RFR(XS): 8230434: [C1, C2] Release barrier for volatile field stores in constructors implemented inconsistently
Doerr, Martin
martin.doerr at sap.com
Mon Sep 2 16:56:30 UTC 2019
Hi,
I have updated the definition of "support_IRIW_for_not_multiple_copy_atomic_cpu" as part of JDK-8229422.
As an additional cleanup, I'd like to use it consistently between C1 and C2 compiler:
http://cr.openjdk.java.net/~mdoerr/8230434_membar_volatile_field_constructor/webrev.00/
I believe the "PPC64_ONLY" in C2 is from a time at which Oracle was maintaining the closed arm port and it should better get removed, now.
Note that this change is only a cleanup, not a functional change because support_IRIW_for_not_multiple_copy_atomic_cpu is only true on PPC64.
But if another platform uses it, it will be inconsistent between C1 and C2. (It could potentially get used by arm in the future.)
Best regards,
Martin
More information about the hotspot-compiler-dev
mailing list