[jdk8u-dev] RFR: 8262017: C2: assert(n != __null) failed: Bad immediate dominator info.

[Martin Balao]

On Wed, 26 Jun 2024 16:12:08 GMT, Francisco Ferrari Bihurriet <fferrari at openjdk.org> wrote:

> Hi, this is a backport of [JDK-8262017: C2: assert(n != __null) failed: Bad immediate dominator info](https://bugs.openjdk.org/browse/JDK-8262017). As explained in the 11u backport (openjdk/jdk11u-dev#156), this also introduces parts from [JDK-8244504](https://bugs.openjdk.org/browse/JDK-8244504 "C2: refactor counted loop code in preparation for long counted loop").
> 
> This backport is not clean because of the changed paths ([JDK-8187443](https://bugs.openjdk.org/browse/JDK-8187443 "Forest Consolidation: Move files to unified layout")) and the removal of the `Compile* C` parameter from the nodes allocation `new` operator ([JDK-8034812](https://bugs.openjdk.org/browse/JDK-8034812 "remove IDX_INIT macro hack in Node class")). Besides the conflicts, new code in `opto/addnode.cpp` had to be adjusted to introduce the mentioned `Compile* C` parameter.
> 
> Finally, the `opto/movenode.hpp` include in `opto/addnode.cpp` had to be removed. `opto/movenode.hpp` doesn't exist in 8u, and `CMoveLNode` is defined in `opto/connode.hpp` instead, which is already included by `opto/addnode.cpp`.
> 
> Apart from those cosmetic adjustments, the change is not substantially different from openjdk/jdk11u-dev at 293d44f6cf07b05ad5e3a8b0c08b244f1697700d.
> 
> 
> ## Testing
> 
> * `hotspot_tier1` showed no regression with a _Linux_ x86-64 slowdebug build
>     * `hotspot/test/compiler/ciReplay/TestSA.sh` was the only failing test, with and without this PR changes
> * In addition, the new `hotspot/test/compiler/rangechecks/TestRangeCheckLimits.java` is passing

Thanks for proposing this backport. Looks good to me.

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

Marked as reviewed by mbalao (Reviewer).

PR Review: https://git.openjdk.org/jdk8u-dev/pull/528#pullrequestreview-2142437351