RFR: 8349764: RISC-V: C1: Improve Class.isInstance intrinsic [v2]

[Gui Cao]

On Wed, 12 Feb 2025 12:01:46 GMT, Hamlin Li <mli at openjdk.org> wrote:

> in this patch, maybe we should consistently use `c_rarg`n or `x1`n rather than mix these 2 types of register names?

Fixed.

> This is also a failure case? Could it bring some benefit to jump to `fail` below or put the `fail` target here (i.e. remove the `fail` below)?

Thanks for your review. In fact I don't know if this is a failure case here, I tried jumping to success or failure first(beq(klass, x17, success)/bne(klass, x17, fail)) and the performance results had little impact.

> An extra empty line above this would help read code.

Fixed.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/23551#discussion_r1953046428
PR Review Comment: https://git.openjdk.org/jdk/pull/23551#discussion_r1953046073
PR Review Comment: https://git.openjdk.org/jdk/pull/23551#discussion_r1953046699