Auto merge of rust-lang#18022 - Veykril:asm-parse, r=Veykril

feat: IDE support for `asm!` expressions

Fixes rust-lang/rust-analyzer#10461, Fixes rust-lang/rust-analyzer#6031 Progresses rust-lang/rust-analyzer#11621

Notably this only works for asm expressions not items yet. Most IDE features work, mainly completions need extra logic still.

Author: bors

src/tools/rust-analyzer/.typos.toml

@@ -15,6 +15,7 @@ extend-ignore-re = [
     '"flate2"',
     "raison d'être",
     "inout",
+    "INOUT",
     "optin"
 ]
 

src/tools/rust-analyzer/Cargo.lock

@@ -2624,6 +2624,7 @@ version = "0.1.0"
 dependencies = [
  "anyhow",
  "directories",
+ "either",
  "flate2",
  "itertools",
  "proc-macro2",

src/tools/rust-analyzer/Cargo.toml

@@ -185,6 +185,7 @@ style = { level = "warn", priority = -1 }
 suspicious = { level = "warn", priority = -1 }
 
 ## allow following lints
+too_long_first_doc_paragraph = "allow"
 # subjective
 single_match = "allow"
 # () makes a fine error in most cases

src/tools/rust-analyzer/crates/hir-def/src/body.rs

@@ -100,7 +100,14 @@ pub struct BodySourceMap {
     field_map_back: FxHashMap<ExprId, FieldSource>,
     pat_field_map_back: FxHashMap<PatId, PatFieldSource>,
 
-    format_args_template_map: FxHashMap<ExprId, Vec<(syntax::TextRange, Name)>>,
+    template_map: Option<
+        Box<(
+            // format_args!
+            FxHashMap<ExprId, Vec<(syntax::TextRange, Name)>>,
+            // asm!
+            FxHashMap<ExprId, Vec<(syntax::TextRange, usize)>>,
+        )>,
+    >,
 
     expansions: FxHashMap<InFile<AstPtr<ast::MacroCall>>, MacroFileId>,
 
@@ -426,7 +433,16 @@ impl BodySourceMap {
         node: InFile<&ast::FormatArgsExpr>,
     ) -> Option<&[(syntax::TextRange, Name)]> {
         let src = node.map(AstPtr::new).map(AstPtr::upcast::<ast::Expr>);
-        self.format_args_template_map.get(self.expr_map.get(&src)?).map(std::ops::Deref::deref)
+        self.template_map.as_ref()?.0.get(self.expr_map.get(&src)?).map(std::ops::Deref::deref)
+    }
+
+    pub fn asm_template_args(
+        &self,
+        node: InFile<&ast::AsmExpr>,
+    ) -> Option<(ExprId, &[(syntax::TextRange, usize)])> {
+        let src = node.map(AstPtr::new).map(AstPtr::upcast::<ast::Expr>);
+        let expr = self.expr_map.get(&src)?;
+        Some(*expr).zip(self.template_map.as_ref()?.1.get(expr).map(std::ops::Deref::deref))
     }
 
     /// Get a reference to the body source map's diagnostics.
@@ -446,11 +462,14 @@ impl BodySourceMap {
             field_map_back,
             pat_field_map_back,
             expansions,
-            format_args_template_map,
+            template_map,
             diagnostics,
             binding_definitions,
         } = self;
-        format_args_template_map.shrink_to_fit();
+        if let Some(template_map) = template_map {
+            template_map.0.shrink_to_fit();
+            template_map.1.shrink_to_fit();
+        }
         expr_map.shrink_to_fit();
         expr_map_back.shrink_to_fit();
         pat_map.shrink_to_fit();
@@ -463,4 +482,13 @@ impl BodySourceMap {
         diagnostics.shrink_to_fit();
         binding_definitions.shrink_to_fit();
     }
+
+    pub fn template_map(
+        &self,
+    ) -> Option<&(
+        FxHashMap<Idx<Expr>, Vec<(tt::TextRange, Name)>>,
+        FxHashMap<Idx<Expr>, Vec<(tt::TextRange, usize)>>,
+    )> {
+        self.template_map.as_deref()
+    }
 }

src/tools/rust-analyzer/crates/hir-def/src/body/lower.rs

@@ -1,6 +1,8 @@
 //! Transforms `ast::Expr` into an equivalent `hir_def::expr::Expr`
 //! representation.
 
