On Sat, 26 Sep 2026 07:47:16 GMT, Shawn Emery <[email protected]> wrote:
> This change-set cleans up the generate_double_keccak_gpr() and > generate_sha3_implCompress_gpr() AArch64 stub generator methods to use a > register set instead of the previous version of the code which hard-coded > these mappings. It also implements register poisoning to future-proof the > code in the keccak_round_gpr() method. > > --------- > - [X] I confirm that I make this contribution in accordance with the [OpenJDK > Interim AI Policy](https://openjdk.org/legal/ai). Looks good. What have you run to test it? src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp line 5127: > 5125: // a0..a24 are respective locals from SHA3.java > 5126: RegSetIterator<Register> regs = (RegSet::range(r0, r28) - > r18_tls).begin(); > 5127: Register a[25]; It's worth keeping the comment here and restating it at line 9103 Suggestion: // a[0]...a[24] correspond to locals a0...a24 from SHA3.java Register a[25]; ------------- PR Comment: https://git.openjdk.org/jdk/pull/33080#issuecomment-5866204150 PR Review Comment: https://git.openjdk.org/jdk/pull/33080#discussion_r4119984122
