RFR: 8351159: Remaining cleanups in cpu/x86 after 32-bit x86 removal
Anton Seoane Ampudia
aseoane at openjdk.org
Mon Oct 27 10:02:33 UTC 2025
On Mon, 27 Oct 2025 08:52:12 GMT, Stefan Karlsson <stefank at openjdk.org> wrote:
>> This PR carries out cleanup on the `cpu` and `os_cpu` directories after the removal of the 32-bit x86 architecture from HotSpot.
>>
>> Mainly, code guarded by `IA32` or `X32`, or `!AMD64` or `!LP64` in a x86 context, has been removed. These guards are now redundant (e.g. checking for 64bit on a x86 context) or the code blocks never executed (e.g. code under `#ifdef X32`). Additionally, x86 AD files have been merged.
>>
>> Please note that this changeset limits itself to changes under `cpu/x86` and `os_cpu/x86`. Other cleanups are filed under a different RFE, and consequently addressed in a different PR: [JDK-8351149](https://github.com/openjdk/jdk/pull/27990).
>>
>> **Testing:** passes tiers 1-5
>
> src/hotspot/cpu/x86/c2_globals_x86.hpp line 49:
>
>> 47: define_pd_global(intx, LoopPercentProfileLimit, 10);
>> 48: define_pd_global(intx, InteriorEntryAlignment, 16);
>> 49: define_pd_global(size_t, NewSizeThreadIncrease, ScaleForWordSize(4*K));
>
> FWIW, I think the comment previously only applied to MaxRAM but now it sounds like it is meant for both:
>
> // Ergonomics related flags
> define_pd_global(uint64_t, MaxRAM, 128ULL*G);
> define_pd_global(intx, RegisterCostAreaRatio, 16000);
>
> Is it meant for both?
I believe so, it is the same in aarch64 and riscv c2 globals files
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/27987#discussion_r2465063690
More information about the hotspot-dev
mailing list