RFR(XS) 8234539 ArchiveRelocationTest.java failed: Archive mapping should always succeed

[Ioi Lam]

Hi Calvin,

Thanks for the review. It turned out that I needed to fix another 
(addr_delta == 0) bug in the code. I've also moved the handling of 
ArchiveRelocationMode==1 in debug builds to 
MetaspaceShared::map_archives(). This way, we can simulate the "mapping 
failure" after all archives have been mapped. This way, we can better 
test the code that unmap the archives after the initial mapping failures.

Here's the updated patch.
http://cr.openjdk.java.net/~iklam/jdk14/8234539-mapping-should-always-succeed.v02/

I am running tier4-rt-cds-relocation multiple times to make sure 8234539 
is no longer triggered on Windows.

Thanks
- Ioi

On 11/22/2019 11:23 AM, Calvin Cheung wrote:
> Hi Ioi,
>
> The fix looks good.
>
> thanks,
>
> Calvin
>
> On 11/21/19 2:58 PM, Ioi Lam wrote:
>> https://bugs.openjdk.java.net/browse/JDK-8234539
>> http://cr.openjdk.java.net/~iklam/jdk14/8234539-mapping-should-always-succeed.v01/ 
>>
>>
>> This bug happens only on Windows. The fix is one-line -- in order to 
>> check
>> whether "This is the second time we try to map the archive(s)", 
>> instead of
>> using (addr_delta != 0), the correct condition is (rs.is_reserved()). 
>> Please
>> see the bug report for details.
>>
>> I also improve the log messages when error happens.
>>
>> Thanks
>> - Ioi