rustc_codegen_ssa/
lib.rs

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

rustc_codegen_ssa/
lib.rs
