rustc_codegen_ssa/
lib.rs

1// tidy-alphabetical-start
2#![allow(internal_features)]
3#![allow(rustc::diagnostic_outside_of_impl)]
4#![allow(rustc::untranslatable_diagnostic)]
5#![doc(html_root_url = "https://doc.rust-lang.org/nightly/nightly-rustc/")]
6#![doc(rust_logo)]
7#![feature(assert_matches)]
8#![feature(box_patterns)]
9#![feature(file_buffered)]
10#![feature(if_let_guard)]
11#![feature(negative_impls)]
12#![feature(rustdoc_internals)]
13#![feature(string_from_utf8_lossy_owned)]
14#![feature(trait_alias)]
15#![feature(try_blocks)]
16#![recursion_limit = "256"]
17// tidy-alphabetical-end
18
19//! This crate contains codegen code that is used by all codegen backends (LLVM and others).
20//! The backend-agnostic functions of this crate use functions defined in various traits that
21//! have to be implemented by each backend.
22
23use std::collections::BTreeSet;
24use std::io;
25use std::path::{Path, PathBuf};
26use std::sync::Arc;
27
28use rustc_data_structures::fx::{FxHashSet, FxIndexMap};
29use rustc_data_structures::unord::UnordMap;
30use rustc_hir::CRATE_HIR_ID;
31use rustc_hir::attrs::{CfgEntry, NativeLibKind};
32use rustc_hir::def_id::CrateNum;
33use rustc_macros::{Decodable, Encodable, HashStable};
34use rustc_metadata::EncodedMetadata;
35use rustc_middle::dep_graph::WorkProduct;
36use rustc_middle::lint::LevelAndSource;
37use rustc_middle::middle::debugger_visualizer::DebuggerVisualizerFile;
38use rustc_middle::middle::dependency_format::Dependencies;
39use rustc_middle::middle::exported_symbols::SymbolExportKind;
40use rustc_middle::ty::TyCtxt;
41use rustc_middle::util::Providers;
42use rustc_serialize::opaque::{FileEncoder, MemDecoder};
43use rustc_serialize::{Decodable, Decoder, Encodable, Encoder};
44use rustc_session::Session;
45use rustc_session::config::{CrateType, OutputFilenames, OutputType, RUST_CGU_EXT};
46use rustc_session::cstore::{self, CrateSource};
47use rustc_session::lint::builtin::LINKER_MESSAGES;
48use rustc_span::Symbol;
49
50pub mod assert_module_sources;
51pub mod back;
52pub mod base;
53pub mod codegen_attrs;
54pub mod common;
55pub mod debuginfo;
56pub mod errors;
57pub mod meth;
58pub mod mir;
59pub mod mono_item;
60pub mod size_of_val;
61pub mod target_features;
62pub mod traits;
63
64rustc_fluent_macro::fluent_messages! { "../messages.ftl" }
65
66pub struct ModuleCodegen<M> {
67    /// The name of the module. When the crate may be saved between
68    /// compilations, incremental compilation requires that name be
69    /// unique amongst **all** crates. Therefore, it should contain
70    /// something unique to this crate (e.g., a module path) as well
71    /// as the crate name and disambiguator.
72    /// We currently generate these names via CodegenUnit::build_cgu_name().
73    pub name: String,
74    pub module_llvm: M,
75    pub kind: ModuleKind,
76    /// Saving the ThinLTO buffer for embedding in the object file.
