Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-8271417

SIGBUS (0x7) at when out of large pages on a NUMA node

    XMLWordPrintable

    Details

    • Subcomponent:
      gc

      Description

      In the benchmark run preflight checks,

      $ numactl --membind=0 --cpunodebind=0 ./jdk-17/bin/java -Xms150g -Xmx150g -Xmn140g -XX:-PrintWarnings -XX:+UseLargePages -XX:+UseNUMA -XX:-UsePerfData -XX:+UseG1GC -XX:+AlwaysPreTouch -XX:ParallelGCThreads=32 -version

      Charlie had a good idea to try -XX:ParallelGCThreads=1 but it crashes the same way.

      I can repro this on x64 and ARM systems with 17-b32.

      ============

      #
      # A fatal error has been detected by the Java Runtime Environment:
      #
      # SIGBUS (0x7) at pc=0x00007f3b4681bdd0, pid=1424059, tid=1424091
      #
      # JRE version: (17.0+32) (build )
      # Java VM: Java HotSpot(TM) 64-Bit Server VM (17-ea+32-LTS-2679, mixed mode, sharing, tiered, compressed class ptrs, g1 gc, linux-amd64)
      # Problematic frame:
      # V [libjvm.so+0xbd8dd0] os::pretouch_memory(void*, void*, unsigned long)+0x20
      #
      # No core dump will be written. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
      #
      #

      --------------- S U M M A R Y ------------

      Command Line: -Xms150g -Xmx150g -Xmn140g -XX:-PrintWarnings -XX:+UseLargePages -XX:+UseNUMA -XX:-UsePerfData -XX:+UseG1GC -XX:+AlwaysPreTouc
      h -XX:ParallelGCThreads=32

      Host: Intel(R) Xeon(R) Gold 6354 CPU @ 3.00GHz, 72 cores, 503G, Oracle Linux Server release 8.4
      Time: Wed Jul 28 19:08:45 2021 UTC elapsed time: 1.026915 seconds (0d 0h 0m 1s)

      --------------- T H R E A D ---------------

      Current thread (0x00007f3b401342f0): GCTaskThread "GC Thread#28" [stack: 0x00007f1431ac1000,0x00007f1431bc1000] [id=1424091]

      Stack: [0x00007f1431ac1000,0x00007f1431bc1000], sp=0x00007f1431bbfdd8, free space=1019k
      Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
      V [libjvm.so+0xbd8dd0] os::pretouch_memory(void*, void*, unsigned long)+0x20
      V [libjvm.so+0xe3848f] GangWorker::loop()+0x5f
      V [libjvm.so+0xe384ef]
      V [libjvm.so+0xd8babe] Thread::call_run()+0xde
      V [libjvm.so+0xbe2521] thread_native_entry(Thread*)+0xe1


      siginfo: si_signo: 7 (SIGBUS), si_code: 2 (BUS_ADRERR), si_addr: 0x00007f2931c00000

      Registers:
      RAX=0x00007f3b46d75010, RBX=0x00007f3b47aac818, RCX=0x0000000000000000, RDX=0x0000000000200000
      RSP=0x00007f1431bbfdd8, RBP=0x00007f1431bbfdf0, RSI=0x00007f2932000000, RDI=0x00007f2931c00000
      R8 =0x0000000000000000, R9 =0x00000000ffffffff, R10=0x0000000000000000, R11=0x0000000000000000
      R12=0x00007f3b47aac800, R13=0x000000000000001d, R14=0x0000000000000001, R15=0x00007f3b401342f0
      RIP=0x00007f3b4681bdd0, EFLAGS=0x0000000000010216, CSGSFS=0x002b000000000033, ERR=0x0000000000000006
        TRAPNO=0x000000000000000e

        Attachments

          Activity

            People

            Assignee:
            Unassigned Unassigned
            Reporter:
            ecaspole Eric Caspole
            Votes:
            0 Vote for this issue
            Watchers:
            8 Start watching this issue

              Dates

              Created:
              Updated:
              Resolved: