| //! Lints in the Rust compiler. |
| //! |
| //! This contains lints which can feasibly be implemented as their own |
| //! AST visitor. Also see `rustc_session::lint::builtin`, which contains the |
| //! definitions of lints that are emitted directly inside the main compiler. |
| //! |
| //! To add a new lint to rustc, declare it here using `declare_lint!()`. |
| //! Then add code to emit the new lint in the appropriate circumstances. |
| //! You can do that in an existing `LintPass` if it makes sense, or in a |
| //! new `LintPass`, or using `Session::add_lint` elsewhere in the |
| //! compiler. Only do the latter if the check can't be written cleanly as a |
| //! `LintPass` (also, note that such lints will need to be defined in |
| //! `rustc_session::lint::builtin`, not here). |
| //! |
| //! If you define a new `EarlyLintPass`, you will also need to add it to the |
| //! `add_early_builtin!` or `add_early_builtin_with_new!` invocation in |
| //! `lib.rs`. Use the former for unit-like structs and the latter for structs |
| //! with a `pub fn new()`. |
| //! |
| //! If you define a new `LateLintPass`, you will also need to add it to the |
| //! `late_lint_methods!` invocation in `lib.rs`. |
| |
| use std::fmt::Write; |
| |
| use ast::token::TokenKind; |
| use rustc_ast::tokenstream::{TokenStream, TokenTree}; |
| use rustc_ast::visit::{FnCtxt, FnKind}; |
| use rustc_ast::{self as ast, *}; |
| use rustc_ast_pretty::pprust::{self, expr_to_string}; |
| use rustc_errors::{Applicability, LintDiagnostic}; |
| use rustc_feature::{AttributeGate, BuiltinAttribute, GateIssue, Stability, deprecated_attributes}; |
| use rustc_hir as hir; |
| use rustc_hir::def::{DefKind, Res}; |
| use rustc_hir::def_id::{CRATE_DEF_ID, DefId, LocalDefId}; |
| use rustc_hir::intravisit::FnKind as HirFnKind; |
| use rustc_hir::{Body, FnDecl, GenericParamKind, PatKind, PredicateOrigin}; |
| use rustc_middle::bug; |
| use rustc_middle::lint::in_external_macro; |
| use rustc_middle::ty::layout::LayoutOf; |
| use rustc_middle::ty::print::with_no_trimmed_paths; |
| use rustc_middle::ty::{self, Ty, TyCtxt, TypeVisitableExt, Upcast, VariantDef}; |
| use rustc_session::lint::FutureIncompatibilityReason; |
| // hardwired lints from rustc_lint_defs |
| pub use rustc_session::lint::builtin::*; |
| use rustc_session::{declare_lint, declare_lint_pass, impl_lint_pass}; |
| use rustc_span::edition::Edition; |
| use rustc_span::source_map::Spanned; |
| use rustc_span::symbol::{Ident, Symbol, kw, sym}; |
| use rustc_span::{BytePos, InnerSpan, Span}; |
| use rustc_target::abi::Abi; |
| use rustc_target::asm::InlineAsmArch; |
| use rustc_trait_selection::infer::{InferCtxtExt, TyCtxtInferExt}; |
| use rustc_trait_selection::traits::misc::type_allowed_to_implement_copy; |
| use rustc_trait_selection::traits::query::evaluate_obligation::InferCtxtExt as _; |
| use rustc_trait_selection::traits::{self}; |
| |
| use crate::errors::BuiltinEllipsisInclusiveRangePatterns; |
| use crate::lints::{ |
| BuiltinAnonymousParams, BuiltinConstNoMangle, BuiltinDeprecatedAttrLink, |
| BuiltinDeprecatedAttrLinkSuggestion, BuiltinDeprecatedAttrUsed, BuiltinDerefNullptr, |
| BuiltinEllipsisInclusiveRangePatternsLint, BuiltinExplicitOutlives, |
| BuiltinExplicitOutlivesSuggestion, BuiltinFeatureIssueNote, BuiltinIncompleteFeatures, |
| BuiltinIncompleteFeaturesHelp, BuiltinInternalFeatures, BuiltinKeywordIdents, |
| BuiltinMissingCopyImpl, BuiltinMissingDebugImpl, BuiltinMissingDoc, BuiltinMutablesTransmutes, |
| BuiltinNoMangleGeneric, BuiltinNonShorthandFieldPatterns, BuiltinSpecialModuleNameUsed, |
| BuiltinTrivialBounds, BuiltinTypeAliasBounds, BuiltinUngatedAsyncFnTrackCaller, |
| BuiltinUnpermittedTypeInit, BuiltinUnpermittedTypeInitSub, BuiltinUnreachablePub, |
| BuiltinUnsafe, BuiltinUnstableFeatures, BuiltinUnusedDocComment, BuiltinUnusedDocCommentSub, |
| BuiltinWhileTrue, InvalidAsmLabel, |
| }; |
| use crate::nonstandard_style::{MethodLateContext, method_context}; |
| use crate::{ |
| EarlyContext, EarlyLintPass, LateContext, LateLintPass, Level, LintContext, |
| fluent_generated as fluent, |
| }; |
| |
| declare_lint! { |
| /// The `while_true` lint detects `while true { }`. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,no_run |
| /// while true { |
| /// |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// `while true` should be replaced with `loop`. A `loop` expression is |
| /// the preferred way to write an infinite loop because it more directly |
| /// expresses the intent of the loop. |
| WHILE_TRUE, |
| Warn, |
| "suggest using `loop { }` instead of `while true { }`" |
| } |
| |
| declare_lint_pass!(WhileTrue => [WHILE_TRUE]); |
| |
| /// Traverse through any amount of parenthesis and return the first non-parens expression. |
| fn pierce_parens(mut expr: &ast::Expr) -> &ast::Expr { |
| while let ast::ExprKind::Paren(sub) = &expr.kind { |
| expr = sub; |
| } |
| expr |
| } |
| |
| impl EarlyLintPass for WhileTrue { |
| #[inline] |
| fn check_expr(&mut self, cx: &EarlyContext<'_>, e: &ast::Expr) { |
| if let ast::ExprKind::While(cond, _, label) = &e.kind |
| && let ast::ExprKind::Lit(token_lit) = pierce_parens(cond).kind |
| && let token::Lit { kind: token::Bool, symbol: kw::True, .. } = token_lit |
| && !cond.span.from_expansion() |
| { |
| let condition_span = e.span.with_hi(cond.span.hi()); |
| let replace = format!( |
| "{}loop", |
| label.map_or_else(String::new, |label| format!("{}: ", label.ident,)) |
| ); |
| cx.emit_span_lint(WHILE_TRUE, condition_span, BuiltinWhileTrue { |
| suggestion: condition_span, |
| replace, |
| }); |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `non_shorthand_field_patterns` lint detects using `Struct { x: x }` |
| /// instead of `Struct { x }` in a pattern. |
| /// |
| /// ### Example |
| /// |
| /// ```rust |
| /// struct Point { |
| /// x: i32, |
| /// y: i32, |
| /// } |
| /// |
| /// |
| /// fn main() { |
| /// let p = Point { |
| /// x: 5, |
| /// y: 5, |
| /// }; |
| /// |
| /// match p { |
| /// Point { x: x, y: y } => (), |
| /// } |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// The preferred style is to avoid the repetition of specifying both the |
| /// field name and the binding name if both identifiers are the same. |
| NON_SHORTHAND_FIELD_PATTERNS, |
| Warn, |
| "using `Struct { x: x }` instead of `Struct { x }` in a pattern" |
| } |
| |
| declare_lint_pass!(NonShorthandFieldPatterns => [NON_SHORTHAND_FIELD_PATTERNS]); |
| |
| impl<'tcx> LateLintPass<'tcx> for NonShorthandFieldPatterns { |
| fn check_pat(&mut self, cx: &LateContext<'_>, pat: &hir::Pat<'_>) { |
| if let PatKind::Struct(ref qpath, field_pats, _) = pat.kind { |
| let variant = cx |
| .typeck_results() |
| .pat_ty(pat) |
| .ty_adt_def() |
| .expect("struct pattern type is not an ADT") |
| .variant_of_res(cx.qpath_res(qpath, pat.hir_id)); |
| for fieldpat in field_pats { |
| if fieldpat.is_shorthand { |
| continue; |
| } |
| if fieldpat.span.from_expansion() { |
| // Don't lint if this is a macro expansion: macro authors |
| // shouldn't have to worry about this kind of style issue |
| // (Issue #49588) |
| continue; |
| } |
| if let PatKind::Binding(binding_annot, _, ident, None) = fieldpat.pat.kind { |
| if cx.tcx.find_field_index(ident, variant) |
| == Some(cx.typeck_results().field_index(fieldpat.hir_id)) |
| { |
| cx.emit_span_lint( |
| NON_SHORTHAND_FIELD_PATTERNS, |
| fieldpat.span, |
| BuiltinNonShorthandFieldPatterns { |
| ident, |
| suggestion: fieldpat.span, |
| prefix: binding_annot.prefix_str(), |
| }, |
| ); |
| } |
| } |
| } |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `unsafe_code` lint catches usage of `unsafe` code and other |
| /// potentially unsound constructs like `no_mangle`, `export_name`, |
| /// and `link_section`. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// #![deny(unsafe_code)] |
| /// fn main() { |
| /// unsafe { |
| /// |
| /// } |
| /// } |
| /// |
| /// #[no_mangle] |
| /// fn func_0() { } |
| /// |
| /// #[export_name = "exported_symbol_name"] |
| /// pub fn name_in_rust() { } |
| /// |
| /// #[no_mangle] |
| /// #[link_section = ".example_section"] |
| /// pub static VAR1: u32 = 1; |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// This lint is intended to restrict the usage of `unsafe` blocks and other |
| /// constructs (including, but not limited to `no_mangle`, `link_section` |
| /// and `export_name` attributes) wrong usage of which causes undefined |
| /// behavior. |
| UNSAFE_CODE, |
| Allow, |
| "usage of `unsafe` code and other potentially unsound constructs" |
| } |
| |
| declare_lint_pass!(UnsafeCode => [UNSAFE_CODE]); |
| |
| impl UnsafeCode { |
| fn report_unsafe( |
| &self, |
| cx: &EarlyContext<'_>, |
| span: Span, |
| decorate: impl for<'a> LintDiagnostic<'a, ()>, |
| ) { |
| // This comes from a macro that has `#[allow_internal_unsafe]`. |
| if span.allows_unsafe() { |
| return; |
| } |
| |
| cx.emit_span_lint(UNSAFE_CODE, span, decorate); |
| } |
| } |
| |
| impl EarlyLintPass for UnsafeCode { |
| fn check_attribute(&mut self, cx: &EarlyContext<'_>, attr: &ast::Attribute) { |
| if attr.has_name(sym::allow_internal_unsafe) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::AllowInternalUnsafe); |
| } |
| } |
| |
| #[inline] |
| fn check_expr(&mut self, cx: &EarlyContext<'_>, e: &ast::Expr) { |
| if let ast::ExprKind::Block(ref blk, _) = e.kind { |
| // Don't warn about generated blocks; that'll just pollute the output. |
| if blk.rules == ast::BlockCheckMode::Unsafe(ast::UserProvided) { |
| self.report_unsafe(cx, blk.span, BuiltinUnsafe::UnsafeBlock); |
| } |
| } |
| } |
| |
| fn check_item(&mut self, cx: &EarlyContext<'_>, it: &ast::Item) { |
| match it.kind { |
| ast::ItemKind::Trait(box ast::Trait { safety: ast::Safety::Unsafe(_), .. }) => { |
| self.report_unsafe(cx, it.span, BuiltinUnsafe::UnsafeTrait); |
| } |
| |
| ast::ItemKind::Impl(box ast::Impl { safety: ast::Safety::Unsafe(_), .. }) => { |
| self.report_unsafe(cx, it.span, BuiltinUnsafe::UnsafeImpl); |
| } |
| |
| ast::ItemKind::Fn(..) => { |
| if let Some(attr) = attr::find_by_name(&it.attrs, sym::no_mangle) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::NoMangleFn); |
| } |
| |
| if let Some(attr) = attr::find_by_name(&it.attrs, sym::export_name) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::ExportNameFn); |
| } |
| |
| if let Some(attr) = attr::find_by_name(&it.attrs, sym::link_section) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::LinkSectionFn); |
| } |
| } |
| |
| ast::ItemKind::Static(..) => { |
| if let Some(attr) = attr::find_by_name(&it.attrs, sym::no_mangle) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::NoMangleStatic); |
| } |
| |
| if let Some(attr) = attr::find_by_name(&it.attrs, sym::export_name) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::ExportNameStatic); |
| } |
| |
| if let Some(attr) = attr::find_by_name(&it.attrs, sym::link_section) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::LinkSectionStatic); |
| } |
| } |
| |
| ast::ItemKind::GlobalAsm(..) => { |
| self.report_unsafe(cx, it.span, BuiltinUnsafe::GlobalAsm); |
| } |
| |
| ast::ItemKind::ForeignMod(ForeignMod { safety, .. }) => { |
| if let Safety::Unsafe(_) = safety { |
| self.report_unsafe(cx, it.span, BuiltinUnsafe::UnsafeExternBlock); |
| } |
| } |
| |
| _ => {} |
| } |
| } |
| |
| fn check_impl_item(&mut self, cx: &EarlyContext<'_>, it: &ast::AssocItem) { |
| if let ast::AssocItemKind::Fn(..) = it.kind { |
| if let Some(attr) = attr::find_by_name(&it.attrs, sym::no_mangle) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::NoMangleMethod); |
| } |
| if let Some(attr) = attr::find_by_name(&it.attrs, sym::export_name) { |
| self.report_unsafe(cx, attr.span, BuiltinUnsafe::ExportNameMethod); |
| } |
| } |
| } |
| |
| fn check_fn(&mut self, cx: &EarlyContext<'_>, fk: FnKind<'_>, span: Span, _: ast::NodeId) { |
| if let FnKind::Fn( |
| ctxt, |
| _, |
| ast::FnSig { header: ast::FnHeader { safety: ast::Safety::Unsafe(_), .. }, .. }, |
| _, |
| _, |
| body, |
| ) = fk |
| { |
| let decorator = match ctxt { |
| FnCtxt::Foreign => return, |
| FnCtxt::Free => BuiltinUnsafe::DeclUnsafeFn, |
| FnCtxt::Assoc(_) if body.is_none() => BuiltinUnsafe::DeclUnsafeMethod, |
| FnCtxt::Assoc(_) => BuiltinUnsafe::ImplUnsafeMethod, |
