Issue rust-lang#5977 - Generalizing RawNullablePointer to RawForbiddenValue

Author: David Rajchenbach-Teller

src/librustc_trans/adt.rs

@@ -82,25 +82,37 @@ pub enum Repr<'tcx> {
     /// General-case enums: for each case there is a struct, and they
     /// all start with a field for the discriminant.
     General(IntType, Vec<Struct<'tcx>>),
-    /// Two cases distinguished by a nullable pointer: the case with discriminant
-    /// `nndiscr` must have single field which is known to be nonnull due to its type.
-    /// The other case is known to be zero sized. Hence we represent the enum
-    /// as simply a nullable pointer: if not null it indicates the `nndiscr` variant,
-    /// otherwise it indicates the other case.
-    RawNullablePointer {
-        nndiscr: Disr,
-        nnty: Ty<'tcx>,
-        nullfields: Vec<Ty<'tcx>>
+    /// Two cases distinguished by a known-to-be-forbidden value.
+    ///
+    /// Example: `Option<&T>` (a `&T` cannot be null)
+    /// Example: `Option<char>` (a `char` is large enough to hold 2^32 - 1,
+    ///           but this value is forbidden by definition)
+    /// Example: `Result<&T, ()>` (a `&T` cannot be null)
+    ///
+    /// One of the cases (the "unit case") must be known to be
+    /// zero-sized (e.g. `None`).  The other case (the "payload case")
+    /// must be known to be a single field that cannot adopt a
+    /// specific value (in the above examples, 0 for `&T` or 2^32 - 1
+    /// for `char`).
+    ///
+    /// We may safely represent the enum by its payload case and
+    /// differentiate between cases by checking for the forbidden
+    /// value.
+    RawForbiddenValue {
+        /// Unit case (e.g. `None` or `Either((), ())`)
+        unit_fields: Vec<Ty<'tcx>>,
+        /// Case holding a payload: the constructor
+        payload_discr: Disr,
+        /// Case holding a payload: the type
+        payload_ty: Ty<'tcx>,
+        /// A value that the payload can never hold.
+        forbidden_value: ValueRef,
     },
     /// Two cases distinguished by a nullable pointer: the case with discriminant
     /// `nndiscr` is represented by the struct `nonnull`, where the `discrfield`th
     /// field is known to be nonnull due to its type; if that field is null, then
     /// it represents the other case, which is inhabited by at most one value
     /// (and all other fields are undefined/unused).
-    ///
-    /// For example, `std::option::Option` instantiated at a safe pointer type
-    /// is represented such that `None` is a null pointer and `Some` is the
-    /// identity function.
     StructWrappedNullablePointer {
         nonnull: Struct<'tcx>,
         nndiscr: Disr,
@@ -217,18 +229,26 @@ fn represent_type_uncached<'a, 'tcx>(cx: &CrateContext<'a, 'tcx>,
             }
 
             if cases.len() == 2 && hint == attr::ReprAny {
-                // Nullable pointer optimization
-                let mut discr = 0;
-                while discr < 2 {
+                // Two cases, so it might be possible to turn this
+                // into a `RawForbiddenValue` or a
+                // `StructWrappedNullablePointer`, if all conditions
+                // are met.
+                for discr in 0 .. 2 {
                     if cases[1 - discr].is_zerolen(cx, t) {
+                        // One of the cases has zero length. We are on the right track.
                         let st = mk_struct(cx, &cases[discr].tys,
                                            false, t);
+
+                        // For the moment, we can only apply these
+                        // optimizations to safe pointers.
                         match cases[discr].find_ptr(cx) {
                             Some(ref df) if df.len() == 1 && st.fields.len() == 1 => {
-                                return RawNullablePointer {
-                                    nndiscr: Disr::from(discr),
-                                    nnty: st.fields[0],
-                                    nullfields: cases[1 - discr].tys.clone()
+                                let payload_ty = st.fields[0];
+                                return RawForbiddenValue {
+                                    payload_discr: Disr::from(discr),
+                                    payload_ty: payload_ty,
+                                    forbidden_value: C_null(type_of::sizing_type_of(cx, payload_ty)),
+                                    unit_fields: cases[1 - discr].tys.clone()
                                 };
                             }
                             Some(mut discrfield) => {
@@ -243,8 +263,13 @@ fn represent_type_uncached<'a, 'tcx>(cx: &CrateContext<'a, 'tcx>,
                             }
                             None => {}
                         }
+                        // No need to continue the loop. If both cases
+                        // have zero length, we can apply neither
+                        // `RawForbiddenValue` nor
+                        // `StructWrappedNullablePointer`.
+                        break;
+
                     }
-                    discr += 1;
                 }
             }
 
