RFR: 8167194: [JVMCI] no reliable mechanism for querying JVMCI system properties

Vladimir Kozlov vladimir.kozlov at oracle.com
Wed Oct 5 20:21:37 UTC 2016


Looks good.

thanks,
Vladimir

On 10/5/16 1:03 PM, Doug Simon wrote:
>
>> On 05 Oct 2016, at 21:10, Doug Simon <doug.simon at oracle.com> wrote:
>>
>>>
>>> On 05 Oct 2016, at 20:49, Vladimir Kozlov <vladimir.kozlov at oracle.com> wrote:
>>>
>>> On 10/5/16 11:34 AM, Doug Simon wrote:
>>>> I chose system property because these properties are set with the command line system property syntax.
>>>
>>> You mean "were set"? It use normal VM flag now.
>>
>> No. The properties *are* still system properties (e.g., -Dgraal.PrintCompilation=true). The only non-standard way about how we use them is that they are accessed via VM.getSavedProperties() instead of System.getProperty() so that non-trusted code cannot influence JVMCI or Graal.
>>>
>>>> I’d actually prefer shortening it to just “properties” if you think it doesn’t risk any ambiguity.
>>>
>>> For me "shortening" is better - few bytes less in VM size :)
>>
>> Ok, I’ll create a new webrev.
>
> http://cr.openjdk.java.net/~dnsimon/8167194.v2/
>
> -Doug
>


More information about the hotspot-compiler-dev mailing list