RFR (S/M) expose L1_data_cache_line_size for diagnostic/sanity checks (8049717)

[David Holmes]

Hi Dan,

I think the performance comment has lost a bit of context. May I suggest 
just a slight change to the original that puts it into historical 
perspective:

  // Performance concern:
  // OrderAccess::storestore() calls release() which at one time stored 0
  // into the global volatile OrderAccess::Dummy variable.  This store was
  // unnecessary for correctness. Many threads storing into a common 
location
  // causes considerable cache migration or "sloshing" on large SMP systems.
  // As such, I avoided using OrderAccess::storestore(). In some cases
  // OrderAccess::fence() -- which incurs local latency on the executing 
processor
  //  -- is a better choice as it scales on SMP systems.  See
  // http://blogs.sun.com/dave/entry/biased_locking_in_hotspot for a 
discussion
  // of coherency costs.  Note that all our current reference platforms 
provide
  // strong ST-ST order, so the issue is moot on IA32, x64, and SPARC.


Though I also note that there is no a single OrderAccess call in 
synchronizer.cpp :)

Thanks,
David

On 15/07/2014 3:47 AM, Daniel D. Daugherty wrote:
> Greetings,
>
> I've tweaked the fix based on David H's feedback.
>
> Here is the URL for the third round webrev:
>
> http://cr.openjdk.java.net/~dcubed/8049717-webrev/2-jdk9-hs-rt/
>
> The following files changed between round 1 and round 2:
>
> src/share/vm/runtime/objectMonitor.cpp
> src/share/vm/runtime/synchronizer.cpp
>
> Thanks, in advance, for any comments, questions or suggestions.
>
> Dan
>
>
> On 7/11/14 2:14 PM, Daniel D. Daugherty wrote:
>> Greetings,
>>
>> I've tweaked the fix based on Vladimir's feedback.
>>
>> Here is the URL for the second round webrev:
>>
>> http://cr.openjdk.java.net/~dcubed/8049717-webrev/1-jdk9-hs-rt/
>>
>> The following files changed between round 0 and round 1:
>>
>> src/cpu/sparc/vm/vm_version_sparc.cpp
>> src/cpu/x86/vm/vm_version_x86.cpp
>> src/cpu/x86/vm/vm_version_x86.hpp
>> src/share/vm/runtime/objectMonitor.cpp
>> src/share/vm/runtime/synchronizer.cpp
>>
>> Thanks, in advance, for any comments, questions or suggestions.
>>
>> Dan
>>
>>
>> On 7/9/14 10:42 AM, Daniel D. Daugherty wrote:
>>> Greetings,
>>>
>>> I have the fix for the following bug ready for JDK9 RT_Baseline:
>>>
>>>     JDK-8049717 expose L1_data_cache_line_size for diagnostic/sanity
>>>                 checks
>>>     https://bugs.openjdk.java.net/browse/JDK-8049717
>>>
>>> Here is the URL for the webrev:
>>>
>>> http://cr.openjdk.java.net/~dcubed/8049717-webrev/0-jdk9-hs-rt/
>>>
>>> This fix is a standalone piece from my Contended Locking reorder
>>> and cache-line bucket. I've split it off as an independent bug fix
>>> in order to make the reorder and cache-line bucket more clear.
>>>
>>> Testing:
>>>
>>> - JPRT test jobs
>>> - manual testing of the new output via existing options:
>>>   -XX:+UnlockExperimentalVMOptions -XX:SyncKnobs=Verbose=1
>>>   -XX:+ExecuteInternalVMTests -XX:+VerboseInternalVMTests
>>> - Aurora Adhoc nsk.sajdi and vm.parallel_class_loading as part of
>>>   testing for my Contended Locking reorder and cache-line bucket
>>>
>>> Thanks, in advance, for any comments, questions or suggestions.
>>>
>>> Dan
>>>
>>>
>>>
>>>
>>
>>
>>
>