Skip to content
Branch: master
Find file Copy path
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
320 lines (292 sloc) 12.2 KB
use crate::check::regionck::RegionCtxt;
use crate::hir;
use crate::hir::def_id::DefId;
use rustc::infer::outlives::env::OutlivesEnvironment;
use rustc::infer::{self, InferOk, SuppressRegionErrors};
use rustc::middle::region;
use rustc::traits::{ObligationCause, TraitEngine, TraitEngineExt};
use rustc::ty::subst::{Subst, SubstsRef, UnpackedKind};
use rustc::ty::{self, Ty, TyCtxt};
use crate::util::common::ErrorReported;
use syntax_pos::Span;
/// This function confirms that the `Drop` implementation identified by
/// `drop_impl_did` is not any more specialized than the type it is
/// attached to (Issue #8142).
/// This means:
/// 1. The self type must be nominal (this is already checked during
/// coherence),
/// 2. The generic region/type parameters of the impl's self type must
/// all be parameters of the Drop impl itself (i.e., no
/// specialization like `impl Drop for Foo<i32>`), and,
/// 3. Any bounds on the generic parameters must be reflected in the
/// struct/enum definition for the nominal type itself (i.e.
/// cannot do `struct S<T>; impl<T:Clone> Drop for S<T> { ... }`).
pub fn check_drop_impl(tcx: TyCtxt<'_>, drop_impl_did: DefId) -> Result<(), ErrorReported> {
let dtor_self_type = tcx.type_of(drop_impl_did);
let dtor_predicates = tcx.predicates_of(drop_impl_did);
match dtor_self_type.sty {
ty::Adt(adt_def, self_to_impl_substs) => {
_ => {
// Destructors only work on nominal types. This was
// already checked by coherence, but compilation may
// not have been terminated.
let span = tcx.def_span(drop_impl_did);
&format!("should have been rejected by coherence check: {}", dtor_self_type));
fn ensure_drop_params_and_item_params_correspond<'tcx>(
tcx: TyCtxt<'tcx>,
drop_impl_did: DefId,
drop_impl_ty: Ty<'tcx>,
self_type_did: DefId,
) -> Result<(), ErrorReported> {
let drop_impl_hir_id = tcx.hir().as_local_hir_id(drop_impl_did).unwrap();
// check that the impl type can be made to match the trait type.
tcx.infer_ctxt().enter(|ref infcx| {
let impl_param_env = tcx.param_env(self_type_did);
let tcx = infcx.tcx;
let mut fulfillment_cx = TraitEngine::new(tcx);
let named_type = tcx.type_of(self_type_did);
let drop_impl_span = tcx.def_span(drop_impl_did);
let fresh_impl_substs = infcx.fresh_substs_for_item(drop_impl_span, drop_impl_did);
let fresh_impl_self_ty = drop_impl_ty.subst(tcx, fresh_impl_substs);
let cause = &ObligationCause::misc(drop_impl_span, drop_impl_hir_id);
match infcx
.at(cause, impl_param_env)
.eq(named_type, fresh_impl_self_ty)
Ok(InferOk { obligations, .. }) => {
fulfillment_cx.register_predicate_obligations(infcx, obligations);
Err(_) => {
let item_span = tcx.def_span(self_type_did);
"Implementations of Drop cannot be specialized"
"Use same sequence of generic type and region \
parameters that is on the struct/enum definition",
return Err(ErrorReported);
if let Err(ref errors) = fulfillment_cx.select_all_or_error(&infcx) {
// this could be reached when we get lazy normalization
infcx.report_fulfillment_errors(errors, None, false);
return Err(ErrorReported);
let region_scope_tree = region::ScopeTree::default();
// NB. It seems a bit... suspicious to use an empty param-env
// here. The correct thing, I imagine, would be
// `OutlivesEnvironment::new(impl_param_env)`, which would
// allow region solving to take any `a: 'b` relations on the
// impl into account. But I could not create a test case where
// it did the wrong thing, so I chose to preserve existing
// behavior, since it ought to be simply more
// conservative. -nmatsakis
let outlives_env = OutlivesEnvironment::new(ty::ParamEnv::empty());
/// Confirms that every predicate imposed by dtor_predicates is
/// implied by assuming the predicates attached to self_type_did.
fn ensure_drop_predicates_are_implied_by_item_defn<'tcx>(
tcx: TyCtxt<'tcx>,
drop_impl_did: DefId,
dtor_predicates: &ty::GenericPredicates<'tcx>,
self_type_did: DefId,
self_to_impl_substs: SubstsRef<'tcx>,
) -> Result<(), ErrorReported> {
let mut result = Ok(());
// Here is an example, analogous to that from
// `compare_impl_method`.
// Consider a struct type:
// struct Type<'c, 'b:'c, 'a> {
// x: &'a Contents // (contents are irrelevant;
// y: &'c Cell<&'b Contents>, // only the bounds matter for our purposes.)
// }
// and a Drop impl:
// impl<'z, 'y:'z, 'x:'y> Drop for P<'z, 'y, 'x> {
// fn drop(&mut self) { self.y.set(self.x); } // (only legal if 'x: 'y)
// }
// We start out with self_to_impl_substs, that maps the generic
// parameters of Type to that of the Drop impl.