| }; |
| self.report_unsafe(cx, span, decorator); |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `missing_docs` lint detects missing documentation for public items. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// #![deny(missing_docs)] |
| /// pub fn foo() {} |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// This lint is intended to ensure that a library is well-documented. |
| /// Items without documentation can be difficult for users to understand |
| /// how to use properly. |
| /// |
| /// This lint is "allow" by default because it can be noisy, and not all |
| /// projects may want to enforce everything to be documented. |
| pub MISSING_DOCS, |
| Allow, |
| "detects missing documentation for public members", |
| report_in_external_macro |
| } |
| |
| pub struct MissingDoc; |
| |
| impl_lint_pass!(MissingDoc => [MISSING_DOCS]); |
| |
| fn has_doc(attr: &ast::Attribute) -> bool { |
| if attr.is_doc_comment() { |
| return true; |
| } |
| |
| if !attr.has_name(sym::doc) { |
| return false; |
| } |
| |
| if attr.value_str().is_some() { |
| return true; |
| } |
| |
| if let Some(list) = attr.meta_item_list() { |
| for meta in list { |
| if meta.has_name(sym::hidden) { |
| return true; |
| } |
| } |
| } |
| |
| false |
| } |
| |
| impl MissingDoc { |
| fn check_missing_docs_attrs( |
| &self, |
| cx: &LateContext<'_>, |
| def_id: LocalDefId, |
| article: &'static str, |
| desc: &'static str, |
| ) { |
| // Only check publicly-visible items, using the result from the privacy pass. |
| // It's an option so the crate root can also use this function (it doesn't |
| // have a `NodeId`). |
| if def_id != CRATE_DEF_ID && !cx.effective_visibilities.is_exported(def_id) { |
| return; |
| } |
| |
| let attrs = cx.tcx.hir().attrs(cx.tcx.local_def_id_to_hir_id(def_id)); |
| let has_doc = attrs.iter().any(has_doc); |
| if !has_doc { |
| cx.emit_span_lint(MISSING_DOCS, cx.tcx.def_span(def_id), BuiltinMissingDoc { |
| article, |
| desc, |
| }); |
| } |
| } |
| } |
| |
| impl<'tcx> LateLintPass<'tcx> for MissingDoc { |
| fn check_crate(&mut self, cx: &LateContext<'_>) { |
| self.check_missing_docs_attrs(cx, CRATE_DEF_ID, "the", "crate"); |
| } |
| |
| fn check_item(&mut self, cx: &LateContext<'_>, it: &hir::Item<'_>) { |
| // Previously the Impl and Use types have been excluded from missing docs, |
| // so we will continue to exclude them for compatibility. |
| // |
| // The documentation on `ExternCrate` is not used at the moment so no need to warn for it. |
| if let hir::ItemKind::Impl(..) | hir::ItemKind::Use(..) | hir::ItemKind::ExternCrate(_) = |
| it.kind |
| { |
| return; |
| } |
| |
| let (article, desc) = cx.tcx.article_and_description(it.owner_id.to_def_id()); |
| self.check_missing_docs_attrs(cx, it.owner_id.def_id, article, desc); |
| } |
| |
| fn check_trait_item(&mut self, cx: &LateContext<'_>, trait_item: &hir::TraitItem<'_>) { |
| let (article, desc) = cx.tcx.article_and_description(trait_item.owner_id.to_def_id()); |
| |
| self.check_missing_docs_attrs(cx, trait_item.owner_id.def_id, article, desc); |
| } |
| |
| fn check_impl_item(&mut self, cx: &LateContext<'_>, impl_item: &hir::ImplItem<'_>) { |
| let context = method_context(cx, impl_item.owner_id.def_id); |
| |
| match context { |
| // If the method is an impl for a trait, don't doc. |
| MethodLateContext::TraitImpl => return, |
| MethodLateContext::TraitAutoImpl => {} |
| // If the method is an impl for an item with docs_hidden, don't doc. |
| MethodLateContext::PlainImpl => { |
| let parent = cx.tcx.hir().get_parent_item(impl_item.hir_id()); |
| let impl_ty = cx.tcx.type_of(parent).instantiate_identity(); |
| let outerdef = match impl_ty.kind() { |
| ty::Adt(def, _) => Some(def.did()), |
| ty::Foreign(def_id) => Some(*def_id), |
| _ => None, |
| }; |
| let is_hidden = match outerdef { |
| Some(id) => cx.tcx.is_doc_hidden(id), |
| None => false, |
| }; |
| if is_hidden { |
| return; |
| } |
| } |
| } |
| |
| let (article, desc) = cx.tcx.article_and_description(impl_item.owner_id.to_def_id()); |
| self.check_missing_docs_attrs(cx, impl_item.owner_id.def_id, article, desc); |
| } |
| |
| fn check_foreign_item(&mut self, cx: &LateContext<'_>, foreign_item: &hir::ForeignItem<'_>) { |
| let (article, desc) = cx.tcx.article_and_description(foreign_item.owner_id.to_def_id()); |
| self.check_missing_docs_attrs(cx, foreign_item.owner_id.def_id, article, desc); |
| } |
| |
| fn check_field_def(&mut self, cx: &LateContext<'_>, sf: &hir::FieldDef<'_>) { |
| if !sf.is_positional() { |
| self.check_missing_docs_attrs(cx, sf.def_id, "a", "struct field") |
| } |
| } |
| |
| fn check_variant(&mut self, cx: &LateContext<'_>, v: &hir::Variant<'_>) { |
| self.check_missing_docs_attrs(cx, v.def_id, "a", "variant"); |
| } |
| } |
| |
| declare_lint! { |
| /// The `missing_copy_implementations` lint detects potentially-forgotten |
| /// implementations of [`Copy`] for public types. |
| /// |
| /// [`Copy`]: https://doc.rust-lang.org/std/marker/trait.Copy.html |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// #![deny(missing_copy_implementations)] |
| /// pub struct Foo { |
| /// pub field: i32 |
| /// } |
| /// # fn main() {} |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Historically (before 1.0), types were automatically marked as `Copy` |
| /// if possible. This was changed so that it required an explicit opt-in |
| /// by implementing the `Copy` trait. As part of this change, a lint was |
| /// added to alert if a copyable type was not marked `Copy`. |
| /// |
| /// This lint is "allow" by default because this code isn't bad; it is |
| /// common to write newtypes like this specifically so that a `Copy` type |
| /// is no longer `Copy`. `Copy` types can result in unintended copies of |
| /// large data which can impact performance. |
| pub MISSING_COPY_IMPLEMENTATIONS, |
| Allow, |
| "detects potentially-forgotten implementations of `Copy`" |
| } |
| |
| declare_lint_pass!(MissingCopyImplementations => [MISSING_COPY_IMPLEMENTATIONS]); |
| |
| impl<'tcx> LateLintPass<'tcx> for MissingCopyImplementations { |
| fn check_item(&mut self, cx: &LateContext<'_>, item: &hir::Item<'_>) { |
| if !cx.effective_visibilities.is_reachable(item.owner_id.def_id) { |
| return; |
| } |
| let (def, ty) = match item.kind { |
| hir::ItemKind::Struct(_, ast_generics) => { |
| if !ast_generics.params.is_empty() { |
| return; |
| } |
| let def = cx.tcx.adt_def(item.owner_id); |
| (def, Ty::new_adt(cx.tcx, def, ty::List::empty())) |
| } |
| hir::ItemKind::Union(_, ast_generics) => { |
| if !ast_generics.params.is_empty() { |
| return; |
| } |
| let def = cx.tcx.adt_def(item.owner_id); |
| (def, Ty::new_adt(cx.tcx, def, ty::List::empty())) |
| } |
| hir::ItemKind::Enum(_, ast_generics) => { |
| if !ast_generics.params.is_empty() { |
| return; |
| } |
| let def = cx.tcx.adt_def(item.owner_id); |
| (def, Ty::new_adt(cx.tcx, def, ty::List::empty())) |
| } |
| _ => return, |
| }; |
| if def.has_dtor(cx.tcx) { |
| return; |
| } |
| |
| // If the type contains a raw pointer, it may represent something like a handle, |
| // and recommending Copy might be a bad idea. |
| for field in def.all_fields() { |
| let did = field.did; |
| if cx.tcx.type_of(did).instantiate_identity().is_unsafe_ptr() { |
| return; |
| } |
| } |
| if ty.is_copy_modulo_regions(cx.tcx, cx.param_env) { |
| return; |
| } |
| if type_implements_negative_copy_modulo_regions(cx.tcx, ty, cx.param_env) { |
| return; |
| } |
| if def.is_variant_list_non_exhaustive() |
| || def.variants().iter().any(|variant| variant.is_field_list_non_exhaustive()) |
| { |
| return; |
| } |
| |
| // We shouldn't recommend implementing `Copy` on stateful things, |
| // such as iterators. |
| if let Some(iter_trait) = cx.tcx.get_diagnostic_item(sym::Iterator) |
| && cx |
| .tcx |
| .infer_ctxt() |
| .build() |
| .type_implements_trait(iter_trait, [ty], cx.param_env) |
| .must_apply_modulo_regions() |
| { |
| return; |
| } |
| |
| // Default value of clippy::trivially_copy_pass_by_ref |
| const MAX_SIZE: u64 = 256; |
| |
| if let Some(size) = cx.layout_of(ty).ok().map(|l| l.size.bytes()) { |
| if size > MAX_SIZE { |
| return; |
| } |
| } |
| |
| if type_allowed_to_implement_copy( |
| cx.tcx, |
| cx.param_env, |
| ty, |
| traits::ObligationCause::misc(item.span, item.owner_id.def_id), |
| ) |
| .is_ok() |
| { |
| cx.emit_span_lint(MISSING_COPY_IMPLEMENTATIONS, item.span, BuiltinMissingCopyImpl); |
| } |
| } |
| } |
| |
| /// Check whether a `ty` has a negative `Copy` implementation, ignoring outlives constraints. |
| fn type_implements_negative_copy_modulo_regions<'tcx>( |
| tcx: TyCtxt<'tcx>, |
| ty: Ty<'tcx>, |
| param_env: ty::ParamEnv<'tcx>, |
| ) -> bool { |
| let trait_ref = ty::TraitRef::new(tcx, tcx.require_lang_item(hir::LangItem::Copy, None), [ty]); |
| let pred = ty::TraitPredicate { trait_ref, polarity: ty::PredicatePolarity::Negative }; |
| let obligation = traits::Obligation { |
| cause: traits::ObligationCause::dummy(), |
| param_env, |
| recursion_depth: 0, |
| predicate: pred.upcast(tcx), |
| }; |
| |
| tcx.infer_ctxt().build().predicate_must_hold_modulo_regions(&obligation) |
| } |
| |
| declare_lint! { |
| /// The `missing_debug_implementations` lint detects missing |
| /// implementations of [`fmt::Debug`] for public types. |
| /// |
| /// [`fmt::Debug`]: https://doc.rust-lang.org/std/fmt/trait.Debug.html |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// #![deny(missing_debug_implementations)] |
| /// pub struct Foo; |
| /// # fn main() {} |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Having a `Debug` implementation on all types can assist with |
| /// debugging, as it provides a convenient way to format and display a |
| /// value. Using the `#[derive(Debug)]` attribute will automatically |
| /// generate a typical implementation, or a custom implementation can be |
| /// added by manually implementing the `Debug` trait. |
| /// |
| /// This lint is "allow" by default because adding `Debug` to all types can |
| /// have a negative impact on compile time and code size. It also requires |
| /// boilerplate to be added to every type, which can be an impediment. |
| MISSING_DEBUG_IMPLEMENTATIONS, |
| Allow, |
| "detects missing implementations of Debug" |
| } |
| |
| #[derive(Default)] |
| pub(crate) struct MissingDebugImplementations; |
| |
| impl_lint_pass!(MissingDebugImplementations => [MISSING_DEBUG_IMPLEMENTATIONS]); |
| |
| impl<'tcx> LateLintPass<'tcx> for MissingDebugImplementations { |
| fn check_item(&mut self, cx: &LateContext<'_>, item: &hir::Item<'_>) { |
| if !cx.effective_visibilities.is_reachable(item.owner_id.def_id) { |
| return; |
| } |
| |
| match item.kind { |
| hir::ItemKind::Struct(..) | hir::ItemKind::Union(..) | hir::ItemKind::Enum(..) => {} |
| _ => return, |
| } |
| |
| // Avoid listing trait impls if the trait is allowed. |
| let (level, _) = cx.tcx.lint_level_at_node(MISSING_DEBUG_IMPLEMENTATIONS, item.hir_id()); |
| if level == Level::Allow { |
| return; |
| } |
| |
| let Some(debug) = cx.tcx.get_diagnostic_item(sym::Debug) else { return }; |
| |
| let has_impl = cx |
| .tcx |
| .non_blanket_impls_for_ty(debug, cx.tcx.type_of(item.owner_id).instantiate_identity()) |
| .next() |
| .is_some(); |
| if !has_impl { |
| cx.emit_span_lint(MISSING_DEBUG_IMPLEMENTATIONS, item.span, BuiltinMissingDebugImpl { |
| tcx: cx.tcx, |
| def_id: debug, |
| }); |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `anonymous_parameters` lint detects anonymous parameters in trait |
| /// definitions. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,edition2015,compile_fail |
| /// #![deny(anonymous_parameters)] |
| /// // edition 2015 |
| /// pub trait Foo { |
| /// fn foo(usize); |
| /// } |
| /// fn main() {} |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// This syntax is mostly a historical accident, and can be worked around |
| /// quite easily by adding an `_` pattern or a descriptive identifier: |
| /// |
| /// ```rust |
| /// trait Foo { |
| /// fn foo(_: usize); |
| /// } |
| /// ``` |
| /// |
| /// This syntax is now a hard error in the 2018 edition. In the 2015 |
| /// edition, this lint is "warn" by default. This lint |
| /// enables the [`cargo fix`] tool with the `--edition` flag to |
