[llvm] [RISCV] Make Zicclsm imply unaligned scalar and vector access (PR #108551)

[Yingwei Zheng via llvm-commits]

dtcxzyw wrote:

> Zicclsm makes no guarantees about performance of unaligned access. It only says they won't trap. They can be emulated in the kernel and take thousands of cycles.

See also https://patchew.org/linux/20240625005001.37901-1-jesse@rivosinc.com/20240625005001.37901-2-jesse@rivosinc.com/

> Note:
> This introduces a new extension name for this feature.
> This requires misaligned support for all regular load and store
> instructions (including scalar and vector) but not AMOs or other
> specialized forms of memory access. Even though mandated, misaligned
> loads and stores might execute extremely slowly. Standard software
> distributions should assume their existence only for correctness,
> not for performance.

https://github.com/llvm/llvm-project/pull/108551