mir-opt: Eliminate dead statements even if they are used by debuginfos

dianqk · dianqk · commit 458ad936b720 · 2025-08-10T22:48:15.000+08:00
diff --git a/compiler/rustc_codegen_ssa/src/mir/statement.rs b/compiler/rustc_codegen_ssa/src/mir/statement.rs
@@ -127,34 +127,42 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                     // Drop unsupported projections.
                     // FIXME: Add a test case.
                     place.projection.iter().all(|p| p.can_use_in_debuginfo()) &&
-                    // Only pointers can calculate addresses.
-                    bx.type_kind(bx.val_ty(place_ref.val.llval)) == TypeKind::Pointer
+                        // Only pointers can calculate addresses.
+                        bx.type_kind(bx.val_ty(place_ref.val.llval)) == TypeKind::Pointer
+                });
+                let assign_ref = match (place_ref, place.is_indirect_first_projection()) {
+                    (Some(place_ref), false) => {
+                        Some((place_ref.val, place_ref.layout, place.projection.as_slice()))
+                    }
+                    (Some(place_ref), true) => {
+                        let projected_ty = place_ref
+                            .layout
+                            .ty
+                            .builtin_deref(true)
+                            .unwrap_or_else(|| bug!("deref of non-pointer {:?}", place_ref));
+                        let layout = bx.cx().layout_of(projected_ty);
+                        Some((place_ref.val, layout, &place.projection[1..]))
+                    }
+                    _ => None,
+                };
+                let (val, layout, projection) = assign_ref.unwrap_or_else(|| {
+                    // If the address cannot be computed, use poison to indicate that the value has been optimized out.
+                    let ty = self.monomorphize(self.mir.local_decls[*dest].ty);
+                    let layout = bx.cx().layout_of(ty);
+                    let to_backend_ty = bx.cx().immediate_backend_type(layout);
+                    let place_ref =
+                        PlaceRef::new_sized(bx.cx().const_poison(to_backend_ty), layout);
+                    (place_ref.val, layout, [].as_slice())
                 });
-                let (val, layout, projection) =
-                    match (place_ref, place.is_indirect_first_projection()) {
-                        (Some(place_ref), false) => {
-                            (place_ref.val, place_ref.layout, place.projection.as_slice())
-                        }
-                        (Some(place_ref), true) => {
-                            let projected_ty =
-                                place_ref.layout.ty.builtin_deref(true).unwrap_or_else(|| {
-                                    bug!("deref of non-pointer {:?}", place_ref)
-                                });
-                            let layout = bx.cx().layout_of(projected_ty);
-                            (place_ref.val, layout, &place.projection[1..])
-                        }
-                        _ => {
-                            // If the address cannot be computed, use poison to indicate that the value has been optimized out.
-                            let ty = self.monomorphize(self.mir.local_decls[*dest].ty);
-                            let layout = bx.cx().layout_of(ty);
-                            let to_backend_ty = bx.cx().immediate_backend_type(layout);
-                            let place_ref =
-                                PlaceRef::new_sized(bx.cx().const_poison(to_backend_ty), layout);
-                            (place_ref.val, layout, [].as_slice())
-                        }
-                    };
                 self.debug_new_value_to_local(bx, *dest, val, layout, projection);
             }
+            StmtDebugInfo::InvalidAssign(local) => {
+                let ty = self.monomorphize(self.mir.local_decls[*local].ty);
+                let layout = bx.cx().layout_of(ty);
+                let to_backend_ty = bx.cx().immediate_backend_type(layout);
+                let place_ref = PlaceRef::new_sized(bx.cx().const_poison(to_backend_ty), layout);
+                self.debug_new_value_to_local(bx, *local, place_ref.val, layout, &[]);
+            }
         }
     }
 
diff --git a/compiler/rustc_middle/src/mir/pretty.rs b/compiler/rustc_middle/src/mir/pretty.rs
@@ -866,6 +866,9 @@ impl Debug for StmtDebugInfo<'_> {
             StmtDebugInfo::AssignRef(local, place) => {
                 write!(fmt, "{local:?} = &{place:?}")
             }
+            StmtDebugInfo::InvalidAssign(local) => {
+                write!(fmt, "{local:?} = &?")
+            }
         }
     }
 }
diff --git a/compiler/rustc_middle/src/mir/statement.rs b/compiler/rustc_middle/src/mir/statement.rs
@@ -498,6 +498,21 @@ impl<'tcx> PlaceRef<'tcx> {
         })
     }
 
