431 pub fn provide(providers: &mut Providers) {
 432     encoder::provide(providers);
 433     decoder::provide(providers);
 434 }
 435
 436 trivially_parameterized_over_tcx! {
 437     VariantData,
 438     RawDefId,
 439     TraitImpls,
 440     IncoherentImpls,
 441     CrateRoot,
 442     CrateDep,
 443 }