RFR (XS): 8220345: Use appropriate type for G1RemSetScanState::IsDirtyRegionState

Kim Barrett kim.barrett at oracle.com
Fri Mar 8 21:04:18 UTC 2019


> On Mar 8, 2019, at 3:37 PM, Thomas Schatzl <thomas.schatzl at oracle.com> wrote:
> 
> Hi,
> 
> On Fri, 2019-03-08 at 14:59 -0500, Kim Barrett wrote:
>> The size of bool is implementation defined (see C++ 5.3.3).  Our
>> atomics support doesn't cover all sizes, though I think we're okay
>> here for currently supported platforms.
>> 
>> OTOH, while there is old code that is "careful" about this, and I was
>> being careful about it, I've noticed there are a number of more
>> recent places using atomic bool.  (I may be guilty of some myself
>> now.)
>> 
>> So I'm not sure how much importance to attach to the potential
>> problem this change is introducing.  It's far from the first, so
>> maybe that ship has sailed.
> 
> I considered whether this is an issue, and I am aware of that bool is
> implementation defined, but thought there wouldn't be any issue.

I’m okay with saying the atomic bool ship has sailed and we can use them,
at least until such time as we get a build failure on some new (and perhaps
peculiar) platform.



More information about the hotspot-gc-dev mailing list