| /// automatically transition old code from the 2015 edition to 2018. The |
| /// tool will run this lint and automatically apply the |
| /// suggested fix from the compiler (which is to add `_` to each |
| /// parameter). This provides a completely automated way to update old |
| /// code for a new edition. See [issue #41686] for more details. |
| /// |
| /// [issue #41686]: https://github.com/rust-lang/rust/issues/41686 |
| /// [`cargo fix`]: https://doc.rust-lang.org/cargo/commands/cargo-fix.html |
| pub ANONYMOUS_PARAMETERS, |
| Warn, |
| "detects anonymous parameters", |
| @future_incompatible = FutureIncompatibleInfo { |
| reason: FutureIncompatibilityReason::EditionError(Edition::Edition2018), |
| reference: "issue #41686 <https://github.com/rust-lang/rust/issues/41686>", |
| }; |
| } |
| |
| declare_lint_pass!( |
| /// Checks for use of anonymous parameters (RFC 1685). |
| AnonymousParameters => [ANONYMOUS_PARAMETERS] |
| ); |
| |
| impl EarlyLintPass for AnonymousParameters { |
| fn check_trait_item(&mut self, cx: &EarlyContext<'_>, it: &ast::AssocItem) { |
| if cx.sess().edition() != Edition::Edition2015 { |
| // This is a hard error in future editions; avoid linting and erroring |
| return; |
| } |
| if let ast::AssocItemKind::Fn(box Fn { ref sig, .. }) = it.kind { |
| for arg in sig.decl.inputs.iter() { |
| if let ast::PatKind::Ident(_, ident, None) = arg.pat.kind { |
| if ident.name == kw::Empty { |
| let ty_snip = cx.sess().source_map().span_to_snippet(arg.ty.span); |
| |
| let (ty_snip, appl) = if let Ok(ref snip) = ty_snip { |
| (snip.as_str(), Applicability::MachineApplicable) |
| } else { |
| ("<type>", Applicability::HasPlaceholders) |
| }; |
| cx.emit_span_lint( |
| ANONYMOUS_PARAMETERS, |
| arg.pat.span, |
| BuiltinAnonymousParams { suggestion: (arg.pat.span, appl), ty_snip }, |
| ); |
| } |
| } |
| } |
| } |
| } |
| } |
| |
| /// Check for use of attributes which have been deprecated. |
| #[derive(Clone)] |
| pub struct DeprecatedAttr { |
| // This is not free to compute, so we want to keep it around, rather than |
| // compute it for every attribute. |
| depr_attrs: Vec<&'static BuiltinAttribute>, |
| } |
| |
| impl_lint_pass!(DeprecatedAttr => []); |
| |
| impl DeprecatedAttr { |
| pub fn new() -> DeprecatedAttr { |
| DeprecatedAttr { depr_attrs: deprecated_attributes() } |
| } |
| } |
| |
| impl EarlyLintPass for DeprecatedAttr { |
| fn check_attribute(&mut self, cx: &EarlyContext<'_>, attr: &ast::Attribute) { |
| for BuiltinAttribute { name, gate, .. } in &self.depr_attrs { |
| if attr.ident().map(|ident| ident.name) == Some(*name) { |
| if let &AttributeGate::Gated( |
| Stability::Deprecated(link, suggestion), |
| name, |
| reason, |
| _, |
| ) = gate |
| { |
| let suggestion = match suggestion { |
| Some(msg) => { |
| BuiltinDeprecatedAttrLinkSuggestion::Msg { suggestion: attr.span, msg } |
| } |
| None => { |
| BuiltinDeprecatedAttrLinkSuggestion::Default { suggestion: attr.span } |
| } |
| }; |
| cx.emit_span_lint(DEPRECATED, attr.span, BuiltinDeprecatedAttrLink { |
| name, |
| reason, |
| link, |
| suggestion, |
| }); |
| } |
| return; |
| } |
| } |
| if attr.has_name(sym::no_start) || attr.has_name(sym::crate_id) { |
| cx.emit_span_lint(DEPRECATED, attr.span, BuiltinDeprecatedAttrUsed { |
| name: pprust::path_to_string(&attr.get_normal_item().path), |
| suggestion: attr.span, |
| }); |
| } |
| } |
| } |
| |
| fn warn_if_doc(cx: &EarlyContext<'_>, node_span: Span, node_kind: &str, attrs: &[ast::Attribute]) { |
| use rustc_ast::token::CommentKind; |
| |
| let mut attrs = attrs.iter().peekable(); |
| |
| // Accumulate a single span for sugared doc comments. |
| let mut sugared_span: Option<Span> = None; |
| |
| while let Some(attr) = attrs.next() { |
| let is_doc_comment = attr.is_doc_comment(); |
| if is_doc_comment { |
| sugared_span = |
| Some(sugared_span.map_or(attr.span, |span| span.with_hi(attr.span.hi()))); |
| } |
| |
| if attrs.peek().is_some_and(|next_attr| next_attr.is_doc_comment()) { |
| continue; |
| } |
| |
| let span = sugared_span.take().unwrap_or(attr.span); |
| |
| if is_doc_comment || attr.has_name(sym::doc) { |
| let sub = match attr.kind { |
| AttrKind::DocComment(CommentKind::Line, _) | AttrKind::Normal(..) => { |
| BuiltinUnusedDocCommentSub::PlainHelp |
| } |
| AttrKind::DocComment(CommentKind::Block, _) => { |
| BuiltinUnusedDocCommentSub::BlockHelp |
| } |
| }; |
| cx.emit_span_lint(UNUSED_DOC_COMMENTS, span, BuiltinUnusedDocComment { |
| kind: node_kind, |
| label: node_span, |
| sub, |
| }); |
| } |
| } |
| } |
| |
| impl EarlyLintPass for UnusedDocComment { |
| fn check_stmt(&mut self, cx: &EarlyContext<'_>, stmt: &ast::Stmt) { |
| let kind = match stmt.kind { |
| ast::StmtKind::Let(..) => "statements", |
| // Disabled pending discussion in #78306 |
| ast::StmtKind::Item(..) => return, |
| // expressions will be reported by `check_expr`. |
| ast::StmtKind::Empty |
| | ast::StmtKind::Semi(_) |
| | ast::StmtKind::Expr(_) |
| | ast::StmtKind::MacCall(_) => return, |
| }; |
| |
| warn_if_doc(cx, stmt.span, kind, stmt.kind.attrs()); |
| } |
| |
| fn check_arm(&mut self, cx: &EarlyContext<'_>, arm: &ast::Arm) { |
| if let Some(body) = &arm.body { |
| let arm_span = arm.pat.span.with_hi(body.span.hi()); |
| warn_if_doc(cx, arm_span, "match arms", &arm.attrs); |
| } |
| } |
| |
| fn check_pat(&mut self, cx: &EarlyContext<'_>, pat: &ast::Pat) { |
| if let ast::PatKind::Struct(_, _, fields, _) = &pat.kind { |
| for field in fields { |
| warn_if_doc(cx, field.span, "pattern fields", &field.attrs); |
| } |
| } |
| } |
| |
| fn check_expr(&mut self, cx: &EarlyContext<'_>, expr: &ast::Expr) { |
| warn_if_doc(cx, expr.span, "expressions", &expr.attrs); |
| |
| if let ExprKind::Struct(s) = &expr.kind { |
| for field in &s.fields { |
| warn_if_doc(cx, field.span, "expression fields", &field.attrs); |
| } |
| } |
| } |
| |
| fn check_generic_param(&mut self, cx: &EarlyContext<'_>, param: &ast::GenericParam) { |
| warn_if_doc(cx, param.ident.span, "generic parameters", ¶m.attrs); |
| } |
| |
| fn check_block(&mut self, cx: &EarlyContext<'_>, block: &ast::Block) { |
| warn_if_doc(cx, block.span, "blocks", block.attrs()); |
| } |
| |
| fn check_item(&mut self, cx: &EarlyContext<'_>, item: &ast::Item) { |
| if let ast::ItemKind::ForeignMod(_) = item.kind { |
| warn_if_doc(cx, item.span, "extern blocks", &item.attrs); |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `no_mangle_const_items` lint detects any `const` items with the |
| /// [`no_mangle` attribute]. |
| /// |
| /// [`no_mangle` attribute]: https://doc.rust-lang.org/reference/abi.html#the-no_mangle-attribute |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// #[no_mangle] |
| /// const FOO: i32 = 5; |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Constants do not have their symbols exported, and therefore, this |
| /// probably means you meant to use a [`static`], not a [`const`]. |
| /// |
| /// [`static`]: https://doc.rust-lang.org/reference/items/static-items.html |
| /// [`const`]: https://doc.rust-lang.org/reference/items/constant-items.html |
| NO_MANGLE_CONST_ITEMS, |
| Deny, |
| "const items will not have their symbols exported" |
| } |
| |
| declare_lint! { |
| /// The `no_mangle_generic_items` lint detects generic items that must be |
| /// mangled. |
| /// |
| /// ### Example |
| /// |
| /// ```rust |
| /// #[no_mangle] |
| /// fn foo<T>(t: T) { |
| /// |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// A function with generics must have its symbol mangled to accommodate |
| /// the generic parameter. The [`no_mangle` attribute] has no effect in |
| /// this situation, and should be removed. |
| /// |
| /// [`no_mangle` attribute]: https://doc.rust-lang.org/reference/abi.html#the-no_mangle-attribute |
| NO_MANGLE_GENERIC_ITEMS, |
| Warn, |
| "generic items must be mangled" |
| } |
| |
| declare_lint_pass!(InvalidNoMangleItems => [NO_MANGLE_CONST_ITEMS, NO_MANGLE_GENERIC_ITEMS]); |
| |
| impl<'tcx> LateLintPass<'tcx> for InvalidNoMangleItems { |
| fn check_item(&mut self, cx: &LateContext<'_>, it: &hir::Item<'_>) { |
| let attrs = cx.tcx.hir().attrs(it.hir_id()); |
| let check_no_mangle_on_generic_fn = |no_mangle_attr: &ast::Attribute, |
| impl_generics: Option<&hir::Generics<'_>>, |
| generics: &hir::Generics<'_>, |
| span| { |
| for param in |
| generics.params.iter().chain(impl_generics.map(|g| g.params).into_iter().flatten()) |
| { |
| match param.kind { |
| GenericParamKind::Lifetime { .. } => {} |
| GenericParamKind::Type { .. } | GenericParamKind::Const { .. } => { |
| cx.emit_span_lint(NO_MANGLE_GENERIC_ITEMS, span, BuiltinNoMangleGeneric { |
| suggestion: no_mangle_attr.span, |
| }); |
| break; |
| } |
| } |
| } |
| }; |
| match it.kind { |
| hir::ItemKind::Fn(.., generics, _) => { |
| if let Some(no_mangle_attr) = attr::find_by_name(attrs, sym::no_mangle) { |
| check_no_mangle_on_generic_fn(no_mangle_attr, None, generics, it.span); |
| } |
| } |
| hir::ItemKind::Const(..) => { |
| if attr::contains_name(attrs, sym::no_mangle) { |
| // account for "pub const" (#45562) |
| let start = cx |
| .tcx |
| .sess |
| .source_map() |
| .span_to_snippet(it.span) |
| .map(|snippet| snippet.find("const").unwrap_or(0)) |
| .unwrap_or(0) as u32; |
| // `const` is 5 chars |
| let suggestion = it.span.with_hi(BytePos(it.span.lo().0 + start + 5)); |
| |
| // Const items do not refer to a particular location in memory, and therefore |
| // don't have anything to attach a symbol to |
| cx.emit_span_lint(NO_MANGLE_CONST_ITEMS, it.span, BuiltinConstNoMangle { |
| suggestion, |
| }); |
| } |
| } |
| hir::ItemKind::Impl(hir::Impl { generics, items, .. }) => { |
| for it in *items { |
| if let hir::AssocItemKind::Fn { .. } = it.kind { |
| if let Some(no_mangle_attr) = |
| attr::find_by_name(cx.tcx.hir().attrs(it.id.hir_id()), sym::no_mangle) |
| { |
| check_no_mangle_on_generic_fn( |
| no_mangle_attr, |
| Some(generics), |
| cx.tcx.hir().get_generics(it.id.owner_id.def_id).unwrap(), |
| it.span, |
| ); |
| } |
| } |
| } |
| } |
| _ => {} |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `mutable_transmutes` lint catches transmuting from `&T` to `&mut |
| /// T` because it is [undefined behavior]. |
| /// |
| /// [undefined behavior]: https://doc.rust-lang.org/reference/behavior-considered-undefined.html |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// unsafe { |
| /// let y = std::mem::transmute::<&i32, &mut i32>(&5); |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Certain assumptions are made about aliasing of data, and this transmute |
| /// violates those assumptions. Consider using [`UnsafeCell`] instead. |
| /// |
| /// [`UnsafeCell`]: https://doc.rust-lang.org/std/cell/struct.UnsafeCell.html |
| MUTABLE_TRANSMUTES, |
| Deny, |
| "transmuting &T to &mut T is undefined behavior, even if the reference is unused" |
| } |
| |
| declare_lint_pass!(MutableTransmutes => [MUTABLE_TRANSMUTES]); |
| |
| impl<'tcx> LateLintPass<'tcx> for MutableTransmutes { |
| fn check_expr(&mut self, cx: &LateContext<'_>, expr: &hir::Expr<'_>) { |
| if let Some((&ty::Ref(_, _, from_mutbl), &ty::Ref(_, _, to_mutbl))) = |
| get_transmute_from_to(cx, expr).map(|(ty1, ty2)| (ty1.kind(), ty2.kind())) |
| { |
| if from_mutbl < to_mutbl { |
| cx.emit_span_lint(MUTABLE_TRANSMUTES, expr.span, BuiltinMutablesTransmutes); |
| } |
| } |
| |
| fn get_transmute_from_to<'tcx>( |
| cx: &LateContext<'tcx>, |
| expr: &hir::Expr<'_>, |
| ) -> Option<(Ty<'tcx>, Ty<'tcx>)> { |
| let def = if let hir::ExprKind::Path(ref qpath) = expr.kind { |
| cx.qpath_res(qpath, expr.hir_id) |
| } else { |
| return None; |
| }; |
| if let Res::Def(DefKind::Fn, did) = def { |
| if !def_id_is_transmute(cx, did) { |
| return None; |
| } |
| let sig = cx.typeck_results().node_type(expr.hir_id).fn_sig(cx.tcx); |
| let from = sig.inputs().skip_binder()[0]; |
| let to = sig.output().skip_binder(); |
| return Some((from, to)); |
| } |
| None |
| } |
| |
| fn def_id_is_transmute(cx: &LateContext<'_>, def_id: DefId) -> bool { |
| cx.tcx.is_intrinsic(def_id, sym::transmute) |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `unstable_features` lint detects uses of `#![feature]`. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// #![deny(unstable_features)] |