77    pub thin_lto_buffer: Option<Vec<u8>>,
78}
79
80impl<M> ModuleCodegen<M> {
81    pub fn new_regular(name: impl Into<String>, module: M) -> Self {
82        Self {
83            name: name.into(),
84            module_llvm: module,
85            kind: ModuleKind::Regular,
86            thin_lto_buffer: None,
87        }
88    }
89
90    pub fn new_allocator(name: impl Into<String>, module: M) -> Self {
91        Self {
92            name: name.into(),
93            module_llvm: module,
94            kind: ModuleKind::Allocator,
95            thin_lto_buffer: None,
96        }
97    }
98
99    pub fn into_compiled_module(
100        self,
101        emit_obj: bool,
102        emit_dwarf_obj: bool,
103        emit_bc: bool,
104        emit_asm: bool,
105        emit_ir: bool,
106        outputs: &OutputFilenames,
107        invocation_temp: Option<&str>,
108    ) -> CompiledModule {
109        let object = emit_obj
110            .then(|| outputs.temp_path_for_cgu(OutputType::Object, &self.name, invocation_temp));
111        let dwarf_object =
112            emit_dwarf_obj.then(|| outputs.temp_path_dwo_for_cgu(&self.name, invocation_temp));
113        let bytecode = emit_bc
114            .then(|| outputs.temp_path_for_cgu(OutputType::Bitcode, &self.name, invocation_temp));
115        let assembly = emit_asm
116            .then(|| outputs.temp_path_for_cgu(OutputType::Assembly, &self.name, invocation_temp));
117        let llvm_ir = emit_ir.then(|| {
118            outputs.temp_path_for_cgu(OutputType::LlvmAssembly, &self.name, invocation_temp)
119        });
120
121        CompiledModule {
122            name: self.name,
123            object,
124            dwarf_object,
125            bytecode,
126            assembly,
127            llvm_ir,
128            links_from_incr_cache: Vec::new(),
129        }
130    }
131}
132
133#[derive(Debug, Encodable, Decodable)]
134pub struct CompiledModule {
135    pub name: String,
136    pub object: Option<PathBuf>,
137    pub dwarf_object: Option<PathBuf>,
138    pub bytecode: Option<PathBuf>,
139    pub assembly: Option<PathBuf>, // --emit=asm
140    pub llvm_ir: Option<PathBuf>,  // --emit=llvm-ir, llvm-bc is in bytecode
141    pub links_from_incr_cache: Vec<PathBuf>,
142}
143
144impl CompiledModule {
145    /// Call `emit` function with every artifact type currently compiled
146    pub fn for_each_output(&self, mut emit: impl FnMut(&Path, OutputType)) {
147        if let Some(path) = self.object.as_deref() {
148            emit(path, OutputType::Object);
149        }
150        if let Some(path) = self.bytecode.as_deref() {
151            emit(path, OutputType::Bitcode);
152        }
153        if let Some(path) = self.llvm_ir.as_deref() {
154            emit(path, OutputType::LlvmAssembly);
155        }
156        if let Some(path) = self.assembly.as_deref() {
157            emit(path, OutputType::Assembly);
158        }
159    }
160}
161
162pub(crate) struct CachedModuleCodegen {
163    pub name: String,
164    pub source: WorkProduct,
165}
166
167#[derive(Copy, Clone, Debug, PartialEq, Encodable, Decodable)]
168pub enum ModuleKind {
169    Regular,
170    Allocator,
171}
172
173bitflags::bitflags! {
174    #[derive(Debug, Clone, Copy, PartialEq, Eq)]
175    pub struct MemFlags: u8 {
176        const VOLATILE = 1 << 0;
177        const NONTEMPORAL = 1 << 1;
178        const UNALIGNED = 1 << 2;
179    }
180}
181
182#[derive(Clone, Debug, Encodable, Decodable, HashStable)]
183pub struct NativeLib {
184    pub kind: NativeLibKind,
185    pub name: Symbol,
186    pub filename: Option<Symbol>,
187    pub cfg: Option<CfgEntry>,
188    pub verbatim: bool,
189    pub dll_imports: Vec<cstore::DllImport>,
190}
191
192impl From<&cstore::NativeLib> for NativeLib {
193    fn from(lib: &cstore::NativeLib) -> Self {
194        NativeLib {
195            kind: lib.kind,
196            filename: lib.filename,
197            name: lib.name,
198            cfg: lib.cfg.clone(),
199            verbatim: lib.verbatim.unwrap_or(false),
200            dll_imports: lib.dll_imports.clone(),
201        }
202    }
203}
204
205/// Misc info we load from metadata to persist beyond the tcx.
