[Bug 2678] New: java process stuck with 100% CPU when running jflex on ARM32 (works with -Xint or Oracle's)

[bugzilla-daemon at icedtea.classpath.org]

http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=2678

            Bug ID: 2678
           Summary: java process stuck with 100% CPU when running jflex on
                    ARM32 (works with -Xint or Oracle's)
           Product: IcedTea
           Version: 2.6.1
          Hardware: arm
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P5
         Component: IcedTea
          Assignee: gnu.andrew at redhat.com
          Reporter: chewi at gentoo.org
                CC: unassigned at icedtea.classpath.org

Running jflex on ARM32 results in a stuck java process that consumes 100% CPU.
It seems to get stuck somewhere in the "emit" stage, after it outputs "Writing
code to Foo.java".

I have observed this when trying to build jflex on Gentoo Linux. jflex uses a
prebuilt jar to bootstrap itself and you can reproduce the problem by manually
running jflex on the provided LexScan.flex file.

java -jar ../../../lib/jflex-1.6.0.jar --skel skeleton.nested LexScan.flex

Gentoo has jflex 1.6.0 but I have also tried 1.6.1. It works fine if I specify
-Xint. Oracle JDK 1.8.0.60 also works. Using taskset to restrict the process to
one core doesn't help. I am currently trying this with IcedTea 2.6.1 but I did
observe this on an earlier version this year.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/distro-pkg-dev/attachments/20151017/55073a8d/attachment-0001.html>