提交 3dbf9691 编写于 作者: F Felix S. Klock II

For FRU, evaluate field expressions (into scratch temps) before base expression.

Fix #23112.
上级 1fe8f221
......@@ -1494,8 +1494,45 @@ pub fn trans_adt<'a, 'blk, 'tcx>(mut bcx: Block<'blk, 'tcx>,
// panic occur before the ADT as a whole is ready.
let custom_cleanup_scope = fcx.push_custom_cleanup_scope();
// First we trans the base, if we have one, to the dest
if let Some(base) = optbase {
if ty::type_is_simd(bcx.tcx(), ty) {
// Issue 23112: The original logic appeared vulnerable to same
// order-of-eval bug. But, SIMD values are tuple-structs;
// i.e. functional record update (FRU) syntax is unavailable.
//
// To be safe, double-check that we did not get here via FRU.
assert!(optbase.is_none());
// This is the constructor of a SIMD type, such types are
// always primitive machine types and so do not have a
// destructor or require any clean-up.
let llty = type_of::type_of(bcx.ccx(), ty);
// keep a vector as a register, and running through the field
// `insertelement`ing them directly into that register
// (i.e. avoid GEPi and `store`s to an alloca) .
let mut vec_val = C_undef(llty);
for &(i, ref e) in fields {
let block_datum = trans(bcx, &**e);
bcx = block_datum.bcx;
let position = C_uint(bcx.ccx(), i);
let value = block_datum.datum.to_llscalarish(bcx);
vec_val = InsertElement(bcx, vec_val, value, position);
}
Store(bcx, vec_val, addr);
} else if let Some(base) = optbase {
// Issue 23112: If there is a base, then order-of-eval
// requires field expressions eval'ed before base expression.
// First, trans field expressions to temporary scratch values.
let scratch_vals: Vec<_> = fields.iter().map(|&(i, ref e)| {
let datum = unpack_datum!(bcx, trans(bcx, &**e));
(i, datum)
}).collect();
debug_location.apply(bcx.fcx);
// Second, trans the base to the dest.
assert_eq!(discr, 0);
match ty::expr_kind(bcx.tcx(), &*base.expr) {
......@@ -1514,31 +1551,14 @@ pub fn trans_adt<'a, 'blk, 'tcx>(mut bcx: Block<'blk, 'tcx>,
}
}
}
}
debug_location.apply(bcx.fcx);
if ty::type_is_simd(bcx.tcx(), ty) {
// This is the constructor of a SIMD type, such types are
// always primitive machine types and so do not have a
// destructor or require any clean-up.
let llty = type_of::type_of(bcx.ccx(), ty);
// keep a vector as a register, and running through the field
// `insertelement`ing them directly into that register
// (i.e. avoid GEPi and `store`s to an alloca) .
let mut vec_val = C_undef(llty);
for &(i, ref e) in fields {
let block_datum = trans(bcx, &**e);
bcx = block_datum.bcx;
let position = C_uint(bcx.ccx(), i);
let value = block_datum.datum.to_llscalarish(bcx);
vec_val = InsertElement(bcx, vec_val, value, position);
// Finally, move scratch field values into actual field locations
for (i, datum) in scratch_vals.into_iter() {
let dest = adt::trans_field_ptr(bcx, &*repr, addr, discr, i);
bcx = datum.store_to(bcx, dest);
}
Store(bcx, vec_val, addr);
} else {
// Now, we just overwrite the fields we've explicitly specified
// No base means we can write all fields directly in place.
for &(i, ref e) in fields {
let dest = adt::trans_field_ptr(bcx, &*repr, addr, discr, i);
let e_ty = expr_ty_adjusted(bcx, &**e);
......
......@@ -12,11 +12,12 @@ struct S { f0: String, f1: int }
pub fn main() {
let s = "Hello, world!".to_string();
let _s = S {
let s = S {
f0: s.to_string(),
..S {
f0: s,
f1: 23
}
};
assert_eq!(s.f0, "Hello, world!");
}
......@@ -15,8 +15,9 @@ struct S {
pub fn main() {
let s = "Hello, world!".to_string();
let _s = S {
let s = S {
f1: s.to_string(),
f0: s
};
assert_eq!(s.f0, "Hello, world!");
}
// Copyright 2015 The Rust Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
// Checks that functional-record-update order-of-eval is as expected
// even when no Drop-implementations are involved.
use std::sync::atomic::{Ordering, AtomicUsize, ATOMIC_USIZE_INIT};
struct W { wrapped: u32 }
struct S { f0: W, _f1: i32 }
pub fn main() {
const VAL: u32 = 0x89AB_CDEF;
let w = W { wrapped: VAL };
let s = S {
f0: { event(0x01); W { wrapped: w.wrapped + 1 } },
..S {
f0: { event(0x02); w},
_f1: 23
}
};
assert_eq!(s.f0.wrapped, VAL + 1);
let actual = event_log();
let expect = 0x01_02;
assert!(expect == actual,
"expect: 0x{:x} actual: 0x{:x}", expect, actual);
}
static LOG: AtomicUsize = ATOMIC_USIZE_INIT;
fn event_log() -> usize {
LOG.load(Ordering::SeqCst)
}
fn event(tag: u8) {
let old_log = LOG.load(Ordering::SeqCst);
let new_log = (old_log << 8) + tag as usize;
LOG.store(new_log, Ordering::SeqCst);
}
// Copyright 2015 The Rust Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
// Checks that struct-literal expression order-of-eval is as expected
// even when no Drop-implementations are involved.
use std::sync::atomic::{Ordering, AtomicUsize, ATOMIC_USIZE_INIT};
struct W { wrapped: u32 }
struct S { f0: W, _f1: i32 }
pub fn main() {
const VAL: u32 = 0x89AB_CDEF;
let w = W { wrapped: VAL };
let s = S {
_f1: { event(0x01); 23 },
f0: { event(0x02); w },
};
assert_eq!(s.f0.wrapped, VAL);
let actual = event_log();
let expect = 0x01_02;
assert!(expect == actual,
"expect: 0x{:x} actual: 0x{:x}", expect, actual);
}
static LOG: AtomicUsize = ATOMIC_USIZE_INIT;
fn event_log() -> usize {
LOG.load(Ordering::SeqCst)
}
fn event(tag: u8) {
let old_log = LOG.load(Ordering::SeqCst);
let new_log = (old_log << 8) + tag as usize;
LOG.store(new_log, Ordering::SeqCst);
}
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册