RFR: 8333840: C2 SuperWord: wrong result for MulAddS2I when inputs permuted [v5]

[Emanuel Peter]

On Thu, 13 Jun 2024 14:05:38 GMT, Christian Hagedorn <chagedorn at openjdk.org> wrote:

>> Emanuel Peter has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains 13 commits:
>> 
>>  - manual merge with master
>>  - review changes for Vladimir
>>  - more refactoring
>>  - improve naming
>>  - move code
>>  - rm isa_pack_input_or_null, leave that to a refactoring
>>  - refactoring / simplification
>>  - refactor into is_muladds2i_pack_with_pack_inputs
>>  - fix done, need refactoring
>>  - more partial fix and more test
>>  - ... and 3 more: https://git.openjdk.org/jdk/compare/5d2a19de...2155d18b
>
> src/hotspot/share/opto/superword.cpp line 2851:
> 
>> 2849: }
>> 2850: 
>> 2851: Node_List* PackSet::strided_pack_input_at_index_or_null(const Node_List* pack, const int index, const int stride, const int offset) const {
> 
> Do you plan to use this method at other places as well? Otherwise, you could hardcode `stride = 2` in the context of `MulAddS2I`.

Yes, I plan to use it with `stride = 1` and `offset = 0`. That would be a normal input pack. I will refactor in a future RFE.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/19619#discussion_r1638326502