RFR(S): 8146011: sun/management/jmxremote/bootstrap/CustomLauncherTest crash at assert(stack_size)

[Coleen Phillimore]

I've reproduced this and I think this is the fix:

Summary: We were setting stack_overflow_limit before initialization 
completed which may change the stack base for some solaris systems with 
unlimited stack

open webrev at http://cr.openjdk.java.net/~coleenp/8146011.01/
bug link https://bugs.openjdk.java.net/browse/JDK-8146011

Tested with sun/management/jmxremote/bootsrap tests with high stack 
limit.  Also, running with JPRT.

Thanks,
Coleen

On 12/23/15 3:43 AM, Lindenmaier, Goetz wrote:
> Hi,
>
> This should fix the given issue as it is closer to the original code.
> I had replaced
>     thread->stack_base() - thread->stack_size() + ((StackShadowPages + StackRedPages + StackYellowPages + StackReservedPages) * vm_page_size())
> by
>     stack_overflow_limit()
> which is supposed to point at the address computed above.
>
> This change reverts this back to
>       thread->stack_base() - thread->stack_size() + JavaThread::stack_guard_zone_size() + JavaThread::stack_shadow_zone_size()
>
> Unfortunately I don't have further information about the bug or test to have a closer look.
>
> Please review this change.  Please check it solves the issue reported.
> I please need a sponsor.
> http://cr.openjdk.java.net/~goetz/webrevs/8146011-fixStack/webrrev.00/
>
> Best regards,
>    Goetz.