Auto merge of #102570 - cjgillot:deagg-debuginfo, r=oli-obk

Perform simple scalar replacement of aggregates (SROA) MIR opt

This is a re-open of #85796

I copied the debuginfo implementation (first commit) from `@eddyb's` own SROA PR.

This pass replaces plain field accesses by simple locals when possible.
To be eligible, the replaced locals:
- must not be enums or unions;
- must not be used whole;
- must not have their address taken.

The storage and deinit statements are duplicated on each created local.

cc `@tmiasko` who reviewed the former version of this PR.

Author: bors

compiler/rustc_codegen_gcc/src/debuginfo.rs

@@ -4,16 +4,25 @@ use rustc_codegen_ssa::traits::{DebugInfoBuilderMethods, DebugInfoMethods};
 use rustc_middle::mir;
 use rustc_middle::ty::{Instance, PolyExistentialTraitRef, Ty};
 use rustc_span::{SourceFile, Span, Symbol};
-use rustc_target::abi::Size;
 use rustc_target::abi::call::FnAbi;
+use rustc_target::abi::Size;
+use std::ops::Range;
 
 use crate::builder::Builder;
 use crate::context::CodegenCx;
 
 impl<'a, 'gcc, 'tcx> DebugInfoBuilderMethods for Builder<'a, 'gcc, 'tcx> {
     // FIXME(eddyb) find a common convention for all of the debuginfo-related
     // names (choose between `dbg`, `debug`, `debuginfo`, `debug_info` etc.).
-    fn dbg_var_addr(&mut self, _dbg_var: Self::DIVariable, _scope_metadata: Self::DIScope, _variable_alloca: Self::Value, _direct_offset: Size, _indirect_offsets: &[Size]) {
+    fn dbg_var_addr(
+        &mut self,
+        _dbg_var: Self::DIVariable,
+        _scope_metadata: Self::DIScope,
+        _variable_alloca: Self::Value,
+        _direct_offset: Size,
+        _indirect_offsets: &[Size],
+        _fragment: Option<Range<Size>>,
+    ) {
         unimplemented!();
     }
 
@@ -31,32 +40,64 @@ impl<'a, 'gcc, 'tcx> DebugInfoBuilderMethods for Builder<'a, 'gcc, 'tcx> {
 }
 
 impl<'gcc, 'tcx> DebugInfoMethods<'tcx> for CodegenCx<'gcc, 'tcx> {
-    fn create_vtable_debuginfo(&self, _ty: Ty<'tcx>, _trait_ref: Option<PolyExistentialTraitRef<'tcx>>, _vtable: Self::Value) {
+    fn create_vtable_debuginfo(
+        &self,
+        _ty: Ty<'tcx>,
+        _trait_ref: Option<PolyExistentialTraitRef<'tcx>>,
+        _vtable: Self::Value,
+    ) {
         // TODO(antoyo)
     }
 
-    fn create_function_debug_context(&self, _instance: Instance<'tcx>, _fn_abi: &FnAbi<'tcx, Ty<'tcx>>, _llfn: RValue<'gcc>, _mir: &mir::Body<'tcx>) -> Option<FunctionDebugContext<Self::DIScope, Self::DILocation>> {
+    fn create_function_debug_context(
+        &self,
+        _instance: Instance<'tcx>,
+        _fn_abi: &FnAbi<'tcx, Ty<'tcx>>,
+        _llfn: RValue<'gcc>,
+        _mir: &mir::Body<'tcx>,
+    ) -> Option<FunctionDebugContext<Self::DIScope, Self::DILocation>> {
         // TODO(antoyo)
         None
     }
 
-    fn extend_scope_to_file(&self, _scope_metadata: Self::DIScope, _file: &SourceFile) -> Self::DIScope {
+    fn extend_scope_to_file(
+        &self,
+        _scope_metadata: Self::DIScope,
+        _file: &SourceFile,
+    ) -> Self::DIScope {
         unimplemented!();
     }
 
     fn debuginfo_finalize(&self) {
         // TODO(antoyo)
     }
 
-    fn create_dbg_var(&self, _variable_name: Symbol, _variable_type: Ty<'tcx>, _scope_metadata: Self::DIScope, _variable_kind: VariableKind, _span: Span) -> Self::DIVariable {
+    fn create_dbg_var(
+        &self,
+        _variable_name: Symbol,
+        _variable_type: Ty<'tcx>,
+        _scope_metadata: Self::DIScope,
+        _variable_kind: VariableKind,
+        _span: Span,
+    ) -> Self::DIVariable {
         unimplemented!();
     }
 
