Skip to content

Treat f16 and f128 as floats in constant evaluation #15897

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Treat f16 and f128 as floats in constant evaluation #15897

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 1 addition & 7 deletions clippy_lints/src/neg_multiply.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -62,16 +62,10 @@ impl<'tcx> LateLintPass<'tcx> for NegMultiply {
}

fn check_mul(cx: &LateContext<'_>, mul_expr: &Expr<'_>, lit: &Expr<'_>, exp: &Expr<'_>) {
const F16_ONE: u16 = 1.0_f16.to_bits();
const F128_ONE: u128 = 1.0_f128.to_bits();
if let ExprKind::Lit(l) = lit.kind
&& matches!(
consts::lit_to_mir_constant(&l.node, cx.typeck_results().expr_ty_opt(lit)),
Constant::Int(1)
| Constant::F16(F16_ONE)
| Constant::F32(1.0)
| Constant::F64(1.0)
| Constant::F128(F128_ONE)
Constant::Int(1) | Constant::F16(1.0) | Constant::F32(1.0) | Constant::F64(1.0) | Constant::F128(1.0)
)
&& cx.typeck_results().expr_ty(exp).is_numeric()
{
Expand Down
72 changes: 47 additions & 25 deletions clippy_utils/src/consts.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ use crate::{clip, is_direct_expn_of, sext, sym, unsext};

use rustc_abi::Size;
use rustc_apfloat::Float;
use rustc_apfloat::ieee::{Half, Quad};
use rustc_apfloat::ieee::Quad;
use rustc_ast::ast::{LitFloatType, LitKind};
use rustc_hir::def::{DefKind, Res};
use rustc_hir::{BinOpKind, Block, ConstBlock, Expr, ExprKind, HirId, PatExpr, PatExprKind, QPath, TyKind, UnOp};
Expand Down Expand Up @@ -38,16 +38,14 @@ pub enum Constant {
Char(char),
/// An integer's bit representation.
Int(u128),
/// An `f16` bitcast to a `u16`.
// FIXME(f16_f128): use `f16` once builtins are available on all host tools platforms.
F16(u16),
/// An `f16`.
F16(f16),
/// An `f32`.
F32(f32),
/// An `f64`.
F64(f64),
/// An `f128` bitcast to a `u128`.
// FIXME(f16_f128): use `f128` once builtins are available on all host tools platforms.
F128(u128),
F128(f128),
/// `true` or `false`.
Bool(bool),
/// An array of constants.
Expand Down Expand Up @@ -169,8 +167,7 @@ impl Hash for Constant {
i.hash(state);
},
Self::F16(f) => {
// FIXME(f16_f128): once conversions to/from `f128` are available on all platforms,
f.hash(state);
f64::from(f).to_bits().hash(state);
},
Self::F32(f) => {
f64::from(f).to_bits().hash(state);
Expand All @@ -179,7 +176,7 @@ impl Hash for Constant {
f.to_bits().hash(state);
},
Self::F128(f) => {
f.hash(state);
f.to_bits().hash(state);
},
Self::Bool(b) => {
b.hash(state);
Expand Down Expand Up @@ -281,14 +278,9 @@ impl Constant {
self
}

fn parse_f16(s: &str) -> Self {
let f: Half = s.parse().unwrap();
Self::F16(f.to_bits().try_into().unwrap())
}

fn parse_f128(s: &str) -> Self {
let f: Quad = s.parse().unwrap();
Self::F128(f.to_bits())
Self::F128(f128::from_bits(f.to_bits()))
}

pub fn new_numeric_min<'tcx>(tcx: TyCtxt<'tcx>, ty: Ty<'tcx>) -> Option<Self> {
Expand Down Expand Up @@ -394,18 +386,20 @@ impl Constant {

pub fn is_pos_infinity(&self) -> bool {
match *self {
// FIXME(f16_f128): add f16 and f128 when constants are available
Constant::F16(x) => x == f16::INFINITY,
Constant::F32(x) => x == f32::INFINITY,
Constant::F64(x) => x == f64::INFINITY,
Constant::F128(x) => x == f128::INFINITY,
_ => false,
}
}

