Heads-up: jdk/jdk CTW test failures

[Zhengyu Gu]

Saw this before, likely called nmethod::oops_marking_prologue/epilogue 
twice. Which repo?

Thanks,

-Zhengyu

On 5/18/19 8:27 AM, Aleksey Shipilev wrote:
> Lots of CTW failures in jdk/jdk with Shenandoah enabled. Seems to be problematic with Traversal:
> 
> $ CONF=linux-x86_64-server-fastdebug make run-test TEST=applications/ctw/modules/
> TEST_VM_OPTS="-XX:+UnlockExperimentalVMOptions -XX:+UnlockDiagnosticVMOptions -XX:+UseShenandoahGC
> -XX:-TieredCompilation -XX:+ShenandoahVerifyOptoBarriers -XX:ShenandoahGCHeuristics=traversal"
> 
> Sample hs_errs:
>   http://cr.openjdk.java.net/~shade/shenandoah/crashes/ctw-traversal/
> 
> #
> # A fatal error has been detected by the Java Runtime Environment:
> #
> #  Internal Error (/home/shade/trunks/jdk-jdk/src/hotspot/cpu/x86/relocInfo_x86.cpp:56), pid=25735,
> tid=25748
> #  guarantee(*(uint32_t*) disp == CompressedOops::encode((oop)x)) failed: instructions must match
> 
> ...
> V  [libjvm.so+0x1559b99]  Relocation::pd_set_data_value(unsigned char*, long, bool)+0x369
> V  [libjvm.so+0x9dcfb3]  CompiledMethod::verify_oop_relocations()+0x133
> V  [libjvm.so+0x138f962]  nmethod::oops_do_marking_epilogue()+0x72
> V  [libjvm.so+0x1694541]  ShenandoahTraversalGC::init_traversal_collection()+0x381
> V  [libjvm.so+0x163e8c1]  ShenandoahHeap::entry_init_traversal()+0x101
> V  [libjvm.so+0x16c618a]  VM_ShenandoahInitTraversalGC::doit()+0x2a
> V  [libjvm.so+0x18d7b21]  VM_Operation::evaluate()+0x1f1
> V  [libjvm.so+0x19080f0]  VMThread::evaluate_operation(VM_Operation*) [clone .constprop.64]+0x180
> V  [libjvm.so+0x1908b4c]  VMThread::loop()+0x71c
> V  [libjvm.so+0x190919a]  VMThread::run()+0xca
> V  [libjvm.so+0x1813b66]  Thread::call_run()+0xf6
> V  [libjvm.so+0x141c40e]  thread_native_entry(Thread*)+0x10e
>