-    fn dbg_scope_fn(&self, _instance: Instance<'tcx>, _fn_abi: &FnAbi<'tcx, Ty<'tcx>>, _maybe_definition_llfn: Option<RValue<'gcc>>) -> Self::DIScope {
+    fn dbg_scope_fn(
+        &self,
+        _instance: Instance<'tcx>,
+        _fn_abi: &FnAbi<'tcx, Ty<'tcx>>,
+        _maybe_definition_llfn: Option<RValue<'gcc>>,
+    ) -> Self::DIScope {
         unimplemented!();
     }
 
-    fn dbg_loc(&self, _scope: Self::DIScope, _inlined_at: Option<Self::DILocation>, _span: Span) -> Self::DILocation {
+    fn dbg_loc(
+        &self,
+        _scope: Self::DIScope,
+        _inlined_at: Option<Self::DILocation>,
+        _span: Span,
+    ) -> Self::DILocation {
         unimplemented!();
     }
 }

compiler/rustc_codegen_llvm/src/debuginfo/mod.rs

@@ -39,6 +39,7 @@ use smallvec::SmallVec;
 use std::cell::OnceCell;
 use std::cell::RefCell;
 use std::iter;
+use std::ops::Range;
 
 mod create_scope_map;
 pub mod gdb;
@@ -163,12 +164,14 @@ impl<'ll> DebugInfoBuilderMethods for Builder<'_, 'll, '_> {
         variable_alloca: Self::Value,
         direct_offset: Size,
         indirect_offsets: &[Size],
+        fragment: Option<Range<Size>>,
     ) {
-        // Convert the direct and indirect offsets to address ops.
+        // Convert the direct and indirect offsets and fragment byte range to address ops.
         // FIXME(eddyb) use `const`s instead of getting the values via FFI,
         // the values should match the ones in the DWARF standard anyway.
         let op_deref = || unsafe { llvm::LLVMRustDIBuilderCreateOpDeref() };
         let op_plus_uconst = || unsafe { llvm::LLVMRustDIBuilderCreateOpPlusUconst() };
+        let op_llvm_fragment = || unsafe { llvm::LLVMRustDIBuilderCreateOpLLVMFragment() };
         let mut addr_ops = SmallVec::<[u64; 8]>::new();
 
         if direct_offset.bytes() > 0 {
@@ -182,6 +185,13 @@ impl<'ll> DebugInfoBuilderMethods for Builder<'_, 'll, '_> {
                 addr_ops.push(offset.bytes() as u64);
             }
         }
+        if let Some(fragment) = fragment {
+            // `DW_OP_LLVM_fragment` takes as arguments the fragment's
+            // offset and size, both of them in bits.
+            addr_ops.push(op_llvm_fragment());
+            addr_ops.push(fragment.start.bits() as u64);
+            addr_ops.push((fragment.end - fragment.start).bits() as u64);
+        }
 
         unsafe {
             // FIXME(eddyb) replace `llvm.dbg.declare` with `llvm.dbg.addr`.

compiler/rustc_codegen_llvm/src/llvm/ffi.rs

@@ -2210,6 +2210,7 @@ extern "C" {
     ) -> &'a DILocation;
     pub fn LLVMRustDIBuilderCreateOpDeref() -> u64;
     pub fn LLVMRustDIBuilderCreateOpPlusUconst() -> u64;
+    pub fn LLVMRustDIBuilderCreateOpLLVMFragment() -> u64;
 
     #[allow(improper_ctypes)]
     pub fn LLVMRustWriteTypeToString(Type: &Type, s: &RustString);

compiler/rustc_codegen_ssa/src/mir/debuginfo.rs

@@ -14,6 +14,8 @@ use super::operand::{OperandRef, OperandValue};
 use super::place::PlaceRef;
 use super::{FunctionCx, LocalRef};
 
+use std::ops::Range;
+
 pub struct FunctionDebugContext<S, L> {
     pub scopes: IndexVec<mir::SourceScope, DebugScope<S, L>>,
 }
@@ -25,14 +27,18 @@ pub enum VariableKind {
 }
 
 /// Like `mir::VarDebugInfo`, but within a `mir::Local`.
-#[derive(Copy, Clone)]
+#[derive(Clone)]
 pub struct PerLocalVarDebugInfo<'tcx, D> {
     pub name: Symbol,
     pub source_info: mir::SourceInfo,
 