+    /// Return the place accessed locals that include the base local.
+    pub fn accessed_locals(self) -> impl Iterator<Item = Local> {
+        std::iter::once(self.local).chain(self.projection.iter().filter_map(|proj| match proj {
+            ProjectionElem::Index(local) => Some(*local),
+            ProjectionElem::Deref
+            | ProjectionElem::Field(_, _)
+            | ProjectionElem::ConstantIndex { .. }
+            | ProjectionElem::Subslice { .. }
+            | ProjectionElem::Downcast(_, _)
+            | ProjectionElem::OpaqueCast(_)
+            | ProjectionElem::UnwrapUnsafeBinder(_)
+            | ProjectionElem::Subtype(_) => None,
+        }))
+    }
+
     /// Generates a new place by appending `more_projections` to the existing ones
     /// and interning the result.
     pub fn project_deeper(
@@ -1029,4 +1044,5 @@ impl<'tcx> ops::DerefMut for StmtDebugInfos<'tcx> {
 #[derive(Clone, TyEncodable, TyDecodable, HashStable, TypeFoldable, TypeVisitable)]
 pub enum StmtDebugInfo<'tcx> {
     AssignRef(Local, Place<'tcx>),
+    InvalidAssign(Local),
 }
diff --git a/compiler/rustc_middle/src/mir/visit.rs b/compiler/rustc_middle/src/mir/visit.rs
@@ -406,6 +406,13 @@ macro_rules! make_mir_visitor {
                             location
                         );
                     },
+                    StmtDebugInfo::InvalidAssign(local) => {
+                        self.visit_local(
+                            $(& $mutability)? *local,
+                            PlaceContext::NonUse(NonUseContext::VarDebugInfo),
+                            location
+                        );
+                    }
                 }
             }
 
diff --git a/compiler/rustc_mir_dataflow/src/debuginfo.rs b/compiler/rustc_mir_dataflow/src/debuginfo.rs
@@ -5,16 +5,16 @@ use rustc_middle::mir::*;
 /// Return the set of locals that appear in debuginfo.
 pub fn debuginfo_locals(body: &Body<'_>) -> DenseBitSet<Local> {
     let mut visitor = DebuginfoLocals(DenseBitSet::new_empty(body.local_decls.len()));
-    visitor.visit_body(body);
+    for debuginfo in body.var_debug_info.iter() {
+        visitor.visit_var_debug_info(debuginfo);
+    }
     visitor.0
 }
 
 struct DebuginfoLocals(DenseBitSet<Local>);
 
 impl Visitor<'_> for DebuginfoLocals {
-    fn visit_local(&mut self, local: Local, place_context: PlaceContext, _: Location) {
-        if place_context == PlaceContext::NonUse(NonUseContext::VarDebugInfo) {
-            self.0.insert(local);
-        }
+    fn visit_local(&mut self, local: Local, _: PlaceContext, _: Location) {
+        self.0.insert(local);
     }
 }
diff --git a/compiler/rustc_mir_dataflow/src/impls/liveness.rs b/compiler/rustc_mir_dataflow/src/impls/liveness.rs
@@ -282,9 +282,6 @@ impl<'a, 'tcx> Analysis<'tcx> for MaybeTransitiveLiveLocals<'a> {
         if let Some(destination) =
             Self::can_be_removed_if_dead(&statement.kind, &self.always_live, &self.debuginfo_locals)
             && !state.contains(destination.local)
-            // FIXME: We can eliminate the statement, but we'll need the statements it depends on
-            // for debuginfos. We need a way to handle this.
-            && !self.debuginfo_locals.contains(destination.local)
         {
             // This store is dead
             return;
diff --git a/compiler/rustc_mir_transform/src/simplify.rs b/compiler/rustc_mir_transform/src/simplify.rs
@@ -663,6 +663,22 @@ impl<'tcx> MutVisitor<'tcx> for LocalUpdater<'tcx> {
         self.tcx
     }
 
+    fn visit_statement_debuginfo(
+        &mut self,
+        stmt_debuginfo: &mut StmtDebugInfo<'tcx>,
+        location: Location,
+    ) {
+        match stmt_debuginfo {
+            StmtDebugInfo::AssignRef(local, place) => {
+                if place.as_ref().accessed_locals().any(|local| self.map[local].is_none()) {
+                    *stmt_debuginfo = StmtDebugInfo::InvalidAssign(*local);
+                }
+            }
+            StmtDebugInfo::InvalidAssign(_) => {}
+        }
+        self.super_statement_debuginfo(stmt_debuginfo, location);
+    }
+
     fn visit_local(&mut self, l: &mut Local, _: PlaceContext, _: Location) {
         *l = self.map[*l].unwrap();
     }
diff --git a/compiler/rustc_mir_transform/src/strip_debuginfo.rs b/compiler/rustc_mir_transform/src/strip_debuginfo.rs
@@ -1,5 +1,6 @@
 use rustc_middle::mir::*;
 use rustc_middle::ty::TyCtxt;
+use rustc_mir_dataflow::debuginfo::debuginfo_locals;
 use rustc_session::config::MirStripDebugInfo;
 