pub fn is_neg_infinity(&self) -> bool {
match *self {
// FIXME(f16_f128): add f16 and f128 when constants are available
Constant::F16(x) => x == f16::NEG_INFINITY,
Constant::F32(x) => x == f32::NEG_INFINITY,
Constant::F64(x) => x == f64::NEG_INFINITY,
Constant::F128(x) => x == f128::NEG_INFINITY,
_ => false,
}
}
Expand All @@ -420,14 +414,14 @@ pub fn lit_to_mir_constant(lit: &LitKind, ty: Option<Ty<'_>>) -> Constant {
LitKind::Char(c) => Constant::Char(c),
LitKind::Int(n, _) => Constant::Int(n.get()),
LitKind::Float(ref is, LitFloatType::Suffixed(fty)) => match fty {
// FIXME(f16_f128): just use `parse()` directly when available for `f16`/`f128`
FloatTy::F16 => Constant::parse_f16(is.as_str()),
FloatTy::F16 => Constant::F16(is.as_str().parse().unwrap()),
FloatTy::F32 => Constant::F32(is.as_str().parse().unwrap()),
FloatTy::F64 => Constant::F64(is.as_str().parse().unwrap()),
// FIXME(f16_f128): just use `parse()` directly when available for `f128`
FloatTy::F128 => Constant::parse_f128(is.as_str()),
},
LitKind::Float(ref is, LitFloatType::Unsuffixed) => match ty.expect("type of float is known").kind() {
ty::Float(FloatTy::F16) => Constant::parse_f16(is.as_str()),
ty::Float(FloatTy::F16) => Constant::F16(is.as_str().parse().unwrap()),
ty::Float(FloatTy::F32) => Constant::F32(is.as_str().parse().unwrap()),
ty::Float(FloatTy::F64) => Constant::F64(is.as_str().parse().unwrap()),
ty::Float(FloatTy::F128) => Constant::parse_f128(is.as_str()),
Expand Down Expand Up @@ -932,6 +926,7 @@ impl<'tcx> ConstEvalCtxt<'tcx> {
}
}