+mod asm;
+
 use std::mem;
 
 use base_db::CrateId;
@@ -35,8 +37,8 @@ use crate::{
             FormatPlaceholder, FormatSign, FormatTrait,
         },
         Array, Binding, BindingAnnotation, BindingId, BindingProblems, CaptureBy, ClosureKind,
-        Expr, ExprId, InlineAsm, Label, LabelId, Literal, LiteralOrConst, MatchArm, Movability,
-        OffsetOf, Pat, PatId, RecordFieldPat, RecordLitField, Statement,
+        Expr, ExprId, Label, LabelId, Literal, LiteralOrConst, MatchArm, Movability, OffsetOf, Pat,
+        PatId, RecordFieldPat, RecordLitField, Statement,
     },
     item_scope::BuiltinShadowMode,
     lang_item::LangItem,
@@ -693,10 +695,7 @@ impl ExprCollector<'_> {
                 }
             }
             ast::Expr::UnderscoreExpr(_) => self.alloc_expr(Expr::Underscore, syntax_ptr),
-            ast::Expr::AsmExpr(e) => {
-                let e = self.collect_expr_opt(e.expr());
-                self.alloc_expr(Expr::InlineAsm(InlineAsm { e }), syntax_ptr)
-            }
+            ast::Expr::AsmExpr(e) => self.lower_inline_asm(e, syntax_ptr),
             ast::Expr::OffsetOfExpr(e) => {
                 let container = Interned::new(TypeRef::from_ast_opt(&self.ctx(), e.ty()));
                 let fields = e.fields().map(|it| it.as_name()).collect();
@@ -1848,7 +1847,7 @@ impl ExprCollector<'_> {
             },
             syntax_ptr,
         );
-        self.source_map.format_args_template_map.insert(idx, mappings);
+        self.source_map.template_map.get_or_insert_with(Default::default).0.insert(idx, mappings);
         idx
     }
 
@@ -2061,6 +2060,7 @@ impl ExprCollector<'_> {
             is_assignee_expr: false,
         })
     }
+
     // endregion: format
 
