RFR(M) 8191375 Add high-level jtreg VMProps to filter out CDS tests

Jiangli Zhou jiangli.zhou at oracle.com
Thu Feb 15 19:18:05 UTC 2018


> On Feb 15, 2018, at 11:17 AM, Ioi Lam <ioi.lam at oracle.com> wrote:
> 
> 
> 
> On 2/15/18 11:11 AM, Jiangli Zhou wrote:
>> Hi Ioi,
>> 
>> HeapRegion is ambiguous. How about renaming WB_areHeapRegionsSupportedByCDS() to WB_isJavaHeapArchiveSupportedByCDS() or just WB_isJavaHeapArchiveSupported()? 
>> 
>> And rename vm.cds.archived.heap to vm.cds.archived.java.heap, vmCDSForArchivedHeap() to vmCDSForArchivedJavaHeap()?
>> 
> OK, I will change that.
>> You can remove the following line from OpenArchiveRegion.java.
>>  29  * @requires (vm.gc=="null")
>> 
> This one is needed because the test uses -XX:+UseParallelGC.

Ok.

Thanks,

Jiangli

> 
> Thanks
> - Ioi
> 
>> The rest looks good.
>> 
>> Thanks,
>> Jiangli
>> 
>>> On Feb 15, 2018, at 4:55 AM, Ioi Lam <ioi.lam at oracle.com <mailto:ioi.lam at oracle.com>> wrote:
>>> 
>>> https://bugs.openjdk.java.net/browse/JDK-8191375 <https://bugs.openjdk.java.net/browse/JDK-8191375>
>>> http://cr.openjdk.java.net/~iklam/jdk11/8191375-cds.tests.vmprop.v01/ <http://cr.openjdk.java.net/~iklam/jdk11/8191375-cds.tests.vmprop.v01/>
>>> 
>>> Summary:
>>> 
>>> 1. Added the new property "@require vm.cds.archived.heap". This replaces various
>>>    ad-hoc settings like the following, which is hard to maintain:
>>> 
>>>    * Feature support: G1GC only, compressed oops/kptrs, 64-bit os, not on windows
>>>    * @requires (sun.arch.data.model != "32") & (os.family != "windows")
>>>    * @requires vm.cds
>>>    * @requires vm.gc.G1
>>> 
>>> 
>>> 2. Also did some clean up such as removing duplicated lines of "@require vm.cds"
>>>    in some tests
>>> 
>>> 3. Removed explicit settings of -XX:-UseCompressedOops, -XX:+UseGCGC, etc, that
>>>    are now implied by "@require vm.cds" or "@require vm.cds.archived.heap"
>>> 
>>> 4. Removed unnecessary '@require vm.gc == "null"' from a few test cases.
>>>    (It is only needed by tests that explicitly choose GC types, but somehow it's
>>>    been leaked to other tests due to copy-paste).
>>> 
>>> I've tested locally with jtreg:
>>> 
>>>   - no explicit setting of GC
>>>   - explicitly set GC with -vmoptions:-XX:+UseG1GC
>>>   - explicitly set GC with -vmoptions:-XX:+UseParallelGC
>>> 
>>> Testing with mach5 with tiers 1-3 now.
>>> 
>>> Thanks
>>> - Ioi
>> 
> 



More information about the hotspot-runtime-dev mailing list