On Wed, 25 Nov 2020 14:07:04 GMT, Rémi Forax <github.com+828220+fo...@openjdk.org> wrote:
>> Jim Laskey has updated the pull request incrementally with one additional >> commit since the last revision: >> >> 8248862: Implement Enhanced Pseudo-Random Number Generators >> >> Changes to RandomGeneratorFactory requested by @PaulSandoz > > src/java.base/share/classes/java/util/random/RandomGeneratorFactory.java line > 335: > >> 333: ctorBytes = tmpCtorBytes; >> 334: ctorLong = tmpCtorLong; >> 335: ctor = tmpCtor; > > This one is a volatile write, so the idea is that ctorBytes and ctorLong does > not need to be volatile too, which i think is not true if there is a code > somewhere that uses ctorBytes or ctorLong without reading ctor. > This code is too smart for me, i'm pretty sure it is wrong too on non TSO cpu. The 2 uses call ensureConstructors, which calls this method, which reads ctor. The memory model guarantees this to be safe, even on non tso hardware. https://shipilev.net/blog/2016/close-encounters-of-jmm-kind/#pitfall-avoiding ------------- PR: https://git.openjdk.java.net/jdk/pull/1292
