RFR: 8333258: C2: high memory usage in PhaseCFG::insert_anti_dependences() [v2]

[Vladimir Kozlov]

On Tue, 25 Jun 2024 15:20:20 GMT, Roland Westrelin <roland at openjdk.org> wrote:

> And we push uses of store_n only if it is a MergeMem

You are right, I missed `op == Op_MergeMem ` check.

Yes, I thought about loop since we check for Phi node. But we have `worklist_visited` check for `MergeMem` which will not allow to push the same `mem1` on work list. It look like you are right - there should be only sequential elements in work list for the same memory.
I withdraw my comment.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/19791#discussion_r1653069980