core: Change RHS operand of bitshift operators to int32 #349

Merged
sb10q merged 2 commits from issue-336 into master 2023-11-09 12:16:21 +08:00
6 changed files with 105 additions and 16 deletions

View File

@ -298,8 +298,40 @@ impl<'ctx, 'a> CodeGenContext<'ctx, 'a> {
(Operator::BitOr, _) => self.builder.build_or(lhs, rhs, "or").into(), (Operator::BitOr, _) => self.builder.build_or(lhs, rhs, "or").into(),
(Operator::BitXor, _) => self.builder.build_xor(lhs, rhs, "xor").into(), (Operator::BitXor, _) => self.builder.build_xor(lhs, rhs, "xor").into(),
(Operator::BitAnd, _) => self.builder.build_and(lhs, rhs, "and").into(), (Operator::BitAnd, _) => self.builder.build_and(lhs, rhs, "and").into(),
(Operator::LShift, _) => self.builder.build_left_shift(lhs, rhs, "lshift").into(),
Outdated
Review

It's become unclear here to which side of the operand signed refers to. I suggest adding a comment here to clarify what is going on.

It's become unclear here to which side of the operand ``signed`` refers to. I suggest adding a comment here to clarify what is going on.
(Operator::RShift, _) => self.builder.build_right_shift(lhs, rhs, true, "rshift").into(), // Sign-ness of bitshift operators are always determined by the left operand
(Operator::LShift, signed) | (Operator::RShift, signed) => {
Outdated
Review

RHS is always 32, no?

RHS is always 32, no?
// RHS operand is always 32 bits
assert_eq!(rhs.get_type().get_bit_width(), 32);
Outdated
Review

AFAICT this will break silently with negative shifts.

Python disallows them:

>>> 1 << (-1)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: negative shift count

So the RHS type should perhaps be uint32 below instead of int32, and we should check that the compiler reports a sensible error if the user enters a negative value there.

But I worry that uint32 would require explicit casts and a lot of typing, e.g. 1 << uint32(1) instead of just 1 << 1?

Alternatively we can simply support negative shifts since this is just a strict superset of Python as far as I can tell. But the zero-extend here should be replaced with a sign extend.

AFAICT this will break silently with negative shifts. Python disallows them: ``` >>> 1 << (-1) Traceback (most recent call last): File "<stdin>", line 1, in <module> ValueError: negative shift count ``` So the RHS type should perhaps be uint32 below instead of int32, and we should check that the compiler reports a sensible error if the user enters a negative value there. But I worry that uint32 would require explicit casts and a lot of typing, e.g. ``1 << uint32(1)`` instead of just ``1 << 1``? Alternatively we can simply support negative shifts since this is just a strict superset of Python as far as I can tell. But the zero-extend here should be replaced with a sign extend.
let common_type = lhs.get_type();
let rhs = if common_type.get_bit_width() > 32 {
if signed {
self.builder.build_int_s_extend(rhs, common_type, "")
Outdated
Review

Do we ever do this? Are there any int types smaller than 32-bit?

Do we ever do this? Are there any int types smaller than 32-bit?
} else {
self.builder.build_int_z_extend(rhs, common_type, "")
}
Outdated
Review

Is LLVM able to optimize this away when the RHS is uint32?

Is LLVM able to optimize this away when the RHS is uint32?
Outdated
Review

Or a constant?

Or a constant?

Yes, all test cases can be constant-folded using optimizations.

