On Thu, 5 Feb 2026 11:51:18 GMT, Ferenc Rakoczi <[email protected]> wrote:

>> Ben Perez has updated the pull request incrementally with one additional 
>> commit since the last revision:
>> 
>>   Created subroutine for 32 bit vector multiplication
>
> src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp line 7820:
> 
>> 7818:     //   IntegerPolynomialP521:  19 = 8 + 8 + 2 + 1
>> 7819:     //   P521OrderField:         19 = 8 + 8 + 2 + 1
>> 7820:     // Special Cases 5, 10, 14, 16, 19
> 
> Add a comment in the Java code that the intrinsic can only be used for these 
> lengths. I would also change the Java code to use an intermediate method that 
> has an assert checking the allowed lengths and calls the @IntrinsicCandidate 
> conditionalAssign() method (this is an easy change since there is only one 
> caller in the current JVM code).

Good catch - I just noticed that on the x86 side there's a "default" branch so 
I could do that for aarch64 as well. Checking on the java side might be simpler 
though

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

PR Review Comment: https://git.openjdk.org/jdk/pull/27946#discussion_r2771123481

Reply via email to