206///
207/// Note: though `CrateNum` is only meaningful within the same tcx, information within `CrateInfo`
208/// is self-contained. `CrateNum` can be viewed as a unique identifier within a `CrateInfo`, where
209/// `used_crate_source` contains all `CrateSource` of the dependents, and maintains a mapping from
210/// identifiers (`CrateNum`) to `CrateSource`. The other fields map `CrateNum` to the crate's own
211/// additional properties, so that effectively we can retrieve each dependent crate's `CrateSource`
212/// and the corresponding properties without referencing information outside of a `CrateInfo`.
213#[derive(Debug, Encodable, Decodable)]
214pub struct CrateInfo {
215    pub target_cpu: String,
216    pub target_features: Vec<String>,
217    pub crate_types: Vec<CrateType>,
218    pub exported_symbols: UnordMap<CrateType, Vec<(String, SymbolExportKind)>>,
219    pub linked_symbols: FxIndexMap<CrateType, Vec<(String, SymbolExportKind)>>,
220    pub local_crate_name: Symbol,
221    pub compiler_builtins: Option<CrateNum>,
222    pub profiler_runtime: Option<CrateNum>,
223    pub is_no_builtins: FxHashSet<CrateNum>,
224    pub native_libraries: FxIndexMap<CrateNum, Vec<NativeLib>>,
225    pub crate_name: UnordMap<CrateNum, Symbol>,
226    pub used_libraries: Vec<NativeLib>,
227    pub used_crate_source: UnordMap<CrateNum, Arc<CrateSource>>,
228    pub used_crates: Vec<CrateNum>,
229    pub dependency_formats: Arc<Dependencies>,
230    pub windows_subsystem: Option<String>,
231    pub natvis_debugger_visualizers: BTreeSet<DebuggerVisualizerFile>,
232    pub lint_levels: CodegenLintLevels,
233    pub metadata_symbol: String,
234}
235
236/// Target-specific options that get set in `cfg(...)`.
237///
238/// RUSTC_SPECIFIC_FEATURES should be skipped here, those are handled outside codegen.
239pub struct TargetConfig {
240    /// Options to be set in `cfg(target_features)`.
241    pub target_features: Vec<Symbol>,
242    /// Options to be set in `cfg(target_features)`, but including unstable features.
243    pub unstable_target_features: Vec<Symbol>,
244    /// Option for `cfg(target_has_reliable_f16)`, true if `f16` basic arithmetic works.
245    pub has_reliable_f16: bool,
246    /// Option for `cfg(target_has_reliable_f16_math)`, true if `f16` math calls work.
247    pub has_reliable_f16_math: bool,
248    /// Option for `cfg(target_has_reliable_f128)`, true if `f128` basic arithmetic works.
249    pub has_reliable_f128: bool,
250    /// Option for `cfg(target_has_reliable_f128_math)`, true if `f128` math calls work.
251    pub has_reliable_f128_math: bool,
252}
253
254#[derive(Encodable, Decodable)]
255pub struct CodegenResults {
256    pub modules: Vec<CompiledModule>,
257    pub allocator_module: Option<CompiledModule>,
258    pub crate_info: CrateInfo,
259}
260
261pub enum CodegenErrors {
262    WrongFileType,
263    EmptyVersionNumber,
264    EncodingVersionMismatch { version_array: String, rlink_version: u32 },
265    RustcVersionMismatch { rustc_version: String },
266    CorruptFile,
267}
268
269pub fn provide(providers: &mut Providers) {
270    crate::back::symbol_export::provide(providers);
271    crate::base::provide(providers);
272    crate::target_features::provide(providers);
273    crate::codegen_attrs::provide(providers);
274    providers.queries.global_backend_features = |_tcx: TyCtxt<'_>, ()| vec![];
275}
276
277/// Checks if the given filename ends with the `.rcgu.o` extension that `rustc`
278/// uses for the object files it generates.
