RFR 8231264: Disable biased-locking and deprecate all flags related to biased-locking
Andrew Haley
aph at redhat.com
Sun Nov 24 17:09:41 UTC 2019
On 11/22/19 11:49 AM, Erik Österlund wrote:
> So yeah, loom. There is that. That is what I wanted to add to this
> conversation.
Hmm. I'm working on Loom, and the last thing I want is for the Loom
patch to be a performance regression for OpenJDK as a whole. That
would be a *terrible* message, apart from any other consideration.
Certainly there are problems when running Loom virtual threads with
biased locking enabled, but that is another matter.
Maybe there are other ways to solve the problem. We could arrange it
so that any attempt to create a virtual Thread caused a mass bias
revocation, for example. If we also set BiasedLockingStartupDelay to,
say 4000ms, any application which started a virtual Thread before that
time would effectively prevent biased locking from ever being used.
--
Andrew Haley (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671
More information about the hotspot-runtime-dev
mailing list