/* * Copyright (c) 2019, Oracle and/or its affiliates. All rights reserved. * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. */ /** * @test * @bug 8230062 * @summary The IR of this test contains a reduction chain which corresponds to a pack in which the 2nd last element has a usage outside of the optimized loop. * * @run main/othervm -Xbatch -XX:CompileCommand=compileonly,compiler.loopopts.superword.SuperWordIntermediateUse::test * compiler.loopopts.superword.SuperWordIntermediateUse */ package compiler.loopopts.superword; public class SuperWordIntermediateUse { private int iFld; private int[] iArr = new int[1024]; public void test() { int local = 4; /** * Before unrolling this loop: * iFld: AddI 1 = -85 + Phi 1 * local: MulI 2 = Phi 1 * LoadI 3 * * This loop is now unrolled. 'local' is a reduction. The loop is first copied: * iFldCopy: AddI C1 = -85 + Phi C1 * localCopy: MulI C2 = Phi C1 * LoadI C3 * * iFld: AddI 1 = -85 + Phi 1 * local: MulI 2 = Phi 1 * LoadI 3 * * Then, the unnecessary nodes like phis are removed from the original loop by igvn: * (iFldCopy: AddI C1 = -85 + Phi C1) field store optimized away * localCopy: MulI C2 = Phi C1 * LoadI C3 * * iFld: AddI 1 = -85 + MulI C2 * local: MulI 2 = MulI C2 * LoadI 3 -> Input into Phi C1 * * As a result AddI 1 has an input from MulI C2 which isn't the last operation in the reduction chain * Phi C1 -> MulI C2 -> MulI 2 -> Phi C1 and therefore not the last element in a pack. * Additionally, AddI 1 does not belong to the loop being optimized: The store node for iFld is put outside of the loop being optimized. * This triggers the assertion bug when unrolled at least 4 times which creates packs of at least size 4. */ for (int i = 0; i < 1024; i++) { iFld = -85; iFld = iFld + local; local = local * iArr[i]; iArr[i] = 3; // Just used to trigger SuperWord optimization } } public static void main(String[] strArr) { SuperWordIntermediateUse instance = new SuperWordIntermediateUse(); for (int i = 0; i < 1000; i++) { instance.test(); } } }