279pub fn looks_like_rust_object_file(filename: &str) -> bool {
280    let path = Path::new(filename);
281    let ext = path.extension().and_then(|s| s.to_str());
282    if ext != Some(OutputType::Object.extension()) {
283        // The file name does not end with ".o", so it can't be an object file.
284        return false;
285    }
286
287    // Strip the ".o" at the end
288    let ext2 = path.file_stem().and_then(|s| Path::new(s).extension()).and_then(|s| s.to_str());
289
290    // Check if the "inner" extension
291    ext2 == Some(RUST_CGU_EXT)
292}
293
294const RLINK_VERSION: u32 = 1;
295const RLINK_MAGIC: &[u8] = b"rustlink";
296
297impl CodegenResults {
298    pub fn serialize_rlink(
299        sess: &Session,
300        rlink_file: &Path,
301        codegen_results: &CodegenResults,
302        metadata: &EncodedMetadata,
303        outputs: &OutputFilenames,
304    ) -> Result<usize, io::Error> {
305        let mut encoder = FileEncoder::new(rlink_file)?;
306        encoder.emit_raw_bytes(RLINK_MAGIC);
307        // `emit_raw_bytes` is used to make sure that the version representation does not depend on
308        // Encoder's inner representation of `u32`.
309        encoder.emit_raw_bytes(&RLINK_VERSION.to_be_bytes());
310        encoder.emit_str(sess.cfg_version);
311        Encodable::encode(codegen_results, &mut encoder);
312        Encodable::encode(metadata, &mut encoder);
313        Encodable::encode(outputs, &mut encoder);
314        encoder.finish().map_err(|(_path, err)| err)
315    }
316
317    pub fn deserialize_rlink(
318        sess: &Session,
319        data: Vec<u8>,
320    ) -> Result<(Self, EncodedMetadata, OutputFilenames), CodegenErrors> {
321        // The Decodable machinery is not used here because it panics if the input data is invalid
322        // and because its internal representation may change.
323        if !data.starts_with(RLINK_MAGIC) {
324            return Err(CodegenErrors::WrongFileType);
325        }
326        let data = &data[RLINK_MAGIC.len()..];
327        if data.len() < 4 {
328            return Err(CodegenErrors::EmptyVersionNumber);
329        }
330
331        let mut version_array: [u8; 4] = Default::default();
332        version_array.copy_from_slice(&data[..4]);
333        if u32::from_be_bytes(version_array) != RLINK_VERSION {
334            return Err(CodegenErrors::EncodingVersionMismatch {
335                version_array: String::from_utf8_lossy(&version_array).to_string(),
336                rlink_version: RLINK_VERSION,
337            });
338        }
339
340        let Ok(mut decoder) = MemDecoder::new(&data[4..], 0) else {
341            return Err(CodegenErrors::CorruptFile);
342        };
343        let rustc_version = decoder.read_str();
344        if rustc_version != sess.cfg_version {
345            return Err(CodegenErrors::RustcVersionMismatch {
346                rustc_version: rustc_version.to_string(),
347            });
348        }
349
350        let codegen_results = CodegenResults::decode(&mut decoder);
351        let metadata = EncodedMetadata::decode(&mut decoder);
352        let outputs = OutputFilenames::decode(&mut decoder);
353        Ok((codegen_results, metadata, outputs))
354    }
355}
356
357/// A list of lint levels used in codegen.
358///
359/// When using `-Z link-only`, we don't have access to the tcx and must work
360/// solely from the `.rlink` file. `Lint`s are defined too early to be encodeable.
361/// Instead, encode exactly the information we need.
362#[derive(Copy, Clone, Debug, Encodable, Decodable)]
363pub struct CodegenLintLevels {
364    linker_messages: LevelAndSource,
365}
366
367impl CodegenLintLevels {
368    pub fn from_tcx(tcx: TyCtxt<'_>) -> Self {
369        Self { linker_messages: tcx.lint_level_at_node(LINKER_MESSAGES, CRATE_HIR_ID) }
370    }
371}
rustc_codegen_ssa/lib.rs

rustc_codegen_ssa/
lib.rs
