Troubles with Shenandoah

[Aleksey Shipilev]

On 4/7/19 11:09 PM, Aleksey Shipilev wrote:
> On 4/7/19 10:57 PM, Simone Bordet wrote:
>> On Sun, Apr 7, 2019 at 7:52 PM Simone Bordet <simone.bordet at gmail.com> wrote:
>>> Will try the fastdebug build and report back.
>>
>> I got a crash with fastdebug, attached hs_err.
>>
>> Let me know if it's helpful.
>> I'm interested in the details of the failure, if the file has enough
>> information.
> 
> I think it has.
> 
> Signal is raised trying to access 0x00007f006b8a54fc:
>   siginfo: si_signo: 11 (SIGSEGV), si_code: 2 (SEGV_ACCERR), si_addr: 0x00007f006b8a54fc
> 
> ...which looks to come from %rbx:
>   RBX=0x00007f006b8a54fc: <offset 0x00000000000e04fc> in /lib/x86_64-linux-gnu/libnsl.so.1 at
> 0x00007f006b7c5000
> 
> ...and Instructions section disassembly shows this:
>  https://onlinedisassembler.com/odaweb/uV5P7D1q/0
> 
>  00000016 48 bb fc 54 8a 6b 00 7f 00 00    movabs rbx,0x7f006b8a54fc
>  00000020 80 3c 0b 00                      cmp    BYTE PTR [rbx+rcx*1],0x0 <-- SEGV here
> 
> It looks like bad constant oop in the generated code. -XX:ScavengeRootsInCode=0 might be a plausible
> workaround.

FYI, this one is actually:
  https://bugs.openjdk.java.net/browse/JDK-8222838

...and it is fixed and backported to 12u, 11u, 8u.

-Aleksey