RFR: 8292867: RISC-V: Simplify weak CAS return value handling

Fei Yang fyang at openjdk.org
Thu Aug 25 02:55:19 UTC 2022


On Wed, 24 Aug 2022 19:23:29 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> I see that current weak CAS implementations in `macroAssembler_riscv.cpp` return `0` on success, and then invert it in `riscv.ad` to match the Java's `boolean` result. I think we can just return the final value from `MacroAssembler`.
> 
> Additional testing:
>  - [ ] Linux RISC-V fastdebug `java/lang/invoke/VarHandles/`
>  - [ ] Linux RISC-V fastdebug `compiler/unsafe`

Looks fine. I checked that we covered all the callsites of the macro assembler function.

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

Marked as reviewed by fyang (Reviewer).

PR: https://git.openjdk.org/jdk/pull/10006


More information about the hotspot-dev mailing list