rust-clippy/src/mutex_atomic.rs
Manish Goregaokar c9342d0121 fmt clippy
2016-01-04 09:56:12 +05:30

77 lines
2.9 KiB
Rust

//! Checks for uses of Mutex where an atomic value could be used
//!
//! This lint is **warn** by default
use rustc::lint::{LintPass, LintArray, LateLintPass, LateContext};
use rustc_front::hir::Expr;
use syntax::ast;
use rustc::middle::ty;
use rustc::middle::subst::ParamSpace;
use utils::{span_lint, MUTEX_PATH, match_type};
/// **What it does:** It `Warn`s on usages of `Mutex<X>` where an atomic will do
///
/// **Why is this bad?** Using a Mutex just to make access to a plain bool or reference sequential is shooting flies with cannons. `std::atomic::AtomicBool` and `std::atomic::AtomicPtr` are leaner and faster.
///
/// **Known problems:** This lint cannot detect if the Mutex is actually used for waiting before a critical section.
///
/// **Example:** `let x = Mutex::new(&y);`
declare_lint! {
pub MUTEX_ATOMIC,
Warn,
"using a Mutex where an atomic value could be used instead"
}
/// **What it does:** It `Warn`s on usages of `Mutex<X>` where `X` is an integral type.
///
/// **Why is this bad?** Using a Mutex just to make access to a plain integer sequential is shooting flies with cannons. `std::atomic::usize` is leaner and faster.
///
/// **Known problems:** This lint cannot detect if the Mutex is actually used for waiting before a critical section.
///
/// **Example:** `let x = Mutex::new(0usize);`
declare_lint! {
pub MUTEX_INTEGER,
Allow,
"using a Mutex for an integer type"
}
impl LintPass for MutexAtomic {
fn get_lints(&self) -> LintArray {
lint_array!(MUTEX_ATOMIC, MUTEX_INTEGER)
}
}
pub struct MutexAtomic;
impl LateLintPass for MutexAtomic {
fn check_expr(&mut self, cx: &LateContext, expr: &Expr) {
let ty = cx.tcx.expr_ty(expr);
if let ty::TyStruct(_, subst) = ty.sty {
if match_type(cx, ty, &MUTEX_PATH) {
let mutex_param = &subst.types.get(ParamSpace::TypeSpace, 0).sty;
if let Some(atomic_name) = get_atomic_name(mutex_param) {
let msg = format!("Consider using an {} instead of a Mutex here. If you just want the locking \
behaviour and not the internal type, consider using Mutex<()>.",
atomic_name);
match *mutex_param {
ty::TyUint(t) if t != ast::TyUs => span_lint(cx, MUTEX_INTEGER, expr.span, &msg),
ty::TyInt(t) if t != ast::TyIs => span_lint(cx, MUTEX_INTEGER, expr.span, &msg),
_ => span_lint(cx, MUTEX_ATOMIC, expr.span, &msg),
};
}
}
}
}
}
fn get_atomic_name(ty: &ty::TypeVariants) -> Option<(&'static str)> {
match *ty {
ty::TyBool => Some("AtomicBool"),
ty::TyUint(_) => Some("AtomicUsize"),
ty::TyInt(_) => Some("AtomicIsize"),
ty::TyRawPtr(_) => Some("AtomicPtr"),
_ => None,
}
}