RFR: 8168295: [JVMCI] -XX:+JVMCIPrintProperties should exit after printing
Christian Thalinger
cthalinger at twitter.com
Wed Oct 19 19:30:44 UTC 2016
> On Oct 19, 2016, at 8:13 AM, Doug Simon <doug.simon at oracle.com> wrote:
>
>>
>> On 19 Oct 2016, at 19:36, Christian Thalinger <cthalinger at twitter.com> wrote:
>>
>>
>>> On Oct 19, 2016, at 7:28 AM, Doug Simon <doug.simon at oracle.com> wrote:
>>>
>>>
>>>> On 19 Oct 2016, at 19:14, Christian Thalinger <cthalinger at twitter.com> wrote:
>>>>
>>>>
>>>>> On Oct 19, 2016, at 6:40 AM, Vladimir Kozlov <vladimir.kozlov at oracle.com> wrote:
>>>>>
>>>>> I thought it was intentional :)
>>>>>
>>>>> Original output was similar to JVM flags output:
>>>>
>>>> That was the whole point.
>>>
>>> Sure, but there’s no particularly good reason for it is there? They are not printed by the same flag and they are system properties, not VM flags.
>>
>> Sure. I’m not against the change. I’m just saying that we did that because we wanted to keep the output as similar as possible. And we still could do that.
>
> It would clash with the format used by Graal:
>
> [JVMCI properties]
> String jvmci.Compiler = null
> Selects the system compiler.
> Boolean jvmci.InitTimer = false
> Specifies if initialization timing is enabled.
> Boolean jvmci.PrintConfig = false
> Prints VM configuration available via JVMCI.
> String jvmci.TraceMethodDataFilter = null
> Enables tracing of profiling info when read by JVMCI.
> Empty value: trace all methods
> Non-empty value: trace methods whose fully qualified name contains the value.
> [Graal properties]
> graal.ASMInstructionProfiling = null [String]
> Enables instruction profiling on assembler level. Valid values are a
> comma separated list of supported instructions. Compare with subclasses
> of Assembler.InstructionCounter.
> graal.AlwaysInlineIntrinsics = false [Boolean]
> Unconditionally inline intrinsics
> graal.AlwaysInlineVTableStubs = false [Boolean]
> graal.BenchmarkCountersDumpDynamic = true [Boolean]
>
> Since you’re not against the change, I assume you won’t mind if we go with the new format.
Yes, go ahead. Looks good.
>
>>
>>>
>>>>
>>>>>
>>>>> bool UseCompressedOops = true {lp64_product} {ergonomic}
>>>>> bool UseConcMarkSweepGC = false {product} {default}
>>>>>
>>>>> But on other hand new output looks more clean and I like it.
>>>>>
>>>>> thanks,
>>>>> Vladimir
>>>>>
>>>>> On 10/19/16 6:55 AM, Doug Simon wrote:
>>>>>> Please review this small change that fixes an accidental omission from https://bugs.openjdk.java.net/browse/JDK-8167194. The -XX:+JVMCIPrintProperties flag should cause the VM to exit (like the -Djvmci.PrintFlags system property removed by 8167194 used to). Unlike -XX:+PrintFlagsFinal and -XX:+PrintFlagsInitial, it doesn’t make sense to continue execution after printing the help message triggered by -XX:+JVMCIPrintProperties.
>>>>>>
>>>>>> Include in this patch is a minor change in the format of the help output to match that of Graal[1]. Instead of:
>>>>>>
>>>>>> [JVMCI properties]
>>>>>> String jvmci.Compiler = null
>>>>>> Selects the system compiler.
>>>>>> Boolean jvmci.InitTimer = false
>>>>>> Specifies if initialization timing is enabled.
>>>>>> Boolean jvmci.PrintConfig = false
>>>>>> Prints VM configuration available via JVMCI.
>>>>>> String jvmci.TraceMethodDataFilter = null
>>>>>> Enables tracing of profiling info when read by JVMCI.
>>>>>> Empty value: trace all methods
>>>>>> Non-empty value: trace methods whose fully qualified name contains the value.
>>>>>>
>>>>>> the output is now:
>>>>>>
>>>>>> [JVMCI properties]
>>>>>> jvmci.Compiler = null [String]
>>>>>> Selects the system compiler.
>>>>>> jvmci.InitTimer = false [Boolean]
>>>>>> Specifies if initialization timing is enabled.
>>>>>> jvmci.PrintConfig = false [Boolean]
>>>>>> Prints VM configuration available via JVMCI.
>>>>>> jvmci.TraceMethodDataFilter = null [String]
>>>>>> Enables tracing of profiling info when read by JVMCI.
>>>>>> Empty value: trace all methods
>>>>>> Non-empty value: trace methods whose fully qualified name contains the value.
>>>>>>
>>>>>> The new format makes it easier to scan the available options since the option name is not aligned with the help message.
>>>>>>
>>>>>>
>>>>>> http://cr.openjdk.java.net/~dnsimon/8168295/
>>>>>> https://bugs.openjdk.java.net/browse/JDK-8168295
>>>>>>
>>>>>> -Doug
>>>>>>
>>>>>> [1] https://github.com/graalvm/graal-core/commit/096f2fd304c6937623abc9483fc02862b4fa909e
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/attachments/20161019/8854f4a4/attachment-0001.html>
More information about the hotspot-compiler-dev
mailing list