rust-clippy/clippy_lints/src/transmuting_null.rs

use crate::consts::{constant_context, Constant};
use crate::utils::{match_qpath, span_lint};
use if_chain::if_chain;
use rustc::hir::{Expr, ExprKind};
use rustc::lint::{in_external_macro, LateContext, LateLintPass, LintArray, LintContext, LintPass};
use rustc::{declare_tool_lint, lint_array};
use syntax::ast::LitKind;

declare_clippy_lint! {
    /// **What it does:** Checks for transmute calls which would receive a null pointer.
    ///
    /// **Why is this bad?** Transmuting a null pointer is undefined behavior.
    ///
    /// **Known problems:** Not all cases can be detected at the moment of this writing.
    /// For example, variables which hold a null pointer and are then fed to a `transmute`
    /// call, aren't detectable yet.
    ///
    /// **Example:**
    /// ```rust
    /// let null_ref: &u64 = unsafe { std::mem::transmute(0 as *const u64) };
    /// ```
    pub TRANSMUTING_NULL,
    correctness,
    "transmutes from a null pointer to a reference, which is undefined behavior"
}

#[derive(Copy, Clone)]
pub struct Pass;

impl LintPass for Pass {
    fn get_lints(&self) -> LintArray {
        lint_array!(TRANSMUTING_NULL,)
    }

    fn name(&self) -> &'static str {
        "TransmutingNull"
    }
}

const LINT_MSG: &str = "transmuting a known null pointer into a reference.";

impl<'a, 'tcx> LateLintPass<'a, 'tcx> for Pass {
    fn check_expr(&mut self, cx: &LateContext<'a, 'tcx>, expr: &'tcx Expr) {
        if in_external_macro(cx.sess(), expr.span) {
            return;
        }

        if_chain! {
            if let ExprKind::Call(ref func, ref args) = expr.node;
            if let ExprKind::Path(ref path) = func.node;
            if match_qpath(path, &["std", "mem", "transmute"]);
            if args.len() == 1;

            then {

                // Catching transmute over constants that resolve to `null`.
                let mut const_eval_context = constant_context(cx, cx.tables);
                if_chain! {
                    if let ExprKind::Path(ref _qpath) = args[0].node;
                    let x = const_eval_context.expr(&args[0]);
                    if let Some(constant) = x;
                    if let Constant::RawPtr(ptr_value) = constant;
                    if ptr_value == 0;
                    then {
                        span_lint(
                            cx,
                            TRANSMUTING_NULL,
                            expr.span,
                            LINT_MSG)
                    }
                }

                // Catching:
                // `std::mem::transmute(0 as *const i32)`
                if_chain! {
                    if let ExprKind::Cast(ref inner_expr, ref _cast_ty) = args[0].node;
                    if let ExprKind::Lit(ref lit) = inner_expr.node;
                    if let LitKind::Int(0, _) = lit.node;
                    then {
                        span_lint(
                            cx,
                            TRANSMUTING_NULL,
                            expr.span,
                            LINT_MSG)
                    }
                }

                // Catching:
                // `std::mem::transmute(std::ptr::null::<i32>())`
                if_chain! {
                    if let ExprKind::Call(ref func1, ref args1) = args[0].node;
                    if let ExprKind::Path(ref path1) = func1.node;
                    if match_qpath(path1, &["std", "ptr", "null"]);
                    if args1.len() == 0;
                    then {
                        span_lint(
                            cx,
                            TRANSMUTING_NULL,
                            expr.span,
                            LINT_MSG)
                    }
                }

                // FIXME:
                // Also catch transmutations of variables which are known nulls.
                // To do this, MIR const propagation seems to be the better tool.
                // Whenever MIR const prop routines are more developed, this will
                // become available. As of this writing (25/03/19) it is not yet.
            }
        }
    }
}