Re: [meta-java] Trying to update openjdk-7 to icedtea 2.5.4 and jdk75b13

[Sven Ebenfeld <sven.ebenfeld@gmail.com>]

Hi Otavio,

Am 03.04.2015 um 03:56 schrieb Otavio Salvador:
> Hello Sven,
> 
> On Thu, Apr 2, 2015 at 5:25 PM, Sven Ebenfeld <sven.ebenfeld@gmail.com> wrote:
>> I've managed to build and run OpenJDK-7-75b13 on an armv7 target but now
>> I run into an issue reported all around in the net:
>> http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=1851
>>
>> I'm adressing the same issue with the quartz library on my target as in
>> the last comment. For this specific case it looks like the segfault is
>> thrown on a call to Object.clone() within the quartz library.
>> Unfortunately I don't find any bugfix or other hint for fixing that
>> issue. Other sites state, that "-XX:+PrintCommandLineFlags" works around
>> that issue. This is right but it makes the VM painfully slow. So it
>> looks like either the ThumbEE or the whole ASM optimizations won't be
>> used with that option.
>> http://blog.timmattison.com/archives/2014/08/29/deal-with-os-linux-zero-dot-cpp-related-jvm-crashes-without-using-the-oracle-jvm/
>>
>> I've looked for solutions a lot but didn't find anything helpful, yet.
>> Any ideas?
>>
>> If anyone wants to play around with my current work. Please have a look
>> at my fork on github https://github.com/esven/meta-java/tree/jdk_update
> 
> It would be good to hook a gdb on this before the failure and try to
> see if something can be identified which might help fixing the root
> cause of the issue?
> 

Running with gdb gives the following output. I can't install debug
symbols for java atm. I will do this next week. But at least gdb tells
us which address it wants to access.

08:01:07.529 [main] INFO  org.quartz.impl.StdSchedulerFactory - Quartz
scheduler 'BoxClient Scheduler' initialized from default file in current
working dir: 'quartz.properties'
08:01:07.531 [main] INFO  org.quartz.impl.StdSchedulerFactory - Quartz
scheduler version: 2.2.1
08:01:07.539 [main] INFO  org.quartz.core.QuartzScheduler - JobFactory
set to: com.in2soft.remote.client.scheduler.SchedulerJobFactory@1b487e4
Cannot access memory at address 0x28

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb6064460 (LWP 1801)]
Cannot access memory at address 0x28
0xb6af138c in ?? ()
   from /usr/lib/jvm/java-7-openjdk/jre/lib/arm/server/libjvm.so
(gdb) bt
#0  0xb6af138c in ?? ()
   from /usr/lib/jvm/java-7-openjdk/jre/lib/arm/server/libjvm.so
#1  0x00000000 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb)


Any clues?