| /// #![feature(test)] |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// In larger nightly-based projects which |
| /// |
| /// * consist of a multitude of crates where a subset of crates has to compile on |
| /// stable either unconditionally or depending on a `cfg` flag to for example |
| /// allow stable users to depend on them, |
| /// * don't use nightly for experimental features but for, e.g., unstable options only, |
| /// |
| /// this lint may come in handy to enforce policies of these kinds. |
| UNSTABLE_FEATURES, |
| Allow, |
| "enabling unstable features" |
| } |
| |
| declare_lint_pass!( |
| /// Forbids using the `#[feature(...)]` attribute |
| UnstableFeatures => [UNSTABLE_FEATURES] |
| ); |
| |
| impl<'tcx> LateLintPass<'tcx> for UnstableFeatures { |
| fn check_attribute(&mut self, cx: &LateContext<'_>, attr: &ast::Attribute) { |
| if attr.has_name(sym::feature) |
| && let Some(items) = attr.meta_item_list() |
| { |
| for item in items { |
| cx.emit_span_lint(UNSTABLE_FEATURES, item.span(), BuiltinUnstableFeatures); |
| } |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `ungated_async_fn_track_caller` lint warns when the |
| /// `#[track_caller]` attribute is used on an async function |
| /// without enabling the corresponding unstable feature flag. |
| /// |
| /// ### Example |
| /// |
| /// ```rust |
| /// #[track_caller] |
| /// async fn foo() {} |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// The attribute must be used in conjunction with the |
| /// [`async_fn_track_caller` feature flag]. Otherwise, the `#[track_caller]` |
| /// annotation will function as a no-op. |
| /// |
| /// [`async_fn_track_caller` feature flag]: https://doc.rust-lang.org/beta/unstable-book/language-features/async-fn-track-caller.html |
| UNGATED_ASYNC_FN_TRACK_CALLER, |
| Warn, |
| "enabling track_caller on an async fn is a no-op unless the async_fn_track_caller feature is enabled" |
| } |
| |
| declare_lint_pass!( |
| /// Explains corresponding feature flag must be enabled for the `#[track_caller]` attribute to |
| /// do anything |
| UngatedAsyncFnTrackCaller => [UNGATED_ASYNC_FN_TRACK_CALLER] |
| ); |
| |
| impl<'tcx> LateLintPass<'tcx> for UngatedAsyncFnTrackCaller { |
| fn check_fn( |
| &mut self, |
| cx: &LateContext<'_>, |
| fn_kind: HirFnKind<'_>, |
| _: &'tcx FnDecl<'_>, |
| _: &'tcx Body<'_>, |
| span: Span, |
| def_id: LocalDefId, |
| ) { |
| if fn_kind.asyncness().is_async() |
| && !cx.tcx.features().async_fn_track_caller |
| // Now, check if the function has the `#[track_caller]` attribute |
| && let Some(attr) = cx.tcx.get_attr(def_id, sym::track_caller) |
| { |
| cx.emit_span_lint( |
| UNGATED_ASYNC_FN_TRACK_CALLER, |
| attr.span, |
| BuiltinUngatedAsyncFnTrackCaller { label: span, session: &cx.tcx.sess }, |
| ); |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `unreachable_pub` lint triggers for `pub` items not reachable from other crates - that |
| /// means neither directly accessible, nor reexported, nor leaked through things like return |
| /// types. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// #![deny(unreachable_pub)] |
| /// mod foo { |
| /// pub mod bar { |
| /// |
| /// } |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// The `pub` keyword both expresses an intent for an item to be publicly available, and also |
| /// signals to the compiler to make the item publicly accessible. The intent can only be |
| /// satisfied, however, if all items which contain this item are *also* publicly accessible. |
| /// Thus, this lint serves to identify situations where the intent does not match the reality. |
| /// |
| /// If you wish the item to be accessible elsewhere within the crate, but not outside it, the |
| /// `pub(crate)` visibility is recommended to be used instead. This more clearly expresses the |
| /// intent that the item is only visible within its own crate. |
| /// |
| /// This lint is "allow" by default because it will trigger for a large |
| /// amount existing Rust code, and has some false-positives. Eventually it |
| /// is desired for this to become warn-by-default. |
| pub UNREACHABLE_PUB, |
| Allow, |
| "`pub` items not reachable from crate root" |
| } |
| |
| declare_lint_pass!( |
| /// Lint for items marked `pub` that aren't reachable from other crates. |
| UnreachablePub => [UNREACHABLE_PUB] |
| ); |
| |
| impl UnreachablePub { |
| fn perform_lint( |
| &self, |
| cx: &LateContext<'_>, |
| what: &str, |
| def_id: LocalDefId, |
| vis_span: Span, |
| exportable: bool, |
| ) { |
| let mut applicability = Applicability::MachineApplicable; |
| if cx.tcx.visibility(def_id).is_public() && !cx.effective_visibilities.is_reachable(def_id) |
| { |
| if vis_span.from_expansion() { |
| applicability = Applicability::MaybeIncorrect; |
| } |
| let def_span = cx.tcx.def_span(def_id); |
| cx.emit_span_lint(UNREACHABLE_PUB, def_span, BuiltinUnreachablePub { |
| what, |
| suggestion: (vis_span, applicability), |
| help: exportable, |
| }); |
| } |
| } |
| } |
| |
| impl<'tcx> LateLintPass<'tcx> for UnreachablePub { |
| fn check_item(&mut self, cx: &LateContext<'_>, item: &hir::Item<'_>) { |
| // Do not warn for fake `use` statements. |
| if let hir::ItemKind::Use(_, hir::UseKind::ListStem) = &item.kind { |
| return; |
| } |
| self.perform_lint(cx, "item", item.owner_id.def_id, item.vis_span, true); |
| } |
| |
| fn check_foreign_item(&mut self, cx: &LateContext<'_>, foreign_item: &hir::ForeignItem<'tcx>) { |
| self.perform_lint(cx, "item", foreign_item.owner_id.def_id, foreign_item.vis_span, true); |
| } |
| |
| fn check_field_def(&mut self, _cx: &LateContext<'_>, _field: &hir::FieldDef<'_>) { |
| // - If an ADT definition is reported then we don't need to check fields |
| // (as it would add unnecessary complexity to the source code, the struct |
| // definition is in the immediate proximity to give the "real" visibility). |
| // - If an ADT is not reported because it's not `pub` - we don't need to |
| // check fields. |
| // - If an ADT is not reported because it's reachable - we also don't need |
| // to check fields because then they are reachable by construction if they |
| // are pub. |
| // |
| // Therefore in no case we check the fields. |
| // |
| // cf. https://github.com/rust-lang/rust/pull/126013#issuecomment-2152839205 |
| // cf. https://github.com/rust-lang/rust/pull/126040#issuecomment-2152944506 |
| } |
| |
| fn check_impl_item(&mut self, cx: &LateContext<'_>, impl_item: &hir::ImplItem<'_>) { |
| // Only lint inherent impl items. |
| if cx.tcx.associated_item(impl_item.owner_id).trait_item_def_id.is_none() { |
| self.perform_lint(cx, "item", impl_item.owner_id.def_id, impl_item.vis_span, false); |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `type_alias_bounds` lint detects bounds in type aliases. |
| /// |
| /// ### Example |
| /// |
| /// ```rust |
| /// type SendVec<T: Send> = Vec<T>; |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Trait and lifetime bounds on generic parameters and in where clauses of |
| /// type aliases are not checked at usage sites of the type alias. Moreover, |
| /// they are not thoroughly checked for correctness at their definition site |
| /// either similar to the aliased type. |
| /// |
| /// This is a known limitation of the type checker that may be lifted in a |
| /// future edition. Permitting such bounds in light of this was unintentional. |
| /// |
| /// While these bounds may have secondary effects such as enabling the use of |
| /// "shorthand" associated type paths[^1] and affecting the default trait |
| /// object lifetime[^2] of trait object types passed to the type alias, this |
| /// should not have been allowed until the aforementioned restrictions of the |
| /// type checker have been lifted. |
| /// |
| /// Using such bounds is highly discouraged as they are actively misleading. |
| /// |
| /// [^1]: I.e., paths of the form `T::Assoc` where `T` is a type parameter |
| /// bounded by trait `Trait` which defines an associated type called `Assoc` |
| /// as opposed to a fully qualified path of the form `<T as Trait>::Assoc`. |
| /// [^2]: <https://doc.rust-lang.org/reference/lifetime-elision.html#default-trait-object-lifetimes> |
| TYPE_ALIAS_BOUNDS, |
| Warn, |
| "bounds in type aliases are not enforced" |
| } |
| |
| declare_lint_pass!(TypeAliasBounds => [TYPE_ALIAS_BOUNDS]); |
| |
| impl TypeAliasBounds { |
| pub(crate) fn affects_object_lifetime_defaults(pred: &hir::WherePredicate<'_>) -> bool { |
| // Bounds of the form `T: 'a` with `T` type param affect object lifetime defaults. |
| if let hir::WherePredicate::BoundPredicate(pred) = pred |
| && pred.bounds.iter().any(|bound| matches!(bound, hir::GenericBound::Outlives(_))) |
| && pred.bound_generic_params.is_empty() // indeed, even if absent from the RHS |
| && pred.bounded_ty.as_generic_param().is_some() |
| { |
| return true; |
| } |
| false |
| } |
| } |
| |
| impl<'tcx> LateLintPass<'tcx> for TypeAliasBounds { |
| fn check_item(&mut self, cx: &LateContext<'_>, item: &hir::Item<'_>) { |
| let hir::ItemKind::TyAlias(hir_ty, generics) = item.kind else { return }; |
| |
| // There must not be a where clause. |
| if generics.predicates.is_empty() { |
| return; |
| } |
| |
| // Bounds of lazy type aliases and TAITs are respected. |
| if cx.tcx.type_alias_is_lazy(item.owner_id) { |
| return; |
| } |
| |
| // FIXME(generic_const_exprs): Revisit this before stabilization. |
| // See also `tests/ui/const-generics/generic_const_exprs/type-alias-bounds.rs`. |
| let ty = cx.tcx.type_of(item.owner_id).instantiate_identity(); |
| if ty.has_type_flags(ty::TypeFlags::HAS_CT_PROJECTION) |
| && cx.tcx.features().generic_const_exprs |
| { |
| return; |
| } |
| |
| // NOTE(inherent_associated_types): While we currently do take some bounds in type |
| // aliases into consideration during IAT *selection*, we don't perform full use+def |
| // site wfchecking for such type aliases. Therefore TAB should still trigger. |
| // See also `tests/ui/associated-inherent-types/type-alias-bounds.rs`. |
| |
| let mut where_spans = Vec::new(); |
| let mut inline_spans = Vec::new(); |
| let mut inline_sugg = Vec::new(); |
| |
| for p in generics.predicates { |
| let span = p.span(); |
| if p.in_where_clause() { |
| where_spans.push(span); |
| } else { |
| for b in p.bounds() { |
| inline_spans.push(b.span()); |
| } |
| inline_sugg.push((span, String::new())); |
| } |
| } |
| |
| let mut ty = Some(hir_ty); |
| let enable_feat_help = cx.tcx.sess.is_nightly_build(); |
| |
| if let [.., label_sp] = *where_spans { |
| cx.emit_span_lint(TYPE_ALIAS_BOUNDS, where_spans, BuiltinTypeAliasBounds { |
| in_where_clause: true, |
| label: label_sp, |
| enable_feat_help, |
| suggestions: vec![(generics.where_clause_span, String::new())], |
| preds: generics.predicates, |
| ty: ty.take(), |
| }); |
| } |
| if let [.., label_sp] = *inline_spans { |
| cx.emit_span_lint(TYPE_ALIAS_BOUNDS, inline_spans, BuiltinTypeAliasBounds { |
| in_where_clause: false, |
| label: label_sp, |
| enable_feat_help, |
| suggestions: inline_sugg, |
| preds: generics.predicates, |
| ty, |
| }); |
| } |
| } |
| } |
| |
| pub(crate) struct ShorthandAssocTyCollector { |
| pub(crate) qselves: Vec<Span>, |
| } |
| |
| impl hir::intravisit::Visitor<'_> for ShorthandAssocTyCollector { |
| fn visit_qpath(&mut self, qpath: &hir::QPath<'_>, id: hir::HirId, _: Span) { |
| // Look for "type-parameter shorthand-associated-types". I.e., paths of the |
| // form `T::Assoc` with `T` type param. These are reliant on trait bounds. |
| if let hir::QPath::TypeRelative(qself, _) = qpath |
| && qself.as_generic_param().is_some() |
| { |
| self.qselves.push(qself.span); |
| } |
| hir::intravisit::walk_qpath(self, qpath, id) |
| } |
| } |
| |
| declare_lint! { |
| /// The `trivial_bounds` lint detects trait bounds that don't depend on |
| /// any type parameters. |
| /// |
| /// ### Example |
| /// |
| /// ```rust |
| /// #![feature(trivial_bounds)] |
| /// pub struct A where i32: Copy; |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Usually you would not write a trait bound that you know is always |
| /// true, or never true. However, when using macros, the macro may not |
| /// know whether or not the constraint would hold or not at the time when |
