trans: When coercing to Box<Trait>
or Box<[T]>
, leave datum in its original L-/R-value state.
This fixes a subtle issue where temporaries were being allocated (but not necessarily initialized) to the (parent) terminating scope of a match expression; in particular, the code to zero out the temporary emitted by `datum.store_to` is only attached to the particular match-arm for that temporary, but when going down other arms of the match expression, the temporary may falsely appear to have been initialized, depending on what the stack held at that location, and thus may have its destructor erroneously run at the end of the terminating scope. Test cases to appear in a follow-up commit. Fix #20055
This commit is contained in:
parent
e365e4c054
commit
326443105b
@ -420,9 +420,15 @@ fn apply_adjustments<'blk, 'tcx>(bcx: Block<'blk, 'tcx>,
|
||||
let tcx = bcx.tcx();
|
||||
|
||||
let datum_ty = datum.ty;
|
||||
// Arrange cleanup
|
||||
let lval = unpack_datum!(bcx,
|
||||
datum.to_lvalue_datum(bcx, "unsize_unique_vec", expr.id));
|
||||
|
||||
debug!("unsize_unique_vec expr.id={} datum_ty={} len={}",
|
||||
expr.id, datum_ty.repr(tcx), len);
|
||||
|
||||
// We do not arrange cleanup ourselves; if we already are an
|
||||
// L-value, then cleanup will have already been scheduled (and
|
||||
// the `datum.store_to` call below will emit code to zero the
|
||||
// drop flag when moving out of the L-value). If we are an R-value,
|
||||
// then we do not need to schedule cleanup.
|
||||
|
||||
let ll_len = C_uint(bcx.ccx(), len);
|
||||
let unit_ty = ty::sequence_element_type(tcx, ty::type_content(datum_ty));
|
||||
@ -433,7 +439,7 @@ fn apply_adjustments<'blk, 'tcx>(bcx: Block<'blk, 'tcx>,
|
||||
let base = PointerCast(bcx,
|
||||
base,
|
||||
type_of::type_of(bcx.ccx(), datum_ty).ptr_to());
|
||||
bcx = lval.store_to(bcx, base);
|
||||
bcx = datum.store_to(bcx, base);
|
||||
|
||||
Store(bcx, ll_len, get_len(bcx, scratch.val));
|
||||
DatumBlock::new(bcx, scratch.to_expr_datum())
|
||||
@ -455,13 +461,16 @@ fn apply_adjustments<'blk, 'tcx>(bcx: Block<'blk, 'tcx>,
|
||||
};
|
||||
let result_ty = ty::mk_uniq(tcx, ty::unsize_ty(tcx, unboxed_ty, k, expr.span));
|
||||
|
||||
let lval = unpack_datum!(bcx,
|
||||
datum.to_lvalue_datum(bcx, "unsize_unique_expr", expr.id));
|
||||
// We do not arrange cleanup ourselves; if we already are an
|
||||
// L-value, then cleanup will have already been scheduled (and
|
||||
// the `datum.store_to` call below will emit code to zero the
|
||||
// drop flag when moving out of the L-value). If we are an R-value,
|
||||
// then we do not need to schedule cleanup.
|
||||
|
||||
let scratch = rvalue_scratch_datum(bcx, result_ty, "__uniq_fat_ptr");
|
||||
let llbox_ty = type_of::type_of(bcx.ccx(), datum_ty);
|
||||
let base = PointerCast(bcx, get_dataptr(bcx, scratch.val), llbox_ty.ptr_to());
|
||||
bcx = lval.store_to(bcx, base);
|
||||
bcx = datum.store_to(bcx, base);
|
||||
|
||||
let info = unsized_info(bcx, k, expr.id, unboxed_ty, |t| ty::mk_uniq(tcx, t));
|
||||
Store(bcx, info, get_len(bcx, scratch.val));
|
||||
|
Loading…
Reference in New Issue
Block a user