New lint [inefficient_pow]

Author: Centri3

CHANGELOG.md

@@ -4901,6 +4901,7 @@ Released 2018-09-13
 [`index_refutable_slice`]: https://rust-lang.github.io/rust-clippy/master/index.html#index_refutable_slice
 [`indexing_slicing`]: https://rust-lang.github.io/rust-clippy/master/index.html#indexing_slicing
 [`ineffective_bit_mask`]: https://rust-lang.github.io/rust-clippy/master/index.html#ineffective_bit_mask
+[`inefficient_pow`]: https://rust-lang.github.io/rust-clippy/master/index.html#inefficient_pow
 [`inefficient_to_string`]: https://rust-lang.github.io/rust-clippy/master/index.html#inefficient_to_string
 [`infallible_destructuring_match`]: https://rust-lang.github.io/rust-clippy/master/index.html#infallible_destructuring_match
 [`infinite_iter`]: https://rust-lang.github.io/rust-clippy/master/index.html#infinite_iter

clippy_lints/src/declared_lints.rs

@@ -345,6 +345,7 @@ pub(crate) static LINTS: &[&crate::LintInfo] = &[
     crate::methods::GET_LAST_WITH_LEN_INFO,
     crate::methods::GET_UNWRAP_INFO,
     crate::methods::IMPLICIT_CLONE_INFO,
+    crate::methods::INEFFICIENT_POW_INFO,
     crate::methods::INEFFICIENT_TO_STRING_INFO,
     crate::methods::INSPECT_FOR_EACH_INFO,
     crate::methods::INTO_ITER_ON_REF_INFO,

clippy_lints/src/methods/inefficient_pow.rs

@@ -0,0 +1,43 @@
+use clippy_utils::{diagnostics::span_lint_and_sugg, is_from_proc_macro, source::snippet_opt};
+use rustc_ast::{LitIntType, LitKind};
+use rustc_errors::Applicability;
+use rustc_hir::{Expr, ExprKind};
+use rustc_lint::LateContext;
+
+use super::INEFFICIENT_POW;
+
+#[expect(
+    clippy::cast_possible_truncation,
+    clippy::cast_precision_loss,
+    clippy::cast_sign_loss,
+    clippy::float_cmp
+)]
+pub(super) fn check<'tcx>(cx: &LateContext<'tcx>, expr: &'tcx Expr<'tcx>, recv: &Expr<'_>, arg: &Expr<'_>) {
+    if let ExprKind::Lit(kind) = recv.kind
+        && let LitKind::Int(val, suffix) = kind.node
+        && let power_used = f32::log2(val as f32)
+        // Precision loss means it's not a power of two
+        && power_used == (power_used as u32) as f32
+        // `0` would be `1.pow()`, which we shouldn't lint (but should be disallowed in a separate lint)
+        && power_used != 0.0
+        && let power_used = power_used as u32
+        && !is_from_proc_macro(cx, expr)
+        && let Some(arg_snippet) = snippet_opt(cx, arg.span)
+    {
+        let suffix = match suffix {
+            LitIntType::Signed(int) => int.name_str(),
+            LitIntType::Unsigned(int) => int.name_str(),
+            LitIntType::Unsuffixed => "",
+        };
+
+        span_lint_and_sugg(
+            cx,
+            INEFFICIENT_POW,
+            expr.span,
+            "inefficient `.pow()`",
+            "use `<<` instead",
+            format!("1{suffix} << ({arg_snippet} * {power_used})"),
+            Applicability::MaybeIncorrect,
+        );
+    }
+}

clippy_lints/src/methods/mod.rs

@@ -32,6 +32,7 @@ mod get_first;
 mod get_last_with_len;
 mod get_unwrap;
 mod implicit_clone;
+mod inefficient_pow;
 mod inefficient_to_string;
 mod inspect_for_each;
 mod into_iter_on_ref;
@@ -3444,6 +3445,37 @@ declare_clippy_lint! {
     "`format!`ing every element in a collection, then collecting the strings into a new `String`"
 }
 