Yes, all test cases can be constant-folded using optimizations.
} else {
rhs
};
let rhs_gez = self.builder.build_int_compare(IntPredicate::SGE, rhs, common_type.const_zero(), "");
self.make_assert(
generator,
rhs_gez,
"ValueError",
"negative shift count",
[None, None, None],
self.current_loc
);
match *op {
Operator::LShift => self.builder.build_left_shift(lhs, rhs, "lshift").into(),
Operator::RShift => self.builder.build_right_shift(lhs, rhs, signed, "rshift").into(),
_ => unreachable!()
}
}
(Operator::FloorDiv, true) => self.builder.build_int_signed_div(lhs, rhs, "floordiv").into(), (Operator::FloorDiv, true) => self.builder.build_int_signed_div(lhs, rhs, "floordiv").into(),
(Operator::FloorDiv, false) => self.builder.build_int_unsigned_div(lhs, rhs, "floordiv").into(), (Operator::FloorDiv, false) => self.builder.build_int_unsigned_div(lhs, rhs, "floordiv").into(),
(Operator::Pow, s) => integer_power(generator, self, lhs, rhs, s).into(), (Operator::Pow, s) => integer_power(generator, self, lhs, rhs, s).into(),
@ -1085,6 +1117,9 @@ pub fn gen_binop_expr<'ctx, 'a, G: CodeGenerator>(
Ok(Some(ctx.gen_int_ops(generator, op, left_val, right_val, true).into())) Ok(Some(ctx.gen_int_ops(generator, op, left_val, right_val, true).into()))
} else if ty1 == ty2 && [ctx.primitives.uint32, ctx.primitives.uint64].contains(&ty1) { } else if ty1 == ty2 && [ctx.primitives.uint32, ctx.primitives.uint64].contains(&ty1) {
Ok(Some(ctx.gen_int_ops(generator, op, left_val, right_val, false).into())) Ok(Some(ctx.gen_int_ops(generator, op, left_val, right_val, false).into()))
} else if [Operator::LShift, Operator::RShift].contains(op) {
let signed = [ctx.primitives.int32, ctx.primitives.int64].contains(&ty1);
Ok(Some(ctx.gen_int_ops(generator, op, left_val, right_val, signed).into()))
} else if ty1 == ty2 && ctx.primitives.float == ty1 { } else if ty1 == ty2 && ctx.primitives.float == ty1 {
Ok(Some(ctx.gen_float_ops(op, left_val, right_val).into())) Ok(Some(ctx.gen_float_ops(op, left_val, right_val).into()))
} else if ty1 == ctx.primitives.float && ty2 == ctx.primitives.int32 { } else if ty1 == ctx.primitives.float && ty2 == ctx.primitives.int32 {

View File

@ -2,7 +2,7 @@ use crate::typecheck::typedef::TypeEnum;
use super::type_inferencer::Inferencer; use super::type_inferencer::Inferencer;
use super::typedef::Type; use super::typedef::Type;
use nac3parser::ast::{self, Expr, ExprKind, Stmt, StmtKind, StrRef}; use nac3parser::ast::{self, Constant, Expr, ExprKind, Operator::{LShift, RShift}, Stmt, StmtKind, StrRef};
use std::{collections::HashSet, iter::once}; use std::{collections::HashSet, iter::once};
impl<'a> Inferencer<'a> { impl<'a> Inferencer<'a> {
@ -107,11 +107,28 @@ impl<'a> Inferencer<'a> {
self.check_expr(value, defined_identifiers)?; self.check_expr(value, defined_identifiers)?;
self.should_have_value(value)?; self.should_have_value(value)?;
} }
ExprKind::BinOp { left, right, .. } => { ExprKind::BinOp { left, op, right } => {
self.check_expr(left, defined_identifiers)?; self.check_expr(left, defined_identifiers)?;
self.check_expr(right, defined_identifiers)?; self.check_expr(right, defined_identifiers)?;
self.should_have_value(left)?; self.should_have_value(left)?;
self.should_have_value(right)?; self.should_have_value(right)?;
// Check whether a bitwise shift has a negative RHS constant value
if *op == LShift || *op == RShift {
if let ExprKind::Constant { value, .. } = &right.node {
let rhs_val = match value {
Constant::Int(v) => v,
_ => unreachable!(),
};
if *rhs_val < 0 {
return Err(format!(
"shift count is negative at {}",
right.location
));
}
}
}
} }
ExprKind::UnaryOp { operand, .. } => { ExprKind::UnaryOp { operand, .. } => {
self.check_expr(operand, defined_identifiers)?; self.check_expr(operand, defined_identifiers)?;