 /// Conditionally remove some of the VarDebugInfo in MIR.
@@ -30,6 +31,22 @@ impl<'tcx> crate::MirPass<'tcx> for StripDebugInfo {
                     if place.local.as_usize() <= body.arg_count && place.local != RETURN_PLACE,
             )
         });
+
+        let debuginfo_locals = debuginfo_locals(body);
+        for data in body.basic_blocks.as_mut_preserves_cfg() {
+            for stmt in data.statements.iter_mut() {
+                stmt.debuginfos.retain(|debuginfo| match debuginfo {
+                    StmtDebugInfo::AssignRef(local, _) | StmtDebugInfo::InvalidAssign(local) => {
+                        debuginfo_locals.contains(*local)
+                    }
+                });
+            }
+            data.after_last_stmt_debuginfos.retain(|debuginfo| match debuginfo {
+                StmtDebugInfo::AssignRef(local, _) | StmtDebugInfo::InvalidAssign(local) => {
+                    debuginfo_locals.contains(*local)
+                }
+            });
+        }
     }
 
     fn is_required(&self) -> bool {
diff --git a/tests/codegen-llvm/debuginfo-dse.rs b/tests/codegen-llvm/debuginfo-dse.rs
@@ -1,4 +1,4 @@
-//@ compile-flags: -Copt-level=3 -g -Zverify-llvm-ir
+//@ compile-flags: -Copt-level=3 -g -Zverify-llvm-ir -Zmerge-functions=disabled
 //@ revisions: CODEGEN OPTIMIZED
 //@[CODEGEN] compile-flags: -Cno-prepopulate-passes
 // ignore-tidy-linelength
@@ -9,6 +9,13 @@
 #[derive(Clone, Copy)]
 pub struct Foo(i32, i64, i32);
 
+#[repr(C)]
+pub struct Bar<'a> {
+    a: i32,
+    b: i64,
+    foo: &'a Foo,
+}
+
 #[no_mangle]
 fn r#ref(ref_foo: &Foo) -> i32 {
     // CHECK-LABEL: define {{.*}} i32 @ref
@@ -74,11 +81,20 @@ pub fn fragment(fragment_v1: Foo, mut fragment_v2: Foo) -> Foo {
     fragment_v2
 }
 
+#[no_mangle]
+pub fn deref(bar: Bar) -> i32 {
+    // CHECK-LABEL: define {{.*}} i32 @deref
+    // We are unable to represent dereference within this expression.
+    // CHECK: #dbg_value(ptr poison, [[VAR_deref_dead:![0-9]+]], !DIExpression()
+    let deref_dead = &bar.foo.2;
+    bar.a
+}
+
 #[no_mangle]
 pub fn tuple(foo: (i32, &Foo)) -> i32 {
     // CHECK-LABEL: define {{.*}} i32 @tuple
-    // CHECK-SAME: (i32 {{.*}}, ptr {{.*}} [[ARG_tuple_foo_1:%.*]])
-    // CHECK: #dbg_value(ptr [[ARG_tuple_foo_1]], [[VAR_tuple_dead:![0-9]+]], !DIExpression(DW_OP_plus_uconst, 16, DW_OP_stack_value)
+    // Although there is no dereference here, there is a dereference in the MIR.
+    // CHECK: #dbg_value(ptr poison, [[VAR_tuple_dead:![0-9]+]], !DIExpression()
     let tuple_dead = &foo.1.2;
     foo.1.0
 }
diff --git a/tests/mir-opt/dead-store-elimination/ref.rs b/tests/mir-opt/dead-store-elimination/ref.rs
@@ -11,9 +11,7 @@ pub fn tuple(v: (i32, &Foo)) -> i32 {
     // CHECK-LABEL: fn tuple
     // CHECK: debug _dead => [[dead:_[0-9]+]];
     // CHECK: bb0:
-    // FIXME: Preserve `tmp` for debuginfo, but we can merge it into the debuginfo.
-    // CHECK: [[tmp:_[0-9]+]] = deref_copy (_1.1: &Foo);
-    // CHECK-NEXT: DBG: [[dead]] = &((*[[tmp]]).2: i32)
+    // CHECK: DBG: [[dead]] = &((*_3).2: i32)
     let _dead = &v.1.c;
     v.1.a
 }
diff --git a/tests/mir-opt/dead-store-elimination/ref.tuple.DeadStoreElimination-initial.diff b/tests/mir-opt/dead-store-elimination/ref.tuple.DeadStoreElimination-initial.diff
@@ -13,7 +13,7 @@
   