| /// generating the code. Currently, the compiler does not alert you if the |
| /// constraint is always true, and generates an error if it is never true. |
| /// The `trivial_bounds` feature changes this to be a warning in both |
| /// cases, giving macros more freedom and flexibility to generate code, |
| /// while still providing a signal when writing non-macro code that |
| /// something is amiss. |
| /// |
| /// See [RFC 2056] for more details. This feature is currently only |
| /// available on the nightly channel, see [tracking issue #48214]. |
| /// |
| /// [RFC 2056]: https://github.com/rust-lang/rfcs/blob/master/text/2056-allow-trivial-where-clause-constraints.md |
| /// [tracking issue #48214]: https://github.com/rust-lang/rust/issues/48214 |
| TRIVIAL_BOUNDS, |
| Warn, |
| "these bounds don't depend on an type parameters" |
| } |
| |
| declare_lint_pass!( |
| /// Lint for trait and lifetime bounds that don't depend on type parameters |
| /// which either do nothing, or stop the item from being used. |
| TrivialConstraints => [TRIVIAL_BOUNDS] |
| ); |
| |
| impl<'tcx> LateLintPass<'tcx> for TrivialConstraints { |
| fn check_item(&mut self, cx: &LateContext<'tcx>, item: &'tcx hir::Item<'tcx>) { |
| use rustc_middle::ty::ClauseKind; |
| |
| if cx.tcx.features().trivial_bounds { |
| let predicates = cx.tcx.predicates_of(item.owner_id); |
| for &(predicate, span) in predicates.predicates { |
| let predicate_kind_name = match predicate.kind().skip_binder() { |
| ClauseKind::Trait(..) => "trait", |
| ClauseKind::TypeOutlives(..) | |
| ClauseKind::RegionOutlives(..) => "lifetime", |
| |
| // `ConstArgHasType` is never global as `ct` is always a param |
| ClauseKind::ConstArgHasType(..) |
| // Ignore projections, as they can only be global |
| // if the trait bound is global |
| | ClauseKind::Projection(..) |
| // Ignore bounds that a user can't type |
| | ClauseKind::WellFormed(..) |
| // FIXME(generic_const_exprs): `ConstEvaluatable` can be written |
| | ClauseKind::ConstEvaluatable(..) => continue, |
| }; |
| if predicate.is_global() { |
| cx.emit_span_lint(TRIVIAL_BOUNDS, span, BuiltinTrivialBounds { |
| predicate_kind_name, |
| predicate, |
| }); |
| } |
| } |
| } |
| } |
| } |
| |
| declare_lint_pass!( |
| /// Does nothing as a lint pass, but registers some `Lint`s |
| /// which are used by other parts of the compiler. |
| SoftLints => [ |
| WHILE_TRUE, |
| NON_SHORTHAND_FIELD_PATTERNS, |
| UNSAFE_CODE, |
| MISSING_DOCS, |
| MISSING_COPY_IMPLEMENTATIONS, |
| MISSING_DEBUG_IMPLEMENTATIONS, |
| ANONYMOUS_PARAMETERS, |
| UNUSED_DOC_COMMENTS, |
| NO_MANGLE_CONST_ITEMS, |
| NO_MANGLE_GENERIC_ITEMS, |
| MUTABLE_TRANSMUTES, |
| UNSTABLE_FEATURES, |
| UNREACHABLE_PUB, |
| TYPE_ALIAS_BOUNDS, |
| TRIVIAL_BOUNDS |
| ] |
| ); |
| |
| declare_lint! { |
| /// The `ellipsis_inclusive_range_patterns` lint detects the [`...` range |
| /// pattern], which is deprecated. |
| /// |
| /// [`...` range pattern]: https://doc.rust-lang.org/reference/patterns.html#range-patterns |
| /// |
| /// ### Example |
| /// |
| /// ```rust,edition2018 |
| /// let x = 123; |
| /// match x { |
| /// 0...100 => {} |
| /// _ => {} |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// The `...` range pattern syntax was changed to `..=` to avoid potential |
| /// confusion with the [`..` range expression]. Use the new form instead. |
| /// |
| /// [`..` range expression]: https://doc.rust-lang.org/reference/expressions/range-expr.html |
| pub ELLIPSIS_INCLUSIVE_RANGE_PATTERNS, |
| Warn, |
| "`...` range patterns are deprecated", |
| @future_incompatible = FutureIncompatibleInfo { |
| reason: FutureIncompatibilityReason::EditionError(Edition::Edition2021), |
| reference: "<https://doc.rust-lang.org/nightly/edition-guide/rust-2021/warnings-promoted-to-error.html>", |
| }; |
| } |
| |
| #[derive(Default)] |
| pub struct EllipsisInclusiveRangePatterns { |
| /// If `Some(_)`, suppress all subsequent pattern |
| /// warnings for better diagnostics. |
| node_id: Option<ast::NodeId>, |
| } |
| |
| impl_lint_pass!(EllipsisInclusiveRangePatterns => [ELLIPSIS_INCLUSIVE_RANGE_PATTERNS]); |
| |
| impl EarlyLintPass for EllipsisInclusiveRangePatterns { |
| fn check_pat(&mut self, cx: &EarlyContext<'_>, pat: &ast::Pat) { |
| if self.node_id.is_some() { |
| // Don't recursively warn about patterns inside range endpoints. |
| return; |
| } |
| |
| use self::ast::PatKind; |
| use self::ast::RangeSyntax::DotDotDot; |
| |
| /// If `pat` is a `...` pattern, return the start and end of the range, as well as the span |
| /// corresponding to the ellipsis. |
| fn matches_ellipsis_pat(pat: &ast::Pat) -> Option<(Option<&Expr>, &Expr, Span)> { |
| match &pat.kind { |
| PatKind::Range( |
| a, |
| Some(b), |
| Spanned { span, node: RangeEnd::Included(DotDotDot) }, |
| ) => Some((a.as_deref(), b, *span)), |
| _ => None, |
| } |
| } |
| |
| let (parentheses, endpoints) = match &pat.kind { |
| PatKind::Ref(subpat, _) => (true, matches_ellipsis_pat(subpat)), |
| _ => (false, matches_ellipsis_pat(pat)), |
| }; |
| |
| if let Some((start, end, join)) = endpoints { |
| if parentheses { |
| self.node_id = Some(pat.id); |
| let end = expr_to_string(end); |
| let replace = match start { |
| Some(start) => format!("&({}..={})", expr_to_string(start), end), |
| None => format!("&(..={end})"), |
| }; |
| if join.edition() >= Edition::Edition2021 { |
| cx.sess().dcx().emit_err(BuiltinEllipsisInclusiveRangePatterns { |
| span: pat.span, |
| suggestion: pat.span, |
| replace, |
| }); |
| } else { |
| cx.emit_span_lint( |
| ELLIPSIS_INCLUSIVE_RANGE_PATTERNS, |
| pat.span, |
| BuiltinEllipsisInclusiveRangePatternsLint::Parenthesise { |
| suggestion: pat.span, |
| replace, |
| }, |
| ); |
| } |
| } else { |
| let replace = "..="; |
| if join.edition() >= Edition::Edition2021 { |
| cx.sess().dcx().emit_err(BuiltinEllipsisInclusiveRangePatterns { |
| span: pat.span, |
| suggestion: join, |
| replace: replace.to_string(), |
| }); |
| } else { |
| cx.emit_span_lint( |
| ELLIPSIS_INCLUSIVE_RANGE_PATTERNS, |
| join, |
| BuiltinEllipsisInclusiveRangePatternsLint::NonParenthesise { |
| suggestion: join, |
| }, |
| ); |
| } |
| }; |
| } |
| } |
| |
| fn check_pat_post(&mut self, _cx: &EarlyContext<'_>, pat: &ast::Pat) { |
| if let Some(node_id) = self.node_id { |
| if pat.id == node_id { |
| self.node_id = None |
| } |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `keyword_idents_2018` lint detects edition keywords being used as an |
| /// identifier. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,edition2015,compile_fail |
| /// #![deny(keyword_idents_2018)] |
| /// // edition 2015 |
| /// fn dyn() {} |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Rust [editions] allow the language to evolve without breaking |
| /// backwards compatibility. This lint catches code that uses new keywords |
| /// that are added to the language that are used as identifiers (such as a |
| /// variable name, function name, etc.). If you switch the compiler to a |
| /// new edition without updating the code, then it will fail to compile if |
| /// you are using a new keyword as an identifier. |
| /// |
| /// You can manually change the identifiers to a non-keyword, or use a |
| /// [raw identifier], for example `r#dyn`, to transition to a new edition. |
| /// |
| /// This lint solves the problem automatically. It is "allow" by default |
| /// because the code is perfectly valid in older editions. The [`cargo |
| /// fix`] tool with the `--edition` flag will switch this lint to "warn" |
| /// and automatically apply the suggested fix from the compiler (which is |
| /// to use a raw identifier). This provides a completely automated way to |
| /// update old code for a new edition. |
| /// |
| /// [editions]: https://doc.rust-lang.org/edition-guide/ |
| /// [raw identifier]: https://doc.rust-lang.org/reference/identifiers.html |
| /// [`cargo fix`]: https://doc.rust-lang.org/cargo/commands/cargo-fix.html |
| pub KEYWORD_IDENTS_2018, |
| Allow, |
| "detects edition keywords being used as an identifier", |
| @future_incompatible = FutureIncompatibleInfo { |
| reason: FutureIncompatibilityReason::EditionError(Edition::Edition2018), |
| reference: "issue #49716 <https://github.com/rust-lang/rust/issues/49716>", |
| }; |
| } |
| |
| declare_lint! { |
| /// The `keyword_idents_2024` lint detects edition keywords being used as an |
| /// identifier. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,edition2015,compile_fail |
| /// #![deny(keyword_idents_2024)] |
| /// // edition 2015 |
| /// fn gen() {} |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Rust [editions] allow the language to evolve without breaking |
| /// backwards compatibility. This lint catches code that uses new keywords |
| /// that are added to the language that are used as identifiers (such as a |
| /// variable name, function name, etc.). If you switch the compiler to a |
| /// new edition without updating the code, then it will fail to compile if |
| /// you are using a new keyword as an identifier. |
| /// |
| /// You can manually change the identifiers to a non-keyword, or use a |
| /// [raw identifier], for example `r#gen`, to transition to a new edition. |
| /// |
| /// This lint solves the problem automatically. It is "allow" by default |
| /// because the code is perfectly valid in older editions. The [`cargo |
| /// fix`] tool with the `--edition` flag will switch this lint to "warn" |
| /// and automatically apply the suggested fix from the compiler (which is |
| /// to use a raw identifier). This provides a completely automated way to |
| /// update old code for a new edition. |
| /// |
| /// [editions]: https://doc.rust-lang.org/edition-guide/ |
| /// [raw identifier]: https://doc.rust-lang.org/reference/identifiers.html |
| /// [`cargo fix`]: https://doc.rust-lang.org/cargo/commands/cargo-fix.html |
| pub KEYWORD_IDENTS_2024, |
| Allow, |
| "detects edition keywords being used as an identifier", |
| @future_incompatible = FutureIncompatibleInfo { |
| reason: FutureIncompatibilityReason::EditionError(Edition::Edition2024), |
| reference: "issue #49716 <https://github.com/rust-lang/rust/issues/49716>", |
| }; |
| } |
| |
| declare_lint_pass!( |
| /// Check for uses of edition keywords used as an identifier. |
| KeywordIdents => [KEYWORD_IDENTS_2018, KEYWORD_IDENTS_2024] |
| ); |
| |
| struct UnderMacro(bool); |
| |
| impl KeywordIdents { |
| fn check_tokens(&mut self, cx: &EarlyContext<'_>, tokens: &TokenStream) { |
| // Check if the preceding token is `$`, because we want to allow `$async`, etc. |
| let mut prev_dollar = false; |
| for tt in tokens.trees() { |
| match tt { |
| // Only report non-raw idents. |
| TokenTree::Token(token, _) => { |
| if let Some((ident, token::IdentIsRaw::No)) = token.ident() { |
| if !prev_dollar { |
| self.check_ident_token(cx, UnderMacro(true), ident, ""); |
| } |
| } else if let Some((ident, token::IdentIsRaw::No)) = token.lifetime() { |
| self.check_ident_token( |
| cx, |
| UnderMacro(true), |
| ident.without_first_quote(), |
| "'", |
| ); |
| } else if token.kind == TokenKind::Dollar { |
| prev_dollar = true; |
| continue; |
| } |
| } |
| TokenTree::Delimited(.., tts) => self.check_tokens(cx, tts), |
| } |
| prev_dollar = false; |
| } |
| } |
| |
| fn check_ident_token( |
| &mut self, |
| cx: &EarlyContext<'_>, |
| UnderMacro(under_macro): UnderMacro, |
| ident: Ident, |
| prefix: &'static str, |
| ) { |
| let (lint, edition) = match ident.name { |
| kw::Async | kw::Await | kw::Try => (KEYWORD_IDENTS_2018, Edition::Edition2018), |
| |
| // rust-lang/rust#56327: Conservatively do not |
| // attempt to report occurrences of `dyn` within |
| // macro definitions or invocations, because `dyn` |
| // can legitimately occur as a contextual keyword |
| // in 2015 code denoting its 2018 meaning, and we |
| // do not want rustfix to inject bugs into working |
| // code by rewriting such occurrences. |
| // |
| // But if we see `dyn` outside of a macro, we know |
| // its precise role in the parsed AST and thus are |
| // assured this is truly an attempt to use it as |
| // an identifier. |
| kw::Dyn if !under_macro => (KEYWORD_IDENTS_2018, Edition::Edition2018), |
| |
| kw::Gen => (KEYWORD_IDENTS_2024, Edition::Edition2024), |