+declare_clippy_lint! {
+    /// ### What it does
+    /// Checks for calls to `.pow()` that get the power of two of `n`, power of four, etc.
+    ///
+    /// ### Why is this bad?
+    /// It's not, but it's not optimized down to a simple `<<`, thus, it's slower. This lint is
+    /// available for when that additional performance is absolutely necessary, at the cost of
+    /// readability.
+    ///
+    /// ### Known issues
+    /// If the linted `pow` would overflow, the suggested `<<` will give an incorrect value with
+    /// overflow checks off. In these cases, `pow` will return 0 whilst `<<` will continue on as
+    /// usual. It may also fail to compile if `arithmetic_overflow` is denied. If this happens,
+    /// it likely means the original code was incorrect regardless as it would always return `0`.
+    ///
+    /// ### Example
+    /// ```rust,ignore
+    /// let _ = 2u32.pow(n);
+    /// let _ = 32u32.pow(n);
+    /// ```
+    /// Use instead:
+    /// ```rust
+    /// let _ = 1u32 << n;
+    /// let _ = 1u32 << (n * 5);
+    /// ```
+    #[clippy::version = "1.72.0"]
+    pub INEFFICIENT_POW,
+    restriction,
+    "usage of `.pow()` when `<<` is faster"
+}
+
 declare_clippy_lint! {
     /// ### What it does
     /// Checks for usage of `.skip(0)` on iterators.
@@ -3602,6 +3634,7 @@ impl_lint_pass!(Methods => [
     FORMAT_COLLECT,
     STRING_LIT_CHARS_ANY,
     ITER_SKIP_ZERO,
+    INEFFICIENT_POW,
 ]);
 
 /// Extracts a method call name, args, and `Span` of the method name.
@@ -4008,6 +4041,9 @@ impl Methods {
                         unnecessary_lazy_eval::check(cx, expr, recv, arg, "or");
                     }
                 },
+                ("pow", [arg]) => {
+                    inefficient_pow::check(cx, expr, recv, arg);
+                }
                 ("push", [arg]) => {
                     path_buf_push_overwrite::check(cx, expr, arg);
                 },

tests/ui/inefficient_pow.fixed

@@ -0,0 +1,54 @@
+//@run-rustfix
+//@aux-build:proc_macros.rs:proc-macro
+#![allow(
+    arithmetic_overflow,
+    clippy::erasing_op,
+    clippy::identity_op,
+    clippy::no_effect,
+    unused
+)]
+#![warn(clippy::inefficient_pow)]
+
+#[macro_use]
+extern crate proc_macros;
+
+fn main() {
+    _ = 1i32 << (1 * 1);
+    _ = 1i32 << (0 * 1);
+    _ = 1i32 << (3 * 1);
+    _ = 1i32 << (100 * 1);
+    _ = 1i32 << (3 * 2);
+    _ = 1i32 << (3 * 5);
+    _ = 1i32 << (3 * 6);
+    let n = 3;
+    _ = 1i32 << (n * 6);
+    // Overflows, so wrong return value, but that's fine. `arithmetic_overflow` will deny this anyway
+    _ = 1i32 << (3 * 12);
+    _ = 1i32 << (3 * 16);
+    // Don't lint
+    _ = 0i32.pow(3);
+    _ = 1i32.pow(3);
+    external! {
+        _ = 2i32.pow(1);
+        _ = 2i32.pow(0);
+        _ = 2i32.pow(3);
+        _ = 2i32.pow(100);
+        _ = 4i32.pow(3);
+        _ = 32i32.pow(3);
+        _ = 64i32.pow(3);
+        _ = 4096i32.pow(3);
+        _ = 65536i32.pow(3);
+    }
+    with_span! {
+        span
+        _ = 2i32.pow(1);
+        _ = 2i32.pow(0);
+        _ = 2i32.pow(3);
+        _ = 2i32.pow(100);
+        _ = 4i32.pow(3);
+        _ = 32i32.pow(3);
+        _ = 64i32.pow(3);
+        _ = 4096i32.pow(3);
+        _ = 65536i32.pow(3);
+    }
+}

