On Fri, 22 Sep 2023 16:44:07 GMT, Andrew Haley <a...@openjdk.org> wrote:
> I dunno Robbin, is this really the sort of stuff we should be doing? Binutils > is looking less useful as a base for hsdis. I see your problem, but I can't > help wondering if rather than trying to keep up with Binutils it'd be a > better use of time to work on keeping Capstone up to date. I agree, we shouldn't be doing this. The hsdis-binutils should probably be in binutils, similar to perf jvmti agent in kernel tools/perf/jvmti. Capstone is not worked on by any rv folks, it had a init push of 'base' isa 4y ago and left at that. Just this year it will around 5-10 new extension to the ISA, with major ones as vector. Which means it would be us hotspot de. that needs to everything in capstone. If so it would be easier to add the disassembler to hotspot; we can unittest that you added the disassembler part for your new instruction in the assembler. ------------- PR Comment: https://git.openjdk.org/jdk/pull/15138#issuecomment-1733056380