     /// `DIVariable` returned by `create_dbg_var`.
     pub dbg_var: Option<D>,
 
+    /// Byte range in the `dbg_var` covered by this fragment,
+    /// if this is a fragment of a composite `VarDebugInfo`.
+    pub fragment: Option<Range<Size>>,
+
     /// `.place.projection` from `mir::VarDebugInfo`.
     pub projection: &'tcx ty::List<mir::PlaceElem<'tcx>>,
 }
@@ -145,7 +151,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
             Some(per_local) => &per_local[local],
             None => return,
         };
-        let whole_local_var = vars.iter().find(|var| var.projection.is_empty()).copied();
+        let whole_local_var = vars.iter().find(|var| var.projection.is_empty()).cloned();
         let has_proj = || vars.iter().any(|var| !var.projection.is_empty());
 
         let fallback_var = if self.mir.local_kind(local) == mir::LocalKind::Arg {
@@ -187,6 +193,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                     name,
                     source_info: decl.source_info,
                     dbg_var,
+                    fragment: None,
                     projection: ty::List::empty(),
                 })
             }
@@ -199,7 +206,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
         let name = if bx.sess().fewer_names() {
             None
         } else {
-            Some(match whole_local_var.or(fallback_var) {
+            Some(match whole_local_var.or(fallback_var.clone()) {
                 Some(var) if var.name != kw::Empty => var.name.to_string(),
                 _ => format!("{:?}", local),
             })
@@ -249,7 +256,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
             LocalRef::UnsizedPlace(_) => return,
         };
 
-        let vars = vars.iter().copied().chain(fallback_var);
+        let vars = vars.iter().cloned().chain(fallback_var);
 
         for var in vars {
             let Some(dbg_var) = var.dbg_var else { continue };
@@ -312,9 +319,16 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                 bx.store(place.llval, alloca.llval, alloca.align);
 
                 // Point the debug info to `*alloca` for the current variable
-                bx.dbg_var_addr(dbg_var, dbg_loc, alloca.llval, Size::ZERO, &[Size::ZERO]);
+                bx.dbg_var_addr(dbg_var, dbg_loc, alloca.llval, Size::ZERO, &[Size::ZERO], None);
             } else {
-                bx.dbg_var_addr(dbg_var, dbg_loc, base.llval, direct_offset, &indirect_offsets);
+                bx.dbg_var_addr(
+                    dbg_var,
+                    dbg_loc,
+                    base.llval,
+                    direct_offset,
+                    &indirect_offsets,
+                    None,
+                );
             }
         }
     }
@@ -382,6 +396,10 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                         let ty = self.monomorphize(c.ty());
                         (ty, VariableKind::LocalVariable)
                     }
+                    mir::VarDebugInfoContents::Composite { ty, fragments: _ } => {
+                        let ty = self.monomorphize(ty);
+                        (ty, VariableKind::LocalVariable)
+                    }
                 };
 
                 self.cx.create_dbg_var(var.name, var_ty, dbg_scope, var_kind, span)
@@ -393,6 +411,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                         name: var.name,
                         source_info: var.source_info,
                         dbg_var,
+                        fragment: None,
                         projection: place.projection,
                     });
                 }
@@ -407,10 +426,48 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                                 bx,
                             );
 
-                            bx.dbg_var_addr(dbg_var, dbg_loc, base.llval, Size::ZERO, &[]);
+                            bx.dbg_var_addr(dbg_var, dbg_loc, base.llval, Size::ZERO, &[], None);
                         }
                     }
                 }
+                mir::VarDebugInfoContents::Composite { ty, ref fragments } => {
+                    let var_ty = self.monomorphize(ty);
+                    let var_layout = self.cx.layout_of(var_ty);
+                    for fragment in fragments {
+                        let mut fragment_start = Size::ZERO;
+                        let mut fragment_layout = var_layout;
+
+                        for elem in &fragment.projection {
+                            match *elem {
+                                mir::ProjectionElem::Field(field, _) => {
+                                    let i = field.index();
+                                    fragment_start += fragment_layout.fields.offset(i);
+                                    fragment_layout = fragment_layout.field(self.cx, i);
+                                }
+                                _ => span_bug!(
+                                    var.source_info.span,
+                                    "unsupported fragment projection `{:?}`",
+                                    elem,
+                                ),
+                            }
+                        }
+
+                        let place = fragment.contents;
+                        per_local[place.local].push(PerLocalVarDebugInfo {
+                            name: var.name,
+                            source_info: var.source_info,
+                            dbg_var,
+                            fragment: if fragment_layout.size == var_layout.size {
+                                // Fragment covers entire variable, so as far as
+                                // DWARF is concerned, it's not really a fragment.
+                                None
+                            } else {
+                                Some(fragment_start..fragment_start + fragment_layout.size)
+                            },
+                            projection: place.projection,
+                        });
+                    }
+                }
             }
         }
         Some(per_local)

compiler/rustc_codegen_ssa/src/traits/debuginfo.rs

@@ -6,6 +6,8 @@ use rustc_span::{SourceFile, Span, Symbol};
 use rustc_target::abi::call::FnAbi;
 use rustc_target::abi::Size;
 