View File

@ -96,11 +96,13 @@ pub fn impl_binop(
let (ty, var_id) = unifier.get_fresh_var_with_range(other_ty, Some("N".into()), None); let (ty, var_id) = unifier.get_fresh_var_with_range(other_ty, Some("N".into()), None);
(ty, Some(var_id)) (ty, Some(var_id))
}; };
let function_vars = if let Some(var_id) = other_var_id { let function_vars = if let Some(var_id) = other_var_id {
vec![(var_id, other_ty)].into_iter().collect::<HashMap<_, _>>() vec![(var_id, other_ty)].into_iter().collect::<HashMap<_, _>>()
} else { } else {
HashMap::new() HashMap::new()
}; };
for op in ops { for op in ops {
fields.insert(binop_name(op).into(), { fields.insert(binop_name(op).into(), {
( (
@ -224,7 +226,7 @@ pub fn impl_bitwise_arithmetic(unifier: &mut Unifier, store: &PrimitiveStore, ty
/// LShift, RShift /// LShift, RShift
pub fn impl_bitwise_shift(unifier: &mut Unifier, store: &PrimitiveStore, ty: Type) { pub fn impl_bitwise_shift(unifier: &mut Unifier, store: &PrimitiveStore, ty: Type) {
impl_binop(unifier, store, ty, &[ty], ty, &[ast::Operator::LShift, ast::Operator::RShift]) impl_binop(unifier, store, ty, &[store.int32, store.uint32], ty, &[ast::Operator::LShift, ast::Operator::RShift]);
} }
/// Div /// Div
@ -295,6 +297,7 @@ pub fn set_primitives_magic_methods(store: &PrimitiveStore, unifier: &mut Unifie
uint64: uint64_t, uint64: uint64_t,
.. ..
} = *store; } = *store;
/* int ======== */ /* int ======== */
for t in [int32_t, int64_t, uint32_t, uint64_t] { for t in [int32_t, int64_t, uint32_t, uint64_t] {
impl_basic_arithmetic(unifier, store, t, &[t], t); impl_basic_arithmetic(unifier, store, t, &[t], t);

View File

@ -116,6 +116,7 @@ impl RecordField {
} }
} }
/// Category of variable and value types.
#[derive(Clone)] #[derive(Clone)]
pub enum TypeEnum { pub enum TypeEnum {
TRigidVar { TRigidVar {
@ -123,6 +124,8 @@ pub enum TypeEnum {
name: Option<StrRef>, name: Option<StrRef>,
loc: Option<Location>, loc: Option<Location>,
}, },
/// A type variable.
TVar { TVar {
id: u32, id: u32,
// empty indicates this is not a struct/tuple/list // empty indicates this is not a struct/tuple/list
@ -132,21 +135,41 @@ pub enum TypeEnum {
name: Option<StrRef>, name: Option<StrRef>,
loc: Option<Location>, loc: Option<Location>,
}, },
/// A tuple type.
TTuple { TTuple {
/// The types of elements present in this tuple.
ty: Vec<Type>, ty: Vec<Type>,
}, },
/// A list type.
TList { TList {
/// The type of elements present in this list.
ty: Type, ty: Type,
}, },
/// An object type.
TObj { TObj {
/// The [DefintionId] of this object type.
obj_id: DefinitionId, obj_id: DefinitionId,
/// The fields present in this object type.
///
/// The key of the [Mapping] is the identifier of the field, while the value is a tuple
/// containing the [Type] of the field, and a `bool` indicating whether the field is a
/// variable (as opposed to a function).
fields: Mapping<StrRef, (Type, bool)>, fields: Mapping<StrRef, (Type, bool)>,
/// Mapping between the ID of type variables and the [Type] representing the type variables
/// of this object type.
params: VarMap, params: VarMap,
}, },
TVirtual { TVirtual {
ty: Type, ty: Type,
}, },
TCall(Vec<CallId>), TCall(Vec<CallId>),
/// A function type.
TFunc(FunSignature), TFunc(FunSignature),
} }
@ -294,11 +317,16 @@ impl Unifier {
self.get_fresh_var_with_range(&[], None, None) self.get_fresh_var_with_range(&[], None, None)
} }
/// Returns a fresh [type variable][TypeEnum::TVar] with no associated range.
///
/// This type variable can be instantiated by any type.
pub fn get_fresh_var(&mut self, name: Option<StrRef>, loc: Option<Location>) -> (Type, u32) { pub fn get_fresh_var(&mut self, name: Option<StrRef>, loc: Option<Location>) -> (Type, u32) {
self.get_fresh_var_with_range(&[], name, loc) self.get_fresh_var_with_range(&[], name, loc)
} }
/// Get a fresh type variable. /// Returns a fresh [type variable][TypeEnum::TVar] with the range specified by `range`.
///
/// This type variable can be instantiated by any type present in `range`.
pub fn get_fresh_var_with_range( pub fn get_fresh_var_with_range(
&mut self, &mut self,
range: &[Type], range: &[Type],