// self_to_impl_substs = {'c => 'z, 'b => 'y, 'a => 'x}
// Applying this to the predicates (i.e., assumptions) provided by the item
// definition yields the instantiated assumptions:
// ['y : 'z]
// We then check all of the predicates of the Drop impl:
// ['y:'z, 'x:'y]
// and ensure each is in the list of instantiated
// assumptions. Here, `'y:'z` is present, but `'x:'y` is
// absent. So we report an error that the Drop impl injected a
// predicate that is not present on the struct definition.
let self_type_hir_id = tcx.hir().as_local_hir_id(self_type_did).unwrap();
let drop_impl_span = tcx.def_span(drop_impl_did);
// We can assume the predicates attached to struct/enum definition
// hold.
let generic_assumptions = tcx.predicates_of(self_type_did);
let assumptions_in_impl_context = generic_assumptions.instantiate(tcx, &self_to_impl_substs);
let assumptions_in_impl_context = assumptions_in_impl_context.predicates;
// An earlier version of this code attempted to do this checking
// via the traits::fulfill machinery. However, it ran into trouble
// since the fulfill machinery merely turns outlives-predicates
// 'a:'b and T:'b into region inference constraints. It is simpler
// just to look for all the predicates directly.
assert_eq!(dtor_predicates.parent, None);
for (predicate, _) in &dtor_predicates.predicates {
// (We do not need to worry about deep analysis of type
// expressions etc because the Drop impls are already forced
// to take on a structure that is roughly an alpha-renaming of
// the generic parameters of the item definition.)
// This path now just checks *all* predicates via the direct
// lookup, rather than using fulfill machinery.
// However, it may be more efficient in the future to batch
// the analysis together via the fulfill , rather than the
// repeated `contains` calls.
if !assumptions_in_impl_context.contains(&predicate) {
let item_span = tcx.hir().span(self_type_hir_id);
"The requirement `{}` is added only by the Drop impl.",
"The same requirement must be part of \
the struct/enum definition",
result = Err(ErrorReported);
/// This function confirms that the type
/// expression `typ` conforms to the "Drop Check Rule" from the Sound
/// Generic Drop RFC (#769).
/// ----
/// The simplified (*) Drop Check Rule is the following:
/// Let `v` be some value (either temporary or named) and 'a be some
/// lifetime (scope). If the type of `v` owns data of type `D`, where
/// * (1.) `D` has a lifetime- or type-parametric Drop implementation,
/// (where that `Drop` implementation does not opt-out of
/// this check via the `may_dangle`
/// attribute), and
/// * (2.) the structure of `D` can reach a reference of type `&'a _`,
/// then 'a must strictly outlive the scope of v.
/// ----
/// This function is meant to by applied to the type for every
/// expression in the program.
/// ----
/// (*) The qualifier "simplified" is attached to the above
/// definition of the Drop Check Rule, because it is a simplification
/// of the original Drop Check rule, which attempted to prove that
/// some `Drop` implementations could not possibly access data even if
/// it was technically reachable, due to parametricity.
/// However, (1.) parametricity on its own turned out to be a
/// necessary but insufficient condition, and (2.) future changes to
/// the language are expected to make it impossible to ensure that a
/// `Drop` implementation is actually parametric with respect to any
/// particular type parameter. (In particular, impl specialization is
/// expected to break the needed parametricity property beyond
/// repair.)
/// Therefore, we have scaled back Drop-Check to a more conservative
/// rule that does not attempt to deduce whether a `Drop`
/// implementation could not possible access data of a given lifetime;
/// instead Drop-Check now simply assumes that if a destructor has
/// access (direct or indirect) to a lifetime parameter, then that
/// lifetime must be forced to outlive that destructor's dynamic
/// extent. We then provide the `may_dangle`
/// attribute as a way for destructor implementations to opt-out of
/// this conservative assumption (and thus assume the obligation of
/// ensuring that they do not access data nor invoke methods of
/// values that have been previously dropped).
pub fn check_safety_of_destructor_if_necessary<'a, 'tcx>(
rcx: &mut RegionCtxt<'a, 'tcx>,
ty: Ty<'tcx>,
span: Span,
body_id: hir::HirId,
scope: region::Scope,
) -> Result<(), ErrorReported> {
debug!("check_safety_of_destructor_if_necessary typ: {:?} scope: {:?}",
ty, scope);
let parent_scope = match rcx.region_scope_tree.opt_encl_scope(scope) {
Some(parent_scope) => parent_scope,
// If no enclosing scope, then it must be the root scope
// which cannot be outlived.
None => return Ok(()),
let parent_scope = rcx.tcx.mk_region(ty::ReScope(parent_scope));
let origin = || infer::SubregionOrigin::SafeDestructor(span);
let cause = &ObligationCause::misc(span, body_id);
let infer_ok =, rcx.fcx.param_env).dropck_outlives(ty);
debug!("dropck_outlives = {:#?}", infer_ok);
let kinds = rcx.fcx.register_infer_ok_obligations(infer_ok);
for kind in kinds {
match kind.unpack() {
UnpackedKind::Lifetime(r) => rcx.sub_regions(origin(), parent_scope, r),
UnpackedKind::Type(ty) => rcx.type_must_outlive(origin(), ty, parent_scope),
UnpackedKind::Const(_) => {
// Generic consts don't add constraints.
You can’t perform that action at this time.