       bb0: {
 -         StorageLive(_2);
-          _3 = deref_copy (_1.1: &Foo);
+-         _3 = deref_copy (_1.1: &Foo);
 -         _2 = &((*_3).2: i32);
 +         // DBG: _2 = &((*_3).2: i32);
           _4 = deref_copy (_1.1: &Foo);
diff --git a/tests/mir-opt/inline/inline_shims.drop.Inline.panic-abort.diff b/tests/mir-opt/inline/inline_shims.drop.Inline.panic-abort.diff
@@ -16,12 +16,10 @@
 +             let mut _9: *mut A;
 +             let mut _10: usize;
 +             scope 3 (inlined Vec::<A>::as_mut_ptr) {
-+                 let mut _11: &alloc::raw_vec::RawVec<A>;
 +                 scope 4 (inlined alloc::raw_vec::RawVec::<A>::ptr) {
-+                     let mut _12: &alloc::raw_vec::RawVecInner;
 +                     scope 5 (inlined alloc::raw_vec::RawVecInner::ptr::<A>) {
 +                         scope 6 (inlined alloc::raw_vec::RawVecInner::non_null::<A>) {
-+                             let mut _13: std::ptr::NonNull<u8>;
++                             let mut _11: std::ptr::NonNull<u8>;
 +                             scope 7 (inlined Unique::<u8>::cast::<A>) {
 +                                 scope 8 (inlined NonNull::<u8>::cast::<A>) {
 +                                     scope 9 (inlined NonNull::<u8>::as_ptr) {
@@ -41,15 +39,15 @@
 +                 }
 +             }
 +             scope 14 (inlined drop_in_place::<[A]> - shim(Some([A]))) {
-+                 let mut _14: usize;
-+                 let mut _15: *mut A;
-+                 let mut _16: bool;
++                 let mut _12: usize;
++                 let mut _13: *mut A;
++                 let mut _14: bool;
 +             }
 +         }
 +     }
 +     scope 15 (inlined drop_in_place::<Option<B>> - shim(Some(Option<B>))) {
-+         let mut _17: isize;
-+         let mut _18: isize;
++         let mut _15: isize;
++         let mut _16: isize;
 +     }
   
       bb0: {
@@ -64,20 +62,16 @@
 +         StorageLive(_8);
 +         StorageLive(_9);
 +         StorageLive(_11);
-+         StorageLive(_12);
-+         StorageLive(_13);
-+         _13 = copy (((((*_6).0: alloc::raw_vec::RawVec<A>).0: alloc::raw_vec::RawVecInner).0: std::ptr::Unique<u8>).0: std::ptr::NonNull<u8>);
-+         _9 = copy _13 as *mut A (Transmute);
-+         StorageDead(_13);
-+         StorageDead(_12);
++         _11 = copy (((((*_6).0: alloc::raw_vec::RawVec<A>).0: alloc::raw_vec::RawVecInner).0: std::ptr::Unique<u8>).0: std::ptr::NonNull<u8>);
++         _9 = copy _11 as *mut A (Transmute);
 +         StorageDead(_11);
 +         _10 = copy ((*_6).1: usize);
 +         _8 = *mut [A] from (copy _9, copy _10);
 +         StorageDead(_9);
++         StorageLive(_12);
++         StorageLive(_13);
 +         StorageLive(_14);
-+         StorageLive(_15);
-+         StorageLive(_16);
-+         _14 = const 0_usize;
++         _12 = const 0_usize;
 +         goto -> bb4;
       }
   
@@ -89,33 +83,33 @@
           StorageLive(_5);
           _5 = copy _2;
 -         _0 = drop_in_place::<Option<B>>(move _5) -> [return: bb2, unwind unreachable];
-+         StorageLive(_17);
-+         _17 = discriminant((*_5));
-+         switchInt(move _17) -> [0: bb5, otherwise: bb6];
++         StorageLive(_15);
++         _15 = discriminant((*_5));
++         switchInt(move _15) -> [0: bb5, otherwise: bb6];
       }
   