@@ -432,6 +457,8 @@ impl<'tcx> Case<'tcx> {
         mk_struct(cx, &self.tys, false, scapegoat).size == 0
     }
 
+    /// Find a safe pointer that may be used to discriminate in a
+    /// RawForbiddenValue or StructWrappedNullablePointer.
     fn find_ptr<'a>(&self, cx: &CrateContext<'a, 'tcx>) -> Option<DiscrField> {
         for (i, &ty) in self.tys.iter().enumerate() {
             if let Some(mut path) = find_discr_field_candidate(cx.tcx(), ty, vec![]) {
@@ -643,7 +670,7 @@ pub fn incomplete_type_of<'a, 'tcx>(cx: &CrateContext<'a, 'tcx>,
 pub fn finish_type_of<'a, 'tcx>(cx: &CrateContext<'a, 'tcx>,
                                 r: &Repr<'tcx>, llty: &mut Type) {
     match *r {
-        CEnum(..) | General(..) | RawNullablePointer { .. } => { }
+        CEnum(..) | General(..) | RawForbiddenValue { .. } => { }
         Univariant(ref st) | StructWrappedNullablePointer { nonnull: ref st, .. } =>
             llty.set_struct_body(&struct_llfields(cx, st, false, false),
                                  st.packed)
@@ -659,8 +686,8 @@ fn generic_type_of<'a, 'tcx>(cx: &CrateContext<'a, 'tcx>,
            r, name, sizing, dst);
     match *r {
         CEnum(ity, _, _) => ll_inttype(cx, ity),
-        RawNullablePointer { nnty, .. } =>
-            type_of::sizing_type_of(cx, nnty),
+        RawForbiddenValue { payload_ty, .. } =>
+            type_of::sizing_type_of(cx, payload_ty),
         StructWrappedNullablePointer { nonnull: ref st, .. } => {
             match name {
                 None => {
@@ -756,7 +783,7 @@ pub fn trans_switch<'blk, 'tcx>(bcx: Block<'blk, 'tcx>,
                                 -> (BranchKind, Option<ValueRef>) {
     match *r {
         CEnum(..) | General(..) |
-        RawNullablePointer { .. } | StructWrappedNullablePointer { .. } => {
+        RawForbiddenValue{ .. } | StructWrappedNullablePointer { .. } => {
             (BranchKind::Switch, Some(trans_get_discr(bcx, r, scrutinee, None, range_assert)))
         }
         Univariant(..) => {
@@ -771,7 +798,7 @@ pub fn is_discr_signed<'tcx>(r: &Repr<'tcx>) -> bool {
         CEnum(ity, _, _) => ity.is_signed(),
         General(ity, _) => ity.is_signed(),
         Univariant(..) => false,
-        RawNullablePointer { .. } => false,
+        RawForbiddenValue { payload_ty, .. } => payload_ty.is_signed(),
         StructWrappedNullablePointer { .. } => false,
     }
 }
@@ -792,10 +819,9 @@ pub fn trans_get_discr<'blk, 'tcx>(bcx: Block<'blk, 'tcx>, r: &Repr<'tcx>,
                        range_assert)
         }
         Univariant(..) => C_u8(bcx.ccx(), 0),
-        RawNullablePointer { nndiscr, nnty, .. } =>  {
-            let cmp = if nndiscr == Disr(0) { IntEQ } else { IntNE };
-            let llptrty = type_of::sizing_type_of(bcx.ccx(), nnty);
-            ICmp(bcx, cmp, Load(bcx, scrutinee), C_null(llptrty), DebugLoc::None)
+        RawForbiddenValue { payload_discr, forbidden_value, .. } =>  {
+            let cmp = if payload_discr == Disr(0) { IntEQ } else { IntNE };
+            ICmp(bcx, cmp, Load(bcx, scrutinee), forbidden_value, DebugLoc::None)
         }
         StructWrappedNullablePointer { nndiscr, ref discrfield, .. } => {
             struct_wrapped_nullable_bitdiscr(bcx, nndiscr, discrfield, scrutinee)
@@ -856,7 +882,7 @@ pub fn trans_case<'blk, 'tcx>(bcx: Block<'blk, 'tcx>, r: &Repr, discr: Disr)
         Univariant(..) => {
             bug!("no cases for univariants or structs")
         }
-        RawNullablePointer { .. } |
+        RawForbiddenValue { .. } |
         StructWrappedNullablePointer { .. } => {
             assert!(discr == Disr(0) || discr == Disr(1));
             C_bool(bcx.ccx(), discr != Disr(0))
@@ -881,10 +907,9 @@ pub fn trans_set_discr<'blk, 'tcx>(bcx: Block<'blk, 'tcx>, r: &Repr<'tcx>,
         Univariant(_) => {
             assert_eq!(discr, Disr(0));
         }
-        RawNullablePointer { nndiscr, nnty, ..} => {
-            if discr != nndiscr {
-                let llptrty = type_of::sizing_type_of(bcx.ccx(), nnty);
