RFR: 8323497: On x64, use 32-bit immediate moves for narrow klass base if possible [v7]

[Aleksey Shipilev]

On Fri, 23 May 2025 08:03:34 GMT, Thomas Stuefe <stuefe at openjdk.org> wrote:

>> On x64, we always use the long form of mov immediate to load the klass base into a register. If the klass base fits into 32 bits, we could use the short form and save four instruction bytes. 
>> 
>> Before: mov uses 10 instruction bytes:
>> 
>> 
>>    35  ;; decode_klass_not_null
>>    36   0x00007f8b089e51c4:   movabs $0x82000000,%r11
>>    37   0x00007f8b089e51ce:   add    %r11,%r10
>> 
>> 
>> Now: mov uses 6 instruction bytes:
>> 
>> 
>>    35  ;; decode_klass_not_null
>>    36   0x00007fbe609e51c4:   mov    $0x82000000,%r11d
>>    37   0x00007fbe609e51ca:   add    %r11,%r10
>> 
>> 
>> Note that this optimization does not depend on zero-based addressing, and therefore we change class space reservation: we now always look in low-address regions first.
>> 
>> ----------
>> 
>> Tests: tier1 (GHA), tier 2 on x64 linux
>
> Thomas Stuefe has updated the pull request incrementally with four additional commits since the last revision:
> 
>  - also for encode_klass
>  - fix test
>  - back to unconditional 4g, but with better comment
>  - feedback alexey

This looks fine to me conceptually. I am going to run some tests too. Cosmetics:

src/hotspot/cpu/x86/compressedKlass_x86.cpp line 30:

> 28: #include "oops/compressedKlass.hpp"
> 29: #include "utilities/globalDefinitions.hpp"
> 30: #include "memory/metaspace.hpp"

`make test TEST=sources/` would likely complain about unsorted order.

src/hotspot/cpu/x86/compressedKlass_x86.cpp line 40:

> 38: 
> 39:   // Unconditionally attempting to reserve in lower 4G first makes always sense:
> 40:   // -CDS -COH: try to get unscaled mode (zero base zero shift)

Suggestion:

  // -CDS -COH: Try to get unscaled mode (zero base, zero shift)

src/hotspot/cpu/x86/macroAssembler_x86.cpp line 5418:

> 5416:     shrq(r, CompressedKlassPointers::shift());
> 5417:   }
> 5418:   BLOCK_COMMENT("} encode_and_move_klass_not_null");

Copy-paste nit: this is still `encode_klass_not_null`.

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

PR Review: https://git.openjdk.org/jdk/pull/17340#pullrequestreview-2863599036
PR Review Comment: https://git.openjdk.org/jdk/pull/17340#discussion_r2104069315
PR Review Comment: https://git.openjdk.org/jdk/pull/17340#discussion_r2104068645
PR Review Comment: https://git.openjdk.org/jdk/pull/17340#discussion_r2104066124