+use std::ops::Range;
+
 pub trait DebugInfoMethods<'tcx>: BackendTypes {
     fn create_vtable_debuginfo(
         &self,
@@ -72,6 +74,9 @@ pub trait DebugInfoBuilderMethods: BackendTypes {
         direct_offset: Size,
         // NB: each offset implies a deref (i.e. they're steps in a pointer chain).
         indirect_offsets: &[Size],
+        // Byte range in the `dbg_var` covered by this fragment,
+        // if this is a fragment of a composite `DIVariable`.
+        fragment: Option<Range<Size>>,
     );
     fn set_dbg_loc(&mut self, dbg_loc: Self::DILocation);
     fn insert_reference_to_gdb_debug_scripts_section_global(&mut self);

compiler/rustc_llvm/llvm-wrapper/RustWrapper.cpp

@@ -1111,6 +1111,10 @@ extern "C" uint64_t LLVMRustDIBuilderCreateOpPlusUconst() {
   return dwarf::DW_OP_plus_uconst;
 }
 
+extern "C" int64_t LLVMRustDIBuilderCreateOpLLVMFragment() {
+  return dwarf::DW_OP_LLVM_fragment;
+}
+
 extern "C" void LLVMRustWriteTypeToString(LLVMTypeRef Ty, RustStringRef Str) {
   RawRustStringOstream OS(Str);
   unwrap<llvm::Type>(Ty)->print(OS);

compiler/rustc_middle/src/mir/mod.rs

@@ -1071,14 +1071,67 @@ pub enum VarDebugInfoContents<'tcx> {
     /// based on a `Local`, not a `Static`, and contains no indexing.
     Place(Place<'tcx>),
     Const(Constant<'tcx>),
+    /// The user variable's data is split across several fragments,
+    /// each described by a `VarDebugInfoFragment`.
+    /// See DWARF 5's "2.6.1.2 Composite Location Descriptions"
+    /// and LLVM's `DW_OP_LLVM_fragment` for more details on
+    /// the underlying debuginfo feature this relies on.
+    Composite {
+        /// Type of the original user variable.
+        ty: Ty<'tcx>,
+        /// All the parts of the original user variable, which ended
+        /// up in disjoint places, due to optimizations.
+        fragments: Vec<VarDebugInfoFragment<'tcx>>,
+    },
 }
 
 impl<'tcx> Debug for VarDebugInfoContents<'tcx> {
     fn fmt(&self, fmt: &mut Formatter<'_>) -> fmt::Result {
         match self {
             VarDebugInfoContents::Const(c) => write!(fmt, "{}", c),
             VarDebugInfoContents::Place(p) => write!(fmt, "{:?}", p),
+            VarDebugInfoContents::Composite { ty, fragments } => {
+                write!(fmt, "{:?}{{ ", ty)?;
+                for f in fragments.iter() {
+                    write!(fmt, "{:?}, ", f)?;
+                }
+                write!(fmt, "}}")
+            }
+        }
+    }
+}
+
+#[derive(Clone, TyEncodable, TyDecodable, HashStable, TypeFoldable, TypeVisitable)]
+pub struct VarDebugInfoFragment<'tcx> {
+    /// Where in the composite user variable this fragment is,
+    /// represented as a "projection" into the composite variable.
+    /// At lower levels, this corresponds to a byte/bit range.
+    // NOTE(eddyb) there's an unenforced invariant that this contains
+    // only `Field`s, and not into `enum` variants or `union`s.
+    // FIXME(eddyb) support this for `enum`s by either using DWARF's
+    // more advanced control-flow features (unsupported by LLVM?)
+    // to match on the discriminant, or by using custom type debuginfo
+    // with non-overlapping variants for the composite variable.
+    pub projection: Vec<PlaceElem<'tcx>>,
+
+    /// Where the data for this fragment can be found.
+    // NOTE(eddyb) There's an unenforced invariant that this `Place` is
+    // contains no indexing (with a non-constant index).
+    pub contents: Place<'tcx>,
+}
+
+impl Debug for VarDebugInfoFragment<'_> {
+    fn fmt(&self, fmt: &mut Formatter<'_>) -> fmt::Result {
+        for elem in self.projection.iter() {
+            match elem {
+                ProjectionElem::Field(field, _) => {
+                    write!(fmt, ".{:?}", field.index())?;
+                }
+                _ => bug!("unsupported fragment projection `{:?}`", elem),
+            }
         }
+
+        write!(fmt, " => {:?}", self.contents)
     }
 }