RFR: 8352893: C2: OrL/INode::add_ring optimize (x | -1) to -1 [v2]

[Emanuel Peter]

On Mon, 31 Mar 2025 14:02:45 GMT, Manuel Hässig <duke at openjdk.org> wrote:

>> # Issue Summary
>> 
>> The `add_ring()` implementations of `OrINode` and `OrLNode` are missing the optimization that an or with a value where all bits are ones (since we have signed integers in this case `~0 == -1`) will always yield all zeroes.
>> 
>> # Changes
>> 
>> This PR makes the following straight forward changes:
>>  - `Or(I|L)Node::add_ring()` returns `-1` if one of the two inputs is `-1`.
>>  - Add `Or(I|L)` nodes to the IR framework.
>>  - Add a regression IR test for the implemented optimization.
>> 
>> # Testing
>> 
>> - [Github Actions](https://github.com/mhaessig/jdk/actions/runs/14110978686)
>> - Ran tier1 through tier3 and Oracle internal testing
>
> Manuel Hässig has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Fix bug number in regression test

@mhaessig It looks good, just two comments below :)

test/hotspot/jtreg/compiler/integerArithmetic/TestOrSaturate.java line 32:

> 30:  /*
> 31:   * @test
> 32:   * @bug 8352839

Suggestion:

  * @bug 8352893

Werchstabeverbuechslig ;)

test/hotspot/jtreg/compiler/integerArithmetic/TestOrSaturate.java line 47:

> 45:     @Run(test = {"testL", "testI", "testDelayed"})
> 46:     public static void check() {
> 47:         for (int i = 0; i < WARMUP; i++) {

Do you actually need the WARMUP here? It does not look like this is a standalone test, so I thought the `@Run` actually gets called many times. Not sure if that is correct...

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

Changes requested by epeter (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/24289#pullrequestreview-2729391754
PR Review Comment: https://git.openjdk.org/jdk/pull/24289#discussion_r2021095658
PR Review Comment: https://git.openjdk.org/jdk/pull/24289#discussion_r2021099459