tests/ui/inefficient_pow.rs

@@ -0,0 +1,54 @@
+//@run-rustfix
+//@aux-build:proc_macros.rs:proc-macro
+#![allow(
+    arithmetic_overflow,
+    clippy::erasing_op,
+    clippy::identity_op,
+    clippy::no_effect,
+    unused
+)]
+#![warn(clippy::inefficient_pow)]
+
+#[macro_use]
+extern crate proc_macros;
+
+fn main() {
+    _ = 2i32.pow(1);
+    _ = 2i32.pow(0);
+    _ = 2i32.pow(3);
+    _ = 2i32.pow(100);
+    _ = 4i32.pow(3);
+    _ = 32i32.pow(3);
+    _ = 64i32.pow(3);
+    let n = 3;
+    _ = 64i32.pow(n);
+    // Overflows, so wrong return value, but that's fine. `arithmetic_overflow` will deny this anyway
+    _ = 4096i32.pow(3);
+    _ = 65536i32.pow(3);
+    // Don't lint
+    _ = 0i32.pow(3);
+    _ = 1i32.pow(3);
+    external! {
+        _ = 2i32.pow(1);
+        _ = 2i32.pow(0);
+        _ = 2i32.pow(3);
+        _ = 2i32.pow(100);
+        _ = 4i32.pow(3);
+        _ = 32i32.pow(3);
+        _ = 64i32.pow(3);
+        _ = 4096i32.pow(3);
+        _ = 65536i32.pow(3);
+    }
+    with_span! {
+        span
+        _ = 2i32.pow(1);
+        _ = 2i32.pow(0);
+        _ = 2i32.pow(3);
+        _ = 2i32.pow(100);
+        _ = 4i32.pow(3);
+        _ = 32i32.pow(3);
+        _ = 64i32.pow(3);
+        _ = 4096i32.pow(3);
+        _ = 65536i32.pow(3);
+    }
+}

tests/ui/inefficient_pow.stderr

@@ -0,0 +1,64 @@
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:16:9
+   |
+LL |     _ = 2i32.pow(1);
+   |         ^^^^^^^^^^^ help: use `<<` instead: `1i32 << (1 * 1)`
+   |
+   = note: `-D clippy::inefficient-pow` implied by `-D warnings`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:17:9
+   |
+LL |     _ = 2i32.pow(0);
+   |         ^^^^^^^^^^^ help: use `<<` instead: `1i32 << (0 * 1)`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:18:9
+   |
+LL |     _ = 2i32.pow(3);
+   |         ^^^^^^^^^^^ help: use `<<` instead: `1i32 << (3 * 1)`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:19:9
+   |
+LL |     _ = 2i32.pow(100);
+   |         ^^^^^^^^^^^^^ help: use `<<` instead: `1i32 << (100 * 1)`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:20:9
+   |
+LL |     _ = 4i32.pow(3);
+   |         ^^^^^^^^^^^ help: use `<<` instead: `1i32 << (3 * 2)`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:21:9
+   |
+LL |     _ = 32i32.pow(3);
+   |         ^^^^^^^^^^^^ help: use `<<` instead: `1i32 << (3 * 5)`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:22:9
+   |
+LL |     _ = 64i32.pow(3);
+   |         ^^^^^^^^^^^^ help: use `<<` instead: `1i32 << (3 * 6)`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:24:9
+   |
+LL |     _ = 64i32.pow(n);
+   |         ^^^^^^^^^^^^ help: use `<<` instead: `1i32 << (n * 6)`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:26:9
+   |
+LL |     _ = 4096i32.pow(3);
+   |         ^^^^^^^^^^^^^^ help: use `<<` instead: `1i32 << (3 * 12)`
+
+error: inefficient `.pow()`
+  --> $DIR/inefficient_pow.rs:27:9
+   |
+LL |     _ = 65536i32.pow(3);
+   |         ^^^^^^^^^^^^^^^ help: use `<<` instead: `1i32 << (3 * 16)`
+
+error: aborting due to 10 previous errors
+