     fn lang_path(&self, lang: LangItem) -> Option<Path> {

src/tools/rust-analyzer/crates/hir-def/src/body/lower/asm.rs

@@ -0,0 +1,254 @@
+//! Lowering of inline assembly.
+use hir_expand::name::Name;
+use intern::Symbol;
+use rustc_hash::{FxHashMap, FxHashSet};
+use syntax::{
+    ast::{self, HasName, IsString},
+    AstNode, AstPtr, AstToken, T,
+};
+use tt::{TextRange, TextSize};
+
+use crate::{
+    body::lower::{ExprCollector, FxIndexSet},
+    hir::{AsmOperand, AsmOptions, Expr, ExprId, InlineAsm, InlineAsmRegOrRegClass},
+};
+
+impl ExprCollector<'_> {
+    pub(super) fn lower_inline_asm(
+        &mut self,
+        asm: ast::AsmExpr,
+        syntax_ptr: AstPtr<ast::Expr>,
+    ) -> ExprId {
+        let mut clobber_abis = FxIndexSet::default();
+        let mut operands = vec![];
+        let mut options = AsmOptions::empty();
+
+        let mut named_pos: FxHashMap<usize, Symbol> = Default::default();
+        let mut named_args: FxHashMap<Symbol, usize> = Default::default();
+        let mut reg_args: FxHashSet<usize> = Default::default();
+        for piece in asm.asm_pieces() {
+            let slot = operands.len();
+            let mut lower_reg = |reg: Option<ast::AsmRegSpec>| {
+                let reg = reg?;
+                if let Some(string) = reg.string_token() {
+                    reg_args.insert(slot);
+                    Some(InlineAsmRegOrRegClass::Reg(Symbol::intern(string.text())))
+                } else {
+                    reg.name_ref().map(|name_ref| {
+                        InlineAsmRegOrRegClass::RegClass(Symbol::intern(&name_ref.text()))
+                    })
+                }
+            };
+
+            let op = match piece {
+                ast::AsmPiece::AsmClobberAbi(clobber_abi) => {
+                    if let Some(abi_name) = clobber_abi.string_token() {
+                        clobber_abis.insert(Symbol::intern(abi_name.text()));
+                    }
+                    continue;
+                }
+                ast::AsmPiece::AsmOptions(opt) => {
+                    opt.asm_options().for_each(|opt| {
+                        options |= match opt.syntax().first_token().map_or(T![$], |it| it.kind()) {
+                            T![att_syntax] => AsmOptions::ATT_SYNTAX,
+                            T![may_unwind] => AsmOptions::MAY_UNWIND,
+                            T![nomem] => AsmOptions::NOMEM,
+                            T![noreturn] => AsmOptions::NORETURN,
+                            T![nostack] => AsmOptions::NOSTACK,
+                            T![preserves_flags] => AsmOptions::PRESERVES_FLAGS,
+                            T![pure] => AsmOptions::PURE,
+                            T![raw] => AsmOptions::RAW,
+                            T![readonly] => AsmOptions::READONLY,
+                            _ => return,
+                        }
+                    });
+                    continue;
+                }
+                ast::AsmPiece::AsmOperandNamed(op) => {
+                    let name = op.name().map(|name| Symbol::intern(&name.text()));
+                    if let Some(name) = &name {
+                        named_args.insert(name.clone(), slot);
+                        named_pos.insert(slot, name.clone());
+                    }
+                    let Some(op) = op.asm_operand() else { continue };
+                    (
+                        name.map(Name::new_symbol_root),
+                        match op {
+                            ast::AsmOperand::AsmRegOperand(op) => {
+                                let Some(dir_spec) = op.asm_dir_spec() else {
+                                    continue;
+                                };
+                                let Some(reg) = lower_reg(op.asm_reg_spec()) else {
+                                    continue;
+                                };
+                                if dir_spec.in_token().is_some() {
+                                    let expr = self.collect_expr_opt(
+                                        op.asm_operand_expr().and_then(|it| it.in_expr()),
+                                    );
+                                    AsmOperand::In { reg, expr }
+                                } else if dir_spec.out_token().is_some() {
+                                    let expr = self.collect_expr_opt(
+                                        op.asm_operand_expr().and_then(|it| it.in_expr()),
+                                    );
+                                    AsmOperand::Out { reg, expr: Some(expr), late: false }
+                                } else if dir_spec.lateout_token().is_some() {
+                                    let expr = self.collect_expr_opt(
+                                        op.asm_operand_expr().and_then(|it| it.in_expr()),
+                                    );
+                                    AsmOperand::Out { reg, expr: Some(expr), late: true }
+                                } else if dir_spec.inout_token().is_some() {
+                                    let Some(op_expr) = op.asm_operand_expr() else { continue };
+                                    let in_expr = self.collect_expr_opt(op_expr.in_expr());
+                                    let out_expr =
+                                        op_expr.out_expr().map(|it| self.collect_expr(it));
+                                    match out_expr {
+                                        Some(out_expr) => AsmOperand::SplitInOut {
+                                            reg,
+                                            in_expr,
+                                            out_expr: Some(out_expr),
+                                            late: false,
+                                        },
+                                        None => {
+                                            AsmOperand::InOut { reg, expr: in_expr, late: false }
+                                        }
+                                    }
+                                } else if dir_spec.inlateout_token().is_some() {
+                                    let Some(op_expr) = op.asm_operand_expr() else { continue };
+                                    let in_expr = self.collect_expr_opt(op_expr.in_expr());
+                                    let out_expr =
+                                        op_expr.out_expr().map(|it| self.collect_expr(it));
