[jdk16] RFR: 8259641: C2: assert(early->dominates(LCA)) failed: early is high enough

[Christian Hagedorn]

On Thu, 14 Jan 2021 08:05:25 GMT, Roland Westrelin <roland at openjdk.org> wrote:

> For the fix of 8258393 that causes this regression, I wrote that the
> goal was:
> 
> "I propose to be less conservative in anti-dependence analysis for
> Phis. For a Phi, when computing the LCA, I think it's sufficient to
> only consider region's inputs that we actually reach by following the
> memory edges and that's what I propose here."
> 
> But that's not what the logic I added does. It checks whether a Phi
> input should be taken into account by checking whether it's in the
> worklist but the worklist doesn't include mem, the memory state from
> which we started.

That looks good to me!

test/hotspot/jtreg/compiler/loopopts/TestBrokenAntiDependenceWithPhi.java line 30:

> 28:  *
> 29:  * @run main/othervm -Xcomp -XX:CompileOnly=TestBrokenAntiDependenceWithPhi TestBrokenAntiDependenceWithPhi
> 30:  *

Empty line can be removed

test/hotspot/jtreg/compiler/loopopts/TestBrokenAntiDependenceWithPhi.java line 32:

> 30:  *
> 31:  */
> 32: 

You could move the test into the `compiler.loopopts` package.

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

Marked as reviewed by chagedorn (Reviewer).

PR: https://git.openjdk.java.net/jdk16/pull/121