#[expect(clippy::too_many_lines)]
fn binop(&self, op: BinOpKind, left: &Expr<'_>, right: &Expr<'_>) -> Option<Constant> {
let l = self.expr(left)?;
let r = self.expr(right);
Expand Down Expand Up @@ -1003,7 +998,20 @@ impl<'tcx> ConstEvalCtxt<'tcx> {
},
_ => None,
},
// FIXME(f16_f128): add these types when binary operations are available on all platforms
(Constant::F16(l), Some(Constant::F16(r))) => match op {
BinOpKind::Add => Some(Constant::F16(l + r)),
BinOpKind::Sub => Some(Constant::F16(l - r)),
BinOpKind::Mul => Some(Constant::F16(l * r)),
BinOpKind::Div => Some(Constant::F16(l / r)),
BinOpKind::Rem => Some(Constant::F16(l % r)),
BinOpKind::Eq => Some(Constant::Bool(l == r)),
BinOpKind::Ne => Some(Constant::Bool(l != r)),
BinOpKind::Lt => Some(Constant::Bool(l < r)),
BinOpKind::Le => Some(Constant::Bool(l <= r)),
BinOpKind::Ge => Some(Constant::Bool(l >= r)),
BinOpKind::Gt => Some(Constant::Bool(l > r)),
_ => None,
},
(Constant::F32(l), Some(Constant::F32(r))) => match op {
BinOpKind::Add => Some(Constant::F32(l + r)),
BinOpKind::Sub => Some(Constant::F32(l - r)),
Expand Down Expand Up @@ -1032,6 +1040,20 @@ impl<'tcx> ConstEvalCtxt<'tcx> {
BinOpKind::Gt => Some(Constant::Bool(l > r)),
_ => None,
},
(Constant::F128(l), Some(Constant::F128(r))) => match op {
BinOpKind::Add => Some(Constant::F128(l + r)),
BinOpKind::Sub => Some(Constant::F128(l - r)),
BinOpKind::Mul => Some(Constant::F128(l * r)),
BinOpKind::Div => Some(Constant::F128(l / r)),
BinOpKind::Rem => Some(Constant::F128(l % r)),
BinOpKind::Eq => Some(Constant::Bool(l == r)),
BinOpKind::Ne => Some(Constant::Bool(l != r)),
BinOpKind::Lt => Some(Constant::Bool(l < r)),
BinOpKind::Le => Some(Constant::Bool(l <= r)),
BinOpKind::Ge => Some(Constant::Bool(l >= r)),
BinOpKind::Gt => Some(Constant::Bool(l > r)),
_ => None,
},
(l, r) => match (op, l, r) {
(BinOpKind::And, Constant::Bool(false), _) => Some(Constant::Bool(false)),
(BinOpKind::Or, Constant::Bool(true), _) => Some(Constant::Bool(true)),
Expand All @@ -1053,10 +1075,10 @@ pub fn mir_to_const<'tcx>(tcx: TyCtxt<'tcx>, val: ConstValue, ty: Ty<'tcx>) -> O
(ConstValue::Scalar(Scalar::Int(int)), _) => match ty.kind() {
ty::Bool => Some(Constant::Bool(int == ScalarInt::TRUE)),
ty::Uint(_) | ty::Int(_) => Some(Constant::Int(int.to_bits(int.size()))),
ty::Float(FloatTy::F16) => Some(Constant::F16(int.into())),
ty::Float(FloatTy::F16) => Some(Constant::F16(f16::from_bits(int.into()))),
ty::Float(FloatTy::F32) => Some(Constant::F32(f32::from_bits(int.into()))),
ty::Float(FloatTy::F64) => Some(Constant::F64(f64::from_bits(int.into()))),
ty::Float(FloatTy::F128) => Some(Constant::F128(int.into())),
ty::Float(FloatTy::F128) => Some(Constant::F128(f128::from_bits(int.into()))),
ty::RawPtr(_, _) => Some(Constant::RawPtr(int.to_bits(int.size()))),
_ => None,
},
Expand All @@ -1076,10 +1098,10 @@ pub fn mir_to_const<'tcx>(tcx: TyCtxt<'tcx>, val: ConstValue, ty: Ty<'tcx>) -> O
let range = alloc_range(offset + size * idx, size);
let val = alloc.read_scalar(&tcx, range, /* read_provenance */ false).ok()?;
res.push(match flt {
FloatTy::F16 => Constant::F16(val.to_u16().discard_err()?),
FloatTy::F16 => Constant::F16(f16::from_bits(val.to_u16().discard_err()?)),
FloatTy::F32 => Constant::F32(f32::from_bits(val.to_u32().discard_err()?)),
FloatTy::F64 => Constant::F64(f64::from_bits(val.to_u64().discard_err()?)),
FloatTy::F128 => Constant::F128(val.to_u128().discard_err()?),
FloatTy::F128 => Constant::F128(f128::from_bits(val.to_u128().discard_err()?)),
});
}
Some(Constant::Vec(res))
Expand Down
2 changes: 2 additions & 0 deletions clippy_utils/src/lib.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,6 @@
#![feature(box_patterns)]
#![feature(f128)]
#![feature(f16)]
#![feature(if_let_guard)]
#![feature(macro_metavar_expr)]
#![feature(never_type)]
Expand Down
2 changes: 0 additions & 2 deletions tests/ui/arithmetic_side_effects.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -164,11 +164,9 @@ pub fn association_with_structures_should_not_trigger_the_lint() {

pub fn hard_coded_allowed() {
let _ = 1f16 + 1f16;
//~^ arithmetic_side_effects
let _ = 1f32 + 1f32;
let _ = 1f64 + 1f64;
let _ = 1f128 + 1f128;
//~^ arithmetic_side_effects

let _ = Saturating(0u32) + Saturating(0u32);
let _ = String::new() + "";
Expand Down
Loading