Java 1.7/1.8 JIT Compiler broken?

Question

I have a problem with some code from GlazedList 1.8 that causes a SIGSEGV when running under java 1.8_05/64 bit/FC20 & Windows 8.

I have the disassembled output (-XX:+UnlockDiagnosticVMOptions '-XX:CompileCommand=print,*BoyerMooreCaseInsensitiveTextSearchStrategy.indexOf' see below) but I no clue on how to debug it.

So any help with debugging the code or a hint to where to ask for help is appreciated.

The disassembled code is more than 30000 chars. long so you will have to go here https://java.net/jira/browse/GLAZEDLISTS-564?focusedCommentId=378982&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-378982 to read the code

A fatal error has been detected by the Java Runtime Environment:

SIGSEGV (0xb) at pc=0x00007fdc2d93bcfc, pid=12092, tid=140582414018304

JRE version: Java(TM) SE Runtime Environment (8.0_05-b13) (build 1.8.0_05-b13) Java VM: Java HotSpot(TM) 64-Bit Server VM (25.5-b02 mixed mode linux-amd64 compressed oops)

Problematic frame:

J 12756 C2

ca.odell.glazedlists.impl.filter.BoyerMooreCaseInsensitiveTextSearchStrategy.indexOf(Ljava/lang/String;)I (147 bytes) @ 0x00007fdc2d93bcfc [0x00007fdc2d93baa0+0x25c]

Answer 1

This is indeed a JIT compiler bug. I've verified that it exists in JDK 7u67, 8u11 as well as in the latest JDK 9 sources. Here is the reduced test case:

public class CharArrayCrash {
    static char[] pattern0 = {0};
    static char[] pattern1 = {1};

    static void test(char[] array) {
        if (pattern1 == null) return;

        int i = 0;
        int pos = 0;
        char c = array[pos];

        while (i >= 0 && (c == pattern0[i] || c == pattern1[i])) {
            i--;
            pos--;
            if (pos != -1) {
                c = array[pos];
            }
        }
    }

    public static void main(String[] args) {
        for (int i = 0; i < 1000000; i++) {
            test(new char[1]);
        }
    }
}

The crash happens on array access instruction where the array offset is illegal (0xffffffff).
Seems like JIT incorrectly reorders decrement and array load instructions.

Anyway, I've submitted the bug report to Oracle: https://bugs.openjdk.java.net/browse/JDK-8054478