8230392: Define AArch64 as MULTI_COPY_ATOMIC

[Andrew Haley]

On 1/24/20 1:23 PM, David Holmes wrote:
> On 24/01/2020 7:06 pm, Andrew Haley wrote:
>> On 1/23/20 10:41 PM, David Holmes wrote:
>>> This does make me wonder whether other lock-free code in the VM needs
>>> special handling for non-CPU_MULTI_COPY_ATOMIC ??
>>
>> I don't think we should do exactly that. I would have thoght that
>> where an algorithm needs anything stronger than acquire/release
>> consistency we should use explicit sequentially-consistent loads and
>> stores.
> 
> What I meant was, "I wonder whether other lock-free code in the VM is 
> assuming multi-copy-atomicity?".

Yes, I did understand that. What I meant to say is that even if we could
find it (which we probably can't) we shouldn't special-case it.

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