Request for review (S) 6780143: hs203t003 hits SIGSEGV/EXCEPTION_ACCESS_VIOLATION with -XX:+UseCompressedOops

Vladimir Kozlov vladimir.kozlov at oracle.com
Fri Nov 19 12:04:29 PST 2010


Good.

Vladimir

Coleen Phillimore wrote:
> 
> It's not used later in the code so it didn't need to be saved.
> 
> Thanks!
> Coleen
> 
> Vladimir Kozlov wrote:
>> Coleen,
>>
>> Looks goor but could you explain why you removed saving RCX in 
>> templateTable_x86_64.cpp?
>>
>> Thanks,
>> Vladimir
>>
>> Coleen Phillimore wrote:
>>>
>>> Thanks to Tom for pointing out what was really killing r12, I have a 
>>> new webrev with a better fix and better summary.
>>>
>>> Summary: Using r12 as temporary register around call_VM trashes 
>>> heapbase because call_VM doesn't always return immediately to 
>>> following code (eg forward_exception, popframe and early return 
>>> support).
>>>
>>> open webrev at http://cr.openjdk.java.net/~coleenp/6780143/   (hit 
>>> reload)
>>> bug link at http://bugs.sun.com/view_bug.do?bug_id=6780143
>>>
>>> Thanks,
>>> Coleen
>>>
>>> Tom Rodriguez wrote:
>>>> The changes look ok but what code was killing r12?  All the existing 
>>>> reinit_heapbase calls are because we're using the callee-saved-ness 
>>>> of r12 to keep something else alive but that's no true here.
>>>>
>>>> tom
>>>>
>>>> On Nov 18, 2010, at 12:20 PM, coleen phillimore wrote:
>>>>
>>>>  
>>>>> Summary: missing reinit_heapbase() when popframe support jumps back 
>>>>> into interpreter from C code.
>>>>>
>>>>> open webrev at http://cr.openjdk.java.net/~coleenp/6780143/
>>>>> bug link at http://bugs.sun.com/view_bug.do?bug_id=6780143
>>>>>
>>>>> Tested by running our jvmti tests including the failing test.
>>>>>
>>>>> Coleen
>>>>>     
>>>>
>>>>   


More information about the hotspot-runtime-dev mailing list