| |
| _ => return, |
| }; |
| |
| // Don't lint `r#foo`. |
| if ident.span.edition() >= edition |
| || cx.sess().psess.raw_identifier_spans.contains(ident.span) |
| { |
| return; |
| } |
| |
| cx.emit_span_lint(lint, ident.span, BuiltinKeywordIdents { |
| kw: ident, |
| next: edition, |
| suggestion: ident.span, |
| prefix, |
| }); |
| } |
| } |
| |
| impl EarlyLintPass for KeywordIdents { |
| fn check_mac_def(&mut self, cx: &EarlyContext<'_>, mac_def: &ast::MacroDef) { |
| self.check_tokens(cx, &mac_def.body.tokens); |
| } |
| fn check_mac(&mut self, cx: &EarlyContext<'_>, mac: &ast::MacCall) { |
| self.check_tokens(cx, &mac.args.tokens); |
| } |
| fn check_ident(&mut self, cx: &EarlyContext<'_>, ident: Ident) { |
| if ident.name.as_str().starts_with('\'') { |
| self.check_ident_token(cx, UnderMacro(false), ident.without_first_quote(), "'"); |
| } else { |
| self.check_ident_token(cx, UnderMacro(false), ident, ""); |
| } |
| } |
| } |
| |
| declare_lint_pass!(ExplicitOutlivesRequirements => [EXPLICIT_OUTLIVES_REQUIREMENTS]); |
| |
| impl ExplicitOutlivesRequirements { |
| fn lifetimes_outliving_lifetime<'tcx>( |
| tcx: TyCtxt<'tcx>, |
| inferred_outlives: impl Iterator<Item = &'tcx (ty::Clause<'tcx>, Span)>, |
| item: LocalDefId, |
| lifetime: LocalDefId, |
| ) -> Vec<ty::Region<'tcx>> { |
| let item_generics = tcx.generics_of(item); |
| |
| inferred_outlives |
| .filter_map(|(clause, _)| match clause.kind().skip_binder() { |
| ty::ClauseKind::RegionOutlives(ty::OutlivesPredicate(a, b)) => match *a { |
| ty::ReEarlyParam(ebr) |
| if item_generics.region_param(ebr, tcx).def_id == lifetime.to_def_id() => |
| { |
| Some(b) |
| } |
| _ => None, |
| }, |
| _ => None, |
| }) |
| .collect() |
| } |
| |
| fn lifetimes_outliving_type<'tcx>( |
| inferred_outlives: impl Iterator<Item = &'tcx (ty::Clause<'tcx>, Span)>, |
| index: u32, |
| ) -> Vec<ty::Region<'tcx>> { |
| inferred_outlives |
| .filter_map(|(clause, _)| match clause.kind().skip_binder() { |
| ty::ClauseKind::TypeOutlives(ty::OutlivesPredicate(a, b)) => { |
| a.is_param(index).then_some(b) |
| } |
| _ => None, |
| }) |
| .collect() |
| } |
| |
| fn collect_outlives_bound_spans<'tcx>( |
| &self, |
| tcx: TyCtxt<'tcx>, |
| bounds: &hir::GenericBounds<'_>, |
| inferred_outlives: &[ty::Region<'tcx>], |
| predicate_span: Span, |
| item: DefId, |
| ) -> Vec<(usize, Span)> { |
| use rustc_middle::middle::resolve_bound_vars::ResolvedArg; |
| |
| let item_generics = tcx.generics_of(item); |
| |
| bounds |
| .iter() |
| .enumerate() |
| .filter_map(|(i, bound)| { |
| let hir::GenericBound::Outlives(lifetime) = bound else { |
| return None; |
| }; |
| |
| let is_inferred = match tcx.named_bound_var(lifetime.hir_id) { |
| Some(ResolvedArg::EarlyBound(def_id)) => inferred_outlives |
| .iter() |
| .any(|r| matches!(**r, ty::ReEarlyParam(ebr) if { item_generics.region_param(ebr, tcx).def_id == def_id.to_def_id() })), |
| _ => false, |
| }; |
| |
| if !is_inferred { |
| return None; |
| } |
| |
| let span = bound.span().find_ancestor_inside(predicate_span)?; |
| if in_external_macro(tcx.sess, span) { |
| return None; |
| } |
| |
| Some((i, span)) |
| }) |
| .collect() |
| } |
| |
| fn consolidate_outlives_bound_spans( |
| &self, |
| lo: Span, |
| bounds: &hir::GenericBounds<'_>, |
| bound_spans: Vec<(usize, Span)>, |
| ) -> Vec<Span> { |
| if bounds.is_empty() { |
| return Vec::new(); |
| } |
| if bound_spans.len() == bounds.len() { |
| let (_, last_bound_span) = bound_spans[bound_spans.len() - 1]; |
| // If all bounds are inferable, we want to delete the colon, so |
| // start from just after the parameter (span passed as argument) |
| vec![lo.to(last_bound_span)] |
| } else { |
| let mut merged = Vec::new(); |
| let mut last_merged_i = None; |
| |
| let mut from_start = true; |
| for (i, bound_span) in bound_spans { |
| match last_merged_i { |
| // If the first bound is inferable, our span should also eat the leading `+`. |
| None if i == 0 => { |
| merged.push(bound_span.to(bounds[1].span().shrink_to_lo())); |
| last_merged_i = Some(0); |
| } |
| // If consecutive bounds are inferable, merge their spans |
| Some(h) if i == h + 1 => { |
| if let Some(tail) = merged.last_mut() { |
| // Also eat the trailing `+` if the first |
| // more-than-one bound is inferable |
| let to_span = if from_start && i < bounds.len() { |
| bounds[i + 1].span().shrink_to_lo() |
| } else { |
| bound_span |
| }; |
| *tail = tail.to(to_span); |
| last_merged_i = Some(i); |
| } else { |
| bug!("another bound-span visited earlier"); |
| } |
| } |
| _ => { |
| // When we find a non-inferable bound, subsequent inferable bounds |
| // won't be consecutive from the start (and we'll eat the leading |
| // `+` rather than the trailing one) |
| from_start = false; |
| merged.push(bounds[i - 1].span().shrink_to_hi().to(bound_span)); |
| last_merged_i = Some(i); |
| } |
| } |
| } |
| merged |
| } |
| } |
| } |
| |
| impl<'tcx> LateLintPass<'tcx> for ExplicitOutlivesRequirements { |
| fn check_item(&mut self, cx: &LateContext<'tcx>, item: &'tcx hir::Item<'_>) { |
| use rustc_middle::middle::resolve_bound_vars::ResolvedArg; |
| |
| let def_id = item.owner_id.def_id; |
| if let hir::ItemKind::Struct(_, hir_generics) |
| | hir::ItemKind::Enum(_, hir_generics) |
| | hir::ItemKind::Union(_, hir_generics) = item.kind |
| { |
| let inferred_outlives = cx.tcx.inferred_outlives_of(def_id); |
| if inferred_outlives.is_empty() { |
| return; |
| } |
| |
| let ty_generics = cx.tcx.generics_of(def_id); |
| let num_where_predicates = hir_generics |
| .predicates |
| .iter() |
| .filter(|predicate| predicate.in_where_clause()) |
| .count(); |
| |
| let mut bound_count = 0; |
| let mut lint_spans = Vec::new(); |
| let mut where_lint_spans = Vec::new(); |
| let mut dropped_where_predicate_count = 0; |
| for (i, where_predicate) in hir_generics.predicates.iter().enumerate() { |
| let (relevant_lifetimes, bounds, predicate_span, in_where_clause) = |
| match where_predicate { |
| hir::WherePredicate::RegionPredicate(predicate) => { |
| if let Some(ResolvedArg::EarlyBound(region_def_id)) = |
| cx.tcx.named_bound_var(predicate.lifetime.hir_id) |
| { |
| ( |
| Self::lifetimes_outliving_lifetime( |
| cx.tcx, |
| // don't warn if the inferred span actually came from the predicate we're looking at |
| // this happens if the type is recursively defined |
| inferred_outlives |
| .iter() |
| .filter(|(_, span)| !predicate.span.contains(*span)), |
| item.owner_id.def_id, |
| region_def_id, |
| ), |
| &predicate.bounds, |
| predicate.span, |
| predicate.in_where_clause, |
| ) |
| } else { |
| continue; |
| } |
| } |
| hir::WherePredicate::BoundPredicate(predicate) => { |
| // FIXME we can also infer bounds on associated types, |
| // and should check for them here. |
| match predicate.bounded_ty.kind { |
| hir::TyKind::Path(hir::QPath::Resolved(None, path)) => { |
| let Res::Def(DefKind::TyParam, def_id) = path.res else { |
| continue; |
| }; |
| let index = ty_generics.param_def_id_to_index[&def_id]; |
| ( |
| Self::lifetimes_outliving_type( |
| // don't warn if the inferred span actually came from the predicate we're looking at |
| // this happens if the type is recursively defined |
| inferred_outlives.iter().filter(|(_, span)| { |
| !predicate.span.contains(*span) |
| }), |
| index, |
| ), |
| &predicate.bounds, |
| predicate.span, |
| predicate.origin == PredicateOrigin::WhereClause, |
| ) |
| } |
| _ => { |
| continue; |
| } |
| } |
| } |
| _ => continue, |
| }; |
| if relevant_lifetimes.is_empty() { |
| continue; |
| } |
| |
| let bound_spans = self.collect_outlives_bound_spans( |
| cx.tcx, |
| bounds, |
| &relevant_lifetimes, |
| predicate_span, |
| item.owner_id.to_def_id(), |
| ); |
| bound_count += bound_spans.len(); |
| |
| let drop_predicate = bound_spans.len() == bounds.len(); |
| if drop_predicate && in_where_clause { |
| dropped_where_predicate_count += 1; |
| } |
| |
| if drop_predicate { |
| if !in_where_clause { |
| lint_spans.push(predicate_span); |
| } else if predicate_span.from_expansion() { |
| // Don't try to extend the span if it comes from a macro expansion. |
| where_lint_spans.push(predicate_span); |
| } else if i + 1 < num_where_predicates { |
| // If all the bounds on a predicate were inferable and there are |
| // further predicates, we want to eat the trailing comma. |
| let next_predicate_span = hir_generics.predicates[i + 1].span(); |
| if next_predicate_span.from_expansion() { |
| where_lint_spans.push(predicate_span); |
| } else { |
| where_lint_spans |
| .push(predicate_span.to(next_predicate_span.shrink_to_lo())); |
| } |
| } else { |
| // Eat the optional trailing comma after the last predicate. |
| let where_span = hir_generics.where_clause_span; |
| if where_span.from_expansion() { |
| where_lint_spans.push(predicate_span); |
| } else { |
| where_lint_spans.push(predicate_span.to(where_span.shrink_to_hi())); |
| } |
| } |
| } else { |
| where_lint_spans.extend(self.consolidate_outlives_bound_spans( |
| predicate_span.shrink_to_lo(), |
| bounds, |
| bound_spans, |
| )); |
| } |
| } |
| |
| // If all predicates in where clause are inferable, drop the entire clause |
| // (including the `where`) |
| if hir_generics.has_where_clause_predicates |
| && dropped_where_predicate_count == num_where_predicates |
| { |
| let where_span = hir_generics.where_clause_span; |
| // Extend the where clause back to the closing `>` of the |
| // generics, except for tuple struct, which have the `where` |
| // after the fields of the struct. |
| let full_where_span = |
| if let hir::ItemKind::Struct(hir::VariantData::Tuple(..), _) = item.kind { |
| where_span |
| } else { |
| hir_generics.span.shrink_to_hi().to(where_span) |
| }; |
| |
| // Due to macro expansions, the `full_where_span` might not actually contain all |
| // predicates. |
| if where_lint_spans.iter().all(|&sp| full_where_span.contains(sp)) { |
| lint_spans.push(full_where_span); |
| } else { |
| lint_spans.extend(where_lint_spans); |
| } |
| } else { |
| lint_spans.extend(where_lint_spans); |
| } |
| |
| if !lint_spans.is_empty() { |
| // Do not automatically delete outlives requirements from macros. |
| let applicability = if lint_spans.iter().all(|sp| sp.can_be_used_for_suggestions()) |
| { |
| Applicability::MachineApplicable |
| } else { |
| Applicability::MaybeIncorrect |
| }; |
| |
| // Due to macros, there might be several predicates with the same span |
| // and we only want to suggest removing them once. |
| lint_spans.sort_unstable(); |
| lint_spans.dedup(); |
| |
| cx.emit_span_lint( |
| EXPLICIT_OUTLIVES_REQUIREMENTS, |
| lint_spans.clone(), |
| BuiltinExplicitOutlives { |
| count: bound_count, |
| suggestion: BuiltinExplicitOutlivesSuggestion { |
| spans: lint_spans, |
| applicability, |
| }, |
| }, |
| ); |
| } |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `incomplete_features` lint detects unstable features enabled with |
| /// the [`feature` attribute] that may function improperly in some or all |
| /// cases. |
| /// |
| /// [`feature` attribute]: https://doc.rust-lang.org/nightly/unstable-book/ |
| /// |
| /// ### Example |
| /// |
| /// ```rust |
| /// #![feature(generic_const_exprs)] |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Although it is encouraged for people to experiment with unstable |
| /// features, some of them are known to be incomplete or faulty. This lint |
| /// is a signal that the feature has not yet been finished, and you may |
| /// experience problems with it. |
| pub INCOMPLETE_FEATURES, |
| Warn, |
| "incomplete features that may function improperly in some or all cases" |
| } |
| |
| declare_lint! { |
| /// The `internal_features` lint detects unstable features enabled with |
| /// the [`feature` attribute] that are internal to the compiler or standard |
| /// library. |
| /// |
| /// [`feature` attribute]: https://doc.rust-lang.org/nightly/unstable-book/ |
| /// |
| /// ### Example |
| /// |
| /// ```rust |
| /// #![feature(rustc_attrs)] |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// These features are an implementation detail of the compiler and standard |
| /// library and are not supposed to be used in user code. |
| pub INTERNAL_FEATURES, |
| Warn, |
| "internal features are not supposed to be used" |
