Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-8024830

SEGV in org.apache.lucene.codecs.compressing.CompressingTermVectorsReader.get

    Details

    • Subcomponent:
    • Resolved In Build:
      b59
    • CPU:
      x86
    • OS:
      linux
    • Verification:
      Not verified

      Backports

        Description


        Someone tweeted a link to a Lucene crash with 7u40 (hs24.0-b56):

        https://issues.apache.org/jira/browse/LUCENE-5212

        The issue in the Apache JIRA has a hs_err attached.




        1. assembl_crush.txt
          595 kB
          Vladimir Kozlov
        2. cpuinfo.0
          4 kB
          Igor Ignatyev
        3. cpuinfo.1
          20 kB
          Igor Ignatyev
        4. cpuinfo.2
          4 kB
          Igor Ignatyev
        5. hs_err_pid12412.log
          52 kB
          Igor Ignatyev
        6. hs_err_pid26503.log
          29 kB
          Vladimir Kozlov
        7. hs_err_pid32714.log
          58 kB
          Vladimir Kozlov

          Issue Links

            Activity

            Hide
            kvn Vladimir Kozlov added a comment -
            It is very hard to get small test (to make vector spill in particular stack slot). I used lucene tests to verify the fix.
            After installing lucene test suite and setting up proxies as described above I run next script in lucene/core directory.
            Note, you need to run 64-bit PRODUCT VM on SandyBridge x86 (it is important for vectors generation):

            $ cat run.sh
            #!/bin/sh
            COUNT=1
            while [ $COUNT -le 20 ]
            do
              echo "COUNT: " $COUNT
              $JAVA_HOME/bin/java -d64 -version
              \rm -rf ../../.caches/
              ant test -Dtests.seed=43A1116E7F98BED4 -Dtests.jvms=1 -Dargs="-d64 -XX:-UseCompressedOops -XX:+UseParallelGC -XX:-TieredCompilation"
              if [ -d ../build/core/test/J0 ]
              then
                echo "FAILED!!!!!!!!!!!!!!!!"
                echo "COUNT: " $COUNT
                break
              fi
              COUNT=`expr $COUNT + 1`
            done

            You should hit one of the failure (crush or "CheckIndex failed" test error) in few iterations.
            Show
            kvn Vladimir Kozlov added a comment - It is very hard to get small test (to make vector spill in particular stack slot). I used lucene tests to verify the fix. After installing lucene test suite and setting up proxies as described above I run next script in lucene/core directory. Note, you need to run 64-bit PRODUCT VM on SandyBridge x86 (it is important for vectors generation): $ cat run.sh #!/bin/sh COUNT=1 while [ $COUNT -le 20 ] do   echo "COUNT: " $COUNT   $JAVA_HOME/bin/java -d64 -version   \rm -rf ../../.caches/   ant test -Dtests.seed=43A1116E7F98BED4 -Dtests.jvms=1 -Dargs="-d64 -XX:-UseCompressedOops -XX:+UseParallelGC -XX:-TieredCompilation"   if [ -d ../build/core/test/J0 ]   then     echo "FAILED!!!!!!!!!!!!!!!!"     echo "COUNT: " $COUNT     break   fi   COUNT=`expr $COUNT + 1` done You should hit one of the failure (crush or "CheckIndex failed" test error) in few iterations.
            Hide
            kvn Vladimir Kozlov added a comment -
            Customer verified the fix:

            I confirm that this patch fixes the problem encountered in
            LUCENE-5212. I've tested svn rev. 1523179 (trunk) against jdk8-b114
            with and without Vladimir's patch. Without the patch the test sequence
            ends about 50% of the time in a sigsegv. With the patch all executions
            ended without any errors.
            Show
            kvn Vladimir Kozlov added a comment - Customer verified the fix: I confirm that this patch fixes the problem encountered in LUCENE-5212. I've tested svn rev. 1523179 (trunk) against jdk8-b114 with and without Vladimir's patch. Without the patch the test sequence ends about 50% of the time in a sigsegv. With the patch all executions ended without any errors.
            Hide
            hgupdate HG Updates added a comment -
            URL: http://hg.openjdk.java.net/hsx/hotspot-comp/hotspot/rev/1dcea64e9f00
            User: kvn
            Date: 2013-11-11 22:40:40 +0000
            Show
            hgupdate HG Updates added a comment - URL: http://hg.openjdk.java.net/hsx/hotspot-comp/hotspot/rev/1dcea64e9f00 User: kvn Date: 2013-11-11 22:40:40 +0000
            Hide
            hgupdate HG Updates added a comment -
            URL: http://hg.openjdk.java.net/hsx/hsx25/hotspot/rev/1dcea64e9f00
            User: amurillo
            Date: 2013-11-15 18:50:51 +0000
            Show
            hgupdate HG Updates added a comment - URL: http://hg.openjdk.java.net/hsx/hsx25/hotspot/rev/1dcea64e9f00 User: amurillo Date: 2013-11-15 18:50:51 +0000
            Hide
            afomin Alexander Fomin added a comment -
            SQE OK to take the fix into CPU14_02
            Show
            afomin Alexander Fomin added a comment - SQE OK to take the fix into CPU14_02

              People

              • Assignee:
                kvn Vladimir Kozlov
                Reporter:
                alanb Alan Bateman
              • Votes:
                0 Vote for this issue
                Watchers:
                15 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: