8144856 : fix assert in CompiledStaticCall::set_to_interpreted

Jamsheed C m jamsheed.c.m at oracle.com
Tue May 10 06:10:32 UTC 2016


Thank you Vladimir.
Best Regards,
Jamsheed

On 5/9/2016 11:58 PM, Vladimir Kozlov wrote:
> Jamsheed, your changes looks good.
>
> Thanks,
> Vladimir
>
> On 5/9/16 4:58 AM, Jamsheed C m wrote:
>> Hi Goetz,
>>
>> Thank you for looking at this.
>>
>> On 5/9/2016 2:22 PM, Lindenmaier, Goetz wrote:
>>> Hi Jamsheed,
>>>
>>> Do you want to avoid that data/destination are changed concurrently
>>> between the two compares? Or do you just want to save costs of the 
>>> call?
>> This change was done in x86 as part of JDK-8062493
>> <https://bugs.openjdk.java.net/browse/JDK-8062493>.  from the comment in
>> the change, it says its meant for first case. but, i couldn't find a
>> case in current hotspot code where this could happen.
>> may be its a precautionary change. ( Chris can help here).
>>>
>>> To assure the first, you need to make the two local variables volatile.
>>> Else the compilers can add reloads after inlining data() and 
>>> jump_destination().
>>> We observed this with the Visual Studio compiler and xlC on aix.
>> i have made the two local variables as volatile.
>> revised webrev: http://cr.openjdk.java.net/~jcm/8144856/webrev.01/
>>
>> Best Regards,
>> Jamsheed
>>>
>>> Besides that the change looks good.
>>>
>>> Best regards,
>>>   Goetz.
>>>
>>>
>>>
>>>> -----Original Message-----
>>>> From: hotspot-compiler-dev [mailto:hotspot-compiler-dev-
>>>> bounces at openjdk.java.net] On Behalf Of Jamsheed C m
>>>> Sent: Freitag, 6. Mai 2016 13:21
>>>> To: hotspot compiler<hotspot-compiler-dev at openjdk.java.net>
>>>> Subject: RFR: 8144856 : fix assert in 
>>>> CompiledStaticCall::set_to_interpreted
>>>>
>>>>
>>>> Hi,
>>>>
>>>> Request for review for an assert cleanup change.
>>>> bug id :https://bugs.openjdk.java.net/browse/JDK-8144856
>>>> webrev:http://cr.openjdk.java.net/~jcm/8144856/webrev.00/
>>>>
>>>> Best Regards,
>>>> Jamsheed
>>>>
>>



More information about the hotspot-compiler-dev mailing list