| } |
| |
| declare_lint_pass!( |
| /// Check for used feature gates in `INCOMPLETE_FEATURES` in `rustc_feature/src/unstable.rs`. |
| IncompleteInternalFeatures => [INCOMPLETE_FEATURES, INTERNAL_FEATURES] |
| ); |
| |
| impl EarlyLintPass for IncompleteInternalFeatures { |
| fn check_crate(&mut self, cx: &EarlyContext<'_>, _: &ast::Crate) { |
| let features = cx.builder.features(); |
| features |
| .declared_lang_features |
| .iter() |
| .map(|(name, span, _)| (name, span)) |
| .chain(features.declared_lib_features.iter().map(|(name, span)| (name, span))) |
| .filter(|(&name, _)| features.incomplete(name) || features.internal(name)) |
| .for_each(|(&name, &span)| { |
| if features.incomplete(name) { |
| let note = rustc_feature::find_feature_issue(name, GateIssue::Language) |
| .map(|n| BuiltinFeatureIssueNote { n }); |
| let help = |
| HAS_MIN_FEATURES.contains(&name).then_some(BuiltinIncompleteFeaturesHelp); |
| |
| cx.emit_span_lint(INCOMPLETE_FEATURES, span, BuiltinIncompleteFeatures { |
| name, |
| note, |
| help, |
| }); |
| } else { |
| cx.emit_span_lint(INTERNAL_FEATURES, span, BuiltinInternalFeatures { name }); |
| } |
| }); |
| } |
| } |
| |
| const HAS_MIN_FEATURES: &[Symbol] = &[sym::specialization]; |
| |
| declare_lint! { |
| /// The `invalid_value` lint detects creating a value that is not valid, |
| /// such as a null reference. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,no_run |
| /// # #![allow(unused)] |
| /// unsafe { |
| /// let x: &'static i32 = std::mem::zeroed(); |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// In some situations the compiler can detect that the code is creating |
| /// an invalid value, which should be avoided. |
| /// |
| /// In particular, this lint will check for improper use of |
| /// [`mem::zeroed`], [`mem::uninitialized`], [`mem::transmute`], and |
| /// [`MaybeUninit::assume_init`] that can cause [undefined behavior]. The |
| /// lint should provide extra information to indicate what the problem is |
| /// and a possible solution. |
| /// |
| /// [`mem::zeroed`]: https://doc.rust-lang.org/std/mem/fn.zeroed.html |
| /// [`mem::uninitialized`]: https://doc.rust-lang.org/std/mem/fn.uninitialized.html |
| /// [`mem::transmute`]: https://doc.rust-lang.org/std/mem/fn.transmute.html |
| /// [`MaybeUninit::assume_init`]: https://doc.rust-lang.org/std/mem/union.MaybeUninit.html#method.assume_init |
| /// [undefined behavior]: https://doc.rust-lang.org/reference/behavior-considered-undefined.html |
| pub INVALID_VALUE, |
| Warn, |
| "an invalid value is being created (such as a null reference)" |
| } |
| |
| declare_lint_pass!(InvalidValue => [INVALID_VALUE]); |
| |
| /// Information about why a type cannot be initialized this way. |
| pub struct InitError { |
| pub(crate) message: String, |
| /// Spans from struct fields and similar that can be obtained from just the type. |
| pub(crate) span: Option<Span>, |
| /// Used to report a trace through adts. |
| pub(crate) nested: Option<Box<InitError>>, |
| } |
| impl InitError { |
| fn spanned(self, span: Span) -> InitError { |
| Self { span: Some(span), ..self } |
| } |
| |
| fn nested(self, nested: impl Into<Option<InitError>>) -> InitError { |
| assert!(self.nested.is_none()); |
| Self { nested: nested.into().map(Box::new), ..self } |
| } |
| } |
| |
| impl<'a> From<&'a str> for InitError { |
| fn from(s: &'a str) -> Self { |
| s.to_owned().into() |
| } |
| } |
| impl From<String> for InitError { |
| fn from(message: String) -> Self { |
| Self { message, span: None, nested: None } |
| } |
| } |
| |
| impl<'tcx> LateLintPass<'tcx> for InvalidValue { |
| fn check_expr(&mut self, cx: &LateContext<'tcx>, expr: &hir::Expr<'_>) { |
| #[derive(Debug, Copy, Clone, PartialEq)] |
| enum InitKind { |
| Zeroed, |
| Uninit, |
| } |
| |
| /// Test if this constant is all-0. |
| fn is_zero(expr: &hir::Expr<'_>) -> bool { |
| use hir::ExprKind::*; |
| use rustc_ast::LitKind::*; |
| match &expr.kind { |
| Lit(lit) => { |
| if let Int(i, _) = lit.node { |
| i == 0 |
| } else { |
| false |
| } |
| } |
| Tup(tup) => tup.iter().all(is_zero), |
| _ => false, |
| } |
| } |
| |
| /// Determine if this expression is a "dangerous initialization". |
| fn is_dangerous_init(cx: &LateContext<'_>, expr: &hir::Expr<'_>) -> Option<InitKind> { |
| if let hir::ExprKind::Call(path_expr, args) = expr.kind { |
| // Find calls to `mem::{uninitialized,zeroed}` methods. |
| if let hir::ExprKind::Path(ref qpath) = path_expr.kind { |
| let def_id = cx.qpath_res(qpath, path_expr.hir_id).opt_def_id()?; |
| match cx.tcx.get_diagnostic_name(def_id) { |
| Some(sym::mem_zeroed) => return Some(InitKind::Zeroed), |
| Some(sym::mem_uninitialized) => return Some(InitKind::Uninit), |
| Some(sym::transmute) if is_zero(&args[0]) => return Some(InitKind::Zeroed), |
| _ => {} |
| } |
| } |
| } else if let hir::ExprKind::MethodCall(_, receiver, ..) = expr.kind { |
| // Find problematic calls to `MaybeUninit::assume_init`. |
| let def_id = cx.typeck_results().type_dependent_def_id(expr.hir_id)?; |
| if cx.tcx.is_diagnostic_item(sym::assume_init, def_id) { |
| // This is a call to *some* method named `assume_init`. |
| // See if the `self` parameter is one of the dangerous constructors. |
| if let hir::ExprKind::Call(path_expr, _) = receiver.kind { |
| if let hir::ExprKind::Path(ref qpath) = path_expr.kind { |
| let def_id = cx.qpath_res(qpath, path_expr.hir_id).opt_def_id()?; |
| match cx.tcx.get_diagnostic_name(def_id) { |
| Some(sym::maybe_uninit_zeroed) => return Some(InitKind::Zeroed), |
| Some(sym::maybe_uninit_uninit) => return Some(InitKind::Uninit), |
| _ => {} |
| } |
| } |
| } |
| } |
| } |
| |
| None |
| } |
| |
| fn variant_find_init_error<'tcx>( |
| cx: &LateContext<'tcx>, |
| ty: Ty<'tcx>, |
| variant: &VariantDef, |
| args: ty::GenericArgsRef<'tcx>, |
| descr: &str, |
| init: InitKind, |
| ) -> Option<InitError> { |
| let mut field_err = variant.fields.iter().find_map(|field| { |
| ty_find_init_error(cx, field.ty(cx.tcx, args), init).map(|mut err| { |
| if !field.did.is_local() { |
| err |
| } else if err.span.is_none() { |
| err.span = Some(cx.tcx.def_span(field.did)); |
| write!(&mut err.message, " (in this {descr})").unwrap(); |
| err |
| } else { |
| InitError::from(format!("in this {descr}")) |
| .spanned(cx.tcx.def_span(field.did)) |
| .nested(err) |
| } |
| }) |
| }); |
| |
| // Check if this ADT has a constrained layout (like `NonNull` and friends). |
| if let Ok(layout) = cx.tcx.layout_of(cx.param_env.and(ty)) { |
| if let Abi::Scalar(scalar) | Abi::ScalarPair(scalar, _) = &layout.abi { |
| let range = scalar.valid_range(cx); |
| let msg = if !range.contains(0) { |
| "must be non-null" |
| } else if init == InitKind::Uninit && !scalar.is_always_valid(cx) { |
| // Prefer reporting on the fields over the entire struct for uninit, |
| // as the information bubbles out and it may be unclear why the type can't |
| // be null from just its outside signature. |
| |
| "must be initialized inside its custom valid range" |
| } else { |
| return field_err; |
| }; |
| if let Some(field_err) = &mut field_err { |
| // Most of the time, if the field error is the same as the struct error, |
| // the struct error only happens because of the field error. |
| if field_err.message.contains(msg) { |
| field_err.message = format!("because {}", field_err.message); |
| } |
| } |
| return Some(InitError::from(format!("`{ty}` {msg}")).nested(field_err)); |
| } |
| } |
| field_err |
| } |
| |
| /// Return `Some` only if we are sure this type does *not* |
| /// allow zero initialization. |
| fn ty_find_init_error<'tcx>( |
| cx: &LateContext<'tcx>, |
| ty: Ty<'tcx>, |
| init: InitKind, |
| ) -> Option<InitError> { |
| let ty = cx.tcx.try_normalize_erasing_regions(cx.param_env, ty).unwrap_or(ty); |
| |
| use rustc_type_ir::TyKind::*; |
| match ty.kind() { |
| // Primitive types that don't like 0 as a value. |
| Ref(..) => Some("references must be non-null".into()), |
| Adt(..) if ty.is_box() => Some("`Box` must be non-null".into()), |
| FnPtr(..) => Some("function pointers must be non-null".into()), |
| Never => Some("the `!` type has no valid value".into()), |
| RawPtr(ty, _) if matches!(ty.kind(), Dynamic(..)) => |
| // raw ptr to dyn Trait |
| { |
| Some("the vtable of a wide raw pointer must be non-null".into()) |
| } |
| // Primitive types with other constraints. |
| Bool if init == InitKind::Uninit => { |
| Some("booleans must be either `true` or `false`".into()) |
| } |
| Char if init == InitKind::Uninit => { |
| Some("characters must be a valid Unicode codepoint".into()) |
| } |
| Int(_) | Uint(_) if init == InitKind::Uninit => { |
| Some("integers must be initialized".into()) |
| } |
| Float(_) if init == InitKind::Uninit => Some("floats must be initialized".into()), |
| RawPtr(_, _) if init == InitKind::Uninit => { |
| Some("raw pointers must be initialized".into()) |
| } |
| // Recurse and checks for some compound types. (but not unions) |
| Adt(adt_def, args) if !adt_def.is_union() => { |
| // Handle structs. |
| if adt_def.is_struct() { |
| return variant_find_init_error( |
| cx, |
| ty, |
| adt_def.non_enum_variant(), |
| args, |
| "struct field", |
| init, |
| ); |
| } |
| // And now, enums. |
| let span = cx.tcx.def_span(adt_def.did()); |
| let mut potential_variants = adt_def.variants().iter().filter_map(|variant| { |
| let definitely_inhabited = match variant |
| .inhabited_predicate(cx.tcx, *adt_def) |
| .instantiate(cx.tcx, args) |
| .apply_any_module(cx.tcx, cx.param_env) |
| { |
| // Entirely skip uninhabited variants. |
| Some(false) => return None, |
| // Forward the others, but remember which ones are definitely inhabited. |
| Some(true) => true, |
| None => false, |
| }; |
| Some((variant, definitely_inhabited)) |
| }); |
| let Some(first_variant) = potential_variants.next() else { |
| return Some( |
| InitError::from("enums with no inhabited variants have no valid value") |
| .spanned(span), |
| ); |
| }; |
| // So we have at least one potentially inhabited variant. Might we have two? |
| let Some(second_variant) = potential_variants.next() else { |
| // There is only one potentially inhabited variant. So we can recursively |
| // check that variant! |
| return variant_find_init_error( |
| cx, |
| ty, |
| first_variant.0, |
| args, |
| "field of the only potentially inhabited enum variant", |
| init, |
| ); |
| }; |
| // So we have at least two potentially inhabited variants. If we can prove that |
| // we have at least two *definitely* inhabited variants, then we have a tag and |
| // hence leaving this uninit is definitely disallowed. (Leaving it zeroed could |
| // be okay, depending on which variant is encoded as zero tag.) |
| if init == InitKind::Uninit { |
| let definitely_inhabited = (first_variant.1 as usize) |
| + (second_variant.1 as usize) |
| + potential_variants |
| .filter(|(_variant, definitely_inhabited)| *definitely_inhabited) |
| .count(); |
| if definitely_inhabited > 1 { |
| return Some(InitError::from( |
| "enums with multiple inhabited variants have to be initialized to a variant", |
| ).spanned(span)); |
| } |
| } |
| // We couldn't find anything wrong here. |
| None |
| } |
| Tuple(..) => { |
| // Proceed recursively, check all fields. |
| ty.tuple_fields().iter().find_map(|field| ty_find_init_error(cx, field, init)) |
| } |
| Array(ty, len) => { |
| if matches!(len.try_to_target_usize(cx.tcx), Some(v) if v > 0) { |
| // Array length known at array non-empty -- recurse. |
| ty_find_init_error(cx, *ty, init) |
| } else { |
| // Empty array or size unknown. |
| None |
| } |
| } |
| // Conservative fallback. |
| _ => None, |
| } |
| } |
| |
| if let Some(init) = is_dangerous_init(cx, expr) { |
| // This conjures an instance of a type out of nothing, |
| // using zeroed or uninitialized memory. |
| // We are extremely conservative with what we warn about. |
| let conjured_ty = cx.typeck_results().expr_ty(expr); |
| if let Some(err) = with_no_trimmed_paths!(ty_find_init_error(cx, conjured_ty, init)) { |
| let msg = match init { |
| InitKind::Zeroed => fluent::lint_builtin_unpermitted_type_init_zeroed, |
| InitKind::Uninit => fluent::lint_builtin_unpermitted_type_init_uninit, |
| }; |
| let sub = BuiltinUnpermittedTypeInitSub { err }; |
