[llvm] ISel/AArch64: custom lower vector ISD::[L]LRINT (PR #89035)

[David Green via llvm-commits]

https://github.com/davemgreen commented:

Thanks, this does looks good to me. We usually add Neon codegen before adding SVE and fixed-length SVE on top to help keep things simple, but as the code in this case is the same for both I think this should be OK all at once.

I would recommend moving where/what gets marked as Custom.

The other thing to consider is that as `long` can be i32 on some platforms, there could be i32 variants as well as the i64's versions. Considering the code isn't specific to a certain size I believe that should be OK.

https://github.com/llvm/llvm-project/pull/89035