On Wed, 31 Aug 2022 08:05:03 GMT, Aleksey Shipilev <sh...@openjdk.org> wrote:
>> It's awkward to manage this across various hardware, esp. if the hardware is >> less mature (perhaps combined with a less mature HotSpot implementation). >> The approach looks reasonable, but it would be nice if we could just >> increase the retries and waiting values on weak LL/SC hardware or at least >> follow up with something like that later on. >> >> As a quick experiment could we annotate fields with `@Contended` and see if >> that reduces the number of retries? > > @PaulSandoz, @TheRealMDoerr -- are you okay with this? @shipilev yes, seems reasonable, since spurious failures are rare on other platforms. Good effort digging into the RISC-V issues. ------------- PR: https://git.openjdk.org/jdk/pull/9889