       bb2: {
-+         StorageDead(_16);
-+         StorageDead(_15);
 +         StorageDead(_14);
++         StorageDead(_13);
++         StorageDead(_12);
 +         StorageDead(_8);
 +         StorageDead(_10);
 +         drop(((*_4).0: alloc::raw_vec::RawVec<A>)) -> [return: bb1, unwind unreachable];
 +     }
 + 
 +     bb3: {
-+         _15 = &raw mut (*_8)[_14];
-+         _14 = Add(move _14, const 1_usize);
-+         drop((*_15)) -> [return: bb4, unwind unreachable];
++         _13 = &raw mut (*_8)[_12];
++         _12 = Add(move _12, const 1_usize);
++         drop((*_13)) -> [return: bb4, unwind unreachable];
 +     }
 + 
 +     bb4: {
-+         _16 = Eq(copy _14, copy _10);
-+         switchInt(move _16) -> [0: bb3, otherwise: bb2];
++         _14 = Eq(copy _12, copy _10);
++         switchInt(move _14) -> [0: bb3, otherwise: bb2];
 +     }
 + 
 +     bb5: {
-+         StorageDead(_17);
++         StorageDead(_15);
           StorageDead(_5);
           return;
 +     }
diff --git a/tests/mir-opt/inline_coroutine_body.run2-{closure#0}.Inline.panic-abort.diff b/tests/mir-opt/inline_coroutine_body.run2-{closure#0}.Inline.panic-abort.diff
@@ -224,16 +224,12 @@
 +         StorageLive(_50);
 +         StorageLive(_42);
 +         StorageLive(_43);
-+         StorageLive(_44);
-+         StorageLive(_47);
 +         _45 = copy (_19.0: &mut std::future::Ready<()>);
-+         StorageDead(_47);
 +         StorageLive(_48);
 +         _48 = Option::<()>::None;
 +         _43 = copy ((*_45).0: std::option::Option<()>);
 +         ((*_45).0: std::option::Option<()>) = copy _48;
 +         StorageDead(_48);
-+         StorageDead(_44);
 +         StorageLive(_49);
 +         _49 = discriminant(_43);
 +         switchInt(move _49) -> [0: bb11, 1: bb12, otherwise: bb5];
diff --git a/tests/mir-opt/inline_coroutine_body.run2-{closure#0}.Inline.panic-unwind.diff b/tests/mir-opt/inline_coroutine_body.run2-{closure#0}.Inline.panic-unwind.diff
diff --git a/tests/mir-opt/pre-codegen/dead_on_invalid_place.invalid_place.PreCodegen.after.mir b/tests/mir-opt/pre-codegen/dead_on_invalid_place.invalid_place.PreCodegen.after.mir
diff --git a/tests/mir-opt/pre-codegen/dead_on_invalid_place.rs b/tests/mir-opt/pre-codegen/dead_on_invalid_place.rs
diff --git a/tests/mir-opt/pre-codegen/slice_filter.variant_a-{closure#0}.PreCodegen.after.mir b/tests/mir-opt/pre-codegen/slice_filter.variant_a-{closure#0}.PreCodegen.after.mir
diff --git a/tests/mir-opt/pre-codegen/vec_deref.vec_deref_to_slice.PreCodegen.after.panic-abort.mir b/tests/mir-opt/pre-codegen/vec_deref.vec_deref_to_slice.PreCodegen.after.panic-abort.mir
diff --git a/tests/mir-opt/pre-codegen/vec_deref.vec_deref_to_slice.PreCodegen.after.panic-unwind.mir b/tests/mir-opt/pre-codegen/vec_deref.vec_deref_to_slice.PreCodegen.after.panic-unwind.mir

Original file line number	Diff line number	Diff line change
`@@ -866,6 +866,9 @@ impl Debug for StmtDebugInfo<'_> {`
`866`	`866`	`StmtDebugInfo::AssignRef(local, place) => {`
`867`	`867`	`write!(fmt, "{local:?} = &{place:?}")`
`868`	`868`	`}`
	`869`	`+ StmtDebugInfo::InvalidAssign(local) => {`
	`870`	`+ write!(fmt, "{local:?} = &?")`
	`871`	`+ }`
`869`	`872`	`}`
`870`	`873`	`}`
`871`	`874`	`}`
Original file line number	Diff line number	Diff line change
`@@ -406,6 +406,13 @@ macro_rules! make_mir_visitor {`
`406`	`406`	`location`
`407`	`407`	`);`
`408`	`408`	`},`
	`409`	`+ StmtDebugInfo::InvalidAssign(local) => {`
	`410`	`+ self.visit_local(`
	`411`	`+ $(& $mutability)? *local,`
	`412`	`+ PlaceContext::NonUse(NonUseContext::VarDebugInfo),`
	`413`	`+ location`
	`414`	`+ );`
	`415`	`+ }`
`409`	`416`	`}`
`410`	`417`	`}`
`411`	`418`