-                Store(bcx, C_null(llptrty), val);
+        RawForbiddenValue { payload_discr, forbidden_value, ..} => {
+            if discr != payload_discr {
+                Store(bcx, forbidden_value, val);
             }
         }
         StructWrappedNullablePointer { nndiscr, ref discrfield, .. } => {
@@ -936,7 +961,7 @@ pub fn trans_field_ptr_builder<'blk, 'tcx>(bcx: &BlockAndBuilder<'blk, 'tcx>,
         General(_, ref cases) => {
             struct_field_ptr(bcx, &cases[discr.0 as usize], val, ix + 1, true)
         }
-        RawNullablePointer { nndiscr, ref nullfields, .. } |
+        RawForbiddenValue { payload_discr: nndiscr, unit_fields: ref nullfields, .. } |
         StructWrappedNullablePointer { nndiscr, ref nullfields, .. } if discr != nndiscr => {
             // The unit-like case might have a nonzero number of unit-like fields.
             // (e.d., Result of Either with (), as one side.)
@@ -947,10 +972,10 @@ pub fn trans_field_ptr_builder<'blk, 'tcx>(bcx: &BlockAndBuilder<'blk, 'tcx>,
             if bcx.is_unreachable() { return C_undef(ty.ptr_to()); }
             bcx.pointercast(val.value, ty.ptr_to())
         }
-        RawNullablePointer { nndiscr, nnty, .. } => {
-            assert_eq!(ix, 0);
-            assert_eq!(discr, nndiscr);
-            let ty = type_of::type_of(bcx.ccx(), nnty);
+        RawForbiddenValue { payload_discr, payload_ty, .. } => {
+            assert_eq!(ix, 0); // By definition, the payload of RawForbiddenValue has a single field.
+            assert_eq!(discr, payload_discr);
+            let ty = type_of::type_of(bcx.ccx(), payload_ty);
             if bcx.is_unreachable() { return C_undef(ty.ptr_to()); }
             bcx.pointercast(val.value, ty.ptr_to())
         }
@@ -1102,12 +1127,12 @@ pub fn trans_const<'a, 'tcx>(ccx: &CrateContext<'a, 'tcx>, r: &Repr<'tcx>, discr
             let contents = build_const_struct(ccx, st, vals);
             C_struct(ccx, &contents[..], st.packed)
         }
-        RawNullablePointer { nndiscr, nnty, .. } => {
-            if discr == nndiscr {
-                assert_eq!(vals.len(), 1);
+        RawForbiddenValue { payload_discr, forbidden_value, .. } => {
+            if discr == payload_discr {
+                assert_eq!(vals.len(), 1); // By definition, the payload has only a single field.
                 vals[0]
             } else {
-                C_null(type_of::sizing_type_of(ccx, nnty))
+                forbidden_value
             }
         }
         StructWrappedNullablePointer { ref nonnull, nndiscr, .. } => {
@@ -1203,14 +1228,14 @@ fn roundup(x: u64, a: u32) -> u64 { let a = a as u64; ((x + (a - 1)) / a) * a }
 ///
 /// (Not to be confused with `common::const_get_elt`, which operates on
 /// raw LLVM-level structs and arrays.)
-pub fn const_get_field(r: &Repr, val: ValueRef, _discr: Disr,
+pub fn const_get_field(r: &Repr, val: ValueRef, discr: Disr,
                        ix: usize) -> ValueRef {
     match *r {
         CEnum(..) => bug!("element access in C-like enum const"),
         Univariant(..) => const_struct_field(val, ix),
         General(..) => const_struct_field(val, ix + 1),
-        RawNullablePointer { .. } => {
-            assert_eq!(ix, 0);
+        RawForbiddenValue { .. } => {
+            assert_eq!(ix, 0); // By definition, the payload only has a single field.
             val
         },
         StructWrappedNullablePointer{ .. } => const_struct_field(val, ix)

src/librustc_trans/debuginfo/metadata.rs

@@ -1295,36 +1295,36 @@ impl<'tcx> EnumMemberDescriptionFactory<'tcx> {
                     ]
                 }
             }
-            adt::RawNullablePointer { nndiscr: non_null_variant_index, nnty, .. } => {
+            adt::RawForbiddenValue { payload_discr: payload_variant_index, payload_ty, .. } => {
                 // As far as debuginfo is concerned, the pointer this enum
                 // represents is still wrapped in a struct. This is to make the
                 // DWARF representation of enums uniform.
 