+                                    match out_expr {
+                                        Some(out_expr) => AsmOperand::SplitInOut {
+                                            reg,
+                                            in_expr,
+                                            out_expr: Some(out_expr),
+                                            late: false,
+                                        },
+                                        None => {
+                                            AsmOperand::InOut { reg, expr: in_expr, late: false }
+                                        }
+                                    }
+                                } else {
+                                    continue;
+                                }
+                            }
+                            ast::AsmOperand::AsmLabel(l) => {
+                                AsmOperand::Label(self.collect_block_opt(l.block_expr()))
+                            }
+                            ast::AsmOperand::AsmConst(c) => {
+                                AsmOperand::Const(self.collect_expr_opt(c.expr()))
+                            }
+                            ast::AsmOperand::AsmSym(s) => {
+                                let Some(path) =
+                                    s.path().and_then(|p| self.expander.parse_path(self.db, p))
+                                else {
+                                    continue;
+                                };
+                                AsmOperand::Sym(path)
+                            }
+                        },
+                    )
+                }
+            };
+            operands.push(op);
+        }
+
+        let mut mappings = vec![];
+        let mut curarg = 0;
+        if !options.contains(AsmOptions::RAW) {
+            // Don't treat raw asm as a format string.
+            asm.template()
+                .filter_map(|it| Some((it.clone(), self.expand_macros_to_string(it)?)))
+                .for_each(|(expr, (s, is_direct_literal))| {
+                    let Ok(text) = s.value() else {
+                        return;
+                    };
+                    let template_snippet = match expr {
+                        ast::Expr::Literal(literal) => match literal.kind() {
+                            ast::LiteralKind::String(s) => Some(s.text().to_owned()),
+                            _ => None,
+                        },
+                        _ => None,
+                    };
+                    let str_style = match s.quote_offsets() {
+                        Some(offsets) => {
+                            let raw = usize::from(offsets.quotes.0.len()) - 1;
+                            // subtract 1 for the `r` prefix
+                            (raw != 0).then(|| raw - 1)
+                        }
+                        None => None,
+                    };
+
+                    let mut parser = rustc_parse_format::Parser::new(
+                        &text,
+                        str_style,
+                        template_snippet,
+                        false,
+                        rustc_parse_format::ParseMode::InlineAsm,
+                    );
+                    parser.curarg = curarg;
+
+                    let mut unverified_pieces = Vec::new();
+                    while let Some(piece) = parser.next() {
+                        if !parser.errors.is_empty() {
+                            break;
+                        } else {
+                            unverified_pieces.push(piece);
+                        }
+                    }
+
+                    curarg = parser.curarg;
+
+                    let to_span = |inner_span: rustc_parse_format::InnerSpan| {
+                        is_direct_literal.then(|| {
+                            TextRange::new(
+                                inner_span.start.try_into().unwrap(),
+                                inner_span.end.try_into().unwrap(),
+                            ) - TextSize::from(str_style.map(|it| it + 1).unwrap_or(0) as u32 + 1)
+                        })
+                    };
+                    for piece in unverified_pieces {
+                        match piece {
+                            rustc_parse_format::Piece::String(_) => {}
+                            rustc_parse_format::Piece::NextArgument(arg) => {
+                                // let span = arg_spans.next();
+
+                                let (operand_idx, _name) = match arg.position {
+                                    rustc_parse_format::ArgumentIs(idx)
+                                    | rustc_parse_format::ArgumentImplicitlyIs(idx) => {
+                                        if idx >= operands.len()
+                                            || named_pos.contains_key(&idx)
+                                            || reg_args.contains(&idx)
+                                        {
+                                            (None, None)
+                                        } else {
+                                            (Some(idx), None)
+                                        }
+                                    }
+                                    rustc_parse_format::ArgumentNamed(name) => {
+                                        let name = Symbol::intern(name);
+                                        (
+                                            named_args.get(&name).copied(),
+                                            Some(Name::new_symbol_root(name)),
+                                        )
+                                    }
+                                };
+
+                                if let Some(operand_idx) = operand_idx {
+                                    if let Some(position_span) = to_span(arg.position_span) {
+                                        mappings.push((position_span, operand_idx));
+                                    }
+                                }
+                            }
+                        }
+                    }
+                })
+        };
+        let idx = self.alloc_expr(
+            Expr::InlineAsm(InlineAsm { operands: operands.into_boxed_slice(), options }),
+            syntax_ptr,
+        );
+        self.source_map.template_map.get_or_insert_with(Default::default).1.insert(idx, mappings);
+        idx
+    }
+}