RFR (S) 8218140: Build failures after JDK-8218041 (Assorted wrong/missing includes)
Stefan Karlsson
stefan.karlsson at oracle.com
Thu Jan 31 15:43:02 UTC 2019
On 2019-01-31 16:23, Aleksey Shipilev wrote:
> On 1/31/19 2:30 PM, Stefan Karlsson wrote:
>> On 2019-01-31 13:40, Aleksey Shipilev wrote:
>>> ...and shared files have to include that thread.inline.hpp then.
>>
>> This is really awkward. Some platforms require thread.hpp while others require thread.inline.hpp. I
>> think we should enforce the same include requirement for all platforms, otherwise these kind of
>> problems will happen again.
>
> Yes, it is awkward.
>
>> We "recently" removed the orderAccess_<platform>.inline.hpp to allow OrderAcess to be used in
>> headers. Maybe it's time to simply move the code above to thread.hpp? That would remove this source
>> of compile errors.
>
> I am thinking the reverse: push #ifs around the definition into the method body instead. This would
> ensure we use thread.inline.hpp where it makes sense to.
Not sure what you mean. Maybe we're saying the same thing. My proposal
is this (untested):
http://cr.openjdk.java.net/~stefank/8218140/webrev.alt.01/
But let's make current repository buildable
> first, and it would also put thread.inline.hpp includes where appropriate.
Sure.
StefanK
>
> -Aleksey
>
>
More information about the hotspot-dev
mailing list