RFR (XL) [8u20]: Tests on Intel RTM instructions for locks
Filipp Zhinkin
filipp.zhinkin at oracle.com
Mon Apr 14 17:46:57 UTC 2014
Thank you, Vladimir.
Filipp.
On 04/14/2014 09:53 PM, Vladimir Kozlov wrote:
> Looks good.
>
> Thanks,
> Vladimir
>
> On 4/14/14 9:30 AM, Filipp Zhinkin wrote:
>> Hi all,
>>
>> I'd like to backport tests on Intel RTM instructions for locksinto 8u20.
>> In 8u20 all RTM-related flags are experimental, butin 9 three flags
>> are product,
>> so tests were updated to take that difference into account.
>>
>> Also, we need to backport 8007270, because it includes changes in
>> testlibrary
>> required by 8039497.
>>
>> Here is detailed description of changesets to be backported in the
>> same order in
>> which it should be integrated:
>>
>> - 8007270: Make IsMethodCompilable test work with tiered.
>> Tests were updated, because in 8u20 some of these changes were
>> done as part
>> of 8027124:
>> http://cr.openjdk.java.net/~iignatyev/fzhinkin/rtm/8u20/8007270/webrev.00/
>>
>>
>> - 8039497: Testlibrary should be updated to provide information about
>> all
>> VM types as well as access to Unsafe.
>> This fix applies without any changes, so webrev is the same as for 9:
>> http://cr.openjdk.java.net/~iignatyev/fzhinkin/rtm/8039497/webrev.00/
>>
>> - 8039499: Add all common classes used by tests on RTM support to
>> testlibrary
>> RTMTestBase::executeRTMTest was updated to pass
>> '-XX:+UnlockExperimentalVMOptions' before '-XX:+UseRTMLocking':
>> http://cr.openjdk.java.net/~iignatyev/fzhinkin/rtm/8u20/8039499/webrev.00/
>>
>>
>> - 8037860: Add tests to cover Intel RTM instructions support
>> This fix applies without any changes, so webrev is the same as for 9:
>> http://cr.openjdk.java.net/~iignatyev/fzhinkin/rtm/8037860/webrev.01/
>>
>> - 8039496:Add sanity tests on RTM-related command line options
>> Tests on 'product-in-9' options were updated to expect thatthese
>> options
>> are experimental in 8u20:
>> http://cr.openjdk.java.net/~iignatyev/fzhinkin/rtm/8u20/8039496/webrev.00/
>>
>>
>> Thanks,
>> Filipp.
More information about the hotspot-compiler-dev
mailing list