RFR: 8332689: RISC-V: Use load instead of trampolines

[Robbin Ehn]

On Fri, 31 May 2024 06:02:56 GMT, Fei Yang <fyang at openjdk.org> wrote:

> Interesting. I will take a look and play with it on my machines.

Thanks. I don't expected this will be done before RDP1, so take your time.

Note that this also somewhat fixes issues with out of order machines spectulation.
I.e. the **dest** in the trampoline may very well be I-fetched and decoded which may trigger pipeline flush or similar if it's decoded as something crazy to the CPU. (obviously very implementation dependent)

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

PR Comment: https://git.openjdk.org/jdk/pull/19453#issuecomment-2141304728