[jdk11u-dev] RFR: 8307572: AArch64: Vector registers are clobbered by some macroassemblers [v2]

[Ningsheng Jian]

On Tue, 11 Jul 2023 09:29:26 GMT, Andrew Haley <aph at openjdk.org> wrote:

>> Ningsheng Jian has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains two additional commits since the last revision:
>> 
>>  - Merge 'origin/jdk11u-dev' into jdk11u-8307572-backport to trigger GHA
>>  - Backport 33d9a857308eed53e06b448691910bc8aa2f8fc9
>>    
>>    jdk11 also has the bug of JDK-8307572, so I propose to backport the fix
>>    to jdk11u. This is not a clean backport. Main changes:
>>    
>>    1) Definition of reg_class v4_reg..v7_reg, and operand
>>    vRegD_V4..vRegD_V7 are missing in jdk11u aarch64.ad, which was
>>    introduced in JDK-8214527. Added those definitions in this patch, as we
>>    need to claim them to be killed in match rules.
>>    
>>    2) JDK-8274243 (not a valid backport for 11u) changed
>>    MacroAssembler::encode_iso_array() a lot, but the bug of invalid use of
>>    v4/v5 still exists before and after that patch. Fixed the conflicts by
>>    using 11u naming convention.
>>    
>>    No new jtreg test failure with both release and fastdebug build. Test
>>    case added in JDK-8307572 passes now, which fails without the backport.
>
> Marked as reviewed by aph (Reviewer).

Thank you @theRealAph for the review!

All checks are clean with jdk11u-fix-yes approval in JBS. Can someone please help to sponsor? Thanks!

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

PR Comment: https://git.openjdk.org/jdk11u-dev/pull/2032#issuecomment-1631946170