RFR: 8331711: G1 doesn't need pre write barrier for stores from new allocated objects [v2]
Erik Österlund
eosterlund at openjdk.org
Thu May 9 09:31:52 UTC 2024
On Thu, 9 May 2024 04:47:53 GMT, Liang Mao <lmao at openjdk.org> wrote:
> > Did you check how many of the stores where g1_can_remove_pre_barrier said false and you would have said true, were elided anyway during store capturing (cf. InitializeNode::capture_store), or as part of G1BarrierSetC2::eliminate_gc_barrier? In other words, how many barriers are you eliding, that were not in fact already elided, just a bit later on?
>
>
>
> There is only 1 store that g1_can_remove_pre_barrier return false and was elided by this PR in JBB.
Okay. That's what I expected. Given that we are about to remove all of this code in favour of more robust late barrier expansion, I feel like we can live without that one extra store barrier for now.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/19098#discussion_r1595206949
More information about the hotspot-gc-dev
mailing list