| cx.emit_span_lint(INVALID_VALUE, expr.span, BuiltinUnpermittedTypeInit { |
| msg, |
| ty: conjured_ty, |
| label: expr.span, |
| sub, |
| tcx: cx.tcx, |
| }); |
| } |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `deref_nullptr` lint detects when an null pointer is dereferenced, |
| /// which causes [undefined behavior]. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,no_run |
| /// # #![allow(unused)] |
| /// use std::ptr; |
| /// unsafe { |
| /// let x = &*ptr::null::<i32>(); |
| /// let x = ptr::addr_of!(*ptr::null::<i32>()); |
| /// let x = *(0 as *const i32); |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Dereferencing a null pointer causes [undefined behavior] even as a place expression, |
| /// like `&*(0 as *const i32)` or `addr_of!(*(0 as *const i32))`. |
| /// |
| /// [undefined behavior]: https://doc.rust-lang.org/reference/behavior-considered-undefined.html |
| pub DEREF_NULLPTR, |
| Warn, |
| "detects when an null pointer is dereferenced" |
| } |
| |
| declare_lint_pass!(DerefNullPtr => [DEREF_NULLPTR]); |
| |
| impl<'tcx> LateLintPass<'tcx> for DerefNullPtr { |
| fn check_expr(&mut self, cx: &LateContext<'tcx>, expr: &hir::Expr<'_>) { |
| /// test if expression is a null ptr |
| fn is_null_ptr(cx: &LateContext<'_>, expr: &hir::Expr<'_>) -> bool { |
| match &expr.kind { |
| rustc_hir::ExprKind::Cast(expr, ty) => { |
| if let rustc_hir::TyKind::Ptr(_) = ty.kind { |
| return is_zero(expr) || is_null_ptr(cx, expr); |
| } |
| } |
| // check for call to `core::ptr::null` or `core::ptr::null_mut` |
| rustc_hir::ExprKind::Call(path, _) => { |
| if let rustc_hir::ExprKind::Path(ref qpath) = path.kind { |
| if let Some(def_id) = cx.qpath_res(qpath, path.hir_id).opt_def_id() { |
| return matches!( |
| cx.tcx.get_diagnostic_name(def_id), |
| Some(sym::ptr_null | sym::ptr_null_mut) |
| ); |
| } |
| } |
| } |
| _ => {} |
| } |
| false |
| } |
| |
| /// test if expression is the literal `0` |
| fn is_zero(expr: &hir::Expr<'_>) -> bool { |
| match &expr.kind { |
| rustc_hir::ExprKind::Lit(lit) => { |
| if let LitKind::Int(a, _) = lit.node { |
| return a == 0; |
| } |
| } |
| _ => {} |
| } |
| false |
| } |
| |
| if let rustc_hir::ExprKind::Unary(rustc_hir::UnOp::Deref, expr_deref) = expr.kind { |
| if is_null_ptr(cx, expr_deref) { |
| cx.emit_span_lint(DEREF_NULLPTR, expr.span, BuiltinDerefNullptr { |
| label: expr.span, |
| }); |
| } |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `named_asm_labels` lint detects the use of named labels in the |
| /// inline `asm!` macro. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// # #![feature(asm_experimental_arch)] |
| /// use std::arch::asm; |
| /// |
| /// fn main() { |
| /// unsafe { |
| /// asm!("foo: bar"); |
| /// } |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// LLVM is allowed to duplicate inline assembly blocks for any |
| /// reason, for example when it is in a function that gets inlined. Because |
| /// of this, GNU assembler [local labels] *must* be used instead of labels |
| /// with a name. Using named labels might cause assembler or linker errors. |
| /// |
| /// See the explanation in [Rust By Example] for more details. |
| /// |
| /// [local labels]: https://sourceware.org/binutils/docs/as/Symbol-Names.html#Local-Labels |
| /// [Rust By Example]: https://doc.rust-lang.org/nightly/rust-by-example/unsafe/asm.html#labels |
| pub NAMED_ASM_LABELS, |
| Deny, |
| "named labels in inline assembly", |
| } |
| |
| declare_lint! { |
| /// The `binary_asm_labels` lint detects the use of numeric labels containing only binary |
| /// digits in the inline `asm!` macro. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,ignore (fails on non-x86_64) |
| /// #![cfg(target_arch = "x86_64")] |
| /// |
| /// use std::arch::asm; |
| /// |
| /// fn main() { |
| /// unsafe { |
| /// asm!("0: jmp 0b"); |
| /// } |
| /// } |
| /// ``` |
| /// |
| /// This will produce: |
| /// |
| /// ```text |
| /// error: avoid using labels containing only the digits `0` and `1` in inline assembly |
| /// --> <source>:7:15 |
| /// | |
| /// 7 | asm!("0: jmp 0b"); |
| /// | ^ use a different label that doesn't start with `0` or `1` |
| /// | |
| /// = help: start numbering with `2` instead |
| /// = note: an LLVM bug makes these labels ambiguous with a binary literal number on x86 |
| /// = note: see <https://github.com/llvm/llvm-project/issues/99547> for more information |
| /// = note: `#[deny(binary_asm_labels)]` on by default |
| /// ``` |
| /// |
| /// ### Explanation |
| /// |
| /// An [LLVM bug] causes this code to fail to compile because it interprets the `0b` as a binary |
| /// literal instead of a reference to the previous local label `0`. To work around this bug, |
| /// don't use labels that could be confused with a binary literal. |
| /// |
| /// This behavior is platform-specific to x86 and x86-64. |
| /// |
| /// See the explanation in [Rust By Example] for more details. |
| /// |
| /// [LLVM bug]: https://github.com/llvm/llvm-project/issues/99547 |
| /// [Rust By Example]: https://doc.rust-lang.org/nightly/rust-by-example/unsafe/asm.html#labels |
| pub BINARY_ASM_LABELS, |
| Deny, |
| "labels in inline assembly containing only 0 or 1 digits", |
| } |
| |
| declare_lint_pass!(AsmLabels => [NAMED_ASM_LABELS, BINARY_ASM_LABELS]); |
| |
| #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| enum AsmLabelKind { |
| Named, |
| FormatArg, |
| Binary, |
| } |
| |
| impl<'tcx> LateLintPass<'tcx> for AsmLabels { |
| fn check_expr(&mut self, cx: &LateContext<'tcx>, expr: &'tcx hir::Expr<'tcx>) { |
| if let hir::Expr { |
| kind: hir::ExprKind::InlineAsm(hir::InlineAsm { template_strs, options, .. }), |
| .. |
| } = expr |
| { |
| // asm with `options(raw)` does not do replacement with `{` and `}`. |
| let raw = options.contains(InlineAsmOptions::RAW); |
| |
| for (template_sym, template_snippet, template_span) in template_strs.iter() { |
| let template_str = template_sym.as_str(); |
| let find_label_span = |needle: &str| -> Option<Span> { |
| if let Some(template_snippet) = template_snippet { |
| let snippet = template_snippet.as_str(); |
| if let Some(pos) = snippet.find(needle) { |
| let end = pos |
| + snippet[pos..] |
| .find(|c| c == ':') |
| .unwrap_or(snippet[pos..].len() - 1); |
| let inner = InnerSpan::new(pos, end); |
| return Some(template_span.from_inner(inner)); |
| } |
| } |
| |
| None |
| }; |
| |
| // diagnostics are emitted per-template, so this is created here as opposed to the outer loop |
| let mut spans = Vec::new(); |
| |
| // A semicolon might not actually be specified as a separator for all targets, but |
| // it seems like LLVM accepts it always. |
| let statements = template_str.split(|c| matches!(c, '\n' | ';')); |
| for statement in statements { |
| // If there's a comment, trim it from the statement |
| let statement = statement.find("//").map_or(statement, |idx| &statement[..idx]); |
| |
| // In this loop, if there is ever a non-label, no labels can come after it. |
| let mut start_idx = 0; |
| 'label_loop: for (idx, _) in statement.match_indices(':') { |
| let possible_label = statement[start_idx..idx].trim(); |
| let mut chars = possible_label.chars(); |
| |
| let Some(start) = chars.next() else { |
| // Empty string means a leading ':' in this section, which is not a |
| // label. |
| break 'label_loop; |
| }; |
| |
| // Whether a { bracket has been seen and its } hasn't been found yet. |
| let mut in_bracket = false; |
| let mut label_kind = AsmLabelKind::Named; |
| |
| // A label can also start with a format arg, if it's not a raw asm block. |
| if !raw && start == '{' { |
| in_bracket = true; |
| label_kind = AsmLabelKind::FormatArg; |
| } else if matches!(start, '0' | '1') { |
| // Binary labels have only the characters `0` or `1`. |
| label_kind = AsmLabelKind::Binary; |
| } else if !(start.is_ascii_alphabetic() || matches!(start, '.' | '_')) { |
| // Named labels start with ASCII letters, `.` or `_`. |
| // anything else is not a label |
| break 'label_loop; |
| } |
| |
| for c in chars { |
| // Inside a template format arg, any character is permitted for the |
| // puproses of label detection because we assume that it can be |
| // replaced with some other valid label string later. `options(raw)` |
| // asm blocks cannot have format args, so they are excluded from this |
| // special case. |
| if !raw && in_bracket { |
| if c == '{' { |
| // Nested brackets are not allowed in format args, this cannot |
| // be a label. |
| break 'label_loop; |
| } |
| |
| if c == '}' { |
| // The end of the format arg. |
| in_bracket = false; |
| } |
| } else if !raw && c == '{' { |
| // Start of a format arg. |
| in_bracket = true; |
| label_kind = AsmLabelKind::FormatArg; |
| } else { |
| let can_continue = match label_kind { |
| // Format arg labels are considered to be named labels for the purposes |
| // of continuing outside of their {} pair. |
| AsmLabelKind::Named | AsmLabelKind::FormatArg => { |
| c.is_ascii_alphanumeric() || matches!(c, '_' | '$') |
| } |
| AsmLabelKind::Binary => matches!(c, '0' | '1'), |
| }; |
| |
| if !can_continue { |
| // The potential label had an invalid character inside it, it |
| // cannot be a label. |
| break 'label_loop; |
| } |
| } |
| } |
| |
| // If all characters passed the label checks, this is a label. |
| spans.push((find_label_span(possible_label), label_kind)); |
| start_idx = idx + 1; |
| } |
| } |
| |
| for (span, label_kind) in spans { |
| let missing_precise_span = span.is_none(); |
| let span = span.unwrap_or(*template_span); |
| match label_kind { |
| AsmLabelKind::Named => { |
| cx.emit_span_lint(NAMED_ASM_LABELS, span, InvalidAsmLabel::Named { |
| missing_precise_span, |
| }); |
| } |
| AsmLabelKind::FormatArg => { |
| cx.emit_span_lint(NAMED_ASM_LABELS, span, InvalidAsmLabel::FormatArg { |
| missing_precise_span, |
| }); |
| } |
| // the binary asm issue only occurs when using intel syntax on x86 targets |
| AsmLabelKind::Binary |
| if !options.contains(InlineAsmOptions::ATT_SYNTAX) |
| && matches!( |
| cx.tcx.sess.asm_arch, |
| Some(InlineAsmArch::X86 | InlineAsmArch::X86_64) | None |
| ) => |
| { |
| cx.emit_span_lint(BINARY_ASM_LABELS, span, InvalidAsmLabel::Binary { |
| missing_precise_span, |
| span, |
| }) |
| } |
| // No lint on anything other than x86 |
| AsmLabelKind::Binary => (), |
| }; |
| } |
| } |
| } |
| } |
| } |
| |
| declare_lint! { |
| /// The `special_module_name` lint detects module |
| /// declarations for files that have a special meaning. |
| /// |
| /// ### Example |
| /// |
| /// ```rust,compile_fail |
| /// mod lib; |
| /// |
| /// fn main() { |
| /// lib::run(); |
| /// } |
| /// ``` |
| /// |
| /// {{produces}} |
| /// |
| /// ### Explanation |
| /// |
| /// Cargo recognizes `lib.rs` and `main.rs` as the root of a |
| /// library or binary crate, so declaring them as modules |
| /// will lead to miscompilation of the crate unless configured |
| /// explicitly. |
| /// |
| /// To access a library from a binary target within the same crate, |
| /// use `your_crate_name::` as the path instead of `lib::`: |
| /// |
| /// ```rust,compile_fail |
| /// // bar/src/lib.rs |
| /// fn run() { |
| /// // ... |
| /// } |
| /// |
| /// // bar/src/main.rs |
| /// fn main() { |
| /// bar::run(); |
| /// } |
| /// ``` |
| /// |
| /// Binary targets cannot be used as libraries and so declaring |
| /// one as a module is not allowed. |
| pub SPECIAL_MODULE_NAME, |
| Warn, |
| "module declarations for files with a special meaning", |
| } |
| |
| declare_lint_pass!(SpecialModuleName => [SPECIAL_MODULE_NAME]); |
| |
| impl EarlyLintPass for SpecialModuleName { |
| fn check_crate(&mut self, cx: &EarlyContext<'_>, krate: &ast::Crate) { |
| for item in &krate.items { |
| if let ast::ItemKind::Mod( |
| _, |
| ast::ModKind::Unloaded | ast::ModKind::Loaded(_, ast::Inline::No, _), |
| ) = item.kind |
| { |
| if item.attrs.iter().any(|a| a.has_name(sym::path)) { |
| continue; |
| } |
| |
| match item.ident.name.as_str() { |
| "lib" => cx.emit_span_lint( |
| SPECIAL_MODULE_NAME, |
| item.span, |
| BuiltinSpecialModuleNameUsed::Lib, |
| ), |
| "main" => cx.emit_span_lint( |
| SPECIAL_MODULE_NAME, |
| item.span, |
| BuiltinSpecialModuleNameUsed::Main, |
| ), |
| _ => continue, |
| } |
| } |
| } |
| } |
| } |