2024-08-14 20:41:00 +02:00
|
|
|
// Copyright 2024 the V8 project authors. All rights reserved.
|
|
|
|
// Use of this source code is governed by a BSD-style license that can be
|
|
|
|
// found in the LICENSE file.
|
|
|
|
|
2025-04-29 08:03:15 +02:00
|
|
|
// Flags: --allow-natives-syntax --turbolev --turbofan
|
2024-08-14 20:41:00 +02:00
|
|
|
|
|
|
|
function dematerialized_double_array(n) {
|
|
|
|
let arr = [ 1.1, 2.2, 3.3 ];
|
|
|
|
if (n < 0) {
|
|
|
|
// We won't execute this code when collecting feedback, which means that
|
|
|
|
// Maglev will do an unconditional deopt here (and {arr} will thus have no
|
|
|
|
// use besides the frame state for this deopt).
|
|
|
|
n + 5;
|
|
|
|
return arr;
|
|
|
|
}
|
|
|
|
return n + 2;
|
|
|
|
}
|
|
|
|
|
|
|
|
%PrepareFunctionForOptimization(dematerialized_double_array);
|
|
|
|
assertEquals(7, dematerialized_double_array(5));
|
|
|
|
assertEquals(7, dematerialized_double_array(5));
|
|
|
|
|
|
|
|
%OptimizeFunctionOnNextCall(dematerialized_double_array);
|
|
|
|
assertEquals(7, dematerialized_double_array(5));
|
|
|
|
assertEquals([1.1, 2.2, 3.3], dematerialized_double_array(-1));
|