                 // First create a description of the artificial wrapper struct:
-                let non_null_variant = &adt.variants[non_null_variant_index.0 as usize];
-                let non_null_variant_name = non_null_variant.name.as_str();
+                let payload_variant = &adt.variants[payload_variant_index.0 as usize];
+                let payload_variant_name = payload_variant.name.as_str();
 
                 // The llvm type and metadata of the pointer
-                let non_null_llvm_type = type_of::type_of(cx, nnty);
-                let non_null_type_metadata = type_metadata(cx, nnty, self.span);
+                let payload_llvm_type = type_of::type_of(cx, payload_ty);
+                let payload_type_metadata = type_metadata(cx, payload_ty, self.span);
 
                 // The type of the artificial struct wrapping the pointer
                 let artificial_struct_llvm_type = Type::struct_(cx,
-                                                                &[non_null_llvm_type],
+                                                                &[payload_llvm_type],
                                                                 false);
 
                 // For the metadata of the wrapper struct, we need to create a
                 // MemberDescription of the struct's single field.
                 let sole_struct_member_description = MemberDescription {
-                    name: match non_null_variant.kind {
+                    name: match payload_variant.kind {
                         ty::VariantKind::Tuple => "__0".to_string(),
                         ty::VariantKind::Struct => {
-                            non_null_variant.fields[0].name.to_string()
+                            payload_variant.fields[0].name.to_string()
                         }
                         ty::VariantKind::Unit => bug!()
                     },
-                    llvm_type: non_null_llvm_type,
-                    type_metadata: non_null_type_metadata,
+                    llvm_type: payload_llvm_type,
+                    type_metadata: payload_type_metadata,
                     offset: FixedMemberOffset { bytes: 0 },
                     flags: FLAGS_NONE
                 };
@@ -1334,13 +1334,13 @@ impl<'tcx> EnumMemberDescriptionFactory<'tcx> {
                                                       .get_unique_type_id_of_enum_variant(
                                                           cx,
                                                           self.enum_type,
-                                                          &non_null_variant_name);
+                                                          &payload_variant_name);
 
                 // Now we can create the metadata of the artificial struct
                 let artificial_struct_metadata =
                     composite_type_metadata(cx,
                                             artificial_struct_llvm_type,
-                                            &non_null_variant_name,
+                                            &payload_variant_name,
                                             unique_type_id,
                                             &[sole_struct_member_description],
                                             self.containing_scope,
@@ -1349,11 +1349,11 @@ impl<'tcx> EnumMemberDescriptionFactory<'tcx> {
 
                 // Encode the information about the null variant in the union
                 // member's name.
-                let null_variant_index = (1 - non_null_variant_index.0) as usize;
-                let null_variant_name = adt.variants[null_variant_index].name;
+                let unit_variant_index = (1 - payload_variant_index.0) as usize;
+                let unit_variant_name = adt.variants[unit_variant_index].name;
                 let union_member_name = format!("RUST$ENCODED$ENUM${}${}",
                                                 0,
-                                                null_variant_name);
+                                                unit_variant_name);
 
                 // Finally create the (singleton) list of descriptions of union
                 // members.
@@ -1607,7 +1607,7 @@ fn prepare_enum_metadata<'a, 'tcx>(cx: &CrateContext<'a, 'tcx>,
         adt::CEnum(inttype, _, _) => {
             return FinalMetadata(discriminant_type_metadata(inttype))
         },
-        adt::RawNullablePointer { .. }           |
+        adt::RawForbiddenValue { .. }           |
         adt::StructWrappedNullablePointer { .. } |
         adt::Univariant(..)                      => None,
         adt::General(inttype, _) => Some(discriminant_type_metadata(inttype)),

src/librustc_trans/disr.rs

@@ -8,6 +8,7 @@
 // option. This file may not be copied, modified, or distributed
 // except according to those terms.
 
+/// Representation of single value in a C-style enum.
 #[derive(Debug, Eq, PartialEq, Copy, Clone)]
 pub struct Disr(pub u64);