Request for Approval: Backport of 8047212 : runtime/ParallelClassLoading/bootstrap/random/inner-complex assert(ObjectSynchronizer::verify_objmon_isinpool(inf)) failed: monitor is invalid

[Andrew Hughes]

On 20/08/2019 03:43, Yangfei (Felix) wrote:
> Hi,
> 
> 
> 
> May I got review for the backport of 8146792 to 8u master repo please?
> 
> 
> 
> Webrev: http://cr.openjdk.java.net/~fyang/8047212-8u-backport/webrev.00/
> 
> 
> 
>        Bug: https://bugs.openjdk.java.net/browse/JDK-8047212
> 
>        JDK9 Changeset: http://hg.openjdk.java.net/jdk9/jdk9/hotspot/rev/a0c7a69277da
> 
> 
> 
> This bug can always be reproduced by running jcstress test on our 64/128-core aarch64 server platform with 8u aarch64 fastdebug build.
> 
> 
> 
> Error messge:
> 
> 
> 
> #
> 
> # A fatal error has been detected by the Java Runtime Environment:
> 
> #
> 
> #  Internal Error (/home/yangfei/openjdk8u-aarch64/hotspot/src/share/vm/runtime/synchronizer.cpp:1252), pid=16581, tid=0x0000ffff23c7f200
> 
> #  assert(ObjectSynchronizer::verify_objmon_isinpool(inf)) failed: monitor is invalid
> 
> #
> 
> # JRE version: OpenJDK Runtime Environment (8.0) (build 1.8.0-internal-fastdebug-yangfei_2019_08_18_14_24-b00)
> 
> # Java VM: OpenJDK 64-Bit Server VM (25.71-b00-fastdebug mixed mode linux-aarch64 compressed oops)
> 
> # Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
> 
> #
> 
> # If you would like to submit a bug report, please visit:
> 
> #   http://bugreport.java.com/bugreport/crash.jsp
> 
> #
> 
> 
> 
> ---------------  T H R E A D  ---------------
> 
> 
> 
> Current thread (0x0000ffff24123000):  JavaThread "worker1" daemon [_thread_in_Java, id=16798, stack(0x0000ffff23a7f000,0x0000ffff23c80000)]
> 
> 
> 
> Stack: [0x0000ffff23a7f000,0x0000ffff23c80000],  sp=0x0000ffff23c7dac0,  free space=2042k
> 
> Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
> 
> V  [libjvm.so+0x1117444]  VMError::report_and_die()+0x144
> 
> V  [libjvm.so+0x703ba4]  report_vm_error(char const*, int, char const*, char const*)+0x7c
> 
> V  [libjvm.so+0x1032a78]  ObjectSynchronizer::inflate(Thread*, oop)+0xbc8
> 
> V  [libjvm.so+0x1036b40]  ObjectSynchronizer::slow_exit(oop, BasicLock*, Thread*)+0x110
> 
> V  [libjvm.so+0xeeb340]  SharedRuntime::complete_monitor_unlocking_C(oopDesc*, BasicLock*)+0x130
> 
> J 73% C2 org.openjdk.jcstress.tests.seqcst.sync.L1_S1__S1__S1__S2_L2__Test_jcstress.actor3()Ljava/lang/Void; (166 bytes) @ 0x0000ffff782c4318 [0x0000ffff782c2
> 
> f80+0x1398]
> 
> C  0x0000ffff2e872650
> 
> 
> 
>    As the patch changes shared code, it has to be backported to 8u master repo before it goes to 8u-aarch64 repo.
> 
>       Due to the use of 'PaddedEnd' in jdk9 or higher version, some trivial tweaks are necessary for the backport.
> 
>    As we changed to use ordered access for global `gBlockList`, risk should be low.
> 
>    Jtreg tested with 8u x86_64 fastdebug build.  Also passed two round of jcstress test with 8u aarch64 fastdebug build.
> 
> 
> 
> Thanks for your help,
> 
> Felix
> 

PaddedEnd is introduced by JDK-8049737, which is part of JEP-143, so I
agree it makes sense to workaround this change rather than backport it.
The code is essentially the same, minus the wrapping of the
ObjectMonitor in PaddedEnd.

Some minor issues:

* Empty line lost between:

+        oop obj = (oop)mid->object();

and

+        if (obj == NULL) {

* Empty line lost between:

+        deflated = deflate_monitor(mid, obj, &FreeHead, &FreeTail);

and

+        if (deflated) {

Both are in synchronizer.cpp and cause the diffs to not line up when
compared with the original 9u changeset.

Thanks,
-- 
Andrew :)

Senior Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

PGP Key: ed25519/0xCFDA0F9B35964222 (hkp://keys.gnupg.net)
Fingerprint = 5132 579D D154 0ED2 3E04  C5A0 CFDA 0F9B 3596 4222