View File

@ -41,10 +41,10 @@ def u64_max() -> uint64:
return ~uint64(0) return ~uint64(0)
def i64_min() -> int64: def i64_min() -> int64:
return int64(1) << int64(63) return int64(1) << 63
def i64_max() -> int64: def i64_max() -> int64:
return ~(int64(1) << int64(63)) return ~(int64(1) << 63)
def test_u32_bnot(): def test_u32_bnot():
output_uint32(~uint32(0)) output_uint32(~uint32(0))

View File

@ -37,7 +37,9 @@ def test_int32():
output_int32(a ^ b) output_int32(a ^ b)
output_int32(a & b) output_int32(a & b)
output_int32(a << b) output_int32(a << b)
output_int32(a << uint32(b))
output_int32(a >> b) output_int32(a >> b)
output_int32(a >> uint32(b))
output_float64(a / b) output_float64(a / b)
a += b a += b
output_int32(a) output_int32(a)
@ -74,7 +76,9 @@ def test_uint32():
output_uint32(a ^ b) output_uint32(a ^ b)
output_uint32(a & b) output_uint32(a & b)
output_uint32(a << b) output_uint32(a << b)
output_uint32(a << int32(b))
output_uint32(a >> b) output_uint32(a >> b)
output_uint32(a >> int32(b))
output_float64(a / b) output_float64(a / b)
a += b a += b
output_uint32(a) output_uint32(a)
@ -108,8 +112,10 @@ def test_int64():
output_int64(a | b) output_int64(a | b)
output_int64(a ^ b) output_int64(a ^ b)
output_int64(a & b) output_int64(a & b)
output_int64(a << b) output_int64(a << int32(b))
output_int64(a >> b) output_int64(a << uint32(b))
output_int64(a >> int32(b))
output_int64(a >> uint32(b))
output_float64(a / b) output_float64(a / b)
a += b a += b
output_int64(a) output_int64(a)
@ -127,9 +133,9 @@ def test_int64():
output_int64(a) output_int64(a)
a &= b a &= b
output_int64(a) output_int64(a)
a <<= b a <<= int32(b)
output_int64(a) output_int64(a)
a >>= b a >>= int32(b)
output_int64(a) output_int64(a)
def test_uint64(): def test_uint64():
@ -143,8 +149,8 @@ def test_uint64():
output_uint64(a | b) output_uint64(a | b)
output_uint64(a ^ b) output_uint64(a ^ b)
output_uint64(a & b) output_uint64(a & b)
output_uint64(a << b) output_uint64(a << uint32(b))
output_uint64(a >> b) output_uint64(a >> uint32(b))
output_float64(a / b) output_float64(a / b)
a += b a += b
output_uint64(a) output_uint64(a)
@ -162,9 +168,9 @@ def test_uint64():
output_uint64(a) output_uint64(a)
a &= b a &= b
output_uint64(a) output_uint64(a)
a <<= b a <<= uint32(b)
output_uint64(a) output_uint64(a)
a >>= b a >>= uint32(b)
output_uint64(a) output_uint64(a)
class A: class A: