On Fri, 8 May 2026 01:33:19 GMT, Dean Long <[email protected]> wrote:
>> Coleen Phillimore has updated the pull request incrementally with two
>> additional commits since the last revision:
>>
>> - include another file.
>> - Some forward declarations might do the trick.
>
> src/hotspot/cpu/arm/macroAssembler_arm.hpp line 33:
>
>> 31: class ciInlineKlass;
>> 32: class SigEntry;
>> 33: class VMRegPair;
>
> These could all be safely moved to macroAssembler_common.hpp I think. But I
> see this is how the other CPUs do it.
macroAssembler_common.hpp is included inside MacroAssembler declaration, so
these forward declarations would/might be nested classes instead.
> src/hotspot/share/adlc/output_h.cpp line 1634:
>
>> 1632: instr->is_ideal_call() != Form::JAVA_LEAF) {
>> 1633: // MachConstantBase goes behind arguments, but before jvms.
>> 1634: fprintf(fp,"assert(tf() && tf()->domain_cc(), \"\"); return
>> tf()->domain_cc()->cnt();");
>
> This seems like a separate fix unrelated to non-Oracle platforms.
One of the non-Oracle platforms included this code and I guess the others
didn't.
-------------
PR Review Comment:
https://git.openjdk.org/valhalla/pull/2396#discussion_r3208911532
PR Review Comment:
https://git.openjdk.org/valhalla/pull/2396#discussion_r3208927958
