RFR(M): 8209420: Track membars for volatile accesses so they can be properly optimized

Andrew Haley aph at redhat.com
Tue Aug 14 09:23:04 UTC 2018


On 08/14/2018 09:22 AM, Roland Westrelin wrote:
> 
>>> 1) the store saves a value that was just loaded from the same address
>>
>> That doesn't sound right.  If the accesses are volatile then every
>> core should see them in the same total order.  This IRIW problem should
>> be detectable in jcstress.  I'll do a run.
> 
> As replied to Andrew D. only volatile stores to non escaping objects
> should be affected. FWIW, I did a full jcstress run.

OK.

-- 
Andrew Haley
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671


More information about the hotspot-compiler-dev mailing list