Red Hat Bugzilla – Bug 827453
FTBFS: *** glibc detected *** /usr/lib/jvm/java-1.7.0-openjdk-18.104.22.168.ppc64/jre/bin/java: invalid fastbin entry (free): 0x00000ffea801a7c0
Last modified: 2012-12-18 15:04:20 EST
Description of problem:
eclipse-4.2.0-0.20.I201205031800.fc18 fails to build on PPC64:
[java] [javac] Compiling 36 source files to /builddir/build/BUILD/eclipse-build-d747eda66dae13c133f56952466ff0ca30da99bf/build/eclipse-4.2.0-I20120503-1800-src/plugins/org.eclipse.core.runtime.compatibility/@dot
[java] *** glibc detected *** /usr/lib/jvm/java-1.7.0-openjdk-22.214.171.124.ppc64/jre/bin/java: invalid fastbin entry (free): 0x00000ffea801a7c0 ***
[java] *** glibc detected *** /usr/lib/jvm/java-1.7.0-openjdk-126.96.36.199.ppc64/jre/bin/java: malloc(): memory corruption (fast): 0x00000ffea801a7d0 ***
[java] ======= Backtrace: =========
Steps to Reproduce:
1. ppc-koji build --scratch f18 eclipse-4.2.0-0.20.I201205031800.fc18.src.rpm
Created attachment 588474 [details]
build.log, just in case koji deletes the log after a certain time
Reassigning to openjdk as it's a jvm crash happening on ppc only
Assigning to Chris. Chris, looks like a PPC specific crash. Can you please take a look?
Is there a hs_err_pid*log somewhere for this?
Is it repeatable ?
Thanks for any info... (There are a ton of problems with zip/jar files being read while also being written... but can't tell if thats whats going on here.)
This appears to be the same issue as with 560232
There is unfortunately not much the JVM can do here as memory mapping provides a significant performance advantage. If applications expect loaded jar/zip files to change, they should supply the -Dsun.zip.disableMemoryMapping=true switch to the appropriate java invocation. Therefore the fix for this is to add -Dsun.zip.disableMemoryMapping=true in the build.xml file for the java task that is failing.
Switching component to Eclipse.
Closing the bug as eclipse builds no longer trigger the ppc crash.