RFR: 8341471: Reversed field layout caused by unstable sorting [v3]

[Julian Waters]

On Tue, 8 Oct 2024 14:13:09 GMT, Fei Gao <fgao at openjdk.org> wrote:

>> For class Test:
>> 
>>     public class Test {
>>         char a000;
>>         char a001;
>>         char a002;
>>         char a003;
>>         char a004;
>>         char a005;
>>         char a006;
>>         char a007;
>>         char a008;
>>         char a009;
>>         char a00a;
>>         char a00b;
>>     }
>> 
>> 
>> We found its field layout on macOS was:
>> 
>> 
>> Layout of class Test
>>   Instance fields:
>>    @0 12/- RESERVED
>>    @12 "a00b" C 2/2 REGULAR
>>    @14 "a001" C 2/2 REGULAR
>>    @16 "a002" C 2/2 REGULAR
>>    @18 "a003" C 2/2 REGULAR
>>    @20 "a004" C 2/2 REGULAR
>>    @22 "a005" C 2/2 REGULAR
>>    @24 "a006" C 2/2 REGULAR
>>    @26 "a007" C 2/2 REGULAR
>>    @28 "a008" C 2/2 REGULAR
>>    @30 "a009" C 2/2 REGULAR
>>    @32 "a00a" C 2/2 REGULAR
>>    @34 "a000" C 2/2 REGULAR
>> 
>> 
>> `a000` was put in the end while `a00b` was put in the beginning.
>> 
>> Fields get sorted according to size in [1]. `qsort()` on macOS reverses the order of fields with the same size. We should extend the comparison function to preserve the order on macOS, as we did on Windows.
>> 
>> Tier 1-3 passed on macOS.
>> 
>> [1] https://github.com/openjdk/jdk/blob/3ee94e040a7395d11a294a6b660d707c97f188f8/src/hotspot/share/classfile/fieldLayoutBuilder.cpp#L102
>
> Fei Gao has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Add comments for if branch

In my opinion it was sufficient to use it only for macOS and Windows, but I don't mind it for all platforms

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

Marked as reviewed by jwaters (Committer).

PR Review: https://git.openjdk.org/jdk/pull/21382#pullrequestreview-2356509189