[llvm] [AArch64] Enable masked load/store for Streaming-SVE with -march=armv8-a+sme (PR #163133)

[Shikhar Jain via llvm-commits]

ShikharjQUIC wrote:

> Are you trying to enable auto-vectorization for streaming[-compatible] functions?

Hi
This PR aims to emit/enable masked load/store (LD/ST) instructions when a function runs in streaming mode(with scalable pragma).
This especially help the loops with control flow where predication depends on vector loads/stores.
Please refer to [this bug report](https://github.com/llvm/llvm-project) which I have raised — this PR resolves this issue.  




https://github.com/llvm/llvm-project/pull/163133